article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
ever since the work by dressler et al . ( 1980 ) presenting the first morphology - density relation in local galaxy clusters , the projected number density of galaxies ( the number of galaxies per unit area or volume ) has been a primary tool for investigating the relation between galaxy properties and their environment .
thirty years on , this tool has been applied to galaxies in clusters , groups and the general field , from the local to the high - z universe , to study the systematic variations of galaxy morphologies , colors , current star formation activity , stellar masses and other galaxy properties with environment ( postman & geller 1984 , dressler et al .
1997 , hashimoto et al .
1998 , lewis et al .
2002 , gomez et al .
2003 , blanton et al .
2003 , kauffmann et al .
2004 , balogh et al .
2004 , hogg et al .
2004 , postman et al .
2005 , baldry et al .
2006 , cucciati et al . 2006 , cooper et al .
2007 , elbaz et al .
2007 , cooper et al .
2008 , tasca et al .
2009 , bolzonella et al .
2009 , to name a few ) .
more and more sophisticated methods to characterize the `` local '' galaxy density have been scrutinized , and by now a variety of density estimators have been employed in the literature , including the stellar mass density ( wolf et al .
2009 ) , the dark matter density field ( gray et al .
2004 ) , galaxy counts within a fixed metric projected radius and a fixed recession velocity difference ( blanton et al .
2003 , hogg et al .
2004 , kauffmann et al . 2004 ) and the projected @xmath3th - nearest neighbour distance , with the optimal @xmath3 varying with environment and survey characteristics , sometimes expressed as an overdensity with respect to the median density in the survey ( eg . see cooper et al . 2005 and kovac et al .
2009 for thorough discussions of different methods ) .
widely different methods explore different `` environmental '' conditions and can probe very different physical scales , in some cases over several mpc , which are far from being a measure of the `` local '' environment as in the traditional , close neighbour studies in clusters .
nevertheless , all studies employing the galaxy density share the goal of uncovering the environmental dependence of galaxy properties , and have successfully proved it to be strong .
in contrast , little attention has so far been devoted to _ how and why density conditions change with redshift_. observationally , density distributions in clusters at different redshifts have never been compared . in the field , an overdensity is usually measured , factoring out the evolution of the mean density of the universe , and the evolution of the distribution of environmental conditions beyond that has never been investigated .
theoretical efforts have so far focused on uncovering the physical origin of the observed relations between galaxy properties and density , or have been compared with observations to test the model consistency , or to study the relation between density and halo mass ( e.g. kauffmann et al .
2004 , berlind et al .
2005 , baldry et al .
2006 , elbaz et al .
2008 , gonzalez & padilla 2009 )
. a knowledge of the density distribution of galaxies as a function of redshift , and as a function of system mass in virialized systems , would be very useful to assess the changes of local environment experienced by galaxies , but it is currently lacking . in this paper , we analyze the evolution of galaxy densities in clusters , starting from the observed density distribution in distant and nearby clusters .
we employ the most traditional of all density estimators , the 10 ( or 3 ) closest neighbours density , as in the original dressler 1980 work and in most subsequent cluster studies .
we compare observations with millennium simulation results in order to explain the observed change of the cluster density distribution with redshift , and to relate the evolution in galaxy number density with that of the matter density .
all cluster velocity dispersions @xmath4 are given in the rest frame .
we assume a @xmath5cdm cosmology with ( @xmath6 , @xmath7 , @xmath8 ) = ( 73 @xmath9 , 0.25 , 0.75 ) .
in this paper we exploit two samples of galaxy clusters : the eso distant cluster survey ( hereafter , ediscs ) at high - z , and a low - z sample from the sdss .
we use the projected local galaxy densities and the total number of members in these clusters as key observables for our study . the eso distant cluster survey ( hereafter , ediscs )
is a photometric and spectroscopic survey of galaxies in 20 fields containing galaxy clusters at @xmath10 .
an overview of the goals and strategy of this survey is given in white et al .
( 2005 ) who present the sample selection and the optical ground based photometry . for all 20 fields
ediscs has obtained deep optical photometry with fors2/vlt , near - ir photometry with sofi / ntt , multislit spectroscopy with fors2/vlt , and mpg / eso 2.2/wfi wide field imaging in @xmath11 .
acs / hst mosaic imaging in @xmath12 of 10 of the highest redshift clusters has also been acquired ( desai et al .
2007 ) , and a number of multiwavelength follow - up studies have been conducted ( see poggianti et al .
2009 for an overview ) .
spectra of @xmath13 galaxies per cluster field were obtained for 18 out of the 20 cluster fields .
the spectroscopic selection , observations , and catalogs are presented in halliday et al . (
2004 ) and milvang - jensen et al .
( 2008 ) , together with the cluster velocity dispersions .
spectroscopic completeness functions and [ oii ] line properties as a function of cluster mass are discussed in poggianti et al .
( 2006 ) . in this paper
we use the projected local galaxy densities computed by poggianti et al .
( 2008 ) , who present the star formation - density and morphology - density relations .
our sample consists of 18 clusters and groups at @xmath14 covering a wide range of velocity dispersions ( @xmath15 ) , as listed in table 1 . at low - z
, we use a compilation of 23 clusters and groups at @xmath16 drawn from the sdss , spanning a similar range of velocity dispersions to ediscs . the sample selection , completeness and data used
are fully described in poggianti et al .
projected local densities were computed by poggianti et al .
( 2008 ) , in a similar way as it was done for ediscs ( see 4 ) .
.list of clusters.[tbl1 ] [ cols="<,<,^ , < " , ] col .
( 1 ) : cluster name .
( 2 ) : short cluster name .
( 3 ) cluster redshift .
( 4 ) cluster velocity dispersion .
redshifts and velocity dispersions are taken from halliday et al .
( 2004 ) and milvang - jensen et al .
clusters with an asterisk do not have local density measurements , and only their numbers of members are used in this paper .
we make use of the millennium simulation ( hereafter ms ; springel et al .
the ms follows @xmath17 particles of mass @xmath18 within a comoving box of size @xmath19mpc on a side and with a spatial resolution of @xmath20kpc .
we extracted 90 haloes at @xmath21 and 90 at @xmath22 within the simulation box , uniformly distributed in log(mass ) between @xmath23 and @xmath24 .
these are @xmath25 masses ( hereafter `` halo masses '' ) , computed from the simulations as follows .
we define @xmath26 of a fof - halo as the radius of a sphere which is centered on the most bound particle of the group and has an overdensity of 200 with respect to the critical density of the universe af the redshift considered .
the enclosed mass is defined as @xmath25 . for 10-neighbours density measurements , only haloes with more than 10 galaxies brighter than @xmath27 were considered ( 65 haloes at @xmath21 and 62 at @xmath22 ) .
dark matter haloes were populated using the semi - analytic model presented in de lucia & blaizot ( 2007 ) , which are publicly available ( see also croton et al .
2006 ) . when deriving quantities to be compared with observations ( see next section ) , in order to take into account projection effects of galaxies close , but not belonging , to the halo , for each halo we considered a box 6mpc on a side , centred on the most bound particle . for each halo
, we considered all galaxies within 2 @xmath1 radii from the central galaxy to compute the projected velocity dispersion along the x , y , and z axes using the same bi - weight estimator that was used for the ediscs clusters , and used the mean of these projected velocity dispersions as the velocity dispersion of the system .
in this section we describe the basic ingredients of our study , and the nomenclature adopted throughout the paper . for simplicity , in the following we will refer to all galaxy systems of any @xmath4 and mass as `` clusters '' , because the distinction between clusters and groups is irrelevant in this paper .
our analysis is performed within @xmath1 , defined as the radius delimiting a sphere with interior mean density 200 times the critical density .
@xmath1 is often assumed to be approximately equal to the cluster virial radius , since the radius corresponding to a mean interior overdensity of @xmath28 has been shown by n - body simulations to `` accurately demarcate the virialized region of the dm halo , which is in approximate dynamical equilibrium , from the exterior , where material is still falling in '' ( cole & lacey 1996 ) .
since most observational studies use @xmath1 to compare galaxy properties in clusters at different redshifts , or in clusters and groups of different masses , we use @xmath1 to investigate how the density felt by those galaxies that are usually compared observationally changes with redshift and halo mass . assuming the virial theorem is valid , from the evolution of the critical density with redshift
, @xmath1 can be estimated from the observed line - of - sight velocity dispersion as ( finn et al .
2005 ) : @xmath29 similarly , from the virial theorem , the cluster mass can be estimated from the velocity dispersion as @xmath30 ( finn et al . 2005 ) . in this paper ,
observed cluster masses are therefore virialized total masses based on observed velocity dispersions derived from spectroscopy . for ediscs ,
the latter are in rather good agreement with the velocity dispersions obtained from the weak lensing analysis of clowe et al .
( 2006 ) , as shown in milvang - jensen et al .
( 2008 ) .
in the following , we refer to an output value of the simulations either as a `` simulated '' or a `` sim - observed '' quantity : a _ sim - observed _ quantity ( number of galaxies , velocity dispersion , projected local density etc . )
is computed from the simulation with the same method that would be used _
observationally_. for example , the sim - observed number of members within @xmath1 is the number _ in projection _ on the xy plane within @xmath1 ( derived from the velocity dispersion using eqn.(1 ) ) and with a velocity along the z axis within 3@xmath4 from the cluster velocity .
no attempt is made to reproduce the cluster selection strategy adopted by ediscs , nor the geometrical constraints in constructing the masks etc .
a _ simulated _ quantity gives the `` true '' value provided by the simulation .
in the example above , this would be the actual number of galaxies inside the sphere defined by the halo @xmath26 radius .
the difference between simulated and sim - observed quantities is fundamental , as we will show in the paper .
obviously , only sim - observed quantities can be directly compared with observations , but simulated quantities are very useful to address projection issues and other possible observational biases . all densities in this paper are computed using _ proper _ ( not comoving ) quantities , i.e. proper areas and volumes , in @xmath31 and @xmath32 .
this is motivated by the fact that , in order to study the dependence of galaxy properties on the density of the local environment , what matters are gravitational and vicinity effects , and therefore proper distances between galaxies .
projected local galaxy densities were computed from the observations as described in detail in poggianti et al .
( 2008 ) using the circular area that in projection on the sky encloses the 10 closest galaxies brighter than an absolute v magnitude @xmath27 . from the simulation ,
we compute both the sim - observed projected 2d local density and the physical 3d density .
sim - observed projected local densities are computed for each galaxy `` member '' ( sim - observed member ) of a simulated cluster as for observations : we consider as members all galaxies brighter than a limit @xmath33 ( either -20 or -19.4 , see below ) within a projected radius @xmath1 from the `` bcg '' ( the central galaxy of the halo ) and within 3@xmath4 in velocity from the mean cluster velocity .
the projected local density @xmath34 ( gal/@xmath35 ) is calculated from the ratio @xmath3/area where area is the circular area encompassing the @xmath3 projected closest neighbours brighter than @xmath33 in the xy ( `` sky '' ) plane , with @xmath36 unless otherwise stated .
the `` physical '' local density @xmath37 ( gal/@xmath38 ) is meant to measure the 3d number density of a region around each galaxy .
it is computed as the ratio @xmath3/volume , where the volume is the spherical volume encompassing the @xmath3 closest neighbours in 3d brighter than @xmath33 , with @xmath36 unless otherwise stated .
the 10 closest neighbours to each galaxy are those with the 10 smallest values of distance @xmath39 from the galaxy , where @xmath40 are the distances along each axis in the simulation . when computing the densities from the simulation , all neighbours above the magnitude cut were considered , also if outside @xmath26 , to avoid edge effects at the radius @xmath26 , similarly to what was done observationally .
observationally , the projected local density distribution in low - z clusters is shifted towards lower values compared to clusters of similar masses at higher redshift .
this is shown in fig .
[ uffa555 ] , and was mentioned in poggianti et al .
this could also have been inferred if someone had compared previous low - z ( dressler 1980 , lewis et al .
2002 ) and high - z ( dressler et al .
1997 , postman et al .
2005 ) cluster observations . to our knowledge ,
the evolution of the observed density distribution , its origin and consequences have not been discussed in the past . it may be argued that this effect could be due to density measurements at high - z suffering from a higher contamination of interlopers , that can artificially inflate the observed densities , but we will show this is not the case . in order to understand the significance of the density evolution , we investigate how the observed 2d projected density distribution at different redshifts is related to the physical 3d density distribution in the simulation , and how both are expected to change with redshift in clusters .
observations and simulations are compared in fig .
[ uffa555 ] for galaxies brighter than @xmath41 . only haloes with @xmath42 are used in this comparison , to match the masses of ediscs and sdss clusters that dominate the observed density distribution .
the observed projected density distribution ( red histograms ) shifts by about 0.4dex from @xmath21 to @xmath22 , corresponding to a decrease by a factor @xmath43 .
sim - observed projected densities evolve by the same amount ( black histograms ) , and they follow rather well the observed projected density distributions both at high- and low - z .
the corresponding 3d physical densities computed from the simulation are shown in the left panel of fig .
[ uffa55 ] .
the evolution of the physical 3d densities is even stronger than that of the 2d distribution .
physical densities for @xmath27 were on average higher by 0.5 dex , that is a factor @xmath44 , in distant clusters than today .
therefore , the evolution in projected density is not merely a projection effect with @xmath45 , but corresponds to physically evolving conditions in 3d galaxy number density .
it is also interesting to note that , when accounting for the average observed shift and rescaling to the same number of galaxies , the 2d and 3d density distributions at @xmath22 closely match the respective distributions at @xmath21 ( see dashed histograms in fig .
[ uffa55 ] ) . after number rescaling
, the high - z distribution can be obtained with good approximation from the low - z histogram simply multiplying each density by a factor @xmath44 in 3d , and 2.5 in 2d , since the shape of the distribution does not change .
importantly , we find that both the 2d and the 3d density distributions do not vary strongly with cluster mass , as shown in fig .
[ 4masses ] . , the radius whose interior mean density is equal to 500 times the critical density . ]
the @xmath36 neighbours physical density distribution are rather similar at all halo masses above @xmath46 ( top panels ) , covering the same range of densities .
the binned distributions shown in fig . [ 4masses ] do not overlap with each other for always less than 12% of the galaxies . at lower halo masses , @xmath47 ,
corresponding to the lowest mass groups in ediscs , using @xmath36 is inappropriate because the number of luminous members is less than 10 for many systems . using densities computed for @xmath48 ( bottom panels in fig .
[ 4masses ] ) , the density distributions remain rather similar to those of more massive clusters .
galaxies in haloes with masses @xmath49 cover the same density range of galaxies in more massive haloes , and the 3d density distributions do not overlap in less than 15% of the galaxies . remarkably , this means that the distributions of local densities to the 3 closest neighbours are quite similar for galaxies in clusters over two orders of magnitude in mass , suggesting that there is an approximately equal distribution of dense and less dense regions from groups to massive clusters when the `` really local '' density is considered .
the intuitive belief that more massive clusters are denser environments proves to be wrong .
actually , we see that the lowest halo mass bin has the largest 3d high - density tail . so far we have used a fixed @xmath27 galaxy magnitude limit at both redshifts , but it is physically more meaningful to adopt an @xmath33 limit evolving with redshift , taking into account the fact that a galaxy luminosity evolves as the galaxy stars become older .
we consider passive evolution , that is the evolution of an integrated spectrum due to simple aging of old stellar populations , without the addition of any new star . using van dokkum & franx ( 2001 ) or , equivalently , fritz et al .
( 2007 ) spectrophometric models for a single episode of star formation occurred at @xmath45 between 1.5 and 3 for a salpeter imf and solar metallicity , the evolution of @xmath33 between @xmath21 and @xmath50 is 0.5 - 0.7mag . between @xmath51 and @xmath22 .
if we arbitrarily assume the uncertainty to be the same in @xmath52 and in @xmath53 , the change in luminosity over our redshift range due to the imf is still much smaller than that obtained lowering the star formation redshift to e.g. @xmath51 . thus , the uncertainty in the passively evolving magnitude limit is dominated by the assumption in the formation redshift of most stars .
the great majority of cluster galaxies have very old mass - weighted stellar ages ( e.g. fritz et al . 2010 , in prep . )
therefore a high formation redshift range as the one adopted here is appropriate . ]
therefore , we consider an evolving magnitude limit going from @xmath27 at @xmath21 to @xmath54 at @xmath22 .
passive evolution produces a stronger decreasing luminosity evolution than any other star formation history .
therefore , assuming a passively evolving magnitude limit , we are sure to include in the density calculation at @xmath22 all galaxies that would be included at @xmath21 , and possibly some more galaxies that do not make it into the sample at @xmath21 .
this yields a sort of lower limit on ( but is probably close to ) the evolution of the density distribution as it would be measured using a fixed galaxy stellar mass limit and if galaxy masses did not evolve significantly . on the contrary ,
the non - evolving @xmath27 used above excludes from the low - z density calculation all those galaxies that have faded below the limit by @xmath22 but were included at @xmath21 .
therefore , it provides an upper limit for the evolution of the density distribution for a mass - limited galaxy sample with non - evolving galaxy masses .
unfortunately , the sdss spectroscopy is not deep enough to reach @xmath54 in our sdss clusters . adopting a brighter , passively evolving limit at both redshifts would result in too few galaxies in high - z clusters to allow a robust density determination .
hence , we are forced to investigate the passive evolution case only in the simulation , mimicking the way a sufficiently deep observational survey at low - z would be treated to compare with the high - z results .
the fact that the observed @xmath55 density distributions are well reproduced in the simulation , as is the number of cluster members shown in the next section , makes us confident that the simulated densities should be reliable also down to 0.5 magnitude fainter . with a passively evolving galaxy magnitude limit ,
the shifts of the 2d and 3d density distributions in the simulation are smaller , but still conspicuous ( fig .
[ uffa55bis ] ) . after number rescaling
, the high - z 3d and 2d density distributions match the @xmath22 distributions if densities decrease by a factor @xmath56 ( 0.25dex ) in 3d and 1.6 ( 0.2dex ) in 2d .
the analysis of the simulations presented so far demonstrates that , within @xmath1 and computing local densities from the 10 or 3 closest neighbours , a ) distant clusters were denser environments than low - z clusters , in the sense that , at @xmath21 , cluster galaxies had on average @xmath57 times ( @xmath58dex , @xmath59dex ) more luminous / massive neighbors than at @xmath22 , and b ) the distribution of physical densities is rather similar in haloes over two orders of magnitude in mass .
in the analysis presented above , densities represent _ number densities _ of galaxies , that means number of neighbours of an individual galaxy down to a certain magnitude , per unit of projected area or volume . in the following we want to establish *
how the evolution in number density of individual galaxies corresponds to the evolution of the _ global average number density _ over the whole cluster ( total number of members per unit volume @xmath0 ) . in doing this ,
we investigate separately the evolution of @xmath60 within @xmath1 and the evolution of the volume @xmath52 enclosed by @xmath1 .
the question we wish to address is how the change of the 3d number density with redshift is related to the evolution of @xmath0 , and @xmath60 and @xmath52 separately , and how and why these evolve .
note that @xmath0 can be estimated observationally , by estimating a cluster radius from either @xmath4 or an alternative mass measurement , and counting the number of members within this radius .
* how the _ number _ density evolution is related to the mean _ matter _ density evolution in clusters , that is the evolution of the mean total , or dm , mass per unit volume . in principle , mass density and number density are different measurements of environment , and are expected to influence physical processes in different ways : for example , number density is more relevant for galaxy - galaxy interactions and mergers , while mass density is more relevant for other effects , such as the cluster tidal field .
we wish to understand if number densities and mass densities evolve in similar or different ways , to gain a more complete picture of the evolution of environmental conditions .
the number of galaxies per cluster observed in the ediscs and sloan datasets within @xmath1 is shown as a function of the observed velocity dispersion in fig .
[ vssigma ] .
this is compared at each redshift with the sim - observed number of galaxies as a function of sim - observed velocity dispersion . from fig .
[ vssigma ] , it can be seen that the sim - observed numbers agree quite well with the observed numbers at both redshifts , although the sim - observed numbers at @xmath22 tend to be slightly higher than the observed values by about 0.1dex .
the number of galaxies increases with velocity dispersion at both redshifts ( higher velocity dispersion clusters have more members , obviously ) . for a fixed @xmath41 limit , the number of members decreases from @xmath61 to @xmath62 ( the number of galaxies down to a fixed magnitude limit in a cluster of a given @xmath4 was higher in the past ) , as shown by the offset in the right panel between the high - z best - fit line and the @xmath22 points . since clusters with the same observed velocity dispersion but different redshift have different masses ( cf .
4 ) , we remove this effect by plotting the number of galaxies versus cluster mass in fig .
[ vsmass ] , see bottom panels .
the trends remain similar : both observationally and theoretically , the number of galaxies increases linearly with mass in a log - log plot , and , for a fixed galaxy magnitude limit , it decreases towards lower redshift .
to understand if the number decline at lower redshift is simply due to galaxies dropping off the sample at @xmath22 because they fade below the fixed magnitude cut , in fig .
[ vssigma ] and fig .
[ vsmass ] we also show the sim - observed number of cluster members for a passively evolving magnitude limit @xmath63 at @xmath22 ( blue points ) . as explained in 5 , we can not do the same observationally , because the sdss spectroscopy is not deep enough . with the passively evolving limit , the numbers at @xmath22 in the bottom right panel of fig .
[ vsmass ] are only 0.057dex lower than the best fit @xmath21 correlation shown by the solid line ( @xmath64 ) ) .
as we will discuss later in this section , this residual small mismatch is caused by the way masses are derived , and the effect disappears when `` true '' simulated masses of haloes are considered ( top right panel in fig . [ vsmass ] ) . to conclude , both observations and simulations show a mild evolution in the @xmath60 vs @xmath65 relation with redshift for a fixed magnitude limit .
when passive evolution is taken into account , the number of members in a cluster of a given mass does not evolve between @xmath66 and @xmath22 .
a similar result was found by lin et al .
( 2006 ) , who , based on cluster mass estimates derived from the x - ray temperature and observed number counts , concluded that for passive evolution the @xmath67 relation shows no sign of evolution out to @xmath68 .
this result implies that the evolution in the number density distributions for the passive evolution case in cluster samples with the same cluster mass distribution at different redshifts ( fig .
[ uffa55bis ] ) is not due to a higher number of members at @xmath21 , but must originate mostly from the `` expanded size '' of local clusters compared to distant ones , that is , to the evolution of the cluster volume . at this point
it is useful to recall some textbook knowledge about how and why the volume and the mean matter density of clusters evolve , and how this evolution simply stems from using @xmath1 to define both of them .
due to the expansion of the universe , a unit volume today corresponds to a volume smaller by a factor @xmath69 at redshift @xmath45 ( peebles 1993 ) .
therefore , the physical 3d density ( number of galaxies per unit volume ) of non - evolving objects locked into the hubble flow changes with redshift as @xmath70 . galaxies and clusters of galaxies , however , are far from being unevolving objects , with clusters accreting new galaxies and groups .
moreover , collapsed structures such as clusters should have broken off from the hubble flow , at a time when their density exceeded a fixed multiple of the critical density at that redshift ( peebles 1993 , cole & lacey 1996 ) . since we only include galaxies within the cluster @xmath1 , _ by definition _ the mean _ matter _ density within this radius is 200 times the critical density at that redshift .
the critical density evolves with @xmath45 as : @xmath71 where @xmath72 is the gravitational constant @xmath73 . as a consequence , the mean matter density (= mass per unit volume ) in clusters of _ any mass _ , within a radius @xmath74 goes as : @xmath75 where @xmath76 and @xmath77 in our cosmology .
this means that _ the mean matter density in clusters at a given redshift is the same for all clusters_. moreover , it means that _ the mean matter density evolves in the same way for clusters of all masses _ , by a factor @xmath78 between @xmath79 and @xmath80 .
this has a number of implications , all consequences of using @xmath1 as the radius delimiting the cluster , among which : * the mean matter density within a cluster @xmath1 at any redshift can be computed from eqn.(3 ) : the most distant cluster at @xmath79 has a density @xmath81 times higher than a cluster at @xmath80 . in our case
, any cluster at @xmath21 ( of any mass ) has a matter density @xmath82 , and any cluster at @xmath22 has @xmath83 , a factor of 1.774 less dense .
the larger the redshift difference , the stronger the evolution : between @xmath84 and @xmath22 , the matter density evolves by a factor 2.750(4.656 ) .
* equation ( 3 ) implies that two clusters of the same mass @xmath85 but different redshifts @xmath79 and @xmath80 have different volumes : the most distant cluster at @xmath79 has a volume @xmath86 times smaller than the other ( 1.774 times smaller at @xmath21 compared to @xmath22 ) .
the ratio of the volumes of two equal mass clusters at different redshifts is therefore invariant with the cluster mass , and only depends on redshift , while , at any given redshift , the volume is linearly proportional to the mass ( eqn .
3 ) : a cluster twice as massive has twice as large a volume . and volume @xmath52 , separately , change in an evolving individual cluster(halo ) , considering that on average @xmath65 grows with @xmath45 by a factor that is larger at larger masses , as shown by numerous theoretical works ( eg .
lacey & cole 1993 ) .
simulations show that the average mass growth between @xmath21 and @xmath22 is about 1.5 times for @xmath87 ( mass at @xmath21 ) , about 1.9 for @xmath88 , 2.35 for @xmath89 , and 3.0 for @xmath90 ( the mass growth for finer intervals of cluster masses can be found in table 4 of poggianti et al .
the independence of @xmath91 from the cluster mass ( eqn .
3 ) implies that , although the mass growth rate depends on cluster mass , the parallel evolution in cluster volume ( radius ) compensates the mass dependence . ] the number of cluster members divided by the cluster mass @xmath92 is presented in fig .
[ noverm ] as a function of cluster mass .
the sim - observed quantities agree well with the ediscs and sdss datapoints ( top left and right panels ) .
both observations and simulation show a strong trend of declining n / m with mass , with a shift to lower n / m at lower @xmath45 when the @xmath27 limit is used . accounting for passive evolution ( blue points in the bottom left panel )
, the @xmath22 trend almost overlaps with that at @xmath21 , as was already shown in fig .
[ vsmass ] .
the observed and predicted declining @xmath92 trend with mass seems to suggest that more massive clusters have a lower number of galaxies per unit of cluster mass .
we will now show that this is just an illusory result due to the way the mass is computed `` observationally '' .
in fact , we find there is a discrepancy in the simulations between the halo masses and the mass derived from the sim - observed velocity dispersion ( fig .
[ massmass ] ) .
masses derived from the sim - observed velocity dispersion underestimate the halo masses by up to an order of magnitude in low - mass groups ( @xmath93 ) .
they yield halo mass values with good approximation at intermediate masses ( @xmath94 ) , and slightly overestimate the halo masses for massive clusters ( @xmath95 ) .
the correlation is @xmath96 , where both masses are in units of @xmath97 .
the top panels of fig .
[ vsmass ] shows that , using halo masses , the correlation between number of galaxies and cluster mass steepens ( @xmath98 ) ) , and passively evolved points at @xmath22 follow the same exact @xmath67 correlation found at @xmath21 with no offset ( top right panel , blue points ) .
the mass discrepancy corresponds to a discrepancy in both @xmath4 and cluster radius : the sim - observed velocity dispersion and sim - observed @xmath1 deviate from the velocity dispersion obtained from the halo mass and from the theoretical @xmath26 radius , underestimating them at low values and overestimating them at high values ( middle and right panels of fig .
[ massmass ] ) .
the mismatch between halo masses and cluster mass estimators based on velocity dispersion was studied in detail by biviano et al .
( 2006 ) , whose results agree very well with ours . based on cosmological hydrodynamical simulations , they studied clusters with masses above @xmath99 considering several observational effects such as the presence of interlopers and subclusters , sample size , incompleteness and different tracers of gravitational potential ( different types of galaxies and dm particles ) . for masses below @xmath99 , where we find the strongest mismatch between the two types of masses , we are not aware of any previous work we can compare with . the bottom right panel of fig .
[ noverm ] presents the number of cluster members per unit cluster mass obtained using the halo mass in the simulation instead of the sim - observed mass derived from the sim - observed velocity dispersion .
the decline with cluster mass disappears : the ratio between number of members and cluster mass is constant with mass and , when passive evolution is included , is constant also with redshift .
there appear to be , on average , approximately 20 galaxies brighter than a passively evolving @xmath22 limit @xmath54 for each @xmath99 of cluster mass , for any mass , at both redshifts .
this constancy is remarkable , also because it applies over a two order magnitude range of cluster mass , and regardless of redshift at least up to @xmath21 .
the fact that simulations are able to reproduce the observed @xmath100 vs @xmath101 trends ( upper panels of fig .
[ noverm ] ) , and that such trends disappear when adopting the halo mass instead of the sim - observed mass ( hence , when simply removing projection effects that bias the observational estimate of the halo mass ) supports the validity of the constancy of @xmath92 , regardless of the prescriptions of the semi - analytic model .
this constancy , and the small scatter around the mean value , especially in massive clusters ( @xmath102 for haloes with @xmath103 , and @xmath104 for @xmath105 ) , suggests that measuring @xmath60 should be a powerful method to estimate the cluster mass , free from the systematic mismatch between `` true '' halo mass and @xmath4-based mass estimate . recall that @xmath60 can be observationally estimated as the number of members brighter than our passively evolving limit within @xmath1 , if an estimate of @xmath1 is available .
the constancy of @xmath92 should allow to start from a first guess for @xmath60 within a fixed metric radius ( see also andreon & hurn 2009 ) , derive the mass , the corresponding @xmath1 and @xmath60 within this radius , iterating the procedure until convergence at @xmath106 is reached .
further investigation , combining simulations and observations , would be valuable to assess the utility of this method , comparing other observed cluster samples , at different redshifts , with simulations as in fig .
[ noverm ] ) .
this should test the model capability to reproduce the observed @xmath60 and , most importantly , compare the mass estimate precision with that of other methods , especially those based on other definitions of richness that do not consider the @xmath92 constancy found in this paper ( see e.g. popesso et al .
2005 , gladders et al .
2007 , rozo et al .
2009 , andreon & hurn 2009 ) . since our simulations show that for passive evolution and using halo masses , the average @xmath92 is constant both with @xmath45 and @xmath65 ( fig .
[ noverm ] ) , and given that @xmath107 ( from eqn(3 ) ) , the number of galaxies per unit volume goes as : @xmath108 where @xmath109 .
therefore , _ both the mean matter density _ ( mass per unit volume , eqn . 3 ) and the average _ global number density _ ( number of members per unit volume , eqn .
4 ) depend only on redshift , not on cluster mass , and are higher in more distant clusters by the same factor @xmath81 with @xmath110 .
we have seen that between @xmath21 and @xmath22 this factor is equal to 1.774 ( 0.25dex ) in our cosmology .
this is in striking agreement with , and accounts for , the 0.25dex shift of the 3d individual number density distribution shown in fig .
[ uffa55bis ] , and it must be ultimately responsible for the higher projected number densities observed at higher redshift ( fig . [ uffa555 ] ) . to summarize , our results show that higher - z clusters are denser than lower - z clusters , both in individual and global number density of galaxies , and in mass .
the constancy of the average _ mass _ density with cluster mass , and its evolution with redshift , is simply a consequence of the evolution of the critical density and of considering as `` cluster '' only the region within @xmath1 which approximates the sphere of influence of cluster gravity against the hubble flow .
not at all obvious was the fact that also the _ global number _
density should follow the same laws , should be independent of cluster mass and should evolve as the mass density .
we have demonstrated that this is true because the number of galaxies per unit of cluster mass @xmath92 is invariant _ both with cluster mass and redshift _ , when passive evolution is taken into account and when masses are unaffected by observational biases .
far from obvious was also the evolution of the projected and the 3d galaxy number density distributions , whose shape does not change with redshift and whose mean evolves exactly as the global number density .
interestingly , these conclusions apply to any virialized system regardless of its mass , thus should be applicable from the most massive clusters to the least massive groups .
it may appear counterintuitive , but we have just shown that the _ average _ surrounding number and mass density seen by a galaxy in any virialized region is the same at a given redshift , regardless whether the galaxy belong to a small group , an intermediate mass or a massive cluster . in the previous section , we have also shown that not only the average density , but also the number density _ distribution _ is approximately the same in the virialized regions of all halo masses explored . as far as densities are concerned , on large and unbiased cluster samples , the mass of the system does not matter , epoch does . however , it is fundamental to keep in mind that , while the density average and , approximately , distributions are independent of cluster mass , there is a scatter in the predicted n / m ( and therefore n / v and 3d distribution ) _ at any given mass _ , a scatter that increases towards lower halo masses , as visible in fig .
[ noverm ] and testified by the larger scatter ( 0.22 versus 0.10 ) in the @xmath100 relations given earlier in this section .
cluster - to - cluster density variations at a given mass are therefore not described in this formalism , but should be important for individual clusters expecially below @xmath99 , and likely will depend on the specific cluster growth history .
\1 ) the observed distribution of projected local densities in high - z clusters is shifted to higher values compared to the low - z distribution , and is reproduced by simulations .
based on the 3d density distributions obtained from the simulations , we find that this is due to high - z clusters being denser , in physical space , than their local counterparts .
galaxies in distant clusters are therefore on average closer to each other than galaxies in local clusters .
\2 ) the shift to higher individual galaxy number densities at higher redshifts is consistent with the globally higher number of galaxies per unit volume on the cluster scale that we obtain from the way the cluster region is defined and from the simulation results .
the number of galaxies per unit volume is higher by approximately a factor 1.8 at @xmath21 , and a factor 4.7 at @xmath111 , compared to @xmath22 .
\3 ) we find that the global number density @xmath0 evolves because of the combination of two effects : the dm mass per unit volume ( the mass density ) increases at higher redshift , simply because the cluster @xmath1 radius is defined to enclose a density that is 200 times the critical density , while simulations show that the average number of galaxies per unit of dm mass @xmath92 is constant both with redshift and cluster mass , being always @xmath112 counting only galaxies brighter than a passively evolving @xmath113 at @xmath22 .
the constancy of @xmath92 is found when considering the dm halo mass , but does not persist when using sim - observed masses derived from the velocity dispersion as it is done observationally , because the latter strongly underestimate the system mass at low masses , and slightly overestimate it at high masses . in the case of sim - observed masses derived from velocity dispersions ,
simulations show a decline of @xmath92 with mass , in agreement with the observed relations both at high- and low - z .
\4 ) from the previous point , it stems that both the mass density and the global number density evolve as the critical density and do not depend on cluster mass , therefore they are the same for clusters of all masses , at a given redshift .
moreover , the distribution of physical number densities obtained from the simulations does not vary strongly with cluster mass .
hence , contrary to the most intuitive belief of more massive systems being denser , mentioned in many observational studies , the most massive and the least massive clusters or groups are on average equally dense , at a given redshift .
it is important to keep in mind that these conclusions apply to _ virialized regions _ and for the density estimators adopted in this paper , that is for a method using the 10 ( and , we have verified , also 3 ) closest neighbours .
these conclusions can not be blindly applied to other types of density estimates that probe physically different scales , nor can be expected to hold in unvirialized regions of the universe . for example , the observed shift in 3d densities with redshift can not simply be assumed to be valid also when density is measured as number counts within a 8 mpc sphere or a 2mpc@xmath1142mpc@xmath114500@xmath115 cylinder in a general field survey .
moreover , we note that in principle these results may depend on the semi - analytic model employed , but this can only be ascertained repeating the analysis with other models .
our results highlights two main aspects : \a ) the strong evolution in the average density conditions experienced by galaxies in clusters .
more distant systems are much denser environments , both in number and in dm ( @xmath2 total)mass .
\b ) the impressive homogeneity in the density of clusters regardless of mass , at a given redshift . not only
the dm mass density , but also the average number density of galaxies and , approximately , the distribution of physical number densities are the same in the virialized regions of clusters and groups of any mass , at a given redshift .
cluster - to - cluster variations _ at a given mass _ , however , may be important for individual systems , especially at masses below @xmath97 .
bmp thanks the alexander von humboldt foundation and the max planck instituut fur extraterrestrische physik in garching for a very pleasant and productive staying during which the work presented in this paper was carried out .
the millennium simulation data bases used in this paper and the web application providing online access to them were constructed as part of the activities of the german astrophysical virtual observatory .
bmp acknowledges financial support from asi contract i/016/07/0 .
gdl acknowledges financial support from the european research council under the european community s seventh framework programme ( fp7/2007 - 2013)/erc grant agreement n. 202781 .
andreon , s. , hurn , m.a . ,
2009 , mnras , submitted baldry , i.k . ,
balogh , m.l .
, bower , r.g . ,
glazebrook , k. , nichol , r.c . , bamford , s.p .
, budavari , t. , 2006 , mnras , 373 , 469 balogh , m.l . , baldry , i.k . , nichol , r. , miller , c. , bower , r. , glazebrook , k. , 2004 , apj , 615 , l101 berlind , a.a . , blanton , m.r . , hogg , d.w . , weinberg , d.h . , dave , r. , eisenstein , d.j . , katz , n. , 2005 , apj , 629 , 625 biviano , a. , murante , g. , borgani , s. , diaferio , a. , dolag , k. , girardi , m. , 2006 , a&a , 456 , 23 blanton , m.r . , et al . , 2003 , apj , 594 , 186 bolzonella , m. , et al . , 2009 , a&a in press ( arxiv : 0907.0013 ) clowe , d. , et al . , 2006 , a&a , 451 , 395 cole , s. , lacey , c. , 1996 , mnras , 281 , 716 conroy , c. , gunn , j.e . , white , m. , 2009 , apj , 699 , 486 cooper , m.c . , newman , j.a . ,
madgwick , d.s . ,
gerke , b.f .
, yan , r. , davis , m. , 2005 , apj , 634 , 833 cooper , m.c . , newman , j.a . , weiner , b.j .
, et al . , 2007 , mnras , 376 , 1445 cooper , m.c . , newman , j.a . , weiner , b.j .
, et al . , 2008 , mnras , 383 , 1058 croton , d.j . , springel , v. , white , s.d.m .
, de lucia , g. , frenk , c.s . , gao , l. , jenkins , a. , kauffmann , g. , navarro , j.f . ,
yoshida , n. , 2006 , mnras , 365 , 11 cucciati , o. , iovino , a. , marinoni , c. , et al . , 2006 ,
a&a , 458 , 39 de lucia , g. , blaizot , j. , 2007 , mnras , 375 , 2 desai , v. , et al . , 2007 ,
apj , 660 , 1151 dressler , a. 1980 , apj , 236 , 351 dressler , a. , oemler , a. , couch , w.j . , smail , i. , ellis , r.s . , barger , a. , butcher , h. , poggianti , b.m . , sharples , r.m . , 1997 ,
apj , 490 , 577 elbaz , d. , daddi , e. , le borgne , d. , et al . , 2007 , a&a , 468 , 33 finn , r.a . , zaritsky , d. , mccarthy , d.w . ,
poggianti , b.m . ,
rudnick , g. , halliday , c. , milvang - jensen , b. , pello , r. , simard , l. , 2005 , apj , 630 , 206 fritz , j. , et al . , 2007 , a&a , 470 , 137 gladders , m.d .
, et al . , 2007 , apj , 655 , 128 gomez , p.l . ,
nichol , r.c . ,
miller , c.j .
, et al . , 2003 , apj , 584 , 210 gonzalez , r.e . , padilla , n.d . , 2009 , mnras , 397 , 1498 gray , m.e . ,
wolf , c. , meisenheimer , k. , taylor , a. , dye , s. , borch , a. , kleinheinrich , m. , 2004 , mnras , 347 , l73 halliday , c. et al .
, 2004 , a&a , 427 , 397 hashimoto , y. , oemler , a. , lin , h. , tucker , d.l .
, 1998 , apj , 499 , 589 hogg , d.w .
, et al . , 2004 , apj , 601 , l29 kauffmann , g. , white , s.d.m . ,
heckman , t.m .
, menard , b. , brinchmann , j. , charlot , s. , tremonti , c. , brinkmann , j. , 2004 , mnras , 353 , 713 kovac , k. , et al . , 2009 ,
apj in press ( arxiv:0903.3409 ) lacey , c , cole , s. , 1993 , mnras , 262 , 627 lewis , i. , balogh , m. , de propris , r. et al .
, 2002 , mnras , 334 , 673 lin , y .- t . ,
mohr , j.j . ,
gonzalez , a.h . ,
stanford , a. , 2006 , apjl , 650 , l99 milvang - jensen , b. , et al .
, 2008 , a&a , 482 , 419 peebles , p.j.e . ,
1993 , principles of physical cosmology , princeton university press , princeton poggianti , b.m . , et al . , 2006 ,
apj , 642 , 188 poggianti , b.m .
et al . , 2008 , apj , 684 , 888 poggianti , b.m .
et al . , 2009 , apj , 693 , 112 popesso , p , biviano , a. , bohringer , h. , romaniello , m. , voges , w. , 2005 , a&a , 433 , 431 postman , m. , geller , m.j .
, 1984 , apj , 281 , 95 postman , m. et al . , 2005 , apj , 623 , 721 rozo , e. , et al .
2009 , apj , 699 , 768 springel , v. , white , s.d.m . ,
jenkins , a. , et al .
2005 , nature , 435 , 629 tasca , l.a.m . , et al . , 2009 , a&a , 503 , 379 van dokkum , p.g .
, franx , m. , 2001 , apj , 553 , 90 white , s.d .
et al . , 2005 ,
a&a , 444 , 365 wolf , c. et al .
, 2009 , mnras , 393 , 1302 | we show that , observationally , the projected local density distribution in high - z clusters is shifted towards higher values compared to clusters at lower redshift . to search for the origin of this evolution
, we analyze a sample of haloes selected from the millennium simulation and populated using semi - analytic models , investigating the relation between observed projected density and physical 3d density , using densities computed from the 10 and 3 closest neighbours .
both observationally and in the simulations , we study the relation between number of cluster members and cluster mass , and number of members per unit of cluster mass . we find that the observed evolution of projected densities reflects a shift to higher values of the physical 3d density distribution . in turn , this must be related with the globally higher number of galaxies per unit of cluster volume @xmath0 in the past .
we show that the evolution of @xmath0 is due to a combination of two effects : a ) distant clusters were denser in dark matter ( dm ) simply because the dm density within @xmath1 ( @xmath2 the cluster virial radius ) is defined to be a fixed multiple of the critical density of the universe , and b ) the number of galaxies per unit of cluster dm mass is remarkably constant both with redshift and cluster mass if counting galaxies brighter than a passively evolving magnitude limit .
our results highlight that distant clusters were much denser environments than today s clusters , both in galaxy number and mass , and that the density conditions felt by galaxies in virialized systems do not depend on the system mass .
[ firstpage ] galaxies : clusters : general galaxies : evolution galaxies : statistics galaxies : interactions |
in the past decade , the formation of rough surfaces under far - from - equilibrium conditions was a central theme in statistical physics .
the application of self - affine fractals and scaling methods was essential to the progress that has been made towards the understanding of these non - equilibrium phenomena . within this context
, the standard tools used to describe various self - affine structures observed in disordered surface growth are the roughness @xmath4 and the growth @xmath8 exponents @xcite .
the central goal of this approach is provide information about the correlations between fluctuations of a space and/or time varying property .
theoretical modeling of self - affine growth processes frequently used some of the models investigated in critical phenomena , e.g. , directed percolation , random field ising @xcite and sine - gordon models @xcite .
these efforts faced an old problem : thirty years after the renormalization group theory of critical phenomena , the quantitatively accurate prediction of the location and characteristics of phase transitions still constitute a challenging and controversial question @xcite .
this is particularly true when one considers systems with random quenched disorder ( e.g. , random field ising and potts models , spin glasses and structural glasses @xcite ) , which exhibit non - trivial features such as extremely slow dynamics , aging , ergodicity breaking , complex energy landscapes , etc . on the other hand , the inverse problem , i.e. , using the roughness exponents to study
the main features of the phase diagram of equilibrium spin models , has not been much explored up to now . in 1997 ,
de sales _ et al . _
@xcite mapped cellular automata ( ca ) configurations on solid - on - solid like profiles and used the roughness exponent @xmath4 to classify the elementary wolfram ca rules .
later , they also shown that this exponent can be used to detect the frozen - active transition in the one dimensional domany - kinzel ca ( dkca ) @xcite without any reference to order parameters or response functions . as we mentioned before , beyond the roughness exponent @xmath4 , the growth exponent @xmath8 is another critical index used to describe roughening processes in the surface growth context .
recently , atman and moreira @xcite determined the exponent @xmath8 for the growth process generated by the spatiotemporal patterns of the dkca .
the value of @xmath8 exhibits a cusp at the frozen - active frontier and , if one observes the difference configuration between two dkca replicas , also at the active - chaotic critical frontier .
the advantage of this method to find the phase diagram of the dkca is that it is not necessary to wait for the system `` thermalizes '' , a process which often dispends a lot of computational time . in this paper
we extended the roughness exponent analysis to other standard models of statistical mechanics .
specifically , we study the @xmath0-state potts model ( with @xmath1 and @xmath2 ) , the simplest locally interacting statistical model exhibiting both first and second order static phase transitions .
we study also the @xmath3 clock model for which a kosterlitz - thouless type phase transition is observed . finally , since any connectivity problem can be studied by starting with pure random percolation and then adding interactions , we apply the roughness method to random - site percolation . in section 2
we describe the models and define the mapping between spin configurations and walk profiles . in section 3 , using the mapping between spin states and profiles , we characterize the phases of the random - site percolation problem , potts and clock models by the roughness exponent @xmath4 .
finally , we conclude and indicate future directions of this work in section 4 .
the @xmath0-state potts ferromagnet consists of spin variables @xmath9 which may take on @xmath0 discrete values @xmath10 and are coupled by the dimensionless hamiltonian @xmath11 where @xmath12 is the kronecker delta function .
the @xmath13-state clock model is defined by the hamiltonian @xmath14 in which each spin can assume @xmath13 discrete values @xmath15 . in eqs .
( [ potts ] ) and ( [ clock ] ) the sums in @xmath16 runs over the lattice sites and their nearest - neighbors , @xmath17 , @xmath18 is the temperature , @xmath19 and @xmath20 is the coupling constant .
we simulated the potts model with @xmath1 and @xmath2 and the clock model with @xmath3 states , on square lattices of @xmath21 sites imposing periodic boundary conditions .
for updating the spins we use a sequential monte carlo heat - bath process . in random - site percolation ,
one randomly occupies a fraction @xmath13 of the sites of a @xmath22-dimensional lattice ( @xmath23 : empty site ; @xmath24 : occupied site ) .
when @xmath13 is small , the pair connectedness length scale @xmath25 is short , comparable to the lattice constant @xmath26 .
however , when @xmath13 approaches @xmath27 , there occur fluctuations in the characteristic size of clusters on all scales from @xmath26 to @xmath25 , which diverges as @xmath28 .
each feature of thermal critical phenomena has a corresponding analog in percolation , so that the percolation problem is called a geometric or connectivity critical phenomena . for site percolation on the square lattice
the critical probability is @xmath29 @xcite . in the present work we mainly focused on the numerical study of the self - affine profiles generated from the configurations in the ordered and disordered phases of the potts and clock models and in the site - percolation problem .
as shown in a previous work @xcite the spin states can be mapped on random walk - like profiles and the correlations present in them can be measured using the roughness exponent .
the simplest method to generate walk profiles from the spin configurations at a time @xmath30 is a 1:1 mapping in which each spin state @xmath31 is associated to a step ( to the right or to the left ) of a one - dimensional walk .
specifically , to an unique spin configuration @xmath32 corresponds a spatial profile @xmath33 , given by the sequence of the walker displacements @xmath34 after @xmath35 unit steps defined as @xmath36 where @xmath37 for the @xmath38 and @xmath39 potts models , @xmath40 ( @xmath41 ) if @xmath42 ( if @xmath43 ) for the @xmath3 clock model , and @xmath44 for the ising ( @xmath45 ) model and site - percolation . after obtained the profiles by this mappings , the roughness exponent @xmath4 was calculated by determining the average standard deviation of parts of the profiles with various scales @xmath46 . at site @xmath35 , in the scale @xmath46
, the rms displacement fluctuation is given by : @xmath47 with @xmath48 the roughness in the scale @xmath46 is given by @xmath49 the roughness @xmath50 can distinguish two possible types of profiles .
if it is random or even exhibits a finite correlation length extending up to a characteristic range ( such as in a markov chain ) , then @xmath51 as in a normal random walk .
in contrast , if the profile has infinitely long - range correlations ( no characteristic length ) , then its roughness will be described by a power law scaling such as @xmath52 with @xmath53 .
the case @xmath54 implies that the profile presents persistent correlations , i.e. , a given displacement sequence ( increasing or decreasing ) is likely to be close to another of the same type . on the other hand ,
profiles with @xmath55 are anticorrelated , which means that displacement sequences containing a great fraction of steps to the right are more likely to alternate with another one in which steps to the left are predominant and vice versa .
the exponent @xmath4 is restricted to the interval @xmath56 $ ] and is related to the fractal dimension @xmath22 of the profile by @xmath57 @xcite .
before discuss the results , we shall make a very brief review of the equilibrium phase diagrams of the @xmath0-state potts and @xmath3 clock models . the square lattice @xmath0-state potts ferromagnet presents a second order phase transition for @xmath58 and a first order one for @xmath59 at the critical temperatures @xmath60 ( in units of @xmath61 ) .
the @xmath13-state clock model interpolates between the ising ( @xmath62 ) and the xy ( @xmath63 ) models . for @xmath64
one expects the emergence of a soft spin - wave phase between the ordered , low temperature , and the disordered , high temperature phases @xcite . for the @xmath3 case ,
the spin - wave phase is limited by the transition temperatures @xmath65 and @xmath66 @xcite . in fig .
1 we show typical walk profiles generated by spin configurations of the ising model ( @xmath45 ) . from data similar to those in fig . 1 for different temperatures
, we obtained the behavior of the roughness exponents @xmath4 as shown in fig .
2 , for the ising model with @xmath67 and 128 .
these results correspond to averages over typically @xmath68 random initial configurations taken after thermalization .
the roughness exponent exhibits an abrupt fall from @xmath69 to @xmath6 at the temperature @xmath70 which agrees with the static critical value @xmath71 as we increase the system size we can note that the change of @xmath4 at @xmath72 becomes sharper , suggesting a step function for the @xmath73 curve at the thermodynamical limit , and a high - temperature value of the exponent @xmath4 tending to @xmath74 . indeed , the inset in fig .
2 shows that the width @xmath75 goes to zero as @xmath76 increases , where @xmath77 is arbitrarily defined as the difference between the observed upper temperature @xmath78 for which @xmath79 and the temperature @xmath80 for which @xmath4 has fallen to half its maximum ( @xmath81 ) .
we also observe that , in this same limit , the temperature @xmath78 tends to the exact critical temperature @xmath82 ( @xmath83 , @xmath84 and @xmath85 ) . in fig .
3 we show similar behaviors for the @xmath38 and @xmath39 potts model with @xmath86 .
the abrupt falls in the exponent @xmath4 are located at the temperatures @xmath87 and @xmath88 which are also in good agreement with the static critical values @xmath89 and @xmath90 the values @xmath5 observed in the low temperature phases correspond to flat profiles reflecting the existence of long - range order ( magnetization ) . in contrast , the values @xmath6 obtained in the high temperature phases characterize random walk profiles , as expected for disordered spin configurations .
finally , any qualitative difference is observed among the @xmath73 curves across the critical surfaces corresponding to second ( @xmath38 ) or first ( @xmath39 ) order phase transitions . in fig .
4 we show a typical log - log plot of @xmath91 for the ising model , whose fitted slopes give the roughness exponent values .
we observe , in general , the existence of two distinct linear portions in the curve , whose intersection point defines a particular length scale @xmath92 .
the value of @xmath93 seems to be a measure of the average size of the spin islands or magnetized micro - domains limited by the correlation length @xmath94 .
indeed , at the ferromagnetic phase , for small scales ( @xmath95 ) there is resolution to see the local spin fluctuations around the smooth profile , which leads to @xmath96 . for large scales ( @xmath97 ) the profile is flat and the values @xmath5 reflect the long - range order . at high temperatures
@xmath93 rapidly decreases to the lattice constant @xmath98 , which agrees with the fact that such crossover is not observed in a random profile .
an additional support to the relationship between @xmath92 and @xmath99 is provided by fig . 5 which shows that @xmath93 ( for the ising model ) has a peak near the temperature @xmath72 . the height and sharpness of this peak increase with the system size .
a finite size analysis for the ising model shows that the maximum value of the crossover length @xmath93 ( which occurs near the critical point ) scales linearly with the system size @xmath100 , similarly to the correlation length @xmath99 .
in fact , fig .
5 suggests that @xmath93 diverges at @xmath101 as @xmath102 for the ising model .
moreover , the behavior of @xmath92 near @xmath82 is similar to the @xmath103 ising model correlation length , whose exact expression @xmath104 is known for @xmath105 @xcite ( see fig .
5 ) . here , the dual inverse temperature @xmath106 is given by @xmath107 .
a recent work by kantelhardt et al .
@xcite provides the strongest support to the close relation between the length scales @xmath92 and @xmath99 that we suggest here based on our limited numerical evidence .
these authors demonstrated that the detrended fluctuation analysis ( dfa ) , which we used here in zero - order , can detect crossovers in the observed long - range correlation behavior of data series .
they analyzed artificial data with a crossover from long - range correlations ( @xmath108 ) for @xmath109 to uncorrelated behavior for @xmath110 or vice - versa .
their dfa results clearly revealed the crossover and provided estimated crossover lengths @xmath111 always larger than the real @xmath112 by a systematic deviation which increases with the detrending order @xmath113 . also , the estimated @xmath111 were less accurate for @xmath4 close to @xmath114 .
the support for our suggestion comes from the analogy between the artificial series generated by fourier transforms studied in @xcite and our data series build from spin configurations in which only thermal correlations , extending up to the scale @xmath99 ( the correlation length ) , are present .
the existence of two distinct linear portions in the plots of @xmath91 is also observed in the potts model . in fig .
6 it is shown the behavior of @xmath93 as function of temperature for the potts model with @xmath38 and 7 . we can note the sharp peaks around the temperatures @xmath115 and @xmath116 . in fig .
7 we show typical walk profiles generated by spin configurations of the @xmath3 clock model at three temperatures : @xmath117 , located in the ordered phase , @xmath118 , in the spin - wave phase and @xmath119 in the disordered one . the roughness exponent @xmath4 as a function of the temperature for the @xmath3 clock model is shown in fig .
it suggests the existence of three distinct phases , namely , a low temperature flat ( @xmath5 ) phase with long - range order and extending up to @xmath120 , a high temperature , disordered and rough ( @xmath6 ) phase for @xmath121 , and an intermediate , rough ( @xmath122 ) phase on the temperature range @xmath123 .
these transition temperatures are in good agreement with those obtained for the static transitions between the ferromagnetic , paramagnetic and soft spin - wave phases of this model @xcite . the behavior of @xmath92 as a function of temperature is shown in fig .
@xmath124 exhibits a sharp peak around @xmath125 , the critical temperature separating the paramagnetic and the spin - wave phases , as well as a sudden jump near @xmath126 , where the transition between the ferromagnetic and spin - wave phase occurs .
but , the main feature of fig .
9 is that , in the spin - wave phase ( @xmath123 ) , @xmath92 has a high and almost constant value as expected for a kosterlitz - thouless `` critical '' phase characterized by an infinite correlation length @xmath99 at all those temperatures @xcite . finally , the roughness exponent method applied to the random site - percolation problem results in a constant value @xmath127 for the roughness exponent over the entire range of the probability @xmath13 .
this result is expected since in site - percolation a fraction @xmath13 of the lattice sites is randomly occupied , generating random profiles in both phases .
in addition , it is not observed a length scale @xmath92 in the log - log plots of @xmath128 , as occurred in the magnetic models .
in this study we have shown that spin configurations for the potts ( with @xmath1 and 7 states ) and @xmath3 clock models exhibit distinct self - affine characteristics , measured by the roughness exponent @xmath4 .
the low temperature phases of the potts model correspond to flat ( @xmath5 ) profiles , whereas the high temperature phases are associated to rough ( @xmath6 ) ones . for the @xmath3 clock model , in addition to the flat ( ferromagnetic ) and rough ( paramagnetic ) profiles , an intermediate rough ( @xmath7 ) phase , associated to a soft spin - wave one , is found .
the transition temperatures between the different roughness regimes are in good agreement with the static critical temperatures of these models .
our results show that the roughness exponent method is able to detect equilibrium phase transitions and provides accurate numerical determination of the critical surfaces without any reference to thermodynamical potentials , order parameters or response functions .
in contrast , the roughness exponent method applied to the random site - percolation problem results in a constant value @xmath127 over the entire range of the probability @xmath13 , showing that the roughness exponent @xmath4 is insensitive to detect geometric critical phenomena .
the same analysis can be applied to damage - spreading phase transitions by focusing on the difference configurations between two replicas of the system , as done by atman and moreira @xcite . moreover , using the growth exponent @xmath8 , we can determine the phase diagram of spin models without any thermalization , leading to an impressive gain in simulations speed .
we are extending our simulations in order to reliably estimate the value of the exponent @xmath129 and compare this value with the known critical exponents @xmath130 for the ising model .
we conjecture that the exponent @xmath129 controlling the divergence of @xmath93 is the same as the correlation length exponent @xmath130 for this model .
another problem we can address using this roughness method is the location of phase transitions in disordered models such as random field and ising spin glass models . for spin glasses , exchange ( parallel tempering )
techniques greatly improve traditional monte carlo algorithms , but the sizes and temperatures accessible to simulations are still insufficient to clearly solve several important questions @xcite . in particular
the value of the critical temperature for the @xmath131 edwards - anderson model is controversial .
apart from the application of the roughness method to spin models , a central issue is the nature of the relationship between the correlations measured on the spin - state profiles by the @xmath4 or @xmath8 exponents and the traditional two - point correlation functions @xmath132 .
we intend to examine this question more carefully in a future work .
a. l. barabsi and h. e. stanley , in _ fractal concepts in surface growth _
( cambridge university press , cambridge , 1995 ) . , edited by f.
family and t. vicsek ( world scientific , singapore , 1991 ) .
p. meakin , in _ fractals , scaling and growth far from equilibrium _
( cambridge university press , cambridge , 1998 ) .
j. toner and d. p. divincenzo , phys .
b * 41 * , 632 ( 1990 ) .
m. plischke and b. bergersen , in _ equilibrium statistical physics _ , ( world scientific , singapore , 1994 ) .
, edited by a. p. young ( world scientific , singapore , 1998 ) . j. a. de sales , m. l. martins and j. g. moreira , physica a * 245 * , 461 ( 1997 ) .
j. a. de sales , m. l. martins and j. g. moreira , j. phys .
a * 32 * , 885 ( 1999 ) .
a. p. f. atman and j. g. moreira , eur .
j. b * 16 * , 501 ( 2000 ) .
r. m. ziff , phys .
* 56 * , 545 ( 1986 )
. s. elitzur , r. b. pearson and j. shigemitsu , phys .
d * 19 * , 3698 ( 1979 ) .
e. seiler , i. o. stamatescu , a. patrascioiu and v. linke , nucl .
b * 305 * 623 ( 1988 ) .
b. m. mccoy and t. t. wu , in _ the two - dimensional ising model _
( harvard uni . press , cambridge , 1973 ) ; c. itzykson and j. m. drouffe , in _ statistical field theory vol .
( cambridge univ . press ,
cambridge , 1989 ) .
j. w. kantelhardt , e. koscielny - bunde , henio h. a. rego , s. havlin and a. bunde , preprint cond - mat/0102214 ( 2001 ) | we present a numerical study of the self - affine profiles obtained from configurations of the @xmath0-state potts ( with @xmath1 and @xmath2 ) and @xmath3 clock models as well as from the occupation states for site - percolation on the square lattice .
the first and second order static phase transitions of the potts model are located by a sharp change in the value of the roughness exponent @xmath4 characterizing those profiles . the low temperature phase of the potts model corresponds to flat ( @xmath5 ) profiles , whereas its high temperature phase is associated to rough ( @xmath6 ) ones . for the @xmath3 clock model , in addition to the flat ( ferromagnetic ) and rough ( paramagnetic ) profiles , an intermediate rough ( @xmath7 ) phase - associated to a soft spin - wave
one - is observed .
our results for the transition temperatures in the potts and clock models are in agreement with the static values , showing that this approach is able to detect the phase transitions in these models directly from the spin configurations , without any reference to thermodynamical potentials , order parameters or response functions .
finally , we show that the roughness exponent @xmath4 is insensitive to geometric critical phenomena .
jos arnaldo redinz and marcelo lobato martins + pacs numbers : 05.10.-a , 05.50.+q , 05.40.-a , 05.70.fh + key - words : phase transitions , roughness exponent , lattice spin models + electronic adressses : [email protected] ; [email protected] 2 [ ] |
# 1#1 = 6.5 in = 6.5 in = 9.5 in = 9.5 in = -.5 in = -.3 in plus .5pt minus .5pt
# 1 almost a hundred years ago , lorentz calculated the self force on a charged particle , that is , the force exerted on the particle by its own radiation field . in an expansion in powers of the radius @xmath3 of the particle
, he found for the force @xmath4 @xmath5 where @xmath6 is the position vector of the particle , @xmath7 is a function of @xmath3 that goes like @xmath8 for small @xmath3 , the @xmath9 coefficient is independent of @xmath3 ( @xmath10 is the charge of the particle ) and the higher terms all go to zero as @xmath11 . in order to deal with the problem of the divergence of @xmath7 as @xmath11 , lorentz invented mass renormalization . applying newton s second law @xmath12 or , with @xmath13 fixed and @xmath11 , @xmath14 a finite equation of motion .
however , the divergence problem returns through the back door : ( [ eq:1.3 ] ) has a run - away solution @xmath15 with a run - away time of @xmath16 , @xmath17 seconds for electrons .
a lorentz covariant equation which reduces to ( [ eq:1.3 ] ) in the instantaneous rest system of the particle was proposed by dirac ; however , the run - away problem was not removed .
of course , in the relativistic model it is the momentum , not the velocity , that runs away .
the clearest algebraic insight into the problem , both classical and quantum , is obtained by finding an upper half - plane pole in the laplace transform @xmath18 of @xmath19 : @xmath20 here @xmath21 , with @xmath22 positive and large enough to make the integral converge . in appendix
a we take that path and , following moniz and sharp , show that in a suitably cut - off classical theory there are no run - aways , no matter how small the cut - off radius , provided @xmath23 , the bare mass , is positive we then show that the mass renormalization , carried out for @xmath11 as described above , leads to run - aways . in appendix b , we obtain the same result in a non - relativistic quantum theory .
thus quantum effects per se do not cure the disease , nor does the addition of a confining oscillator potential . in appendices
a and b we considered , and in the following we consider , the simplest problem that shows the run - away effect : a charged ( in the limit ) point particle , which we shall call an electron , is in a wave - packet state with approximate momentum @xmath24 and energy @xmath25 for times @xmath26 . at @xmath27 , a weak spatially uniform external electric field @xmath28 is switched on , and we follow the electron s motion , calculating the mean value of the coordinate @xmath6 , or of the velocity @xmath29 . the solution for @xmath18 is given for classical motion in ( [ eq : a.19 ] ) of appendix a and for the non - relativistic quantum case by ( [ eq : b.16 ] ) of appendix b : @xmath30 here @xmath31 is the cut - off function , which must go to zero as @xmath32 fast enough to make the integral converge .
@xmath33 is the laplace transform of @xmath34 , @xmath35 and @xmath36 in the quantum case represents the mean value of the variable .
the absence ( in ( [ eq:1.6 ] ) of an upper half @xmath37 plane pole in the cut - off theory , and its emergence following lorentz s renormalization procedure , are shown in appendix a. the presence of such a pole in the laplace transform produces an exponential run - away in time , via the inversion formula @xmath38 the upper half plane pole appears at a frequency @xmath39 , where @xmath40 and @xmath41 is the fine structure constant .
we point out here that for weak coupling the energy @xmath42 is deep in the relativistic region where the equations used to describe the system are certainly not valid . since relativistic spin @xmath43 quantum electrodynamics is known to hold accurately in that energy region , we presume that there will be no analogous pole in that theory . in the following
we formulate a way of studying that issue , and indeed find no analogous pole in the relativistic spin @xmath43 theory .
however , if a pole of the type suggested by landau , abrikosov and khalatnikov exists , either in the photon green s function or in the electron photon forward scattering amplitude , there would be a new type of run - away with the characteristic frequency @xmath44 , high enough so that there appears to be no way in which present day experiments can rule it out .
of course , such a pole is already a symptom of a diseased theory . in the following sections we will be calculating the mean value of the electron velocity , to lowest order in the external field . of course
, the limitation of the electron s velocity to @xmath45 does not automatically occur in this approximation .
however , if the electron momentum does run away , the limitation of the velocity to @xmath45 will appear as a non - linear effect of the external field . correspondingly
, the linear approximation to a run - away momentum would be a run - away velocity , for which we search .
we work in the schroedinger representation with a state vector @xmath46 which satisfies the equation @xmath47 here @xmath48 is the electron - photon hamiltonian , and @xmath49 the interaction hamiltonian with the spatially uniform external field : @xmath50 where @xmath51 and @xmath52 is the electron current density @xmath53 where @xmath54 have their usual meaning .
we take for the velocity operator @xmath55 where @xmath56 is the charge density ) . ]
@xmath57 for @xmath58 , the electron is in a state @xmath59 which satisfies the equation @xmath60 @xmath59 may be expressed as a superposition of stationary states of momentum @xmath24 and spin @xmath61 : @xmath62 where @xmath63 \label{eq:2.9}\ ] ] the state vector normalization is @xmath64 so that the wave - packet state @xmath59 is normalized to one with @xmath65 the mean value of the velocity vector in the state ( [ eq:2.8 ] ) is @xmath66 the matrix element of @xmath67 is @xmath68 since @xmath69 , there is no form factor or anomalous magnetic term .
the spinors @xmath70 satisfy the dirac equation @xmath71 the matrix - element in ( [ eq:2.13 ] ) has the value @xmath72 giving for the mean value ( [ eq:2.12 ] ) @xmath73 we can , in this calculation and others that involve momentum conserving operators , dispense with the wave packet @xmath74 , replace the state @xmath59 by @xmath75 , and leave the @xmath76 integral undone , thereby eliminating the delta function in ( [ eq:2.13 ] ) , and replacing the wave packet average in ( [ eq:2.16 ] ) by the single value @xmath77 . from here on we will assume that this has been done . for the real electron , the wave - packet average must be taken .
we solve ( [ eq:2.1 ] ) for small @xmath78 by expanding in powers of @xmath78 .
with @xmath79 we have @xmath80 and @xmath81 so that @xmath82 as just described , we replace @xmath59 by @xmath83 and leave the @xmath76 integral in ( [ eq:2.5 ] ) undone .
the mean value of the velocity is then given by @xmath84 with @xmath85 for @xmath86 and @xmath87 understood to be zero for @xmath58 .
we make ( [ eq:2.22 ] ) more explicit : @xmath88 where @xmath89 . to compare with the results from appendices a and b we take the laplace transform of ( [ eq:2.23 ] ) .
we note first that the laplace transform of @xmath90 is @xmath91 ( where @xmath92 is the laplace transform of @xmath93 ) so that the laplace transform in ( [ eq:2.23 ] ) is @xmath94 \phi_p \right ) a^e_k(\omega ) ~~. \label{eq:2.26}\end{aligned}\ ] ] if we took the expression ( [ eq:2.26 ] ) at face value , we would immediately conclude that there could be no upper half plane singularities in @xmath95 , since @xmath48 has only real eigenvalues .
however , since the theory is not finite , we are not justified in drawing that conclusion .
we must rather deal with the mass and charge renormalized theory , which in practise means order by order perturbation theory in @xmath41 . in the next section
we will show how the expression @xmath95 in ( [ eq:2.26 ] ) can be calculated order by order from feynman diagrams .
before we turn to that , we exploit ( [ eq:2.26 ] ) to calculate the zeroth order ( in the fine structure constant ) motion of the electron in the weak external field .
there are two classes of intermediate states : one particle ( which makes no contribution ) , and three particles ( two electrons and a positron ) .
one finds for @xmath95 @xmath96 \ , a^e_k(\omega ) ~~. \label{eq:2.27}\ ] ] here @xmath97 is the positive energy projection operator with momentum @xmath98 , @xmath99 the negative energy projection operator of momentum @xmath98 ( corresponding to positron momentum @xmath100 ) : @xmath101 the first term in ( [ eq:2.27 ] ) is symmetric in @xmath102 and @xmath103 , and hence gives zero when @xmath104 . to calculate the second term , we need @xmath105 so that @xmath106 we see in ( [ eq:2.31 ] ) the structure that will guide us in the search for run - away poles .
the amplitudes that are odd in interchange of @xmath103 and @xmath102 are odd in @xmath37 , to all orders in @xmath41 ; those that are even in @xmath107 interchange are even in @xmath37 to all orders in @xmath41 .
higher order odd amplitudes in spinor electrodynamics go like @xmath108 ( at high @xmath37 ) except for logarithms ; higher order even amplitudes go like @xmath109 except for logarithms .
there is therefore no way for a second order term to be of the same order as a zeroth order term ( the sign of a possible pole ) at @xmath110 , as in the non - relativistic case ; rather , it must be of order @xmath111 .
indeed , the pole@xmath112 conjectured by landau , abrikosov and khalatnikov would arise in just that way .
we translate ( [ eq:2.31 ] ) into time dependence : @xmath113 we close the @xmath37 integration in the upper or lower half plane as called for , and find @xmath114 \right\ } e^e_k ( t ' ) dt ' \nonumber \\ & & \hbox{\qquad } + e \
, \sigma_{kl } \ , { 1\over e } \int_0^t \sin [ 2e(t'-t ) ] e^e_k ( t ' ) dt ' ~~. \label{eq:2.33}\end{aligned}\ ] ] we recognize in ( [ eq:2.33 ] ) the classical relativistic first order equation for @xmath115 .
this follows by neglecting the zitterbewegung like sine and cosine averages of the external electric field a valid neglect for fields whose characteristic time is much longer than @xmath116 ( which is @xmath117 seconds for electrons ) .
finally , we wish here to find the @xmath37 dependence of @xmath118 at high @xmath37 . from ( [ eq:2.26 ] )
, it is @xmath119 \phi_p \right ) \ ,
a^e_k(\omega ) \label{eq:2.34}\\ & = & - { 2i e a^e_k(\omega)\over\omega } \left ( \phi_p ~,~ \psi^{\dagger}(x ) \
, \sigma^{\ell k } \psi(x ) \phi_p \right ) ~~. \label{eq:2.35}\end{aligned}\ ] ] the matrix element in ( [ eq:2.35 ] ) is of course finite in zeroth order , as shown in ( [ eq:2.31 ] ) . in second order
, it would be finite except for vacuum polarization effects .
we shall see in section iii that the correctly renormalized feynman amplitude for @xmath120 implies that the expression ( [ eq:2.26 ] ) must carry an explicit factor @xmath121 , where @xmath122 is the photon propagator renormalization constant @xmath123 with @xmath124 a momentum cut - off .
the finiteness of ( [ eq:2.26 ] ) together with the divergence in @xmath125 implies a high @xmath37 dependence , @xmath126 using the technique discussed in section iii , we can calculate the high @xmath37 behavior of @xmath118 explicitly , and find agreement with ( [ eq:2.37 ] ) .
we consider first the mean value of the velocity operator , without the accelerating electric field .
the stationary state @xmath127 is given as @xmath128 where @xmath129 is the bare ( or interaction representation ) one particle state , and @xmath130 is the unitary operator which transforms that state in time : @xmath131 @xmath132 is the interaction hamiltonian expressed in terms of interaction representation operators : @xmath133~~. \label{eq:3.4}\ ] ] the free hamiltonian @xmath134 is expressed in terms of the correct mass of the electron ; hence the necessity for the subtraction of @xmath135 in ( [ eq:3.4 ] ) .
we wish to calculate the velocity operator in the schroedinger state @xmath136 .
it is @xmath137 here we recognize the operator @xmath138 so that @xmath139 is the velocity expressed in the interaction representation , and @xmath140 where the @xmath141 matrix @xmath142 , and where we have used the unitarity of @xmath143 and its group property : @xmath144 the @xmath141 matrix is diagonal on one particle states . since there is no self - energy correction , it must be a phase , which is canceled by disconnected diagrams .
in ( [ eq:3.7 ] ) we see @xmath145 expressed in a way which allows us to make use of the dyson technique and feynman diagrams .
the 2nd order result is expressed in the diagrams shown in fig .
1 . @xmath146 @xmath147 fig . 1 here
the initial and final lines represent the wave functions @xmath148 ; the dot @xmath149 represents the operator @xmath150 ; and so forth . the diagrams ( 1b ) and ( 1c ) have the @xmath7 correction subtracted from them , and are multiplied by @xmath151 to reproduce the correct normalization of the incoming state .
the sum of all such diagrams is @xmath152 since @xmath153 .
thus , @xmath154 is both ultra - violet and infra - red finite , and equal to its expected free particle value , @xmath155 the closed fermion loop diagrams cancel because @xmath156 and @xmath56 carry the same charge renormalization .
we turn next to the effect of the external field as given by ( [ eq:2.23 ] ) .
we consider the feynman amplitude @xmath157 which we can readily calculate , order by order , following dyson : @xmath158 @xmath159 in ( [ eq:3.11 ] ) and ( [ eq:3.12 ] ) is the forward scattering amplitude of a photon of four momentum @xmath160 on an electron of momentum @xmath24 . by considering the form ( [ eq:3.11 ] )
we see that @xmath161 which has the same structure as ( [ eq:2.26 ] ) except that @xmath162 in the second term of ( [ eq:3.13 ] ) is in the lower half plane , whereas @xmath37 in ( [ eq:2.26 ] ) is in the upper half plane .
we therefore obtain the correct formula ( [ eq:2.26 ] ) from ( [ eq:3.12 ] ) by continuing @xmath163 into the upper half plane and setting it equal to @xmath37 .
the use of feynman diagrams makes clear the finiteness and high @xmath163 behavior of ( [ eq:2.26 ] ) .
we illustrate with the 2nd order ( in @xmath41 ) terms . the diagrams ( with the exception of vacuum polarization )
are : @xmath164 -8pt @xmath165 fig . 2 in figs .
( 2a ) and ( 2c ) , the self - mass is subtracted , and the remainder diagrams multiplied by @xmath151 . in fig .
( 2b ) , the self - mass is subtracted . in what remains ,
the ultra - violet divergences cancel ( using @xmath166 again ) by adding @xmath167 the infra - red divergence cancels by adding @xmath168 .
the high @xmath163 dependence of each of the diagrams ( 1b ) , ( 1d ) , ( 1e ) and ( 1f ) is @xmath169 .
however , when the four diagrams are added the logarithms cancel , leading to a high @xmath37 dependence in the odd amplitudes @xmath170 which therefore can not in weak coupling produce the upper half plane singularity which classically leads to the run - away solutions .
note however that the above simple asymptotic calculation only shows a cancellation in the @xmath107 odd amplitude .
the logarithms in the @xmath107 even amplitudes fail to cancel , leaving the possibility of a singularity of order @xmath171 in the even amplitudes , and a run - away @xmath172 in the even amplitudes .
we should call this a landau run - away , as opposed to a lorentz run - away .
the second order results of these calculations are given in appendix c. the relativistic theory brings in a new effect , the electron - positron vacuum polarization .
to take this into account we must add two more diagrams to those of fig . 2 : fig .
3 we study the effect of the diagrams of fig . 3 on the renormalization of ( [ eq:2.26 ] )
we replace @xmath173 by @xmath174 where @xmath175 is the bare charge and @xmath176 the lepton current : @xmath177 eq .
( [ eq:2.26 ] ) becomes @xmath178 \phi_p \right ) a_k^{e , u } \label{eq:3.15}\ ] ] where @xmath179 is the unrenormalized external vector potential .
in the combination @xmath180 the numerator operator will produce a factor @xmath181 , where @xmath182 is the finite photon propagation function , normalized to @xmath183 as @xmath184 .
the denominator is @xmath122 .
the calculation of the numerator matrix element will automatically generate the factor @xmath122 ; the denominator must be put in by hand . on the right side ,
the unrenormalized external field is produced by an external current with a charge @xmath175 .
the renormalized external field would be produced by a charge @xmath10 .
therefore @xmath185 , where @xmath186 is the renormalized external potential .
the matrix element containing @xmath187 will therefore produce a finite factor @xmath188 ; correspondingly the matrix element of @xmath189 will produce a renormalization factor @xmath190 . we may therefore rewrite ( [ eq:3.15 ] ) as the finite expression @xmath191 \phi_p \right ) a_k^e \ \ , \label{eq:3.16}\ ] ] justifying the discussion following eq .
( [ eq:2.35 ] ) .
it is straightforward to show directly from the diagrams in fig . 3 that the logarithmic dependence of @xmath192 at high @xmath193 translates into the expected logarithmic behavior of @xmath95 , as given in ( [ eq:2.36 ] ) and ( [ eq:2.37 ] ) .
( see appendix c. ) it is possible that the series of radiative corrections to @xmath194 sum to a pole , as conjectured by landau , abrikosov and khalatnikov@xmath112 .
if so , the pole is at space like @xmath193 , _
i.e. _ @xmath195 and @xmath196 , producing a new class of run - away solutions .
we emphasize that the above discussion is specifically for spin @xmath151 particles .
what happens to the classical run - away solutions in the relativistic quantum electrodynamics of a spin zero particle is an interesting question , but has not been dealt with here .
* acknowledgment * i wish to thank arthur kerman for arousing my interest in this subject , and to thank him , kenneth johnson and al mueller for very helpful discussions .
the vector potential is transverse , and given by @xmath204 where @xmath205 is defined by fourier transforming ( [ eq : a.8 ] ) : @xmath206 here , @xmath207 we solve ( [ eq : a.9 ] ) for @xmath208 with the boundary condition ( for convenience only ) @xmath209 the solution is @xmath210 so with @xmath211 this is an exact integral equation for the electron motion . if we replace @xmath212 by @xmath213 it would be a relativistic equation were it not for the cut - off function @xmath214 .
if we set @xmath215 , the equation becomes divergent ( near @xmath216 ) and has no solution . in view of these difficulties
it makes no sense to study this equation further , or to try to make it relativistic .
however , in the weak field limit we can and will solve the equation exactly , in order to study the way the run - away solutions appear . in this limit @xmath217 and @xmath218
are of first order in @xmath199 and @xmath219 is of second order .
therefore the equation becomes @xmath220 and can be solved by laplace transform .
remembering our boundary conditions @xmath221 , we have , for @xmath18 , @xmath222 where @xmath223
so we find @xmath224 @xmath18 in ( [ eq : a.19 ] ) has no upper half - plane singularities for @xmath225 .
this follows from the observation that the imaginary part of @xmath226 is @xmath227 which is different from zero in the upper half @xmath37 plane unless @xmath228 . here
@xmath229 and @xmath22 are the real and imaginary parts of @xmath37 . when @xmath228 , the denominator in ( [ eq : a.19 ] ) is always positive .
thus there are no run - aways .
renormalization changes that .
replace the integral @xmath230 by @xmath231 where @xmath232 and note that the integral in ( [ eq : a.21 ] ) is now convergent in the limit @xmath233 .
in that limit , the denominator @xmath234 in ( [ eq : a.19 ] ) becomes @xmath235 and the run - away pole appears at @xmath236
in this appendix we show that the classical formula ( [ eq : a.19 ] ) holds exactly , including arbitrary external field strength , in a non - relativistic quantum theory in which we neglect recoil .
the neglect of recoil has an internal consistency , since recoil can only be properly taken into account in a fully relativistic theory .
the point of this exercise is to show that quantum theory in itself does not cure the run - away problem , since it almost identically reproduces the classical effect .
we can solve this model exactly since the neglect of recoil makes the heisenberg equations of motion linear , and hence soluble , with the same solution as the classical case .
note that the neglect of recoil , _
i.e. _ of the @xmath237 dependence of @xmath238 , is exactly the approximation we made following ( [ eq : a.16 ] ) to arrive at a simply soluble classical problem there .
the hamiltonian is @xmath239 where @xmath240 and @xmath241 are radiation oscillators : @xmath242 = { 1\over i } \delta_{k k ' } \delta_{\lambda \lambda ' } \label{eq : b.2}\ ] ] and @xmath243 a cut - off function can be introduced here , as in the classical calculation of appendix a. we save space and time by simply inserting it into the final answer , ( [ eq : b.17 ] ) .
the coulomb hamiltonian @xmath244 is independent of @xmath6 and hence does not enter into the equations of motion .
these are @xmath245 although ( [ eq : b.5])([eq : b.8 ] ) are operator equations , we will in what follows only need mean values .
we therefore can set all operators to zero at @xmath27 , reproducing the classical boundary condition of appendix a. with that understanding , we carry out a laplace transform .
as before , we call @xmath246 for any variable @xmath247 .
the resulting equations are @xmath248 with solution @xmath249 or , with the introduction of the cut - off function , @xmath250 } ~~ , \label{eq : b.17}\ ] ] the same formula we found for the classical theory .
of course @xmath18 here represents the mean value of the operator .
note finally that adding a harmonic binding potential does not remove the run - away pole .
the amplitude we must calculate is @xmath251 taken in the forward scattering limit , @xmath252 , and evaluated by continuing @xmath253 to the upper half plane , as shown in eq .
( [ eq:3.13 ] ) . the possibility of a landau like pole will appear in lowest order by the coherent positive addition to ( [ eq : c.2 ] ) of a term coming from the next order in @xmath41 , and going like @xmath256 or @xmath257 at large @xmath37 .
the possible asymptotic covariant amplitudes are limited by the conservation law @xmath258 .
those that contribute to @xmath259 in the next order are @xmath260 which goes like @xmath261 at large @xmath37 ; @xmath262 and @xmath263 both of which go like @xmath264 at large @xmath37 .
one finds easily , to order @xmath265 , @xmath266 the next order correction comes from the diagrams in fig . 2 and fig .
3 . it is , again to order @xmath264 , and keeping only terms with a factor log @xmath267 at large @xmath37 : @xmath268 the first term in ( 4.7 ) comes from the feynman diagrams in fig . 3 , which in a different context signal the landau ghost . finally , we point out two curiosities that emerge on inspecting eq .
( [ eq : c.7 ] ) .
first , since @xmath269 , we see that the leading term in @xmath261 has no first order logarithmic radiative correction , as implied earlier by eq .
( [ eq:2.37 ] ) .
second , we see that the radiative corrections to the @xmath264 terms in the amplitude @xmath270 coming from the diagram of fig
. 2 precisely cancel those coming from the landau diagram of fig .
3 . whether these are more than a numerical accident is not known to the author | the existence of run - away solutions in classical and non - relativistic quantum electrodynamics is reviewed .
it is shown that the less singular high energy behavior of relativistic spin @xmath0 quantum electrodynamics precludes an analogous behavior in that theory . however ,
a landau - like anomalous pole in the photon propagation function or in the electron - massive photon foward scattering amplitude would generate a new run - away , characterized by an energy scale @xmath1 .
this contrasts with the energy scale @xmath2 associated with the classical and non - relativistic quantum run - aways . |
a closed orientable surface has a topological invariant known as the euler characteristic @xmath2=@xmath3 , where the genus @xmath4 is the number of holes in the surface .
nuclei as we now know them have the topology of a sphere with @xmath2=2 .
wheeler suggested that under appropriate conditions the nuclear fluid may assume a toroidal shape ( @xmath2=0 ) . using the liquid - drop model @xcite and the rigid - body moment of inertia @xcite
, it was shown that a toroidal nucleus , endowed with an angular momentum @xmath5=@xmath6 aligned about its symmetry @xmath7-axis beyond a threshold , is stable against the breathing deformation in which the major radius @xmath8 contracts and expands .
the rotating liquid - drop nuclei can also be stable against sausage instabilities ( know also as plateau - rayleigh instabilities , in which the torus breaks into smaller fragments ) , when the same mass flow is maintained across the meridian to lead to high-@xmath6 isomers within an angular momentum window @xcite .
recently ichikawa _ et al . _
@xcite found that toroidal high - spin isomer with @xmath5=60@xmath9 may be in a local energy minimum in the excited states of @xmath10ca .
they used a cranked skyrme - hartree - fock ( hf ) method starting from the initial ring configuration of 10 alpha particles . in ref .
@xcite it was found from the time - dependent hf and the random - phase approximation that a collective rotation about the axis perpendicular to symmetry axis of the toroidal isomer @xmath10ca(@xmath5=60@xmath9 ) can result in a pure collective precession motion . in our previous study
@xcite , we found that rotating toroidal nuclei have general occurrences and we located 18 even - even , @xmath11=@xmath12 , high - spin toroidal isomeric states : @xmath13si(@xmath5=44@xmath9 ) , @xmath14s(@xmath5=48 , 66@xmath9 ) , @xmath15ar(@xmath5=56 , 72 , 92@xmath9 ) , @xmath10ca(@xmath5=60 , 82@xmath9 ) , @xmath16ti(@xmath5=68 , 88 , 112@xmath9 ) , @xmath17cr(@xmath5=72 , 98 , 120@xmath9 ) , and @xmath18fe(@xmath5=52 , 80 , 104 , 132@xmath9 ) in the region 16@xmath0a@xmath052 .
subsequent to the work of @xcite , ichikawa _ et al .
_ in ref .
@xcite investigated the existence of toroidal isomers and their precession motions for nuclei with 28@xmath0a@xmath052 .
they also obtained high - spin toroidal isomers in @xmath15ar , @xmath10ca , @xmath16ti , @xmath17cr , and @xmath18fe , confirming the general occurrence of high - spin toroidal isomers in this mass region in @xcite .
( color online . ) the total hfb energy of @xmath19 mg , @xmath13si , @xmath14s , @xmath15ar , @xmath10ca , @xmath16ti , @xmath17cr , and @xmath18fe as a function of the quadrupole moment @xmath20 for the case of @xmath5=0 , taken from ref .
axially - symmetric toroidal configurations are indicated by full circles , axially - asymmetric toroidal configurations by full squares , and configurations with a spherical topology by open circles .
some toroidal density distributions are displayed . ] to study the occurrence of high - spin toroidal isomers we use a three - step method @xcite , ( see also @xcite ) .
first , using the skyrme - hartree - fock - bogoliubov ( hfb ) model with the quadrupole moment constraints on @xmath20 , we look for those oblate configurations with toroidal nuclear density distributions , as shown in fig .
the energies of axially - symmetric toroidal configurations as a function of @xmath20 lie on a slope .
this indicates that the magnitudes of the shell corrections are not sufficient to stabilize the tori against the tendency to return to the topology of a sphere .
we next take these toroidal configurations as the initial configurations in @xmath20-constrained cranking skyrme - hf calculations . for a non - collectively rotating toroidal nucleus around the symmetry
@xmath7-axis with aligned angular momentum , @xmath5=@xmath6 , we use a lagrange multiplier @xmath21 to describe the constraint @xmath22=@xmath23=@xmath24 , where @xmath25 is the @xmath7-component of @xmath26-th single - particle total angular momentum .
when we locate the configurations which lie close to a local minimum for each quantized value of angular momentum , @xmath5=@xmath6 , we repeat the cranked hf calculations without the @xmath20 constraint to find the high - spin toroidal isomeric states in free convergence in the last step . results of this method in the case of @xmath18fe are shown in fig .
[ fig2 ] .
( color online . ) the excitation energy @xmath27 of high - spin toroidal states of @xmath18fe as a function of @xmath20 for different @xmath5=@xmath6 about the symmetry axis .
the locations of isomeric toroidal energy minima are indicated by the star symbols . ] from the quantum mechanical point of view , the non - collective rotation around the symmetry axis corresponds only to particle - hole ( p - h ) excitations in the axially - symmetric , @xmath5=0 , nucleus .
we plot in fig .
[ fig3 ] the @xmath5=0 toroidal proton - quasiparticle state energies as a function of the quadrupole moment @xmath20 obtained in skm*-hfb model for @xmath18fe .
the toroidal quasiparticle states are labelled by asymptotic quantum numbers @xmath28\omega$ ] .
it is clear from fig .
[ fig3 ] , and also fig .
1(a ) in ref .
@xcite , that the low - lying states possess the nodal quantum number of @xmath29=0 .
the occupation numbers of 14 , 18 , 22 , and 26 in fig .
[ fig3 ] indicate the toroidal shell - gaps , as in fig .
1(a ) of @xcite .
the occupation of all levels below the fermi energy of this even - even nucleus lead to the state with @xmath5=0 , while the 2p-2h and 3p-3h particle - hole excitation result to high - spin toroidal isomeric states with @xmath6=26 and 40@xmath9 , respectively .
( color online . )
the toroidal proton - quasiparticle energies as a function of the quadrupole moment @xmath20 obtained in the skm*-hfb model for @xmath18fe with @xmath5=0 .
they are labelled by @xmath28\omega$ ] , with even parity levels as solid lines , and odd parity levels as dashed lines .
starting from the @xmath5=0 configuration , the 2p-2h and 3p-3h excitations shown in the plot ( with holes as open circles and particles in solid circular points ) lead to non - collective rotations with a total @xmath6=26 and 40@xmath9 , respectively .
the vertical dashed lines coincide with the quadrupole deformations of two first toroidal isomers ( fig .
[ fig2 ] ) .
the horizontal ( red ) dashed line represents zero quasiparticle energy for @xmath5=0 . ]
.[table1 ] the quantized values of aligned angular momentum @xmath5=@xmath22 for different particle - hole excitations of the states with @xmath30=0 , where parameter @xmath31=0,1,@xmath32 . [ cols="^,<,<",options="header " , ] table [ table1 ] gives the simple rules to calculate the aligned angular momentum @xmath5=@xmath22 for different p - h excitations from states with @xmath30=0 , for neutron number @xmath11 ( or proton number @xmath12 ) , such that ( @xmath11 or @xmath12 mod 4)= 0 or 2 .
for nucleus @xmath18fe with @xmath11=@xmath12=26 and a parameter @xmath31=6 , one can find from table [ table1 ] the aligned angular momentum @xmath22=2@xmath3314 , 2@xmath3326 , 2@xmath3340 , 2@xmath3352 , 2@xmath3366@xmath9 for 1- , 2- , 3- , 4- , 5-p - h excitations , ( see also fig .
1(b ) in ref .
@xcite ) . with our method to study the occurrence of toroidal high - spin isomers described above
, all the above p - h states turn out to be toroidal isomeric states as shown in fig .
[ fig2 ] , with the exception of the ( 1p-1h ) @xmath22=28@xmath9 state that is apparently below the threshold for a non - collectively rotating toroidal nucleus .
( color online . )
the density distributions of the isomeric toroidal states of @xmath18fe with @xmath5= 52 , 80 , 104 , and 132@xmath9 as a cut in the radial direction @xmath34 .
the values of the isomeric quadrupole moments @xmath20 are indicated .
for the first isomeric toroidal state with @xmath5= 52@xmath9 we show the major radius @xmath8 and the minor radius @xmath35 ( see text ) .
the dash - dot curve shows the density distribution in the ground state of @xmath18fe . ]
we plot in fig .
[ fig4 ] the density distributions of the isomeric toroidal states of @xmath18fe ( presented in fig .
[ fig2 ] ) with @xmath5= 52 , 80 , 104 , and 132@xmath9 as a cut in the radial direction @xmath34 .
one notes that when the aligned angular momentum @xmath5 increases , the maximum toroidal density @xmath36 decreases from 0.134 to 0.123 @xmath37 and the major radius @xmath8 increases from 7.39 to 8.20 fm .
only the minor radius @xmath35 , defined as a half width at half maximum ( hwhm ) of the toroidal distribution , stays constant at @xmath381.37 fm .
for comparison , we also show the total density distribution of @xmath18fe in its ground state ( dash - dot curve ) with @xmath36=0.173 @xmath37 which is distinctly larger than @xmath36 of toroidal distributions .
the above - mentioned features are typical for all high - spin toroidal isomeric states of even - even , @xmath12=@xmath11 nuclei with 28@xmath0a@xmath052 , ( see table 1 in ref .
@xcite ) . in fig .
[ fig5 ] we plot the total energy of all found isomers as a function of @xmath39 .
one can see that the total energies of isomers obtained by the same p - h excitation show linear dependence on @xmath39 .
there appears to be a regular pattern on the systematics of high - spin toroidal isomers from which properties on the nuclear fluid in the exotic toroidal shape may be extracted .
( color online . )
the total energy of the isomeric toroidal states of @xmath13si , @xmath14s , @xmath15ar , @xmath10ca , @xmath16ti , @xmath17cr , and @xmath18fe of different @xmath5 values as a function of @xmath39 . ]
non - collective nuclear rotations with a large angular momentum provide a favorable environment for the nuclear matter to redistribute itself .
our exploration of the density distribution of these nuclei with cranked self - consistent skyrme - hartree - fock approach in the region of 28@xmath0@xmath1@xmath052 reveals the possibility of toroidal - shape high - spin isomers at their local energy minima , when the angular momenta are greater than some large thresholds .
the particle - hole nature of these non - collective rotational states , the locations of these local energy minima , the magnitudes of the non - collective angular momenta , and the geometrical properties of these isomeric states have been systematically evaluated for further theoretical and experimental explorations . | nuclei under non - collective rotation with a large angular momentum above some threshold can assume a toroidal shape . in our previous work , we showed by using cranked skyrme - hartree - fock approach that even - even , n = z , high - k , toroidal isomeric states may have general occurrences for light nuclei with 28@xmath0@xmath1@xmath052 .
we present here some additional results and systematics on the particle - hole nature of these high - spin toroidal isomers . |
the observation of the discovered higgs boson @xmath5 with decaying into a pair of tau leptons @xcite provided more evidence for the existence of at least one @xmath6 doublet scalar field in addition to various other measurements of the properties of @xmath5 at the cern large hadron collider ( lhc ) .
then , the natural question is `` how many doublet fields are there in the higgs sector ? '' in general , multi - higgs doublet models ( mhdms ) can reproduce the predictions of the standard model ( sm ) composed of only one higgs doublet , and so we should take into account the possibility of the existence of multi - doublets . on the other hand ,
multi - doublet fields arise in many models beyond the sm . as the most familiar example , in the minimal supersymmetric sm ( mssm ) the higgs sector is extended to have two doublets because of the requirement of gauge anomaly cancellation .
in addition , the multi - doublet structure is required to have an additional source of cp - violation , which is necessary to realize the successful scenario of electroweak baryogenesis @xcite . moreover ,
additional doublets are often introduced in various loop - induced neutrino mass models , such as the model by zee @xcite and by ma @xcite etc .
the point is that the structure of mhdms depends on new physics scenarios , e.g. , in the mssm the so - called type - ii yukawa interaction is adopted .
therefore , by studying the phenomenology of mhdms with a bottom - up approach , we can narrow down the possible scenarios of new physics beyond the sm .
one of the most important issues when mhdms are discussed is the possibility of flavour changing neutral currents ( fcncs ) mediated by neutral higgs bosons , which appear at the tree level in general .
the simplest way to avoid such fcncs is to realize a yukawa lagrangian where each type of fermion , i.e. , charged leptons , up - type and down - type quarks couples to only one scalar doublet field .
this is usually achieved by imposing discrete symmetries in the higgs sector , and is so - called `` natural flavour conservation '' ( nfc ) @xcite .
depending on how doublet fields couple to each type of fermions , there are four ( five ) independent types of yukawa interactions in models with two ( more than two ) doublets @xcite . in this paper , we discuss the deviation in the couplings of @xmath5 to weak bosons ( @xmath3 and @xmath7 ) and fermions @xmath8 from the sm prediction in the mhdms with nfc . in ref .
@xcite , it has been clarified that the four types of yukawa interactions in two higgs doublet models ( 2hdms ) can be determined by measuring the correlation between the deviations in the charged lepton yukawa couplings @xmath9 and the down - type quark yukawa couplings @xmath10 as long as there is a non - zero deviation in the @xmath3 couplings .
now , we extend this discussion to models with @xmath11 higgs doublet fields , and we investigate how the pattern of the deviation can be different in 2hdms and in mhdms with @xmath12 . this paper is organized as follow . in sec .
ii , we define the mhdms with nfc .
we then give the general expressions for the @xmath3 and @xmath13 couplings .
we show more explicit forms of these couplings in the 2hdms and the three higgs doublet models ( 3hdms ) as simple examples . in sec .
iii , we give numerical results of the deviation in the couplings of @xmath5 especially for the @xmath1 and @xmath2 couplings in the 2hdms and the 3hdms .
conclusions are given in sec .
we consider models with @xmath11 higgs doublet fields and @xmath14 parameters . in ref .
@xcite , the maximally allowed number of doublets has been given as a function of a mass difference among higgs bosons . ]
@xmath15 @xmath16 with hypercharge is given by @xmath17 with @xmath18 being the third component of the isospin . ]
@xmath19 and vacuum expectation values ( vevs ) @xmath20 which are taken to be real . throughout this paper , we assume that the higgs sector has neither explicit nor spontaneous cp - violation for simplicity . in order to extract couplings among physical higgs bosons and gauge bosons or fermions , it is convenient to define the so - called higgs basis @xcite , where only one of the @xmath11 doublets has the vev @xmath21 which is related to the fermi constant @xmath22 by @xmath23 gev .
the higgs basis is defined via the @xmath24 orthogonal matrix @xmath25 by @xmath26 the doublets @xmath27 and @xmath28 @xmath29 appearing in the right - hand side of eq .
( [ hb ] ) are expressed as @xmath30 where @xmath31 and @xmath32 are the nambu - goldstone bosons which are absorbed into the longitudinal components of the @xmath33 and @xmath34 bosons , respectively . in eq .
( [ hb ] ) , the matrix @xmath25 is expressed in terms of @xmath35 angles , so that the @xmath11 vevs are translated into @xmath21 and the @xmath35 angles . by using @xmath25 ,
each of vevs is expressed as @xmath36 and the sum rule : @xmath37 follows from this equation . in the higgs basis ,
the @xmath35 pairs of singly - charged states @xmath38 , @xmath35 cp - odd states @xmath39 and @xmath11 cp - even states @xmath40 are not mass eigenstates in general .
their mass eigenstates can be defined by introducing @xmath41 unitarity matrices @xmath42 and @xmath43 for the singly - charged states and cp - odd states and an @xmath24 unitary matrix @xmath44 for the cp - even states as @xmath45 we identify the @xmath46 state as the discovered higgs boson with a mass of 125 gev .
let us move on to the construction of the kinetic term and the yukawa lagrangian for the scalar doublets in the higgs basis . without loss of generality , the kinetic term is expressed as follows : @xmath47 where @xmath48 is the covariant derivative for the doublet higgs fields . from the above lagrangian
, the gauge - gauge - scalar type interaction is obtained from the first term , because only @xmath27 contains the vev @xmath21 .
thus , we can extract the couplings of @xmath0 with weak bosons as @xmath49 where we take @xmath50 .
we note that from the second term in eq .
( [ kin ] ) , we obtain the scalar - scalar - gauge type interactions which are proportional to the derivative of a scalar field . these interaction terms can be important when we consider production and decay of extra higgs bosons .
next , we give the yukawa lagrangian . in the scenario based on nfc , the yukawa lagrangian takes the following form : @xmath51 where @xmath52 are any one of @xmath15 , and @xmath53 with @xmath54 being the second pauli matrix . here , we do not explicitly show the flavour indices .
the lagrangian given in eq .
( [ yuk ] ) is naturally realized by imposing discrete symmetries such as @xmath55 for models with @xmath11 higgs doublet fields .
under such a discrete symmetry , all the @xmath11 doublets can have a distinct charge from each other .
depending on the charge assignment for the right - handed fermions ) . ] , there appear five independent types of yukawa interactions .
they can be defined as follows : @xmath56 [ types ] a similar classification of the yukawa interactions has been presented in ref .
we note that the type - ii , type - x and type - y ( type - z ) are realized for @xmath57 ( @xmath12 ) .
the yukawa lagrangian can be rewritten in the higgs basis as @xmath58 + \text{h.c .
} , \end{aligned}\ ] ] where @xmath59 , @xmath60 and @xmath61 are respectively the mass matrix for up - type quarks , down - type quarks and charged leptons , which are given by @xmath62 for @xmath63 .
the @xmath64 factors are determined by the matrix elements of @xmath25 as follows : @xmath65 in terms of @xmath66 and @xmath64 , the yukawa couplings for @xmath0 are expressed by @xmath67 \times y_{hff}^{\text{sm}}. \label{yhff}\end{aligned}\ ] ] in order to express the deviations in the higgs boson couplings , we introduce the scaling factor for the gauge couplings by @xmath68 and for the yukawa couplings by @xmath69 .
we can express @xmath70 using @xmath71 by @xmath72 from the above expression , it is seen that @xmath70 become unity by taking the limit of @xmath73 , and this limit is the so - called `` alignment limit '' @xcite .
we thus can regard @xmath0 as the sm - like higgs boson for the case with @xmath74 .
because of the existence of the alignment limit , the mhdms can safely reproduce predictions in the sm .
notice here that in mhdms , @xmath75 must be smaller or equal to unity at the tree level multiplets such as triplets , @xmath76 is possible at the tree level @xcite . ] .
on the other hand , @xmath77 can be both smaller and larger than unity due to the second term of eq .
( [ kappaf ] ) .
.@xmath64 factors appearing in eq .
( [ yhff ] ) for each type of yukawa interaction in the 2hdm and the 3hdm . in the 2hdm ,
the @xmath78 factors are rewritten by @xmath79 . [
cols="^,^,^,^,^,^,^,^,^,^ " , ] based on the general discussion given in the previous subsection , we here consider two simple examples , i.e. , the case for @xmath80 ( 2hdms ) and for @xmath81 ( 3hdms ) . in the 2hdms
, the matrix @xmath25 takes the @xmath82 form as @xmath83 where @xmath84 .
we introduce shorthand notations for the trigonometric functions as @xmath85 ( @xmath86 ) @xmath87 . from eqs .
( [ xif ] ) and ( [ beta ] ) , we can find that @xmath78 ( notice that the superscript 2 does not mean the square of @xmath88 ) is taken to be either @xmath89 or @xmath90 depending on the type of yukawa interaction and the type of fermion .
if we fix @xmath91 , all the @xmath92 factors are fixed for each type of yukawa interaction and for each type of fermion as shown in table [ ratios ] .
in addition , the matrix @xmath44 also takes @xmath82 form with an angle independent of @xmath93 .
conventionally , this angle is expressed by @xmath94 ( see e.g. , @xcite ) , so that @xmath44 is written as @xmath95 because of the orthogonal property of the @xmath44 matrix , @xmath96 appearing in eq .
( [ kappaf ] ) is simply written by @xmath97 with @xmath98 $ ] , and thus @xmath70 is determined by @xmath99 , @xmath75 and @xmath100 .
then , @xmath70 can be rewritten by @xmath101 where @xmath102 and @xmath103 .
the far - right hand side is valid for @xmath104 which is supported by the current measurement at the lhc @xcite .
we note that because of the @xmath105 term of eq .
[ kappaf2 ] , we can take the so - called `` wrong sign limit '' @xcite defined by @xmath106 .
the case with this limit or this regime provides a phenomenologically interesting scenario , where loop induced decay rates such as @xmath107 and @xmath108 can be modified from the sm prediction due to effects of interference even having the same or similar value of the decay rate of @xmath109 and @xmath110 as those in the sm at the tree level .
( [ kappaf2 ] ) tells us the following important fact .
as long as @xmath111 , the sign of @xmath112 is determined by the sign of @xmath113 . in this case , the prediction of @xmath114 in the four types of yukawa interactions appears in the four different quadrant on the @xmath115@xmath116 plane for a fixed @xmath100 which can be determined by measuring the sign of @xmath117 .
namely for @xmath118 , the sign of @xmath114 is predicted by @xmath119 , @xmath120 , @xmath121 and @xmath122 in the type - i , -ii , -x and -y yukawa interaction , respectively , while for @xmath123 these signs are flipped as compared to the former case .
therefore , by the precise measurements of @xmath115 and @xmath116 , we can determine the type of yukawa interaction in the 2hdm .
although this statement is valid for the case with @xmath124 for all @xmath125 as mentioned above , the case with at least one of @xmath126 being smaller than @xmath127 is phenomenologically quite difficult to realize in the type - ii , -x and -y 2hdm , which is explained following . in order to achieve , the condition @xmath128 is satisfied unless we take @xmath129 which is disfavored by the various @xmath130 physics constraints mainly due to the enhancement of the top yukawa coupling .
see , e.g. , @xcite for the constraints on the parameter space from @xmath130 physics in 2hdms . ]
@xmath131 , we need a rather large value of @xmath99 if @xmath132 . for example , when @xmath133 is given to be @xmath134 , @xmath99 should be larger than about 6 ( 14 ) to satisfy @xmath135 .
the important point is that such a case gives a huge deviation in some of yukawa couplings in the type - ii , -x , and -y 2hdm .
in fact in the type - ii 2hdm , @xmath136 for the case of @xmath137 , @xmath138 and @xmath139 .
similarly in the type - x ( type - y ) 2hdm , @xmath115 @xmath140 becomes the above value , and it goes without saying that such a huge deviation has already been excluded by the current data at the lhc @xcite .
if @xmath141 is realized , only the type - i 2hdm provides a phenomenologically acceptable scenario .
next , we consider the 3hdms . in this case , the matrix @xmath25 takes the @xmath142 form , and its one of the explicit forms is given by @xcite @xmath143 where @xmath144 and @xmath145 .
a similar but different parameterization of the three vevs has also been given in ref .
@xcite . using this notation ,
each of @xmath64 factors are expressed as in table [ ratios ] , where we assign @xmath91 , and @xmath146 ( if these are different from @xmath147 ) in the type - ii , -x and -y interactions . for the type - z , we assign @xmath148 . unlike the 2hdms , the second term of eq .
( [ kappaf ] ) is not simply determined by @xmath75 , so that to get the prediction of @xmath70 , we need to further input @xmath149 and @xmath150 under the constraint of @xmath151 .
we here discuss critical differences of the higgs boson couplings between in the 2hdms and in the 3hdms .
first , as it is immediately seen , the type - z yukawa interaction is only realized in 3hdms . in other words in 2hdms
, there is at least one pair of @xmath70 and @xmath152 ( @xmath153 ) with @xmath154 as seen in ( [ types ] ) , while the type - z can provide the prediction of @xmath155 , @xmath156 and @xmath157 .
next , in 2hdms the prediction of @xmath158 with @xmath159 or @xmath160 is not allowed as explained by the following .
first of all , @xmath161 is realized in the type - x and type - y 2hdm .
however , in these types , e.g. , in the type - x , @xmath162 and @xmath163 .
therefore for a fixed @xmath100 , the sign of @xmath115 is opposite to that of @xmath116 as long as @xmath128 . on the other hand , this is not the case for the 3hdms as it will numerically be shown in the next section .
in this section , we show the correlation between @xmath164 and @xmath165 in the 2hdms and in the 3hdms .
we choose the following parameters as inputs : @xmath166 where @xmath167 $ ] . in these input parameters for the 3hdm
, @xmath168 is derived by @xmath169 ^ 2}$ ] .
for all the calculations below , we scan the value of @xmath170 in the range from @xmath171 to @xmath172 . in order to clarify the possible allowed predictions in the 3hdms which can not be explained in the 2hdms
, we take into account the one - loop electroweak corrections to the yukawa couplings in the 2hdms based on the on - shell renormalization scheme according to refs .
@xcite . for the one - loop calculation
, we scan the parameters of the 2hdm by @xmath173 gev , @xmath174 and @xmath175 with @xmath176 , where @xmath177 is the mass of extra higgs bosons ( here we assume that all the extra higgs bosons are degenerate in mass ) , and @xmath178 @xcite is a dimensionful parameter which is irrelevant to the higgs vev ( see , e.g. , @xcite for the detailed explanation about these parameters ) . for the sm inputs
, we use the following values @xcite @xmath179 we take into account theoretical constraints on the model parameters from the perturbative unitarity @xcite , the vacuum stability @xcite , and the triviality with the criterion that the landau pole does not appear below 3 tev . we note that the amount of qcd corrections to the @xmath2 coupling can be a sub percent level in the sm @xcite , which are not included in our calculation . and @xmath165 plane in the type - i , -ii , -x and -y 2hdms in the case of @xmath180 .
the solid curves show the tree level prediction .
each dot on these curves denotes the prediction of the fixed value of @xmath99 , where its value is written beside the dot .
the gray shaded regions show the one - loop corrected results .
the left and right panels respectively show the case for @xmath181 and @xmath182 .
the largest ( green ) dotted ellipse shows the 2@xmath100 error of the measurement of @xmath164 and @xmath165 from the current lhc data extracted from ref .
@xcite , while the middle ( cyan ) and smallest ( orange ) ones respectively show the expected @xmath183 accuracy of the measurement at the lhc with the collision energy of 14 tev and the integrated luminosity of 300 fb@xmath184 and at the ilc with the collision energy of 500 gev and the integrated luminosity of 500 fb@xmath184 ) extracted from @xcite . , title="fig:",width=264 ] and @xmath165 plane in the type - i , -ii , -x and -y 2hdms in the case of @xmath180 .
the solid curves show the tree level prediction .
each dot on these curves denotes the prediction of the fixed value of @xmath99 , where its value is written beside the dot .
the gray shaded regions show the one - loop corrected results .
the left and right panels respectively show the case for @xmath181 and @xmath182 .
the largest ( green ) dotted ellipse shows the 2@xmath100 error of the measurement of @xmath164 and @xmath165 from the current lhc data extracted from ref .
@xcite , while the middle ( cyan ) and smallest ( orange ) ones respectively show the expected @xmath183 accuracy of the measurement at the lhc with the collision energy of 14 tev and the integrated luminosity of 300 fb@xmath184 and at the ilc with the collision energy of 500 gev and the integrated luminosity of 500 fb@xmath184 ) extracted from @xcite . ,
title="fig:",width=264 ] first , we show the difference of the prediction of @xmath164 and @xmath165 in the 2hdms with four types of yukawa interaction . in fig .
[ fig1 ] , we show the correlation between @xmath164 and @xmath165 in the four types of yukawa interaction in the 2hdm .
we here take @xmath185 , @xmath174 and @xmath186 in the left ( right ) panel .
the solid curves show the tree level prediction , while the shaded regions do the one - loop corrected result .
we can see that the predictions of the four types of yukawa interaction are well separated in the @xmath164@xmath165 plane even taking into account the one - loop correction to the yukawa couplings . in this figure
, we display the @xmath183 error of the measurement of @xmath164 and @xmath165 at the lhc run-1 experiment errors of @xmath187 and @xmath188 from ref .
@xcite , but we do not use their central values . ] by the green dotted ellipse , where data from the atlas and cms collaborations are combined @xcite . in addition , we also show the expected @xmath183 accuracy of the measurement of @xmath164 and @xmath165 at the lhc with 14 tev and 300 fb@xmath184 ( denoted as lhc300 ) and at the international linear collider ( ilc ) with 500 gev and 500 fb@xmath184 ( denoted as ilc500 ) .
although the current measurement ( lhc run-1 ) is not enough accurate to separate the four types of yukawa interaction in the configuration with @xmath189 , we may be able to discriminate these types of yukawa interaction by the future measurements at the lhc300 and the ilc500 .
we note that for a larger ( smaller ) value of @xmath133 , the distance of each prediction in the four types of yukawa interaction becomes to be more ( less ) spread . as the extreme case ,
when @xmath190 is taken , all the predictions converge at @xmath191 . and
@xmath165 in the 2hdm and in the 3hdm in the case of @xmath180 .
the left - upper , right - upper , left - lower and right - lower panel respectively shows the case for ( type - x , @xmath123 ) , ( type - y , @xmath123 ) , ( type - x , @xmath192 ) and ( type - y , @xmath192 ) . similar to fig .
[ fig1 ] , the black curve shows the tree level prediction , and the gray shaded region does the one - loop corrected result in the 2hdm .
the red , blue , green and magenta curve respectively shows the tree level prediction in the 3hdm with @xmath193 , 2 , 3 and 5 in the 3hdm .
for all the results in the 3hdm , we take @xmath194 .
the description of the three ellipses is the same as those in fig .
, title="fig:",width=264 ] and @xmath165 in the 2hdm and in the 3hdm in the case of @xmath180 .
the left - upper , right - upper , left - lower and right - lower panel respectively shows the case for ( type - x , @xmath123 ) , ( type - y , @xmath123 ) , ( type - x , @xmath192 ) and ( type - y , @xmath192 ) . similar to fig .
[ fig1 ] , the black curve shows the tree level prediction , and the gray shaded region does the one - loop corrected result in the 2hdm .
the red , blue , green and magenta curve respectively shows the tree level prediction in the 3hdm with @xmath193 , 2 , 3 and 5 in the 3hdm .
for all the results in the 3hdm , we take @xmath194 .
the description of the three ellipses is the same as those in fig .
, title="fig:",width=264 ] + and @xmath165 in the 2hdm and in the 3hdm in the case of @xmath180 .
the left - upper , right - upper , left - lower and right - lower panel respectively shows the case for ( type - x , @xmath123 ) , ( type - y , @xmath123 ) , ( type - x , @xmath192 ) and ( type - y , @xmath192 ) . similar to fig .
[ fig1 ] , the black curve shows the tree level prediction , and the gray shaded region does the one - loop corrected result in the 2hdm .
the red , blue , green and magenta curve respectively shows the tree level prediction in the 3hdm with @xmath193 , 2 , 3 and 5 in the 3hdm .
for all the results in the 3hdm , we take @xmath194 .
the description of the three ellipses is the same as those in fig .
, title="fig:",width=264 ] and @xmath165 in the 2hdm and in the 3hdm in the case of @xmath180 .
the left - upper , right - upper , left - lower and right - lower panel respectively shows the case for ( type - x , @xmath123 ) , ( type - y , @xmath123 ) , ( type - x , @xmath192 ) and ( type - y , @xmath192 ) . similar to fig .
[ fig1 ] , the black curve shows the tree level prediction , and the gray shaded region does the one - loop corrected result in the 2hdm .
the red , blue , green and magenta curve respectively shows the tree level prediction in the 3hdm with @xmath193 , 2 , 3 and 5 in the 3hdm .
for all the results in the 3hdm , we take @xmath194 .
the description of the three ellipses is the same as those in fig .
, title="fig:",width=264 ] next , we compare the prediction of @xmath164 and @xmath165 in the 2hdm and in the 3hdm . because in the type - i and type - ii case , tree level predictions in both the 2hdm and the 3hdm are given on the line of @xmath195 , it is difficult to see the difference between these two models .
we thus compare the prediction in the type - x and type - y yukawa interaction . in fig .
[ fig2 ] , we show the correlation between @xmath164 and @xmath165 in the 2hdm and in the 3hdm in the case of @xmath180 .
the results for the type - x ( type - y ) are given in the left ( right ) panel , while those in the case for @xmath196 @xmath197 are displayed in the upper ( lower ) panel .
clearly , we can find the region which can not be drawn by the one - loop corrected prediction in the 2hdm , but can be explained in the 3hdm .
we note that in the 3hdms with a larger ( smaller ) value of @xmath198 , each of the parabolas becomes that with a larger ( small ) curvature through the same point given in the 2hdm denoted by the black dot . and @xmath165 in the type - z 3hdm in the case of @xmath180 .
the left ( right ) panel shows the case for @xmath199 ( @xmath200 ) .
the value of @xmath198 is taken to be 1 ( solid curve ) , 2 ( dashed curve ) and 1/2 ( dotted curve ) .
the description of the three ellipses is the same as that in fig .
, title="fig:",width=264 ] and @xmath165 in the type - z 3hdm in the case of @xmath180 .
the left ( right ) panel shows the case for @xmath199 ( @xmath200 ) .
the value of @xmath198 is taken to be 1 ( solid curve ) , 2 ( dashed curve ) and 1/2 ( dotted curve ) .
the description of the three ellipses is the same as that in fig .
, title="fig:",width=264 ] finally , we show the prediction in the type - z 3hdm on the @xmath164 and @xmath165 plane .
the left and right panel show the case for @xmath199 and @xmath200 , respectively .
we can see that the predictions with @xmath201 ( @xmath202 ) and @xmath203 are allowed for the case of @xmath204 ( @xmath205 ) , which are not allowed in the 2hdms at the tree level .
even if we take into account the one - loop corrections to the yukawa coupling , only a few percent level of the difference between @xmath164 and @xmath165 is allowed as we see in fig .
therefore , if the higgs boson couplings are measured to be in this region , i.e. , @xmath201 ( @xmath202 ) and @xmath203 , it can be evidence for the mhdms with @xmath12 .
we have studied the sm - like higgs boson couplings with weak bosons @xmath3 and fermions @xmath13 in mhdms with nfc . we have presented the generic expression for the scaling factors of @xmath75 and @xmath70 in the model with @xmath11 doublet scalar fields . as simple concrete examples
, we have discussed the 2hdms and the 3hdms , and have presented the formulae of these scaling factors in terms of the ratios of the higgs vevs and the matrix elements of the diagonalization matrix for the cp - even scalar states for each type of yukawa interaction .
we then have shown the correlation between @xmath164 and @xmath165 in the 2hdms and in the 3hdms under the assumption that the @xmath3 couplings deviate from the sm prediction by 1% .
it has been clarified that there are predictions in the @xmath164 and @xmath165 plane in the 3hdms , which can not be explained in the 2hdms even when taking into account the one - loop corrections to the yukawa coupling .
we would like to mention that such a region can also be explained within a 2hdm if we relax the framework of nfc such as the so - called type - iii 2hdm .
however , in such a model fcncs mediated by neutral higgs bosons are naturally induced , so that measurements at flavour experiments also become important to discriminate the models with nfc and those without nfc . finally , we briefly comment on the case with cp - violation in the higgs sector which is assumed not being occurred in this paper . if there is a non - zero physical cp - violating phase in the higgs potential , cp - even and cp - odd component scalar fields
are mixed with each other . through the mixing
, the yukawa coupling for the sm - like higgs boson contains a term proportional to the @xmath206 matrix , which can modify various physical quantities such as decay rates and cross sections of the sm - like higgs boson from the sm prediction even if @xmath207 is taken .
the collider phenomenology with such a cp - violating effect has been studied in refs .
@xcite in the 2hdms .
in addition , the method to extract the cp - even component ( without @xmath206 ) and the cp - odd component ( with @xmath206 ) of the @xmath208 coupling has been proposed in ref .
@xcite by measuring the @xmath208 cross section weighted by an operator constructed from the ( anti-)top quark momentum .
the author is grateful to shinya kanemura for fruitful discussions .
he also thanks to andrew akeroyd for careful reading of the manuscript .
this work was supported by a jsps postdoctoral fellowships for research abroad .
g. aad _ et al . _ [ atlas collaboration ] , jhep * 1504 * , 117 ( 2015 ) [ arxiv:1501.04943 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , jhep * 1405 * , 104 ( 2014 ) [ arxiv:1401.5041 [ hep - ex ] ] .
n. turok and j. zadrozny , phys .
lett . * 65 * , 2331 ( 1990 ) .
a. i. bochkarev , s. v. kuzmin and m. e. shaposhnikov , phys .
d * 43 * , 369 ( 1991 ) .
a. e. nelson , d. b. kaplan and a. g. cohen , nucl . phys .
b * 373 * , 453 ( 1992 ) .
a. zee , phys .
lett.b * 93 * , 389 ( 1980 ) [ erratum - ibid.b * 95 * , 461 ( 1980 ) ] .
y. grossman , nucl .
b * 426 * , 355 ( 1994 ) .
s. kanemura , k. tsumura , k. yagyu and h. yokoya , phys . rev .
d * 90 * , 075001 ( 2014 ) [ arxiv:1406.3294 [ hep - ph ] ] .
s. davidson and h. e. haber , phys .
d * 72 * , 035004 ( 2005 ) [ phys .
d * 72 * , 099902 ( 2005 ) ] .
[ hep - ph/0504050 ] .
p. m. ferreira , l. lavoura and j. p. silva , phys .
b * 688 * , 341 ( 2010 ) [ arxiv:1001.2561 [ hep - ph ] ] .
bhupal dev and a. pilaftsis , jhep * 1412 * , 024 ( 2014 ) erratum : [ jhep * 1511 * , 147 ( 2015 ) ] [ arxiv:1408.3405 [ hep - ph ] ]
. s. kanemura , m. kikuchi and k. yagyu , phys .
d * 88 * , 015020 ( 2013 ) [ arxiv:1301.7303 [ hep - ph ] ] .
g. c. branco , p. m. ferreira , l. lavoura , m. n. rebelo , m. sher and j. p. silva , phys .
rept . * 516 * , 1 ( 2012 ) [ arxiv:1106.0034 [ hep - ph ] ] .
j. f. gunion , h. e. haber , g. l. kane and s. dawson , front .
* 80 * , 1 ( 2000 ) .
g. aad _ et al . _ [ atlas and cms collaborations ] , jhep * 1608 * , 045 ( 2016 ) [ arxiv:1606.02266 [ hep - ex ] ] .
p. m. ferreira , j. f. gunion , h. e. haber and r. santos , phys .
d * 89 * , no . 11 , 115003 ( 2014 ) [ arxiv:1403.4736 [ hep - ph ] ] . f. mahmoudi and o. stal , phys . rev .
d * 81 * , 035016 ( 2010 ) .
[ arxiv:0907.1791 [ hep - ph ] ] .
t. enomoto and r. watanabe , jhep * 1605 * , 002 ( 2016 ) [ arxiv:1511.05066 [ hep - ph ] ] .
s. kanemura , m. kikuchi and k. yagyu , phys .
b * 731 * , 27 ( 2014 ) [ arxiv:1401.0515 [ hep - ph ] ] .
s. kanemura , m. kikuchi and k. yagyu , nucl .
b * 896 * , 80 ( 2015 ) [ arxiv:1502.07716 [ hep - ph ] ] .
s. kanemura , y. okada , e. senaha and c .- p .
yuan , phys .
d * 70 * , 115002 ( 2004 ) [ hep - ph/0408364 ] .
k. a. olive _ et al .
_ [ particle data group collaboration ] , chin . phys .
c * 38 * , 090001 ( 2014 ) .
s. dawson _
_ , arxiv:1310.8361 [ hep - ex ] .
a. barroso , p. m. ferreira , r. santos and j. p. silva , phys . rev .
d * 86 * , 015022 ( 2012 ) [ arxiv:1205.4247 [ hep - ph ] ] .
d. fontes , j. c. romo , r. santos and j. p. silva , phys .
d * 92 * , no . 5 , 055014 ( 2015 ) [ arxiv:1506.06755 [ hep - ph ] ] . v. keus , s. f. king , s. moretti and k. yagyu , jhep * 1604 * , 048 ( 2016 ) [ arxiv:1510.04028 [ hep - ph ] ] .
j. f. gunion and x. g. he , phys . rev .
lett . * 76 * , 4468 ( 1996 ) [ hep - ph/9602226 ] . | we investigate the deviation in the couplings of the standard model ( sm ) like higgs boson ( @xmath0 ) with a mass of 125 gev from the prediction of the sm in multi - doublet models within the framework where flavour changing neutral currents at the tree level are naturally forbidden .
after we present the general expressions for the modified gauge and yukawa couplings for @xmath0 , we show the correlation between the deviation in the yukawa coupling for the tau lepton @xmath1 and that for the bottom quark @xmath2 under the assumption of a non - zero deviation in the @xmath3 @xmath4 couplings in two higgs doublet models ( 2hdms ) and three higgs doublet models ( 3hdms ) as simple examples .
we clarify the possible allowed prediction of the deviations in the 3hdms which can not be explained in the 2hdms even taking into account the one - loop electroweak corrections to the yukawa coupling . |
despite their exotic nature , general relativity ( gr ) tells us that black holes are remarkably simple objects , at least as far as the external universe is concerned . in astrophysical settings ( where charge neutrality is essentially guaranteed and hence the spacetime
is described by the kerr metric ; kerr 1963 ) , an isolated black hole is fully characterized by just its mass ( @xmath2 ) and angular momentum ( usually specified in terms of the dimensionless spin parameter @xmath3 , where @xmath4 is the angular momentum and the cosmic censorship hypothesis ensures that @xmath5 ) . for gravitational processes , the mass acts as a trivial scaling factor for distances , timescales , and energies .
this directly underlies the fact that accreting black holes across the entire mass scale ( from those in x - ray binaries [ bh - xrbs ] to active galactic nuclei [ agn ] ) have very similar spectral and temporal characteristics , once one scales the energies and timescales appropriately .
spin , however , is a much more interesting quantity . depending upon the spin
, the kerr metric can possess qualitatively different properties that have astrophysical significance .
for example , for slowly rotating black holes , the innermost stable circular orbit ( isco ) is far from the ergosphere ; accreting matter will be well into its ballistic
death plunge " before encountering the ergosphere and the strong frame dragging effects .
but for @xmath6 , the isco moves into the ergosphere raising the possibility that the accretion disk can tap into and radiate the rotational energy of the black hole ( agol & krolik 2000 ) .
indeed , the current paradigm is that relativistic jets from black holes are powered by the magnetic extraction of rotational energy from a central kerr black hole ( blandford & znajek 1977 ) .
clearly , measuring black hole spin is of great interest .
it is a necessary step in any kerr - metric based tests of strong - field gr .
spin measurements are also vital if we are to understand the astrophysical consequences of spin and , particularly , give observational grounding to theoretical notions such as the spin - driving of relativistic jets .
measuring spin is a challenging endeavour , however to find the signatures of spin , we must search for and characterize ( with high precision ) signals from within a few gravitational radii of the black hole . in this chapter ,
i review the progress that has been made in measuring black hole spins using relativistic x - ray reflection spectroscopy ( a.k.a .
broad iron line spectroscopy ) . with this one technique ,
we can study black hole spin across the mass range , from the stellar - mass black holes in x - ray binaries to the supermassive black holes in active galactic nuclei ( agn ) .
a major advantage of this method is that we require no knowledge of the black hole mass or its distance in order to derive the black hole spin .
neither do we require knowledge of the accretion disk inclination indeed , x - ray reflection spectroscopy allows us to measure the inclination concurrently with the black hole spin .
in contrast to the continuum fitting ( cf ) method for measuring spins ( mcclintock & narayan , this volume ) , the relativistic reflection method has been developed , evolved and debated by several groups over the past few years , producing a literature that can be rather bewildering .
hence , in this chapter , i review the assumptions on which the technique rests , summarize the current best practices " for employing this technique , and describe how careful consideration of the astrophysical consistency ( and , in some cases , instrumental issues compromising the data ) can resolve discrepancies in published spins . reflecting my own biases ,
much of this discussion will be dedicated to supermassive black holes ( smbhs ) , but i shall also describe recent work on stellar - mass black holes in bh - xrbs and the address the concordance between relativistic reflection and cf spin measurements .
i shall end by discussing recent results on relativistic reverberation .
this chapter is organized as follows .
section [ sec : methodology ] lays out the assumptions and methodology of the relativistic reflection spin measurements .
the results for smbhs are discussed in section [ sec : smbh ] , using two case studies to highlight important systematic issues that can affect spin measurements .
stellar - mass black holes in bh - xrbs are discussed in section [ sec : xrb ] .
the dramatic progress that has been made recently in characterizing relativistic reverberation is summarized in section [ sec : reverberation ] .
the casual reader who is interested more in the current status of spin results rather than details of the methodology may wish to skip sections [ sec : nuts ] and [ sec : agn_cases ] . unless otherwise stated , all spins quoted in this review are given with their 90% error ranges .
in both spectroscopic methods for determining black hole spin ( cf and relativistic reflection ) , the central assumption is that the accretion disk remains geometrically - thin , optically - thick , and radiatively efficient down to the isco .
as the accretion flow crosses over the isco , the inward radial velocity rapidly increases , the flow quickly becomes super - sonic and super - alfvenic ( hence ballistic ) and , due to conservation of mass flux , the density of the flow rapidly drops .
this basic behavior is confirmed by magnetohydrodynamic simulations of thin accretion disks ( reynolds & fabian 2008 ; penna et al .
as discussed below , the result is that the isco serves as the inner - edge " of the various observables , including the x - ray reflection features that are our focus here .
the ability to measure spin then follows directly from the fact that the location of the isco ( in units of gravitational radii @xmath7 ) is a simple function of spin ( bardeen , press & teukolsky 1972 ) .
this central assumption is likely to be valid for sources accreting at modest rates , with bolometric luminosities in the range @xmath8 ( where @xmath9 is the usual eddington luminosity ) .
this includes the thermal dominant ( a.k.a .
high / soft ) state the of bh - xrbs , as well as the more luminous broad - optical - line seyfert galaxies and radio - galaxies .
thermal emission from the accretion disk is clearly visible in these classes of sources , in the soft x - ray band for bh - xrbs and the optical / uv / euv bands for agn .
generically , however , the spectra of these sources also show a hard x - ray power - law tail which must originate from some structure external to the accretion disk , either a disk corona or the base of a jet .
this geometry is key to the reflection method for spin determination .
the x - ray source irradiates the underlying accretion disk with a hard x - ray continuum . outside of the isco , where the accretion disk is optically - thick and still contains combined metal ions
, the irradiated surface emits a complex spectrum consisting of atomic emission features sitting on a compton - backscattered continuum ( fig .
[ fig : reflection ] , left ) .
this is the _ x - ray reflection spectrum _ , and provided that the disk photosphere is not too highly photoionized , the most prominent feature is the iron - k@xmath10 emission line ( rest - frame energy @xmath11kev depending upon ionization state ) .
doppler shifts and gravitational redshifts , that become progressively stronger as one considers radii in the disk closer to the black hole , broaden and skew the observed features in a well - defined manner ( fig .
[ fig : reflection ] , right ) . within the isco , the density of the accretion flow plummets and it will be fully ionized by any significant x - ray irradiation of the flow .
hence , the region within the isco will not contribute to the observed atomic signatures ( although there may still be a compton - reflected continuum component arising from within the isco ) .
in other words , the isco defines the inner edge to the region producing spectrally structured " x - ray reflection . given these basics ,
how do we proceed at an operational level to analyze a real spectrum and extract black hole spin ? to isolate the signatures of relativistic reflection needed for the spin measurement , we must understand the spectrum of the primary continuum , the effects of absorption , and the presence of other reprocessing signatures ( e.g. , emission from an optically - thin photoionized wind , or x - ray reflection from distant structures such as the companion star in a bh - xrb or the molecular torus in an agn ) . here , i shall focus on the methodology developed for ( non - blazar , compton - thin ) agn , mentioning a few operational differences with bh - xrbs towards the end of the section . in a disk - dominated agn ,
the primary x - ray continuum in the crucial 250kev range is very well approximated by a power - law .
this is readily understood in terms of a separation of energy - scales .
the x - rays are believed to be produced by the thermal comptonization of soft optical / uv photons ( @xmath12ev ) by an electron population that is hot ( @xmath13kev ) .
for photon energies @xmath14 satisfying @xmath15 , comptonization generically produces a power - law .
the most common and obvious complexities observed in agn spectra are ( i ) a narrow iron - k@xmath10 emission line at 6.4kev due to reflection / fluorescence by distant and low - velocity material , and ( ii ) absorption by cold and photoionized gas along the line - of - sight to the central x - ray source .
thus , to our power - law spectral model , we add a model of cold , distant x - ray reflection ( e.g. , using the pexmon model which self - consistently includes the iron - k@xmath10 , iron - k@xmath16 , and nickel - k@xmath10 lines as well as the compton reflected continuum ; nandra et al .
2007 ) . to treat absorption ,
we add successive absorption components ( each with its own ionization parameter and column density ; computing using a photoionization code such as xstar ; kallman & bautista 2001 ) until no further improvement in the fit to the data is found .
the final step before dealing with the relativistic reflection is to add a phenomenological component ( a blackbody or additional comptonization component ) to treat the soft excess that many agn display . in approximately 3050% of type-1 agn
, applying this procedure to data with sufficient signal - to - noise will leave unmodelled residuals indicative of a broad / redshifted iron line , excess compton reflection , i.e. a hump at @xmath17kev ( nandra et al . 2007 ) .
these are the signatures of relativistic reflection that we seek to characterize although , as discussed below , there are alternative explanations for the soft excess that can lead to important ambiguities .
thus , we add to the spectral model a component describing an ionized reflection spectrum ( the current industry standard is reflionx ; ross & fabian 2005 ) that has been convolved by a
disk - line transfer function " encoding the doppler / gravitational redshifts from the accretion disk .
the available and commonly employed disk - line convolution models ( e.g. , relconv ; dauser et al .
2010 ) are based on gr ray - tracing calculations within the kerr metric , assuming a razor thin accretion disk lying in the equator of the rotating black hole .
in addition to the black hole spin @xmath18 and the disk - inclination @xmath19 , the disk - line transfer function depends upon the intensity of the irradiating flux as a function of radius , @xmath20 , that , of course , depends upon the nature of the primary x - ray source .
it is sometimes said that the lack of knowledge of the location and structure of the x - ray source is a major limitation for studies of relativistic reflection .
in fact , from the point of view of spin measurements , all we require is a parameterized form of @xmath20 that can then be constrained by the data themselves .
empirically , it is found that a broken - powerlaw is a good description of this irradiation profile , giving up to five additional fit parameters , the inner / break / outer radii @xmath21 and the two power - law indices @xmath22 defined such that in the relevant patch we have @xmath23 .
as detailed above , we identify @xmath24 with the isco .
furthermore , provided that @xmath25 , the spectrum will be insensitive to a sufficiently large @xmath26 .
thus , the total set of parameters describing the relativistic reflection is @xmath27 . the final spectral model describing the continuum source , the distant reprocessing , absorption components , and relativistic reflection from the accretion disk , is then fitted to the data ( with all interesting spectral parameters , including those describing the continuum form and the absorption components allowed to be free ) and constraints on the spin can be obtained .
there are two aspects of this methodology that must be underscored .
firstly , in many situations , the error bar on the derived spin will be driven as much by uncertainties in the continuum / absorber parameters as by uncertainties in the form of the relativistic reflection itself .
thus , it is crucial to allow all relevant spectral parameters to vary in the final spectral analysis , including those describing the continuum and absorption .
secondly , when attempting to measure black hole spin , one must model the full ionized reflection spectrum .
phenomenological models in which isolated broad iron lines are added to the primary continuum are useful for initial explorations of spectra , but are inadequate for characterizing the subtle spectral signatures of spin .
a very similar procedure is adopted in the case of bh - xrbs , although there are a few differences compared with the case for agn driven by the fact that xrb accretion disks are hotter , the astrophysical environments are cleaner and , typically , the signal - to - noise ( s / n ) in the datasets is much better .
absorption is much less of a concern in studies of the disk reflection in bh - xrbs .
strong outflows are confined to strongly thermally - dominated states ( of potential concern for cf spin measurements ) but are weak or absent during states that display prominent reflection ( miller et al .
2006 , 2008 ; king et al .
2011 , 2012 ) .
even when present , the outflows tend to have significantly higher ionization states ( such that iron has a fully stripped l - shell ) and , hence , absorption tends not to introduce broad - band curvature that can confuse studies of the broad iron line .
on the other hand , the thermal disk spectrum in bh - xrb is now in the x - ray band and must be modeled if one hopes to robustly probe the broad iron line and other disk reflection signatures .
furthermore , the disk reflection features ( and especially the iron - k lines ) have significant broadening even in the rest frame from compton scattering in the hot disk atmosphere . while this additional broadening is captured by the ionized reflection models ( see top model in left panel of fig .
[ fig : reflection ] ) , there is some uncertainty in the degree of comptonization that can affect spin determinations ( see discussion of xte j1550564 in section [ sec : xrb ] ) .
of course , the s / n is often extremely high for bh - xrbs .
indeed , one must start to consider bright source " instrumental issues such as errors in the calibration of the effective area and photon pile - up .
miller et al .
( 2010 ) has explicitly shown that pile - up tends to make a broad line narrower and , hence , always acts to decrease the inferred spin .
we return to this point in the discussion of cygnus x-1 in section [ sec : xrb ] .
we end this section on methodology by highlighting the complementarities and strengths of the relativistic reflection method in comparison to the cf method . as discussed by mcclintock & narayan ( this volume ) , the cf method , in its essence , uses the stefan - boltzamann law to measure the emitting area of the accretion disk ( employing a novikov & thorne [ 1974 ] emission profile ) and hence the size ( in physical units ) of the isco .
to apply this technique requires accurate optical / ir measurements of mass ( to convert the isco into units of gravitational radii , @xmath28 ) , distance ( in order to derive luminosities ) , and inclination , as well as a trustworthy absolute x - ray flux calibration .
by contrast , the relativistic transfer function underlying the reflection method is completely independent of mass , i.e. , all radii are naturally scaled in units @xmath29 , and one never needs to know the explicit size of @xmath29 . hence , mass uncertainties are irrelevant to reflection - based spins . in addition , since we are interested purely in spectral shape , the distance to the source and the absolute flux is also irrelevant ( although , for agn , the cosmological redshift is important to know )
. finally , while the inclination of the inner disk is an important quantity , it can be directly measured along with spin from the x - ray spectrum .
the first quantitative attempt to constrain black spin was based on modeling of the broad iron line profile in the seyfert galaxy mcg6 - 30 - 15 .
a long _ asca _ observation of this source had caught its transition to the so - called deep minimum state " in which iwasawa et al .
( 1996 ) noticed that the iron line was too broadened and redshifted to be explained by a non - rotating black hole .
subsequent modeling of this line profile by dabrowski et al .
( 1997 ) concluded that @xmath30 under the assumption that the irradiation profile of the reflection follows a novikov - thorne profile . in its original form , however , this analysis was not completely robust reynolds & begelman ( 1997 ) showed that , in a lamp - post geometry , the combination of light - bending effects and a small penetration of the line - emitting region inside of the isco could render the deep minimum line profile consistent with schwarzschild ( @xmath31 ) geometry .
young , ross & fabian ( 1998 ) showed that the need for rapid spin in mcg6 - 30 - 15 was restored when one considered the full reflection continuum , showing that the reynolds & begelman [ 1997 ] scenario predicted a strong , and unobserved , absorption edge from ionized iron .
this was the first realization that modeling the full reflection spectrum , not just the iron line , was crucial for spin determinations . the first spin - analysis to follow the modern methodology ( section [ sec : methodology ] ) was by brenneman & reynolds ( 2006 ) , also for mcg6 - 30 - 15 .
using a long and high s / n _ xmm - newton _ observation , they confirmed the rapid spin in mcg6 - 30 - 15 , obtaining a formal limit of @xmath32 ( although finite disk - thickness effects can relax this limit to @xmath33 ; reynolds & fabian [ 2008 ] ) .
wider application of these techniques to other agn had to await the availability of sufficiently deep observations . over the past few years
, the number of useful datasets has expanded , leading to an explosion in spin studies .
we now discuss two case studies that demonstrate the quality , as well as possible ambiguities , of reflection - based smbh spin measurements .
we will focus on results from one particular program , the _ suzaku agn spin survey _ ( sass ) , a cycle 46 _ suzaku _ key program ( pi : reynolds ) .
a posterchild for sass is the bright seyfert 1.5 galaxy ngc 3783 ( @xmath34 ) . despite the complex ( multi - zone ) photoionized absorber and the presence of significant distant reflection , the relativistic reflection signatures from the inner accretion disk could be isolated and the spin constrained to be @xmath35 ( fig . [
fig : ngc3783_spin ] ; brenneman et al . 2011
; reis et al .
2012 ; reynolds et al . 2012 ) . figure [ fig : ngc3783_res ] shows the residuals when comparing the data with the best fitting spectral model ( @xmath36 ) and a model in which we have forced a non - spinning black hole ( and , for physical consistency , @xmath37 ) but allowed all other spectral parameters to fit .
we see that the non - spinning model is over - predicting the spectrum in the 56kev band .
there is a clear way to understand this . in this spectrum , which includes hard - band data ( up to 45kev ) ,
the overall fractional amount of reflection is well determined .
the question then arises whether the observed iron line is consistent with that reflection .
consistency of the expected line strength with the data demands that the iron line is strongly broadened and hence that the black hole is rapidly spinning .
if a non - spinning black hole is imposed in the spectral fit , the iron line is too narrow " and , given that its strength is set by the overall amount of reflection , one over - predicts the 56kev flux .
the strength of the iron line also depends upon the iron abundance .
thus , from the same considerations described in the previous paragraph , we would expect the spectral fit to show a statistically correlation between the spin and the iron abundance .
our analysis of ngc 3783 finds exactly such a correlation ( fig .
[ fig : ngc3783_fe ] , left ) . in a free - fit , the spectral analysis of ngc 3783 clearly demands iron abundances that are 24@xmath38 the cosmic average , in line with the metallicity enhancement often inferred within seyfert nuclei on the basis of the optical / uv line ratios ( e.g. , warner et al .
2004 ; nagao et al .
however , if we simply impose solar iron abundances , the spectral model is a poorer fit to the data ( @xmath39 ) and we essentially lose the ability to diagnose spin ( fig . [
fig : ngc3783_fe ] , right ) .
this exercise clearly demonstrates the need to permit non - solar iron abundance in the spectral fit if one is interested in measuring spin .
we also note that , as discussed in detail in reynolds et al .
( 2012 ) , robust spin measurements in absorbed agn such as ngc 3783 absolutely require tabulated absorption models with adequate resolution in the @xmath40-dimension . while ngc 3783 is an example of an object in which we can obtain a clean and robust spin result , some objects can display spectral complexity that conspires with the medium - resolution of today s spectrometers in order to produce ambiguities in the measured spin .
these ambiguities can be particularly severe if hard x - ray data is unavailable .
we illustrate this with another of the sass objects , the luminous seyfert galaxy fairall 9 ( @xmath41 ) .
this is an example of a bare " seyfert nucleus in the sense that every pointed x - ray observation shows it to be free of intrinsic absorption , either neutral or ionized ( although _ rxte _ monitoring does show unusual x - ray dips " that may correspond to transient compton - thick absorption events ; lohfink et al .
analysis of earlier _ suzaku _ data by schmoll et al .
( 2009 ) found an obvious broad iron line with a red - wing extending down to 5kev or possibly lower and , by imposing some restrictions on the irradiation profile , was able to estimated the black hole spin to be @xmath42 .
however , the broad line in fairall 9 is weaker than in ngc 3783 or mcg6 - 30 - 15 and , while it lacks absorption , there are some additional spectral complications . due to the higher inclination of the accretion disk , the blue - peak of the broad line is blended with narrow fe25/fe26 emission lines from circumnuclear photoionized material .
there is also a smooth and variable soft excess below 2kev ( fig .
[ fig : fairall9 ] , left ) . to deal with such complexity , lohfink et al .
( 2012b ) accumulated all available _ xmm - newton _ and _ suzaku _ data ( taken over the course of several years ) and applied the technique of multi - epoch fitting in which physical arguments are used to tie certain spectral parameters across epochs , but all other spectral parameters are allowed to fit freely to the spectrum from each epoch .
specifically , they tied the black hole spin , accretion disk inclination , iron abundance , and the strength of the narrow iron line components ( which originate from parsec scales and hence will be stable between different observations ) .
this procedure finds two acceptable solutions . in the first solution ,
the soft excess is described by an additional continuum component ( modeled as thermal comptonization from a moderately hot @xmath43kev plasma ) and the relativistic disk reflection , driven by the modestly broad iron line , implies a modestly spinning black hole ( @xmath44 , completely consistent with the previous analysis of schmoll et al .
interestingly , such a solution demands a very high iron abundance ( @xmath45 ) which would drive one to consider somewhat exotic phenomena such as radiative - levitation to enhance the photospheric iron content of the inner disk ( reynolds et al .
however , a second solution exists which is a slightly worse description of the data but still statistically acceptable overall . here
, the soft excess is described by a highly - ionized ( @xmath46 ) relativistic reflection component from the inner accretion disk . in order to model the modest breadth of the broad iron line ,
the same disk must have low ionization patches that become more dominant at larger radii .
this solution has an iron abundance that is slightly sub - solar , but the need to produce a very smooth soft excess from a structured reflection spectrum drives the fit to a rapidly - spinning black hole ( @xmath6 ) .
these two solutions predict very different behavior above 20kev ( fig .
[ fig : fairall9 ] , right ) . in principle , the pin instrument on _ suzaku _ should have distinguished these models .
unfortunately , due to pointing and aspect problems during the observation ( discussed in lohfink et al .
2012b ) , the pin data can not be normalized with respect to the xis data and , together with the low s / n of fairall 9 in the pin detector , provides no additional constraints .
this experience underscores the importance of data above 10kev even when one is attempting to extract spin from signatures below 10kev .
we look forward to _ nustar _ and _ astro - h _ studies of fairall 9 to squash these ambiguities .
lcccl object & mass ( @xmath47 ) & spin & mass / spin references + mrk335 & @xmath48 & @xmath49 & pe04/wa13 + iras 005217054 & & @xmath50 & /ta12 + tons180 & @xmath51 & @xmath52 & zw05/wa13 + fairall 9 & @xmath53 & @xmath54 & pe04/lo12 + mrk359 & @xmath55 & @xmath56 & zw05/wa13 + mrk1018 & @xmath57 & @xmath58 & be11/wa13 + 1h0419 - 577 & @xmath59 & @xmath60 & zw05/wa13 + ark120 & @xmath61 & @xmath62 & pe04/wa13 + swift j0501.9 - 3239 & & @xmath63 & /wa13 + 1h0707 - 495 & @xmath64 & @xmath65 & zw05/zo10 + mrk79 & @xmath66 & @xmath67 & pe04/ga11 + mrk110 & @xmath68 & @xmath60 & pe04/wa13 + ngc3783 & @xmath69 & @xmath70 & pe04/br11 + ngc4051 & @xmath71 & @xmath63 & pe04/pa12 + rbs1124 & & @xmath65 & /wa13 + iras132243809 & @xmath72 & @xmath73 & go12/fa13 + mcg6 - 30 - 15 & @xmath74 & @xmath32 & mc05/br06 + mrk841 & @xmath75 & @xmath76 & zw05/wa13 + swift j2127.4 + 5654 & @xmath77 & @xmath78 & ma08/mi09 + ark564 & @xmath55 & @xmath79 & zw05/wa13 + @xmath80note that patrick et al .
( 2012 ) find a spin of @xmath81 , clearly discrepant with the results of brenneman et al .
( 2011 ) report here .
however , this object has a high column density warm absorber .
the modeling of this absorber , specifically the resolution of the table - models used to characterize the absorber , appears to be the root cause of the spin discrepancy .
[ tab : spinsummary ] the past couple of years has seen an explosion in the number of published smbh spin measurements based on relativistic reflection spectra , including two major works that use the public data archives to compile samples of smbh spins ( patrick et al .
2012 ; walton et al .
drawing on the published refereed literature at the time of writing ( january 2013 ) , table [ tab : spinsummary ] compiles the 20 smbh / agn spin measurements that satisfy the following quality control " criteria : 1 .
spin measurement are based on fits of the fully ionized reflection spectrum ( not just an isolated broad iron line ) . in practice ,
all of these measurements have employed the ross & fabian ( 2005 ) model reflionx .
the iron abundance characterizing the disk reflection is allowed to be a free parameter in the spectral fit .
this is important given the abundance - spin partial degeneracy noted in section [ sec : agn_cases ] .
the inclination of the accretion disk is allowed to be a free parameter in the spectral fit and can be constrained . in cases where the inclination can not be constrained ,
the spin constraint must be driven by some aspect of the spectrum other than the broad iron line ( e.g. , the soft excess ) and one must be concerned about lack of uniqueness of the spectral model .
the irradiation index @xmath16 is allowed to be a free parameter and converges on a value @xmath82 . in cases where the best - fit implies @xmath83 ,
the x - ray reflection is dominated by the _ outer _ accretion disk and so we would not expect to be sensitive to spin .
in addition to these quality - control criteria , we have rejected two sources from the walton et al .
( 2013 ) sample ; mrk 509 has a _
suzaku_-based spin value that is not robust to the treatment of the pin normalization , and pds 456 has a spin solution that appears to be double - valued .
we have included fairall 9 , listing its intermediate spin value from lohfink et al .
( 2012b ; based on the strict implementation of the methodology discussed in section [ sec : methodology ] ) , but we note the ambiguity of this particular spin value ( see section [ sec : agn_cases ] ) . while this is far from a well - defined or complete sample of objects , we can already draw some interesting conclusions .
firstly , there are a large number of objects with rapidly spinning black holes . for the most rapid spinners ( @xmath84 ) ,
the numbers must be taken with a pinch of salt the reported spins are the formal results ( with statistical errors ) that arise from the application of our standard methodology which assumes a razor thin keplerian accretion disk . using the radiation - pressure disk theory of shakura & sunyaev ( 1973 ) , we expect the inner disks of these seyfert galaxies , most of which are between 130% of their eddington luminosities , to have thicknesses of @xmath85 .
hence , especially for sources with rather large eddington ratios , we may expect finite disk thickness effects at the isco to become relevant when the isco becomes small , i.e. for large spins .
further work is needed to fully characterize this systematic error , but preliminary considerations by reynolds & fabian ( 2008 ) suggested that finite - thickness effects could relax the limits on the true " spin value in these extreme cases to @xmath84 . and spin @xmath18 from the sample listed in table [ tab : spinsummary ] . reflecting the conventions in the primary literature , all masses are marked with @xmath86 error bars whereas spins are marked with @xmath87 error ranges .
when no error estimate is available for the mass , we have assumed an error of @xmath88 .
, scaledwidth=80.0% ] still , it seems clear that a significant number of smbhs in these agn are rapidly spinning ( @xmath89 ) , at least in the mass range @xmath90 .
this suggests that the last mass doubling of these smbhs occurred via coherent accretion and not chaotic accretion ( king & pringle 2006 ) or smbh - smbh major merger ( volonteri et al . 2005 ) .
interestingly , as shown in fig .
[ fig : smbh_spins ] there are hints that the most massive black holes in this sample ( @xmath91 ) as well as the least massive black holes ( @xmath92 ) may have more modest spins .
if these trends hold up in more rigorous analyses , they would provide direct evidence for the increased role of chaotic accretion and/or major mergers at these two extreme ends of the smbh mass spectrum .
the second noteworthy point is that all of the agn in table [ tab : spinsummary ] are radio - quiet , and many of them are hosted by late - type ( i.e. , spiral disk ) galaxies .
this appears to conclusively reject the notion that black hole spin drives the radio - quiet / radio - loud dichotomy and its association with host galaxy type ( sikora , stawarz & lasota 2007 ) .
the first black hole broad iron line was discovered by barr , white & page ( 1985 ) in the _ exosat _ spectrum of the bh - xrb cygnus x-1 .
these authors interpreted the line broadening as the effects of compton scattering of iron emission line photons in the hot disk atmosphere .
it was several years before this result was re - interpreted as broadening due to doppler / gravitational shifts in the innermost regions of the black hole accretion disk ( fabian et al .
1989 ) .
due to bright - source limitations by ccd spectrometers , relativistic reflection studies in bh - xrb lagged those of agn during the _ asca _ era .
however , this situation changed with the launch of _ chandra _ , _ xmm - newton _ at the turn of the millennium .
the first robust sign of spin in the reflection spectrum of a bh - xrb was found in the _ xmm - newton_/epic - pn data of xte j1650500 ( miller et al .
2002 ) , and quantitative constraints on spin using an early version of the methodology described in
[ sec : methodology ] were derived for the bh - xrb gx3394 ( miller et al .
2004 ) . to measure the spin of a stellar mass black hole in a bh - xrbs
, one must confront several challenges .
typically , the crucial 210kev part of the x - ray spectrum contains contributions from both the thermal ( blackbody ) disk emission and the comptonized hard x - ray tail . when applying disk - reflection methods to determine spin
, one must model the thermal disk emission and be cognizant of the fact that any mis - modeling of the continuum may affect the inferred spin . similarly
, when applying the cf method , one attempts to select spectral states that are overwhelmingly dominated by the thermal disk emission but any remaining comptonization of that emission ( beyond that already included in the disk atmosphere models ) will skew the spin results .
additional complications are caused , ironically , by the fact that these sources are so bright . in many of today s more capable x - ray ( ccd ) spectrometers , special instrumental modes are required in order to observe such sources while avoiding telemetry saturation as well as photon - pileup and other more subtle instrumental effects .
thus , unlike the case for agn , one is often in the realm where systematic uncertainties related to both the spectral models as well as the calibration of the spectral data dominate statistical ( photon ) errors ; there is the danger that one will end up with very precise but inaccurate spin measures .
these concerns are valid for both the reflection- and cf - methods .
once these challenges are overcome , however , bh - xrbs offer a tremendous reward the possibility of measuring spin using multiple techniques and thereby validate the underlying assumptions of all spin measurements .
for this reason , bh - xrbs have deservedly received significant attention .
lccl object & spin from reflection & spin from cf & references + 4u 1543475 & @xmath93 & @xmath94 & mi09/sh06 + cygnus x-1 & @xmath95 & @xmath96 & fa12/go11 + gx3394 & @xmath97 & & mi09/ + grs1915@xmath98105 & @xmath99 & @xmath96 & bl09/mc06 + gro j165540 & @xmath100 & @xmath67 & rei09/sh06 + lmc x 1 & @xmath101 & @xmath102 & st12/go09 + maxi j1836194 & @xmath103 & & rei12/ + sax j1711.63808 & @xmath104 & & mi09/ + swift j1753.50127 & @xmath105 & & rei09/ + xte j1550564 & @xmath106 & @xmath107 & mi09/st11 + xte j1650500 & @xmath108 & & mi09/ + xte j1652453 & @xmath109 & & hi11/ + xte j1752223 & @xmath110 & & rei11/ + xte j1908@xmath98094 & @xmath111 & & mi09/ + @xmath112we note that mi09 quote a spin value of @xmath113 .
however , it appears likely that the _ xmm - newton _
data on which this measurement rested were affected by photon - pileup ( j.m.miller , private communication ) .
see further discussion in section [ sec : xrb ] + @xmath114we quote the high - spin model of bl09 which employs a broad - band fit of a broken power - law primary continuum and a corresponding reflection continuum modeled with a blurred pexriv ( magdziarz & zdziarski 1995 ) .
the more self - consistent reflionx model ( ross & fabian 2005 ) could not be adequately applied to the broad - band data due to the complex continuum .
+ @xmath115result combines the work of mi09 and st11 to develop a composite error range that encompasses systematic uncertainties of the reflection model .
see discussion in section [ sec : xrb ] .
[ tab : xrbspin ] table [ tab : xrbspin ] draws together the bh - xrbs with published spins based on the relativistic reflection method . of the six objects with both reflection- and cf - based spin measurements ,
there is general concordance for four .
clear discrepancies remain for two objects , 4u 1543475 and gro j165540 , the origin of which remain unclear .
two of the systems in which there are concordant spin measurements deserve additional discussion . for cygnus x-1
, there is now general agreement between reflection - based measurements ( fabian et al .
2012 ) and cf - measurements ( gou et al . 2011 ) that the black hole is a rapid spinner , @xmath30 . again , this is the formal result and the finite disk thickness considerations discussed in section [ sec : smbh_current ] may weaken this limit to @xmath84 .
however , it must be noted that early spin measurements using the reflection method suggested a small spin ( @xmath113 ) .
it now seems apparent that these early _ xmm - newton _ data on which this measurement was based were inflicted by photon - pileup effects ( j.m.miller , private communication ) that tend to narrow line features and hence decrease inferred spins ( miller et al .
analysis of more recent _ xmm - newton _ data using modified timing mode to handle the high count rate found two solutions for spin ( duro et al .
assuming an irradiation profile described by a single power - law , the duro et al .
analysis suggested either an intermediate spin with a steep irradiation profile ( @xmath116 ) or a rapid spin with a standard irradiation profile ( @xmath117 ) .
guided by the realization that steep profiles correspond to either extreme light - bending effects or inner - disk torques , both of which require rapid spins , these authors rejected the intermediate spin solution
. finally , fabian et al . (
2012 ) examined _ suzaku _ data and showed that rapid spin is favored unambiguously if one adopts a broken power - law form for the irradiation profile , a form that is verified by a non - parametric determination of the irradiation .
xte j1550564 is another interesting case .
the cf method yields a rather broad range of allowable spins , @xmath118 at the 90% confidence level ( steiner et al .
miller et al . ( 2009 ) apply the reflection method to _ asca_/gis data for this object , using the standard reflionx ( ross & fabian 2005 ) ionized reflection model , and find a spin slightly above the upper end of the cf range , @xmath119 .
however , motivated by the fact that , for the highly ionized disks found in bh - xrbs , the iron line can be appreciably compton broadened , steiner et al .
( 2011 ) re - examined this object using a very high - density , hot , variant of the reflection model ( refbhb ; ross & fabian 2007 ) .
they find that the additional compton broadening of the iron line present in refbhb leads to a slightly lower best fit spin , @xmath120 .
the fact that the spin does not change dramatically implies that the broadening of the reflection spectrum is dominated by the disk dynamics ( doppler / gravitational ) , with compton - broadening having a noticeable but sub - dominant effect . given that the true density and vertical structure of the x - ray photosphere is still uncertain , we take these two reflection results as bracketing the possible systematic error introduced by the reflection model , giving us the range of @xmath121 that we quote in table [ tab : xrbspin ] .
we end this section with a brief discussion of the astrophysical implications .
given that it is difficult to conceive of an astrophysical scenario whereby a stellar - mass black hole in a binary system can accrete a substantial fraction of its mass , we conclude that the measured spin must be close to the birth spin of the black hole .
furthermore , irrespective of whether we use the reflection- or cf - based spin measures , all of the stellar mass black holes examined to date have spin parameters that greatly exceed estimates for the natal spin parameters of neutron stars ( @xmath122 , miller , miller & reynolds 2011 ) .
this constitutes direct evidence for differences in the core - collapse supernovae that have produced the black hole and neutron star populations .
one possibility is that core collapse generically forms rapidly - spinning compact objects but , in the case of neutron stars , there is speedy reduction in the angular momentum from either gravitational wave loses due to @xmath123-mode instabilities ( andersson et al .
1999 ) or a magnetic propeller mechanism .
we conclude this chapter with a brief discussion of an exciting frontier , the recent discovery of relativistic iron line reverberation . a key aspect of the geometry
underlying the x - ray reflection picture is that the hard x - ray source is external to the optically - thick flow i.e. , there is a finite distance between the emission site for the hard x - rays and the x - ray reflector .
it is straightforward to see that variations in the luminosity of the primary x - ray source will drive variations in the strength of the observed x - ray reflection with a time - delay related to the light travel time from the source to the disk .
the full energy - dependent transfer function that relates continuum fluctuations with changes in the broad iron line encodes a tremendous amount of information including the location of the x - ray source and the spin parameter of the black hole ( reynolds et al .
for this reason , relativistic iron line reverberation has been a focus of proposals for future high - throughput x - ray observatories such as _ loft _ ( feroci et al . 2012 ) .
while measurements of the full transfer function must await future observatories , recent analyses of _ xmm - newton _ data have already found evidence for time delays between the continuum emission and relativistic reflection .
the narrow - line seyfert-1 galaxy 1h0707495 has a strong soft excess that can be explained as soft reflection from an ionized relativistic disk .
this interpretation is bolstered by spectral structure in the soft excess that resembles a relativistically - broadened iron - l line ( fabian et al .
fabian et al .
( 2009 ) discovered that , at intermediate timescales , the soft - excess lags the 24kev continuum band by @xmath124s ( also see detailed follow - up work by zoghbi et al .
since the continuum source itself always displays a pattern of hard - lags - soft ( hard lags " ) , this pattern of soft - lags - hard ( soft lags " ) is best interpreted as reverberation delays due to a displacement of the variable x - ray source from the disk ( however , see alternative interpretation of these soft - lags by l.miller et al . 2010 , and a discussion of the problems with this alternative model by zoghbi , uttley & fabian 2011 ) . since the initial discovery of soft - lags in 1h0707495 , they have been found in a number of other agn ( zoghbi & fabian 2011 ; de marco et al . 2012 ; fabian et al . 2012b ; kara et al .
2013 ) .
very recently , zoghbi et al .
( 2012 , 2013 ) has used similar techniques to find reverberation lags in the broad iron lines of ngc 4151 , ngc 7314 and mcg5 - 23 - 16 ( some of the brightest seyfert-1 galaxies in the iron - k band ) .
a skewed broad iron line can be clearly seen when the lags are examined as a function of energy ( see ngc4151 results in fig .
[ fig : ngc4151 ] ) .
furthermore , the shape of this lag spectrum " is a function of fourier frequency , with lower - frequency variations showing a narrower line with longer time - lags .
this is exactly as expected from reverberation theory , suggesting that we are gaining our first glimpse at the structure of the transfer function .
putting this together suggests that the variable x - ray source is located above the accretion disk , possibly on the black hole spin axis , at a height of a few@xmath125 .
this positive detection of iron line reverberation with _ xmm - newton _ shows conclusively that future observatories , such as _ loft _ , will be able to make tremendous progress in studies of strong gravity using relativistic reverberation in agn .
x - ray reflection signatures from the inner accretion disk of bh - xrbs and agn give us one of the most direct probe of black hole spin . in this chapter ,
i have reviewed the assumptions that underlie this technique as well as the detailed methodology as applied to both stellar - mass black hole systems and supermassive black holes .
the key points to take away from this discussion are : 1 .
relativistic reflection spectroscopy can be applied to derive spin across the full mass range of astrophysical black holes .
a major strength of this method is that one does not need to know the black hole mass or the distance of the system in order to derive spin .
furthermore , one does not need prior knowledge of the inclination of the inner accretion disk ( it is a measured quantity along with the spin ) .
there has been an explosion of smbh spin measurements in recent years .
applying some objective quality - control criteria to results in the literature , we obtain a list of 19 agn with smbh spin measurements .
it is clear that a large fraction of smbhs are rapidly - spinning , although there are tentative hints of a more slowly spinning population at high ( @xmath0 ) and low ( @xmath1 ) mass .
stellar - mass black holes in bh - xrbs provide the invaluable opportunity to measure spin with multiple techniques , thereby allowing us to validate the basic assumptions of the various methodologies . comparing relativistic reflection results to those derived from continuum fitting
, there is generally concordance ( exceptions being 4u 1543475 and gro j165540 ) .
the recent discovery of relativistic reverberation , especially of time delays between continuum fluctuations and the broad iron - k line , has opened a new frontier .
in addition to definitively confirming the broad iron line paradigm , this proves that future large - area x - ray observatories will make tremendous progress in studies of strong gravity using strong - field reverberation mapping . | i review the current status of x - ray reflection ( a.k.a .
broad iron line ) based black hole spin measurements .
this is a powerful technique that allows us to measure robust black hole spins across the mass range , from the stellar - mass black holes in x - ray binaries to the supermassive black holes in active galactic nuclei . after describing the basic assumptions of this approach
, i lay out the detailed methodology focusing on best practices " that have been found necessary to obtain robust results . reflecting my own biases ,
this review is slanted towards a discussion of supermassive black hole ( smbh ) spin in active galactic nuclei ( agn ) .
pulling together all of the available _ xmm - newton _ and _ suzaku _ results from the literature that satisfy objective quality control criteria , it is clear that a large fraction of smbhs are rapidly - spinning , although there are tentative hints of a more slowly spinning population at high ( @xmath0 ) and low ( @xmath1 ) mass .
i also engage in a brief review of the spins of stellar - mass black holes in x - ray binaries . in general ,
reflection - based and continuum - fitting based spin measures are in agreement , although there remain two objects ( gro j165540 and 4u 1543475 ) for which that is not true .
i end this review by discussing the exciting frontier of relativistic reverberation , particularly the discovery of broad iron line reverberation in _ xmm - newton _
data for the seyfert galaxies ngc 4151 , ngc 7314 and mcg5 - 23 - 16 .
as well as confirming the basic paradigm of relativistic disk reflection , this detection of reverberation demonstrates that future large - area x - ray observatories such as _ loft _ will make tremendous progress in studies of strong gravity using relativistic reverberation in agn .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
the response to external electric and magnetic fields provides a fundamental tool for studying and altering the properties of materials with numerous attendant applications .
in particular higher - order responses allow for ` manipulating light with light . '
thus , there is considerable interest in identifying molecular systems with large non - linear responses .
one approach in this direction is based on push - pull systems , i.e. , chain - like molecules with an electron - donor group at one end and an electron - acceptor group at the other ( see fig . [ fig01 ] ) . when the backbone is a @xmath0-conjugated oligomer the @xmath0 electrons of the backbone may respond easily to perturbations like those of the substituents and/or external fields . due to the donor and acceptor groups a large electron transfer , and , accordingly ,
a large dipole moment can occur and one may hope for large responses of the dipole moment to external fields . for these @xmath0-conjugated systems , each circle in fig .
[ fig01 ] could be , for example , a vinylene group , a phenylene group , a methinimine group , or combinations of those .
if the push - pull system is sufficiently large , we may split it into three parts , i.e. , a left ( l ) , a central ( c ) , and a right ( r ) part as shown in fig .
[ fig01 ] .
electrons of the central part are assumed to be so far from the terminations that they do not feel the latter ( or , more precisely , the effects of the terminations are exponentially decaying in the central part ) .
the dipole moment , @xmath1 , is useful in quantifying the response of the system to an external electric field , @xmath2 here , @xmath3 is the @xmath4th component ( i.e. , @xmath5 , @xmath6 , or @xmath7 ) of the external field with the frequency @xmath8 and @xmath9 is the frequency of the response of the molecule to the field .
the @xmath10 summations go over all the frequencies of the applied field .
@xmath11 is the dipole moment in the absence of the field which vanishes for @xmath12 .
moreover , @xmath13 is the linear polarizability , and @xmath14 , @xmath15 , @xmath16 are the first , second , @xmath16 hyperpolarizability .
sum rules require that these quantities can be non - zero only if the frequency of the response , @xmath9 , equals the sum of the frequencies ( eventually multiplied by @xmath17 ) , i.e. , for @xmath18 we require @xmath19 . in the present paper we focus on static external fields , in which case @xmath20 .
furthermore , we shall study a neutral system , although our arguments also are valid for charged systems as long as the extra charge is localized to the terminations .
we let @xmath21 be the ( field - dependent ) total charge density ( i.e. , the sum of the nuclear and electronic charge densities ) , and choose the long axis to be @xmath7 .
then the component of the total dipole moment that is of interest here , namely @xmath7 , is given by ( omitting its argument , @xmath9 ) @xmath22 where we have split the integral into contributions from the left , central , and right regions of the chain .
the central region consists of identical neutral units .
we can , therefore , write @xmath23 where @xmath24 is the number of units in c and @xmath25 is the @xmath7 component of the dipole moment of one of these units . in order to evaluate the other two contributions to the total dipole moment in eq .
( [ eqn02 ] ) we define a ` typical ' center for each term , i.e. , @xmath26 and @xmath27 ( these could , e.g. , be the center of mass of the right and left parts , respectively ) , and let @xmath28 and @xmath29 be the @xmath7 components of these vectors .
since the chain is neutral we , then , obtain @xmath30 the first term on the right hand side describes the contribution to the dipole moment associated with electron transfer from one end to the other .
this term grows linearly with chain length ( due to @xmath31 ) as does the term in eq .
( [ eqn03 ] ) . on the other hand , the last two terms in eq .
( [ eqn04 ] ) describe local dipole moments that arise from the electron distributions within the two terminal regions and they are independent of the chain length .
this discussion suggests that donor / acceptor ( = d / a ) substitution at the ends of long chains may change the charge distribution in r and l so as to strongly enhance the dipole moment and , consequently , produce a particularly large change in the dipole moment when the system is exposed to an external electric field .
therefore , very many studies have been devoted to push - pull systems as a function of increasing length ( see , e.g. , [ ] ) .
not only the electrons but also the structure ( phonons ) will respond to a static electric field .
we will demonstrate that , for sufficiently long chains , the electronic response per unit of a push - pull system ( with structural relaxation taken into account ) becomes independent of the donor and acceptor groups , implying that the materials properties can not be improved upon substitution .
our mathematical arguments for this finding are presented in the next section , and in sec .
[ sec03 ] we illustrate and analyse the results through calculations on a model system .
the particular case of inversion symmetry is discussed in sec .
[ sec04 ] where we also make a comparison with previous results .
finally , a summary is provided in sec .
[ sec05 ] .
the arguments we present are related to those originally given by vanderbilt and king - smith for an extended system in the absence of an external field .
they argued that the permanent polarization ( i.e. dipole moment per unit length ) is a bulk property.@xcite very recently , kudin _
et al._@xcite proved that the permanent polarization is quantized for d / a substituted systems .
neither of these works considered the induced polarization or the structural relaxation due to an external field .
finally , in a recent paper we presented some of the arguments behind the present work but did not analyze the predictions as we do here using a model system.@xcite
by replacing some ( groups of ) atoms with others at the chain ends , the electronic orbitals with components near the ends will change .
since the set of electronic orbitals is orthonormal , all other orbitals will change as well .
accordingly , the charge distribution may change everywhere due to the substitutions .
when an electrostatic field is applied as well , each orbital will respond to the field . since the orbitals will have changed due to the substitution ,
so will their responses to the field .
furthermore , the structural responses due to the field will also depend on the substitution at the ends .
therefore , the dipole moment can depend upon both the substitution and the field . from these arguments
there is no reason to believe that @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath16 ( with @xmath36 being the number of repeated units ) will be independent of the substitution .
however , we shall argue here that the charge @xmath37 in eq .
( [ eqn04 ] ) can change , at most , by an integral number of elementary units for different d / a substitutions at fixed external static field .
our proof is a generalization of arguments due to vanderbilt and king - smith@xcite ( see also [ ] ) , and was previously proposed by the present authors.@xcite it will be verified here by calculations on a model system and given a thorough analysis on that basis . for a given system ( with specified geometry ) , and
value of the external field , we transform the set of occupied orbitals into a set of orthonormal , localized functions .
those functions ascribed to c will be similar to the wannier functions of the infinite periodic system .
the localized orbitals will be centered in one region , but may have tails sticking into another region .
we assume that the terminal regions are large enough so that any functions centered therein , which differ from those of c , are exponentially vanishing in c. on the other hand , those functions ascribed to c , but centered on units next to l or r , will likely have tails extending into those regions .
the density matrix can then be written in block - diagonal form with three blocks , one for each of the three regions .
since the density matrix is idempotent , each block will be so , too , and there will be an integral number of electrons associated with each of the three sets of functions .
that is to say , the number of electrons associated with the functions centered in the two end regions is integral .
accordingly , any non - integral part of @xmath38 is associated with the tails of the functions in c that extend into r , which , per construction , is independent of the terminations , i.e. , also of d / a substitution .
we conclude that , for different terminations , @xmath38 can change only by an integer .
this is valid for long chains and all fields .
therefore , the electronic response per unit of the chains to the field , with or without nuclear response , is independent of termination .
the only possible change for different terminations is that @xmath38 may jump by an integer for different field strengths .
in fact , our numerical studies on a hckel - type model will confirm this prediction .
of course , in ab initio calculations , there may also be a jump due to changing the basis set or the method ( e.g. hartree - fock vs. kohn - sham dft ) .
in order to explore in detail the predictions from above we studied a hckel like model for long , finite ( ab)@xmath39 chains . in our model , we use a basis set of orthonormal atomic orbitals ( aos ) with one ao per atom .
the system has one electron per atom , and the nuclei are given charges of @xmath40 whereas the electronic charge is set equal to @xmath17 .
( all quantities are expressed in atomic units in this paper . )
given that @xmath41 is the ao of the @xmath42th atom ( @xmath43 ) and @xmath44 is the kohn - sham or fock single - electron hamiltonian we assume that only @xmath45 , @xmath46 , and @xmath47 are non - vanishing with values @xmath48\nonumber\\ \langle \chi_j\vert\hat h\vert \chi_{j+2}\rangle&= & -[t_2-\alpha_2(z_{j+2}-z_j ) ] .
\label{eqn06}\end{aligned}\ ] ] here @xmath49 is the position of the @xmath50th atom .
different donor and acceptor groups are modeled by modifying the on - site energies of the terminating atoms and/or the terminating hopping integrals , @xmath51+t_l\nonumber\\ \langle \chi_{4k+1}\vert\hat h\vert \chi_{4k+2}\rangle&= & -[t_1-\alpha_1(z_{4k+2}-z_{4k+1})]+t_r . \label{eqn07}\end{aligned}\ ] ] finally , we assume that @xmath52 in order to analyse the results we , first , define a reference structure for which the position of the @xmath42th atom is @xmath53 here @xmath54 is the length of the unit cell for an infinite , periodic system with the same electronic interactions and no external field .
subsequently , we define for each atom @xmath55 the total energy is written as the sum over occupied orbital energies ( multiplied by 2 due to spin degeneracy ) augmented by a harmonic term in the nearest- and the next - nearest - neighbour bond lengths , @xmath56 @xmath57 is the strength of the electrostatic field . for the infinite , periodic chain without an external field
, the lowest total energy corresponds to a certain lattice constant @xmath54 and @xmath58 the force constants @xmath59 and @xmath60 are determined so that @xmath54 and @xmath61 take certain chosen values . with @xmath62 being
the @xmath63th orbital ( ordered according to increasing orbital energy ) we calculate the mulliken charge on the @xmath42th atom for field @xmath57 as @xmath64 which leads to the dipole moment @xmath65 the charge transfer is given through @xmath66 we also define @xmath67 where @xmath68 is the charge for the infinite , periodic chain in the absence of the field . @xmath69
quantifies the effects on the charge distribution of the push - pull chain due to including the field , whereas @xmath70 includes effects both from the field and from the terminations .
note that @xmath71 gives the field - independent effect of the terminations .
finally , it turns out to be useful to define the center and width of the @xmath63th orbital according to @xmath72^{1/2 } , \label{eqn15}\end{aligned}\ ] ] which is consistent with eq .
( [ eqn07a ] ) .
we performed calculations for six different terminations specified by @xmath73 .
the results are summarized in figs .
[ fig02 ] , [ fig02a ] , [ fig04 ] , [ fig05 ] , and [ fig06 ] . since our model is that of a finite chain with two different types of atoms , a and b , the mulliken charges in the central region take two values .
this is clearly recognized in the presentation of @xmath74 in fig .
[ fig02 ] for @xmath75 . in fig .
[ fig02 ] it is also seen that near the ends , the mulliken charges differ from the values of the inner part and , moreover , these charges depend sensitively on the terminations . for the field strength @xmath75 these findings
are only marginally modified compared to those of a vanishing field ( not shown ) . from @xmath76
we see that the combination of electrostatic field and termination leads to an internal polarization of each unit in c. actually , @xmath76 shows a reduced internal polarization compared to @xmath77 .
thus , terminating the chain reduces the effect of the field in that regard . whereas @xmath77 contains information about the field - induced charge redistributions
, @xmath76 contains additional information about the ( field - dependent ) effects of the terminations . for @xmath75
the field - induced charge redistributions are smaller near the terminations than in the central parts . for the larger field , @xmath78 , in fig .
[ fig02a ] the identification of the central region becomes much more difficult and , as we shall see below , electrons are transferred from one end to the other .
moreover , in this case the field perturbs the system so strongly that the effects of the field are stronger than those of the terminations .
this can be seen from the fact that @xmath76 and @xmath77 are very similar .
the structure also depends upon the termination . for the intermediate field of fig .
[ fig02 ] ( and for zero field as well ) the atomic coordinate @xmath79 is nearly constant in c but varies considerably near the ends where its value depends on the termination , as was the case for the atomic charges .
for the higher field in fig .
[ fig02a ] it appears as if no central region can be identified from this parameter .
however , the fact that @xmath80 is essentially linear for the innermost atoms implies that there is a well - defined , repeating structure in c with a lattice constant differing from that of the field - free case .
[ fig04 ] shows that the charge transfer , @xmath38 , is independent of termination ( though not independent of the field ) , with the exception of jumps by ( even ) integers .
( the integers are even because we have not allowed for spin polarization . )
however , the charge distribution inside r or l does depend on the terminations and , as a consequence , the dipole moment does as well . on the other hand , the variation of @xmath81 as a function of @xmath57 for different terminations follows parallel curves , implying that the ( hyper)polarizabilities are independent of the terminations .
in fact , a least squares fit yields the values ( including maximum deviations ) : @xmath82 , @xmath83 , @xmath84 , and @xmath85 for all six terminations .
as a function of field @xmath86 is discontinuous and the power series expansion is valid only up to the field where the discontinuity occurs .
once such a discontinuity has been passed , the dipole moment depends more strongly on the field .
this means that the only way of increasing the responses of long push - pull systems to dc fields is to design chains for which the integral electron transfers occur at low fields . at a given field
the size of the chain for which jumps in the charge @xmath38 ( i.e. zener tunneling ) take place depends on the terminations ( cf .
[ fig05 ] ) . in the shortest chains ,
for which zener tunneling does not occur , @xmath86 follows parallel curves as a function of chain length , @xmath87 , for different terminations .
this means that the dipole moment and ( hyper)polarizabilities per unit become independent of termination .
however , as seen in fig .
[ fig05 ] , the slope of these curves increase after zener tunneling has taken place , implying that the dipole moment increases . assuming that the field - dependence of the dipole moment likewise increases , this suggests that the polarizability and/or hyperpolarizabilities per unit may increase for d / a substituted systems after an integral number of electrons has been transferred from one end to the other . in fig .
[ fig06 ] we show an example of what happens to the molecular orbitals when the jumps take place .
calculations were performed for field strengths between @xmath88 and @xmath89 in steps of @xmath90 , but in the figure we only show the results for fields where zener tunneling occurs . in all cases ,
the curves vary smoothly as a function of field strength . at the lowest two fields , the occupied orbitals closest to the fermi level have a center in the left part ( @xmath91 ) , whereas the unoccupied orbitals closest to the fermi level are centered in the right part . at the field @xmath92 , two electrons ( one per spin direction ) are transferred from one side to the other , which again happens at a larger field ( @xmath93 ) . in the first case , we observe the occurrence of two new , very localized , orbitals close to ( but not at ) the fermi level . the energetically lower ( i.e occupied ) one
is localized towards to the chain end on the right side while the other ( unoccupied ) is localized towards the chain end on the left side . accompanying
this interchange is a similar interchange of two rather delocalized orbitals , both of which are further away from the fermi level and centered closer to the middle of the chain .
again , at the second electron transfer a pair of new , rather localized , orbitals near ( even closer to ) the fermi level show up towards the chain ends , and also this transfer is accompanied by some reorganization of the other orbitals .
finally , fig .
[ fig06 ] also shows an example of a reorganization of the orbitals , i.e. , for a field around @xmath94 . here
, one localized , occupied orbital interchanges order with an adjacent ( in energy ) more delocalized orbital , but otherwise no further significant changes are observed .
before proceeding to compare with previous results we develop an interesting consequence of our findings with regard to inversion symmetry .
the same arguments can be applied for a system containing a mirror plane perpendicular to the chain axis , but here we shall for the sake of simplicity restrict ourselves to the case of inversion symmetry .
suppose the long oligomer of interest contains a central region made up of units with inversion symmetry .
even if the central part does not have inversion symmetry , it may be possible to create such with the addition of appropriate terminating groups .
this is , for example , the case for oligomers of thienyleneethynylenes and thienylenevinylenes that were studied by geisler _
et al._@xcite many of the systems of interest fall into one of these two categories .
since , according to our findings , d / a substitution can not affect the ( hyper)polarizabilities per unit , the latter must vanish even if the symmetry is not preserved .
for instance , modifying the terminations of the systems of geisler _ et al .
_ so that inversion symmetry no longer exists can not result in a non - vanishing @xmath34 if the chains are sufficiently long .
a large fraction of previous observations are for systems of the type described in the preceding paragraph .
some of these cases are discussed below along with others pertinent to our findings herein .
we now briefly consider , in particular , the works mentioned in the introduction . in their combined experimental and theoretical study on some push - pull oligoenes , meyers _
et al._@xcite observed a ` negligible charge transfer all the way from the donor to the acceptor ' , which implies that @xmath38 is independent of the termination .
on the other hand , in their theoretical study tsunekawa and yamaguchi@xcite examined shorter , nitrogen - containing push - pull oligomers .
they noted that these systems are interesting from the perspective of maximizing @xmath95 , but our results establish that , for such to be true , the systems must be short enough so that our approach is inapplicable .
this serves to highlight the point that apparent , but not real , discrepancies can occur due to shortness of the chain length .
marder _ et al._@xcite presented an approach for unifying the description of linear and nonlinear polarization in organic polymethine dyes .
it has since been shown that their analysis is invalid if phonons are taken into account.@xcite here , however , we emphasize that the conclusions they draw regarding @xmath95 can , again , hold only for systems that are too short for our treatment to apply . clearly , the chain length required for validity of the treatment given here is an important issue . in fig .
[ fig05 ] the dipole moment is converged for chains with some 20 units . however
, this may be an artifact of our simple hckel model . in an experimental study@xcite and in several computational studies,@xcite the second hyperpolarizability per unit
was found to converge considerably slower which , in fact , agrees with our own earlier findings.@xcite thus , when focusing on higher - order non - linear responses quite large chains may be required for the results of the present work to be relevant . in shorter push - pull systems ( for instance those considered by geisler _
et al._@xcite or by morley _
et al._@xcite ) d / a substitution can have an influence on the response . as shown numerically by champagne _
et al._,@xcite @xmath34 also converges relatively slowly as a function of size .
they considered d / a substituted oligomers of polymetheimine [ also called polycarbonitrile , ( chn)@xmath96 .
this system has a zigzag backbone of alternating c and n atoms with alternating bond lengths . without the bond length alternation it would , at least hypothetically , be possible to choose donor and acceptor groups so that the overall system is centrosymmetric .
even if chemical arguments imply that this structure is unrealistic , a non - zero value of @xmath34 for long chains should be ascribed , strictly speaking , to the bond length alternation .
polyphenylenes and polypyridines have been studied by zhang and lu.@xcite they focused on @xmath97 and @xmath98 as a function of the length of a closed ring for each system and applied a finite - field approach in their calculations .
unfortunately , as we have shown earlier ( see , e.g. , [ ] ) , this approach will never converge to the results for the infinite , periodic chain . nevertheless ,
although @xmath34 will vanish for the polyphenylenes , we predict that a non - zero value will occur for both short and long oligomers of the polypyridines .
for the d / a substituted polyenes studied by champagne _
et al._@xcite our analysis confirms their findings , i.e. , that @xmath34 will vanish for sufficiently large chains .
their numerical results indicate that @xmath34 goes through a maximum and that convergence to the infinite chain result for larger @xmath36 is slow .
even the polarizability , @xmath33 , and the permanent dipole moment , @xmath99 , may converge more slowly , as a function of chain length , than predicted by our simple model .
this is , for example , the case for the systems investigated by smith _
et al._@xcite and by kudin _
et al._@xcite in a recent study , botek _
et al._@xcite compared finite oligomers of [ @xmath36]helicenes and [ @xmath36]phenylenes that possess a helical structure for @xmath36 larger than roughly 6 . by making explicit use of the helical symmetry of the central region
we predict that , when those systems are sufficiently long , d / a substitution will not be able to modify the electronic responses to static fields .
the fact that botek _
find changes upon d / a substitution implies that the chains of their study are not converged to the long chain limit .
as long as the applied field is not so strong that an integral number of electrons is transferred from one end to the other , the answer to the question of the title is clearly : there can be no change .
this comes from our mathematical analysis in sec .
[ sec02 ] , which generalizes treatments presented previously by vanderbilt and king - smith@xcite and by kudin _ et al._,@xcite who considered only electronic polarization in the absence of an external electrostatic field .
it is also in agreement with our own earlier prediction.@xcite calculations on a model system confirm the basic result and shed light on the nature of the end - to - end charge transfer .
although the end charges , permanent dipole moment , and structure depend sensitively on the terminations neither the amount of charge transferred nor the ( hyper)polarizabilities per unit do so .
the field and/or chain length at which the charge jumps take place also depend on the terminations .
each jump is associated with an interchange of occupied and unoccupied molecular orbitals that are well - localized in the chain end region .
these orbitals are close to but not at the fermi level .
there is also an accompanying orbital reorganization .
one consequence of our finding is that long unsubstituted chains which have inversion or mirror symmetry , or can be made symmetric by substitution , must have a vanishing first hyperpolarizability per unit .
experimental and theoretical determinations are consistent with this fact , although apparent contradictions can occur for short chains .
this work was supported by the german research council ( dfg ) through project sp439/20 within the spp 1145 .
moreover , one of the authors ( ms ) is very grateful to the international center for materials research , university of california , santa barbara , for generous hospitality . | mathematical arguments are presented that give a unique answer to the question in the title .
subsequently , the mathematical analysis is extended using results of detailed model calculations that , in addition , throw further light on the consequences of the analysis .
finally , through a comparison with various recent studies , many of the latter are given a new interpretation . |
heavy fermion superconductors have been extending for more than two decades the affluence of physical phenomena associated with cooper pair formation.@xcite the competition between magnetism and superconductivity for the same fermi surface of heavily renormalized carriers resulted in observations of unconventional superconductivity@xcite@xmath8shivaramupt3 and raised speculations that spin pairing might be mediated by magnetic interaction.@xcite research in the field was associated with difficulties in sample preparation , sample to sample variation , experimental conditions and ultimately , in the number of examples where relevant physical phenomena can be observed in a clean form . the recently discovered cemin@xmath0 family ( m = ir , rh , co ) of heavy fermion superconductors encapsulates many aspects of important physics in this class of materials .
cerhin@xmath0@xcite superconducts under applied pressures above 17 kbar with t@xmath9around 2 k whereas ceirin@xmath0@xcite and cecoin@xmath0@xcite are ambient pressure superconductors .
cecoin@xmath0 offers clean example of ambient pressure heavy fermion superconductivity with a remarkably high t@xmath1=2.3 k. the intriguing properties of cecoin@xmath10 led to speculation that it may exhibit d - wave superconductivity , kohori@xmath8@xcite@xmath8@xcite and fulde - ferrel - larkin - ovchinnikov state in high magnetic fields.@xcite in order to have more insight into the nature of cecoin@xmath0 we perturbed its superconducting state by substituting la onto the ce site . for the purpose of comparing influences of magnetic and nonmagnetic pair breaking on t@xmath1 suppression
, we also substituted 5% of nd on ce site .
we find that the anisotropy in the upper critical field does not change in the whole concentration range and that the decrease of t@xmath1 with increased la doping can not be explained solely with pressure effects due to unit cell expansion .
in addition , our results present an evidence for an anisotropic order parameter in cecoin@xmath0 .
single crystals of ce@xmath4la@xmath5coin@xmath0 were grown by the self flux method in a manner previously described.@xcite crystals grew as thin plates with the c axis perpendicular to the plate .
removal of in from the surface was performed by etching in concentrated hcl for several hours followed by thorough rinsing in ethanol .
all samples obtained with this process showed no signs of in contamination .
powder x - ray patterns showed that samples crystallized in hocoga@xmath0 structure without any additional peaks introduced by la alloying .
in addition , magnetization measurements provided a more sensitive test of possible presence of magnetically ordered second phases . both as grown and etched samples showed no sign of antiferromagnetic transition of cein@xmath11 .
electrical contacts were made with epotek - h20e silver epoxy .
in - plane resistivity was measured in quantum design mpms and ppms measurement systems from 0.35 to 300 k and in fields up to 90 koe applied parallel and perpendicular to the c - axis .
there is uncertainty in nominal resistivity values associated with sample geometry and uneven surfaces of etched samples .
we measured several samples for each concentration in order to reduce measurement error which allowed us to estimate uncertainties in nominal values as well .
the dimensions of the samples were measured by high precision optical microscope with 10@xmath12 m resolution and average values are presented .
randomly chosen samples within each batch had no difference in their @xmath13 curves .
magnetization measurements were performed in mpms-7 quantum design magnetometer in the magnetic field of 10 kg , applied parallel and perpendicular to c axis .
the results of powder x - ray diffraction measurement taken at room temperature are summarized in table 1 and shown in fig . 1 , together with the unit cell volume of lacoin@xmath0 .
as expected , la doped samples have larger unit cell volume . the volume increase in the concentration range @xmath14
is consistent with expansion of the unit cell as la substitutes ce in accordance with vegard s law .
2 shows the magnetic susceptibility for ce@xmath15nd@xmath16coin@xmath0 , ce@xmath17la@xmath18coin@xmath0 , and cecoin@xmath0 , taken in the applied field of 10koe . in the whole temperature range above t@xmath1 , the substitution of magnetic ce@xmath19 by nonmagnetic la@xmath19 reduces susceptibility values in la doped sample when compared with undoped cecoin@xmath10 .
comparison of high temperature moments through curie - weiss analysis of the polycrystalline susceptibility average at high temperatures shows that approximately 14% ce ions were substituted with la . no quantitative difference from undoped cecoin@xmath0
was detected in high temperature susceptibility of 5% nd doped sample .
low temperature magnetic susceptibility of ce@xmath17la@xmath18coin@xmath0 does not reveal any difference in curie tail from pure material , thus ruling out kondo - hole interpretation of la dilution ( fig . 2
inset).@xcite we also see broadening of the plateau - like feature in @xmath20 in ce@xmath17la@xmath18coin@xmath0 ascribed@xcite to thermal depopulation of ce 4f levels . on the other hand ,
nd impurities contribute to pronounced curie tail at low temperatures .
subtraction of magnetic susceptibility of cecoin@xmath0 from ce@xmath15nd@xmath16coin@xmath0 in the normal state below 10k is consistent with approximately 8% of nd@xmath19 paramagnetic moment , result close to nominal stochiometric value and within rough approximation of our analysis .
temperature dependent electrical resistivities normalized to their value at 300 k for ce@xmath4la@xmath5coin@xmath0 and ce@xmath15nd@xmath16coin@xmath0 are presented in fig . 3a .
there are several key features to notice .
resistivities of all samples are weakly temperature dependant at high temperatures , and they pass through a maximum as temperature is decreased .
this behavior is traditionally interpreted as a crossover from incoherent kondo scattering to coherent bloch states of heavy electrons in the kondo lattice . in the case of cecoin@xmath0
this drop , at least partially , could be attributed to depopulation of crystalline electric field levels .
we observe decrease of @xmath21 for higher la concentrations ( fig .
3a inset ) . at low temperatures
, there is a clear suppression of t@xmath1 as more ce ions are replaced by la ( fig
. 3 inset ) .
the increase of the normal state residual resistivity @xmath22is probably due to disorder which contributes to increased conduction electron scattering . on the other hand ,
the resistive transition width sharpens with la alloying .
it is interesting to note that ce@xmath4la@xmath5coin@xmath0 is not in the well defined fermi liquid regime above t@xmath1 : the @xmath23 curves above t@xmath24 do not show signs of @xmath25 dependence , as it has been reported for cecu@xmath26si@xmath26.@xcite depression of t@xmath1 in cecoin@xmath10 seems to scale with @xmath22values for both magnetic and nonmagnetic dopants , as seen by comparison of the @xmath23 data of ce@xmath27nd@xmath16coin@xmath0 and ce@xmath28la@xmath29coin@xmath0 .
4 shows the anisotropic upper critical field for ce@xmath4la@xmath5coin@xmath10 , normalized to transition temperature in zero field for each value of @xmath7 ( values for @xmath30 were taken from previous report).jpsj70muramatsu3362 the h@xmath2 data were determined as a midpoint between onset and zero in resistivity from @xmath23 curves at constant field and @xmath31 curves at constant temperature .
adding la impurities results in depression of h@xmath32 , however , anisotropy @xmath3 remains at the same value of @xmath33 2 ( inset in fig .
4 ) . uncertainty in our estimate of @xmath34 decreases for higher field data , away from h=0 transition ( t / t@xmath351 ) . assuming that fermi surface properties of doped material do not change substantially in the dilute la limit,@xcite it is reasonable to assume inverse proportionality between @xmath36 and @xmath37 , and therefore values of @xmath38 could be estimated from @xmath39 for the whole doping series ( @xmath40 ) using the the value of constant a from reported @xmath41and @xmath22values for pure material.prl86movshovich5152 we obtain @xmath42 @xmath43 for cecoin@xmath0 without la impurities .
5 shows the ratio of mean free path l@xmath44 to in - plane superconducting coherence length @xmath45 @xmath46 for ce@xmath4la@xmath5coin@xmath0 obtained at t = t@xmath1/2 . in the whole doping range electronic system
is in the clean limit which could explain nearly constant value of @xmath3 .
a comparison of the effects of la substitution on t@xmath1 in cecoin@xmath0 and cecu@xmath47si@xmath26 is shown on fig .
6.@xcite doping results in depression of t@xmath1 in both cases but cecoin@xmath0 is more robust to pair breaking arising from la impurities
. the initial rate of t@xmath48 suppression is smaller than the rate seen in cecu@xmath47si@xmath26 : [ ( 0.056t@xmath1)/(1% of la substitution ) in cecoin@xmath0 vs ( 0.085t@xmath1)/(1% of la substitution in cecu@xmath47si@xmath26 ) ] .
la doping in cecoin@xmath49 is associated with only modest increase in nominal residual resistivity values @xmath39 , shown in fig .
the @xmath39 values for @xmath30 ( @xmath505@xmath51 cm ) in our experiment are in between values reported previously in literature ( 3.1@xmath51cmprl86movshovich5152 and @xmath507@xmath51cm@xcite ) .
the slope of h@xmath2 vs t curve at t@xmath1 can be used to estimate zero temperature orbital critical field @xmath52 using the weak - coupling formula for conventional superconductors in werthamer - helfand - hohenberg model ( whh ) : @xmath53.prwerthamer295 table 1 shows estimates of @xmath54 near t@xmath1 for doped samples , together with previously reported value for @xmath30 for both crystalline directions.@xcite all investigated samples have high initial slopes , as expected in the case of heavy fermion superconductors.@xcite@xmath8@xcite values of @xmath52 decrease with introduction of la impurities ( table 1 ) .
the paramagnetic limiting field @xmath55(where @xmath56 is energy gap at t=0 and @xmath57 is bohr magneton ) for pure cecoin@xmath0 ( t@xmath1=2.3k ) is well below the orbital critical field @xmath52 for either s - wave ( @xmath58),@xcite or d - wave pairing state ( @xmath59),@xcite and our results indicate that this unusual situation is valid for investigated la doping range .
we note that experimental values of upper critical field for ce@xmath4la@xmath5coin@xmath0 ( x=0 - 0.15 ) samples are most likely below the values obtained by applying whh model ( table 1 ) , probably due to polarization of magnetic sublattice due to enhanced internal field along both crystalline axis .
it has recently been reported that t@xmath1 in cecoin@xmath0 increases under applied pressure.@xcite negative chemical pressure should cause some decrease in t@xmath1 . in the lack of better approximation
, we take bulk modulus of cecoin@xmath0 to be the same as the one for cein@xmath11 ( 650kbar),@xcite and we calculate approximate chemical pressure ( p@xmath60 ) for each la concentration using @xmath61650kbar .
the results are shown in table 1 .
depression of t@xmath1 occurs at a rate @xmath62 k / kbar - a slope that is an order of magnitude larger than reported increase of t@xmath1 under hydrostatic pressure .
an order of magnitude difference from pure pressure effect on t@xmath1 is likely to exceed error in estimation of bulk modulus , and therefore points to the conclusion that the pair breaking mechanisms which enter through disorder due to la alloying and increased scattering of cooper pairs are dominant in cecoin@xmath0 .
in contrast to conventional superconductors where nonmagnetic impurities have small effect on t@xmath1 , cooper pairs formed in cecoin@xmath0 are rather sensitive to la doping : 2% of la depresses t@xmath1 to the same value as @xmath505% of nd .
the t@xmath1 suppression induced by the nonmagnetic la substitution in ce@xmath63la@xmath5coin@xmath0 is reminiscent of the pair breaking effect by magnetic impurities.@xcite although various factors may suppress t@xmath1 ( an anisotropic scattering , for example),@xcite we focus here on the scenario of cecoin@xmath0 having an anisotropic gap @xmath64 on the fermi surface .
this scenario is quite likely given the unconventional nature in many heavy - fermion materials .
here t@xmath67 is the critical temperature of the material in the absence of all scattering , @xmath68 is the scattering time by nonmagnetic impurities , and @xmath69 characterizes the gap anisotropy , @xmath70 stands for averaging over fermi surface , and @xmath71 is the digamma function . for a weak gap anisotropy ,
this result is due to hohenberg,@xcite see also later publications.pr131markowitz563@xmath8@xcite it can be shown that in fact eq .
( 1 ) holds for an arbitrary gap anisotropy.@xcite for isotropic @xmath65 , @xmath72 , and we come to anderson s theorem : t@xmath48=t@xmath67 . for pure d - wave order parameter , @xmath73 , and eq .
( 1 ) describes the d - pair breaking by nonmagnetic scattering ( which differs from the abrikosov - gorkov result only by the factor of 2 in the definition of the parameter @xmath74=@xmath75 ) . to analyze the @xmath76 data shown in fig .
6 , one has to relate @xmath6 to the scattering time @xmath68 , a nontrivial connection .
we avoid this difficulty by assuming that the residual resistivity @xmath39 is proportional to 1/@xmath77 .
further , we exclude parameter t@xmath67 from eq .
( 1 ) by writing it for two values of @xmath6 and subtracting the results:@xmath78,\mu _ { _ { 1,2}}=\beta \frac{\rho _ { _ { 1,2}}}{% t_{_{1,2}}}\]]where @xmath79 and @xmath80 is a constant to be determined . writing this equation for two different pairs @xmath81
one can determine the unknown @xmath82 and @xmath80 .
this procedure yields values scattered around @xmath83 and @xmath84 .
hence , we find @xmath85 which implies a strongly anisotropic gap . knowing the value of @xmath80 we can estimate the scattering time using measured resistivities ; for @xmath30 we obtain @xmath86 with the electronic specific heat coefficient@xcite @xmath87 we roughly estimate the fermi velocity @xmath88cm / s .
this would correspond to the mean - free path @xmath89 , a value smaller than expected but within factor of two of our determination of mean free path which is reasonable given the assumptions of average fermi velocity and isotropic scattering . in summary ,
diamagnetic pair breaking effect in cecoin@xmath0 is consistent with picture of strongly anisotropic order parameter .
anisotropy in the upper critical field @xmath90 does not change for x = ( 0 - 0.15 ) in ce@xmath4la@xmath5coin@xmath0 , indicating electronic system in the clean limit .
we thank joerg schmalian for useful discussions and hal sailsbury for help with optical microscope .
this work was carried out at ames laboratory , which is operated for the u.s .
department of energy by iowa state university under contract no .
w-7405 - 82 .
this work was supported by the director for energy research , office of basic energy sciences of the u.s .
department of energy . | we report effects of la substitution on superconducting state of heavy fermion superconductor cecoin@xmath0 , as seen in transport and magnetization measurements .
as opposed to the case of conventional superconductors , pair breaking by nonmagnetic la results in depression of t@xmath1 and indicates strong gap anisotropy .
upper critical field h@xmath2 values decrease with increased la concentration , but the critical field anisotropy , @xmath3 , does not change in the ce@xmath4la@xmath5coin@xmath0 ( @xmath6=0 - 0.15 ) .
the electronic system is in the clean limit for all values of @xmath7 . |
due to the unique properties of the neutrons , their use as an experimental probe for studying atomic ( or molecular ) vibrations and interactions is a rather well - established technique which has contributed to many advances in various scientific areas such as condensed matter physics and chemical sciences @xcite .
notable recent examples have revealed the existence of a magnetic resonance peak in iron - based superconductors @xcite , clarified the connection between bose - einstein condensation and superfluidity in the bosonic liquid @xmath7he @xcite , unravelled the existence of a roton - like signature in non - bosonic liquid @xmath16he @xcite , found a non - gaussian proton momentum distribution in confined water @xcite , .. the momentum @xmath17 and energy @xmath3 landscape that can be probed by neutrons is rather broad , and generally not accessible by a single instrument and multiple reconfigurations of the same instrument @xciteare often necessary .
this is because each neutron instrument is tailored and optimized ( by design and technical limitations ) to only probe a small cross - section of the wider @xmath18 space , making it often necessary to use multiple neutron instruments with overlapping @xmath19 windows and/or other complementary techniques before a particular phenomenon can be fully understood .
one notoriously known limiting factor for reconciling and interpreting neutron scattering data from different instruments is the fact that the energy resolution function of a given instrument depends not only on the instrument parameters but also on the imparted momentum @xmath2 and energy @xmath3 to the sample .
therefore , while qualitative interpretation of the raw ins data is generally possible ( peak positions , dispersive nature of the excitations ) , any rigorous quantitative analysis of ins data ( excitation lifetimes , bec fraction ) requires an accurate knowledge of the resolution function @xmath1 @xcite . to set the background , we note that at small momentum transfers @xmath2 ( @xmath20 @xmath21 ) , inelastic neutron scattering ( ins ) is quite effective in examining molecular re - orientation , diffusion processes and low energy excitations . in those cases , the resolution function
is usually obtained by directly measuring the exact same sample at the lowest possible temperatures where dynamical processes become frozen out on the instrument measurement time window , leaving out only the instrument contributions at or close to the elastic region .
this is particularly true for backscattering instruments where the energy resolution width remains largely constant over the accessible @xmath2 range and over a fairly broad dynamics range close to the elastic peak @xcite .
[ fig : roton_basis ] compares the observed elementary excitation ( or roton ) in superfluid @xmath7he at temperature of 1.7 k to its resolution limited response at 300 mk on the basis neutron spectrometer @xcite at the oak ridge national laboratory s spallation neutron source ( sns ) .
the low temperature measurement faithfully reproduces the asymmetric resolution due to the liquid h moderator source . on most indirect geometry neutron instruments such as basis , or direct geometry instruments in low @xmath2 mode such as the dcs @xcite at the nist center for neutron research , it may at times be sufficient to simply measure a purely incoherent standard such as vanadium in lieu of the low temperature measurements to determine @xmath1 .
unfortunately , no such straightforward practical approach exists when dealing with ins data from direct geometry instruments at high @xmath2 s . at intermediate @xmath2 s ( @xmath22 @xmath21 ) , which is well - suited for studying collective excitation modes such as phonons and magnons , and characteristic excitations such as rotons and molecular crystal fields , there exists few benchmarked software tools that account for resolution contributions in analyzing ins data collected on time - of - flight chopper spectrometers .
the tobyfit software package @xcite which uses a semi - empirical method to approximate the resolution function has been the primary workhorse for analyzing time - of - flight ins data in this regime . on the other end of the @xmath2 spectrum
, the high momentum transfer regime ( @xmath23 @xmath21 ) , which is the subject of this study , the inelastic scattering process of the individual atoms resembles closely the scattering of freely moving particles and the neutron response is characterized by recoil scattering . in these two cases , @xmath1 can either be inferred from measuring calibrated samples for which the scattering function is well known so that it can be numerically deconvoluted from the measurements , or from using ray tracing monte carlo methods with all instrument characteristics as input . in this article
, we present the resolution function of the arcs neutron spectrometer at the spallation neutron source @xcite at the oak ridge national laboratory obtained using each of these two procedures .
the main goal is to properly account for the instrument resolution contributions when analyzing ins data obtained at high @xmath2 s ( i.e. dins ) . using these
differently obtained @xmath1 , we investigate the changes in the resulting average kinetic energy @xmath24 of the atoms in normal liquid @xmath7he , and the fraction of atoms that bose condense ( bec ) in the superfluid phase .
while @xmath24 is directly proportional to the linewidth @xmath11 of the dins signal , the macroscopic number of atoms in the bec state can be inferred from the relative change in intensity in the dins response between the normal and superfluid phases .
our analysis shows consistent @xmath24 results with previous measurements at saturated vapor pressure ( svp ) @xcite and at elevated pressure near the liquid - solid transition line @xcite .
the condensate fraction @xmath13 shows however sensitivity to otherwise marginally different @xmath1 .
he , as observed on the basis neutron spectrometer at the spallation neutron source .
the resolution limited asymmetric sharp peak ( blue circles ) at low temperature is used to represent the intrinsic energy resolution function ( hwhm@xmath25 5.8 @xmath26ev ) for studying the temperature dependence of the roton lifetime at @xmath27 k ( red circles ) . ]
the wide angular range chopper spectrometer arcs @xcite is one of four direct geometry neutron instruments located at the spallation neutron source ( sns ) , oak ridge national laboratory .
this means that it uses a monochromatic beam of neutrons whose energy @xmath28 can be set by the experimenter by varying the rotation speed and phase of a spinning fermi chopper @xcite located before the sample position to probe large area of momentum - energy ( @xmath19 ) space of a subject material .
a model view of the arcs instrument layout is illustrated in fig .
[ fig : layout ] for more details .
the final energy @xmath29 of the neutrons after scattering from the sample is determined by time - of - flight techniques ( tof ) , allowing the energy transfer @xmath3 to the sample to be calculated , @xmath30 .
using kinematic constraints for a given incident energy @xmath28 , the momentum transfer @xmath2 can be conveniently expressed as a function of the energy transfer @xmath3 , and the scattering angle @xmath31 , the angle between the incident and scattered beam ( often called @xmath32 in diffraction methods ) , yielding , @xmath33 where @xmath34 2.017 mev @xmath35 . on arcs ,
this parametric relation allows large region of @xmath19 to be probed , thanks to the large @xmath31 coverage by the detector arrays , @xmath36 .
[ fig : sqe_lhe_exp ] illustrates the large @xmath19 range accessible on arcs with the incident energy @xmath37686 mev we used in our experiment .
this particular two - dimensional ( 2d ) map shows the data collected on bulk liquid @xmath7he under its own vapor pressure ( svp ) ( colored curved strip ) after proper background subtraction .
parts of the inaccessible regions due to kinematic constraints ( parabola cut - off imposed by eq.[eqn : qvsomega ] ) are marked by the sharp edges on the strip at the high @xmath2 and @xmath3 regions .
the solid black line is the ideal free @xmath7he atom recoil line , expressed as @xmath38 .
the background considered here consists of 1 ) the empty al container measured using the same instrumental configuration at 2.5 k and 2 ) an intrinsic weak broad angle - independent ( @xmath39 ) signal centered around 300 mev ( less than 1 - 2% of the overall intensity ) , suggestive of multiple scattering effects .
this broad component was approximated by averaging out the total signal at low angles ( @xmath40 , away from the @xmath7he recoil scattering ) , and extrapolating it uniformly across all angles . below we describe in details the neutron measurements and the mc simulations . in the high @xmath2-regime
also known as impulse approximation limit , the incoming neutron transfers high energy and momentum to the atoms in the sample . in this event ,
the energy transferred to the sample , @xmath3 , is large compared to the collective excitations energies in the material , with very short scattering time ( atto - seconds ) .
the validity of this approximation in neutron scattering is very well - documented @xcite .
the ia effectively treats the scattering event as single atom billiard ball scattering , in which the momentum and energy conservation rules apply to the neutron and target atom pair . as a result ,
the target atom with mass @xmath41 recoils somewhat independently of its neighbors with an energy @xmath42 ( where @xmath43 is the recoil speed ) and the observed scattering intensity is well approximated by the incoherent dynamic structure factor @xmath44 . in most liquids
, this is an excellent approximation for q@xmath45 15 @xmath21 .
the scattering function @xmath46 can be expressed in terms of the longitudinal momentum variable , @xmath47 , such that @xmath48 where @xmath49 is essentially the component of the single particle atomic momentum projected along q and @xmath50 is the longitudinal momentum distribution .
this transformation reduces the separate @xmath2 and @xmath3 dependence of the scattering into a single convenient variable @xmath47 and @xmath50 collapses into a @xmath2-independent distribution that is centered around @xmath51 .
unfortunately , the limits of the ia are not quite achieved experimentally and hence deviations known as final state effects ( fse ) must be taken into considerations when analyzing the data . the fse account for the interactions of the struck @xmath7he atom with its neighbors , as sensed by the neutrons upon scattering off the sample at these high but finite @xmath2 values @xcite . in @xmath7he ,
the fse do not change much with temperature and we use the previously determined fse found at svp @xcite to treat the current dins data .
the fse impose some residual @xmath2 dependence on the observed scattering and so what is measured can be thought of as a convolution between @xmath52 and fse such that @xmath53 .
hence careful analysis of @xmath54 over a sufficiently large range in @xmath2 can yield information on both @xmath50 and fse since j@xmath55 is @xmath2-independent and fse are not . for @xmath7he with @xmath564.002 gram / mol
, we use the factor @xmath571.0443 mev @xmath58 to convert @xmath59 to @xmath60 .
similarly , for protonated compounds
( e.g. h@xmath61o ) commonly studied in dins , it is convenient to use the conversion factor @xmath62 mev @xmath35 since @xmath63= 1 gram / mol . as a side note , it is worthwhile noting that while h@xmath61o related compounds are measurable on arcs using 2 - 3 ev incident energy , the h recoil line is significantly broader than the instrument resolution and not limited by resolution effects .
however , because of the relatively smaller mass of hydrogen compared to @xmath7he , the recoil line does nt extend high enough in @xmath2 and the reliable dins analysis tends to be limited to very small @xmath2 range . to study single proton dynamics over a wider high @xmath2-range on neutron chopper instruments such as arcs or the sister instrument sequoia@xcite
, it may be useful to use deuterated compounds ( d ) to shift the proton recoil line to lower energies and access a wider @xmath2 range . in general ,
the actual observed scattering function @xmath64 is the resolution broadened signal @xmath54 , @xmath65 eq . [ eqn : jqy_conv ] can be computed numerically if @xmath66 and @xmath14 are known . in practice however , only @xmath64 and @xmath14 are directly accessible .
this means that @xmath54 can only be obtained by inverting eq .
[ eqn : jqy_conv ] using deconvolution methods .
cumulant expansion methods@xcite are generally used to obtain analytical expressions for @xmath54 , often up to @xmath67 order depending on the need and the data quality .
as explained above , in cases where a measured @xmath14 is not available , it becomes necessary to use a simulated @xmath14 to extract @xmath50 and fse .
the intrinsic @xmath50 for bulk liquid @xmath7he is well documented with a well known temperature dependent linewidth @xmath68 @xcite . at @xmath69 @xmath21 for example ,
the intrinsic @xmath7he linewidth in energy ( @xmath70 leading gaussian term in the cumulant expansion ) is @xmath71=@xmath72=@xmath73= 18.8 mev at svp . in the present analysis
, we retain the full 6@xmath74 order expansion of @xmath50 and fse , and allow only the 1@xmath75 leading term ( @xmath11 ) and condensate component to vary .
the higher order terms in @xmath50 ( @xmath76 and @xmath77 ) are kept fixed at their previously determined values at svp and at 24 bars @xcite .
he under its own saturated vapor pressure ( svp ) showing the observed instrument broadened @xmath7he recoil signal ( centered at @xmath78 , shown as solid black line ) in the energy - momentum ( @xmath79 ) space using an incident energy of @xmath80 686 mev on the arcs neutron chopper spectrometer . ]
we used arcs in its high energy resolution configuration with @xmath82 mev to gather dins data of normal liquid state ( @xmath83 k ) at svp as a benchmark measurement . without actual superfluid data at svp from arcs , we use instead our previously collected superfluid data at elevated pressure of 24 bars @xcite to determine the condensate fraction .
the data analysis was subsequently performed over the @xmath2-range , @xmath84 @xmath21 with @xmath850.5 @xmath21 , which we found to have the best signal - to - background ratio in the present measurement . at @xmath86 @xmath21 comparing the actual measurement ( filled square symbols ) with the simulated response ( open blue diamonds ) and the different resolution functions @xmath87 .
the blue solid line ( @xmath88 ) represents the simulated resolution function and the black solid line ( @xmath89 ) is the one derived from experiment . the red solid line is the resolution convoluted fit , using @xmath89 .
similar fit is obtained with @xmath88 . ]
[ fig : sqe_cuts ] shows a representative one - dimensional ( 1d ) spectra cut taken around @xmath86 @xmath21 as a function of the longitudinal momentum variable @xmath90 .
this figure compares the net experimental scattering intensity of liquid @xmath7he with the simulated response using the mcvine ray tracing monte carlo , described in ref.@xcite to which we return below .
the red line is the best model fit ( using eq .
[ eqn : jqy_conv ] ) with @xmath91 .
a very similar fit can be also obtained with @xmath88 , but with slightly different fit parameters . to truly appreciate the difference in the quality of the fits and for better comparison
, we undertook a detailed @xmath92 analysis as a function of @xmath2 , and could not again satisfactorily differentiate between the fitting agreement factors @xmath92 , other than to say that both @xmath93 yield consistent fits at most @xmath2 values , as indicated in table [ tbl : chisquared ] for superfluid @xmath7he at 24 bars . .fitting agreement factor @xmath94 ^ 2}{n_p - m}$ ] for superfluid
@xmath7he data at 24 bars versus selected @xmath2 for different resolution function @xmath93 , where the label @xmath4 refers to the extraction method used to determine @xmath14 .
the parameter @xmath95 represents the number of data points and @xmath41 the number of fitting parameters in @xmath96 .
[ cols="^,^,^,^,^,^,^",options="header " , ] [ tbl : sum ] finally , we observe a bose - einstein condensate fraction @xmath13 at 24 bars that is indeed smaller than the reported bulk value of 7.25% at @xmath97 k @xcite , independently of the resolution function used to analyze the data .
the observed @xmath13 values are shown as a function of @xmath2 in fig .
[ fig : n0_hp ] .
while the @xmath13 values derived from @xmath88 tend to be generally smaller than those obtained with @xmath89 in magnitude , the effective @xmath98averaged values are consistent with each other and with previous work @xcite , as summarized in table [ tbl : sum ] .
the relative sensitivity of @xmath13 to resolution effects highlights the importance of reliably and accurately determining the full @xmath87 to better analyze dins data . in the present study
, either@xmath14 provides reliable characteristic parameters for the momentum distribution of liquid @xmath7he which agree with theoretical predictions@xcite , and previous measurements @xcite . of liquid @xmath7he at temperature @xmath83 k and pressure @xmath99 bars .
the elevated pressure leads to an increase in density , which in turn leads to larger kinetic energy and @xmath11 values compared to the bulk liquid values .
the lines represent the @xmath2-averaged values .
the red solid line shows the average bulk liquid @xmath7he value observed on mari at 24 bars @xcite . ]
-dependence of the bose condensate fraction @xmath13 of liquid helium at elevated pressure of @xmath99 bars and temperature @xmath100 k , based on the experimentally extracted ( black solid squares and blue open diamonds ) and calculated ( red open circles ) resolution functions .
the lines represent the corresponding @xmath2-averaged values .
the variation of @xmath13 with @xmath2 observed with @xmath89 are also observed with @xmath101 .
the dashed lines represent the average values . ]
in this comparative study , we have presented different experimental and simulation methods for determining the full resolution function of a direct - geometry time - of - flight spectrometer for the purpose of analyzing dins data .
independently of the methods used , the observed resolution broadened dynamical scattering patterns @xmath102 are consistently similar for liquid @xmath7he .
consequently , the extracted signal width @xmath11 values are consistent with each other and agree well with previous work .
the average bec values ( @xmath13 ) at 24 bars are more sensitive to changes in the resolution lineshape than @xmath11 , varying from for example from @xmath252.85% with @xmath101 to @xmath253.9% with @xmath103 within acceptable uncertainty .
the simulations offer the most direct route for analyzing future dins data at arcs .
while the methods presented are for analyzing dins data for the arcs spectrometer with liquid @xmath7he as the sample , they can be effectively applied to other systems such as h@xmath61 and h@xmath61o measured either at arcs or other neutron spectrometers such as sequoia .
we wish to thank r. senesi , t. prisk , e. mamontov , h. bordallo , f.x .
gallmeier , e. iverson , g.e .
granroth , b. fultz and h. glyde for many valuable stimulating discussions .
we acknowledge the use of the mantid software package @xcite to reduce the neutron data and the nist dave fitting software @xcite for the data analysis .
this work is sponsored by the scientific user facilities division , office of basic energy sciences , u.s .
department of energy . c. andreani , d. colognesi , j. mayers , g. f. reiter , measurement of momentum distribution of light atoms and molecules in condensed matter systems using inelastic neutron scattering , advances in physics 54 ( 2005 ) 377 .
a. d. christianson , e. a. goremychkin , r. osborn , s. rosenkranz , m. d. lumsden , c. d. malliakas , i. s. todorov , h. claus , d. y. chung , m. g. kanatzidis , r. i. bewley , t. guidi , unconventional superconductivity in ba0.6k0.4fe2as2 from inelastic neutron scattering , nature 456 ( 2008 ) 930932 . s. o. diallo , r. t. azuah , d. l. abernathy , j. taniguchi , m. suzuki , j. bossy , n. mulders , h. r. glyde , evidence for a common physical origin of the landau and bec theories of superfluidity , physical review letters 113 ( 215302 ) ( 2014 ) 15 . h. godfrin , m. meschke , h .- j .
lauter , a. sultan , h. m. bhm , e. krotscheck , m. panholzer , observation of a roton collective mode in a two - dimensional fermi liquid , nature 483 ( 2012 ) 576579 .
v. garbuio , c. andreani , s. imberti , a. pietropaolo , g. f. reiter , r. senesi , m. a. ricci , proton quantum coherence observed in water confined in silica nanopores , journal of chemical physics 127 ( 2007 ) 154501 .
a. pietropaolo , c. andreani , a. filabozzi , r. senesi , g. gorini , e. perelli - clippo , m. tardocchi , n. j. rhodes , e. m. schooneveld , dins measurements on vesuvio in the resonance detector configuration : proton mean kinetic energy in water , journal of instrumentation 1 ( 2006 ) p04001 .
p. p. deen , a. vickery , k. h. andersen , a design study of vor : a versatile optimal resolution chopper spectrometer for the ess , european journal of physics : web of conferences 83 ( 2015 ) 03002 .
k. h. andersen , resolution function of the neutron time - of - flight spectrometer mari , nuclear instruments and methods in physics research , section a : accelerators , spectrometers , detectors and associated equipment 371 ( 3 ) ( 1996 ) 472479 .
a. seeger , l. l. daemen , the neutron instrument simulation package , nisp , proc .
spie 5536 , advances in computational methods for x - ray and neutron optics 5536 ( 2004 ) 109123 .
v. anette , l. udby , n. violini , j. voigt , p. p. deen , k. leffman , a monte carlo simulation of neutron instrument resolution functions , journal of the physical society of japan 82 ( sa037 ) ( 2013 ) 19 .
n. tsapatsaris , p. k. willendrup , r. e. lechner , h. n. bordallo , http://www.epj-conferences.org/10.1051/epjconf/20158303015[from basis to miracles : benchmarking and perspectives for high - resolution neutron spectroscopy at the ess ] , epj web of conferences 83 ( 2015 ) 03015 .
e. mamontov , k. w. herwig , a time - of - flight backscattering spectrometer at the spallation neutron source , basis , review of scientific instruments 82 ( 8) ( 2011 ) 010 .
j. r. d. copley , j. c. cook , the disk chopper spectrometer at nist : a new instrument for quasielastic neutron scattering studies , chemical physics 292 ( 2 - 3 ) ( 2003 ) 477485 . t. g. perring , http://tobyfit.isis.rl.ac.uk/[tobyfit : fitting of resolution broadened models to data from time - of - flight chopper spectrometers ] .
d. l. abernathy , m. b. stone , m. j. loguillo , m. s. lucas , o. delaire , x. tang , j. y. y. lin , b. fultz , design and operation of the wide angular - range chopper spectrometer arcs at the spallation neutron source , review of scientific instruments 83 ( 2012 ) 015114 .
h. r. glyde , r. t. azuah , w. g. stirling , condensate , momentum distribution , and final - state effects in liquid 4 he , physical review b 62 ( 21 ) ( 2000 ) 337349 .
s. o. diallo , r. t. azuah , d. l. abernathy , r. rota , j. boronat , h. r. glyde , http://link.aps.org/doi/10.1103/physrevb.85.140505[bose-einstein condensation in liquid 4he near the liquid - solid transition line ] , physical review b 85 ( 14 ) ( 2012 ) 140505 . m. b. stone , j. l. niedziela , d. l. abernathy , l. debeer - schmitt , g. ehlers , o. garlea , g. e. granroth , m. graves - brook , a. i. kolesnikov , a. podlesnyak , b. winn , http://dx.doi.org/10.1063/1.4870050[a comparison of four direct geometry time - of - flight spectrometers at the spallation neutron source ] , review of scientific instruments 85 ( 2014 ) 045113 . j. voigt , n. violini , t. brckel , http://dx.doi.org/10.1016/j.nima.2013.12.036[chopper layout for spectrometers at long pulse neutron sources ] , nuclear instruments and methods in physics research , section a : accelerators , spectrometers , detectors and associated equipment 741 ( 2014 ) 2632 .
h. r. glyde , s. o. diallo , r. t. azuah , o. kirichek , j. w. taylor , atomic momentum distribution and bose - einstein condensation in liquid 4he under pressure , physical review b - condensed matter and materials physics 84 ( 18 ) ( 2011 ) 114 . h. r. glyde , s. o. diallo , r. t. azuah , o. kirichek , j. w. taylor , bose - einstein condensation in liquid he4 under pressure , physical review b - condensed matter and materials physics 83 ( 10 ) ( 2011 ) 47 .
r. senesi , a. i. kolesnikov , c. andreani , measurement of proton momentum distributions using a direct geometry instrument , journal of physics : conference series 571 ( 2014 ) 012007 .
g. f. reiter , j. mayers , j. noreland , momentum - distribution spectroscopy using deep inelastic neutron scattering , physical review b 65 ( 10 ) ( 2002 ) 104305
. h. glyde , excitations in liquid and solid helium , oxford university press , oxford , 1994 .
g. i. watson , http://arxiv.org/abs/cond-mat/9606061[neutron compton scattering ] , j phys condens matter 8 ( 33 ) ( 1996 ) 22 .
r. senesi , http://dx.doi.org/10.1016/j.nima.2011.09.039[direct kinetic energy extraction from neutron compton profiles ] , nuclear instruments and methods in physics research , section a : accelerators , spectrometers , detectors and associated equipment 661 ( 1 ) ( 2012 ) 7076 . j. y. y. lin , a. a. aczel , d. l. abernathy , s. e. nagler , w. j. l. buyers , g. e. granroth , using monte carlo ray tracing simulations to model the quantum harmonic oscillator modes observed in uranium nitride , physical review b 89 ( 14 ) ( 2014 ) 144302 .
j. y. y. lin , h. l. smith , g. e. granroth , d. l. abernathy , m. d. lumsden , b. winn , a. a. aczel , m. aivazis , b. fultz , mcvine - an object oriented monte carlo neutron ray tracing simulation package , nuclear instruments and methods in physics research , section a : accelerators , spectrometers , detectors and associated equipment 810 ( 2016 ) 8699 .
y. yiu , a. a. aczel , g. granroth , d. l. abernathy , s. m.b . , w. j. l. buyers , j. y. y. lin , g. d. samolyuk , g. m. stocks , s. e. nagler , light atom quantum oscillations in uc and us , physical review b 93 ( 2015 ) 014306 .
o. arnold , j. c. bilheux , j. m. borreguero , a. buts , s. i. campbell , l. chapon , m. doucet , n. draper , r. ferraz leal , m. a. gigg , v. e. lynch , a. markvardsen , d. j. mikkelson , r. l. mikkelson , r. miller , k. palmen , p. parker , g. passos , t. g. perring , p. f. peterson , s. ren , m. a. reuter , a. t. savici , j. w. taylor , r. j. taylor , r. tolchenov , w. zhou , j. zikovsky , mantid - data analysis and visualization package for neutron scattering and @xmath26 sr experiments , nuclear instruments and methods in physics research , section a : accelerators , spectrometers , detectors and associated equipment 764 ( 2014 ) 156166 .
loong , s. ikeda , j. m. carpenter , the resolution function of a pulsed - source neutron chopper spectrometer , nuclear instruments and methods in physics research , section a : accelerators , spectrometers , detectors and associated equipment 260 ( 1987 ) 381402 .
d. abernathy , j. niedziela , m. stone , extracting source parameters from beam monitors on a chopper spectrometer , epj web of conferences 83 ( 2015 ) 03001 .
http://mcstas.org/download/components/contrib/sns$\backslash${_}source$\backslash${_}analytic.html[website ] .
http://mcstas.org/download/components/contrib/sns$\backslash${_}source$\backslash${_}analytic.html[http://mcstas.org/download/components/contrib/sns$\backslash$\{_}source$\backslash$\{_}analytic.html ]
p. willendrup , e. farhi , k. lefmann , mcstas 1.7 - a new version of the flexible monte carlo neutron scattering package , physica b : condensed matter 350 ( 1 - 3 suppl .
1 ) ( 2004 ) 735737 .
http://mcstas.org/download/components/optics/fermichopper.html[website ] .
http://mcstas.org/download/components/optics/fermichopper.html s. ikeda , j. m. carpenter , wide - energy - range , high - resolution measurements of neutron pulse shapes of polyethylene moderators , nuclear instruments and methods in physics research section a : accelerators , spectrometers , detectors and associated equipment 239 ( 3 ) ( 1985 ) 536544 .
d. m. ceperley , e. l. pollock , path - integral computation of the low - temperature properties of liquid he4 , physical review letters 56 ( 4 ) ( 1986 ) 351354 .
d. m. ceperley , path integrals in the theory of condensed helium , reviews of modern physics 67 ( 2 ) ( 1995 ) 279355 .
s. moroni , m. boninsegni , condensate fraction in liquid 4he at zero temperature , journal of low temperature physics 136 ( 2004 ) 129137 .
p. e. sokol , bose condensation in liquid helium , neutron news 4 ( 3 ) ( 1993 ) 2226 .
r. t. azuah , l. r. kneller , y. qiu , p. l. w. tregenna - piggott , c. m. brown , j. r. d. copley , r. m. dimeo , dave : a comprehensive software suite for the reduction , visualization , and analysis of low energy neutron spectroscopic data , journal of research of the national institute of standards and technology 114 ( 6 ) ( 2009 ) 341 . | inelastic neutron scattering at high momentum transfers ( i.e. @xmath0 ) , commonly known as deep inelastic neutron scattering ( dins ) , provides direct observation of the momentum distribution of light atoms , making it a powerful probe for studying single - particle motions in liquids and solids . the quantitative analysis of dins data requires an accurate knowledge of the instrument resolution function @xmath1 at each momentum @xmath2 and energy transfer @xmath3 , where the label @xmath4 indicates whether the resolution was experimentally observed @xmath5 or simulated @xmath6 . here
, we describe two independent methods for determining the total resolution function @xmath1 of the arcs neutron instrument at the spallation neutron source , oak ridge national laboratory .
the first method uses experimental data from an archetypical system ( liquid @xmath7he ) studied with dins , which are then numerically deconvoluted using its previously determined intrinsic scattering function to yield @xmath8 .
the second approach uses accurate monte carlo simulations of the arcs spectrometer , which account for all instrument contributions , coupled to a representative scattering kernel to reproduce the experimentally observed response @xmath9 . using a delta function as scattering kernel , the simulation yields a resolution function @xmath10 with comparable lineshape and features as @xmath8 , but somewhat narrower due to the ideal nature of the model . using each of these two @xmath1 separately
, we extract characteristic parameters of liquid @xmath7he such as the intrinsic linewidth @xmath11 ( which sets the atomic kinetic energy @xmath12 ) in the normal liquid and the bose - einstein condensate parameter @xmath13 in the superfluid phase .
the extracted @xmath11 values agree well with previous measurements at saturated vapor pressure ( svp ) as well as at elevated pressure ( 24 bars ) within experimental precision , independently of which @xmath14 is used to analyze the data .
the actual observed @xmath13 values at each @xmath2 vary little with the model @xmath15 , and the effective @xmath2-averaged @xmath13 values are consistent with each other , and with previously reported values .
neutron chopper spectrometer , instrument resolution , monte - carlo simulations , inelastic neutron scattering |
let @xmath1 denote a random walk on @xmath0 , started at a nonrandom point @xmath2 , and denote its range by @xmath3 .
our goal is to give an answer to the following question of barlow and taylor ( * ? ? ?
* problem , p. 145 ) : @xmath4 here , @xmath5 denotes the macroscopic hausdorff dimension of a set @xmath6 , as was defined in @xcite .
we will recall the formal definition and first properties of @xmath7 in [ subsec : dim ] below . in informal terms , @xmath5 measures the `` local dimension of @xmath8 at infinity , '' and describes the large - scale geometry of @xmath8 in a manner that is similar to the way that the ordinary , microscopic , hausdorff dimension of @xmath8 describes the small - scale geometry of @xmath8 . in the case that @xmath9 is a recurrent random walk on @xmath0
, one can answer easily .
recall that a point @xmath10 is _ possible _ for @xmath9 if @xmath11 for some @xmath12 , where @xmath13 denotes the conditional law of @xmath9 given that @xmath14 , as usual .
the collection of all possible points of @xmath9 is an additive subgroup of @xmath0 , whence homomorphic to @xmath15 for some integer @xmath16 , thanks to the structure theory of finitely - generated abelian groups .
therefore , the strong markov property implies that @xmath17 is homomorphic to @xmath15 a.s . ,
and hence we have @xmath18 a.s . @xcite .
in other words , has non - trivial content if and only if @xmath9 is transient .
this explains why barlow and taylor @xcite pose only for transient random walks .
as far as we know , the only positive result about is due to barlow and taylor themselves ( * ? ? ?
7.9 ) . in order to describe their result ,
let @xmath19 denote the green function of @xmath9 .
that is , @xmath20 of course , @xmath21 as well , and the transience of @xmath9 is equivalent to the finiteness of the function @xmath19 on all of @xmath22 .
question was in part motivated by the following positive result .
[ pr : bt ] let @xmath23 .
suppose there exist constants @xmath24 $ ] and @xmath25 such that @xmath26 whenever @xmath27 .
then , @xmath28 a principal goal of this article is to answer question in general .
we do this by following a suggestion of barlow and taylor and introducing a random - walk `` index '' that is equal to @xmath29 .
moreover , as is tacitly implied in @xcite , that index is defined solely in terms of the statistical properties of @xmath9 . that index turns out to be related to the notion of a `` recurrent set '' for @xmath9 .
recall that a set @xmath30 is said to be _ recurrent _ for @xmath9 , under @xmath13 , if the random set @xmath31 is unbounded @xmath13-a.s .
this definition makes sense regardless of whether @xmath32 is random or not . because @xmath9 is transient ,
a necessary condition for the recurrence of @xmath32 is that @xmath32 is unbounded , though this condition is not sufficient , as the following example shows : let @xmath9 denote the simple symmetric walk on @xmath33 , and define @xmath34 , where @xmath35 . by the local central limit theorem , @xmath36 as @xmath37 , hence @xmath38=\sum_{k=0}^\infty p^0\{x_k\in\mathcal{r}_x\}<\infty,\ ] ]
thanks to tonelli s theorem . therefore , @xmath32 is not recurrent for @xmath9 , though it is unbounded .
by @xmath39 we mean the expectation operator for @xmath40 a necessary - and - sufficient condition for the recurrence of a _ nonrandom _ set @xmath32 was found first by it and mckean @xcite in the case that @xmath9 is the simple random walk on @xmath0 for some @xmath41 .
lamperti @xcite discovered a necessary - and - sufficient condition in the case that @xmath9 belongs to a large family of transient random walks on @xmath0 , and in fact many transient markov chains on countable state spaces .
when @xmath9 is a general transient random walk on @xmath0 , or more generally a markov chain on a countable state space , there is also an exact condition for set recurrence , but that condition is more involved ; see bucy @xcite and , more recently , benjamini et al @xcite .
all latter works involve various notions of abstract capacity that are borrowed from probabilistic potential theory .
our answer to is also stated in terms of a sort of abstract capacity condition ; see corollary [ cor : main ] .
one can prove that when the underlying random walk @xmath9 is sufficiently regular for example as in the statement of proposition [ pr : bt]our condition can simplify .
see corollary [ cor : rw : dim ] for instance . still , our index is very difficult to work with in general ; see [ sec : conj ] for a discussion of this shortcoming , and for potential remedies .
we conclude the introduction with an outline of the paper . in
[ sec : backgd ] we include some of the technical prerequisites to reading this paper .
then , in [ sec : fp ] we develop a macroscopic theory of `` fractal percolation '' that is the large - scale analogue of the microscopic theory of fractal percolation @xcite .
our macroscopic extension of the microscopic theory is not entirely trivial , but will ring familiar to many experts . in
[ sec : forest ] we write a forest representation of @xmath0 and use it together with the theory of two - parameter processes @xcite in order to characterize exactly when @xmath17 intersects a piece of a macroscopic fractal percolation set .
this is the truly novel part of the present article , and is likely to have other uses particularly in computing the ordinary and/or large - scale dimension of complex random sets . finally , in
[ sec : mc ] we adapt an elegant replica method of peres @xcite to the present setting in order to compute the dimension of @xmath42 for every recurrent nonrandom set @xmath32 . in the section that follows this discussion [ [ sec : dim_m ] ]
we derive the following simpler and more elegant representation for the macroscopic minkowski dimension of an arbitrary random walk on @xmath0 : @xmath43 in the last section
[ sec : conj ] we state a few remaining open problems and related conjectures .
this section introduces the prerequisite material , necessary for later use . throughout
we follow the original notation of barlow and taylor @xcite by setting @xmath44 for all @xmath45 . for every integer @xmath46 define @xmath47 to be the collection of all dyadic cubes @xmath48 of the form @xmath49 where @xmath50 are integers .
if a cube @xmath48 has the form , then we say that @xmath51 is the _ south - west corner _ of @xmath48 , and @xmath52 is the _ sidelength _ of @xmath48 .
by @xmath53 we mean the collection of all dyadic hypercubes of @xmath0 ; that is , @xmath54 a special role is played by @xmath55 this is the collection of all dyadic cubes of sidelength no smaller than @xmath56 . for every @xmath57 and @xmath58
define @xmath59 where the minimum is taken over all possible coverings of @xmath60 by cubes @xmath61 of sidelength @xmath62 and south - west corner @xmath63 .
all these cubes are elements of @xmath64 .
the _ macroscopic hausdorff dimension _
@xcite of @xmath65 is defined by in place of our @xmath66 , and @xmath67 in place of our @xmath7 .
we prefer @xmath66 as it reminds us that our @xmath66 is the large - scale analogue of besicovitch s @xmath68-dimensional net measures .
and we use for @xmath7 in favor of @xmath67 to distinguish between large - scale and ordinary hausdorff dimension . ]
it is easy to see that if @xmath70 , then @xmath71 the second seemingly - natural inequality was one of the motivations of the theory of @xcite , and does not hold for some of the previously - defined candidates of large - scale dimension in the literature @xcite .
let @xmath9 denote the simple symmetric random walk on @xmath0 where @xmath41 . according to the local central limit theorem ,
@xmath72 as @xmath73
. therefore , proposition [ pr : bt ] applies and implies the very appealing fact that the macroscopic hausdorff dimension of the range of @xmath9 is a.s .
@xmath74 .
barlow and taylor @xcite have proved that macroscopic hausdorff dimension of the range of transient brownian motion is also a.s . 2 , giving further credance to their assertion that @xmath7 is a natural large - scale variation of the classical notion of [ microscopic ] hausdorff dimension , nowadays usually denoted by @xmath67 .
we follow the existing works on related potential theory @xcite , and consider a somewhat more general setting in which our random walk @xmath9 is replaced by a transient markov chain , still denoted by @xmath9 . however , by contrast with the prior works we continue to assume that our markov chain @xmath9 takes values in the special state space @xmath0 for some @xmath75 , and not a general countable state space .
we will continue to write @xmath17 to stand for the range of the markov chain @xmath9 .
that is , @xmath17 is the random set @xmath76 .
let @xmath13 continue to denote the conditional law of @xmath9 , given @xmath14 .
then we say that a random or nonrandom set @xmath30 is _ recurrent _ for @xmath9 under @xmath13 when @xmath42 is unbounded @xmath13-a.s .
we are aware of at least two characterizations of recurrent nonrandom sets for general chains , due to bucy @xcite and benjamini et al @xcite . in order to describe the second characterization , which turns out to be more relevant to our needs ,
let @xmath77 denote the collection of all probability measures on @xmath32 , and @xmath78 the _ martin capacity _ of @xmath32 for the walk started at @xmath79 @xcite .
that is , @xmath80^{-1},\ ] ] where @xmath81 for all @xmath82 .
benjamini et al @xcite have characterized all recurrent sets for transient markov chains on countable state spaces .
if we apply their result to transient markov chains on @xmath0 , then we obtain the following : choose and fix some @xmath79 .
a nonrandom set @xmath30 is recurrent for @xmath9 under @xmath13 if and only if @xmath83 , where the infimum is taken over all cofinite subsets @xmath8 of @xmath32 .
recall that a set @xmath84 is said to be _ cofinite _ when @xmath85 is a bounded set .
the preceding capacity condition for cofinite sets is not so easy to verify in concrete settings .
there is an older result , due to lamperti @xcite , which contains a more easily - applicable characterization of recurrent sets for `` nice '' markov chains .
the following is a slightly different formulation that works specifically for transient chains on @xmath0 .
barlow and taylor ( * ? ? ?
* proposition 8.2 ) state a special case of it by adapting lamperti s method [ see example [ ex : rw : nice ] below ] .
we derive it later on as a corollary of a `` master theorem '' on hitting probabilities of transient chains on @xmath0 [ theorem [ th : rw : fp ] ] .
[ cor : lamperti ] suppose that there exist @xmath79 and a finite constant @xmath86 such that for all @xmath87 and @xmath88 , @xmath89 then , @xmath32 is recurrent for @xmath9 under @xmath13 if and only if @xmath90 .
[ ex : rw : nice ] suppose that @xmath9 is a random walk that satisfies condition of proposition [ pr : bt ] .
it readily follows that @xmath91 for all @xmath92 , and for all @xmath93 '' to mean that there exists a positive and finite constant @xmath94 such that @xmath95 for all @xmath93 . ]
@xmath96 simultaneously for all integers @xmath45 and @xmath97 .
therefore , @xmath98 for all @xmath45 , where @xmath99^{-1}\ ] ] describes the usual random - walk capacity of @xmath84 . ''
is the same capacity form as lamperti s `` @xmath94 '' @xcite and barlow and taylor s `` @xmath100 '' @xcite .
this fact can be found implicitly in bucy @xcite , and might be older still . ]
it is easy to see that the lamperti - type condition also holds in this case since the walk has i.i.d
. increments .
therefore , corollary [ cor : lamperti ] tells us that @xmath32 is recurrent for @xmath9 under @xmath101 iff @xmath102 .
this is proposition 8.2 of @xcite .
we temporarily leave the topic of markov chains and random walks in order to present some basic facts about macroscopic fractal percolation .
let @xmath45 denote a fixed integer , and suppose @xmath103 is a collection of independent random variables , defined on a rich enough probability space @xmath104 , such that each @xmath105 is distributed uniformly between @xmath106 and @xmath107 .
we may define , for all @xmath108 $ ] , @xmath109 then : 1 .
@xmath110 are i.i.d . ; 2 .
@xmath111 and @xmath112 ; and 3 .
@xmath113 if @xmath114 .
for all integers @xmath115 define @xmath116 we also abuse notation to define @xmath117 . in this way , we see for example that @xmath118 is a random set of cubes with sidelength @xmath119 which result from the _ first step _ of a certain branching process ; see figure [ fig : treerep ] .
_ fractal percolation _ on @xmath120 [ with parameter @xmath108 $ ] ] is the random set @xmath121 one can see that this is the usual construction of mandelbrot s fractal percolation @xcite , scaled to take place in the cube @xmath120 .
namely , we may write @xmath120 as a disjoint union of @xmath122 elements of @xmath123 ; each of those elements is selected independently with probability @xmath124 and rejected with probability @xmath125 .
we then write every one of the selected cubes as a disjoint union of @xmath122 elements of @xmath126 ; each resulting subcube is kept / selected or discarded / deselected independently with respective probabilities @xmath124 and @xmath125 ; and we continue .
elementary branching - process theory implies that @xmath127 if and only if @xmath128 ; see also figures [ fig : cubeperc0 ] , [ fig : sim0.8 ] , and [ fig : treerep ] . presently , we are interested in performing fractal percolation in @xmath120 but we will stop the subdivisions after @xmath129 steps . in other words , we are interested in @xmath130 , which is a random , possibly empty , collection of side - one cubes in @xmath131 .
since @xmath113 whenever @xmath114 , we see that @xmath132 a.s . , and hence if @xmath114 , then @xmath133 for all borel sets @xmath134 .
now let us construct all of these fractal percolations on the same probability space so that : 1 .
@xmath130 is a fractal percolation in @xmath120 for every @xmath45 as decsribed earlier , 2 .
@xmath135 are independent . in words , we appeal to the preceding procedure in order to construct the @xmath130 s simultaneously for all @xmath136 , using an independent collection of weights @xmath137 s for every @xmath120 .
by _ macroscopic fractal percolation _ we mean the random set @xmath138 ( 0,0)rectangle(4,4 ) ; ( 0,2)rectangle(2,4 ) ; ( 0,0)rectangle(4,4 ) ; ( 0,2)(4,2 ) ; ( 2,0)(2,4 ) ; ( 0,0)rectangle(4,4 ) ; ( 0,2)rectangle(2,4 ) ; ( 0,1)rectangle(2,2 ) ; ( 2,0)rectangle(3,1 ) ; ( 2,2)rectangle(3,3 ) ; ( 3,3)rectangle(4,4 ) ; ( 0,0)rectangle(4,4 ) ; ( 0,2)(4,2 ) ; ( 2,0)(2,4 ) ; ( 0,1)(4,1 ) ; ( 3,0)(3,4 ) ; ( 2,3)(4,3 ) ; ( 1,0)(1,2 ) ; ( 0,0)rectangle(4,4 ) ; ( 0,2)rectangle(2,4 ) ; ( 0,1)rectangle(2,2 ) ; ( 2,0)rectangle(3,1 ) ; ( 2,2)rectangle(3,3 ) ; ( 3,3)rectangle(4,4 ) ; ( 0,0.5)(0,1)(1,1)(1,0)(0.5,0)(0.5,0.5)(0,0.5 ) ; ( 1,0)rectangle(2,.5 ) ; ( 3,0)(3,.5)(3.5,.5)(3.5,1)(4,1)(4,0)(3,0 ) ; ( 2,1)(2,2)(2.5,2)(2.5,1.5)(3.5,1.5)(3.5,1)(2,1 ) ; ( 3,2)rectangle(3.5,3 ) ; ( 2.5,3.5)rectangle(3,4 ) ; ( 0,0)rectangle(4,4 ) ; ( 0,2)(4,2 ) ; ( 2,0)(2,4 ) ; ( 0,1)(4,1 ) ; ( 3,0)(3,4 ) ; ( 2,3)(4,3 ) ; ( 1,0)(1,2 ) ; ( 0,0.5)(2,.5 ) ( 3,.5)(4,.5 ) ; ( 2,1.5)(4,1.5 ) ; ( 3,2.5)(4,2.5 ) ; ( 2,3.5)(3,3.5 ) ; ( 0.5,0)(.5,1 ) ; ( 1.5,0)(1.5,1 ) ; ( 2.5,1)(2.5,2 ) ( 2.5 , 3)(2.5,4 ) ; ( 3.5,0)(3.5,3 ) ; ( 0,0)rectangle(4,4 ) ; ( 0,2)rectangle(2,4 ) ; ( 0,1)rectangle(2,2 ) ; ( 2,0)rectangle(3,1 ) ; ( 2,2)rectangle(3,3 ) ; ( 3,3)rectangle(4,4 ) ; ( 0,0.5)(0,1)(1,1)(1,0)(0.5,0)(0.5,0.5)(0,0.5 ) ; ( 1,0)rectangle(2,.5 ) ; ( 3,0)(3,.5)(3.5,.5)(3.5,1)(4,1)(4,0)(3,0 ) ; ( 2,1)(2,2)(2.5,2)(2.5,1.5)(3.5,1.5)(3.5,1)(2,1 ) ; ( 3,2)rectangle(3.5,3 ) ; ( 2.5,3.5)rectangle(3,4 ) ; ( 0.25,0.25)rectangle(0.5,0.5 ) ; ( 1.25,0.75)rectangle(1.75,1 ) ; ( 3.25,0.75)rectangle(3.5,0.5 ) ; ( 2.75,2)rectangle(3,1.5 ) ; ( 3.5,1.5)rectangle(3.75,1.75 ) ; ( 3.5,2.5)rectangle(4,3 ) ; ( 2.5,3.75)(2.25,3.75)(2.25,3.25)(2.75 , 3.25)(2.75,3.5)(2.5,3.5)(2.5,3.75 ) ; ( 0,0)rectangle(4,4 ) ; ( 0,2)(4,2 ) ; ( 2,0)(2,4 ) ; ( 0,1)(4,1 ) ; ( 3,0)(3,4 ) ; ( 2,3)(4,3 ) ; ( 1,0)(1,2 ) ; ( 0,0.5)(2,.5 ) ( 3,.5)(4,.5 ) ; ( 2,1.5)(4,1.5 ) ; ( 3,2.5)(4,2.5 ) ; ( 2,3.5)(3,3.5 ) ; ( 0.5,0)(.5,1 ) ; ( 1.5,0)(1.5,1 ) ; ( 2.5,1)(2.5,2 ) ( 2.5 , 3)(2.5,4 ) ; ( 3.5,0)(3.5,3 ) ; ( 3.75,2.5)(3.75,3 ) ; ( 3.5,2.75)(4,2.75 ) ; ( 2.75,1.75)(3,1.75 ) ; for @xmath139 .
shells are independent among shells .
the nested squares delineate the cubical shells sets @xmath140 and @xmath141 . with the exception of @xmath142 all other shells @xmath143 are cubical annuli .
percolation in different cubical annuli results from independent branching processes.,title="fig:",width=245,height=245 ] ( 0,0)rectangle(1,1 ) ; ( 2,3)rectangle(3,4 ) ; ( -2,-2)rectangle(-1,-1 ) ; ( -2,-3)rectangle(-1,-2 ) ; ( 0,-4)rectangle(2,-2 ) ; ( 2,-2)rectangle(3,0 ) ; ( 3,2)rectangle(4,3 ) ; ( -4,1)rectangle(1,3 ) ; ( -3,3)rectangle(-2,4 ) ; ( -4,-2)rectangle(-3,-1 ) ; ( 0,0)rectangle(1,1 ) ; ( 0,0)rectangle(1,1 ) ; ( 0,0)rectangle(1,1 ) ; in -4, ... ,4 ( , -4)(,4 ) ; in -4, ... ,4 ( -4,)(4 , ) ; ( -4,-4)rectangle(4,4 ) ; ( -2,-2)rectangle(2,2 ) ; ( -1,-1)rectangle(1,1 ) ; ( -1,1)(1,1)(1,-1 ) ; ( -2,2)(2,2)(2,-2 ) ; ( -4,4)(4,4)(4,-4 ) ; ( 5,0)(0,0)(0,5 ) ; in -4, ... ,3 in -4, ...
,3 ( , ) circle ( 1.5 mm ) ; ( 0.5,0.5)node[color = sussexp]*a * ; ( 0.5,1.5)node[color = sussexp]*b * ; ( -0.5,1.5)node[color = sussexp]*c * ; ( -1.5,1.5)node[color = sussexp]*d * ; ( -1.5,-1.5)node[color = sussexp]*e * ; ( 3.5,2.5)node[color = sussexp]*g * ; ( 2.5,3.5)node[color = sussexp]*f * ; ( 0.5,2.5)node[color = sussexp]*h * ; ( -1.5,2.5)node[color = sussexp]*i * ; ( -0.5,2.5)node[color = sussexp]*j * ; ( -2.5,3.5)node[color = sussexp]*k * ; ( -3.5,2.5)node[color = sussexp]*l * ; ( -2.5,2.5)node[color = sussexp]*m * ; ( -2.5,1.5)node[color = sussexp]*n * ; ( -3.5,1.5)node[color = sussexp]*o * ; ( -3.5,-1.5)node[color = sussexp]*p * ; ( -1.5,-2.5)node[color = sussexp]*q * ; ( 1.5,-2.5)node[color = sussexp]*t * ; ( 0.5,-2.5)node[color = sussexp]*u * ; ( 0.5,-3.5)node[color = sussexp]*v * ; ( 1.5,-3.5)node[color = sussexp]*w * ; ( 2.5,-0.5)node[color = sussexp]*r * ; ( 2.5,-1.5)node[color = sussexp]*s * ; of course , @xmath144 and @xmath145 .
starting from here , we often assume tacitly that @xmath146 in order to avoid the trivial cases @xmath147 and @xmath148 . in any case , it is easy to deduce our next result .
[ lm : basic ] the following are valid : 1 .
@xmath149 whenever @xmath114 ; 2 .
@xmath150 are independent random sets ; 3 .
@xmath151 is distributed as @xmath152 for every integer @xmath45 .
if @xmath65 and @xmath153 are subsets of @xmath154 , then we say that @xmath65 is a _ recurrent set _ for @xmath153 provided that there exist infinitely many shells @xmath155 such that @xmath156 for all @xmath12 .
we sometimes write `` @xmath157 '' in place of `` @xmath65 is a recurrent set for @xmath153 '' ; clearly , @xmath158 is an equivalence relation between pairs of subsets of @xmath154 .
[ lem:0 - 1 ] if @xmath134 is non random , then @xmath159 or @xmath107 .
let @xmath160 if @xmath161 and @xmath162 otherwise .
then , the @xmath163 s are independent and @xmath164 by kolmogorov s @xmath106-@xmath107 law . a word of notational caution at this point .
we use @xmath165 as shorthand for @xmath166 , since @xmath167 is a random set . however , when we condition on a given configuration from this a.s
. event , @xmath165 reverts to its original meaning .
as we have noted already , the [ microscopic ] fractal percolation set @xmath168 is nonvoid if and only if @xmath128 . the large - scale analogue of becoming nonvoid is to become unbounded .
the following shows that the large - scale result takes a different form than its small - scale counterpart in the critical case @xmath169 .
[ lem : perc : unbdd ] @xmath167 is almost surely unbounded if @xmath170 and it is almost surely bounded if @xmath171 . [ lem:20 ] we will be interested only in @xmath167 when it is unbounded ; the preceding tells us that we want to consider only values of @xmath172 $ ] .
the said condition on @xmath124 will appear several times in the sequel for this very reason . by the borel
cantelli lemma for independent events , @xmath167 is unbounded a.s .
if @xmath173 ; otherwise , @xmath167 is bounded a.s . since @xmath174 , the probability that @xmath151 is nonempty is at most the probability that , in its first @xmath136 generations , a galton
watson branching process with mean branch rate @xmath175 survives .
we shall denote by @xmath176 the number of descendants in the @xmath136-th generation . when @xmath171 , the said galton
watson process is strictly subcritical .
it is well - known that @xmath177 which forms a summable sequence in @xmath136 .
the simple bound @xmath178 ensures that @xmath179 thus , we conclude that @xmath167 is a.s
. bounded when @xmath171 . by the monotonicity of @xmath180
, it remains to prove that @xmath181 is a.s .
one can define fractal percolation on any dyadic cube @xmath182 in much the same way as we defined it on @xmath120 .
next we note that if @xmath183 , then @xmath184 where @xmath185 denotes the event that fractal percolation with @xmath169 on a dyadic cube of side @xmath186 does not become void in @xmath136 steps .
therefore , @xmath187 is bounded below by the probability that a certain critical galton
watson does not become extinct in its first @xmath136 generations .
a well - known theorem of kolmogorov @xcite asserts that for critical galton
watson branching process , @xmath188 where @xmath189 denotes the variance of the offspring distribution ; see also kesten et al @xcite and lyons et al @xcite .
this implies that there exists @xmath190 such that @xmath191 and completes the proof .
the following is the result of an elementary computation .
[ lem : perc : hit ] if @xmath192 for some @xmath193 , then @xmath194 if @xmath195 then
@xmath196 , where @xmath197 according to , @xmath198 when @xmath199 .
both bounds can be achieved : @xmath200 when the most recent common ancestor of @xmath201 and @xmath202 in the branching process is the root @xmath120 ; and @xmath203 when @xmath204 .
we will use the preceding in order to prove the following .
[ th : dim : pi ] for all non random sets @xmath134 , @xmath205:\ , f\operatorname{\sim_{{\rm ( r)}}}\pi_p\right\}.\ ] ] [ rem : dim : pi ] theorem [ th : dim : pi ] can be recast in the following equivalent terms : if @xmath206 then @xmath165 ; otherwise if @xmath207 , then @xmath208 .
the case that @xmath209 is not decided by a dimension criterion . that case can be decided by a more delicate capacity criterion .
however , we will not have need for this refinement , and so will not pursue it here .
define , as in barlow and taylor @xcite , @xmath210 , where @xmath211 denotes the closest point in @xmath0 to @xmath212 , with some procedure in place for breaking ties .
we may observe that @xmath213 is nonempty if and only if @xmath214 is nonempty .
lemma 6.1 of barlow and taylor @xcite ensures that @xmath215 .
therefore , we may assume without loss of generality that @xmath32 is a subset of @xmath0 ; otherwise , we can replace @xmath32 by @xmath216 everywhere throughout the remainder of the proof . from now
on we consider only @xmath30 .
let @xmath45 be an arbitrary integer .
we can find dyadic cubes @xmath217 such that : 1 .
the sidelength @xmath218 @xmath219 of every @xmath61 is at least one ; 2 . @xmath220 ; and 3 .
@xmath221 .
thus , we obtain @xmath222 for the inequality we have used the fact that @xmath223 for every integer @xmath224 . in any case , it follows that @xmath225 the preceding holds for all @xmath108 $ ] .
suppose for the moment that @xmath226 .
then , @xmath227 by the definition of @xmath7 ; and implies that @xmath228 the non recurrence of @xmath32 for @xmath167 follows by the borel cantelli lemma .
we have proved that if @xmath229 , then @xmath32 is not recurrent for @xmath167 .
it now remains to show that @xmath230 from now on we choose and fix an arbitrary @xmath231 .
note that @xmath232 thanks to the definition of @xmath7 . according to theorem 4.2 of barlow and taylor @xcite ,
we can find a sigma - finite measure @xmath233 on @xmath154 and a positive and finite constant @xmath234 , that depends only on the ambient dimension @xmath235 , such that @xmath236 for all integers @xmath45 and all dyadic cubes @xmath237 with sidelength @xmath238 .
[ barlow and taylor @xcite define a measure on @xmath143 , that they refer to as @xmath239 .
the restriction of our @xmath233 to @xmath143 is their measure @xmath239 . ]
normalize @xmath240 in order to conclude the following for all integers @xmath45 : 1 .
@xmath241 ; 2 .
@xmath242 , uniformly for all dyadic cubes @xmath237 with sidelength @xmath238 .
[ this is valid even when @xmath243 , provided that we define @xmath244 . ]
define @xmath245 where @xmath246 . by lemma [ lem : perc : hit ] : ( i ) @xmath247 = \mu({\mathcal{s}}_k ) = 1 $ ] ; and ( ii ) @xmath248 & = \sum_{x , y\in{\mathcal{s}}_k } p^{-{{\rm d}}(x , y)}\,\mu(x)\,\mu(y)\\ & = \sum_{j=0}^{k+1 } p^{-j}\,(\mu\times\mu)\left\ { ( x\,,y)\in{\mathcal{s}}_k^2:\ { { \rm d}}(x\,,y)=j\right\}\\ & { \leqslant}\sum_{j=0}^{k+1 } p^{-j}\,(\mu\times\mu)\left\ { ( x\,,y)\in{\mathcal{s}}_k^2:\ { { \rm d}}(x\,,y){\geqslant}j\right\}. \end{split}\ ] ] for a given @xmath201 , define @xmath249 . because @xmath250 , @xmath251 since @xmath252 , we can find a dyadic cube @xmath253 such that @xmath254 .
therefore , property ( ii ) of the measure @xmath239 ensures that @xmath255 and hence @xmath248 & { \leqslant}\frac{c}{\mathcal{n}_\delta(f\,,{\mathcal{s}}_k)}\cdot \sum_{j=0}^k \big(2^{\delta}p\big)^{-j}{\leqslant}\frac{c\left(1- ( 2^\delta p)^{-1 } \right)^{-1}}{\mathcal{n}_\delta(f\,,{\mathcal{s}}_k)}. \end{split}\ ] ] for the last inequality we have used the hypothesis of ( [ eq : goal : lb ] ) .
the paley - zygmund inequality yields @xmath256)^2/{\mathrm{e}}[\zeta_k^2]$ ] .
therefore , our bounds for the two first moments of @xmath163 lead us to @xmath257 this and together imply that @xmath258 . the independence half of the borel
cantelli lemma implies .
let us close this section by presenting a quick application of theorem [ th : dim : pi ] .
[ co : dim : pi ] for each @xmath108 $ ] , @xmath259 a.s .
this result has content only when @xmath172 $ ] ; see lemma [ lem:20 ] .
in particular , it states that the dimension of fractal percolation is @xmath106 at criticality .
the proof uses the replica argument of peres @xcite without need for essential changes .
more specifically , let @xmath260 denote a fractal percolation set with parameter @xmath261 $ ] such that @xmath167 and @xmath260 are independent .
because @xmath262 has the same distribution as @xmath263 , it follows that @xmath264 thus , by lemma [ lem : perc : unbdd ] , we have @xmath265 we may first condition on @xmath167 and then appeal to theorem [ th : dim : pi ] in order to see that @xmath266 where @xmath267 is the critical constant @xmath261 $ ] such that @xmath268 . because @xmath269 the corollary follows .
we can now deduce barlow and taylor s dimension theorem [ proposition [ pr : bt ] ] from the previous results of this paper .
the following is a standard codimension argument ; see taylor ( * ? ? ? * theorem 4 ) .
barlow and taylor ( * ? ? ?
8.4 ) have observed that , under the conditions of proposition [ pr : bt ] , @xmath270 if @xmath271 and @xmath272 if @xmath273 .
this is an immediate consequence of proposition 8.2 of @xcite , which is a variation of lamperti s test @xcite ; the general form of this form of lamperti s test is in fact corollary [ cor : lamperti ] , which we will prove in due time .
we apply the preceding observation conditionally , with @xmath274 , where @xmath275 $ ] is a fixed parameter . by corollary
[ co : dim : pi ] , @xmath276 a.s .. therefore , @xmath277 if @xmath278 and @xmath279 if @xmath280 . by the hewitt savage 0 - 1 law
, @xmath29 is @xmath13-a.s . a constant .
choose @xmath281 and assume to the contrary that @xmath282 restrict the probability space of the random walk to the full-@xmath13 probability event @xmath283 and fix a realization @xmath284 .
by theorem [ th : dim : pi ] we conclude that @xmath285 for almost all realizations of the random walk .
this gives the desired contradiction since for @xmath286-a.e .
realization of @xmath167 , @xmath277 , while @xmath287 it follows that @xmath288 a.s .
as long as @xmath278 .
an analogous argument shows that @xmath289 whenever @xmath280 .
therefore , we conclude that @xmath290 @xmath13-a.s . where @xmath291 .
if @xmath292 for some integer @xmath45 , then there exists a unique sequence @xmath293 of dyadic sets such that : 1 .
@xmath294 and @xmath295 ; 2 .
@xmath296 for all @xmath297 ; and 3 .
@xmath298 for all @xmath299 .
conversely , if @xmath300 denotes a collection of dyadic cubes such that : 1 .
@xmath301 ; 2 .
@xmath302 for all @xmath297 ; and 3 .
@xmath303 for all @xmath299 ; then there exists a unique point @xmath292 defined unambiguously via @xmath304 [ equivalently , @xmath305 for @xmath306 ( see figure [ fig : treerep ] ) .
moreover , @xmath307 for all @xmath308 .
the preceding describes a bijection between the points in @xmath309 and a certain collection of @xmath310-chains of dyadic cubes .
we can now use this bijection in order to build a directed - tree representation of @xmath309 : the vertices of the tree are comprised of all dyadic cubes @xmath182 whose sidelength is @xmath311 .
for the [ directed ] edges of our tree , we draw an arrow from a vertex @xmath312 to a vertex @xmath313 if and only if there exists an integer @xmath297 and a point @xmath10 such that @xmath314 and @xmath315 . the resulting graph is denoted by @xmath316 .
it is easy to observe the following properties of @xmath316 : 1 .
@xmath316 is a finite rooted tree , the root of @xmath316 being @xmath120 ; 2 .
every ray in @xmath316 has depth @xmath129 ; 3 .
there is a canonical bijection from the rays of @xmath316 to @xmath309 .
( 10,4)(6 , 5)(2 , 4 ) ; ( 3,3.75)(7,4.75 ) ; ( 5,3.25)(9,4.25 ) ; ( 3,4.25)(7,3.25 ) ; ( 5,4.75)(9,3.75 ) ; ( 4,3.5)(8,4.5 ) ; ( 4,4.5)(8,3.5 ) ; ( 2,3)(10,5)node[below left]@xmath202 ; ( 2,5)(10,3)node[below left]@xmath201 ; ( 4,7)(3,4 ) ; ( 4,7)(3.75 , 3.75 ) ; ( 4,7)(4.75,4 ) ; ( 4,7)(4 , 4.25 ) ; ( 5.7,6.5)(4.75,3.5 ) ; ( 5.7,6.5)(5.75 , 3.25 ) ; ( 5.7,6.5)(6 , 3.75 ) ; ( 5.7,6.5)(6.75 , 3.5 ) ; ( 8,7)(7,4 ) ; ( 8,7)(7.75,3.75 ) ; ( 8,7)(8,4.25 ) ; ( 8,7)(9,4 ) ; ( 2,7 )
( 10,7)(6 , 8)(2 , 7 ) ; ( 4,6.5)(8,7.5 ) ; ( 4,7.5)(8,6.5 ) ; ( 6,10)(4,7 ) ; ( 6,10)(5.7,6.5 ) ; ( 6,10)(6,7.5 ) ; ( 6,10)(8,7 ) ; ( 2,10 ) ( 6,9 ) ( 10,10)node[right](6 , 11)(2 , 10 ) ; ( -0.5,1)(-2,0)node[below , color = black]a ; in -2, ... ,1 ( -0.5,1)(,0 ) ; ( , 0 ) circle(1 mm ) ; ( -.5,1 ) circle(1mm)node[above]@xmath317 ; ( 4,1)(4.75,0)node[below ] ? ; ( 7,2)(4,1)(4.25,0)node[below , color = black]b ; ( 7,2)(6,1)(6.75,0)node[below , color = black ] ? ; ( 7,2)(6,1)(5.25,0)node[below , color = black]c ; ( 7,2)(6,1)(5.75,0)node[below , color = black]d ; ( 7,2)(8,1)(7.25,0)node[below , color = black ] ? ;
( 7,2)(8,1)(8.25,0)node[below , color = black]e ; ( 7,2)(10,1)(9.75,0)node[below , color = black ] ? ; ( 4,1 ) ( 3.25,0 ) ; ( 4,1 ) ( 3.75,0 ) ; ( 4,1)(4.25,0 ) ; ( 4,1 ) ( 4.75,0 ) ; ( 3.25,0)circle(1 mm ) ; ( 3.75,0)circle(1 mm ) ; ( 4.25,0)circle(1 mm ) ; ( 4.75,0)circle(1 mm ) ; ( 6,1 ) ( 5.25,0 ) ; ( 6,1 ) ( 5.75,0 ) ; ( 6,1)(6.25,0 ) ; ( 6,1 ) ( 6.75,0 ) ; ( 5.25,0)circle(1 mm ) ; ( 5.75,0)circle(1 mm ) ; ( 6.25,0)circle(1 mm ) ; ( 6.75,0)circle(1 mm ) ; ( 8,1 ) ( 7.25,0 ) ; ( 8,1 ) ( 7.75,0 ) ; ( 8,1)(8.25,0 ) ; ( 8,1 ) ( 8.75,0 ) ; ( 7.25,0)circle(1 mm ) ; ( 7.75,0)circle(1 mm ) ; ( 8.25,0)circle(1 mm ) ; ( 8.75,0)circle(1 mm ) ; ( 10,1 ) ( 9.25,0 ) ; ( 10,1 ) ( 9.75,0 ) ; ( 10,1)(10.25,0 ) ; ( 10,1 )
( 10.75,0 ) ; ( 9.25,0)circle(1 mm ) ; ( 9.75,0)circle(1 mm ) ; ( 10.25,0)circle(1 mm ) ; ( 10.75,0)circle(1 mm ) ; in 2, ... ,5 ( 7,2)(2*,1 ) ; ( 2*,1 ) circle(1 mm ) ; ( 7,2 ) circle(1 mm ) node[above]@xmath318 ; ( 10,3)(4,2)(2.5,1)(2.5 + -1 * 0.12,0)node[below , color = black]f ; ( 10,3)(4,2)(2.5,1)(2.5 + 3 * 0.12,0)node[below , color = black]g ; ( 10,3)(4,2)(3.5,1)(3.5 + 1 * 0.12,0)node[below , color = black]h ; ( 10,3)(8,2)(6.5,1)(6.5 + 1 * 0.12,0)node[below , color = black]i ; ( 10,3)(8,2)(6.5,1)(6.5 + 3 * 0.12,0)node[below , color = black]j ; ( 10,3)(8,2)(7.5,1)(7.5 - 1 * 0.12,0)node[below , color = black]k ; ( 10,3)(8,2)(7.5,1)(7.5 + 1 * 0.12,0)node[below , color = black]l ; ( 10,3)(8,2)(7.5,1)(7.5 + 3 * 0.12,0)node[below , color = black]m ; ( 10,3)(8,2)(8.5,1)(8.5 - 3 * 0.12,0)node[below , color = black]n ; ( 10,3)(8,2)(8.5,1)(8.5 - 1 * 0.12,0)node[below , color = black]o ; ( 10,3)(12,2)(11.5,1)(11.5 + 1 * 0.12,0)node[below , color = black]p ; ( 10,3)(12,2)(13.5,1)(13.5 - 1 * 0.12,0)node[below , color = black]q ; ( 10,3)(16,2)(14.5,1)(14.5 - 1 * 0.12,0)node[below , color = black]r ; ( 10,3)(16,2)(14.5,1)(14.5 + 1 * 0.12,0)node[below , color = black]s ; ( 10,3)(16,2)(16.5,1)(16.5 - 3 * 0.12,0)node[below , color = black]t ; ( 10,3)(16,2)(16.5,1)(16.5 - 1 * 0.12,0)node[below , color = black]u ; ( 10,3)(16,2)(16.5,1)(16.5 + 1 * 0.12,0)node[below , color = black]v ; ( 10,3)(16,2)(16.5,1)(16.5 + 3 * 0.12,0)node[below , color = black]w ; ( 4.5 -3 * 0.14,-0.1 ) ( 4.5 + 3 * 0.14,-0.1 ) ; ( 4.5,-0.1)node[below ] ? ; ( 9.5 -3 * 0.14,-0.1 ) ( 9.5 + 3 * 0.14,-0.1 ) ; ( 9.5,-0.1)node[below ] ? ; ( 10.5 -3 * 0.14,-0.1 ) ( 10.5 + 3 * 0.14,-0.1 ) ; ( 10.5,-0.1)node[below ] ? ; ( 15.5 -3 * 0.14,-0.1 ) ( 15.5 + 3 * 0.14,-0.1 ) ; ( 15.5,-0.1)node[below ] ? ; in 4,8,12,16 ( 10,3)(,2 ) ; in -3,-1,1,3 ( , 2 ) ( + .5 * , 1 ) ; in -3,-1,1,3 ( + .5*,1)(+.5*+.12*,0 ) ; ( + .5*+.12*,0 ) circle(0.8 mm ) ; ( + .5*,1 ) circle(1 mm ) ; ( , 2 ) circle(1 mm ) ; ( 10,3)circle(1mm)node[above]@xmath319 ; [ fig : forest ] because the directed tree @xmath316 is finite for every @xmath45 , we can isometrically embed @xmath316 in @xmath320 such that the vertices of @xmath316 that have maximal depth lie on the real axis .
of course , there are infinitely - many such possible isometric embeddings ; we will choose and fix one [ it will not matter which one ] .
in this way , we can think of every @xmath316 as a finite rooted tree , drawn in @xmath320 , whose vertices of maximal depth lie on the real axis and whose root lies @xmath129 units above the real axis . because every @xmath321 has been coded by the rays of @xmath316 , and since those rays can in turn be coded by their last vertex [ these are vertices of maximal depth ]
, thus we obtain a bijection @xmath322 that maps each point @xmath321 to a certain point @xmath323 on the real axis of @xmath320 .
note that , in this way , @xmath324 can be identified with a finite collection of points on the real line .
the collection @xmath325 is a forest representation of @xmath0 .
we use this representation in order to impose an order relation @xmath326 on @xmath0 as follows : 1 .
if @xmath321 and @xmath327 for two integers @xmath328 such that @xmath329 , then we declare @xmath330 ; 2 .
if @xmath331 for the same integer @xmath45 , and @xmath332 , then we declare @xmath330 .
it can be checked , using only first principles , that @xmath326 is in fact a bona fide total order on @xmath0 .
we might sometimes also write @xmath333 in place of @xmath330 . if we identify @xmath331 with 2 rays of the tree @xmath316 , viewed as a tree drawn in @xmath154 as was described earlier , then @xmath330 iff the ray for @xmath201 lies on , or to the left of , the ray for @xmath202 . and
if @xmath192 and @xmath334 for @xmath335 , then our definition of @xmath330 stems from the fact that we would like to draw the tree @xmath316 to the left of the tree @xmath336 , as we embed the forest @xmath337 , tree by tree , isometrically in @xmath320 .
now we return to markov chains . throughout this section , let @xmath1 denote a transient markov chain on @xmath0 .
this chain is constructed in the usual way : we have a probability space @xmath338 together with a family @xmath339 of probability measures such that under @xmath13 , the markov chain begins at @xmath14 for every @xmath79 .
by @xmath340 we mean the corresponding expectation operator for @xmath13 for all @xmath79 [ @xmath341 .
we assume that the markov chain is independent of the fractal percolations .
the two processes are jointly constructed on @xmath342 .
we write @xmath343 and @xmath344 the corresponding expectation operator [ @xmath345 .
as is well known , the transience of @xmath9 is equivalent to seemingly - simpler condition that @xmath346 for all @xmath10 .
this is because @xmath347 for all @xmath10 ; in fact , one can apply the strong markov property to the first hitting time of @xmath10 in order to see that @xmath348 we define an equivalence relation on @xmath0 as follows : for all @xmath349 we write `` @xmath350 '' when there exists an integer @xmath45 such that @xmath201 and @xmath202 are both in the same shell @xmath143 .
symbol @xmath351 denotes that @xmath201 and @xmath202 are in different shells .
if @xmath239 is a probability measure on @xmath0 , then we define two `` energy forms '' for @xmath239 .
the first form is defined , for every fixed @xmath79 , as @xmath352 where @xmath353 as before .
our second definition of energy requires an additional parameter @xmath108 $ ] , and is defined as follows : @xmath354 note , in particular , that @xmath355 coincides with the _ martin energy of @xmath239 _ @xcite .
finally we define a quantity that can be thought of as a kind of graded martin capacity associated to @xmath9 : for any set @xmath32 , @xmath356 and @xmath357 , define the martin @xmath124-capacity by @xmath358^{-1}.\ ] ] the set function @xmath359 is the same martin capacity that appeared earlier in .
it might help to observe that the martin @xmath124-capacity satisfies the following monotonicity property : @xmath360 the main result of this section can be stated as follows .
[ th : rw : fp ] if @xmath30 is nonrandom , then for all @xmath79 , @xmath361 where @xmath362 is defined by .
theorem [ th : rw : fp ] implies the following .
[ cor : main ] if @xmath30 is recurrent for @xmath9 @xmath13-a.s . , and @xmath363 and @xmath32 are nonrandom , then @xmath364 where @xmath365:{\inf_{\substack{g\subset{\mathbf{z}}^d:\\ g\text { \rm is cofinite}}}}\ c_p(f\cap g\,;a)>0 \bigg\},\ ] ] and @xmath366 is defined in above .
suppose that there exists a number @xmath172 $ ] for which @xmath367 where the infimum is computed over all cofinite sets @xmath84 .
define @xmath368 according to theorem [ th : rw : fp ] , @xmath369 it follows from this and elementary properties of probabilities that @xmath370 this in turn shows that @xmath371 apply lemma [ lem:0 - 1 ] , conditionally on @xmath17 , in order to deduce from the preceding that @xmath372 a.s .
@xmath373 $ ] .
in particular , we apply theorem [ th : dim : pi ] , once again conditionally on @xmath17 , in order to see that @xmath374 optimize over our choice of @xmath124 to see that @xmath375 for the other bound , it is enough to consider the case that @xmath376 $ ] since @xmath377 yields a trivial bound .
thus , we can consider instead a number @xmath378 such that @xmath379 , where once again the infimum is over all cofinite sets @xmath84 .
it is easy to deduce from this choice of @xmath124 and theorem [ th : rw : fp ] that @xmath380 since the random walk @xmath9 is transient , and because @xmath42 is a.s .
recurrent , elementary properties of probabilities imply that @xmath381 .
therefore , we may apply theorem [ th : dim : pi ] , one more time conditionally on @xmath17 , in order to see that @xmath382 .
optimize over our choice of @xmath124 to see that @xmath383 the corollary follows .
theorem [ th : rw : fp ] has a number of other consequences as well .
the following is a universal estimate on the expected martin @xmath124-capacity of the range of the fractal percolation set in a shell @xmath143 .
[ cor : mc1 ] for every point @xmath79 , integers @xmath45 , nonrandom finite sets @xmath384 , and percolation parameters @xmath385 $ ] , @xmath386 { \leqslant}256 c_{pq}(f;a).\ ] ] let @xmath260 denote an independent copy of @xmath387 and denote the corresponding ( independent of @xmath388 ) measure by @xmath389 with corresponding expectation operator @xmath390 . theorem [ th : rw : fp ] ensures that @xmath391 therefore we integrate @xmath392 $ ] in order to see that @xmath393.\ ] ] for the other bound , we recall that @xmath262 has the same law [ @xmath394 as @xmath263 does @xmath395 $ ] . therefore , theorem [ th : rw : fp ] implies that @xmath396 together , and yield @xmath397.$ ] the other bound in the statement follows similarly . the second consequence of theorem [ th : rw : fp ] is a lamperti - type condition on recurrence that was stated in corollary [ cor : lamperti ] .
consider the stopping times @xmath398 defined by @xmath399 where @xmath400 .
theorem [ th : rw : fp ] ensures that @xmath401 for all integers @xmath402 , and @xmath403 . of course ,
@xmath404 if and only if @xmath405 . therefore , if @xmath406 , then the borel cantelli lemma and together imply that @xmath407 .
note that this portion does not require the lamperti condition .
the complementary half of the corollary does .
if , on the other hand , @xmath408 , then ensures that @xmath409 .
a standard second moment argument reduces our problem to showing the existence of a positive constant @xmath410 such that @xmath411 as @xmath412 .
this is what we aim to prove . by the strong markov property , @xmath413\\ & { \leqslant}128 e^a\left [ { \mathds{1}}_{\{t_k<\infty\ } } c_1(f\cap{\mathcal{s}}_{k+l};x_{t_k})\right]\\ & { \leqslant}128 p^a\{t_k<\infty\}\sup_{z\in{\mathbf{z}}^d\cap{\mathcal{s}}_k } c_1(f\cap{\mathcal{s}}_{k+l};z ) .
\end{split}\ ] ] let us observe that , thanks to , there exists a finite constant @xmath414 such that @xmath415 whenever @xmath416 and @xmath417 for integers @xmath418 such that @xmath88 .
thus , it follows readily from the definition of @xmath359 that @xmath419 uniformly for all integers @xmath420 . in accord with , @xmath421
whenever @xmath420 .
similarly , we can appeal to in order to find a finite constant @xmath422 such that @xmath423 as long as @xmath424 .
let @xmath425 . because we can write @xmath426 as @xmath427 , the preceding two displayed bounds together imply that @xmath428 since @xmath429 as @xmath412 , we obtain if @xmath430 for all @xmath431 large .
the corollary follows immediately .
let us mention a final corollary of theorem [ th : rw : fp ] .
that corollary presents a more tractable formula for @xmath432 , valid under the lamperti - type condition .
[ cor : rw : dim ] let @xmath384 and @xmath2 be non random .
then , @xmath433 a.s .
@xmath434 $ ] , where @xmath435:\ \sum_{k=0}^\infty c_p(f\cap{\mathcal{s}}_k;a)<\infty\big\}\\ & = { \inf\big\ { p\in[2^{-d}\,,1]:\ \sum_{k=0}^\infty c_p(f\cap{\mathcal{s}}_k;a)=\infty\big\}. } \end{aligned}\ ] ] if , in addition , holds , then @xmath436$.}\ ] ] if @xmath9 is a random walk that satisfies the conditions of proposition [ pr : bt ] and starts at @xmath106 , then our previous comments in example [ ex : rw : nice ] imply that @xmath437 first we prove the upper bound on @xmath432 .
if @xmath438 for some @xmath275 $ ] , then theorem [ th : dim : pi ] ensures that @xmath439 ; see especially remark [ rem : dim : pi ] .
this , the easy half of the borel cantelli lemma , and theorem [ th : rw : fp ] together imply that @xmath440 .
optimize over @xmath275 $ ] in order to deduce that @xmath383 in the reverse direction we assume that holds , and strive to show that @xmath441 there is nothing to prove if @xmath442 .
therefore , we assume without loss of generality that @xmath443 according to theorem [ th : rw : fp ] , and thanks to the definition of the critical probability @xmath444 , @xmath445 for every @xmath446 $ ] .
that is , @xmath447 next we verify that there exists a uniform positive constant @xmath65 so that @xmath448 by the markov property of @xmath9 and
the particular construction of @xmath167 , @xmath449 theorem [ th : rw : fp ] then implies that @xmath450 where @xmath451 .
we now apply an argument very similar to the one used to produce in order to see that there exists an integer @xmath452 such that @xmath453 as long as @xmath454 . in this way we find that @xmath455 since @xmath456 [ see theorem [ th : rw : fp ] ] , it follows that @xmath457 therefore , theorem [ th : rw : fp ] shows that @xmath458 ^ 2 + 2k_*\mathbb{e}^a(\tau_n).\ ] ] because of the 0 - 1 law [ see lemma [ lem:0 - 1 ] ] , this and together imply that @xmath459 a.s .
@xmath373 $ ] as @xmath412 .
this is another way to state that @xmath372 a.s .
@xmath373 $ ] . theorem [ th : dim : pi]see , in particular , remark [ rem : dim : pi]then implies that @xmath460 since @xmath446 $ ] were arbitrary , the lower bound follows . because @xmath461 is equal to @xmath462 we can assume without loss of generality that @xmath32 is a finite set .
the first inequality of the proposition follows readily by adapting the second - moment argument of benjamini et al @xcite .
the few details follow . for every @xmath463 , there exists a unique positive integer @xmath136 such that @xmath464 .
let @xmath465 for this pairing of @xmath10 and @xmath193
. then we define , for all probability measures @xmath97 , a nonnegative random variable @xmath466 where @xmath467 for every @xmath82 .
the preceding display contains an almost surely well - defined sum because the summands are non negative and @xmath468 a.s .
[ @xmath13 ] for all @xmath115 .
we can therefore rearrange the sum and write @xmath470 because @xmath471 for all @xmath10 , @xmath472 similarly , we compute @xmath473 if @xmath474 for some @xmath97 , then certainly @xmath475 for some @xmath115 . therefore , the paley
zygmund inequality implies that for every @xmath97 , @xmath476 ^ 2}{\mathbb{e}^a(j_{\mu}^2)}\\\notag & { \geqslant}\left [ 2\left\ { i(\mu\,;a)+i_p(\mu\,;a)\right\ } \right]^{-1}. \end{aligned}\ ] ] the left - most quantity does not depend on @xmath97 ; therefore , we may optimize the right - most quantity in over all probability measures @xmath97 in order to see that @xmath477 .
this is the desired lower bound on the hitting probability of the theorem .
next we verify the complementary probability , still assuming without loss of generality that @xmath32 is finite ; that is , @xmath478 for a nonnegative integer @xmath136 that is still held fixed throughout . without loss of generality , we may also assume @xmath479 otherwise , there is nothing to prove . in order to establish the more interesting second inequality of the theorem we will need to introduce some notation .
let @xmath480 denote the sigma - algebra generated by @xmath481 for all @xmath115 .
recall that , because of our forest representation of @xmath0 , we identify every point @xmath482 with a ray in a finite tree @xmath316 , which was described in
[ sec : forest ] .
recall also that @xmath316 has been embedded in @xmath320 so that its deepest vertices lie on the real axis of @xmath320 . in this way
, we can identify every point @xmath482 with a point , which we continue to write as @xmath483 , on the real axis .
for every @xmath484 , let @xmath485 denote the sigma - algebra generated by the fractal - percolation weights @xmath486 for all @xmath487 such that @xmath488 .
similarly , let @xmath489 the sigma - algebra generated by all of the fractal - percolation weights @xmath486 , where @xmath487 satisfies @xmath490 .
if we think of @xmath483 as a maximum - depth vertex of @xmath316 and the latter is embedded in @xmath320 , as was mentioned earlier , then we can think of @xmath485 as the information , on the fractal percolation , on @xmath316 that lies to the left of @xmath483 [ including @xmath483 ] ; this is the `` @xmath492ast '' of @xmath483 .
similarly , we may think of @xmath489 as the information to the right of @xmath483 ; this is the `` @xmath493uture '' of @xmath483 .
next we define two `` 2-parameter martingales , '' @xmath494 and @xmath495 as follows : @xmath496;\qquad v_{m,\rho } : = \mathbb{e}^a[j_{\mu}\,|\,\mathcal{x}_m\vee \mathcal{p}_\rho],\ ] ] for all @xmath497 and @xmath484 .
because our random walk is independent of the fractal percolation , we may write the following after we appeal to independence : @xmath498 the markov property implies the a.s .- inequality , @xmath499 we stress , once again , that the ratio of the green s functions are well defined @xmath13-a.s .
similarly , @xmath500 therefore , with probability one , @xmath501 there are only a countable number of such pairs @xmath502 .
therefore , the previous lower bound on @xmath503 holds , off a single null set , simultaneously for all integers @xmath497 and integral points @xmath484 . in order to simplify the typesetting ,
let us write @xmath504 we might note that , with probability one , @xmath505 simultaneously for all integers @xmath497 .
now we apply an important idea that is , in a different form due to fitzsimmons and salisbury @xcite .
define a @xmath506-valued random variable @xmath507 by @xmath508 where @xmath400 .
@xmath507 is a stopping time with respect to the filtration of the random walk , conditionally on the entire history of the fractal percolation , @xmath286-a.s . on @xmath509 .
consider the event , @xmath510 hypothesis is another way to state @xmath511 .
moreover , implies the following key a.s .
inequality : @xmath512 the preceding is valid @xmath388-a.s . for any probability measure @xmath239 on @xmath32 .
we apply it using the following particular choice : @xmath513 for this particular choice of @xmath97 we obtain the following : @xmath514 ^ 2\,\bigg| g\bigg)\cdot\mathbb{p}^a(g)\\\notag & { \geqslant}\bigg[\mathbb{e}^a\bigg(\sum_{x\in{\mathbf{z}}^d } \frac{g(x_m\,,x)}{g(a\,,x ) } \left\{p^{-{\rm d}(x , x_m ) } { \mathds{1}}_{\{x\leftrightarrow x_m\ } } + { \mathds{1}}_{\{x\not\leftrightarrow x_m\}}\right\}\mu(x)\,\bigg| g\bigg)\bigg]^2\cdot\mathbb{p}^a(g)\\\notag & = \left [ i(\mu\,;a)+i_p(\mu\,;a)\right]^2 \cdot\mathbb{p}^a(g ) , \end{aligned}\ ] ] @xmath388-a.s . , thanks to the cauchy
schwarz inequality and out special choice of @xmath239 in .
[ the conditional expectation is well defined since @xmath511 . ] recall that the forest representation of @xmath0 in [ sec : forest ] identifies @xmath484 with a certain finite subset of the real line . with this in mind , we see that @xmath515 is a 2-parameter martingale under the probability measure @xmath388 , in the sense of cairoli @xcite , with respect to the 2-parameter filtration @xmath516 the latter filtration satisfies the commutation hypothesis ( f4 ) of cairoli @xcite because @xmath517 and @xmath492 are independent ; see khoshnevisan ( * ? ? ?
* , p. 35 ) for a more modern account .
therefore , cairoli s maximal inequality ( * ? ? ?
* corollary 3.5.1 , p. 37 ) implies that @xmath518 this and jensen s inequality together imply that @xmath519 similarly , we can prove that @xmath520 therefore , we may combine these remarks with in the following manner : @xmath521 because of the above bound and , and since @xmath511 [ see ] , it follows that @xmath522 is strictly positive and finite .
therefore , we may resolve using in order to obtain the inequality @xmath523 this completes our proof .
let us recall @xcite that the _ macroscopic upper minkowski dimension _ of a set @xmath524 is defined as in place of our @xmath525 . ]
@xmath526 where @xmath527 is the usual logarithm in base two . in analogy with the usual [ small - scale ] theory of the dimensions ,
one always has @xmath528 for all sets @xmath529 ; see barlow and taylor @xcite .
the minkowski dimension is perhaps the most commonly used notion of large - scale dimension , in some form or another , in part because it is easy to understand and in many cases compute . in the context of random walks
, we have the following elegant formula for the minkowski dimension of the range of a transient random walk on @xmath0 .
[ th : dim_m ] let @xmath9 denote a transient random walk on @xmath0 , with green s function @xmath19 , as before .
then , with probability one , @xmath530 where @xmath531 where @xmath532 .
the proof hinges on the analysis of the _ 0-potential measure _
@xmath533 defined for all @xmath534 . because @xmath9 is transient ,
the set function @xmath535 is a radon borel measure on @xmath154 . since @xmath536 for all @xmath349
, it follows that for all @xmath534 , @xmath537 = \sum_{x\in a } p^0\left\ { x_k = x\text { for some $ k{\geqslant}0$}\right\ } = \frac{u(a)}{g(0\,,0)},\ ] ] thanks to a combination of tonelli s theorem and .
the following simple argument proves immediately the first half of theorem [ th : dim_m ] .
we first prove that , with probability one , @xmath538 the more involved converse bound will be proved later .
chebyshev s inequality and together show that for all real numbers @xmath539 and integers @xmath193 , @xmath540 because @xmath541 for all @xmath349see there exists a finite constant @xmath542 such that @xmath543 for all @xmath193 .
therefore , the sum over @xmath136 of the right - hand side of is always finite when @xmath544 . if @xmath545 is such that the right - hand side of forms a summable series [ indexed by @xmath136 ] , then so does the left - hand side .
the borel
cantelli lemma shows that for any such value of @xmath547 the random variable @xmath548 defined by @xmath549 is a.s .
in particular , @xmath550 for all @xmath193 .
this proves that @xmath551 whence @xmath552 a.s .
for such a @xmath547 .
optimize over all such @xmath547 s in order to see that @xmath553 where @xmath532 .
to finish , note that if @xmath192 then @xmath554 , whence @xmath555 this and together imply . for the remaining , more challenging , direction of theorem [ th : dim_m ] we need to know that the measure @xmath535 is _ volume - doubling_. that is the gist of the following result .
[ pr : voldouble ] @xmath556 for all @xmath115 .
this is a volume - doubling result because @xmath557 .
see khoshnevisan and xiao @xcite for a corresponding result for lvy processes on @xmath154 .
the proposition holds trivially when @xmath558 .
therefore , we will concentrate on the case @xmath12 from now on .
we begin with a familiar series of random - walk computations .
choose and fix an integer @xmath12 and some @xmath10 .
then , we apply the strong markov property at @xmath559 @xmath560 $ ] in order to see that @xmath561.\ ] ] since @xmath562 @xmath101-a.s . on @xmath563 , and @xmath564 ,
this readily yields the `` shifted - ball inequality , '' @xmath565 eq
. becomes obvious if `` @xmath566 '' were replaced by `` @xmath567 . ''
the strong markov property of @xmath9 was needed in order to establish this improvement .
armed with we proceed in a standard way : we can always find @xmath568 integer points @xmath569 such that @xmath570 for all @xmath12 , where the union is a disjoint one .
thus , @xmath571 the proposition follows from this and .
next we rewrite @xmath572see in a slightly different form . we will be ready to complete the proof of theorem [ th : dim_m ] once that task is done .
[ pr : gamma_c ] @xmath573 .
if @xmath192 , then @xmath574 . therefore , @xmath575 therefore , we can infer from that @xmath576 we apply to rewrite @xmath572 once more time : if @xmath577 , then @xmath578 as @xmath37 . if on the other hand @xmath579 , then we can argue by contraposition to see that , for every fixed @xmath580 , @xmath581 for infinitely - many integers @xmath136 .
this means that @xmath582 now we prove the proposition .
the assertion of the proposition is that @xmath583 , where @xmath584
on one hand , implies that @xmath585 if , on the other hand , @xmath586 is an arbitrary finite number , then there exists a finite constant @xmath587 such that @xmath588 for all integers @xmath193 .
in particular , @xmath589 whence follows that @xmath590 as @xmath591 .
since this is true for all @xmath586 , it follows that @xmath592 , as was promised .
it remains to prove that @xmath593 where @xmath572 was defined in .
if @xmath594 , then we are done .
therefore , from now on we assume without loss of generality that @xmath595 define @xmath596 for all @xmath10 [ @xmath597 .
since @xmath598 , tonelli s theorem implies that @xmath599 + 2\mathop{\sum\sum}\limits_{\substack{x , y\in{\mathcal{v}}_n\\x\neq y } } p^0\left\ { \tau(x)<\tau(y)<\infty\right\}\\ & = e^0\left[\operatorname{card}\left(\mathcal{r}_x\cap{\mathcal{v}}_n\right)\right]+ 2 \mathop{\sum\sum}\limits_{\substack{x , y\in{\mathcal{v}}_n\\x\neq y } } p^0\ { \tau(x)<\infty\}p^x\{\tau(y)<\infty\ } , \end{split}\ ] ] thanks to the strong markov property .
if @xmath600 and @xmath12 are held fixed , then @xmath601 since @xmath602 .
therefore , implies that @xmath603 + \frac{2u({\mathcal{v}}_n)u({\mathcal{v}}_{n+1})}{[g(0\,,0)]^2}\\ & \qquad{\leqslant}e^0\left[\operatorname{card}\left(\mathcal{r}_x\cap{\mathcal{v}}_n\right)\right ] + 2^{1 + 2d}\left\{e^0\left[\operatorname{card}\left(\mathcal{r}_x\cap{\mathcal{v}}_n\right)\right]\right\}^2 , \end{split}\ ] ] thanks to and proposition [ pr : voldouble ] . because @xmath604= \frac{u({\mathcal{v}}_n)}{g(0\,,0)},\ ] ] eq . and the paley
zygmund inequality together yield the following : for infinitely many @xmath605 , @xmath606\right\}^2 } { 4e^0\left(\left|\operatorname{card}\left ( \mathcal{r}_x\cap{\mathcal{v}}_n\right)\right|^2\right)}\\ & { \geqslant}\frac{1}{4(1 + 2^{1+d})}\\ & : = \varrho(d ) .
\end{split}\ ] ] the last part holds since implies that @xmath607{\geqslant}1 $ ] for infinitely - many integers @xmath608 .
the preceding displayed inequality and proposition [ pr : gamma_c ] together imply that @xmath609 , with probability at least @xmath610 . since @xmath611 for all @xmath612 , an application of the hewitt
savage 01 law shows the desired result that @xmath613 almost surely .
corollary [ cor : main ] succeeds in yielding a formula for @xmath432 for every recurrent set @xmath384 , though it is difficult to work with that formula .
we do not expect a simple formula for @xmath432 when @xmath32 is a general recurrent set in @xmath0 .
in fact , it is not even easy to decide whether or not a general set @xmath32 is recurrent , as we have seen already .
however , one can hope for simpler descriptions of @xmath432 when @xmath614 . in this section
we conclude with a series of remarks , problems , and conjectures that have these comments in mind .
question was in part motivated by its `` local variation , '' which had been open since the mid - to - late 1960 s @xcite , and possibly earlier .
namely , let @xmath615 be a lvy process in @xmath154 . the local version of asks , `` what is the ordinary hausdorff dimension @xmath67 of the range @xmath616 ? ''
this question was answered several years later by khoshnevisan , xiao , and zhong ( * ? ? ?
* corollary 1.8 ) , who showed among other things that @xmath617 is a.s .
equal to an index that was introduced earlier in pruitt @xcite as part of the solution to the very same question . under a quite mild regularity condition
, it has been shown that the general formula for @xmath617 reduces to the following ( * ? ? ?
* ( 1.19 ) of theorem 1.5 ) : @xmath618 where @xmath619 denotes the 1-potential kernel of @xmath202 .
khoshnevisan and xiao @xcite find an alternative fourier - analytic formula .
if we proceed purely by analogy , then we might guess from and the following formula for the macroscopic hausdorff dimension of the range @xmath17 of our random walk @xmath9 on @xmath0 : @xmath620 in principle , we ought to be able to decide whether or not is correct , based solely on corollary [ cor : main ] .
but we do not know how to do that at this time mainly because it is quite difficult to compute @xmath621 when @xmath9 is a general transient random walk .
instead , we are able to only offer [ op1 ] @xmath622 a.s . for every transient random walk on @xmath0 , where @xmath572 was defined in . because of theorem [ th : dim_m ] ,
conjecture [ op1 ] is equivalent to the assertion that @xmath623 a.s .
it is known that the ordinary [ microscopic ] hausdorff dimension of the range of a lvy process is always equal to its ordinary [ microscopic ] _ lower _ minkowski dimension , and not always the upper minkowski dimension .
if conjecture [ op1 ] were correct , then it would suggest that large - scale dimension theory of random walks is somewhat different from its small - scale counterpart .
our next problem is an attempt to understand this difference better .
barlow and taylor @xcite have introduced two other notions of macroscopic dimension that are related to our present interests .
namely , they define the [ macroscopic ] _ lower minkowski dimension _ of @xmath524 and the _ lower hausdorff dimension _ of @xmath524 respectively as and @xmath624 in place of our @xmath625 and @xmath626 . ]
@xmath627 one has @xmath628 and @xmath629 for all @xmath529 .
it is easy to obtain a nontrivial upper bound for the lower minkowski dimension of @xmath17 , valid for every transient random walk @xmath9 on @xmath0 .
namely , by and fatou s lemma , @xmath630 { \leqslant}\liminf_{n\to\infty } 2^{-n\gamma } u({\mathcal{v}}_n),\ ] ] for every @xmath631 . from this we readily can deduce that @xmath632 we believe that this is a sharp bound , and thus propose the following .
[ op2 ] with probability one , @xmath633 admittedly , we have not tried very hard to prove this , but it seems to be a natural statement .
there are two other good reasons for our interest in conjecture [ op2 ] .
first of all , it suggests that , as far as random walks and their analogous lvy processes are concerned , the more natural choice of `` macroscopic hausdorff dimension '' is @xmath634 and not @xmath7 , in contrast with the proposition of @xcite .
also , if conjecture [ op2 ] were true , then together with theorem [ th : dim_m ] and proposition [ pr : gamma_c ] it would imply that regardless of whether or not conjecture [ op1 ] is true , @xmath29 always lies in the nonrandom interval @xmath635.$ ] the extrema of this interval are typically not hard to compute ; therefore , we at least will have easy - to - compute bounds for @xmath29 .
let us state a third conjecture that is motivated also by conjecture [ op1 ] .
choose and fix an arbitrary integer @xmath612 , and define @xmath636 to be @xmath431 independent copies of a symmetric , transient random walk @xmath9 on @xmath0 whose green s function satisfies the barlow
taylor condition for some @xmath24 $ ] .
we can define an @xmath431-parameter _ additive random walk _ @xmath637 as follows ( * ? ? ?
4 ) : for every @xmath638 , @xmath639 let @xmath640 denote the _ range _ of the random field @xmath641 .
[ op3 ] suppose @xmath642 and @xmath643 .
then for all nonrandom @xmath524 : 1 . if @xmath644 , then @xmath645 is a.s .
unbounded ; 2 . if @xmath646 , then @xmath645 is a.s . bounded .
proposition [ pr : bt ] implies that conjecture [ op3 ] is correct if @xmath431 were replaced by @xmath107 ; the case @xmath643 has eluded our many attempts at solving this problem .
it is possible to adapt the arguments of @xcite in order to derive conjecture [ op1 ] from conjecture [ op3 ] .
we skip the details of that argument . instead
, we conclude with two problems about the `` continuous version '' of corollary [ cor : main ] , which we recall , contained our hausdorff dimension formula for the range of a walk .
let @xmath615 be a transient , but otherwise general , lvy process on @xmath154 whose characteristic exponent is @xmath647 , normalized as @xmath648 to be concrete .
is there a formula for the a.s .- constant quantity @xmath649 that is solely in terms of @xmath647 ?
before we state our last question let us define the _ upper minkowski dimension _ of a set @xmath650 as follows : define @xmath651 to be the union of all dyadic cubes @xmath652 of sidelength one that intersect @xmath65 .
the _ macroscopic upper minkowski dimension _
@xmath653 is defined , via the barlow
taylor upper minkowski dimension , as follows : @xmath654 the same proof that worked for @xmath529 works to show that @xmath655 for all @xmath534 .
although we have not checked all of the details , we believe that the method of proof of theorem [ th : dim_m ] can be adapted to the continuous setting in order to produce @xmath656 where @xmath657 for all borel sets @xmath534 . can one write an expression for @xmath658 solely in terms of @xmath647 ? finally , we mention that conjecture [ op3 ] is likely to have a lvy process version wherein the role of the @xmath659s are replaced by that of isotropic @xmath68-stable lvy processes .
we leave the statement [ and perhaps also a proof ! ] to the interested reader .
alex d. ramos thanks cnpq - brazil and capes , process bex 2176/13 - 0 , for their financial support , and the mathematics department of university of utah for their hospitality during his visit .
99 m. t. barlow and s. j. taylor , fractional dimension of sets in discrete spaces , _ j. phys .
* 22 * ( 1989 ) no . 13 , 26212626 .
martin t. barlow and s. james taylor , defining fractal subsets of @xmath0 , _ proc .
london math .
soc . _ ( 3 ) * 64 * ( 1992 ) 125152 .
itai benjamini , robin pemantle , and yuval peres , martin capacity for markov chains , _ ann .
_ * 23*_(3 ) _ ( 1995 ) 13321346 .
r. s. bucy , recurrent sets , _ ann .
_ * 36*(_2 _ ) ( 1965 ) 535545 .
r. cairoli , sur la convergence des martingales indexes par @xmath660 , _ sminaire de probabilits , xiii _
strasbourg , strasbourg , 1977/78 ) , pp . 162173 , in : _ lecture notes in math .
_ * 721 * , springer , berlin , 1979 .
p. j. fitzsimmons and thomas s. salisbury , capacity and energy for multiparameter markov processes , _ ann .
de li.h.p .
_ * 25*_(3 ) _ ( 1989 ) 325350 .
k. it and h. p. mckean , potentials and random walks , _ ill .
j. math . _
( 1960 ) 119132 .
davar khoshnevisan , _ multiparameter processes : an introduction to random fields _ , springer , new york , 2002 .
davar khoshnevisan and yimin xiao , harmonic analysis of additive lvy processes , _ probab .
rel/ fields _ * 145 * ( 2009 ) 459515 .
davar khoshnevisan and yimin xiao , packing dimension of the range of a lvy process , _ the proc .
* 136*_(7 ) _ ( 2008 ) 25972607 .
davar khoshnevisan , yimin xiao , and yuquang zhong , measuring the range of an additive lvy process , _ ann .
_ * 31*(_(2 ) _ ( 2003 ) 10971141 .
davar khoshnevisan and yimin xiao , weak unimodality of finite measures , and an application to potential theory of additive lvy processes , _ proc .
_ * 131*_(8 ) _ ( 2003 ) 26112616 .
h. kesten , p. ney , and f. spitzer , the galton - watson process with mean one and finite variance , _ teor . verojatnost .
i primenen . _
( 1966 ) 579611 [ also _ th .
appl . _ * 11 * ( 1966 ) 513540 ]
. a. n. kolmogorov , on the solution of a problem in biology , _ izv .
nii mat .
_ * 2*_(1 ) _ ( 1938 ) 712 ( in russian ) .
john lamperti , wiener s test and markov chains , _ j. math . analysis and appl . _ * 6 * ( 1963 ) 5866 .
russel lyons , robin pemantle , and yuval peres , conceptual proofs of @xmath661 criteria for mean behavior of branching processes , _ ann
. probab .
_ * 23*_(3 ) _ ( 1995 ) 11251138 .
benot mandelbrot , _ the fractal geometry of nature _ , w. h. freeman , san francisco , 1983 . j. naudts , dimension of discrete fractal spaces , _ j. phys .
a _ * 21 * ( 1988 ) 447452 . j. naudts , reply to a paper by m. t. barlow and s. j. taylor , _ j. phys . a _ * 22 * ( 1989 ) 26272628 .
yuval peres , remarks on intersection equivalence and capacity equivalence , _ ann .
h. poincar : phys .
_ * 64*_(3 ) _ ( 1996 ) 339347
. w. e. pruitt , the hausdorff dimension of the range of a process with stationary independent increments , _ j. math .
_ * 19 * ( 1969 ) 371378
. s. j. taylor , multiple points for the sample paths of the symmetric stable process , _ z. wahr . verw .
( 1966 ) 247264 . * nicos georgiou * + department of mathematics , university of sussex , sussex house , brighton , bn1 9rh , united kingdom + + * davar khoshnevisan * & * kunwoo kim * + department of mathematics , university of utah , salt lake city , ut 84112 - 0090 + & + * alex d. ramos * + federal university of pernambuco , department of statistics / ccen , av .
luiz freire , s / n , cidade universitria , recife / pe , 50740 - 540 , brazil + | we find formulas for the macroscopic minkowski and hausdorff dimensions of the range of an arbitrary transient walk in @xmath0 .
this endeavor solves a problem of barlow and taylor ( 1991 ) .
+ _ keywords . _
transient random walks , hausdorff dimension , recurrent sets , fractal percolation , capacity .
+ _ ams 2010 subject classification . _ primary 60g50 ; secondary 60j45 , 60j80 . |
the _ cluster algebra _ of fomin - zelevinsky was turned out be ubiquitous in algebraic lie theory and classical invariant theory .
it had been realized under different motivation ( eg . , @xcite ) that the _ upper cluster algebra _
@xcite is a more useful notion than the cluster algebra .
the purpose of this series of papers is to understand when a semi - invariant of quiver representations is an upper cluster algebra , and use the cluster structure solve problems in invariant theory .
one family of such upper cluster algebras found in @xcite is the semi - invariant ring of complete _ triple flags _ introduced in @xcite .
they are _ multigraded _ upper cluster algebras of the _ hive quivers_. the grading is also very useful because the dimension of each graded piece counts certain _ littlewood - richardson coefficient_. on the other hand , many quivers can be projected from the triple flag quivers through exceptional sequences .
in fact , the author does not know any quiver without oriented cycles that can not be projected from a triple flag quiver . as we shall see in this paper that finding the cluster structure of semi - invariant ring of their representations will be ultimately reduced to the case of complete triple flags .
more naturally and generally , one can ask how a graded upper cluster algebra ( or guca in short ) might change into another one under degree ( or weight ) restriction .
we are unable to answer this question in general . however
, if the grading is restricted to the hyperplane @xmath9 cut out by some weight @xmath10 , then we can provide a complete solution under some additional conditions .
there is one special case which is particularly interesting to us .
that is when @xmath11 is a _ projector _ in the graded upper cluster algebra @xmath12 . here
, being a projector simply means that @xmath13 is _ real _ and @xmath14 for some @xmath15 . in this case , the restriction leads to deleting some vertex . for the general cases , we need more restriction on the weights @xmath16 of an extended cluster . in general
, we need to delete a set of frozen vertices , then freeze another set of vertices .
let @xmath6 be an ice quiver , and @xmath17 be its associated upper cluster algebra .
let @xmath18 be a set of vertices of @xmath6 .
we use the notation @xmath19 and @xmath20 for the quiver obtained from @xmath6 by deleting and freezing @xmath18 respectively .
let @xmath21 be a finite quiver without oriented cycles , and @xmath1 be a fixed dimension vector of @xmath2 .
the space @xmath22 of all @xmath1-dimensional representations is acted by the product of special linear groups @xmath23 by the natural base change .
the rings of semi - invariants @xmath0 is by definition the invariant ring @xmath24^{{\operatorname{sl}}_\beta}$ ] .
this algebra is graded by a sublattice of @xmath25 equipped with a bilinear form @xmath26 .
we denote by @xmath27 the weight-@xmath28 component of @xmath0 .
the weight restriction to @xmath29 applied to @xmath0 leads to a change of the quiver @xmath2 and the dimension vector @xmath1 .
this was studied by a. schofield @xcite and derksen - weyman @xcite . to be more precise
, we have that @xmath30 , where @xmath31 is another quiver with one vertex less than @xmath2 , and @xmath32 is a dimension vector of @xmath33 .
the general weight restriction does not necessarily lead to such a change .
however , it does in another special case called _ vertex removal _ with an additional condition on @xmath1 .
we denote the new quiver with dimension vector in this situation by @xmath34 .
put all these together , and we obtain our first two main results .
suppose that @xmath0 is a naturally graded uca @xmath35 , and @xmath36 has full rank . if @xmath37 is real and extremal in the grading cone of @xmath0 , then we have that @xmath38 .
suppose that @xmath0 is a naturally graded uca @xmath39 , and @xmath36 has full rank .
let @xmath40 be a vertex of @xmath2 satisfying and that @xmath41 for all @xmath42 except for a set of frozen vertices @xmath18 , where all @xmath43 ( or all @xmath44 ) .
then we have that @xmath45 , where @xmath46 is _ attached _ to @xmath18 . in @xcite and @xcite , the _ mutation _ of _ quivers with potentials _
is invented to model the cluster algebras .
for an ice quiver with potential @xmath47 , we can associated to it the _ jacobian algebra _ @xmath48 .
let @xmath49 be the set of all _ @xmath50-supported vectors _ in @xmath51 introduced in @xcite .
we say an iqp @xmath47 _ models _ an algebra @xmath12 if the _ generic cluster character _ of @xcite maps
@xmath49 onto a basis of @xmath12 .
our other two main results say that the model is in some sense hereditary under the two operations considered before .
suppose that @xmath52 is a projector in @xmath39 .
if @xmath47 models @xmath39 , then @xmath53 models @xmath54 . under the same assumption as proposition [ p : outerproj ] , we suppose that @xmath47 models @xmath39 . then @xmath55 models @xmath56 .
if in addition @xmath49 is given by lattice points in some polyhedron , then so are @xmath57 and @xmath58 .
a comprehensive application of this theory has been illustrated in @xcite . to invite readers ,
we still provide many other examples .
the first class of examples explains how we get some known cluster structure from @xcite .
they include the total coordinate rings of partial flag varieties and unipotent subgroups of @xmath59 @xcite .
the second class of examples is mildly new .
they are the semi - invariant rings of complete @xmath60-tuple flags for @xmath61 , whose cluster structure can also be obtained from @xcite .
in section [ s : guca ] we recall the graded upper cluster algebra following @xcite .
the new emphasis is on the grading given by a weight configuration of a unimodular lattice . in section
[ s : proj ] , we introduce the crucial map @xmath62 and the definition of a projector .
we prove the precursors of the first two main results
proposition [ p : subalgebra ] and [ p : outerproj ] . in section [ s : si ] , we recall the theory of semi - invariant ring of quiver representations with an emphasis on the two operations orthogonal projection and vertex removal .
we give a characterization of a projector in a semi - invariant ring in proposition [ p : extremal_proj ] .
we prove our first two main results theorem [ t : invguca ] and [ t : removal ] , and provide many examples . in section
[ s : cc ] , we recall the ice quivers with potentials and their representations following @xcite .
we recall the generic cluster character in the current setting following @xcite . in section [
s : model ] , we use the restriction and induction functor to study the relation of the cluster models under the two operations .
we prove the other two main results theorem [ t : frozen_model ] and [ t : hat ] , and provide examples .
our vectors are exclusively row vectors .
all modules are right modules . for a quiver
@xmath2 , we denote by @xmath63 the set of vertices and by @xmath64 the set of arrows . for an arrow @xmath65 ,
we denote by @xmath66 and @xmath67 its tail and head .
arrows are composed from left to right , i.e. , @xmath68 is the path @xmath69 .
unadorned @xmath70 and @xmath71 are all over the base field @xmath72 .
we write @xmath73 for @xmath74 . for any representation @xmath75
, @xmath76 is the dimension vector of @xmath75 . for direct sum of @xmath77 copies of @xmath75
, we write @xmath78 instead of the traditional @xmath79 .
a quiver is just a directed graph with possibly multiple arrows between two vertices . throughout
we assume that quivers have no loops or oriented 2-cycles .
quiver @xmath6 is a quiver , where some vertices are designated as _
mutable _ while the rest are designated _
frozen_. we usually label the quiver such that the first @xmath80 vertices are mutable .
if we require no arrows between frozen vertices , then such a quiver is uniquely determined by its _
@xmath36 . it is a @xmath82 matrix given by @xmath83 the combinatorial data defining a cluster algebra is completely encoded in an ice quiver .
[ d : qmu]let @xmath84 be a mutable vertex of @xmath6 . the _ quiver mutation _
@xmath85 transforms @xmath6 into the new quiver @xmath86 via a sequence of three steps . 1 . for each pair of arrows @xmath87 ,
introduce a new arrow @xmath88 ( unless both @xmath89 and @xmath90 are frozen , in which case do nothing ) ; 2 . reverse the direction of all arrows incident to @xmath84 ; 3 . remove all oriented 2-cycles .
the above recipe can be reformulated in terms of @xmath81-matrix as follows .
let @xmath91 be the @xmath92 matrix obtained from the identity matrix by replacing the @xmath84-th row by a vector @xmath93 where @xmath94 we write @xmath95 for the restriction of @xmath91 to its @xmath96 upper left corner .
then the mutated @xmath81-matrix @xmath97 for @xmath8 is related to the original one by @xmath98 we note that @xmath99 and @xmath100 so the quiver mutation is an _ involution _ and preserves the rank of the @xmath81-matrix . [ d : seeds ] let @xmath101 be a field containing @xmath72 .
a _ seed _ in @xmath101 is a pair @xmath102 consisting of an ice quiver @xmath6 together with a collection @xmath103 , called an _ extended cluster _
, consisting of algebraically independent ( over @xmath72 ) elements of @xmath101 , one for each vertex of @xmath6 .
the elements of @xmath104 associated with the mutable vertices are called _ cluster variables _ ; they form a _ cluster_.
the elements associated with the frozen vertices are called _ frozen variables _ , or _ coefficient variables_. a _ seed mutation _
@xmath85 at a ( mutable ) vertex @xmath84 transforms @xmath102 into the seed @xmath105 defined as follows .
the new quiver is @xmath86 .
the new extended cluster is @xmath106 where the new cluster variable @xmath107 replacing @xmath108 is determined by the _ exchange relation _ @xmath109
we note that the mutated seed @xmath110 contains the same coefficient variables as the original seed @xmath102 .
it is easy to check that one can recover @xmath102 from @xmath110 by performing a seed mutation again at @xmath84 .
two seeds @xmath102 and @xmath111 that can be obtained from each other by a sequence of mutations are called _ mutation - equivalent _ , denoted by @xmath112 . if @xmath6 and @xmath113 are clear from the context , we may just write @xmath114 .
[ d : cluster - algebra ] the _ cluster algebra @xmath115 _ associated to a seed @xmath102 is defined as the subring of @xmath101 generated by all elements of all extended clusters of the seeds mutation - equivalent to @xmath102 .
note that the above construction of @xmath115 depends only , up to a natural isomorphism , on the mutation equivalence class of the initial quiver @xmath6 .
so we may drop @xmath104 and simply write @xmath116 .
an amazing property of cluster algebras is [ t : laurent ] any element of a cluster algebra @xmath115 can be expressed in terms of the extended cluster @xmath104 as a laurent polynomial , which is polynomial in coefficient variables .
since @xmath115 is generated by cluster variables from the seeds mutation equivalent to @xmath102 , theorem [ t : laurent ] can be rephrased as @xmath117 where @xmath118 $ ] .
note that our definition of @xmath119 is slightly different from the original one in @xcite , where the coefficient variables are inverted in @xmath120 $ ] .
the _ upper cluster algebra _ ( or uca in short ) with seed @xmath102 is @xmath121 in general , there may be infinitely many seeds mutation equivalent to @xmath102 .
so the above definition is not very useful to test the membership in an upper cluster algebra .
however , the following theorem allows us to do only finitely many checking .
@xcite we define the upper bounds @xmath122 ( * ? ? ?
* corollary 1.9 ) [ t : bounds ] suppose that @xmath36 has full rank , and @xmath112 , then @xmath123 .
in particular , @xmath124 .
this theorem is originally proved for @xmath125 and @xmath126 with @xmath119 defined there . however ,
if we carefully examine the argument , we find that the result is also valid for our @xmath119 . any upper cluster algebra , being a subring of a field , is an integral domain ( and under our conventions , a @xmath72-algebra ) .
however , it may fail to be noetherian @xcite .
since normality is preserved under localization and intersection , any uca is normal .
the next lemma is useful to identify an uca as a subalgebra of some given noetherian normal domain .
let @xmath104 be an extended cluster in some uca .
if @xmath104 is contained in another commutative ring @xmath127 , we write @xmath128 for the union of the zero locus in @xmath129 defined by all cluster variables in @xmath104 .
[ d : cr1 ] we say that a seed @xmath102 is cluster - regular in codimension 1 ( or cr1 in short ) in @xmath127 if each adjacent cluster @xmath130 , and @xmath131 has codimension @xmath132 in @xmath133 .
we call two elements of @xmath127 _ coprime in codimension 1 _ if the locus of their common zeros has codimension @xmath134 in @xmath135 . by expanding @xmath131
, we can easily see that the codimension condition in definition [ d : cr1 ] is equivalent to that @xmath136 [ l : rca ] let @xmath127 be a @xmath72-algebra and a noetherian normal domain . if @xmath102 is a cr1 seed in @xmath127 , then @xmath137 .
for any @xmath40 in the complement of @xmath131 in @xmath133 , we can find a seed @xmath138 ( either the original one or an adjacent one ) such that any cluster variable in @xmath139 does not vanish at @xmath40 .
we can express any @xmath140 as a laurent polynomial in @xmath139 , which is polynomial in coefficient variables .
since @xmath141 , we conclude that @xmath142 is regular at @xmath40 .
we know from ( * ? ? ?
* corollary 5.22 ) that if @xmath127 is a noetherian domain , then @xmath143 is the integral closure of @xmath127 . by the cr1 assumption , we conclude that @xmath142 is in the integral closure of @xmath127 which is @xmath127 itself by the normality .
let @xmath103 be an ( extended ) cluster .
for a vector @xmath145 , we write @xmath146 for the monomial @xmath147 . for @xmath148 , we set @xmath149 where @xmath150 is the @xmath84-th row of the matrix @xmath36 , and let @xmath151 .
suppose that an element @xmath152 can be written as @xmath153 where @xmath154 is a rational polynomial not divisible by any @xmath155 , and @xmath156 .
if we assume that the matrix @xmath36 has full rank , then the elements @xmath157 are algebraically independent so that the vector @xmath158 is uniquely determined @xcite .
we call the vector @xmath158 the ( extended ) _ @xmath144-vector _ of @xmath142 .
definition implies at once that for two such elements @xmath159 we have that @xmath160 .
so the set @xmath161 of all @xmath144-vectors in @xmath17 forms a sub - semigroup of @xmath162 .
[ l : independent ] assume that the matrix @xmath36 has full rank .
let @xmath163 be a subset of @xmath17 with distinct well - defined @xmath144-vectors .
then @xmath164 is linearly independent over @xmath72 .
let @xmath165 be a _
unimodular _ lattice equipped with a bilinear form @xmath166 , @xmath167 .
the bilinear form may be neither symmetric nor skew - symmetric but must be unimodular , i.e. , the mapping @xmath168 is a @xmath169-module isomorphism .
this kind of lattice is the one discussed throughout this paper .
for any @xmath170 , we write @xmath171 for its left and right _ orthogonal sublattices_. the _ coxeter transformation _
@xmath172 is the unique linear transformation determined by @xmath173 .
note that @xmath174 via the coxeter transformation , allowing us to restrict our discussion to right orthogonal sublattices . for a _ real _ vector @xmath175 , i.e. , @xmath176 , we define two maps @xmath177 called the right and left _ projection _ through @xmath13 . [
d : wtconfig ] a _ weight configuration _
@xmath16 of @xmath178 on an ice quiver @xmath6 is an assignment for each vertex @xmath89 of @xmath6 a ( weight ) vector @xmath179 such that for each mutable vertex @xmath84 , we have that @xmath180 the _ mutation _ @xmath85 also transforms @xmath16 into a weight configuration @xmath181 on the mutated quiver @xmath182 defined as @xmath183 we usually write such a weight configuration as a pair @xmath184 and its mutation by @xmath185 .
by slight abuse of notation , we can view @xmath16 as a matrix whose @xmath89-th row is the weight vector @xmath186 . in this matrix notation ,
the condition is equivalent to @xmath187 is a zero matrix . for any weight configuration of @xmath6
, the mutation can be iterated because of .
let @xmath184 be a weight configuration of a unimodular lattice .
let @xmath52 be a vertex of @xmath6 such that @xmath188 corresponds to a real vector @xmath13 .
we define the ( left ) _ projection _ of the weight configuration @xmath184 through @xmath52 as the pair @xmath189 , where @xmath190 is the full subquiver obtained from @xmath6 by deleting the vertex @xmath52 , and @xmath191 it is easy to see that the projection commutes with mutations away from @xmath52 . given a weight configuration @xmath184 , we can assign a multidegree ( or weight ) to the upper cluster algebra @xmath126 by setting @xmath192 for @xmath193
. then mutation preserves multihomogeneousity .
we say that this upper cluster algebra is @xmath16-graded , and denoted by @xmath194 .
we refer to @xmath195 as a graded seed .
note that the variables in @xmath196 have zero degrees .
so if @xmath142 has a well - defined @xmath144-vector as in , then @xmath142 is homogeneous of degree @xmath197 .
let @xmath198 be a monoid . for simplicity , we assume that @xmath198 is a submonoid inside a unimdoular lattice @xmath199 .
let @xmath200 be a @xmath198-graded @xmath72-algebra , which is also an integral domain .
we do not assume that @xmath200 is noetherian , or have finite - dimensional graded pieces .
we will speak of a @xmath198-graded @xmath72-algebra just as a graded algebra . for any @xmath201
, we define @xmath202 to be the graded subalgebra @xmath203 of @xmath200 .
we denote by @xmath204 the localization of @xmath200 at @xmath205 . for any @xmath205 of real weight @xmath13
, we define the algebra homomorphism @xmath206 by the following @xmath72-linear extension on the homogeneous components : @xmath207 clearly @xmath62 fixes @xmath202 . for any @xmath208 ,
we have that @xmath209 so @xmath62 is surjective .
the homomorphism @xmath62 can be uniquely extended to the quotient fields @xmath210 , which fixes @xmath211 .
we denote the extension still by @xmath62 .
let @xmath212 be the embedding @xmath213 .
[ d : projector ] an element @xmath214 is called a _ projector _ in @xmath200 if @xmath13 is real and @xmath215 .
such an @xmath216 is also called a _ projector _ for @xmath200 . if @xmath52 is a projector , then @xmath62 maps @xmath200 onto @xmath202 .
we call @xmath62 the _ projection through _ @xmath52 .
[ l : projector ] @xmath214 is a projector if and only if for any @xmath217 with @xmath218 , @xmath219 factors as @xmath220 for some @xmath221 . clearly @xmath222 .
for @xmath223 such that @xmath224 , we have that @xmath225 . [
p : projector ] suppose that @xmath13 is a projector for @xmath200 , then we have the following . 1 . if @xmath226 , then @xmath227 for any @xmath228
if any @xmath229 is a unit in @xmath200 , then any @xmath230 is irreducible in @xmath200 .
3 . if @xmath200 is a ufd and @xmath226 contains all units in @xmath200 , then @xmath13 is extremal in the cone @xmath231 .
if @xmath214 is a projector , then for all @xmath232 we have that @xmath233 for some @xmath234 .
but @xmath226 , so @xmath227 . ( 2 ) .
let @xmath52 be a projector in @xmath235 .
suppose that @xmath236 with @xmath237 and @xmath238 .
one of @xmath239 and @xmath1 , say @xmath239 , must have that @xmath218 .
then @xmath240 for some @xmath241 . on the other hand , @xmath242 for some @xmath229 .
then @xmath243 because @xmath200 is a domain .
by assumption @xmath84 is a unit , then so is @xmath244 .
suppose that @xmath13 is not extremal , then there is some @xmath245 such that @xmath246 . by ( 1 )
, we have that @xmath247 for @xmath248 , and some @xmath249 . by ( 2 )
@xmath52 is irreducible , so @xmath247 contradicts the ufd property of @xmath200 . throughout this paper
, we will use @xmath52 for both the vertex @xmath52 and the cluster variable @xmath251 .
we find no potential confusion from this abuse of notation . from now on , the map @xmath62 is always defined on the quotient field @xmath252 of @xmath194 . if @xmath104 is an extended cluster , when we say laurent polynomial in @xmath104 "
, it always implies that it is polynomial in coefficient variables in @xmath104 .
we omit this part if it is not important in the context .
the goal of this subsection is to relate @xmath253 to another upper cluster algebra , where @xmath254 is real .
our first lemma reduces the general case to the case when @xmath52 is frozen . by freezing a set of vertex @xmath46 , we mean that we make @xmath46 frozen .
let @xmath255 be the quiver obtained from @xmath6 by freezing @xmath46 .
[ l : reduce2frozen ] suppose that @xmath36 is of full rank .
let @xmath256 and @xmath257 , then we have that @xmath258 .
we first show that @xmath259 is a subalgebra of @xmath12 . by theorem [ t : bounds ] , it suffices to show that any @xmath260 can be written as a laurent polynomial in @xmath261 .
note that @xmath262 can be written as a laurent polynomial @xmath80 in @xmath104 , which is polynomial in @xmath52 .
but @xmath263 , where @xmath244 is the exchange polynomial .
substituting @xmath52 by @xmath264 , we get the required laurent polynomial in @xmath265 .
since any cluster @xmath104 of @xmath259 is also a cluster of @xmath12 , any element in @xmath12 can be written as a laurent polynomial in @xmath104 but not necessary polynomial in @xmath52 .
so after inverting @xmath52 in @xmath259 , we obtain that @xmath266 .
we conclude that @xmath267 .
[ r : cr1 ] the lemma still holds if the full rank condition is replaced by the condition that @xmath12 is noetherian and @xmath102 is a cr1 seed in @xmath12 . indeed ,
if @xmath102 is a cr1 seed in @xmath12 , then @xmath268 is also a cr1 seed in @xmath12 by . by lemma [ l : rca ]
, we still have that @xmath269 .
the same remark applies to several results depending on lemma [ l : reduce2frozen ] , eg . , proposition [ p : subalgebra ] , corollary [ c : gproj_e ] , theorem [ t : invguca ] , and corollary [ c : model_proj ] .
for a graded seed @xmath195 , we define @xmath270 by @xmath271 for @xmath272 . since @xmath104 is algebraically independent , so is @xmath270 .
we have seen that @xmath273 is a weight configuration , so @xmath274 is a graded seed .
[ l : projmu ] for any sequence of mutations @xmath275 with @xmath276 , we have that @xmath277 and @xmath278 for @xmath272 .
it suffices to show for @xmath279 with @xmath280 .
it is rather clear from the definition that @xmath85 commutes with the deletion at @xmath52 . the map @xmath62 commutes with mutations because @xmath62 is a homomorphism .
[ c : liftcluster ] for any extended cluster @xmath281 of @xmath282 , there is an extended cluster @xmath283 of @xmath284 such that @xmath285 and @xmath286 for @xmath272 .
in particular , @xmath287 .
suppose that @xmath281 is obtained from @xmath270 by a sequence of mutations @xmath275 .
we claim that @xmath288 .
since the mutation is away from @xmath52 , we have that @xmath285 .
by lemma [ l : projmu ] we have that @xmath289 [ p : subalgebra ] suppose that @xmath52 is frozen or @xmath36 has full rank .
let @xmath256 and @xmath290 , then we have that @xmath291 . by lemma [ l : reduce2frozen ]
, we only need to consider the case when @xmath52 is frozen .
we first show that @xmath292 .
let @xmath293 be any extended cluster of @xmath102 .
since @xmath52 is frozen , @xmath294 is an extended cluster of @xmath282 by lemma [ l : projmu ] . for any homogeneous @xmath295 , we write @xmath262 as a laurent polynomial @xmath296 , which is equal to @xmath297 .
let @xmath298 be another extended cluster , and we write @xmath262 as a laurent polynomial @xmath299 . by the definition of @xmath62
, we see that @xmath300 and @xmath301 differ by a factor of some power of @xmath52 .
so @xmath262 lies in the image @xmath302 , which is @xmath303 .
now let @xmath304 be homogeneous .
let @xmath305 be any extended cluster of @xmath306 .
choose an extended cluster @xmath283 as in corollary [ c : liftcluster ] , and then we can write @xmath65 as a laurent polynomial @xmath300 .
we apply the map @xmath307 to @xmath308 , and then @xmath309 is a laurent polynomial in @xmath281 .
we thus obtain the other containment @xmath310 .
[ r : cluster ] we saw from corollary [ c : liftcluster ] that @xmath311 .
however , we have examples to show that under the assumption of proposition [ p : subalgebra ] the inclusion can be strict .
we define a linear function @xmath312 by @xmath313 we define a linear isomorphism @xmath314 by @xmath315 by construction the map @xmath316 is weight - preserving , that is , @xmath317 , and satisfies that @xmath318 ( @xmath212 as defined before definition [ d : projector ] ) .
let @xmath319 be the matrix of the linear transform @xmath316 .
[ l : projb ] if @xmath52 is frozen , then @xmath320 . so we have that @xmath321 and @xmath322 , where @xmath323 and @xmath324 are the @xmath325-variables associated to @xmath102 and @xmath306 .
it suffices to show that the @xmath52-th columns are equal , or equivalently @xmath326 since @xmath327 , we have that @xmath328 .
the required identity is obtained by taking @xmath329 on the both sides . by definition , @xmath330 , where @xmath331 is the @xmath84-th row of @xmath332 .
applying @xmath307 to the above equation , we get that @xmath322 .
let @xmath333 be the linear map defined by forgetting the @xmath52-th coordinate .
then it is a left inverse of @xmath316 and satisfies that @xmath334 .
we say that an upper cluster algebra @xmath17 is _
@xmath144-indexed _ if it has a basis whose elements have well - defined distinct @xmath144-vectors in @xmath161 ( see section [ s : gv ] ) .
[ c : gproj_e ] suppose that @xmath52 is frozen or @xmath36 has full rank .
if @xmath194 is @xmath144-indexed , then @xmath335 has a basis with well - defined @xmath144-vectors in @xmath336 . moreover , if @xmath52 is a projector , then @xmath335 is @xmath144-indexed as well .
suppose that a basis element @xmath142 in @xmath39 can be written as .
then by lemma [ l : projb ] , @xmath337 . by proposition [ p : subalgebra ]
, @xmath54 has a basis with well - defined @xmath144-vectors in @xmath336 .
now suppose that @xmath52 is a projector , then @xmath338 .
we remain to show that @xmath144-vectors can be chosen to be distinct for a basis of @xmath54 .
since elements with well - defined @xmath144-vectors are homogeneous , @xmath339 has a basis with @xmath144-vectors in @xmath340 .
it suffices to show that @xmath341 is injective restricted to @xmath340 . if @xmath342 , then @xmath343 for all @xmath280 . but @xmath344 .
the last statement follows . in general , the cases when @xmath13 is not necessary in @xmath345 is much more complicated .
we can only obtain similar result after imposing more restriction .
however , in what follows we need not assume @xmath13 to be real , so we use @xmath10 instead of @xmath13 .
[ d : attached ] a set of vertices @xmath46 is called _ attached _ to another set of vertices @xmath18 if @xmath346 ( resp .
@xmath347 ) whenever @xmath348 ( resp .
@xmath349 ) .
for such a pair @xmath350 , we form a quiver @xmath351 by deleting @xmath18 and freezing @xmath46 .
let @xmath18 be a set of vertices .
we write @xmath352 for the set of cluster variables @xmath353 in the cluster @xmath104 .
we also write @xmath354 for the restriction of @xmath16 to @xmath355 .
[ p : outerproj ] let @xmath256 with @xmath36 of full rank .
suppose that @xmath10 is a weight such that @xmath356 for all @xmath42 except for a set of frozen vertices @xmath18 , where all @xmath357 @xmath358or all @xmath359 .
then the graded subalgebra @xmath360 is equal to @xmath361 , where @xmath46 is attached to @xmath18 .
the same argument as lemma [ l : reduce2frozen ] shows that @xmath259 is contained in @xmath12 , and thus in @xmath360 .
indeed , by theorem [ t : bounds ] it suffices to show that any @xmath260 can be written as a laurent polynomial in @xmath362 for each @xmath363 .
note that @xmath262 can be written as a laurent polynomial in @xmath104 , which is polynomial in @xmath364 s .
let @xmath244 be the exchange polynomial @xmath365 so that @xmath366 .
we thus get the required laurent polynomial in @xmath265 . conversely , for any homogeneous @xmath367 , we can write @xmath65 as a laurent polynomial @xmath80 in any given extended cluster @xmath368 with polynomial coefficient variables .
we assume that @xmath369 is obtained from @xmath195 by mutations away from @xmath46 .
it is clear that any cluster mutated from the seed @xmath370 is contained in some @xmath283 .
since @xmath46 is attached to @xmath18 , we still have that @xmath371 for all @xmath372 and @xmath357 ( or @xmath44 ) for @xmath373 .
then we see that any coefficient variables in @xmath374 can not appear in @xmath80 because otherwise @xmath375 .
we remain to show that @xmath80 is polynomial in @xmath364 for @xmath363 .
we write @xmath65 as a laurent polynomial @xmath376 in the mutated cluster @xmath362 .
since @xmath46 is attached to @xmath18 , the exchange polynomial @xmath244 contains variables in @xmath374 .
in particular , @xmath377 has the same sign as @xmath378 where @xmath379 .
we decompose @xmath376 as a sum of ( homogeneous ) laurent monomials with no cancelations .
we see that @xmath380 must have non - positive degree in each such monomial because otherwise @xmath375 .
but @xmath381 so @xmath80 must be polynomial in @xmath364 . similar to remark [ r : cr1 ] , we can replace the full rank condition by the condition that @xmath12 is noetherian and @xmath102 is a cr1 seed in @xmath12 . then by @xmath370 is also a cr1 seed in @xmath12 , so by lemma [ l : rca ] we still have that @xmath269 .
the same remark applies to several results depending on proposition [ p : outerproj ] , such as corollary [ c : gprojv ] , theorem [ t : removal ] , theorem [ t : hat ] , and corollary [ c : removal_model ] .
analogous to remark [ r : cluster ] , under the assumption of proposition [ p : outerproj ] , we only have @xmath382 in general . in the situation of proposition [ p : outerproj ] ,
if we define @xmath383 by evaluation of @xmath374 at @xmath384 , then @xmath385 is a left inverse of the embedding @xmath386 .
similar to lemma [ l : projb ] , we have that @xmath387 and @xmath388 , where @xmath324 is the @xmath325-variable for @xmath370 .
let @xmath389 and @xmath390 be the linear maps such that @xmath391 and @xmath392 .
the proof of the following corollary is similar to ( but easier than ) that of corollary [ c : gproj_e ] .
[ c : gprojv ] suppose that we are in the situation of proposition [ p : outerproj ] .
if @xmath39 is @xmath144-indexed , then @xmath56 is @xmath144-indexed as well , and moreover we have that @xmath393 .
let us briefly recall the semi - invariant rings of quiver representations @xcite . let @xmath2 be a finite quiver without oriented cycles .
we fix an algebraically closed field @xmath72 of characteristic zero . for a dimension vector @xmath1 of @xmath2 ,
let @xmath394 be a @xmath1-dimensional vector space @xmath395 .
we write @xmath396 for the @xmath397-th component of @xmath394 .
the space of all @xmath1-dimensional representations is @xmath398 the product of general linear group @xmath399 acts on @xmath22 by the natural base change .
define @xmath400 by @xmath401 we are interested in the rings of semi - invariants @xmath402^{{\operatorname{sl}}_\beta}.\ ] ] the ring @xmath0 has a weight space decomposition @xmath403 where @xmath28 runs through the multiplicative _ characters _ of @xmath404 .
we refer to such a decomposition the @xmath28-grading of @xmath0 .
recall that any character @xmath405 can be identified with a weight vector @xmath406 @xmath407 let us understand these multihomogeneous components @xmath408\mid g(f)=\sigma(g)f , \forall g\in{\operatorname{gl}}_\beta \}.\ ] ] since @xmath2 has no oriented cycles , the degree zero component is the field @xmath72 @xcite . for any projective presentation @xmath409
, we view it as an element in the homotopy category @xmath410 of bounded complexes of projective representations of @xmath2 .
we view each weight @xmath406 as an element in the grothendieck group @xmath411 . concretely , suppose that @xmath412 and @xmath413 for @xmath414 , then @xmath415 . here
, we use the notation @xmath416 for @xmath417 , where @xmath418 is the indecomposable projective representation corresponding to the vertex @xmath397 .
let @xmath28 be a weight such that @xmath419 , and @xmath219 be a projective presentation of weight @xmath28
. in @xcite schofield constructed for each such @xmath219 a ( not necessary non - zero ) element in @xmath27 .
for the construction of @xmath420 , we refer the readers to the original text or ( * ? ? ?
* section 1 ) .
in fact , [ t : inv_span ] @xmath420 s span @xmath421 over the base field @xmath72 .
let @xmath422 be the set of all weights @xmath28 such that @xmath27 is non - empty .
it is known that such a weight @xmath28 corresponds to some dimension vector @xmath423 , i.e. , @xmath424 , where @xmath425 is the _ euler form _ of @xmath2 .
the euler form induces a form @xmath26 on the space of weights satisfying @xmath426 .
the unimodular lattice @xmath25 with this form will be used in our weight configuration later .
the next theorem is an easy consequence of king s stability criterion ( * ? ? ?
* proposition 3.1 ) and theorem [ t : inv_span ] . by a general @xmath1-dimensional representation , we mean in a sufficiently small zariski open subset ( sufficient " here depends on the context ) .
following @xcite , we use the notation @xmath427 to mean that a general @xmath1-dimensional representation has a @xmath428-dimensional subrepresentation .
* theorem 3 ) [ t : cone ] we have @xmath429 in particular , @xmath422 is a pointed saturated monoid .
recall that a dimension vector @xmath239 is called a _
root if @xmath430 .
it is called _ schur _ if @xmath431 for @xmath75 general in @xmath432 . for a real schur root @xmath239
, @xmath433 has a dense orbit for any @xmath434 .
[ l : gsreal ] if @xmath239 is a real schur root , then @xmath435 for any @xmath434 .
* lemma 1 ) [ l : exact ] suppose that we have an exact sequence of representations of @xmath2 @xmath436 if @xmath437 , then as a function on @xmath22 , @xmath438 is , up to a scalar , equal to @xmath439 , where @xmath440 is the minimal presentation of @xmath75 .
if @xmath441 or @xmath442 , then @xmath443 . for two dimension vector @xmath444 , we write @xmath445 if @xmath446 for @xmath447 general in @xmath432 and @xmath22 .
we recall that an _ exceptional sequence of dimension vector _ @xmath448 is a sequence of real schur roots of @xmath2 such that @xmath449 for any @xmath450 .
it is called _ quiver _ if @xmath451 for any @xmath450 .
it is called _ complete _ if @xmath452 .
the _ quiver _
@xmath453 of a quiver exceptional sequence @xmath454 is by definition the quiver with vertices labeled by @xmath455 and @xmath456 arrows from @xmath457 to @xmath455 . according to ( * ? ? ?
* theorem 4.1 ) , @xmath458 . given an exceptional sequence of dimension vector @xmath459 with @xmath460 general in @xmath461
, we consider the ( right ) _ orthogonal subcategory _
@xmath462 defined by @xmath463 according to @xcite , this abelian subcategory is equivalent to the module category of another quiver denoted by @xmath464 .
we say that @xmath2 _ projects _ to @xmath464 through @xmath465 . in this case
there is a unique quiver exceptional sequence @xmath466 of @xmath2 such that the concatenation of two sequences @xmath467 is a ( complete ) exceptional sequence and @xmath468 .
we refer to @xcite for the construction of @xmath454 .
we define a linear isometry ( with respect to the euler forms ) @xmath469 by @xmath470 conversely we define a linear map @xmath471 left inverse to @xmath472 as the composition @xmath473 , where the first map @xmath474 is given by the recursion @xmath475 both maps induce a linear map on the weights , still denoted by @xmath472 and @xmath471 .
we write @xmath476 and @xmath477 for @xmath478 and @xmath479 .
to effectively compute @xmath477 later , we can use the formulae @xmath480 [ t : siemb ] let @xmath13 be a real schur root such that @xmath481 .
then we have that @xmath482 .
[ p : extremal_proj ] let @xmath13 is a real root of @xmath2
. then @xmath483 is extremal in @xmath484 if and only if @xmath485 is a projector for @xmath0 . by ( * ?
* theorem 3.17 ) , any @xmath0 is a ufd .
since @xmath486 and @xmath422 is pointed , the direction @xmath487 follows from proposition [ p : projector].(3 ) .
conversely , let us first show that @xmath13 is in fact schur .
if not , then a general representation @xmath5 in @xmath488 decomposes as @xmath489 . by lemma
[ l : exact ] , we must have that @xmath490 .
since @xmath483 is extremal , the weights of @xmath491 and @xmath492 must lie on @xmath493 .
but @xmath13 is real , so it is clearly impossible . for any dimension vector @xmath239 such that @xmath494 and @xmath495
, we must have that @xmath496 where @xmath497 .
indeed , let @xmath428 be the rank of a general morphism @xmath219 from a general representation @xmath498 to a general representation @xmath499 .
note that @xmath500 .
we claim that @xmath501 .
we first observe that @xmath428 must be a multiple of @xmath13 , say @xmath502 . otherwise by lemma
[ l : exact ] the weights of @xmath503 and @xmath504 lie in @xmath484 contradicting @xmath485 being extremal .
now we have an exact sequence @xmath505 where @xmath5 is the exceptional representation of dimension @xmath13 . by applying @xmath506
, we find that @xmath507 .
since @xmath13 is real schur , a general representation of dimension @xmath508 is isomorphic to @xmath509 .
if @xmath510 , then there is a nontrivial homomorphism from @xmath511 to @xmath512 .
we can thus construct a homomorphism from @xmath513 to @xmath394 with image strictly containing @xmath514 .
so we must have that @xmath515 .
now by theorem [ t : inv_span ] , any semi - invariant function of weight @xmath516 is a linear combination of @xmath420 s , where @xmath219 is the minimal presentation of a general representation in @xmath432 . by lemma [ l : exact ] , any such @xmath420 has a factor @xmath517 in @xmath0 .
we conclude that @xmath485 is a projector from lemma [ l : gsreal ] and [ l : projector ] .
we say that a semi - invariant ring @xmath0 is a _ naturally graded _
upper cluster algebra @xmath35 if @xmath518 and @xmath186 is the @xmath28-weight of @xmath519 under the isomorphism .
[ t : invguca ] suppose that @xmath0 is a naturally graded uca @xmath35 , and @xmath36 has full rank . if @xmath520 is real and extremal in @xmath484 , then @xmath13 is schur and @xmath521 is the naturally graded uca @xmath522 . from the proof of proposition [ p : extremal_proj ] , we have seen that @xmath13 must be schur .
the rest follows from proposition [ p : subalgebra ] , [ p : extremal_proj ] , and theorem [ t : siemb ] . in what follows
, we will always identify @xmath523 with @xmath524 through the linear isometry @xmath525 if available .
so we may keep @xmath526 instead of writing @xmath522 .
we note that if @xmath527 is an exceptional sequence with each @xmath528 extremal in @xmath484 , then each @xmath529 is extremal in @xmath530 .
so theorem [ t : invguca ] can be inductively applied to such an exceptional sequence .
we first recall the main result in @xcite .
consider the triple flag quiver @xmath531 @xmath532 with the _ standard _ dimension vector @xmath533 as indicated above .
let @xmath534 be the unit vector supported on the @xmath397-th vertex of the @xmath65-th arm of @xmath531 .
let @xmath535 be the weight given by @xmath536 for @xmath537 , and @xmath538 .
we use the convention that @xmath539 is the zero vector and @xmath540 for @xmath541 .
it is clear that each @xmath535 is extremal in @xmath542 .
it is proved in @xcite that the semi - invariant ring @xmath543 is the guca @xmath544 , where @xmath545 is the ice hive quiver and @xmath546 .
the _ ice hive quiver @xmath545 of size @xmath77 _ is an ice quiver in the plane with @xmath547 vertices arranged in a triangular grid .
we label the vertices as shown in figure [ fig : hive ] .
note that three vertices @xmath548 and @xmath549 are missing , and all boundary vertices are frozen .
there are three types of arrows : @xmath550 , and @xmath551 .
it is easy to see that @xmath552 has full rank .
@xmath553 [ ex : cflag ] consider the exceptional sequence @xmath554 of @xmath531 .
note that @xmath555 for @xmath556 so the order of @xmath465 does not matter .
we find that the unique quiver exceptional sequence @xmath454 completing @xmath465 is given by @xmath557 where @xmath558 .
the quiver of @xmath454 is displayed below , where @xmath559 and @xmath560 corresponds to the vertices @xmath561 , and @xmath562 respectively .
@xmath563 we denote the quiver of @xmath454 by @xmath564 .
let @xmath565 be the dimension vector given by @xmath566 and @xmath567 .
we can verify by that under the embedding the dimension vector @xmath565 goes to @xmath533 .
it is not hard to see that @xmath568 is the _ total coordinate ring _
@xmath569 of the complete flag variety @xmath570 of the @xmath77-dimensional vector space .
let @xmath571 be the divisor class group of a variety @xmath572 . by definition , @xmath573\in{\operatorname{cl}}(x)}\gamma(x , o_x(d)),\ ] ] where the multiplication is given by multiplying homogeneous sections in the field of rational functions @xmath574 . however , to rigorously prove this fact , we need _ variation of git_. we will write this up in some lecture notes . we observe that the exceptional sequence @xmath465 corresponds to the last row of frozen vertices of @xmath545 .
let @xmath575 be the ice quiver obtained from @xmath545 by deleting the last row of frozen vertices ( see figure [ fig : flats ] ) .
0.3 @xmath576 [ fig : hivef ] 0.3 @xmath577 [ fig : hiveff ] + it follows from theorem [ t : invguca ] that the total coordinate ring of @xmath570 is the guca @xmath578 . here
, the new weight configuration @xmath579 can be computed from .
for example , @xmath580 , which is clearly extremal in @xmath581 . [ ex : ucell ] we sketch how to get the cluster structure of the maximal unipotent subgroup @xmath582 of @xmath59 from the previous example , and leave the details to the readers .
although as an ungraded algebra it is just a polynomial ring , it is certainly non - trivial if the multigrading is taken into account .
we verify that each @xmath583 corresponds to a real schur root , denoted by @xmath584 .
in fact , @xmath585 is an exceptional sequence of @xmath564 .
we find that the unique quiver exceptional sequence @xmath454 completing @xmath465 is given by @xmath586 where @xmath587 .
then we can verify that the arrow matrix of @xmath453 is a strict upper - triangular matrix with all entries equal to one .
let @xmath588 be the dimension vector @xmath589 .
we can verify by that under the embedding the dimension vector @xmath588 goes to @xmath565 .
clearly @xmath590 is the coordinate ring of @xmath582 .
let @xmath591 be the ice quiver obtained from @xmath575 by deleting the right ( or left ) row of frozen vertices .
it follows that the coordinate ring of @xmath582 is the guca @xmath592 .
we want to tell curious readers that we can not project from any vertex of @xmath591 except for @xmath593 because none of the weights in @xmath594 corresponds to a real schur root .
[ ex:6flag ] consider the exceptional sequence @xmath595 of @xmath596 .
we find that the unique quiver exceptional sequence @xmath454 completing @xmath465 is given by @xmath597 ( the order of @xmath65 does not matter ) , where @xmath598 and @xmath599 .
the quiver of @xmath454 is the sextuple flag quiver , where @xmath534 ( @xmath600 ) and @xmath601 ( @xmath600 ) correspond to the vertices on the left and right three flags respectively , and @xmath13 corresponds to the central vertex @xmath77 .
@xmath602 we call such a quiver sextuple flag quiver , denoted by @xmath603 .
let @xmath604 be the dimension vector indicated as above .
we can verify by that under the embedding the dimension vector @xmath604 goes to @xmath605 of @xmath596 .
let @xmath606 be the ice quiver obtained from @xmath607 by deleting the vertices corresponding to @xmath465 .
@xmath608 it follows from theorem [ t : invguca ] that the semi - invariant ring @xmath609 is the guca @xmath610 . here
, @xmath611 can also be easily obtained from .
since it will appear in proposition [ p : multiplelr ] , we list it for readers convenience .
let @xmath612 such that @xmath613 , and similar for @xmath614 and @xmath537 , then @xmath615 let @xmath2 be a quiver , and @xmath1 a dimension vector of @xmath2
. for a fixed vertex @xmath616 , let @xmath617 be the quiver such that @xmath618 , and @xmath619\mid h(a)=t(b)=r\},\ ] ] where @xmath620)=t(a)$ ] and @xmath621)=h(b)$ ] . the _ vertex removal
_ at @xmath40 on @xmath22 is the algebraic morphism @xmath622 sending @xmath75 to @xmath623 defined by @xmath624 ) & = m(ab ) & \text{if } & t(a)\xrightarrow{a } r \xrightarrow{b } h(b).\end{aligned}\ ] ] [ l : projsi ] assume the following condition on @xmath1 @xmath625 the vertex removal @xmath626 induces an embedding @xmath627 mapping @xmath628 onto @xmath629 , where @xmath630 runs through all weights not supported on @xmath40 .
it suffices to show that @xmath631 .
let @xmath632 and @xmath633 . by the first and second fundamental theorem of invariant theory of @xmath634
, we can identify @xmath635 with the rank at most @xmath636 maps in @xmath637 .
but due to our assumption , all maps in @xmath637 have rank at most @xmath636 . by definition , @xmath638^{{\operatorname{sl}}_{{\overline{\beta}}}\times { \operatorname{gl}}(m(r))}$ ] , which is equal to @xmath639^{{\operatorname{sl}}_{{\overline{\beta}}}}$ ] . in other words ,
the @xmath630 in the above lemma is in @xmath9 , where @xmath10 is the weight corresponding to the unit vector @xmath640 .
it follows from proposition [ p : outerproj ] and lemma [ l : projsi ] that [ t : removal ] suppose that @xmath0 is a naturally graded uca @xmath39 , and @xmath36 has full rank .
let @xmath40 be a vertex of @xmath2 satisfying and that @xmath41 for all @xmath42 except for a set of frozen vertices @xmath18 , where all @xmath43 ( or all @xmath44 ) .
then we have that @xmath628 is the naturally graded uca @xmath56 , where @xmath46 is attached to @xmath18 .
finally , we remark that proposition [ p : outerproj ] can also be used to remove arrows .
the outer automorphism group @xmath641 of the quiver @xmath2 is by definition @xmath642 , where @xmath643 is the linear space spanned by arrows from @xmath397 to @xmath614 . in particular , for each arrow @xmath65 of @xmath2 , there is an associated torus @xmath644 in @xmath641 .
since the action of @xmath641 commutes with the action of @xmath404 , the torus @xmath645 induces another grading of the semi - invariant ring .
if the cluster in a seed is homogeneous with respect to the grading induced by @xmath644 , then we can use proposition [ p : outerproj ] to remove the arrow @xmath65 .
we are able to find the cluster structure of total coordinate ring of any partial flag varieties in type @xmath200 .
since the divisor class group of a grassmannian is @xmath169 , its total coordinate ring is just the usual coordinate ring .
hence we should be able to recover the result in @xcite as well .
we can start from the result in example [ ex : cflag ] , then remove some vertices on the long arm of @xmath564 depending on the particular flag variety .
the following hint may be useful .
let @xmath646 for @xmath647 .
we define a sequence of vertices @xmath648 .
this contains all mutable vertices of @xmath649-th row of @xmath575 .
if we want to remove all vertices in @xmath650 , we first apply a sequence mutation in the following order @xmath651 then we are able to apply theorem [ t : removal ] .
to get the cluster structure of complete quintuple flags , we start with example [ ex:6flag ] , then we remove all vertices in a flag in the order @xmath652 .
remove vertices in another flag , and we get a complete quadruple flag .
we shall find that the semi - invariant ring @xmath653 is the guca @xmath654 ; the semi - invariant ring @xmath655 is the guca @xmath656 .
we are not going to give the precise construction of the quivers @xmath657 and @xmath658 because it is rather clear .
when @xmath659 , the quiver @xmath660 and @xmath661 is displayed below .
0.3 @xmath662 0.3 @xmath663 + the quiver @xmath664 can be projected from a triple flag quiver for any @xmath665 .
the semi - invariant ring @xmath666 of any complete @xmath60-tuple flag is an upper cluster algebra .
what we will do in the rest of this subsection is to illustrate in a single example on how to inductively apply the vertex removal technique ( theorem [ t : removal ] ) to find possible cluster structure .
if the method described below fails , we believe that the cluster structure may not exist .
however , the method is not deterministic . *
the method : * suppose that @xmath0 is a guca @xmath39 , and we want to remove a vertex @xmath40 of @xmath2 satisfying let us assume that @xmath1 is a schur root , so the stabilizer in general position of the action is zero - dimensional
. then the krull dimension of @xmath0 is given by @xmath667 we first compute the difference @xmath668 of @xmath669 before and after the vertex removal .
if theorem [ t : removal ] works , then @xmath670 should vanish for all but exactly @xmath668 frozen vertices , where @xmath43 ( or @xmath44 ) .
these @xmath668 vertices are our choice of @xmath18 .
otherwise the method fails , or the cluster structure may not exist for @xmath628 . next we need to seek a suitable sequence of mutations such that after application @xmath671 for all mutable @xmath84 .
if no such sequence exists , then the method fails , or the cluster structure may not exist . if succeeds , we take @xmath351 , where @xmath46 is attached to @xmath18 .
[ ex : t335 ] in this example , we explain how to find a cluster algebra structure of the semi - invariant ring @xmath672 , where @xmath673 is obtained from @xmath674 by removing the @xmath675 s of all three arms , and the @xmath676 s and @xmath677 s of the first two arms .
we also explain why the cluster algebra structure may not exist if we further forget either @xmath676 or @xmath677 of the third arm .
we suggest that readers play this example with b. keller s applet @xcite
. we can directly apply theorem [ t : removal ] to remove all @xmath675 s and one @xmath677 , say of the first arm .
the results correspond to the figure , , , and .
we apply another mutation at @xmath678 to make the quiver easier to draw .
this step is optional .
next , to remove @xmath676 of the first arm , we first mutate at @xmath679 and @xmath680 , then we find that @xmath681 and @xmath682 as required .
so we delete @xmath18 and freeze @xmath46 , and perform an optional mutation at @xmath679 .
similarly , to remove @xmath677 of the second arm , we mutate at @xmath679 and @xmath683 , then we find that @xmath684 and @xmath685 .
so we delete @xmath18 and freeze @xmath46 , and perform an optional mutation at @xmath679 . finally to remove @xmath676 of the first arm ,
we mutate at @xmath679 and @xmath686 , then we find that @xmath687 and @xmath688 .
so we delete @xmath18 and freeze @xmath46 , and perform an optional mutation at @xmath689 .
the results correspond to the figure , , and .
the interested readers can verify using @xcite that the types of the cluster algebras are affine @xmath690 .
0.3 @xmath691 0.3 @xmath692 + 0.3 @xmath693 0.3 @xmath694 + 0.3 @xmath695 0.3 @xmath696 + 0.3 @xmath697 for the last one , the cluster variable associated to a vertex @xmath89 is semi - invariant of weight @xmath698 , where @xmath699 as before and @xmath700 one can verify that they all correspond to all real schur roots .
the semi - invariant ring @xmath672 is this graded upper cluster algebra .
however , if we want to further remove @xmath677 or @xmath676 of the third arm , then we are unable to choose the set @xmath18 because the expected cardinality of @xmath18 is @xmath701 .
even worse , no matter how we mutate from , there are at least @xmath677 vertices @xmath89 such that @xmath702 .
so we suspect that it may not be an upper cluster algebra .
in @xcite and @xcite , the mutation of quivers with potentials is invented to model the cluster algebras .
we emphasis here that ice quivers considered here can have arrows between frozen vertices .
following @xcite , we define a potential @xmath512 on an ice quiver @xmath6 as a ( possibly infinite ) linear combination of oriented cycles in @xmath6 .
more precisely , a _
potential _ is an element of the _ trace space _
@xmath703 $ ] , where @xmath704 is the completion of the path algebra @xmath705 and @xmath706 $ ] is the closure of the commutator subspace of @xmath704 .
the pair @xmath47 is an _ ice quiver with potential _ , or iqp for short . for each arrow @xmath707 , the _ cyclic derivative _
@xmath708 on @xmath709 is defined to be the linear extension of @xmath710 for each potential @xmath512 , its _ jacobian ideal _
@xmath711 is the ( closed two - sided ) ideal in @xmath704 generated by all @xmath712 .
jacobian algebra _
@xmath48 is the quotient algebra @xmath713 .
if @xmath512 is polynomial and @xmath48 is finite - dimensional , then the completion is unnecessary to define @xmath48 .
this is the case throughout this paper .
the key notion introduced in @xcite is the _ mutation _ of quivers with potentials and their decorated representations . for an ice quiver with _ nondegenerate
( see @xcite ) , the mutation in certain sense lifts " the mutation in definition [ d : qmu ] .
since we do not need mutation in an explicit way , we refer readers to the original text .
a _ decorated representation _ of a jacobian algebra @xmath714 is a pair @xmath715 , where @xmath716 , and @xmath717 is a finite - dimensional @xmath718-module .
let @xmath719 be the set of decorated representations of @xmath48 up to isomorphism .
there is a bijection between two additive categories @xmath719 and @xmath720 mapping any representation @xmath75 to its minimal presentation in @xmath721 , and the simple representation @xmath722 of @xmath718 to @xmath723 .
suppose that @xmath724 corresponds to a projective presentation @xmath725 for @xmath726 .
the _ @xmath144-vector _
@xmath727 of a decorated representation @xmath724 is the weight vector of its image in @xmath728 , that is , @xmath729 .
the representation @xmath724 is called _ @xmath144-coherent _ if @xmath730 for all vertices @xmath84 .
[ d : ice ] for an iqp @xmath47 , _ freezing _ a set of vertices @xmath46 is the operation that we set all vertices in @xmath46 to be frozen , but we do not delete arrows between frozen vertices .
we also keep the same potential @xmath512 .
this definition is slightly different from @xcite where we delete arrows between frozen vertices , and thus change potential as well .
this difference can be ignored if we only look at the category of @xmath50-supported representations ( see definition [ d : mu_suprep ] ) .
the reason is that for the two different definitions , the two categories are equivalent and their objects take the same value under the cluster character ( definition [ d : cc ] ) .
however , if we insist on deleting arrows between frozen vertices , that will cause additional trouble in constructing the two functors in section [ s : resind ] .
[ d : mu_suprep ] a decorated representation @xmath724 is called _ @xmath50-supported _ if the supporting vertices of @xmath75 are all mutable .
we denote by @xmath731 the set of all @xmath144-coherent @xmath50-supported decorated representations of @xmath732 .
[ d : cc ] we define the _ cluster character _ @xmath733 by @xmath734 where @xmath735 is the variety parameterizing @xmath736-dimensional quotient representations of @xmath75 , and @xmath737 denotes the topological euler - characteristic .
it turns out ( * ? ? ?
* lemma 5.3 ) that if @xmath738 , then @xmath739 is an element in @xmath126 .
we note that @xmath739 has a well - defined @xmath144-vector , which is @xmath727 .
[ t : cc ] suppose that iqp @xmath47 is non - degenerate and @xmath36 has full rank .
let @xmath740 be a set of @xmath144-coherent @xmath50-supported decorated representations with all distinct @xmath144-vectors , then @xmath741 maps @xmath740 ( bijectively ) to a set of linearly independent elements in the upper cluster algebra @xmath17 .
however , it is not clear to us whether any uca has a basis with distinct @xmath144-vectors . to any @xmath742 we associate the _ reduced _ presentation space @xmath743_+),p([-{{\sf g}}]_+)),\ ] ]
where @xmath744_+$ ] is the vector satisfying @xmath744_+(u ) = \max({{\sf g}}(u),0)$ ] .
we denote by @xmath745 the cokernel of a general presentation in @xmath746 .
reader should be aware that @xmath745 is just a notation rather than a specific representation .
if we write @xmath747 , this simply means that we take a general presentation in @xmath746 , then let @xmath75 to be its cokernel . [
d : mu_supg ] a weight vector @xmath748 is called _ @xmath50-supported _ if @xmath745 is @xmath50-supported .
let @xmath49 be the set of all @xmath50-supported vectors in @xmath749 .
we define the _ generic character _ @xmath750 by @xmath751 it is known ( * ? ? ?
* lemma 5.11 ) that a general presentation in @xmath746 corresponds to a @xmath144-coherent decorated representation .
so the image of @xmath752 is also contained in @xmath753 . in particular , we have that @xmath754 .
[ r : basis ] assume that we are in the situation of theorem [ t : cc ] .
moreover , we assume that @xmath17 is @xmath16-graded with finite - dimensional graded components . let @xmath755 be the set of all @xmath144-vectors in @xmath740 .
if @xmath756 forms a basis of @xmath17 , then so does @xmath757 by lemma [ l : independent ] .
[ d : model ] we say that an iqp @xmath47 _ models _ an algebra @xmath12 if the generic cluster character maps @xmath49 onto a basis of @xmath12 .
if @xmath12 is the upper cluster algebra @xmath17 , then we simply say that @xmath47 is a _
cluster model_. if in addition @xmath49 is given by lattice points in some polyhedron , then we say that the model is _
polyhedral_. in this definition we do not require the iqp to be non - degenerate . if @xmath17 is @xmath16-graded with finite - dimensional graded components , then by remark [ r : basis ] @xmath47 being a cluster model is equivalent to that @xmath17 is @xmath144-indexed and @xmath758 .
for an iqp @xmath47 , _ deleting _ a set of vertices @xmath759 is the operation that we delete all vertices in @xmath18 for @xmath6 , and set all incoming and outgoing arrows of @xmath373 to be zero in @xmath512 .
we denote the new iqp by @xmath760 .
the operation of deleting @xmath18 is the same as _ restricting _ @xmath47 to the vertex set @xmath355 as in ( * ? ? ?
* definition 8.8 ) . in
what follows , we shall mainly consider the case when @xmath18 is a single vertex @xmath52
. the general case can be easily treated by induction . according to (
* proposition 8.9 ) , the restriction induces an algebra epimorphism @xmath761 .
we denote @xmath762 .
let @xmath763 be the restriction of scalar functor induced by @xmath80 . concretely ,
if @xmath764 , then the extension of @xmath765 by zeros to the quiver @xmath6 is clearly a representation of @xmath732 . such a representation is @xmath766 .
conversely , we denote the functor @xmath767 by @xmath768 .
it is well - known @xcite that the functor @xmath768 is left adjoint to the exact functor @xmath769 .
we extend the functor @xmath769 to decorated representations by copying the decorated spaces .
we denote such a map still by @xmath770 . the functor @xmath768 can be naturally extended .
for @xmath771 , we define @xmath772 as @xmath773 .
we note that @xmath774 but @xmath775 is the indecomposable projective @xmath776 in @xmath777 .
so @xmath778 , where @xmath341 is the linear map forgetting the @xmath52-th coordinate .
recall from ( * ? ? ?
* ( 1.13 ) ) that @xmath781 we refer readers to ( * ? ? ?
* section 10 ) for the definition of the map @xmath782 . comparing @xmath724 and @xmath779
, we find that there is no changes on the three terms on the right for @xmath280 .
the last statement is also clear from the equation .
since the composition is right exact , it suffices to show that it is an identity on the indecomposable projective representations . by lemma [ l : gdiff ] ,
for any @xmath786 we have the following presentation @xmath787 for some @xmath788 .
we apply the right exact functor @xmath768 to the exact sequence , and get @xmath789 if @xmath52 is frozen , then @xmath790 for some @xmath50-supported @xmath791 .
the last statement follows .
suppose that @xmath792 and @xmath793 .
then we can view @xmath794 as row vectors with entries in @xmath795 , and @xmath796 can be represented by an @xmath797 matrix @xmath798 with @xmath799 a linear combination of paths from @xmath800 to @xmath801 . from this description
, we see that the functor @xmath802 induces a surjective algebraic map @xmath803 . in particular ,
if @xmath219 is general enough in @xmath804 , then both @xmath739 and @xmath805 take the value of the generic character , where @xmath724 is the decorated representation corresponding to @xmath219 .
[ l : commute_e ] we have the following commutative diagram for @xmath52 frozen .
@xmath806^{c } \ar@{->>}[d]_{{\operatorname{ind}}_e } & { { \overline{{\mathcal{c}}}}}(\delta,{\bold{x}};{\boldsymbol{\sigma } } ) \ar@{->>}[d]_{\ell_e } \\ { \mathcal{r}}ep^{{{\sf g}},\mu}(j_e ) \ar@{^(->}[r]^{c\quad } & { { \overline{{\mathcal{c}}}}}(\delta_e,{\bold{x}}_e;{\boldsymbol{\sigma}}_e ) } } \ ] ] the surjectivity of @xmath62 follows from proposition [ p : subalgebra ] . for @xmath738
, we have that @xmath807 for some @xmath808 because @xmath52 is frozen .
recall from lemma [ l : resind ] that @xmath809 . by the definition of @xmath769 , @xmath810 is isomorphic to @xmath811 restricting to the subquiver @xmath190 . by definition @xmath812 . by lemma [ l : projb ]
we have that @xmath813 on the other hand , @xmath814 since @xmath815 , we conclude that @xmath816 .
if @xmath52 is a projector , then we have the embedding @xmath817 .
however , we want to warn readers that another functor @xmath769 and the embedding do not fit into the above commutative diagram in general , i.e. , @xmath818 .
we first assume that @xmath52 is frozen .
we first show that @xmath819 forms a basis of @xmath820 . by assumption
, there is a subset @xmath740 of @xmath821 with distinct @xmath144-vectors in @xmath822 ( defined after remark [ r : cluster ] ) such that @xmath756 forms a basis of @xmath339 .
moreover , by the remark in the end of section [ s : resind ] , each element in @xmath740 can be chosen to take the value of the generic character . since @xmath52 is a projector , @xmath823 by proposition [ p : subalgebra ] .
so @xmath824 forms a basis of @xmath54 . since @xmath778 , by a similar argument as in corollary [ c : gproj_e ] we see that elements in @xmath825 have distinct @xmath144-vectors .
now we come back to the general situation , that is , @xmath52 is not necessary frozen .
we first claim that @xmath826 spans @xmath54 if @xmath827 spans @xmath39 . for any such @xmath828
, we can always find some @xmath60 sufficiently large such that @xmath829 is not supported on @xmath52 , i.e. , @xmath830 . by the definition of @xmath62 ,
the @xmath144-vector of @xmath831 is equal to @xmath832 .
note that this is also the @xmath144-vector of @xmath833 .
since @xmath834 spans @xmath335 , so is @xmath826 by lemma [ l : independent ] .
finally , we can apply the previous conclusion to @xmath835 .
the quiver @xmath545 can be equipped with a _ rigid _ potential @xmath838 ( see @xcite for detail ) .
it is proved in @xcite that @xmath839 models the semi - invariant ring @xmath543 .
moreover , @xmath840 is given by the lattice points in the hive cone @xmath841 defined as follows .
a maximal straight path in @xmath545 must start from a frozen vertex and end in another frozen vertex , eg .
, from @xmath842 to @xmath843 .
we define @xmath841 by @xmath844 here , subpaths include all trivial paths @xmath845 for @xmath89 frozen , and the notation @xmath846 stand for @xmath847 if @xmath80 is the path @xmath848 . given a weight configuration @xmath16 of a quiver @xmath6 and a convex polyhedral cone @xmath849 , we define the ( not necessarily bounded ) convex polytope @xmath850 as @xmath851 cut out by the hyperplane sections @xmath852 .
it follows from knutson - tao @xcite and derksen - weyman @xcite ( see @xcite ) that the lattice points in @xmath853 counts the littlewood - richardson coefficient @xmath854 .
we refer readers to @xcite for the definition of the partitions @xmath855 , and @xmath856 . by corollary [ c : model_proj ] , @xmath857 is given by lattice points in some polyhedral cone denoted by @xmath858 . similar to the calculation in (
* ? ? ? * section 6 ) , we find that the cone @xmath858 can also be described by . by a calculation similar to (
* ? ? ? * proposition 1 ) , we find that the dimension of each graded piece @xmath859 counts the _ kostka number _ @xmath860 , which is by definition the multiplicity of the irreducible complex representation @xmath861 of @xmath59 in the tensor product @xmath862 .
the partitions @xmath863 and the weight @xmath50 are determined by @xmath28 as follows @xmath864 alternatively , one can see this from the identification in theorem [ t : siemb ] . under the identification @xmath865
, the kostka coefficient can be interpreted as certain littlewood - richardson coefficient .
this interpretation is well - known . by corollary [ c : model_proj ] , @xmath868 is given by lattice points in some polyhedral cone denoted by @xmath869 .
since the semi - invariant ring is just a polynomial ring in @xmath870 variables , we see that the cone is generated by @xmath765 extremal rays .
moreover , the cone @xmath869 can also be described by , and we denote this hyperplane representation by @xmath871 . the cone is related to _
kostant s vector partition function _
@xcite in type @xmath200 as follows .
recall that @xmath873 counts the lattice points in the polytope @xmath874 where @xmath875 is the matrix form by rows of positive roots in type @xmath876 .
it is easy to verify by linear algebra that @xmath877 , so we have that @xmath878 , where @xmath879 if @xmath880 .
since @xmath881 is _ totally unimodular _
, the two polytope have the same number of lattice points .
recall that kostant s formulae @xcite @xmath882 relates any kostka number to an alternating sum of values in the partition function , and steinberg s formulae @xcite @xmath883 relates any littlewood - richardson coefficient to an alternating sum of kostka numbers . here
@xmath884 is the dot action of @xmath885 defined by @xmath886 for @xmath887 the half sum of positive roots in type @xmath200 .
can we see these two formula from the relations among polytopes @xmath888 , and @xmath841 ? by corollary
[ c : model_proj ] , @xmath889 is given by lattice points in some polyhedral cone denoted by @xmath890 . by a similar calculation in (
* ? ? ? * section 6 ) , we find that the cone @xmath891 can also be described by .
similar to ( * ? ? ?
* proposition 1 ) , the dimension of each graded piece @xmath609 counts certain _ multiple _ littlewood - richardson coefficients @xmath892 ( ( * ? ? ?
* section 1 ) ) , where @xmath893 and @xmath863 are determined by the weight @xmath28 similar to @xcite .
suppose that @xmath46 is attached to @xmath18 , and @xmath195 is a graded seed .
if we restrict the seed to @xmath351 , we get a new graded seed @xmath897 as in section [ s : notin ] .
denote @xmath898 . recall that in the situation of proposition [ p : outerproj ]
, we have an epimorphism @xmath899 .
the proof of the following lemma , theorem , and corollary is completely analogous to ( but easier than ) that of lemma [
l : commute_e ] , theorem [ t : frozen_model ] , and corollary [ c : model_proj ] , so we leave the details to the readers .
we have the following commutative diagram under the assumption of proposition [ p : outerproj ] .
@xmath806^{c } \ar@{->>}[d]_{{\operatorname{res}}_{{\bold{v } } } } & { { \overline{{\mathcal{c}}}}}(\delta,{\bold{x}};{\boldsymbol{\sigma } } ) \ar@{->>}[d]_{p_{{\bold{v } } } } \\ { \mathcal{r}}ep^{{{\sf g}},\mu}(j_{{\bold{v } } } ) \ar@{^(->}[r]^{c\quad } & { { \overline{{\mathcal{c}}}}}({\delta_{{\bold{v}}}^{{\bold{u}}}},{\bold{x}}({\widehat{{\bold{v}}}});{\boldsymbol{\sigma}}({\widehat{{\bold{v } } } } ) ) } } \ ] ] 19 m. f. atiyah , i. g. macdonald , _ introduction to commutative algebra , _ addison - wesley publishing co. , reading , mass .- london - don mills , ont .
1969 ix+128 pp .
a. berenstein , s. fomin , a. zelevinsky , _ cluster algebras .
iii . upper bounds and double bruhat cells , _ duke math . j. 126 ( 2005 ) , no .
1 , 152 . h. derksen , j. fei , _ general presentations of algebras , _ adv .
278 ( 2015 ) , 210237 .
h. derksen , j. weyman , _ semi - invariants of quivers and saturation for littlewood - richardson coefficients , _
( 2000 ) , no .
3 , 467479 . h. derksen , j. weyman , _ the combinatorics of quiver representation , _ ann . inst .
fourier ( grenoble ) 61 ( 2011 ) , no .
3 , 10611131 . h. derksen , j. weyman , a. zelevinsky , _ quivers with potentials and their representations i , _ selecta math .
14 ( 2008 ) , no .
1 , 59119 .
h. derksen , j. weyman , a. zelevinsky , _ quivers with potentials and their representations ii , _ j. amer . math .
( 2010 ) , no .
3 , 749790 . m. domokos , a. zubkov , _ semi - invariants of quivers as determinants , _ transform .
groups 6 ( 2001 ) , no .
1 , 924 . j. fei , _
cluster algebras and semi - invariant rings i. triple flags , _ arxiv:1411.4693 .
cluster algebras , invarant theory , and kronecker coefficients i , _ arxiv:1504.02970 . v. fock and a. goncharov , _ moduli spaces of local systems and higher teichmller theory , _ publ
hautes tudes sci . , 103 ( 2006 ) 1211
. s. fomin , p. pylyavskyy , _ tensor diagrams and cluster algebras , _ arxiv:1210.1888
. s. fomin , a. zelevinsky , _ cluster algebras .
i. foundations , _
( 2002 ) , no .
2 , 497529 . s. fomin , a. zelevinsky , _ cluster algebras .
coefficients , _ compos . math .
143 ( 2007 ) , no .
1 , 112164 . c. geiss , b. leclerc , j. schrer , _ auslander algebras and initial seeds for cluster , _ journal of the london mathematical society ( 2 ) 75 ( 2007 ) , 718740 .
c. geiss , b. leclerc , j. schrer , _ partial flag varieties and preprojective algebras , _ ann .
fourier ( grenoble ) 58 ( 2008 ) , no .
3 , 825876 . j. e. humphreys , _ introduction to lie algebras and representation theory , _ graduate texts in mathematics , vol . 9 .
springer - verlag , new york - berlin , 1972 .
b. keller , quiver mutation in java , http://www.math.jussieu.fr/ keller / quivermutation/. a.d .
king , _ moduli of representations of finite - dimensional algebras , _ quart .
oxford ser .
( 2 ) 45 ( 1994 ) , no .
180 , 515530 .
a. knutson , t. tao , _ the honeycomb model of @xmath901 tensor products .
i. proof of the saturation conjecture , _
soc . 12 ( 1999 ) , no .
4 , 10551090 .
p. plamondon , _ generic bases for cluster algebras from the cluster category , _ int math res notices ( 2013 ) 2013 ( 10 ) : 23682420
. v. l. popov , e. b. vinberg , _ invariant theory , in : algebraic geometry .
iv , encyclopaedia of mathematical sciences , _ vol .
55 , springer - verlag , berlin , 1994 , 123284 .
a. schofield , _ semi - invariants of quivers , _ j. london math . soc .
( 2 ) 43 ( 1991 ) , no .
3 , 385395 .
a. schofield , _ general representations of quivers , _ proc .
london math .
( 3 ) 65 ( 1992 ) , no .
1 , 4664 .
d. speyer , _ an infinitely generated upper cluster algebra , _ arxiv:1305.6867 .
a. schofield , m. van den bergh , _ semi - invariants of quivers for arbitrary dimension vectors , _ indag .
12 ( 2001 ) , no .
1 , 125138 .
j. scott , _ grassmannians and cluster algebras , _ proc .
london math .
( 3 ) 92 ( 2006 ) , no .
2 , 345380 .
z. koda , _ noncommutative localization in noncommutative geometry , _ non - commutative localization in algebra and topology , 220313 , london math .
lecture note ser . , 330 , cambridge univ .
press , cambridge , 2006 . | let @xmath0 be the semi - invariant ring of @xmath1-dimensional representations of a quiver @xmath2 .
suppose that @xmath3 projects to another quiver with dimension vector @xmath4 through an exceptional representation @xmath5 .
we show that if @xmath0 is the upper cluster algebra associated to an ice quiver @xmath6 , then @xmath7 is the upper cluster algebra associated to @xmath8 , where @xmath8 is obtained from @xmath6 through simple operations depending on @xmath5 .
we also study the relation of their bases using the quiver with potential model .
[ multiblock footnote omitted ] |
it has been suggested @xcite , that globular clusters may contain central black holes with masses lying along the extension of the correlation between black hole mass and bulge velocity dispersion as seen for galaxies @xcite .
however it has also been suggested that the observational evidence can be otherwise explained @xcite .
these difficulties of interpretation are compounded by the small number of observable stars in the cores of globular clusters .
in contrast to a galactic center , current observations approach the limit when the uncertainty of an aggregate measurement , such as a velocity dispersion , can not be reduced by further observations . once
every star has been observed , there is nothing more to be done .
( see drukier et al .
2003 for a general discussion of issues relating to the observation and interpretation of velocity dispersions in globular clusters . )
this is particularly the case for central black holes , since there is only a relatively small region over which the black hole provides the dominant contribution to the gravitational potential .
most current observational work relating to the possible presence of central black holes in globular clusters has been done using radial velocity measurements to determine the central value and radial gradient of the velocity dispersion . in this _
letter we explore an alternative approach , namely the use of proper motion measurements to identify individual stars whose velocities are strongly influenced by the presence of a central black hole .
we find that existing crude models suggest that this approach may be fruitful , and that the considerable theoretical and observational work that will be required to apply this approach in practice may thus be justified . in 2 , we estimate the number of stars with anomalously high velocities caused by a central black hole , and in 3 we generate a list of the clusters with the highest probability of observing such stars .
we find that for plausible assumptions such stars might well be observable with hst in the most favorable clusters . _
consider a globular cluster with central velocity dispersion @xmath1 , where this is the dispersion outside the radius of influence , @xmath2 , of a black hole with mass @xmath3 , given by @xmath4 we define @xmath5 as the one - dimensional velocity dispersion and @xmath6 as the stellar surface density within @xmath2 .
outside @xmath2 , the influence of the black hole is too small to significantly affect stellar orbits . for simplicity , we assume that the velocities are distributed as a gaussian with dispersion @xmath5 . now , at a radius @xmath7 , the fraction of stars which will have velocities greater than some multiple @xmath8 of the velocity dispersion @xmath1 is given by @xmath9 where @xmath10 is the dimensionality of the observed velocity ( 1 for radial velocities , 2 for proper motions , and 3 for space velocities ) . while the upper limit could ,
in principle , be some multiple of the escape velocity , this is difficult to define without detailed modeling of the structure of the whole cluster . instead , as the complementary error function drops off rapidly with increasing argument , we will just use the infinite upper limit shown .
this will lead to a slight overestimate in the number of expected stars , but this effect is small compared to that due to the uncertainty in @xmath11 discussed below . with this limit ,
equation ( [ e : f ] ) reduces to @xmath12 the total number of stars with `` significantly '' high velocities ( significant being defined here as @xmath8 times the velocity dispersion @xmath1 outside @xmath13 ) is then given by @xmath14 the only existing dynamical investigation of the effects of a central black hole on the structure of a globular cluster is that of @xcite who used the fokker - planck equation to integrate a steady - state , anisotropic distribution function in the vicinity of a black hole .
they found that the projected velocity dispersion and surface density profiles could be approximated , for @xmath15 , by @xmath16 the two terms in equation ( [ e : sigma ] ) deal with the flattening of the velocity dispersion profile near @xmath2 . substituting these into equation ( [ e : n1 ] )
leads , after a suitable rescaling of the integration variable , to @xmath17 where @xmath18 the integral can be evaluated numerically and gives @xmath19 to two figures . by way of comparison , if no black hole is present then , assuming @xmath20 and @xmath21 , @xmath22 for which @xmath23 , and the expected number of stars is given by equation ( [ e : n2 ] ) with @xmath24 in place of @xmath25 .
the galactic black hole mass versus velocity dispersion correlation @xcite is usually given in the form @xmath26 where @xmath1 is a suitable velocity dispersion and @xmath27 km s@xmath28 .
the uncertainty in the power - law slope is the largest source of uncertainty in our estimated number of high - velocity stars .
recent estimates for @xmath11 ( the variation in @xmath29 is small since all studies agree for @xmath30 ) range from @xmath31 ( @xmath32 ) @xcite to @xmath33 ( @xmath34 ) @xcite .
these agree within their quoted errors , but , on extrapolation to the globular cluster regime , the use of @xmath35 predicts 20 times as many high - velocity stars as does the use of @xmath36 .
we therefore give results for both these slopes below .
the original claim for globular clusters @xcite used @xmath35 . substituting equations ( [ e : rh ] ) and
( [ e : mbh ] ) into equation ( [ e : n2 ] ) we arrive at the following @xmath37 where @xmath1 is measured in km s@xmath28 and @xmath38 is the number of _ measurable _ stars per square parsec .
what we can easily observe is not @xmath38 but the central surface density , @xmath39 , so we need to rewrite @xmath38 in terms of @xmath39 for a reasonable globular cluster stellar population . for a defined population of stars ,
let @xmath40 be the cluster luminosity per star in that population in solar units , and let @xmath41 be the fraction of these stars that are usefully measurable . then for @xmath39 in @xmath42 magnitude per square arc second , and taking @xmath43 @xmath44 in this case @xmath45 where @xmath46 , @xmath47 , @xmath48 , and @xmath49 . for our problem ,
@xmath40 is determined by the luminosity function in the core of the cluster in question .
it needs to take into account mass segregation effects and other possible population peculiarities as might be indicated by , for example , color gradients .
the measurable fraction , @xmath41 , is an observational selection effect on the luminosity function .
it will depend on the observational technique to be used , the distance of the cluster , crowding and so forth .
since a complete stellar census is a difficult undertaking , we estimate @xmath40 as follows .
define our population to be all the stars brighter than some magnitude , @xmath50 ( e.g. the expected magnitude limit of the observations ) in a cluster color - magnitude diagram ( cmd ) .
then , @xmath51 the quantities @xmath52 and @xmath53 are the mean luminosities for the bright and faint parts of the cmd ( divided at @xmath50 ) and @xmath54 is the ratio of the number of stars brighter than @xmath50 to the number of those fainter .
@xmath52 can be estimated directly from the cluster cmd , for which purpose we use the cmds in the compilation by @xcite .
@xmath53 and @xmath54 can be estimated from the corresponding theoretical luminosity function ( lf ) .
we use those from the models of @xcite .
one limitation is that these lfs only include stars from the hydrogen burning limit to the tip of the red giant branch , so we must use the cmd to correct the ratio of the number of stars in each part of the lf , @xmath55 , for the contribution of post - rgb stars .
if , brighter than @xmath50 , we see in the cmd @xmath56 stars in the regions covered by the luminosity function , and @xmath57 more evolved stars elsewhere , then @xmath58 by substituting equations ( [ e : nf ] ) and ( [ e : lbar ] ) into equation ( [ e : n3 ] ) , we can now estimate @xmath59 for any given cluster , subject to the selection of @xmath41 .
we proceed to do this in the next section .
lrrrrrrr ngc 6388 & 14.55 & 18.9 & 10.0 & 2.5 & 11.4 & 2.5 & 3.7 + ngc 6441 & 14.99 & 18.0 & 11.7 & 2.0 & 11.3 & 1.4 & 1.9 + ngc 7078 & 14.21 & 14.0 & 10.3 & 1.3 & 11.8 & 1.0 & 1.0 + ngc 6715 & 14.82 & 14.2 & 26.8 & 0.5 & 11.4 : & 0.6 & 0.6 + ngc 5139 & 16.77 & 22.0 & 5.3 & 6.5 & 11.4 : & 0.6 & 1.1 + ngc 104 & 14.43 & 11.5 & 4.5 & 2.1 & 9.8 & 0.4 & 0.3 + ngc 6266 & 15.35 & 14.3 & 6.9 & 2.1 & 10.5 & 0.4 & 0.4 + ngc 2808 & 15.17 & 13.4 & 9.6 & 1.3 & 10.7 & 0.4 & 0.4 + ngc 1851 & 14.15 & 10.4 & 12.1 & 0.6 & 11.2 & 0.3 & 0.2 + ngc 6752 & 15.20 & 12.5 & 4.0 & 2.8 & 11.4 : & 0.3 & 0.2 + ngc 6093 & 15.19 & 12.4 & 10.0 & 1.1 & 12.5 & 0.2 & 0.2 + ngc 5824 & 15.08 & 11.6 & 32.0 & 0.3 & 13.7 & 0.2 & 0.1 +
we present in table [ t : top choices ] some relevant numbers for the 12 clusters most likely to show evidence for a central black hole .
all other clusters are estimated to have at least a factor of three fewer observable high - velocity stars than ngc 5824 .
the value of @xmath2 in the fifth column of the table is calculated assuming @xmath60 and @xmath35 , and is given in arc seconds for the distance in column 4 . note
that while this radius scales as @xmath61 , the radii for @xmath36 are smaller by a factor of approximately 5 .
our estimates of @xmath40 are given in the sixth column of table [ t : top choices ] .
the @xcite cmds cover 9 of our listed clusters .
for those not listed we have used @xmath62 , the mean for the other 9 .
these are marked by a colon after @xmath63 .
we have taken @xmath50 for each cluster such that the division is at @xmath64 , a magnitude or so below the turn - off for these low - metallicity systems .
for the luminosity functions we use a very flat mass - spectral - index @xmath65 of -0.5 , where the salpeter value is 1.35 .
mass segregation is likely to have removed most of the low - mass stars from the cluster center many of the mass functions are actually inverted ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) so this should be a reasonable approximation . in any case , the luminosity - function - dependent term @xmath66 in equation ( [ e : lbar ] ) is roughly 10% that of @xmath52 , so the estimate of @xmath40 is dominated by the stars in the cmd .
the final two columns give the expected number of stars for the two estimates of the @xmath0 relation relative to the number expected for m 15 ( ngc 7078 ) , the cluster with the best studied core .
only two clusters , ngc 6388 and ngc 6441 , are likely to have more fast - moving stars than m 15 for the lower slope .
ngc 5139 ( @xmath67 cen ) would have roughly the same number as m 15 using the higher slope , but the absolute numbers are down by a factor of 24 with respect to the low slope . to complete our estimate
, it remains only to specify @xmath41 to get the number of fast - moving stars we could detect .
we consider two scenarios : radial velocities and proper motions . for radial velocity measurements ,
@xmath41 is likely to be small , as only the brightest , least crowded , central stars are suitable . for the case of m 15 , if we compare the number of stars with radial velocities as compiled by @xcite with the _ hst _ photometric list in @xcite then , with @xmath68 , we find that about 4% of the stars
brighter than @xmath50 have velocities . within @xmath69
this is 7% , but within @xmath70 , where most of the fast - moving stars are to be found , there are no velocities out of the 10 stars available .
thus , for radial velocities in m 15 , @xmath71 at best , and we would expect to see @xmath72 stars , moving with a relative radial velocity @xmath73 if a black hole is present . that no such star has been seen is , therefore , unsurprising and uninformative .
combining the entire dozen most - likely clusters yields perhaps one or two fast - moving stars .
@xmath74 stars .
the number expected is effectively zero . a further problem with using radial velocities to detect the fast - moving stars lies in their radial distribution as given by the integrands in equations ( [ e : i1 ] ) and ( [ e : i2 ] )
after an appropriate coordinate transformation .
we compare these in figure [ f : x ] for @xmath75 and the three values of @xmath10 . in all three cases ,
the fast - moving stars are concentrated well within @xmath2 .
the break point at @xmath76 is the result of the break in @xmath77 at this radius .
note the very different expected radial distributions for the fast - moving stars depending on whether ( _ solid _ ) or not ( _ dotted _ ) there is a black hole .
if there is a black hole , the fast - moving stars should be concentrated in the center of the cluster , with a maximum inside @xmath78 . for radial velocities ,
the high - velocity stars are confined within about @xmath79 or 03 for the nearer clusters .
for the higher - dimensional velocities , the detection region is somewhat broader
. clearly , radial velocities are unlikely to prove the question one way or another . for proper motions ,
the situation is much better , at least for nearby clusters for which such observations are feasible . in the case of ngc 6752 , @xcite measure 153
stars brighter than @xmath50 within @xmath80 of the cluster center . to the same radius , @xcite ,
measure the proper motions of 15 stars .
their detection region only covers about a third of the area within @xmath2 due to gaps in their 1999 data .
these gaps , unfortunately , covers much of the very center of the cluster , so it is difficult to estimate the detection rate within about @xmath81 where we expect to find most of the fast - moving stars .
crowding is bound to be a problem in this region , even for _ hst_. extrapolating from the most crowded regions studied by @xcite and taking into account the gaps in their coverage , we estimate that @xmath82 is appropriate for proper motions . in this case
@xmath83 stars .
if this @xmath0 conjecture is correct , the top three clusters should contain of order 10 high - velocity stars between them with the balance of the 12 clusters in table [ t : top choices ] contributing another 5 or so in total .
in addition , their radial range is double that for radial velocities . on the other hand , @xmath84 stars , and the black holes
will be undetectable by this method .
proper motions have the feature that their uncertainty is inversely proportional to the time baseline and proportional to the distance .
nonetheless , the top three clusters , all with distances between 10 and 12 kpc , should be close enough that , given sufficient data , the black hole hypothesis can be tested , assuming a value of @xmath11 at the lower end of its range . the main limitation at this distance will be the increase in effective crowding , proportional to the distance squared . for the top three clusters in table [ t : top choices ] , crowding will be 10 to 16 times higher than in the ngc 6752 observations of @xcite . in the absence of a black hole ,
the flat extrapolation would lead us to expect to see one or two fast - moving stars within @xmath2 in each of our top three candidates .
these stars will have a radial distribution proportional to radius , and should be found in the vicinity of @xmath2 , not concentrated within @xmath85 as is predicted under the black hole conjecture .
fast - moving stars can have origins other than a black hole , of course .
ejection from the core during core - collapse is one plausible mechanism for producing such stars @xcite , but their velocity vectors will be radial unlike a star in orbit around a black hole .
we caution that the numbers presented here are only estimates and depend on the scalings found by @xcite .
those models are single - mass anisotropic fokker - planck simulations for the steady - state stellar distribution in the vicinity of the black hole .
more modern models , which should include , at the very least , a range of stellar masses and a self - consistent potential , will be needed to fully assess the significance of any fast - moving stars which are observed in these clusters .
the estimates made here also depend on the current central velocity dispersions in the globular clusters .
since globular clusters can lose a large fraction of their mass due to stellar evolution and stellar - dynamical evolution , the numbers provided in table [ t : top choices ] may well be underestimated if the proper velocity dispersion to use in determining black hole masses is the original value , not the current one .
further , the mass of any central black hole will have increased to some extent due to the capture of cluster stars .
using double the current velocity dispersion , for example , would increase @xmath86 by a factor of 16 and @xmath87 by a factor of 39 , in which case the higher slope also predicts significant numbers of observable stars in the top few clusters . given the sensitivity to these effects , obtaining reliable estimates for the numbers of high - velocity stars will require fully evolving models . even in the event that proper - motion studies uncover no fast moving stars
, such models will allow for firm upper limits on the mass of any black hole .
constructing such models , and carrying out the necessary proper motion observations , is no small task , but given the strong interest and controversy currently surrounding this topic , we believe that efforts along these lines should be vigorously pursued .
albrow , m. d. , de marchi , g. , & sahu , k. c. 2002 , , 579 , 660 baumgardt , h. , hut , p. , makino , j. , mcmillan , s. , & portegies zwart , s. 2003a , , 582 , l21 baumgardt , h. , makino , j. , hut , p. , mcmillan , & s. , portegies zwart , s. 2003b , , 589 , l25 cohn , h.n . & kulsrud , r.m .
1978 , , 226 , 1087 de marchi , g. , paresce , f. , & pulone , l. 2000 , , 530 , 351 dull , j. d. , cohn , h. n. , lugger , p. m. , murphy , b. w. , seitzer , p. o. , callanan , p. j. , rutten , r. g. m. , & charles , p. a. 2003 , , 585 , 598 drukier , g.a . , cohn , h.n . , lugger , p.m. , & yong , h. 1999 , , 518 , 233 drukier , g.a . , bailyn , c.d .
, van altena , w.f . , & girard , t.m .
2003 , , 125 , 2559 ferrarese , l. & merritt , d. 2000 , , 539 , l9 gebhardt , k. et al .
2000 , , 539 , l13 gebhardt , k. , rich , r.m .
, & ho , l.c .
2002 , , 578 , l41 gerssen , j. , van der marel , r.p . ,
gebhardt , k. , guhathakurta , p. , peterson , r.c . , & pryor , c. 2002 , , 124 , 3270 harris , w.e .
1996 , , 112 , 1487 merritt , d. & ferrarese , l. 2001 , in asp conf .
249 , the central kiloparsec of starbursts and agns , ed .
knapen , j.k .
beckman , i. shlosman , & t.j .
mahoney ( san francisco : asp ) , 335 merritt , d. , meylan , g. , & mayor , m. 1997 , , 114 , 1074 piotto , g. et al .
2002 , , 391 , 945 pryor , c. & meylan , g. 1993 , in asp conf .
50 , structure and dynamics of globular clusters , ed .
djorgovski & g. meylan , ( san francisco : asp ) , 357 rubenstein , e.p . &
bailyn , c.d .
1997 , , 474 , 701 silvestri et al .
1998 , , 509 , 192 sosin , c. 1997 , , 114 , 1517 tremaine , s. et al .
2002 , , 574 , 740 van der marel , r.p .
, gerssen , j. , guhathakurta , p. , peterson , r.c . , & gebhardt , k. 2002 , , 124 , 3255 | we estimate the number of individual , fast - moving stars observable in globular clusters under the assumption that the clusters contain massive central black holes which follow the galactic @xmath0 relationship .
we find that radial velocity measurements are unlikely to detect such stars , but that proper motion studies could reveal such stars , if they exist , in the most likely clusters .
thus , _ hst _
proper motion studies can test this hypothesis in a few nearby clusters . |
the availability of cold and ultracold dimers provides a unique opportunity to investigate fundamental quantum processes @xcite .
the efforts of many experimental groups are focused on the production of heteronuclear alkali dimers , as they constitute a prototype of an ultracold quantum gas with long - range dipole - dipole interactions . in this context , the photoassociation of lics , nacs , krb , and rbcs has been reported recently @xcite , while feshbach resonances have been observed for lina , lik , lirb , and krb @xcite .
the long - range anisotropic dipole - dipole interaction between these polar molecules gives rise to an intriguing many body physics with new interesting physical phenomena .
in addition , these systems allow for a wide range of applications , such as state resolved chemical reactions @xcite or quantum computing @xcite .
particularly interesting is the study of the influence of external fields on these molecular systems .
external fields play a key role in trapping and cooling processes as well as in the manipulation and control of the dipole - dipole interaction , chemical reactions or collisions @xcite . as an example , the application of an electric field has been predicted to significantly affect the photoassociation process of ultracold polar molecules as well as the subsequent radiative decay cascade @xcite . on the other hand , the internal structure of molecules is significantly affected by an electric field .
the strong field regime is characterized by the appearance of pendular states : the molecule is oriented along the field axis and can be described as a coherent superposition of field - free rotational states . recently , the authors have performed studies of the rovibrational spectra of co and lics in their electronic ground states exposed to an electric field : the orientation and hybridization of the angular motion , the stark shift , and the radiative decay properties have been investigated for a wide range of field strengths and rovibrational levels @xcite . in the present work
we extend our previous studies by performing a comparative analysis of several polar alkali dimers which are of immediate experimental interest .
specifically , we analyze the influence of a static electric field on the @xmath0 electronic ground state of @xmath1li@xmath2na , @xmath1li@xmath3k , @xmath1li@xmath4rb , @xmath5li@xmath6cs , @xmath2na@xmath7k , @xmath2na@xmath6cs , @xmath7k@xmath8rb , and @xmath8rb@xmath6cs .
in the following we consider the field regime where perturbation theory holds for the description of the molecular electronic structure .
the born - oppenheimer approach then provides the following hamiltonian for the nuclear motion of a polar dimer in its electronic ground state exposed to a static electric field @xmath9 where the molecule fixed frame with the origin at the center of mass of the nuclei has been employed , ( @xmath10 ) being the internuclear distance and the euler angles .
@xmath11 , @xmath12 , @xmath13 , @xmath14 , and @xmath15 are the vibrational kinetic energy , the reduced mass of the nuclei , the rotational angular momentum , the field - free electronic potential energy curve ( pec ) , and the electronic dipole moment function ( edmf ) , respectively .
the electric field is taken parallel to the laboratory frame @xmath16-axis with strength @xmath17 .
since we restrict ourselves on a non - relativistic treatment , couplings to electronic states of different spin symmetry do not arise . in the framework of the effective rotor approach @xcite
, the rovibrational hamiltonian ( [ eq : rotvib_hamiltonian ] ) reduces to @xmath18 with @xmath19 , @xmath20 .
@xmath21 and @xmath22 are the field - free vibrational wave function and energy for a given vibrational state @xmath23 with @xmath24 , respectively .
this approximation , which is valid for large parts of the rovibrational spectrum , relies on two assumptions : i ) the rotational and vibrational energy scales should differ by several orders of magnitude and can therefore be separated adiabatically , and ii ) the field influence on the vibrational motion is very small and can consequently be treated by perturbation theory . besides reducing ( [ eq : rotvib_hamiltonian ] ) to an one - dimensional problem , this approach allows us to compare the field influence on several vibrational levels , and even between different molecular species , by means of the parameter @xmath25 which derives from rescaling the hamiltonian ( [ hamil_era ] ) using the rotational constant of a given vibrational state , @xmath26 .
if @xmath27 acquires the same order of magnitude as the rescaled rotational kinetic energy , @xmath28 , the rotational dynamics is severely altered by the interaction with the electric field , while @xmath29 provides an estimation of the field strength needed to show a significant effect .
we remark that in the presence of the electric field , only the magnetic quantum number is retained , whereas the vibrational and rotational ones are no more conserved .
however , for reasons of addressability , we will label the electrically dressed states by means of their field - free vibrational , rotational , and magnetic quantum numbers @xmath30 .
furthermore , in this letter we focus on rotationally non - excited states with corresponding field - free quantum numbers @xmath31 .
the computational method used to solve the full rovibrational schrdinger equation is described in ref .
the pecs of the singlet electronic ground states of the alkali dimers employed here are derived from either experimental or theoretical studies . for the pecs of lina , lik , lics ,
nak , nacs , krb , and rbcs we use very accurate experimental data from spectroscopic analyses @xcite .
the lirb pec is taken from theoretical studies based on a semiempirical method @xcite .
when describing vibrationally highly excited states , special attention has to be payed to the long - range part of the pecs : for lina , lics , nacs , krb , and rbcs , the pecs include van der waals terms , @xmath32 , and an exchange energy term , @xmath33 , see refs .
@xcite , with @xmath34 for lina @xcite .
the long - range part of the nak pec additionally includes damping functions @xmath35 in the van der waals expansion , @xmath36 , and a different expression for the exchange energy , @xmath37 @xcite . for the lirb dimer only the first van der waals term , @xmath38 ,
is considered @xcite .
electronic ground states of lina ( @xmath39 ) , lik ( @xmath40 ) , lirb ( @xmath41 ) , lics ( @xmath42 ) , nak ( @xmath43 ) , nacs(@xmath44 ) , krb ( @xmath45 ) , and rbcs ( @xmath46 ) . ]
the edmfs employed for the lina , lik , lics , nak , nacs , and krb systems are computed by the same semiempirical technique as the lirb pec @xcite .
the rbcs edmf is taken from an ab - initio relativistic valence - bound calculation @xcite . since for the electronic ground state of the polar alkali
dimers the long - range behaviour of the edmf is given by @xmath47 @xcite , this asymptotic function has been fitted to the theoretical data at large internuclear distances .
we note that , when necessary for computational reasons , the short - range behaviour of the edmfs is linearly extrapolated .
the resulting edmfs are plotted in fig .
[ fig : edmf ] as a function of the internuclear distance .
all edmfs are negative and exhibit a broad minimum which is shifted with respect to the minimum of the corresponding pec , _
e.g. _ , by @xmath48 and @xmath49 for the krb and lics dimers , respectively .
the edmfs of lirb , nak , nacs , and krb become positive for @xmath50 albeit approaching zero thereafter ; this change of sign will be reflected in the response of highly excited rovibrational levels to the field .
however , this has to be taken with a grain of salt since ab - initio electronic structure calculations of molecules typically employ basis sets of exponentially localized functions , which do not easily catch the long - range behaviour of the atom - atom interaction .
the lics edmf has the largest absolute value with a maximum of @xmath51 at @xmath52 .
the edmfs of lina , krb , and rbcs show only a weak dependence on @xmath53 for a wide range of internuclear distances and overall comparatively small values for the dipole moment . versus the field - free vibrational quantum number @xmath23 for the same dimers as in fig .
[ fig : edmf ] . ]
let us start by analyzing the behaviour of the parameter @xmath54 as a function of the vibrational quantum number . in fig .
[ fig : omega_nu ] , the modulus of @xmath54 is presented for the @xmath55 levels of all dimers on a logarithmic scale .
the maximum vibrational quantum numbers considered are @xmath56 ( lik ) , 46 ( lina ) , 48 ( lirb ) , 53 ( lics ) , 73 ( nak ) , 85 ( nacs ) , 97 ( krb ) , and 135 ( rbcs ) ; the very last vibrational bands have not been included . for all dimers
, @xmath57 shows a qualitatively similar but quantitatively different behaviour as a function of @xmath23 : after a weakly pronounced maximum , @xmath57 decreases rapidly for high vibrational excitations .
in particular , nak , krb , and rbcs show a plateau - like behaviour indicating a quantitatively similar field impact in many vibrational bands . according to their @xmath57 values
, the considered molecules can be classified into four groups whose response to the field will be of comparable magnitude for a large part of their spectra . possessing the largest @xmath57 values ,
the first group is formed by the nacs and rbcs dimers and their @xmath58 states , followed by the @xmath59 levels of lics and nak . despite the large lics edmf , the corresponding @xmath57 is only approximately half of the @xmath57 belonging to nacs and rbcs : the large reduced masses of the latter compensate their smaller dipole moments .
the third set contains lik , lirb , and krb and their lowest @xmath60 vibrational bands . finally , lina will be least affected by the field which is due to both small @xmath15 and @xmath12 .
we remark that the sign of @xmath54 depends crucially on the corresponding dipole moment : while being generally negative , the change of sign of the nak , nacs , and krb edmfs at long distances causes @xmath61 for their @xmath62 , @xmath63 and @xmath64 levels , respectively .
in particular , the minimum observed for the @xmath65 level of krb can be attributed to this change of sign .
for all molecules considered , the slope of @xmath57 becomes very steep as @xmath23 approaches the dissociation threshold .
consequently @xmath57 drastically varies from its maximum to its minimum ; for example we get for lina @xmath66 versus @xmath67 cm / kv , and for nacs @xmath68 versus @xmath69 cm / kv .
hence , for high - lying vibrational excitations the field strength should be enhanced at least one order of magnitude to obtain similar field - dressed states as on the lower part of the rovibrational spectrum . in order to quantify the spectral changes upon the application of the electric field , we consider the ratio of the stark shift to the rotational spacing in the presence of the field @xmath70 where @xmath71 and @xmath72 are the energies of the @xmath55 and @xmath73 levels at field strength @xmath17
; for high - field seekers we get @xmath74 . if @xmath75 , the field - dressed spectrum shows many similarities to the field - free one . in particular
, this criterion is fulfilled for @xmath76 : in second order perturbation theory the @xmath55 and @xmath73 states are shifted by @xmath77 and @xmath78 @xcite , respectively , compared to the field - free rotational spacing of @xmath79 . in figs .
[ fig : kappa](a ) and ( b ) the parameter @xmath80 is presented as a function of the vibrational quantum number for @xmath81 and @xmath82 kv / cm , respectively ; the selected field strengths are chosen such that the experimentally accessible range within the strong field regime is covered . besides showing high - field seeking character , the behaviour of @xmath80 has many analogies among the considered molecules .
initially , it has a weak dependence on the vibrational excitation , thereby smoothly increasing as @xmath23 is enhanced and passing a shallow maximum ( due to the semilogarithmic scale , this increasing trend appears as a plateau ) . for higher vibrational excitations ,
@xmath80 rapidly decreases as the levels get closer to the dissociation limit : this reduction can be up to one order of magnitude when @xmath23 is enhanced by merely one quantum . the above classification of the considered molecules
is equally observed for @xmath80 .
the lowest values appear for lina , @xmath83 and @xmath84 , for @xmath81 and @xmath82 kv / cm , respectively , _ i.e. _ , the impact of the field on the spectrum is negligible for the smaller of these two field strengths .
in contrast , nacs and rbcs are most affected by the field @xmath81 kv / cm , with @xmath85 for @xmath86 and @xmath87 , respectively .
lics and nak possess values @xmath88 for @xmath89 and @xmath90 , respectively . for lirb and krb we found @xmath91 for the @xmath92 and @xmath93 states , whereas a smaller value is achieved for lik with @xmath94 for @xmath95 .
an increase of the field strength by one order of magnitude ( @xmath96 kv / cm ) produces a significant variation of @xmath80 , which depends on the system and especially on the states under consideration .
indeed , we observe @xmath97 for most levels of lics , nak , nacs , and rbcs .
the levels @xmath98 , and @xmath99 of lik , lirb , and krb satisfy @xmath100 . in order to provide a reasonable scaling ,
the minimum of @xmath80 for krb at @xmath65 has not been included in fig .
[ fig : kappa ] . as a function of the field - free vibrational quantum number for the @xmath55 levels and same molecules as in fig .
[ fig : edmf ] , for ( a ) @xmath81 kv / cm and ( b ) @xmath82 kv / cm . ] the orientation and alignment of the pendular states are characterized by the expectation value @xmath101 and @xmath102 , respectively . the closer @xmath103 is to zero , the stronger is the alignment and the closer the modulus of @xmath101 is to one , the stronger is the orientation of the state along the field .
the expectation value @xmath101 of these levels is presented in figs .
[ fig : cos](a ) and ( b ) for @xmath81 and @xmath82 kv / cm , respectively , as a function of the field - free vibrational quantum number .
the corresponding results for @xmath103 are shown in figs .
[ fig : delta](a ) and ( b ) .
the sign of the effective electric dipole moment , @xmath104 , determines whether the orientation will be parallel or antiparallel to the field direction , _
i.e. _ , @xmath105 or @xmath106 , respectively .
for the moment let us focus on the description of the results for @xmath81 kv / cm , see figs .
[ fig : cos](a ) and [ fig : delta](a ) .
as expected from the behaviour of the parameter @xmath54 , the orientation smoothly increases as @xmath23 is enhanced , reaching a broad maximum and rapidly decreasing thereafter as the state gets close to the dissociation threshold . since the initial increasing trend is not very pronounced for nak , nacs , krb , and rbcs
, @xmath101 exhibits a plateau as a function of @xmath23 followed by a rapid decrease of the orientation .
the strongest orientation is achieved for the nacs and rbcs dimers with @xmath107 , for the @xmath108 and @xmath109 levels , respectively .
the states @xmath89 and @xmath110 of lics and nak , respectively , present a significant orientation as well : @xmath111 . for lik , lirb , and krb
the influence of the field is not very pronounced and the extremal values are @xmath112 , @xmath113 , and @xmath114 for the @xmath115 , @xmath116 , and @xmath117 levels , respectively . for lina as well as for highly vibrationally excited levels of the other molecules , the effect of an electric field of strength @xmath81
kv / cm is negligible .
analogous conclusions are derived from the analysis of @xmath103 , whose behaviour as a function of @xmath23 is very similar to the one previously described for @xmath101 .
the variation of @xmath103 is very smooth as @xmath23 is augmented , and major changes are observed only for high - lying excitations .
the nacs and rbcs dimers show the strongest alignment , with @xmath118 for the @xmath119 and @xmath120 levels , respectively . the other dimers present relative large values with @xmath121 which is the lowest one achieved for lics at @xmath122 , indicating a modest alignment . in particular
, @xmath103 is very close to its field - free value , _
i.e. _ , @xmath123 , for the lina dimer and for all high - lying vibrational levels .
if the field strength is increased to @xmath96 kv / cm , both the orientation and alignment are more pronounced , see figs .
[ fig : cos](b ) and [ fig : delta](b ) . except for the lina molecule
, the strong field regime with the appearance of pendular states is already reached for most of the levels considered .
we observe @xmath124 for the states with @xmath125 , and @xmath126 of the lirb , lics , nak , nacs , and rbcs dimers , respectively .
even more , @xmath127 for the @xmath128 and @xmath129 states of nacs and rbcs , respectively .
the lina molecule still shows a modest orientation , with a minimal value @xmath130 for the @xmath131 state : only if @xmath17 is further increased by one order of magnitude a significant orientation , _ i.e. _ , @xmath132 , is attained for this molecule . for @xmath96
kv / cm , the strong field regime is also manifest in @xmath103 where many levels present a pronounced alignment with @xmath133 .
the smallest values , @xmath134 , are obtained for the @xmath135 and @xmath136 states of nacs and rbcs , respectively .
in contrast , for the lina dimer we find that @xmath137 for any @xmath23 .
finally , let us remark that @xmath101 is positive for the states with @xmath62 , @xmath63 , and @xmath64 of the nak , nacs , and krb dimers , respectively , while being characterized by a weak orientation and alignment at both field strengths . in particular , for krb @xmath138 ( @xmath103 ) achieves a minimum ( maximum ) for the @xmath65 level , _
i.e. _ , for the same state as @xmath57 shows a minimum , cf . fig .
[ fig : omega_nu ] .
but for the expectation value @xmath101 . ]
but for @xmath139 .
] the expectation value @xmath140 provides an estimate of the field - induced hybridization of the angular motion .
since we only consider the rotational ground state in this letter , one yields @xmath141 for @xmath142 .
figures [ fig : j2](a ) and ( b ) illustrate the evolution of @xmath140 with increasing degree of vibrational excitation for @xmath81 and @xmath82 kv / cm , respectively . for both field strengths investigated , @xmath140 presents a similar behaviour for all molecules : it increases as the vibrational excitation is enhanced , passes through a broad maximum , and decreases , approaching zero , thereafter .
the lower parts of the nak , krb , and rbcs spectra show again a plateau - like behaviour of @xmath140 . for @xmath81 kv / cm , the rotational dynamics has mostly @xmath143-wave character , _
e.g. _ , @xmath144 for the @xmath145 level of nacs , @xmath146 for the @xmath147 state of nak , and @xmath148 for the @xmath149 level of krb .
the hybridization of the angular motion is negligible for lina , _
i.e. _ , @xmath150 for any @xmath23 , and @xmath151 of the dynamics is controlled by the @xmath24 wave .
in contrast , the pendular character of the nacs and rbcs dimers is already significant with @xmath152 for the @xmath135 and @xmath153 levels , respectively .
the @xmath154 and @xmath155 states of lics and nak , respectively , satisfy @xmath156 ; analogously we observe for lirb and krb @xmath157 for @xmath158 and @xmath93 .
for the lik states we find @xmath159 for @xmath160 .
because of the high - field seeking character of these levels , @xmath140 is drastically increased for @xmath96 kv / cm , see fig .
[ fig : j2](b ) . except for lina
, strong field effects characterize this expectation value for a wide range of levels .
indeed , @xmath161 for the @xmath162 and @xmath163 states of nacs and rbcs , @xmath164 for those with @xmath165 and @xmath166 of the nak and lics dimers , and @xmath167 for the lowest @xmath168 vibrational states of lik .
they present a rich pendular dynamics , with significant contributions of higher field - free rotational states .
as @xmath23 gets closer to the dissociation limit the relative weight of the @xmath143-wave becomes more important , and @xmath140 is reduced .
in contrast , the field - free @xmath24 level still dominates the lina states with @xmath169 for any @xmath23-value .
finally , let us remark that the effect of the field in terms of @xmath140 for the @xmath65 level of krb is negligible but slightly increases again for the last states .
but for the expectation value @xmath140 .
in conclusion , we have investigated the rovibrational spectra of the electronic ground state @xmath0 of the alkali dimers lina , lik , lirb , lics , nak , nacs , krb , and rbcs in the presence of a static homogeneous electric field . performing a comparative analysis ,
we have considered the lowest rotational excitation within each vibrational band up to the dissociation limit .
the vibrational state - dependent parameter @xmath54 provides a first estimation of the influence of the field on the corresponding dimer and allows a classification of these systems into four groups .
thus , nacs and rbcs are most affected by the application of an electric field and lina is least . in general , vibrationally highly excited states show less field impact whereas already for moderate field strengths pronounced effects are observed for lower - lying levels .
the richness and variety of field - dressed rotational dynamics have been illustrated by analyzing the stark shift of the spectra as well as the orientation , alignment , and hybridization of the angular motion .
this work should serve as a guide to experimentalists interested in electric field effects on cold and ultracold heteronuclear alkali dimers . a natural extension of this work includes the study of the influence of the field on states emerging from higher rotational excitations and/or with nonzero magnetic quantum numbers @xmath170 . | a comparative study of the effect of a static homogeneous electric field on the rovibrational spectra of several polar dimers in their @xmath0 electronic ground state is performed . focusing upon the rotational ground state within each vibrational band ,
results for energies and various expectation values are presented . for moderate field strengths
the electric field - induced energy shifts , orientation , alignment , and angular motion hybridization are analyzed up to high vibrational excitations close to the dissociation threshold . |
in this paper i consider the problem of detecting neutrinos from active galactic nuclei ( agn ) .
recent predictions ( see for example @xcite ) of the diffuse background of neutrinos from the sum of all agn are high enough that water and ice cerenkov detectors under construction such as dumand @xcite , amanda @xcite , lake baikal @xcite and nestor @xcite should be sensitive to muons from the interactions of the neutrinos in the earth @xcite .
most recently , protheroe @xcite has used a new model to calculate the rate of gamma ray and neutrino production in agn jets .
the resulting diffuse neutrino background varies over many orders of magnitude depending upon assumptions made about the relative contributions of various production mechanisms in the agn system .
the minimum estimated flux may be beyond the reach of the current generation of neutrino detectors mentioned above .
i have reassessed the detectability of a wide range of flux predictions using calculations of the neutrino cross sections based on nucleon structure function parameterizations @xcite ( parton distributions ) derived from recent accelerator experiments @xcite .
use of the new parton distributions leads to higher cross sections and a corresponding increased probability of a detectable muon being produced in the rock surrounding a detector .
however , increased neutrino absorption in the earth is an important effect and offsets some of the gains for the longest neutrino paths through the earth .
i present the rates and angular distributions of muons for different agn spectra and for the flux of atmospheric neutrinos @xcite produced when cosmic rays interact with the earth s atmosphere , including neutrinos from charmed meson decay @xcite .
the outline of the paper is as follows .
a brief review of the quark parton model and how it is used to calculate neutrino cross sections using accelerator structure function data ( parton distributions ) is presented .
i then examine two of the most recent sets of parton distributions and look at the implications for the total cross sections . finally , i discuss the use of these cross sections to estimate the rates and angular distributions of neutrino induced muons in large detectors , using a monte carlo simulation of the passage of neutrinos through the earth .
this has the advantage of allowing a precise treatment of the effects of neutral current scattering , which reduces the neutrino energies before they finally interact to produce muons .
the prospects for detection of the agn fluxes above the atmospheric backgrounds are then examined . upon completion of these calculations
, i learned of similar work by gandhi , quigg , reno and sarcevic @xcite , using another recent set of parton distributions @xcite , which yield muon rates consistent with the results presented in this work .
the interaction of a neutrino with a nucleon may be described @xcite in terms of the quark - parton model .
the nucleon is considered to be a collection of quasi - free partons ( quarks and gluons ) which share the nucleon momentum .
a proton consists of three valence quarks ( up , up and down ) surrounded by a sea of quark / antiquark pairs of all flavours .
the description of a weak interaction of a neutrino ( energy @xmath2 ) with a nucleon ( mass @xmath3 ) in the quark - parton model is described in terms of an interaction of the neutrino with any one of the individual quarks ( which carries a fraction @xmath4 of the nucleon momentum ) . in a charged current process
a @xmath5 boson is exchanged from a muon - neutrino to the quark with the neutrino turning into a muon ( energy @xmath6 ) .
a neutral current process involves the exchange of a @xmath7 boson , with the neutrino remaining a neutrino .
the square of the 4-momentum transfer is denoted @xmath8 and the variable @xmath9 ( @xmath10 ) describes the fractional energy transfer .
the cross section for the inclusive processes @xmath11 and @xmath12 can be written in terms of the variables @xmath13 and @xmath14 ( referred to as bjorken scaling variables ) where @xmath4,@xmath15 and @xmath16 are related to the square of the centre of mass energy @xmath17 , by @xmath18 . the cross section for the _ charged current _
interaction @xmath11 is written in terms of three structure functions @xmath19 and @xmath20 , but if the transverse momentum carried by the quarks in the nucleon is negligible then the relation @xmath21 @xcite reduces this dependence to two structure functions @xcite : @xmath22^{2 } \nonumber \\ & & \;\;\;\;\;\;\;\;\;\;\;\;\;\ ; \times \left\{(1-y+\frac{y^{2}}{2 } ) f^{\nu n , \bar{\nu}n}_{2}(x , q^{2 } ) \pm ( y-\frac{y^{2}}{2})xf^{\nu n , \bar{\nu}n}_{3}(x , q^{2 } ) \right\ } \label{eq : sigma_cc}\end{aligned}\ ] ] the @xmath23 sign corresponds to the @xmath24 cross sections .
@xmath25 is the mass of the w - boson and @xmath26 is the fermi constant @xcite .
the cross section in equation [ eq : sigma_cc ] is expressed in natural units where @xmath27 yielding a result in units of gev@xmath28 , to convert to a cross section in @xmath29 we multiply by @xmath30 @xmath31 @xcite .
equation [ eq : sigma_cc ] also holds for @xmath32 .
the structure functions @xmath33 and @xmath34 are defined in terms of quark distributions @xcite : @xmath35 and @xmath36 + [ b - t]\}\end{aligned}\ ] ] where @xmath37 are the sums of the quark / antiquark distributions ( @xmath38 etc ) .
each quark distribution is a function of @xmath4 and @xmath8 describing the number of quarks of that flavour having a fractional momentum in the range @xmath39 , at momentum scale @xmath8 .
note that the @xmath23 sign again corresponds to the @xmath24 cross sections .
the _ neutral current _
cross sections @xmath12 are expressed in the same form as equation [ eq : sigma_cc ] , but with @xmath40 replaced by @xmath41 and the structure functions @xmath42 and @xmath43 written as @xcite @xmath44 \nonumber \\ & + & 2x[(c - s)+(t - b)][\delta^2_1 + \delta^2_3 - \delta^2_2 - \delta^2_4 ] \end{aligned}\ ] ] and @xmath45\ ] ] the @xmath46 factors are the chiral couplings , defined in terms of the weak mixing parameter @xmath47 @xcite @xmath48 for this work the cross sections have been calculated using sets of parton distributions @xcite fitted to the latest hera deep inelastic scattering data @xcite . the h1 and zeus collaborations have extracted the electromagnetic structure function @xmath49 from collisions of 26.7 gev electrons with 820 gev protons .
@xmath49 is the sum of the individual parton distributions weighted by the squares of their electric charges : @xmath50 from the @xmath49 measurement the individual quark distributions can be calculated and used to form the structure functions @xmath51 and @xmath52 needed for the neutrino nucleon cross section calculation .
however , to calculate the cross section in the energy range @xmath53 gev , we need to know the parton distributions at higher @xmath8 than the hera measurements .
qcd theory @xcite describes the evolution of the parton distributions to higher @xmath8 .
i consider two different solutions of the evolution equations , which predict sufficiently different behaviour in the unknown regions of @xmath8 and @xmath4 to give neutrino cross sections up to a factor of two bigger than previous calculations @xcite in the energy range of interest .
martin , roberts and stirling have presented several sets of parton distributions @xcite based on the increasing amount of accelerator data .
their latest parton set ( mrs g " ) @xcite is a parameterisation that includes the recent hera @xmath42 data . the mrs fits supply individual parton distributions @xmath54 which can be used to form @xmath42 and @xmath43 . in their recent work @xcite frichter , mckay and ralston
have effectively fitted a two parameter ( @xmath55 ) solution of the structure function evolution equations to the hera @xmath56 data : @xmath57 @xmath58 is the qcd scale parameter ( taken as @xmath59 in the fmr result ) and @xmath60 and @xmath61 are the @xmath62 moments of the quark - gluon and gluon - gluon splitting functions , evaluated at next to leading order @xcite . in order to obtain @xmath63 , fmr make the assumption that the quark ( @xmath64 ) and anti - quark ( @xmath65 ) distributions of the @xmath66 flavour are equal , and that @xmath67 substituting into @xmath49 and @xmath63 yields @xmath68 and @xmath69 from which we obtain : @xmath70 a least squares minimization to the 1992 and 1993 hera @xmath71 data yields best fit parameters @xmath72 and @xmath73 . to obtain the total cross sections i have integrated equation [ eq : sigma_cc ] using the mrs g and
fmr parton distributions , for both the charged and neutral current forms of @xmath42 and @xmath43 .
the mrs g distributions cover the full range @xmath74 required , but the fmr result applies only at @xmath75 or less . in order to evaluate the fmr cross section over the full @xmath13 range ,
i have used the fmr result for @xmath76 and the mrs g result for @xmath77 .
the mrs g parton distributions are valid down to @xmath78 , which is adequate for the calculation of the cross sections in the energy range considered in this work ( @xmath79 gev ) .
( note that the double - logarithmic approximation @xcite may be used to extrapolate the parton distributions to lower @xmath4 for cross section calculations @xcite where @xmath80gev . )
note also that i have combined a set of _ next - to - leading order _
( nlo ) parton distributions ( mrs g ) calculated in the @xmath81 renormalization convention with the _ leading order _ ( lo ) form of the structure functions @xmath82 and @xmath83 .
the nlo expressions for @xmath82 and @xmath83 ( obtained by convolving the quark and gluon distributions with the wilson coefficients @xcite ) will differ from the lo expressions by terms with coefficients @xmath84 , where @xmath85 is the qcd coupling constant , which decreases as @xmath8 increases . in the regions of high @xmath8 ( @xmath86 ) important for the neutrino energies considered in this work @xmath84 has a value of @xmath87 and the resulting differences between the lo and nlo structure functions , and hence the difference in the resulting cross sections , will be small ( @xmath88 a few percent ) .
given the very large uncertainties in the astrophysical assumptions in this work , this is an acceptable approximation .
the total neutrino / anti - neutrino charged / neutral current cross sections obtained by using the fmr and mrs g parton distributions are shown in figure [ crossfinal ] , along with the @xmath89 neutrino charged current estimate by reno @xcite . above @xmath90 gev neutrino and
anti - neutrino cross sections are equal and the fmr cross sections are roughly 1.5 times bigger than the mrs g predictions and 2 times bigger than the reno @xmath89 prediction .
the recent calculations of gandhi _ et al _ @xcite , using the cteq3 parton distributions @xcite , yield a neutrino charged current cross section lower than using mrs g by about 7% at 10@xmath91 gev , with difference rising to 23% at 10@xmath92 gev .
the cteq3 cross section is bigger than the reno @xmath89 cross section by 1 - 10% in the same energy range .
the reported observations @xcite of uhe @xmath93rays from binary x - ray systems in the early 1980s led to predictions of neutrino fluxes from such objects .
the assumption was that protons accelerated in the vicinity of the compact object would interact with other matter in the system to give both neutral and charged pions , the neutral pions decaying to produce the observed @xmath93rays and the charged pions producing a neutrino flux .
calculations were made @xcite of the expected neutrino flux by scaling the models to the observed @xmath93ray flux . however the non - confirmation of the @xmath93ray observations by much larger and more sensitive air shower arrays such as casa @xcite has led to upper limits on the @xmath93ray fluxes that are orders of magnitude lower than the original detection claims , dampening expectation of observable neutrino fluxes from binary x - ray systems .
attention in recent times has focussed on the production of neutrinos in active galactic nuclei , where particle acceleration may be occuring at accretion shocks in the vicinity of massive black holes .
stecker , done , salamon and sommers ( sdss ) @xcite calculated the neutrino flux resulting from interactions of protons with energy above the threshold energy for photoproduction on uv photons ( @xmath94 ) .
szabo and protheroe ( sp ) @xcite included all the subsequent interactions of the protons after the initial photoproduction .
luminosity functions @xcite ( derived from x - ray and gamma - ray measurements - describing the luminosity of agn as a function of redshift ) and the relationship between the neutrino and x - ray and gamma - ray outputs in a given agn model were used to calculate the neutrino output from the sum of all agn in the universe .
more recently stecker and salamon @xcite revised their original model and calculated diffuse neutrino backgrounds for quasar cores and blazar jets .
protheroe @xcite has calculated the neutrino and gamma - ray fluxes from proton acceleration in blobs of matter in agn jets .
two major production mechanisms were considered .
accelerated protons may leave the blob and interact with the radiation field of the accretion disk ( @xmath95 interactions ) , or interact with each other or with other matter in the vicinity ( @xmath96 interactions ) . the observed time variability of gamma - rays from agn such as mkr421 place constraints on the relative contribution of each mechanism .
the @xmath96 interactions have a long characteristic time scale and therefore can not produce variability of the order of a day .
the @xmath97 contribution , however , can vary rapidly and must be responsible for the time variation . to produce an intensity change of a factor of 2 over a short time scale , the @xmath96 interaction can contribute no more than about 50% to the total output .
the @xmath96 interactions produce the most neutrinos , so the maximum allowable neutrino flux comes from equal contributions of @xmath96 and @xmath97 and the minimum flux from 100% @xmath97 .
the luminosity function @xcite derived from the egret survey and the results from the sas-2 experiment @xcite have been used to produce a diffuse neutrino background estimate .
figure [ agnspec ] shows the resulting neutrino fluxes from a number of different models .
note that the fluxes are multiplied by @xmath98 to reduce the range of the plot .
curves labelled sp ( dot - dashed lines ) are predictions from the earlier szabo and protheroe models .
sdss92 is the stecker _
et al _ @xmath97 flux , while ss95 is the sum of the quasar core and blazar jet contributions in the stecker and salamon model ( dashed lines ) .
the new protheroe model is shown in solid lines for the limiting cases , the minimum expectation coming from @xmath97 interactions only and the maximum from equal contributions from @xmath96 and @xmath97 interactions .
the major backgrounds to searches for agn fluxes come from the decay of kaons , pions and muons produced from the interactions of cosmic rays in the earth s atmosphere .
these fluxes show an angular dependence ; in the horizontal direction the fluxes at the surface are greatest as the kaons , pions and muons traverse more atmosphere and have more time to decay .
the results of lipari @xcite are shown .
another potential background comes from the prompt decay of charmed particles produced in cosmic ray interactions in the atmosphere .
the expected neutrino fluxes should have a flatter energy spectrum and not show an angular dependence as the particles mostly decay before they can interact .
zas , halzen and vzquez @xcite have calculated the expected neutrino fluxes from charmed meson decay under various assumptions of the charm production cross section and have computed muon rates in downward looking detectors .
gaisser , halzen and stanev @xcite conclude that model b is an upper limit to the neutrino fluxes in the region @xmath99 tev .
however , model d predicts higher fluxes at lower neutrino energies and gives overall the highest muon rates in downward looking detectors of the two cases .
i therefore use model d to predict the upper limit on the muon rate from charm production . as a lower limit ,
i take the prediction of thunman , ingelman and gondolo @xcite which gives neutrino fluxes nearly 100 times smaller than the zhv model d. it is clear that there is a large uncertainty in the potential contribution of charmed meson decay to the neutrino induced muon fluxes .
the implications of this on the detectability of the agn fluxes will be addressed in the next section .
cerenkov detectors ( dumand , amanda , nestor , lake baikal ) detect upward going muons from charged current interactions of neutrinos in the earth .
the muons can be produced many kilometres from the detector and still survive to be detected via their cerenkov emissions in water or ice .
calculations of the expected muon fluxes from agn @xcite have been made using various parton distributions and muon energy loss assumptions . in their original cross section paper @xcite , frichter , mckay and
ralston calculated the rates of charged lepton production in a cubic kilometre volume of ice and noted that despite the increased attenuation with their new cross section , the increased interaction probability more than compensated for this , resulting in increased lepton production .
in later work @xcite , they calculated the specific rate expectations for a radio receiver in antarctic ice , detecting radio cerenkov radiation from electrons .
the highest of the szabo and protheroe fluxes @xcite and the stecker _
et al _ flux @xcite were used , the results suggesting these fluxes would be readily detectable with a single antenna buried at a depth of 600 metres .
zas , halzen and vzquez @xcite calculated the detection rates of neutrino induced horizontal air showers and underground muons , again for sdss and sp fluxes .
these calculations included the expected rates from charm induced neutrinos .
the cross sections were calculated with the kmrsb-5 @xcite parton distributions which give total cross sections of about the size of the reno @xmath89 estimate .
most recently , gandhi , quigg , reno and sarcevic @xcite ( gqrs ) , have evaluated the cross sections for a variety of parton distributions and have then used the cteq3 parton distributions @xcite to estimate muon rates from astrophysical sources . due to the relative differences in the final cross sections , the cteq3 partons will yield muon rates closer to those using reno @xmath89 than mrs g. the calculation of muon fluxes induced by neutrino interactions in the earth is complicated by neutrinos undergoing neutral current interactions in the earth ( which reduce their energies but do not remove them from the beam ) before they finally interact to produce a muon .
the final number of neutrinos in a given energy bin after passage through the earth is reduced by both charged and neutral current interactions , but some of the total loss is offset by originally higher energy neutrinos having moved down in energy due to neutral current interactions .
an effective total cross section can be calculated , which is dependent on the original source spectrum .
this approach was followed by berezinsky _
et al _ @xcite where the total cross section was taken as the sum of neutral and charged current cross sections , with a correction term included to account for the source dependent neutral current neutrino regeneration effect .
as an alternative approach to the neutral current effect , i have developed a monte carlo simulation @xcite to model the passage of the neutrinos through the earth , taking into account the relative likelihood of neutral and charged current interactions , and selecting the final neutrino or muon energy from the energy dependent @xmath15 distributions .
a simple weighting system was devised ( and generalized to account for neutrinos that undergo none , one , two or more neutral current scatters before a charged current interaction ) to keep the computing time reasonable .
neutrinos and anti - neutrinos were treated separately using the appropriate cross sections .
the column density of nucleons to be traversed by a neutrino as a function of nadir angle is calculated using the preliminary reference earth model @xcite , which describes how the density of the earth varies with distance from the centre .
the final energy of a muon reaching the detector is determined using the muon propagation code of lipari and stanev @xcite .
the muon energy losses due to bremsstrahlung @xcite , @xmath100 pair production @xcite and photonuclear interactions @xcite are treated stochastically in order to give the correct final fluctuations in muon ranges .
ionization losses are also taken into account .
only muons which are produced within the final tens of kilometres from the detector will actually survive to the detector , so for these calculations i use standard rock
( @xmath101 g @xmath102 ) for the propagation .
i have calculated the expected muon rates using the cross sections reno @xmath89 , fmr and mrs g resulting from the parton distributions described in section [ partons ] which are derived from the latest accelerator structure function data .
the higher cross sections result in greater attenuation of neutrinos within the earth , but this effect is offset by a greater probability of muon production by the surviving neutrinos in the final kilometres of rock before a detector , leading to greater expected muon rates .
it is important to remember that the effective area of a detector will have an energy and arrival angle dependence and that the overall detection rate will depend upon the energy and angular spectrum of muons incident upon the detector .
for example , the effective area for muon detection ( averaged over arrival direction ) @xcite by the dumand - ii detector ( @xmath103m@xmath104 geometric area ) varies from about @xmath105m@xmath104 at 10 gev ( the array does not trigger on all muons ) up to @xmath103m@xmath104 at 5 tev and goes up to @xmath106m@xmath104 at @xmath107 gev ( the array triggers on muons that pass outside the detector volume ) . the muon rates calculated here assuming a spherical detector with a constant 20000m@xmath104 effective area for all arrival directions and all muon energies will therefore be approximate lower bounds on the numbers of muons detected for muons above about 1 tev ( near the energy at which the effective area equals the geometric area ) . due to the increase in effective area at higher energies
, the muon rates for flatter spectra will be enhanced over those of steeper spectra compared to the numbers presented here which are calculated for an energy independent effective area . keeping this in mind
, we examine the predicted muon rates induced by agn neutrinos .
the integral muon rates expected in a downward looking detector ( @xmath108 field of view ) are shown in figure [ intmuonrates ] for the three different cross sections and nine different neutrino spectra of figure [ agnspec ] .
the predicted muon fluxes span several orders of magnitude .
the sp94 flux exceeds the atmospheric background for a muon energy cut of a few gev . at the other extreme
, the p96p@xmath93 does not exceed the atmospheric background until an energy of around 10 tev . the wide variation in the expected number of muons from charm induced neutrinos is evident .
the tig neutrino flux induces a very low rate of muons , around 1/10 of the lowest agn flux ( p96p@xmath93 ) , and should not pose a background problem at any muon energy cut .
however , the zhv d neutrino flux results in a muon rate exceeding that of the sdds92 flux for muon energies up to @xmath88400 gev , and roughly equal to that of the conventional atmospheric and p96p@xmath93 fluxes at a muon energy cut of 10 tev .
this great uncertainty suggests that there would be difficulties in identifying the origins of any observed muons should the lowest agn flux be correct .
the results presented here are not directly comparable to those of gandhi _ et al _ @xcite , as the neutrino source spectra used in both cases are not identical .
gqrs made a broken power law approximation to the sp94 ( b=1 ) neutrino flux ; this yields muon rates less than the sp94 curve in figure [ intmuonrates ] , which comes from using the highest of the sp94 fluxes .
gqrs also calculated muon rates for the ss95 neutrino flux , but only for the quasar core component , whereas i have used the sum of quasar core and blazar jet components .
also , gqrs treat neutral currents by taking two cross sections : the first the sum of charged and neutral current cross sections , which gives a lower limit to the muon rate ; the second the charged current cross section alone , which will give an upper limit to the muon rate . to check the consistency of the two sets of results ,
i have used the forms of the neutrino fluxes as implemented by gqrs with the mrs g and reno @xmath89 cross sections and find that the muon rates obtained are consistent with the differences in cross sections in the two sets of work .
the sp94 and sp92max fluxes have already been excluded by an analysis of the energy distribution of horizontal neutrino - induced muons in the frjus experiment @xcite .
the frjus detector was a fine grain tracking calorimeter operated deep underground , with the ability to accurately determine the energy of the through going muons from the energy deposited from electromagnetic showering . at near horizontal angles the background atmospheric muon flux
is sufficiently suppressed by the rock overburden to allow a measurement of the rate of atmospheric and agn induced muons . the number of muons ( producing an observed energy loss in the detector greater than the maximum observed energy loss ) expected from each agn spectrum has been calculated @xcite taking into account the rock thickness , muon propagation and the increasing efficiency of the detector with increasing muon energy .
the high fluxes sp94 and sp92max predict more muons than the 90% confidence level upper limit of 2.3 muons and are therefore excluded by this experiment . all the other fluxes considered predict less than 2.3 muons and are therefore not excluded by the frjus result .
the angular distributions of the muon fluxes for muon energies above @xmath109 gev and @xmath110 gev are shown in figure [ muonang103 ] for a collecting area of 20000 m@xmath104 , which corresponds to that of the proposed nine string dumand - ii array .
the increases in the overall muon rates predicted by the fmr cross section are not uniform over arrival direction .
it is clear that most of the increases come from the highest nadir angles ( muons coming from angles 0 to @xmath8830 degrees down from the horizon ) where the thickness of the earth traversed by the neutrinos is smaller and attenuation less important while the likelihood of muon production near the detector is greater .
the rate of directly upgoing muons is very similar for all three cross sections considered - the increased rate of muon production in the final kilometres of rock is balanced by increased neutrino attenuation in the earth .
the increase in muon rates comes from arrival directions where surface or shallow detectors are most susceptible to noise contamination from downward going atmospheric muons mis - reconstructed as horizontal or upward going
. however , deep underwater detectors which are better shielded from atmospheric muons will be sensitive to these increased near horizontal fluxes .
for example , the dumand - ii detector is expected to be sensitive to muons from downgoing neutrinos up to @xmath8820 degrees above the horizon .
.[ratetable]integral muon rates in a @xmath103m@xmath104 downward looking detector for nadir angles @xmath111 ( most upgoing ) and @xmath112 ( nearer horizontal ) for various neutrino spectra , cross sections and minimum muon energies . [ cols="^,^,^,^,^,^,^,^ " , ] here we consider only those neutrino predictions that have not been ruled out by the frjus experiment .
the total muons expected for nadir angles greater and less than 60 degrees as a function of muon energy cut and cross section for each neutrino source are summarized in table [ ratetable ] .
it is evident that the highest of these fluxes may be observable over the atmospheric background for a 100 gev muon energy cut ( @xmath88300 - 500 events per year above @xmath882000 background ) but that the smaller fluxes ( 10 - 100 events per year ) would not . for a muon energy cut of 1 tev , the atmospheric background is significantly reduced and the fluxes p96@xmath113 , sp92min and ss95 q+b are now greater than the background and should be observable .
the lower sdss92 flux would produce @xmath8855 - 70 events per year above a 225 background and should be observable over several years but an uncertain contribution from charm induced neutrinos ( @xmath881 - 50 muons ) would complicate matters
. however figure [ muonang103 ] shows that the charm induced muons have a flat angular distribution and this might aid in identification .
the p96@xmath97 is well below the background and given the charm uncertainty , would seem not to be detectable . moving to a 10 tev cut
reduces the atmospheric background to @xmath885 per year , making all the fluxes readily observable , except for the lower bound of p96@xmath97 .
this latter flux , the atmospheric background and the highest charm decay estimate ( zhv d ) produce roughly equal numbers of muons . given that the charm induced muon rate
could actually be practically zero ( as table [ ratetable ] shows ) , this would make a differentiation between charm induced and agn induced muon fluxes difficult , if only a small excess above the expected conventional atmospheric rate was seen .
one could not simply do a counting experiment as the excess above the conventional atmospheric background could come from either of two sources .
an examination of the angular distributions or differential energy spectra of the detected muons would be needed .
the p96@xmath97 , charm d and conventional atmospheric fluxes produce different angular distributions , shown in figure [ 10tevblow ] .
the shape of the atmospheric neutrino induced muon curve reflects primarily the angular distribution of the source neutrinos .
the isotropic charm - induced neutrinos produce a flatter angular distribution ; but with muons resulting from higher energy neutrinos the angular shape shows evidence of some neutrino absorption .
the muons induced by the isotropic p96@xmath97 flux show a strong angular dependence due to contributions from much higher neutrino energies .
the three angular distributions are quite different for nadir angles greater than @xmath114 ( @xmath115 ) . a detector located deep enough such that atmospheric muons are not a problem should be sensitive to these differences , provided enough muons were measured to reveal the angular shape ( requiring many years exposure for a 20000 m@xmath104 detector or a much larger detector ) .
surface or shallow detectors are affected by misreconstruction of downward muons , which restricts their acceptance aperture .
figure [ 10tevblow ] shows that for nadir angles less than @xmath114 ( @xmath116 ) the angular distributions of all three spectra are the same shape ( fairly flat ) .
thus , even using a very large detector / long exposure time but with restricted aperture , the angular distribution alone is not sufficient to determine how much of a small excess of muons was due to agn , because of the large uncertainty in the charm induced neutrino flux . measuring the energy spectrum of the observed muons may allow a differentiation between source spectra .
the p96@xmath97 and charm d fluxes give equal numbers of muons above 10 tev ; however @xmath8895% of the charm d muons are in the energy range 10 - 100 tev .
the p96@xmath97 muons are split roughly equally between the range 10 - 100 tev and the region above 100 tev .
a comparison of the excesses in each of these energy regions may aid identification .
however , we are dealing with very low count rates where statistical fluctuations and the uncertainty in the conventional atmospheric neutrino flux will also play a role . as we have just seen , determining the origin of a small excess of muons above the conventional atmospheric expectation relies on looking for angular or energy distribution characteristics .
the p96@xmath97 and charm d spectra produce sufficiently different muon distributions to allow them to be distinguished .
the integral muon energy spectrum ( figure [ intmuonrates ] ) for the p96@xmath117 flux is similar to the charm d muon spectrum , although it produces an absolute level of muons about a factor of 10 bigger .
also the angular distributions for the p96@xmath117 flux have near the same slopes as the charm d angular distributions .
this shows that the higher the contribution from @xmath96 interactions to the diffuse agn flux , the more the muon energy and angular distributions become similar to those from the charm induced neutrinos .
an agn spectrum made up of only 10% @xmath96 and 90% @xmath97 contributions produces a differential muon spectrum with @xmath8885% of the muons ( above 10 tev ) in the energy range 10 - 100 tev , and an absolute level only a few times bigger than the charm d muon spectrum .
this shows that a small observed muon excess showing angular and energy distribution characteristics typical of the charm induced muon spectra could actually be due to a combination of an agn spectrum with a modest @xmath96 contribution with a low overall flux level ( of order of the p96@xmath97 flux given here ) and a charm spectrum lower than the charm d spectrum considered here .
in this discussion i have considered models which produce low numbers of muons in large detectors .
clearly , there will be unambiguous signatures of agn fluxes if the more optimistic fluxes are correct .
excesses well above those predicted by the charm d flux will be clear evidence of an extra terrestrial origin . the p96@xmath97 flux , although of the same level as the charm d flux , has distinctive characteristics ( flatter muon energy spectrum and steeper angular distribution ) that set it apart from the charm induced muons
. however , agn spectra with only modest @xmath96 contributions produce muon distributions with similar characteristics to the charm induced muons . interpreting a small
observed excess of muons above the conventional atmospheric expectation will require a better understanding of the production of neutrinos from the prompt decay of charmed particles in the atmosphere .
in this work i have calculated the expected agn neutrino induced muon rates in large detectors and have compared them to the backgrounds expected from atmospheric neutrinos ( produced by the decay of pions , kaons , muons and the prompt decay of charmed mesons ) .
a monte carlo simulation has been used which correctly accounts for the effects of neutral current scattering and uses cross sections calculated with the fmr and mrs g parton distributions , which were derived from recent accelerator data .
these higher cross sections result in greater muon rates due to a greater probability of muon production in the final kilometres before a detector , but for incident neutrino directions in the 120 degree aperture about the nadir this effect is largely offset by increased neutrino absorption .
the increases in muon rates come from nadir angles greater than 60 degrees , the region where surface or shallow detectors are most susceptible to atmospheric muon contamination .
the very deepest detectors will be able to observe these increased fluxes .
the lowest neutrino flux considered , due to @xmath97 interactions only in the protheroe agn model , would produce roughly 6 muons per year above 10 tev in a 20000 m@xmath104 detector , which is about the same as that expected from the conventional atmospheric sources and the same as expected from the highest estimates from charmed meson decay .
the lowest neutrino induced muon flux due to charmed meson decay gives practically negligible muon rates , which suggests that telling these sources apart may be difficult
. however , the p96@xmath97 agn flux produces muons with angular and energy characteristics that would allow it to be distinguished from the charm and conventional atmospheric backgrounds . on the other hand , an agn flux with around 10% or more contribution from @xmath96 interactions produces muon angular and energy distributions similar to those of the charm induced muons .
an observation of muons seemingly consistent with the level of the charm d flux could actually be due to an agn spectrum with a modest @xmath96 contribution and an overall flux level around 1/10 th of the p96@xmath117 flux , in combination with a charm induced flux consistent with the tig model . a better understanding of the charm production mechanism in the atmosphere
will then be required to interpret such an observation of a small number of muons above the conventional atmospheric background .
this discussion has centred around a possible worst case scenario ( low agn fluxes and uncertain charm fluxes ) .
the larger fluxes considered in this work would be readily observable in a 20000 m@xmath104 detector , using energy cuts of around 1 - 10 tev .
the planned deployments or extensions of the dumand , amanda , nestor and lake baikal experiments should herald the birth of observational extra - terrestrial neutrino astronomy , if the more optimistic agn models and neutrino fluxes considered in this work are near to correct .
i thank f. steffens , w. melnitchouk , a. thomas and g. frichter for useful discussions on qcd and neutrino cross sections and r. protheroe for clarifying aspects of agn and atmospheric neutrino production .
i thank b. dawson and r. clay for useful discussions , encouragement and careful reading of manuscripts and m. roberts , t. porter and j. reid for their advice throughout the course of this work .
an anonymous referee is also thanked for helpful comments on the original manuscript . | new information on the structure of the nucleon from the hera @xmath0 collider leads to higher neutrino cross sections for the processes @xmath1 needed to calculate the expected rates of astrophysical neutrino induced muons in large detectors either under construction , or in the design stage .
these higher cross sections lead to higher muon rates for arrival angles where neutrino attenuation in the earth is less important .
on the other hand , new estimates of agn neutrino fluxes suggest that the expected muon rates in these detectors may be much lower than previously calculated .
i use the new cross sections to calculate the expected muon rates and angular distributions in large detectors for a variety of agn models and compare these rates with the atmospheric neutrino backrounds ( from both conventional decay channels and the prompt charmed meson decay channels ) .
if the lowest flux estimates are correct , there may be diffculties in determining the origin of a small excess of muons , due to the large uncertainty in the rate of the charmed meson neutrino production channel .
however , the more optimistic agn neutrino fluxes should be detectable in proposed detectors , such as dumand - ii , amanda , nestor and lake baikal . |
in our previous paper @xcite we proposed to use an invariant link formulation to investigate some properties of two dimensional ( @xmath1 ) models in the low temperature limit .
we have argued that this approach is more reliable for calculation of asymptotic expansions of invariant functions in cases when the mermin - wagner theorem forbids spontaneous symmetry breaking in the thermodynamic limit .
we have calculated the first order coefficient of the free energy of @xmath2 model which turned out to coincide with the conventional answer .
the starting point of calculations is the following partition function of @xmath1 @xmath0 model @xmath3 \ , \label{lpf}\ ] ] where the jacobian @xmath4 is a product of @xmath0 delta - functions taken over all plaquettes of @xmath1 lattice @xmath5 \ . \label{jacob}\ ] ] the sum over @xmath6 is sum over all representations of @xmath0 , @xmath7 is the dimension of @xmath6-th representation . the @xmath0 character @xmath8 depends on a product of the link matrices @xmath9 along a closed path independent loops . ] : @xmath10 one of the main technical observations made in @xcite is that at large @xmath11 the @xmath0 delta - function can actually be replaced by the dirac delta - function so that partition function becomes @xmath12 \prod_{p , k}\delta ( \omega_p^k ) \ , \label{lpfd}\ ] ] where @xmath13 is a plaquette angle @xmath14 such a naive replacement is of course plagued by the problem of @xmath15 zero modes for auxiliary fields therefore the delta - function in ( [ lpfd ] ) should be correspondingly regularized .
this can be done , for instance , by introducing the heat - kernel into the expression for the partition function instead of the @xmath0 delta - function .
this procedure is equivalent to introducing a mass term for auxiliary fields .
direct calculations show however that zero modes decouple from the large-@xmath11 expansion in the @xmath2 model @xcite ( we could not generalize that consideration for arbitrary @xmath0 , though ) . in what follows we work with massless green functions omitting zero modes from all lattice sums similarly to @xmath2 case .
all general expressions remain valid if one works with mass regulator term as mentioned above .
since all logarithmic divergences cancel we expect that the convergence to the thermodynamic limit ( tl ) is uniform : in all cases the final result can be expressed in terms of link functions @xmath16 , @xmath17 introduced in @xcite and standard @xmath18-functions . for @xmath0 model the expansion itself
is done precisely like in @xcite for @xmath2 .
we obtain for the first order coefficient of the free energy @xmath19 this result agrees with the result of the conventional pt @xcite .
an essential motivation of our investigations is to study the problem of the uniformity of the low - temperature expansion in @xmath1 models , the question addressed in @xcite . before direct attack of this problem
we have decided we need to make our approach more suitable for practical computations . also , it is necessary to establish some general properties of the expansion , for example the path independence of the correlation functions and the infrared finitness of the expansion .
we address these and some other problems in the present article .
we give a general expression for the second order coefficient of fixed distance correlation function in terms of green functions and calculate this coefficient for distance 1 .
our final result agrees with eq.([c1fe ] ) .
let us consider two - point correlation function @xmath20 where @xmath21 if along the path @xmath22 the link @xmath23 goes in the positive direction and @xmath24 , otherwise .
obviously , it does not depend on a shape of the path @xmath22 and any expansion in the link formulation must respect this property . in large-@xmath11 expansion every coefficient at a given order should be path independent .
this requirement generates certain relations on link green functions thus allowing for a deformation of any given path to any other one .
since this independence is guaranteed only by the jacobian @xmath4 , the check on such independence would be highly desirable since it can show whether we treat the jacobian correctly and whether all contributions are taken into account .
further , we re - analyze the free energy of @xmath25 models .
as is well known , large-@xmath11 expansion in @xmath25 non - abelain models is non - uniform in the volume , therefore our expansion has to explain this feature . as for now
, all our results agree completely with the standard approach to the low - temperature expansion in @xmath1 models .
nevertheless , we find that the expansion in the link formulation also suffers from ambiguities previously found in the standard expansion @xcite .
again , problem arises only for non - abelian models and only starting from @xmath26 .
we mention this example in the discussion .
this paper is organized as follows . in the next section
we describe some properties of link green functions . in section 3
we study the correlation function in @xmath1 models .
we calculate the fixed distance correlation function in the @xmath27 model .
the general expansion of the second order coefficient of the correlation function in @xmath0 model is given in terms of green functions .
in this paper we compute this coefficient for distance 1 .
a proof of the path independence of the first and second order coefficients of the correlation function is given in section 4 . also , we prove here the infrared finitness of the second order coefficient .
section 5 is devoted to one dimensional models .
we analyze the low - temperature expansion of the free energy and explain why this expansion is non - uniform in the volume . in section 6 we study contribution of the holonomy operators to the low - temperature expansion in @xmath1 models and compare it with one dimensional case .
we summarize our results in section 7 .
the main building blocks of the low - temperature expansion in the link formulation are link green functions @xmath16 and @xmath17 introduced in @xcite . in this section
we describe some of their basic properties .
the functions @xmath16 and @xmath17 are defined as @xmath28 @xmath29 where link @xmath30 is defined by a point @xmath31 and a positive direction @xmath32 .
@xmath33 is a `` standard '' green function on the periodic lattice @xmath34 where we have denoted @xmath35 normalization is such that @xmath36 . in the momentum space @xmath16
reads @xmath37 using this representation it is easy to prove the following `` orthogonality '' relations for the link functions @xmath38 where the sum over @xmath39 runs over all links of @xmath1 lattice .
let @xmath22 be some path connecting points @xmath31 and @xmath40 and let @xmath41 be a path dual to the path @xmath22 , i.e consisting of the dual links which are orthogonal to the original links @xmath42 . for simplicity
let us consider a path consisting of links which point only in a positive direction ( see fig.1 for our notation of links on a dual lattice ) .
we then have @xmath43 where @xmath44 let @xmath45 be any closed path
. then @xmath46 where @xmath47 if both link @xmath23 and @xmath48 point in either positive or negative direction and @xmath49 if one ( and only one ) of links points in negative direction .
let us demonstrate this property for the simplest closed path , i.e. for a plaquette
. from fig.1 and definition ( [ gll1 ] ) we have on the dual lattice refer to the thermodynamic limit . ]
@xmath50 we obtain from ( [ clpath ] ) in case of plaquette @xmath51 one sees that @xmath16 satisfies the following equation @xmath52 for any link @xmath48 .
@xmath17 satisfies the lattice laplas equation @xmath53 we shall also use the following equality @xmath54 which is valid for any links @xmath55)-([orthog ] ) and ( [ glleq])-([gllsum ] ) are valid only up to @xmath56 corrections . ] .
throughout this section we work on a dual lattice and follow notations of the previous section . links @xmath57 denote links belonging to the path @xmath41 . in our previous paper
we considered low - temperature expansion of the free energy of the @xmath27 model . here
we want to demonstrate how conventional results for the fixed distance correlation function can be recovered in the link formulation .
we refer to @xcite for expansion of the gibbs measure and calculation of the generating functional in the link formulation . for the fixed - distance correlation function
we get @xmath58 coefficients @xmath59 are given by @xmath60 @xmath61 using orthogonality relation ( [ orthog1 ] ) and eq.([glxy1 ] ) we obtain @xmath62 + o(\beta^{-3 } ) \ . \label{xycorres}\ ] ] this result coincides with that quoted in the literature for @xmath63 model @xcite . in this subsection
we give general expression for the second - order coefficient of the fixed - distance correlation function in terms of green functions .
the generating functional is diagonal in group indices and coincides with that given in @xcite for @xmath2 .
again , the expansion of the gibbs measure and the jacobian is done precisely like in @xcite . expanding ( [ corf ] ) we write down @xmath64
the first coefficient is given by @xmath65 and coincides with the conventional result .
we split the second coefficient into three pieces @xmath66 @xmath67 describes contribution from the second order term of the correlation function and zero order term of the gibbs measure @xmath68 where @xmath69 @xmath70 \ , \label{q12}\ ] ] @xmath71 \ , \label{q13}\ ] ] @xmath72 \ . \label{q14}\ ] ] @xmath73 describes contribution of @xmath74 order from the expansion of correlation function and of @xmath75 order from the expansion of the jacobian .
this `` self - connected '' piece is given by @xmath76 \ , \label{q2}\end{aligned}\ ] ] where @xmath77 finally , there are contributions of the first order terms from the expansion of correlation function , the gibbs measure and the jacobian .
@xmath78 describes the corresponding connected pieces @xmath79 where @xmath80 @xmath81 \ , \label{q32}\end{aligned}\ ] ] @xmath82 we have denoted @xmath83 @xmath84 in all formulae link @xmath85 ( @xmath86 ) refers to one of four links attached to a given site @xmath31 ( @xmath87 ) ( see fig.1 ) .
as it stays , this expression for the second order coefficient is valid for any path if all links @xmath88 point in positive directions .
if one considers a path where some links point in a negative direction , one has to change a sign of the corresponding green functions .
here we compute correlation function for distance 1 . from ( [ suncorf])-([sunc2 ] )
we have @xmath90 where we took into account that @xmath91 and @xmath92 in this case . since only @xmath93 contributes to @xmath67 we
easily find @xmath94 to calculate @xmath78 we first note that one has to put @xmath95 , where @xmath39 is any fixed link . from ( [ q31 ] )
we obtain using ( [ orthog1 ] ) @xmath96 to calculate @xmath97 we make use the translation invariance to write down @xmath98 using again ( [ orthog1])-([orthog2 ] ) we find @xmath99 the same tricks applied to @xmath100 give @xmath101 where @xmath102 @xmath103 @xmath104 and @xmath105 have been calculated in @xcite and equal @xmath106 we thus find @xmath107 it leads to a final result @xmath108 which agrees with result for the free energy , eq.([c1fe ] ) .
we want to prove now that our representation for the correlation function is path independent . in the strong coupling region
such an independence is a simple consequence of the compact invariant integration which is lost in the large-@xmath11 expansion . since , however any coefficient at a given order of the large-@xmath11 expansion has to be path independent the direct proof of this property serves as an additional check of the correctness of the expansion .
the path independence is guaranteed by the jacobian which introduces the constraint on the plaquette matrix @xmath110 .
just this constraint imposes independence of the correlation function of the path and this constraint must be checked at each order of the large-@xmath11 expansion . to show that this constraint is fulfilled we write down @xmath111 the path independence requires @xmath112 first
, we note that the expansion for the correlation function given in section 3.2 remains valid also for @xmath113 where one has to identify links @xmath57 with links belonging to a given plaquette as shown in fig.1 .
besides , one has to change a sign in some green functions . if we write @xmath114 the signs should be changed in @xmath115 , @xmath116 , @xmath117 and @xmath118 .
the equality @xmath119 trivially follows from ( [ clpath ] ) .
we parametrize @xmath120 as @xmath121 since @xmath122 is arbitrary , the coefficients @xmath123 and @xmath124 should be equal to zero independently .
there are two contributions to @xmath123 coming from @xmath67 and from @xmath125 .
we find via direct computation @xmath126 - \nonumber \\ 2\sum_{b_1>b_2>b_3}\left [ g_{b_1b_2}+g_{b_2b_3}+g_{b_1b_3 } + 2(g_{b_1b_2}g_{b_2b_3}+g_{b_1b_3}g_{b_3b_2}+ g_{b_1b_2}g_{b_1b_3 } ) \right ] - \nonumber \\ 4\sum_{b_1>b_2>b_3>b_4}\left [ g_{b_1b_2}g_{b_3b_4}+g_{b_1b_4}g_{b_2b_3}+ g_{b_1b_3}g_{b_2b_4 } \right ] + \nonumber \\ \frac{1}{2}\sum_{b_1b_2}\sum_lg_{b_1l}g_{lb_2 } = 0 .
\label{c1gp}\end{aligned}\ ] ] @xmath124 consists of the following terms @xmath127 where we have introduced obvious definitions . for @xmath128 ( terms at @xmath129 in ( [ q11])-([q14 ] ) ) we get @xmath130 + \nonumber \\ 2\sum_{b_1>b_2>b_3}\left [ g_{b_1b_2}+g_{b_2b_3}+g_{b_1b_3 } + g_{b_1b_2}g_{b_2b_3}+g_{b_1b_3}g_{b_3b_2}+ g_{b_1b_2}g_{b_1b_3 } \right ] + \nonumber \\ 4\sum_{b_1>b_2>b_3>b_4 } \left [ g_{b_1b_2}g_{b_3b_4}+g_{b_1b_4}g_{b_2b_3 } \right ] = -4 \left ( 1+g_{l_1l_2 } \right ) ^2 = -\frac{16}{\pi^2 } \ .
\label{q1p}\end{aligned}\ ] ] @xmath131 is given in ( [ q2 ] ) and equals zero for any closed path .
this can be most easily seen from the cyclic permutations of links under the trace in ( [ gpl ] ) . from ( [ glleq ] )
follows that @xmath132 and @xmath133 only one term survives the summation over @xmath57 .
eq.([dleq ] ) gives @xmath134 performing all summations we finally get @xmath135 one sees that this connected piece cancels exactly contribution of the second order term from the expansion of the correlation function .
we get convinced that @xmath136 as a matter of fact , eqs.([clpath ] ) , ( [ glleq ] ) and ( [ dleq ] ) are short - hand statements of the independence of the correlation function of the path . using these equations one deforms any path expressing some link from a given plaquette through remaining ones .
the generalization of this proof to the product of link matrices along an arbitrary closed path is straightforward . in this subsection
we prove the infrared finitness of @xmath137 .
the only term which includes @xmath33 and hence could logarithmically diverge is @xmath138 and @xmath139 is given in eq.([i1 ] ) .
we want to show that @xmath33 in the last expression can actually be replaced by @xmath140 because all additive constants like logarithmic term in @xmath33 do not contribute to the sums in eq.([q33div ] ) .
thus , one has to prove that @xmath141 indeed , using ( [ orthog1 ] ) @xmath142 can be transformed to @xmath143 \left [ a(b_2b)-a(bb_2 ) \right ] \ , \label{str}\ ] ] where sum over @xmath39 runs over all links of the lattice and @xmath144 the equality @xmath145 follows from ( [ glleq ] ) and ( [ gllsum ] ) .
therefore , @xmath146 and eq.([q33div ] ) is infrared finite .
to analyze @xmath25 @xmath0 model we note that the formula for @xmath137 given in section 3.2 remains valid if we take for the link green function the following expression @xmath147 this equation is a trivial consequence of the fact that in the link formulation the @xmath25 model reduces to one link integrals .
then , it is straightforward to calculate , for example , the first order coefficient of the free energy .
we find @xmath148 which agrees with the expansion of the exact result in the tl . on the other hand
, it is well known that the low - temperature expansion in @xmath25 non - abelian models is non - uniform in the volume , in particular the conventional pt produces result different from the eq.([c1fe1d ] ) . to explain this non - uniformity we remind that on the periodic lattice one has to constraint a holonomy operator if one works in the link formulation .
working with @xmath1 models we have neglected this additional constraint since it seems to us rather unlikely that such global constraint may influence the tl in @xmath1 ( see next section for details ) .
this happens , however in @xmath25 model as we are going to show below .
the partition function is given by @xmath3 \ , \label{lpf1d}\ ] ] where @xmath4 introduces a global constraint on link matrices @xmath149 again , at large @xmath11 we replace the @xmath0 delta - function by the dirac delta - function , i.e. @xmath150 \ , \label{j1dd}\ ] ] where @xmath151 is defined as @xmath152 \ .
\label{cang}\ ] ] we omit all technical details which are exactly the same as in @xmath1 . for the first order coefficient of the free energy we find in the large volume limit
@xmath153 the second term on the right - hand side of the last formula comes from the expansion of @xmath4 and modifies the correct expression ( [ c1fe1d ] ) .
our result ( [ c11d ] ) disagrees with the one given in @xcite .
we think it is because the result of @xcite was obtained using the mass regulator term , i.e. the procedure which is known to give wrong answer even in a finite volume @xcite . to check the correctness of ( [ c11d ] ) we have compared it for @xmath154 with @xmath155 model @xmath156 where the second term comes from the hasenfratz term which survives the tl in @xmath25 .
one sees that results indeed coincide in @xmath157 . ] .
on @xmath1 lattice one should restrict two holonomy operators . in our previous analysis
we have neglected this restriction . since , however this global constraint influences the tl in @xmath25 if the low - temperature expansion is done in a finite volume we think it is instructive to see what happens with holonomies in two - dimensional models .
let @xmath158 be any given path winding around the whole lattice .
@xmath159 and @xmath160 are orthogonal to each other .
one has to introduce two global constraints into the partition function ( [ lpf ] ) @xmath161 \ , \label{j2dd}\ ] ] where @xmath162 is defined as @xmath163 \ . \label{cang2d}\ ] ] there are two types of contributions from @xmath164 .
the first one comes from the expansion of @xmath164 itself .
it is too cumbersome to be given here in full .
this contribution , however can be expressed only through link green functions and is proportional to the linear size of the lattice .
this is a reason why it survives the tl in @xmath25 .
correspondingly , in @xmath1 it vanishes like @xmath165 .
the second type is related to the modification of the generating functional .
namely , one should make the following replacement of the green functions @xmath166
@xmath167 @xmath168 in particular , the corresponding replacements should be made in formulae ( [ suncorf])-([i2 ] ) .
let us take for simplicity such paths @xmath169 which consist of links pointing only in one direction .
the coordinates of the corresponding links on the dual lattice are @xmath170 \ , \ b\in h_1\ ] ] and @xmath171 \ , \ b\in h_2 \ .\ ] ] it is easy to find @xmath172 @xmath173 in general we get for @xmath30 , @xmath174 @xmath175 on the other hand , one finds in the tl @xmath176 which leads to @xmath177 we thus have for new functions @xmath178 @xmath179 @xmath180 one sees that the only new term which could potentially survive the tl is the second term in the last expression .
this term is to be substituted into ( [ q33 ] ) and it leads to the computation of sums of the form @xmath181 the first term vanishes like @xmath165 because of 2 extra deltas .
the second and constant terms equal zero because of ( [ canc ] ) .
moreover , in general it is clear from ( [ gllhres])-([gxxhres ] ) that the holonomies may only contribute to the tl through the constant term in @xmath182 .
this is , however equivalent to non - cancelation of infrared divergences in some higher orders . in all other cases the holonomies
do not survive the tl .
in this paper we used the link representation to develop the low - temperature expansion of some @xmath1 models .
our main concern here was the investigation of a fixed distance correlation function in models with non - abelian @xmath0 global symmetry .
we have calculated the first and second order coefficients of large-@xmath11 expansion of such correlation in terms of green functions and have shown that they coincide with the conventional expansion , at least for the shortest distance .
we have demonstrated how the path independence of the correlation function manifests itself in our expansion and have proven such independence for the first two coefficients via direct calculations .
we have also shown which properties of the expansion guarantee its infrared finitness , at least in lowest orders . moreover , since in our expansion the source of such divergences is exactly localized ( they can only emerge from the expectation values of auxiliary fields which appear in even general power ) we think it should be possible to generalize the proof given in section 4.1 for higher orders .
it could lead to a lattice analog of david s theorem @xcite which states the infrared finitness of the weak coupling expansion of continuum models .
further , we have re - analyzed the low - temperature expansion of @xmath25 models in the link representation . in this representation
one should impose a global constraint on link matrices ( [ jacob1d ] ) .
this global constraint vanishes in the tl if this limit is taken before low - temperature expansion .
however , if expansion is done in a finite volume , the expansion of the holonomy operator , which imposes the global constraint does survive the tl .
it leads to the non - uniformity of the low - temperature expansion in one dimension .
we also have proven that it is not the case in @xmath1 : at least in the lowest orders the holonomies do not survive the tl and there is a good reason to believe that they do not in higher orders as well .
we thus find that the low - temperature expansion performed in the link representation coincides with the conventional pt , at least in the lowest orders .
in fact , it seems that these two expansions have to coincide up to arbitrary order .
if the conventional pt produces the correct asymptotic expansion in a finite volume , any other expansion is bound to reproduce the same coefficients when the volume is fixed . moreover
, this also refers to the tl if these coefficients are infrared finite .
originally ( following arguments given in @xcite ) we thought that the expansion around @xmath183 has somehow better theoretical status .
in particular , if the expansion around @xmath184 fails in the tl as the expansion around the non - existing in the tl saddle point , there seems to be a little doubt that the point @xmath183 is the only saddle point in the tl similarly to the abelian case .
it led us to a conclusion that the expansion in the link representation could be free of ambiguities found in @xcite ( see also discussion in @xcite-@xcite ) and which put under doubts that the conventional pt gives an asymptotic uniform in the volume for non - abelian models .
unfortunately , this our conclusion was wrong as the next example shows .
we consider this example as an superinstanton example @xcite adjusted for the periodic lattice in the link formulation .
since the calculations are very lengthy we only sketch the arguments leaving details for a future @xcite .
let us consider the system on the periodic lattice given by the partition function ( [ lpf ] ) with the only difference : we leave one plaquette , say @xmath185 , unrestricted , i.e. on the dual lattice we have ( @xmath186 ) @xmath187 \ . \label{jacobunr}\ ] ] it has to be clear that this modification should leave no memory in the tl .
for example , it is the case in the high - temperature expansion ( truly , we checked only the lowest order ) .
we are going to argue that while the coefficients of the low - temperature expansion of the @xmath27 model are indeed unchanged in the tl , it is not so for non - abelian models and the trouble starts from @xmath26 , precisely like in the superinstanton example .
the expansion is done precisely like with the standard partition function , the only modification concerns the generating functional .
it can be shown that one has to use the following green functions @xmath188
@xmath189 @xmath190 as the simplest examples , we give the expression for the average plaquette in the @xmath27 model up to the first order @xmath191 and the expression for an average link of the unrestricted plaquette up to the second order @xmath192 one sees that even links of unrestricted plaquette converge to the true tl , though the convergence is slow , like @xmath193 ( similarly to the pt with superinstanton bc ) .
the first order coefficient of the correlation function of the @xmath0 model also converges logarithmically to the tl value ( [ sunc1 ] ) .
however , it is not the case for the second order coefficient .
this can be most easily seen , e.g. from the expression for the function @xmath194 in eq.([dlun ] ) . in the case
when all sums converge to the tl uniformly , one gets in the tl the following expression @xmath195 which shows that there appear new contributions which do not vanish in the tl .
we have not finished calculations of all these contributions for links of restricted plaquette but it seems for us that even in this case one should expect a modification of the standard tl .
o. borisenko , v. kushnir , a. velytsky , hep - lat/9809133 .
y. brihaye , p. rossi , nucl.phys .
b235 ( 1984 ) 226 .
a. patrascioiu , e. seiler , phys.rev.lett .
74 ( 1995 ) 1920 .
s. elitzur , nucl.phys .
b212 ( 1983 ) 501 .
p. hasenfratz , phys.lett .
b141 ( 1984 ) 385 .
f. david , phys.lett .
b69 ( 1980 ) 371 .
f. david , phys.rev.lett . 75
( 1995 ) 2626 .
f. niedermayer , m. niedermaier and p. weisz , questionable and unquestionable in the perturbation theory of non - abelian models , hep - lat/9612002 .
a. patrascioiu , e. seiler , phys.rev .
d57 ( 1998 ) 1394 .
o. borisenko , v. kushnir , in preparation . | we extend our previous analysis to arbitrary two dimensional @xmath0 principal chiral model in a link formulation .
a general expression for the second order coefficient of fixed distance correlation function is given in terms of green functions .
this coefficient is calculated for distance 1 and is proven to be path independent .
we also study the weak coupling expansion of the free energy of one dimensional @xmath0 model and explain why it is non - uniform in the volume .
further , we investigate the contribution of holonomy operators to the low - temperature expansion in two dimensions .
all our results agree with the conventional expansion . nevertheless , we give some example which indicates that the expansion in the link formulation can also suffer from ambiguities previously found in the conventional perturbation theory .
-5 mm * low - temperature expansion in link formulation .
ii * + + + |
local lorentz invariance ( lli ) is an important foundation of modern physics and has been a subject of many stringent experimental and observational tests.@xcite however , a number of theories unifying gravity with the standard model of particle physics suggest possible violation of lorentz symmetry . while the suggested lli - violation energy scale is much larger than the energy currently attainable by particle accelerators , it might be accessible with precision measurements at low energy .
therefore , high - precision tests of lli with photons and particles ( protons , neutrons , electrons ) may provide insight into possible new physics and set limits on various theories .
experimental breakthroughs in atomic , molecular and optical ( amo ) physics , including laser cooling and trapping of atoms , attainment of bose - einstein condensation , optical frequency combs , and quantum control led to extraordinary advances in the control of matter and light . these achievements , coupled with dramatic improvements in precision time and frequency metrology , measurement techniques such as atomic interferometry and magnetometry , and advances in first - principles atomic and molecular theory enabled a plethora of new applications of amo , including novel ways to test the fundamental physics laws .
the availability of trapped ultracold atoms and ions , subject to precise interrogation and control , provided for new opportunities for tests of lorentz symmetry .
a diverse set of amo lorentz - symmetry tests involves experiments with atomic clocks,@xcite other precision spectroscopy measurements,@xcite magnetometers,@xcite electromagnetic cavities,@xcite and quantum information trapped ion technologies.@xcite a cold cs atom clock test of lorentz invariance in the matter sector was carried out in ref . , setting the best limits on the tensor lorentz - violating coefficients for the proton . in 2015 , an experiment with a pair of trapped calcium ions improved bounds on lli - violating standard - model extension ( sme ) coefficients @xmath2 for electrons by a factor of 100 demonstrating the potential of quantum information techniques in the search for physics beyond the standard model.@xcite the same experiment can be interpreted as testing anisotropy in the speed of light , improving a previous such bound@xcite by a factor of 5 , with the sensitivity similar to more recent work reported in ref . .
lorentz - violation tests are analyzed in the context of the phenomenological framework known as the sme , which is an effective field theory that augments the standard - model lagrangian with every possible combination of the standard - model fields that is not term - by - term lorentz invariant , while maintaining gauge invariance , energy - momentum conservation , and observer lorentz invariance of the total action.@xcite violations of lorentz invariance and the einstein equivalence principle in bound electronic states result in a small shift of the hamiltonian that can be described by@xcite @xmath3 where we use atomic units , @xmath4 is the momentum of a bound electron , and @xmath5 is the speed of light .
the parameters @xmath6 , @xmath7 , and @xmath8 are elements in the @xmath9 tensor which characterises hypothetical lorentz violation in the electron sector within the sme.@xcite the nonrelativistic form of the @xmath10 operator is @xmath11 . predicting the energy shift due to lli violation involves the calculation of the expectation value of the above hamiltonian for the atomic states of interest .
therefore , the shift of the ca@xmath0 @xmath12 energy level due to the @xmath9 tensor depends on the values of @xmath13 and @xmath14 matrix elements . the frequency difference ( in hz ) between the shifts of the @xmath15 and @xmath16 substates of the @xmath12 manifold
was calculated in ref .
: @xmath17 the basic idea of the ca@xmath0 experiment is to monitor this energy difference between the magnetic @xmath18 and @xmath19 substates of the @xmath12 @xmath20 manifold over time to set the limit on potential violation of lli .
therefore , the shift of the ca@xmath0 @xmath12 energy level due to the @xmath9 tensor depended only on the value of @xmath14 matrix element , as the contribution of the scalar term canceled for the states of the same @xmath20 manifold .
superposition of two ions prepared in a decoherence - free subspace @xmath21 where @xmath22 represents the @xmath12 state with @xmath23 and @xmath24 for the first and second ion , respectively , allowed elimination of the largest potential systematic uncertainty cased by the fluctuation of the magnetic field .
details of the experiment are given in ref . .
further improvement of lli violation limits calls for a system with a long - lived ( or ground ) state that has a large @xmath25 matrix element .
we have carried out a systematic study of this quantity for various systems and identified general rules for the enhancement of the reduced matrix elements of the @xmath26 operator.@xcite our calculations for @xmath27 states in ba@xmath0 and yb@xmath0 , which are heavier analogues of ca@xmath0 , found only a small increase of the @xmath26 matrix elements in comparison with the ca@xmath0 case .
however , yb@xmath0 has another metastable level , @xmath28 @xmath29 with the @xmath30 matrix element that is over an order of magnitude larger than for the @xmath27 states .
we find that deeper localization of the probe electron leads to enhanced sensitivity to the tensor lorentz violation .
our study has shown that @xmath31 of @xmath320.8 a.u or below for the corresponding electron is a good indicator of the large value of the reduced @xmath26 matrix element .
this condition is satisfied for the @xmath33 hole states , such as yb@xmath0 state considered here , or for highly charged ions with @xmath34 valence electrons and degree of ionization @xmath35 .
the reduced @xmath26 matrix elements in yb@xmath0 and sm@xmath36 are 135 a.u . and 149 a.u .
, respectively , in comparison of 9.3 a.u for the @xmath12 state in ca@xmath0 .
the yb@xmath0 @xmath28 @xmath29 state also has an exceptionally long lifetime on the order of several years.@xcite therefore , the ramsey duration of the proposed experiment with yb@xmath0 is not limited by spontaneous decay during the measurement as in ca@xmath0 case .
the electric - octupole e3 transition between the @xmath28 @xmath37 excited state and the ground state is used as the basis for the optical atomic clock with the single trapped yb@xmath0 ion , which presently has the lowest uncertainty among the all of the optical ion clocks.@xcite yb@xmath0 ions are also used in quantum information research.@xcite as a result , experimental techniques for precision control and manipulation of yb@xmath0 atomic states are particulary well developed making it an excellent candidate for searches of lorentz - violation signature .
we estimated that experiments with the metastable @xmath28 @xmath37 state of yb@xmath0 can reach sensitivities of @xmath38 for the @xmath2 coefficients,@xcite over @xmath39 times more stringent than current best limits.@xcite moreover , the projected sensitivity to the @xmath40 coefficients will be at the level of @xmath41 , below the ratio between the electroweak and planck energy scales.@xcite similar sensitivities may potentially be reached for lv tests with highly charged ions , given future development of experimental techniques for these systems . | a lorentz - symmetry test with ca@xmath0 ions demonstrated the potential of using quantum information inspired technology for tests of fundamental physics .
a systematic study of atomic - system sensitivities to lorentz violation identified the ytterbium ion as an ideal system with high sensitivity as well as excellent experimental controllability .
a test of lorentz - violating physics in the electron - photon sector with yb@xmath0 ions has the potential to reach levels of 10@xmath1 , five orders of magnitude more sensitive than the current best bounds .
similar sensitivities may be also reached with highly charged ions . |
the dynamical state and mass distribution in the central regions of elliptical galaxies provide clues to the formation and evolutionary histories of these galaxies ( see merritt 1999 for a review ) .
consequently , the existence of central black holes has been the target of intense scrutiny ( see kormendy & richstone 1995 for a review ) . in this paper
, we study the elliptical galaxy ngc 3379 using both _ hubble space telescope _ and ground - based photometric and kinematic data in order to understand its central dynamics . at ground - based resolution
ngc 3379 is a prototypical elliptical ( e1 ) galaxy with m@xmath6 ( faber _ et al . _
1997 ) at a distance of 10.4 mpc ( ajhar _ et al . _
lauer _ et al .
_ ( 1995 ) classify it as a `` core '' galaxy a galaxy that has a break in the surface brightness profile but still maintains a rising profile into the smallest measured radius .
faber _ et al .
_ ( 1997 ) hypothesize that core galaxies are associated with massive black holes .
however , few core galaxies have strong black hole detections : m87 ( harms _ et al . _
1994 ) , m84 ( bower _ et al . _
1998 ) , and n4261 ( ferrarese _ et al . _
. these detections result from gas dynamics , not stellar dynamics .
stellar dynamical evidence for central black holes in core galaxies is difficult to obtain since ( 1 ) core galaxies have low surface brightnesses , making kinematic observations difficult , ( 2 ) the lack of rotation complicates the dynamical modeling since velocity anisotropies potentially govern the dynamical support , and ( 3 ) stars near the center travel out to the core radius so their contribution to the central velocity profile is de - weighted ( kormendy 1992 ) .
ngc 3379 was chosen for this study because of its relatively high central surface brightness among core galaxies , and because of its otherwise normal morphological and dynamical structure . the proximity , brightness , and normalcy of ngc 3379 make it the object of many ground - based photometric ( de vaucouleurs & capaccioli 1979 , lauer 1985 , capaccioli _ et al . _ 1987 ,
capaccioli _ et al . _
1991 ) and kinematic ( kormendy 1997 , kormendy 1985 , van der marel _ et al .
_ 1990 , bender _ et al . _
1994 , statler 1994 , statler & smecker - hane 1999 ) studies
. however , previous models for ngc 3379 have yielded inconsistent results . assuming that the departure of the light profile from the @xmath7 law signifies the existence of a stellar disk , capaccioli _ et al .
_ ( 1991 ) find that the most likely model is nearly face - on with an inclination of 31@xmath0 ( intrinsic e5 galaxy ) .
their conclusion was that ngc 3379 and ngc 3115 have the same intrinsic shape but are seen from different viewing angles . in their model for ngc 3379 the central region is nearly oblate , but at about @xmath8 ( @xmath9 ) the model becomes triaxial .
no dynamical information was used in this model .
a major uncertainty is that slight departures from a @xmath7 law , on which their model is based , may be due to tidal effects ( kormendy 1977 ) instead of a disk .
furthermore , there is no reason to expect galaxies to follow a @xmath7 law exactly .
kinematic information is required to determine unambiguously the inclination of a spheroidal system .
van der marel _ et al .
_ ( 1990 ) , using the ground - based velocity data of davies & birkinshaw ( 1988 ) and franx _ et al .
_ ( 1989 ) , conclude that the inclination of ngc 3379 is 60@xmath0 ; however the two datasets are not consistent and give slightly different velocity anisotropy parameters .
van der marel _ et al . _
used 2-integral flattened models with a parametrized form for the anisotropy .
no goodness - of - fit criteria were given , and thus we are not able to compare their results with those of others .
statler ( 1994 ) , using bayesian statistics on the same datasets , prefers models that have inclination greater than 45@xmath0 .
statler rejects capaccioli s conclusion that the galaxy is flattened and triaxial at the 98% confidence level , but his results could accommodate those of van der marel _ et al .
_ statler uses only the mean velocity measured along four position angles ; the radial variation in the velocity or the dispersion are not considered .
furthermore , statler uses data only outside 15 .
however , statler has obtained new , high s / n data ( statler & smecker - hane 1999 ) , which he will present in a future paper .
the studies above use low s / n , sparse kinematic data , or no kinematic data at all . in this paper , we report results from observations of the full line - of - sight velocity distribution ( losvd ) at various position angles from high s / n ground - based data and data from hst .
the data are fit to axisymmetric , three - integral orbit - based models .
richstone _ et al .
_ ( 1999 ) describe the modeling technique in a companion paper .
we exploit the full losvds _ directly _ in the modeling .
alternatively , we could use the moments of the losvds ; however , the full shape of the losvd contains important information on the underlying dynamics , and it is desirable to incorporate the full losvd when the spectra have sufficient signal - to - noise to determine this reliably .
in fact , our result for ngc 3379 critically depends on the shape of the central losvd .
cretton _ et al .
_ ( 1999 ) and van der marel _ et al._(1998 ) present a very similar modeling technique , where the only difference from ours is that we fit the full losvd as opposed to their moment fitting procedure .
there are other programs that also use the higher moments of the losvd as constraints ( dejonghe , 1987 ; gerhard 1993 ; rix _ et al . _
1997 ) . obtaining the full losvd at many position angles is the maximum obtainable kinematic information and has now become routine ( _ e.g. _ bender _ et al .
_ 1994 , carollo _ et al . _ 1995 ,
gebhardt & richstone 1999 ) . in 2 , we describe the photometric and kinematic data for ngc 3379 . in 3
, we describe the models and present results in 4 .
discussion is given in 5 .
we obtained hubble space telescope ( hst ) observations of ngc 3379 , consisting of four 500s f555w ( @xmath10-band ) and four 400s f814w ( @xmath11-band ) wfpc2 images , on 1994 november 19 . centering
the galaxy in the high - resolution pc1 camera ( 0.0455 pixel@xmath12 ) provided a total signal in the central pixel of @xmath13 in both @xmath10 and @xmath11 .
we deconvolved the summed images with 40 lucy - richardson iterations .
the surface brightness , ellipticity , and position angle of the major axis were obtained using the same procedure as in lauer _ et al .
_ ( 1995 ) and are presented in table 1 .
0.2 cm 0.3 cm the hst profile reaches only about 12 from the center and must be extended to larger radii using ground - based photometry .
peletier _ et al .
_ ( 1990 ) have obtained ground - based @xmath14 photometry for ngc 3379 over the radius range 4170 , and capaccioli _ et al . _ ( 1990 ) have @xmath15 photometry out to 500 ; we must therefore estimate the transformations to @xmath10 and @xmath11 using the overlap region with the hst data . the data from peletier _ et al .
_ and capaccioli _ et al . _
agree well , and since peletier _ et al .
_ provide several colors , we use their results .
the colors in the overlap region , based on the average differences there , are @xmath16 and @xmath17 .
we have ignored possible variations in the colors as a function of radius . using these colors ,
we combine the hst and ground - based data to provide the full profile .
the left panel in fig . 1 shows the surface brightness profile in @xmath10 along the major axis from 0.02170 .
the three panels in fig .
2 show the ellipticity , the position angle of the major axis ( measured from north through east ) , and the color @xmath18 as functions of radius along the major axis .
the random uncertainties can be estimated from the local scatter .
the dynamical modeling described below assumes axisymmetry .
we must therefore determine the best axisymmetric density distribution consistent with the observed surface brightness distribution .
strictly speaking , there is none , because , as shown in figure 2 , there are abrupt changes in the ellipticity and position angle at radii less than 1 and larger than 100 .
pa changes in particular are normally attributed to variation in the axis ratios with radius in a triaxial distribution , or to real three - dimensional twists in locally axisymmetric objects in which surfaces of constant density , while axisymmetric , are not co - axial . 0.2 cm 0.3 cm however , for several reasons we believe that , at radii less than 100 , an axisymmetric model will suffice for our purposes .
first , in the middle range of radii , @xmath19 , the ellipticity and pa are fairly constant at 0.1 and 72 @xmath20 5@xmath0 .
second , in the central region there is a dust ring with radius 1 ( discussed below and seen in fig . 5 ) .
the absorption by the dust ring
1% of the galaxy surface brightness is enough to affect estimates of both the ellipticity and the position angle .
furthermore , the shallow central slope ( dlog @xmath21/dlog @xmath22 ) of the inner profile and the small ellipticity ( @xmath23 ) render the position angle at small radii ( @xmath24 ) highly uncertain . even if the pa gradient seen in fig 2
is real and well determined , the amount of light , and therefore the gravitational impact , of the associated departure from axisymmetry is small . finally , at large radii , departures from axisymmetry have little impact on the central gravitational field .
the data thus suggest that ngc 3379 can be reasonably well represented by a spheroid of constant ellipticity .
the deprojection of the surface brightness is then unique for a given inclination .
the deprojection is based on a non - parametric estimate of the density using smoothing splines ( gebhardt _ et al . _
the right panel of fig .
1 plots the luminosity density for three different inclinations ( 90@xmath0 is edge - on ) . for different inclinations , and hence different flattenings , the luminosity density scales as the ratio of the apparent to the true axis ratio .
we did not examine models in which the galaxy is assumed to be axisymmetric but not spheroidal ( _ i.e. _ , with varying ellipticity ) , in part because in this case the deprojection is not unique .
romanowsky & kochanek ( 1997 ) determine the uncertainties in the deprojection of an axisymmetric system for various inclinations , and the uncertainties can be quite large for low inclinations ( close to face - on ) .
we could test the effect of our spheroidal assumption by modeling the most extreme situation : a galaxy which is seen nearly face - on and containing an embedded disk ( similar to the suggestion of capaccioli _ et al . _
the most likely effect is to change to inclination and _ m / l_@xmath3 constraints , and relaxing the spheroidal assumption should have little effect on the measured black hole mass .
this issue , however , will not be addressed here .
ngc 3379 was observed on 1995 , january 23 27 with the 0.21square aperture ( `` 0.25-pair '' ) of the fos .
the spectrum was taken in two visits ; the total integration time was 215.5 minutes . the wavelength range , 45666815 , includes the mg i _ b _ lines at @xmath25 and the na i d lines at the spectrum was electronically quarter - stepped , giving 2064 1/4-diode pixels .
the reciprocal dispersion was 1.09 pixel@xmath12 .
the instrumental velocity dispersion was measured to be @xmath26fwhm/2.35@xmath27 pixels = @xmath28 @xmath29 ( internal error ) .
this width is intrinsic to the instrument and is not strongly affected by how the aperture is illuminated ( keyes _ et al . _ 1995 ) .
we therefore make no aperture illumination corrections to the measured velocity dispersion .
flat fielding and correction for geomagnetically induced motions ( gim ) were done as in kormendy _ et al . _ ( 1996 ) .
the flat - field image was taken with the same 0.25-pair aperture used for ngc 3379 .
the galaxy exposure was divided into 3 and 4 subintegrations for the two visits . because gim shifts between subintegrations are small , unlike the large shifts between galaxy exposures and the flat field , we first averaged all the exposures and used cross - correlation to determine the shift between this average and the flat field .
the measured shift was 2.27 pixel .
each subintegration was multiplied by the flat - field frame shifted by the above amount .
one dead diode was corrected by dividing the affected pixel intensities by 0.8 .
then gim shifts between subintegrations were determined ; most shifts were @xmath30 pixels , with the largest being 0.8 pixels .
each subintegration was then shifted to agree with the subintegration taken closest in time to the comparison spectrum exposure .
finally , the spectra were added , weighted by the exposure times . obtaining the internal kinematic information
requires a deconvolution of the observed galaxy spectrum with a representative set of template stellar spectra .
we deconvolve the spectrum using a maximum - penalized likelihood ( mpl ) estimate that obtains a non - parametric line - of - sight velocity distribution ( losvd ) .
the fitting technique proceeds as follows : we choose an initial velocity profile in bins .
we convolve this profile with a weighted - averaged template ( discussed below ) and calculate the residuals to the galaxy spectrum .
the program varies the velocity profile parameters bin heights and the template weights to provide the best match to the galaxy spectrum .
the mpl technique is similar to that used in saha & williams ( 1994 ) , and merritt ( 1997 ) .
however , in constrast to these authors we simultaneously fit the velocity profile and template weights . we impose smoothness on the losvd by the addition to the @xmath31 of a penalty function , which is the integrated squared second derivative ( details are in merritt 1997 and gebhardt & richstone 1999 ) .
we estimate the best smoothing from bootstrap simulations ( described below ) .
two templates , dsa107 - 442 ( a field k0iii star ) and n188-i69 ( a k2iii star in ngc 188 ) , are used , both observed with the fos .
star n188-i69 is the preferred template , as the fitting routine gives essentially all of the weight to this star .
however , the losvd determined using only the other star is similar , suggesting that the measurement of the losvd is robust .
the results from this maximum likelihood technique were compared to results of a fourier correlation technique ( bender 1990 ) , and no significant differences were found .
the losvds obtained in the two visits by hst agree to within their uncertainties , and also the pointings from the two visits agree to within 0.01 . therefore , we sum the two spectra weighted by exposure time to derive a combined losvd .
a parameterization of the losvd by gauss - hermite moments ( van der marel & franx 1993 , gerhard 1993 ) gives for the first four moments @xmath32 @xmath29 , @xmath33 @xmath29 , @xmath34 , and @xmath35 . the velocity zero - point is arbitrary and will be discussed in 4 .
all error bars correspond to the 68% confidence band .
in the lower part of figure 3 we plot the template spectrum ; the noisy upper line is the averaged fos spectrum for ngc 3379 ( shifted by its redshift ) , and the smooth upper line is the template spectrum convolved with the derived losvd . the continuum has been divided out of both template and galaxy spectra .
the spectrum shown here is a region around the mg i b triplet ( 5175 ) and not the full spectrum obtained by the fos .
0.2 cm 0.3 cm the 68% confidence band for the losvd is determined using a bootstrap approach .
we convolve the template star with the measured losvd to provide the initial galaxy spectrum ( the upper smooth line in fig .
3 ) ; from that initial galaxy spectrum , we then generate 100 realizations and determine the losvd each time .
each realization contains randomly chosen flux values at each wavelength position drawn from a gaussian distribution , with the mean given by the initial galaxy spectrum and the standard deviation given by the root - mean - square of the initial fit .
the average root - mean - square for the continuum - divided spectrum is 0.033 .
the 100 realizations of the losvds provide a distribution of values for every losvd bin , from which we estimate the confidence bands .
0.2 cm -0.8 cm 0.3 cm figure 4 presents the losvd with the 68% confidence band ( solid and dotted lines ) .
the losvd is significantly skewed with a tail to positive velocities a feature also detected using the fourier correlation technique .
the skewness is crucial for the following analysis and must be considered carefully . in an axisymmetric and even a triaxial system ,
the _ central _ losvd is symmetric .
various attempts were made to get rid of the skewness by reducing the data differently ; we used both templates individually , we used flattened and un - flattened spectra , we used the mg b lines and the na doublet separately and combined , different shifts were used to combine the spectra , and the spectra from both peak - ups were reduced separately . in every case ,
the skewness remained .
since the skewness appears to be real , we must conclude that the fos was not centered , that part of the galaxy light is being obscured , that the galaxy is not symmetric near the center ( possibly because of an offset nucleus ) , or any combination of these affects . if the nucleus is offset , we can not successfully model ngc 3379 as an axisymmetric system .
however , there is no evidence to suggest that ngc 3379 has an offset nucleus since the isophotes , except for the dust ring , are concentric .
figure 5 is an image of the central 5 of the galaxy with the spheroidal component subtracted .
we include in fig . 5 the exact position of the fos aperture .
to determine the exact position of the aperture , we have two independent checks : from the values used to peak - up on the galaxy center , and from the acquisition image taken immediately before the galaxy observations . for the peak - up procedure , hst steps by 0.065 and
points the fos at the highest pixel value .
examination of the peak - up numbers reveals that the likely miscentering is 0.033 , since the center of the peak - up numbers lies roughly between two bins in the peak - up array .
this miscentering is also confirmed from the acquisition image which places the fos aperture the same distance from the galaxy center
. 0.2 cm -0.8 cm 0.3 cm in order to explain the losvd skewness , the direction of the aperture offset must correspond appropriately .
hst placed the aperture on the eastern side of the galaxy , along the major axis .
thus we acquire more light from the eastern side relative to the western side .
the ne side of the galaxy has negative rotation velocities ( davies & birkinshaw 1988 ) .
if one side contributes more light and there is significant rotation , the _ mode _ and mean of the distribution shifts towards that side in this case the ne side with negative velocity and the other side still contributes an extensive tail toward positive velocities , thus generating the observed skewness in the losvd .
a further source for the losvd skewness may be that part of the galaxy light is obscured .
a dust ring is clearly visible , inclined by about 45@xmath0 from the major axis ( the solid line in fig . 5 ) .
from the relative depths of the obscuration on opposite sides , it appears that the sw side of the ring is on the near side of the galaxy , while the ne side is on the far side .
this configuration means that the crucial signal from stars near the nucleus is preferentially obscured on the sw side , thus leading to the same effect as the miscentering ; the western side contributes less light than the eastern side causing positive velocity skewness .
the miscentering of the aperture results from two effects , errors in the peak - up procedure and errors due to the dust obscuration . to calculate the miscentering from dust
, we find the best center by using the galaxy light beyond the dust ring and compare this with the fos center .
hst points fos to the highest position in a 0.21aperture on a @xmath36 raster with spacing of 0.065 .
the dust does not appear to have a significant effect on the location of the brightest pixel ; it causes a miscentering of only 0.01 .
however , errors in the peak - up procedure are more significant , and as mentioned above , the miscentering can be 0.033 .
we conclude that the maximum miscentering is 0.04 ( peak - up errors + dust miscentering ) .
we then place the fos aperture in the wfpc2 image at a position 0.04 east from the galaxy center and determine the relative amount of light in the aperture from opposite sides of the galaxy . at this position ,
the light from the ne side of the galaxy is twice as strong as from the sw side .
thus , in the central bin of our model , we include all of the light from the ne side and only half from the sw side .
an alternative is to use the best symmetrized losvd in the modeling and assume equal contributions from both sides of the galaxy . the dashed line in fig .
4 is the maximum penalized likelihood estimate of the best symmetric losvd .
results from both assumptions are discussed below .
ground - based spectroscopic data are available from gebhardt & richstone ( 1999 ) ( along four position angles ) , statler & smecker - hane ( 1999 ) ( along four position angles ) , bender _ et al .
_ ( 1994 ) ( major axis only ) , gonzalez ( 1994 ) ( major and minor axes ) , franx _ et al .
_ ( 1989 ) ( major and minor axes ) , and davies & birkinshaw ( 1988 ) ( major , minor , and 30@xmath0 axes ) . the data from franx _ et al .
_ and davies & birkinshaw have lower s / n and will not be considered here .
the data from the first four sources all have high signal - to - noise ( s / n ) and in general agree with one another ( see gebhardt & richstone for a detailed comparison ) , with the largest disagreement coming from the large radii data . for the data of gebhardt & richstone , the position angles observed and the radial extraction of the galaxy spectra were designed to correspond exactly with the binning used in the modeling ( described in the next section ) , and we use their data .
we use 3-integral axisymmetric models ( richstone _ et al . _
the models are orbit - based ( schwarzschild 1979 ) and constructed using maximum entropy ( richstone & tremaine 1988 ) .
we run a representative set of orbits in a specified potential , and determine the non - negative weights of the orbits to best fit the available data . maximizing the entropy helps to provide a smooth phase - space distribution function .
rix _ et al .
_ ( 1997 ) present a similar code which has been applied to the spherical case ; cretton _ et al . _ ( 1999 ) and van der marel _ et al .
_ ( 1998 ) developed a fully - general axisymmetric code that is also orbit - based .
the difference between our code and theirs is due to how we impose smoothness and how we fit to observational data .
smoothness in our code comes from maximizing entropy , whereas the other authors enforce smoothness of the distribution function directly by minimizing its variation .
the most important difference is that we fit the losvd directly , and both rix _ et al . _ and cretton _ et al .
_ fit moments of the velocity profile . using the losvd
precludes the need to approximate the velocity profile with a parametric function both in the observations and in the orbital libraries .
the surface brightness yields an estimate of the luminosity density distribution as explained above ( 2.1 ) , assuming that the luminosity density is axisymmetric on similar spheroids and that the galaxy has a given inclination . to obtain the mass density distribution , we have made one of two assumptions about the mass - to - light ratio ( _ m / l_@xmath3 ) : ( _ i _ ) constant _ m / l_@xmath3 , ( _ ii _ ) _ m / l_@xmath3 that varies according to the @xmath18 color shown in fig . 2 .
ngc 3379 is redder in the center by @xmath37 .
the analysis of gonzalez ( 1994 ) indicates that this color change is due mainly to metallicity ( _ i.e. _ , no age effects ) . in this case
, models of worthey ( 1994 ) predict that _ m / l_@xmath3 may be as much as 25% higher in the center . the @xmath18 variation in fig .
2 is approximately log - linear , so we vary _ m / l_@xmath3 by 25% from the outer to inner radii following the same functional form . finally , we add a central point mass and compute the potential . using the derived potential
, we follow orbits which sample the available phase space in energy ( @xmath38 ) , angular momentum ( @xmath39 ) , and the third integral ( @xmath40 ) ( richstone _ et al . _
each orbit crosses the equatorial plane 40 times .
binning the orbital distributions both in real space and velocity , and in projected distance and line - of - sight velocity , provides an estimate of the contribution to the losvd from each orbit at each spatial bin .
both the real and projected spatial binning are approximately logarithmic in radius and linear in cos @xmath41 , where @xmath41 is measured from the pole .
this binning scheme provides approximately equal - mass bins . since we are using axisymmetric modeling
, we follow orbits only in @xmath42 and @xmath41 .
we therefore have to choose the azimuthal angle for the projection , and , since all angles are equally likely , we choose that angle at random . for every time step ,
100 azimuthal angles are chosen .
we measure an orbit s contribution on a grid of 80 radial bins , 20 angular bins , and 13 velocity bins compressed to @xmath43 when comparing to the data .
the size of the central bin equals the aperture used in the fos ( 0.21 ) .
the outermost bin ranges from 149 to 200 .
the centers of the five angular bins up from the major axis are at 6@xmath0 , 18@xmath0 , 30@xmath0 , 45@xmath0 , and 72@xmath0 .
the velocity bins are about 100 @xmath29 wide , approximately equal to the spectral resolution of the ground - based observations .
we follow @xmath44 orbits per model depending on inclination and black hole mass .
this number is only for one sign of the angular momentum ; consequently , it must be doubled before comparison to the data .
we flip the losvd at every position in the galaxy to provide the appropriate reversed orbit .
thus , our final model contains @xmath45 orbits .
varying the relative weights of these orbits allows us to determine the best match to the data .
it is desirable to constrain every model bin with an observed losvd . for ngc 3379
, we have ground - based observations of the losvd along four of the five position angles used in the modeling .
in fact , the observations were designed to mimic the model angular bins listed above .
furthermore , the spectra were extracted from the data using the same radial binning scheme as the modeling .
therefore , most model bins correspond _ exactly _ to an observation ; _ i.e. _ , no interpolation of the data or the model is necessary for comparison .
this correspondence breaks down for the ground - based data in the innermost bins because of seeing .
the ground - based data extend out to 64 , about two effective radii .
the total number of bins for which we have an observation is 54 .
the model losvds are calculated at 13 velocity positions with a spacing of 100 @xmath29 , providing at most 13 independent measurements of the losvd per position .
we therefore have a _ maximum _ of 702 ( 54 positions @xmath46 13 losvd bins ) independent observables . however , the actual number is smaller than this , mainly due to the smoothing used in the measurement of the losvds ( discussed in 5 and gebhardt & richstone 1998 ) , which correlates the 13 velocity bins .
thus , we will overestimate the goodness - of - fit since we use the 68% confidence bands of the losvds to calculate it .
the model incorporates seeing by convolving the light distribution for every orbit with the appropriate psf before comparing to the data . for the ground - based data , the estimated psf , approximated as gaussian ,
has a fwhm of 1.5 , including both atmospheric seeing and slit size .
the hst data point was taken in a square aperture of 0.21 .
the psf for hst has a fwhm of around 0.07 , and hence smearing due to the psf is negligible compared to the aperture used .
therefore , for the hst data , only the aperture size was considered with no psf convolution .
we have checked in a number of models that this approximation has no effect . for each model , we match both the light in each of the @xmath47 real - space bins , and the losvds in the 54 locations where we have data .
the technique minimizes the @xmath31 between the model and data losvds .
the light from the model matches the deconvolved spatial brightness profile to a tolerance of better than 1% in each bin , which also ensures that the surface brightness matches better than 1% in any projected light bin .
the match to the total light in each bin has to be done using the real spatial profile and not the surface brightness profile .
if the matching were done in projection , it would admit different density distributions than the one for which the orbits were computed since the deprojection of a flattened , inclined , axisymmetric system is not unique .
for each model , the 702 observables ( 13 losvd bins at the 54 positions ) determine the total @xmath31 , given by @xmath48 where the @xmath49 s are the losvd bin heights for the model and the data and @xmath50 is the uncertainty from the 68% confidence band . as an example ,
figure 6 plots the losvds for both the data and the best model at four positions : the center , a major - axis position near to the center , a position close to one effective radius on the major axis , and a minor - axis position near to the center . judging the goodness - of - fit from the @xmath31 values of the various models is not straightforward because the number of the degrees of freedom is not easy to determine ; the derivation of the losvds includes a smoothing parameter which correlates the values and uncertainties of the velocity bin heights ( gebhardt & richstone 1999 ) . with higher s
/ n data than present , we could reduce the smoothing parameter to lessen this problem .
we do not attempt to estimate directly the actual degrees of freedom and therefore do not have an overall goodness - of - fit measure , but instead we calculate the change in @xmath31 as a function of the three variables black - hole mass , inclination , and _ m / l_@xmath3 .
the lowest @xmath31 value thus provides the best model , with the uncertainties given by the classical estimators for the @xmath51 s
. 0.2 cm -0.6 cm 0.3 cm 0.2 cm 0.3 cm the total @xmath31 s for most of the models have a value near 200 , giving a reduced @xmath31 of 0.3 , assuming one observable equals one degree of freedom .
a reduced @xmath31 near unity would require either that every three losvd bins are correlated by the smoothing , therefore reducing the number of degrees of freedom by three , or that the uncertainties of the bin heights are overestimated by a factor of @xmath52 . in the former case , using @xmath51 as a discriminator for the error of the best - fitting models
will not be affected since the number of degrees of freedom has no influence on the calculation of the @xmath31 differences ; @xmath53 will still correspond to 1-@xmath50 .
however , in the latter case , both the @xmath31 and @xmath51 values would increase by a factor of three , creating a smaller confidence band on the variables ; @xmath54 would then correspond to 1-@xmath50 .
it is likely , however , that the dominant reason for the low reduced @xmath31 is the underestimation of the degrees of freedom since we know _ a priori _ that the smoothing must have an effect . to provide a quick check on smoothing effects
, we refit the losvds for a subset of positions using a much smaller smoothing length .
as expected , the losvd becomes noisier .
measuring the @xmath31 between this new losvd and the model that we fit using the the older losvd , we find that the increase in @xmath31 ranges from 3 - 5 .
thus , we conclude that the dominant effect of the smoothing is to correlate the variables and reduce the number of degrees of freedom .
in retrospect , it would have been better not to have smoothed the losvds and preserve their statistical independence .
note however that our decision to accept @xmath53 as the 1-@xmath50 error is conservative and , if anything , overestimates the errors .
we examined models with inclinations from 29@xmath0 to 90@xmath0(intrinsic e5 to e1 galaxy ) , black holes from zero to @xmath1 , and _
m / l_@xmath3 from 4.0 to 8.0 in solar units .
the @xmath31 is a function of these three variables . for each inclination ,
fig . 7 plots @xmath31 as a function of black - hole mass ( zero black - hole mass is shown as @xmath55 ) , using that _ m / l_@xmath3 which provides the smallest @xmath31 at that black - hole mass .
all inclinations show a minimum @xmath31 at black - hole masses in the range @xmath56 .
the fits clearly prefer the edge - on models ( 90@xmath0 ) .
8 presents contours of @xmath51 as a function of black - hole mass and _ m / l_@xmath3for each inclination .
the points represent locations of the modeled values .
the contours use a two dimensional smoothing spline ( wahba 1980 ) to estimate the @xmath31 values at positions where no models occur . as in wahba
, generalized cross - validation determines the smoothing value ; however , the modeled values are relatively smooth and little smoothing is necessary .
0.4 cm -1 cm 1.0 cm since we have such a large parameter space minimizing several thousand variables we must ensure that the modeling program is finding a true minimum .
furthermore , we require that the numeric noise caused by the minimizer be smaller than the quoted uncertainties of the output parameters ( e.g. , the black hole mass ) .
the two conditions that govern these checks are the sampling densities that we use in parameter space , and the tolerance used to determine convergence in the minimizer .
8 demonstrates the sampling densities for the black - hole mass and the _ m / l_@xmath3 , where each point represents one model .
7 shows the range and number of modeled inclinations .
the smoothness of the @xmath31 contours in both figures demonstrate that we have found the global minimum . as a further check , we have run models with both larger and smaller _ m / l_@xmath3 than those presented , however the fits are significantly worse and we do not include them in the plots .
thus , we conclude that the sampling density provides adequate coverage around the global minimum .
our second concern is whether the tolerance used for the stopping criteria in the minimizer creates significant noise for the parameter estimation .
we must use a stopping tolerance in the program or the time it takes for the @xmath31 to asymptote becomes impractically long .
both figures 7 and 8 provide an estimate of the numeric noise . from the soothness of plots ,
the change in @xmath31 between two adjacent points is much smaller than the global change in @xmath31 .
also , in a handful of models , we iterate till there is no change in @xmath31 relative to machine precision , and find that this asymptotic value is within @xmath57 of the stopping value .
therefore , the noise in @xmath31 generated from the minimization routine is not significant .
the smoothness is the solution is also seen in fig . 9 , where we plot best - fit black hole mass as a function of inclination .
although , the sampling of points is not dense , the fact that there are no abrupt changes signifies we are not subject to numeric noise in the solution . 0.2 cm 0.4 cm 0.2 cm 0.3 cm
as described in 3.1 , we have also run models with _
m / l_@xmath3 increasing 25% log - linearly with radius towards the galaxy center . in this case , the required _
m / l_@xmath3 is systematically lower than when using the constant _ m / l_@xmath3 assumption ; however , the results for the required black hole and inclination are unchanged .
the best model has an inclination of 90@xmath0 , a black hole mass of @xmath58 , and _
m / l_@xmath3@xmath59
. the allowed mass of black holes range from @xmath2 .
the internal dynamics of the best - fitting model are shown in fig .
the top plot shows the internal velocity dispersions of the three components along the major axis .
the model is tangentially biased ( in the @xmath60 direction ) in the smallest bin , radially biased in the mid - range radii , and tangentially biased again at the largest radii .
independent of inclination , fig .
10 well characterizes the orbital distribution for each of the best - fitting models . similarly , the mean velocity in the equatorial plane ( bottom plot of fig .
10 ) has a general shape for the best models ; it shows no systematic trends from @xmath61 ( 20 - 1500 pc ) , increasing significantly in the center due to the presence of the massive black hole . beyond about 60
the model is unconstrained since there is no kinematic information , but we plot the model results there since they represent the maximum entropy configuration . for each inclination
the trend with black - hole mass is as follows : increasing the black - hole mass requires an increased amount of tangential anisotropy in the central bin , a decrease of the radial bias in the mid - range radii , and the orbits at the largest radii to become more isotropic .
the coupling between the black - hole mass and the amount of radial anisotropy in the mid - radii stems from the need to match the same central losvd ; a large central dispersion can be modeled with a large black - hole mass or radial orbits .
therefore , as the black - hole mass increases , the need for radial orbits decreases . in the very central bin , a large black - hole mass induces large rotational velocities and also large dispersion in the @xmath60 direction . at the largest radii ,
the situation is more complex since we have not included a dark halo ; if the projected dispersion is larger than the dispersion obtained assuming isotropy and the stellar mass profile then one can either invoke a dark halo or include tangential anisotropy . in any event ,
it is not clear whether meaningful analysis at large radii can be obtained from models which do not include a dark halo . to demonstrate why the data seem to prefer an edge - on model that contains a black hole
, we must look at the model fits to the data .
figure 11 plots gauss - hermite polynomial estimates for the data and model losvds for all four position angles .
the difference between the edge - on model and the nearly face - on model is almost entirely at large radii ; the edge - on model matches the large - radial data better , specifically for the velocity and velocity dispersion .
in contrast , the difference between the no - black - hole and black - hole model is so subtle that one can barely discriminate those two models in fig .
11 . only in the central bin
does the the black - hole model slightly better estimate the skewness in the profile .
however , fig .
11 is not the optimal way to compare models since the fit uses the full losvd and not moments . instead of plotting all 54 losvds for various models , fig .
12 plots the difference in @xmath31 at each spatial position .
this difference represents the @xmath31 measured for the losvd of a particular model minus that of the best - fit model ( edge - on , black - hole model ) .
the two comparison models are the same as in fig .
11 , the best no - black - hole model ( an edge - on model ) and a nearly face - on black - hole model ( at 29@xmath0 ) . as expected ,
the discrimination between black - hole / no - black - hole comes mainly from the hst / fos data point .
a no - black - hole model fails to match the shape of the central losvd ; however , there is a limit to size of the central black hole since very high black - hole masses models over shoot the dispersions in the central few ground - based measurements . without the hst data point , we would not have been able to determine whether a black hole exists in ngc 3379 .
the discrimination for the inclination comes from the data at larger radii ; here , the edge - on model is much better able to match the projected kinematics . as we discuss later , this inclination constraint is uncertain due to our assumption of constant _ m / l_@xmath3 .
0.2 cm 0.3 cm there is an uncertainty in the absolute wavelength calibration of the hst spectrum . since we do not assume that the central hst losvd is symmetric
, we must determine the velocity zeropoint of this profile relative to the rest frame of the galaxy ( @xmath62 @xmath29 , borne _ et al . _
unfortunately , our original observing strategy did not include an accurate wavelength calibration , since we were concerned mainly with the central dispersion , and thus did not obtain the necessary flanking arc - lamp spectra .
we therefore tried a range of velocity zeropoints , models with velocity offsets from @xmath63 @xmath29 .
as the offset increases , the lower limit on the black - hole mass grows , whereas the upper limit remains the same .
we are not able to fit acceptable models with offsets greater than 15 @xmath29 .
therefore , our allowed range of black hole masses is largest using the 0 @xmath29velocity offset .
all of the figures contain the results using this value .
the models in this paper depend primarily on the input black - hole mass , inclination , and m / l assumption .
a difficulty is that different combinations of these variables can have similar appearances in the projected kinematics .
for example , the inclination and the m / l variation both affect the fit at large radii ; one can easily see how differently inclined models of the same flattened galaxy will have markedly different projected dispersion profiles , and also the dispersion profile is obviously influenced by the inclusion and shape of a dark halo .
thus , by varying the shape and amount of dark halo one may be able to mimic the effect of changing the inclination .
since we have not explored a full range of halo models , we will not make strong claims as to our ability to constrain the inclination of the galaxy .
however , the black hole mass is mainly driven by the data at small radii , and primarily by the hst / fos data point . fig .
9 demonstrates this independence from the inclination , as our best - fit black hole mass does not depend strongly on the inclination .
an even more robust measurement is the orbital structure .
as either the inclination or the black - hole mass varies , we find a similar velocity ellipsoid ; at the smallest bin the velocity ellipsoid is mainly tangential , becoming radial in mid - range radii , and tangential again at the largest radii .
13 plots the radial motion relative to the tangential motion for four models at different inclinations .
determination of the inclination , black - hole mass , and orbital structure is discussed in detail below .
0.3 cm -0.7 cm 0.3 cm the inclination of ngc 3379 has been the subject of some debate .
capaccioli _ et al .
_ ( 1991 ) conclude that ngc 3379 is similar to ngc 3115 but nearly face - on , with some degree of triaxiality .
capaccioli base this conclusion only on the surface brightness profile and use no kinematic data .
in contrast , statler ( 1994 ) concludes that nearly face - on , triaxial models are inconsistent with the data at the 98% confidence limit , strongly contradicting capaccioli s result .
he prefers less - inclined models .
statler has obtained more recent data ( statler & smecker - hane , 1999 ) and a fuller analysis is in preparation . as seen in sect . 4
, our best models prefer an edge - on configuration .
preference for edge - on models arises from two observations ; first , the dispersion profiles along different axes are slightly different , and , in an axisymmetric case , this configuration is not possible for a face - on system .
the difference mainly arises from the inability to match the large radial data , however the face - on model also can not match the shoulder in the dispersion profile at 1020 ( previously noted by statler & smecker - hane , 1999 ) ; the shoulder is more pronounced in the central position angles , and the edge - one models match this configuration well ( see fig .
second , major - axis rotation exists which is not allowed for a face - on configuration of an oblate axisymmetric system .
thus these two restrictions are mainly a result of our axisymmetry assumption .
more general models can be constructed than those presented here .
we could in principle extend the modeling technique to include triaxial distributions but then the parameter space becomes too large for us to search ( each axisymmetric model takes approximately 20 hours on an ultrasparc ) .
second , our assumption that the galaxy is spheroidal could be relaxed .
romanowsky & kochanek ( 1997 ) present a technique which provides a range of different density distributions which similarly project .
third , we have assumed that _ m / l_@xmath3 is independent of radius ( _ i.e. _ , there is no dark halo ) .
ciardullo _ et al . _
( 1993 ) , using radial velocity measurements of planetary nebulae , find no evidence for a dark halo in ngc 3379 out to 3.5 effective radii ( 120 ) ; however their uncertainties are large and could , in fact , allow for some dark halo ( tremblay _ et al . _
1995 ) . in all of our models ,
the last velocity dispersion measurement ( at 64 ) along each position angle is higher than the model velocity dispersion , possibly suggesting the need for an increase in _ m / l_@xmath3 .
at one effective radius ( 34 ) , there is no need for an increase in _
m / l_@xmath3 , which is consistent with previous studies of other galaxies ( kormendy & westphal 1989 , kormendy & richstone 1992 , kormendy _ et al . _
thus , assumptions of axisymmetry , spheroidal distribution , and constant _
m / l_@xmath3 all lead to uncertainties in the best - fit inclination .
detailed modeling including dark halos has been carried out by rix _ et al .
_ ( 1998 ) for ngc 2434 and saglia _ et al . _ ( 1999 ) for ngc 1399 , but only as spherical systems .
only by allowing for axisymmetric or triaxial shapes with a dark halo can we set more realistic limits on inclination .
even though we are not able to place limits on the inclination , we can measure black - hole limits .
the presence of a dark halo will have no effect on the need for a central black hole , or on its mass . the need for the black hole is directly a result of the shape of the central losvd observed with hst . the large dispersion and the skewness
drive the models to require a black hole .
we have also run models assuming a symmetrized central losvd ( the dashed line shown in fig .
the allowable range of black hole masses remains the same for the inclined models in this case .
however , for the edge - on case the model with no black hole fits as well .
the symmetrized losvd has a dispersion of 275 @xmath29 , compared with 289 @xmath29 for the actual losvd , and has less tail weight ( as can be seen in fig .
these differences permit a no - black - hole model .
the central bin most influences the model ; most of the orbits ( @xmath64% for inclined models and @xmath65% for edge - on models ) contribute light to this bin , and , furthermore , many have their pericenter there .
therefore , the characteristics of the central losvd , in particular the high - velocity components , have a strong effect on orbits throughout the galaxy . in the edge - on case with the symmetrized losvd ,
the orbits redistribute themselves in such a way as to create a slightly larger central dispersion ( 275 @xmath29 ) without the need for a black hole , and at the same time maintain a fairly flat projected dispersion profile ; _ i.e. _ , at small radii , the orbits are mainly radial , whereas at large radii they become primarily tangential . however , the actual losvd warrants matching the skewness and the high - velocity tails . with no black hole present ,
the model must use stars at larger radii to create the high - velocity components , thereby limiting its ability to redistribute the orbits at large radii and match the kinematics there . with a black hole
, the black hole provides much of the high - velocity tail of the losvd for orbits locally and gives the model more freedom to better match the larger - radii data .
the symmetrized and skewed losvds are inconsistent with each other with greater than 95% probability , both using an estimate from the binned values and using the gauss - hermite moments . given the arguments in 2.2.1 , we must include the skewness of the central losvd and thus do not consider the symmetrized losvd models further .
relaxing the other two assumptions axisymmetry and spheroidal distribution is unlikely to change the black - hole limits .
the models with no black holes all are extremely poor fits to the data ; therefore , even if we allow for the variation in deprojected densities given by romanowsky & kochanek , the results are unlikely to change .
of course , since the deprojection is unique for the edge - on models , our results are general in the edge - on case .
triaxial models will allow even greater freedom to match the data , however it is not clear whether triaxiality is a viable configuration in the centers of galaxies ( merritt & fridman 1996 , merritt 1999 ) , and we do not consider these models .
the position angle of the dust ring in fig .
5 is 45@xmath0 away from the major axis of the galaxy .
a dust ring in an axisymmetric system with an offset position angle is only neutrally stable but can be stabilized by a small amount of triaxiality . if the galaxy is nearly face - on , then triaxiality is not necessary to stabilize the dust ring ; in the face - on configuration , since the dust ring is close to edge - on ( about 75@xmath0 ) , it must be in an orbit close to polar , and thus can have arbitrary azimuthal angle .
velocity information along the dust ring could provide an important check on the enclosed mass derived from our models .
pastoriza _ et al .
_ ( 1999 ) , using ground - based gas kinematics , assume that the gas disk in ngc 3379 is nearly face - on .
their mass estimate inside 1.3 is @xmath66 , strongly contradicting our result .
however , their large mass is due mainly to their disk inclination assumption .
if one uses the inclination as measured from the dust ring , then the enclosed mass closely equals our value of @xmath58 ; only with hst can we precisely determine the gaseous disk configuration and determine whether the gas is even in a disk at all .
the hst / fos spectrum includes the [ nii ] emissiom , although since it is a single pointing , we do not have the radial information to extract a meaningful dynamical analysis from the gas .
however , the width of the line does provide some information on the underlying mass .
the measured @xmath50 is 200@xmath29 , in agreement with the mass deduced here ; using the mass from pastoriza _ et al .
_ would imply a @xmath50 around 700@xmath29 . to further understand the gas kinematics , we must have spatial information . van der marel _ et al .
_ ( 1990 ) apply 2-integral flattened models for ngc 3379 using constant anisotropy .
they find an acceptable fit with an inclination of 60@xmath0 and anisotropic orbits .
they suffer from the same problem as we do for the inclination estimate ; without inclusion of a dark halo it is not possible to obtain adequate inclination constraints .
furthermore , since our axisymmetric models are fully general while their models specify a form for the distribution function , we can not directly compare our results to theirs .
van der marel _ et al .
_ conclude that there is no evidence to support the need for a third integral using data from the first and second moments of the losvd from davies & birkinshaw ( 1988 ) .
results presented here , which are based on higher s / n data and are determined from the full losvd , show that the best model requires three integrals and is not consistent with a 2-integral distribution function ; the internal dispersions along the long axis in the radial and @xmath41 directions are significantly different , in direct conflict with a 2-integral distribution function . as seen in fig . 13 , the ratio of the @xmath41 to the radial dispersion in our best model have a radial variation from 0.6 to 1.2 , with an average around 0.75 .
for comparison , this ratio is around 0.6 for the spheroid component of our galaxy using rr lyrae halo stars ( layden 1995 ) .
in addition to the measured black - hole mass , the orbital structure appears to be a robust feature as well .
the radial anisotropy profile along the major axis in fig .
13 shows that this profile is nearly independent of input inclination . for all acceptable models ,
the velocity ellipsoid is tangentially biased in the central bin and radially biased in the mid - range radii .
these orbital properties are generally true along any position angle .
this central tangential bias appears in other 3-integral models studied ( gebhardt _ et al . _ 2000 ) and may be a common feature of early - type galaxies .
tangential bias in the central regions around black holes appear to be a common feature in simulations ( quinlan _ et al . _ 1995 ,
quinlan & hernquist 1997 , merritt & quinlan ( 1998 ) , and nakano & makino 1999 ) .
the radial bias in the mid - range radii has also been seen before ; using 3-integral models , both gerhard _ et al . _ ( 1999 ) for ngc 1600 and saglia
_ et al .
_ ( 1999 ) for ngc 1399 find radial motion throughout the main body of the galaxy .
thus , we may be beginning to measure common features of the orbital properties in ellipticals
tangential motion in the central regions and radial motion at mid - range radii . at large radii
, we must include dark halos since they make a significant contribution there .
the next step is to determine dominant evolutionary effects from theoretical models and will require detailed comparisons with n - body simulations .
magorrian _ et al . _ ( 1998 ) , using the same ground - based data , fit two - integral models to the observed second - order velocity moment profiles and require a bh of @xmath67 .
our more general three - integral models would allow an even larger range of bh masses had we similarly used only the losvd s second moments .
when we fit to the full losvds , however , their shape requires that our models be mildly radially anisotropic , reducing the required bh mass by a factor of at least two over magorrian _ et al . _
s result .
the models presented in this paper still contain restrictive assumptions namely constant _ m / l_@xmath3 and spheroidal symmetry requiring yet more models to study black hole properties in complete generality .
kormendy ( 1993 ) and kormendy & richstone ( 1995 ) find that there is a correlation between the black hole mass and the mass of the spheroid . whether this relation is a real correlation or an upper envelope that extends to smaller black - hole masses remains to be seen . as the data quality and the modeling techniques improve
, we should be better able to constrain the relationship between black hole and spheroid mass , and begin to measure correlations with the orbital characteristics .
we thank gary bower , richard green , luis ho , and jason pinkney for detailed discussions about the results .
we thank an anonymous referee who found an error in the previous version of this paper .
we are grateful to c. d. keyes , j. christensen , and j. hayes for help with the data analysis .
this work was supported by hst data analysis funds through grant go02600.0187a and by nserc .
kg is supported by nasa through hubble fellowship grant hf-01090.01 - 97a awarded by the space telescope science institute , which is operated by the association of the universities for research in astronomy , inc .
, for nasa under contract nas 5 - 26555 . rrrrr 0.023 & 14.625 & 1.302 & 0.121 & 92.3 0.046 & 14.739 & 1.314 & 0.121 & 92.3 0.091 & 14.931 & 1.340 & 0.040 & 92.3 0.136 & 14.978 & 1.311 & 0.232 & 92.3 0.182 & 15.047 & 1.322 & 0.181 & 101.8 0.227 & 15.101 & 1.307 & 0.181 & 87.6 0.273 & 15.147 & 1.311 & 0.177 & 87.6 0.318 & 15.184 & 1.313 & 0.149 & 85.8 0.364 & 15.227 & 1.311 & 0.141 & 84.4 0.409 & 15.276 & 1.325 & 0.119 & 84.1 0.455 & 15.288 & 1.311 & 0.119 & 78.8 0.500 & 15.326 & 1.318 & 0.119 & 76.4 0.546 & 15.351 & 1.313 & 0.110 & 64.9 0.566 & 15.368 & 1.326 & 0.083 & 58.1 0.665 & 15.412 & 1.297 & 0.119 & 65.7 0.783 & 15.485 & 1.300 & 0.095 & 58.7 0.921 & 15.578 & 1.299 & 0.089 & 69.6 1.084 & 15.692 & 1.300 & 0.085 & 67.5 1.275 & 15.812 & 1.305 & 0.081 & 70.8 1.500 & 15.933 & 1.299 & 0.101 & 70.9 1.764 & 16.072 & 1.297 & 0.104 & 72.1 2.076 & 16.213 & 1.293 & 0.115 & 73.3 2.442 & 16.365 & 1.295 & 0.117 & 74.0 2.873 & 16.528 & 1.291 & 0.119 & 73.5 3.380 & 16.702 & 1.286 & 0.118 & 73.8 3.977 & 16.895 & 1.284 & 0.111 & 74.0 4.678 & 17.110 & 1.278 & 0.105 & 74.1 5.504 & 17.352 & 1.275 & 0.094 & 74.0 6.475 & 17.596 & 1.272 & 0.088 & 73.8 7.618 & 17.830 & 1.267 & 0.089 & 73.2 8.962 & 18.064 & 1.262 & 0.089 & 73.2 10.544 & 18.313 & 1.257 & 0.090 & 72.1 12.405 & 18.572 & 1.246 & 0.092 & 71.1 14.594 & 18.855 & 1.237 & 0.089 & 69.3 ajhar , e. , lauer , t. , tonry , j. , blakeslee , j. , dressler , a. , holtzman , j. , & postman , m. 1997 , , 114 , 626 bender , r. , 1990 , , 229 , 441 bender , r. , saglia , r.p . , & gerhard , o.e .
1994 , , 269 , 785 borne , k.d . ,
balcells , m. , hoessel , j.g , & mcmaster , m. 1994 , , 435 , 79 bower , g. _ et al . _ 1998 , , 492
, 111 capaccioli , m. , held , e.v .
, & nieto , j.l .
1987 , , 94 , 1519 capaccioli , m. , held , e.v . , lorenz , h. & vietri , m. 1990 , , 99 , 1813 capaccioli , m. , vietri , m. , held , e.v .
, & lorenz , h. 1991 , , 371 , 535 carollo , c.m . , de zeeuw , p.t . ,
van der marel , r.p . , danziger , i.j .
, & qian , e.e .
1995 , , 441 , l25 ciardullo , r. , jacoby , g.h . , &
dejonghe , h.b .
1993 , , 414 , 454 cretton , n. , de zeeuw , p.t .
, van der marel , r.p . , & rix , h .- w .
, 1999 , , accepted davies , r.l .
, & birkinshaw , m. 1988 , , 68 , 409 dejonghe , h.b .
1987 , , 224 , 13 de vaucouleurs , g. , & capaccioli , m. 1979 , , 40 , 699 faber , s.m .
_ et al . _
1997 , , 114 , 1771 ferrarese , l. , ford , h.c . , &
jaffe , w. 1996 , , 470 , 444 franx , m. , illingworth , g.d . , & heckman , t.m .
1989 , , 344 , 613 gebhardt , k. , richstone , d. _ et al . _
1996 , , 112 , 105 gebhardt , k. , richstone , d. 1999 , in preparation gerhard , o.e .
1993 , , 265 , 213 gonzalez , j. 1994 , phd thesis , ucsc harms , r.j .
_ et al . _
1994 , , 435 , l35 keyes , c. d. , _ et al . _ 1995 , fos instrument handbook ( baltimore : stsci )
kormendy , j. 1977 , , 218 , 333 kormendy , j. 1992 , in `` testing the agn paradigm '' , eds .
holt , s.g .
neff , & c.m urry ( ny : aip ) , 23 kormendy , j. 1985 , , 292 , l9 kormendy , j. 1993 , in `` the nearest active galaxies '' , eds
. j. beckman , l. colina , & h. netzer ( madrid ) , 197 kormendy , j. , & westphal , j. 1989 , , 338 , 752 kormendy , j. , & richstone , d. 1992 , , 393 , 559 kormendy , j. , & richstone , d. 1995 , ara&a , 33 , 581 kormendy , j. _ et al . _
1996 , , 459 , l57 kormendy , j. _ et al . _
1997 , , 482 , l139 kormendy , j. , bender , r. , evans , a. , & richstone , d. 1998 , , 115 , 1823 lauer , t.r . 1985 , ,
216 , 429 lauer , t.r .
_ et al . _
1995 , , 110 , 2622 layden , a.c .
1995 , , 110 , 2288 magorrian , j. _ et al . _
1998 , , 115 , 2285 merritt , d. 1997 , , 114 , 228 merritt , d. 1999 , , 111 , 129 merritt , d. , & fridman , t. 1996 , , 460 , 136 nakano , t. , & makino , j. 1999 , , 510 , 155 pastoriza , m. , winge , c. , ferrari , f. , & macchetto , d. 1999 , , accepted peletier , r.f . , davies , r.l . ,
illingworth , g.d . ,
davis , l.e . ,
cawson , m. 1990 , , 100 , 1091 qian , e.e . ,
de zeeuw , p.t . ,
van der marel , r.p . , & hunter , c. 1995 , , 274 , 602 quinlan , g. , hernquist , l. , & sigurdsson , s. 1995 , , 440 , 554 quinlan , g. , & hernquist , l. 1997 , newa , 2 , 533 richstone , d. , gebhardt , k. _ et al . _ 1999 , in preparation richstone , d. , & tremaine , s. 1988 , , 327 , 82 rix , h .- w .
, de zeeuw , p.t . , carollo , c.m . ,
cretton , n. , & van der marel , r.p .
1997 , , in press romanowsky , a. , kochanek , c. 1997 , , 287 , 35 rybicki , g.b .
1987 , in structure and dynamics of elliptical galaxies , iau symposium 127 , ed . de zeeuw , p.t . , ( reidel , dordrecht ) , p.397 ryden , b. 1996 , , 461 , 146 saha , p. , & williams , t.b . 1994 , , 107 , 129 schwarzschild , m. 1979 , , 232 , 236 statler , t.s . 1994 , , 108 , 111 statler , t.s . ,
& smecker - hane , t. 1999 , , feb .
1999 tremblay , b. , & merritt , d. , & williams , t.b .
1995 , , 443 , l5 van der marel , r.p . ,
binney , j.j .
, & davies , r.l . 1990 , , 245 , 582 van der marel , r.p . 1991 , , 253 , 710 van der marel , r.p .
, & franx , m. 1993 , , 407 , 525 van der marel , r.p . , cretton , n. , de zeeuw , p.t . , & rix , h .- w .
1998 , , 493 , 613 van dokkum , p.g . , & franx , m. 1995 , , 110 , 2027 wahba , g. 1980 , technical report no .
595 , univ . of wisconsin worthey , g. 1994 , , 95 , 107 | we fit axisymmetric 3-integral dynamical models to ngc 3379 using the line - of - sight velocity distribution obtained from hst / fos spectra of the galaxy center and ground - based long - slit spectroscopy along four position angles , with the light distribution constrained by wfpc2 and ground - based images .
we have fitted models with inclinations from 29@xmath0 ( intrinsic galaxy type e5 ) to 90@xmath0 ( intrinsic e1 ) and black hole masses from 0 to @xmath1 .
the best - fit black hole masses range from @xmath2 , depending on inclination .
the preferred inclination is 90@xmath0 ( edge - on ) ; however , the constraints on allowed inclination are not very strong due to our assumption of constant _ m / l_@xmath3 .
the velocity ellipsoid of the best model is not consistent with either isotropy or a two - integral distribution function . along the major axis ,
the velocity ellipsoid becomes tangential at the innermost bin , radial in the mid - range radii , and tangential again at the outermost bins .
the rotation rises quickly at small radii due to the presence of the black hole . for the acceptable models , the radial to tangential ( ( @xmath4 dispersion in the mid - range radii ranges from @xmath5 , with the smaller black holes requiring larger radial anisotropy . compared with these 3-integral models ,
2-integral isotropic models overestimate the black hole mass since they can not provide adequate radial motion .
however , the models presented in this paper still contain restrictive assumptions namely assumptions of constant _
m / l_@xmath3 and spheroidal symmetry requiring yet more models to study black hole properties in complete generality . |
field equations that governs our model of consideration are @xmath30 and with frw metric ( the metric of a spatially flat homogeneous and isotropic universe ) @xmath31 they reduce to @xmath32 @xmath33 energy conservation condition reads as @xmath34 to introduce an interaction between dark energy and matter , as already was mentioned in introduction section , ( [ eq : bianchi eq ] ) splits into two following equations @xmath35 and @xmath36 in our case problem solving strategy will start with finding pressure of modified fluid from ( [ eq : inteqm ] ) @xmath37 pressure of van der waals fluid ( [ eq : waals ] ) with ( [ eq : fridmman vlambda ] ) will be expressed by @xmath38 @xmath39 ( [ eq : fridmman vlambda ] ) , ( [ eq : freidmann2 ] ) , ( [ eq : mpressure ] ) and ( [ eq : pwrhom ] ) together allow us to determine a scale factor . from the next section
we will start from the simple case , when interaction between modified fluid and van der waals fluid is absence and then we will present results accounting interaction ( [ eq : ourint ] ) .
[ [ section-1 ] ] non interacting case with @xmath40 simplify equations under consideration and solution could be find very easily .
bellow we will present behavior of @xmath41 and @xmath42 over time ( fig .
[ fig:1]-[fid:3 ] ) .
@xmath43{accel1.jpg } & \includegraphics[width=50 mm]{omega1.jpg } \end{array}$ ] @xmath43{accel4.jpg } & \includegraphics[width=50 mm]{omega4.jpg } \end{array}$ ] @xmath43{accel5.jpg } & \includegraphics[width=50 mm]{omega5.jpg } \end{array}$ ] [ [ section-2 ] ] in this section we will analyze the case when we account interaction between mixture components .
bellow we presented graphical analysis for several cases , with different values of parameters ( fig .
[ fig:4]-[fid:6 ] ) .
@xmath43{accel1_2.jpg } & \includegraphics[width=50 mm]{omega1_2.jpg } \end{array}$ ] @xmath43{accel1_2_1.jpg } & \includegraphics[width=50 mm]{omega1_2_1.jpg } \end{array}$ ] @xmath43{accel2_2.jpg } & \includegraphics[width=50 mm]{omega2_2.jpg } \end{array}$ ] @xmath43{accel3_2.jpg } & \includegraphics[width=50 mm]{omega3_2.jpg } \end{array}$ ] @xmath43{accel3_2_1.jpg } & \includegraphics[width=50 mm]{omega3_2_1.jpg } \end{array}$ ]
we suppose a model of an universe consists of ghost dark energy , van der waals fluid interacting with a modified fluid , which could be result of some interaction ( we could only suppose ) .
interaction between components were assumed to be @xmath1 . by this way
the role of the ghost dark energy were seen by interaction term only . during the investigation
we recover scale factor @xmath5 and present graphical analysis of @xmath41 and eos parameter @xmath42 .
analysis shows that the effects concerning to the presence of a ghost dark energy expressed in interaction term was observed in later stages of the evolution comparing with cases corresponding to absence and presence an interaction of the @xmath44 form between components . from the graphical analysis
we see , that having a situation proposed in this article it is not always true the assumption that dark energy could be the source of the accelerated expansion of the universe .
at least , for some values of the parameters of the model we observe that the mixture behaves as a dark energy with negative eos parameter , while @xmath45 .
for some cases we saw that dark energy really is responsible for accelerated expansion with @xmath46 .
almost the same result we obtain considering a modified fluid not described by the energy density of our consideration , but as a fluid described as @xmath47 . which corresponds to a cosmological constant model with @xmath48 .
here we have following possibilities , first that the modification of a fluid of our consideration could not work with van der waals gas , or probably an interaction of the consideration between components is not realistic and other modification should be done with interaction term .
there is other possibility , which from our opinion is also has right to be , that dark energy is not always was and will be responsible for the accelerated expansion of the universe .
this research activity has been supported by eu fonds in the frame of the program fp7-marie curie initial training network index no.289968 .
urban , a.r .
zhitnitsky , phys .
d 80 , 063001 ( 2009 ) ; + f.r .
urban , a.r .
zhitnitsky , jcap 09 , 018 ( 2009 ) ; + f.r .
urban , a.r .
zhitnitsky , phys .
b 688 , 9 ( 2010 ) ; + f.r .
urban , a.r .
zhitnitsky , nucl .
b 835 , 135 ( 2010 ) ; + n. ohta , phys . lett .
b 695 , 41 ( 2011 ) ; + r.g .
cai , z.l .
tuo , h.b .
zhang , arxiv:1011.3212 .
+ a. sheykhi , a. bagheri , europhys .
95 , 39001 ( 2011 ) ; + e. ebrahimi , a. sheykhi , phys .
b 705 , 19 ( 2011 ) ; + e. ebrahimi , a. sheykhi , int . j. mod
d 20 , 2369 ( 2011 ) ; + a. sheykhi , m. sadegh movahed , gen .
[ doi:10.1007/s10714 - 011 - 1286 - 3 ] . | we consider a model of a universe with ghost dark energy and van der waals fluid interacting with a fluid which was born as a result of interaction between original fluid and some other fluid existing in universe .
we suppose that ghost dark energy has its contribution to the model by an interaction term @xmath0 and we suppose that @xmath1 .
[ [ section ] ] we consider a universe composed of van der waals fluid @xcite , @xcite @xmath2 interacting with a modified fluid model @xcite @xmath3 @xmath4.\ ] ] where @xmath5 is a scale factor , @xmath6 is an interaction coefficient .
the origin of such fluid can be supposed to be an interaction or different types of interactions between an original fluid and other fluids in universe , which are coded in @xmath6 .
generally the origin of this fluid can be anything , but here we will not make assumption and will consider the universe in the presence of this fluid .
we will consider an interaction between last two fluids and we will suppose the presence of a ghost dark energy @xcite , @xcite with energy density given as @xmath7 where @xmath8 is the hubble parameter .
concerning to this component we assume that ghost dark energy with ( [ eq : ghost ] ) will appear and play a role only in interaction term .
dark energy was born in gr to explain experimental data claiming , that the expansion of the universe is accelerated .
dark energy with dark matter are exotic and mysterious components of the universe , capturing a lot of attention and not to be explained still .
generally , it is accepted that dark energy should be described by negative pressure and positive energy density , therefore implying negative eos parameter @xmath9 .
the simplest model is a cosmological constant @xmath10 introduced by einstein , but it suffers with the problem known as cosmological coincidence problem asking why are we living in an epoch in which the densities of dark energy and matter are comparable ? one of the ways to solve the cosmological coincidence problem is to consider the interaction between the components on phenomenological level .
generally , interaction could be considered as a function of energy densities and their derivatives : @xmath11 . to make everything working with units , we only should consider the fact that unit of interaction term @xmath12=\frac{[energy ~ density]}{time}$ ] and assume that unit @xmath13 could be contributed , for instance , from hubble parameter . over years
different models of interactions were proposed and considered , but everything was done on phenomenological level only .
very intensively were considered interactions @xmath14 , @xmath15 , @xmath16 , where @xmath17 is a coupling constant , other form of interactions , where question of time unit were solved with help of first order time derivative @xmath18 , @xmath19 , @xmath20 .
interaction of the general form @xmath21 , where @xmath22 also captured a lot of attention .
interaction between components arose as a result of splitting of the energy conservation , which mathematically can be described as follows : @xmath23 and @xmath24 which could be understood as : there is not energy conservation for the components separately , but due to interaction between all components , energy of whole mixture conserves .
this approach is correct for this moment only .
other type of interaction considered in literature and supported by experimental data is sign - changeable interaction @xcite , @xcite , @xcite .
motivated by @xcite we would like to consider an interaction where energy density of the ghost dark energy appears @xmath25 by this way we would like to make a possibility for other two components to feel an existence of other component which does not appear in field equations .
the mixture of our consideration will be described by @xmath26 and @xmath27 given by @xmath28 and @xmath29 paper organized as follow : after introduction , in next section we will give field equations and a strategy how to solve and find scale factor @xmath5 .
then in two other sections we will work in case when there is not interaction between components and then we will switch on interaction . in section discussion
obtained results will be summarized . |
let @xmath2 be a two - dimensional riemannian torus . by @xmath3 we denote the unique geodesic @xmath4 with the initial condition @xmath5 .
the geodesic flow on the unit tangent bundle @xmath6 is given by @xmath7 .
we also consider geodesics on @xmath1 , where @xmath1 is equipped with the lifted metric .
the topological entropy of a continuous dynamical system represents the exponential growth rate of orbits segments distinguishable with arbitrarily fine but finite precision .
it therefore describes the total exponential orbit complexity by a single number .
note that due to a theorem of a. katok @xcite for @xmath8-flows , @xmath9 , on 3-dimensional spaces positive topological entropy and the existence of a horse - shoe are equivalent .
we consider in this paper the questions which consequences on the behavior of geodesics on @xmath0 we can expect under the assumption of zero topological entropy and which geometrical restrictions on their behavior forbid high complexity of the geodesic flow . more precisely , we formulate necessary and sufficient conditions for zero topological entropy . in order to state the main theorems we have to define the asymptotic direction of geodesic rays , i.e. geodesics @xmath10 note that unlike in most text books in our context a geodesic ray does not have to be minimal .
given a geodesic @xmath11 then we associate to @xmath12 two geodesic rays given by @xmath13 and @xmath14 with @xmath15 .
+ + in the following it will be useful to distinguish the following types of geodesic rays ( see also @xcite ) .
+ [ types of geodesic rays ] + given a complete riemannian metric on @xmath1 .
a geodesic ray @xmath16 is called * _ bounded _ , if the set @xmath17 is bounded . * _ escaping _ , if @xmath18 , where @xmath19 denotes the euclidean norm on @xmath1 . *
_ oscillating _ , if @xmath12 is neither bounded nor escaping .
a geodesic @xmath20 is bounded ( escaping or oscillating ) if both of its geodesic rays @xmath21 and @xmath22 are bounded ( escaping or oscillating ) .
it will turn out that mixed cases of geodesics will not be relevant . furthermore
, their liouville measure is zero , as proved by m. wojtkowski in @xcite .
+ + let @xmath23 be an escaping geodesic ray on @xmath1 .
then , if the limit exists , @xmath24 is called the _ asymptotic direction _ of @xmath12 .
let @xmath25 be the canonical projection onto @xmath26 defined by @xmath27 then we call @xmath28 the _ rotation number _ of @xmath12 .
+ let @xmath29 be a geodesic ray such that its lift @xmath30 provides an escaping geodesic ray for which @xmath31 exists
. then we define @xmath32 and @xmath33 . + a rotation number @xmath34 is called _ rational _ if @xmath35 .
a direction @xmath36 is called _ rational _ if @xmath37 .
+ + 1 .
the definition and the existence of the asymptotic direction are independent of the chosen lift .
2 . a first definition of a rotation number for minimal geodesics induced by the slope of the accompanying euclidean lines goes back to g. a. hedlund ( see @xcite ) and h. m. morse ( see @xcite ) . in @xcite v. bangert presents a definition of the rotation number for minimal geodesics , which coincides with our definition restricted to minimal geodesics .
now we are able to state our first theorem .
+ * theorem i. * _ let @xmath38 be a riemannian metric on @xmath0 with vanishing topological entropy .
then , on the universal covering @xmath1 , every geodesic @xmath12 is escaping without self - intersections and the directions @xmath39 and @xmath40 exist .
furthermore , @xmath41 and @xmath12 lies in a bounded euclidean strip @xmath42 , where @xmath42 is a strip in @xmath1 bounded by two euclidean lines . _
+ + consider the asymptotic direction and the rotation number for a geodesic @xmath43 as functions on the unit tangent bundle defined as @xmath44 and @xmath45 , if they exist .
obviously @xmath46 and @xmath47 . +
* theorem ii . *
_ let @xmath38 be a riemannian metric on @xmath0 with vanishing topological entropy .
then the asymptotic direction @xmath48 and the rotation number @xmath49 are surjective continuous functions invariant under the geodesic flow . _ +
if the constant of motion @xmath50 would be differentiable with non - zero differential almost everywhere , the system would be integrable in the sense of liouville - arnold .
+ > from now on we represent @xmath0 as @xmath51 , where @xmath52 acts on @xmath53 via @xmath54 the non - trivial covering transformations @xmath55 will be called translation elements . sometimes we will denote @xmath55 only by @xmath56 .
a translation element @xmath56 is called primitive if it is not a nontrivial power of another translation element .
two translation elements @xmath56 and @xmath57 are called equivalent if there exist @xmath58 such that @xmath59 by @xmath60 $ ] we denote the corresponding equivalence classes .
a geodesic @xmath20 is called an axis if there exists a nontrivial translation element @xmath56 such that @xmath61 for some @xmath62 and all @xmath63 .
by @xmath64 we will denote the asymptotic direction of the axes of @xmath56 .
an axis @xmath65 of a translation element @xmath66 with @xmath67 is called a non - primitive axis if @xmath68 .
a geodesic @xmath69 is called prime - periodic if its lift is an axis of a primitive translation element @xmath56 . for a geodesic ray @xmath16
we define @xmath70 \mid \ # \{c([0 , \infty ) ) { \;\;\makebox[0pt]{$|$}\makebox[0pt]{$\cap$}\;\;}\eta c([0 , \infty ) ) \ } = \infty \text { for some $ \eta \in [ \tau]$}\}.\ ] ] we will now present a further characterization of the behavior of geodesics on the universal covering in the case of vanishing topological entropy . by intersections we always mean transversal intersections .
+ * theorem iii . * _ let @xmath2 be a riemannian torus with zero topological entropy
. then @xmath71 , for each geodesic ray @xmath16 .
furthermore , geodesics @xmath12 with irrational rotation number intersect their translates @xmath72 only a finite number of times .
geodesics @xmath12 with rational rotation number intersect their translates @xmath72 an infinite number of times at most for @xmath56 with @xmath73 . _
+ + in a previous version of this paper we have had a weaker formulation of the following theorem .
we originally showed applying hyperbolic dynamics and the curve shortening flow that the assumption of the theorem implies zero topological entropy .
v. bangert showed to us how to use variational arguments to even conclude flatness .
we also note that we obtained flatness under the stronger assumption that no geodesic on the universal covering intersects its translate transversally . +
* theorem iv .
* _ let @xmath38 be a riemannian metric on @xmath0 .
then flatness of the metric @xmath38 is equivalent to the condition that no axis @xmath20 on the universal covering intersects any of its translates . _ + + 1 .
the assumption of theorem iv is equivalent to the fact that all axes @xmath20 are axes of primitive elements . 2 .
as the theory of minimal geodesics and the curve shortening flow extend to symmetric finsler metrics , all theorems generalize to symmetric finsler metrics .
3 . m. l. bialy and l. polterovich @xcite and independently v. bangert @xcite present a formalism for minimal geodesics on @xmath0 based on a special class of orbits for monotone twist maps . for this setting
they formulate the notion of rotation number .
however , in general it is not possible to extend their formalism to non - minimal geodesics and arbitrary orbits of twist maps .
+ the study of monotone twist maps with variational methods and many results about their properties mentioned in @xcite and @xcite go back to j. n. mather @xcite and independently to s. aubry and p. y. le daeron @xcite . in @xcite s.
b. angenent studies orbits of monotone twist maps and their intersection properties in the case that these maps have vanishing topological entropy .
he formulates analogious results to parts of the statements of theorem i and theorem iii .
a different approach to understand the relation between the complexity of the geodesic flow and the behavior of geodesics similar to the one in this paper is presented by s. v. bolotin and p. h. rabinowitz in @xcite .
the paper is organized as follows : in the second section we introduce the notion of the topological entropy and discuss the for us relevant properties of the curve shortening flow . in section three
we show that vanishing topological entropy implies the non - existence of self - intersections of lifted geodesics and the non - existence of contractible geodesics on @xmath0 .
section four deals with geometric conditions which imply via the curve shortening flow positive topological entropy . in the case of vanishing topological entropy
we exclude the existence of oscillating geodesic rays .
under the same assumption , in section five we prove the existence and regularity of the rotation number .
in section six we study the intersection properties of geodesics on the universal covering with their translates . in the last section we provide a characterization for flatness of riemannian metrics on @xmath0 .
in particular , we show that no axis on the universal covering intersects its translates iff the torus is flat .
the topological entropy is invariant under topological conjugations and measures as described in the introduction the exponential orbit complexity by a single non - negative number .
the precise meaning becomes apparent in the following definition of topological entropy introduced by r. e. bowen @xcite .
[ topent ] let @xmath74 be a compact metric space , @xmath75 a continuous flow and @xmath76 the dynamical metric defined by @xmath77 for all @xmath78 .
we fix @xmath79 .
a subset @xmath80 is called @xmath81-_separated _ set of @xmath82 with respect to @xmath83 , if for @xmath84 it holds @xmath85 + the topological entropy of @xmath86 is defined as @xmath87 here @xmath88 denotes the maximal cardinality of any @xmath81-separated set of @xmath82 with respect to @xmath83 . for more details and properties of the topological entropy
see for example @xcite or @xcite .
+ + to prove that certain geometric constellations imply positive topological entropy of the geodesic flow on @xmath6 we will use the curve shortening flow on @xmath0 . in the following
we will give the precise definition of this flow and state the properties relevant in this context .
+ + let @xmath89 be a riemannian surface and @xmath90 a family of ( parameterized ) immersed smooth closed curves on @xmath91 .
we consider a continuous local semi - flow @xmath92 with @xmath93 and @xmath94 defined by @xmath95 where @xmath96 denotes the geodesic curvature of @xmath97 and @xmath98 its unit normal vector . this evolution equation defined in ( [ csf ] ) is called the _ curve shortening flow_. + + from the following theorem we will derive the existence of closed geodesics : [ go ] let @xmath91 be a smooth riemannian surface which is convex at infinity , i.e. the convex hull of every compact subset is compact .
let @xmath99 be a smooth curve , embedded in @xmath91 .
then , @xmath100 exists for @xmath101 , for some @xmath102 , satisfying the evolution equation ( [ csf ] ) .
+ if @xmath83 is finite , then @xmath97 converges to a point .
if @xmath83 is infinite , then the curvature of @xmath97 converges to zero in the @xmath103 norm , i.e. there exists a subsequence @xmath104 such that @xmath105 converges to a closed geodesic . the assumption that @xmath91 is convex at infinity ensures that the set of limit curves exists since an evolving curve @xmath97 can not leave a compact set . in our special case of the torus
we will directly show that the embedded curve @xmath97 stays in a compact set on the universal covering @xmath1 of @xmath0 or in a compact set on some unbounded cylinder @xmath106 for all @xmath107 .
then , by excluding that @xmath97 converges to a point we conclude the existence of a closed geodesic . +
as the length @xmath108 of the curve @xmath97 fulfills @xmath109 the length is a decreasing function of @xmath110 .
this is why this flow is called curve shortening .
+ an important fact is that embedded curves never become singular , unless they shrink to a point , as proved by m. a. grayson @xcite .
furthermore we will apply the following consequence of the maximum - principle for parabolic differential equations several times : [ 2seg ] let @xmath111 \to m$ ] be two curve - segments and @xmath112 , @xmath113 solutions of curve shortening which satisfy @xmath114 for all @xmath107 . then the number of intersections of the solutions @xmath97 and @xmath115 is a finite and nonincreasing function of @xmath116 .
the solutions intersect only transversally except at a discrete set of times @xmath117 , and at each @xmath118 the number of intersections of @xmath97 and @xmath115 decreases .
in 1998 j. denvir and r. s. mackay @xcite showed the following result as a conclusion of their study on geodesically convex surfaces of negative euler characteristic .
[ scc ] let @xmath38 be a riemannian metric on @xmath0 with a simple closed contractible geodesic @xmath12 .
then @xmath38 has positive topological entropy .
this theorem allows us to conclude the following lemma : [ 2-loops ] let @xmath38 be a riemannian metric on @xmath0 and @xmath11 a lift of a geodesic with two self - intersections such that @xmath119 and @xmath120 with @xmath121
. then @xmath38 has positive topological entropy .
we use similar ideas as s. angenent in @xcite .
let @xmath122 be a simple closed contractible and smooth curve such that @xmath123 and @xmath124)$ ] do not intersect on @xmath1 and such that @xmath125 $ ] lies in the bounded connected component of @xmath126 .
we apply the curve - shortening flow to @xmath123 and @xmath12 .
note that @xmath124)$ ] is a constant solution .
assume , that @xmath97 shrinks to a point .
then the solution @xmath97 has to pass @xmath124)$ ] under the flow .
consider @xmath127 with @xmath128 ) \not= \emptyset$ ] such that @xmath129 is the first time at which the segments meet . by theorem [ 2seg ] new intersections under the flow
can only appear in the endpoints @xmath130 or @xmath131 .
consider a small @xmath79 such that @xmath132 and @xmath133 .
as @xmath119 and @xmath120 the curve @xmath134 meets the geodesic segment @xmath135)$ ] also at @xmath136 , in contradiction to theorem [ 2seg ] .
hence , as @xmath97 can not pass @xmath124)$ ] it will not shrink to a point .
+ furthermore , @xmath97 will stay in a bounded set @xmath137 for all @xmath110 : we consider two non - equivalent primitive translation elements @xmath138 and two corresponding axes @xmath139 and @xmath140 .
for @xmath141 large enough the curve @xmath123 lies in the open strip between @xmath142 and @xmath143 , and in the open strip between @xmath144 and @xmath145 . and @xmath123 in the proof of lemma [ 2-loops ] .
] the intersection @xmath146 of the closures of these strips defines a region from which @xmath97 can not escape , because since @xmath123 does not intersect any of the bounding minimal geodesics , by theorem [ 2seg ] @xmath97 will not intersect them for all @xmath147 .
+ now we apply m. a. grayson s theorem [ go ] : since @xmath97 will not shrink to a point and stays in a compact set for all times , there exists a simple closed geodesic on @xmath1 . by theorem [ scc ]
@xmath38 has positive topological entropy .
+ in the sequel we will make use of the following standard recurrence theorem ( see for instance @xcite , page 157 ) .
let @xmath148 be a manifold and @xmath149 a probability measure invariant under a continuous flow @xmath150 .
then almost all @xmath151 are recurrent , i.e. there exists a sequence @xmath104 with @xmath152 as @xmath153 such that @xmath154 furthermore , if @xmath155 is recurrent all points on the orbit @xmath156 are recurrent as well .
if @xmath149 is positive on open sets , the set of recurrent points is dense .
[ no - intersections ] let @xmath38 be a riemannian metric on @xmath0 with zero topological entropy .
then no lift of a geodesic has a self - intersection .
let @xmath11 be a lift of a geodesic and assume that @xmath12 has a self - intersection
. then there exist @xmath157 with @xmath158 .
let @xmath159 denote the initial condition of @xmath12 in @xmath160 .
assume first that @xmath12 is recurrent .
then there exists an increasing sequence of times @xmath161 and a sequence @xmath162 of translation elements such that @xmath163 for @xmath153 , see figure [ int ] . for @xmath164
large enough and @xmath104 much larger than @xmath165 there exist @xmath110 near @xmath104 and @xmath166 such that @xmath167 .
hence , @xmath168 . as @xmath169 by lemma [ 2-loops ]
the metric @xmath38 has positive topological entropy , in contradiction to the assumption . and
@xmath170 in the proof of theorem [ no - intersections ] . ]
let @xmath12 be an arbitrary geodesic .
as recurrent geodesics are dense , there exists a sequence of recurrent geodesics @xmath171 with @xmath172 such that @xmath173 converges to @xmath174 .
since @xmath12 has a self - intersection , i.e. @xmath175 , the continuous dependency implies that for @xmath164 large enough the recurrent geodesic @xmath171 has a self - intersection .
then by lemma [ 2-loops ] the metric @xmath38 has positive topological entropy in contradiction to the assumption .
the following theorem is due to v. bangert in @xcite [ bounded ] let @xmath38 be a complete riemannian metric on @xmath1
. then the existence of a bounded geodesic ray @xmath176 implies the existence of a simple closed geodesic .
we note that this result can be also obtained combining curve shortening with arguments from topological dynamics . for more details about this approach
see @xcite where a slightly weaker result has been obtained .
[ bounded2 ] let @xmath38 be a riemannian metric on @xmath0 .
assume there exists a bounded geodesic ray @xmath12 on the universal covering @xmath1 .
then the metric @xmath38 has positive topological entropy .
in this section we present a fundamental geometric constellation on the universal covering of @xmath0 and prove that it implies positive topological entropy for the riemannian metric @xmath38 on @xmath0 . in the further sections we will use this argument several times . a continuous curve @xmath177 , for @xmath178 ,
is called a broken geodesic if there exists a finite set @xmath179 such that @xmath12 is geodesic on @xmath180 .
we call @xmath181 the set of vertices .
[ fundamentallemma ] let @xmath38 be a riemannian metric on @xmath0 and @xmath182 a minimal axis of the translation element @xmath56 .
let @xmath183 \to { \mathbb{r}}^2 $ ] and @xmath184 \to { \mathbb{r}}^2 $ ] be two geodesic segments with endpoints on @xmath65 and @xmath185 , @xmath186 .
assume that there exists a translation element @xmath57 , with @xmath187 such that @xmath188{$|$}\makebox[0pt]{$\cap$}\;\;}c_1([0 , a ] ) \not= \emptyset$ ] and @xmath189{$|$}\makebox[0pt]{$\cap$}\;\;}c_2([0 , b ] ) \not= \emptyset$ ] .
then the metric @xmath38 has positive topological entropy .
we choose @xmath190 such that @xmath191{$|$}\makebox[0pt]{$\cap$}\;\;}c_1([0 , a ] ) \not= \emptyset \quad \text{but } \quad \eta^{k+1 } \alpha({\mathbb{r } } ) { \;\;\makebox[0pt]{$|$}\makebox[0pt]{$\cap$}\;\;}c_1([0 , a ] ) = \emptyset.\ ] ] analogously we choose @xmath192 such that @xmath193{$|$}\makebox[0pt]{$\cap$}\;\;}c_2([0 , b])\not= \emptyset \quad \text{but } \quad \eta^{-(l+1 ) } \alpha({\mathbb{r } } ) \cap c_2([0 , b])= \emptyset.\ ] ] w.l.o.g .
. then @xmath195 ) = \emptyset \quad \text{but } \quad \eta \alpha({\mathbb{r } } ) { \;\;\makebox[0pt]{$|$}\makebox[0pt]{$\cap$}\;\;}\eta^{l+1 } c_2([0 , b ] ) \not= \emptyset.\ ] ] and @xmath196 in the proof of the fundamental lemma [ fundamentallemma ] . ] by @xmath197 we denote the geodesic strip bounded by @xmath65 and @xmath198 . in the next step
we use the segments @xmath199 and @xmath200 to construct suitable broken geodesics @xmath201 and @xmath202 which intersect .
we distinguish two cases : 1 .
there exists @xmath203 such that @xmath204 and @xmath200 intersect . in this case
we will denote @xmath205 and @xmath206 .
2 . for all @xmath203 the segments @xmath204 and @xmath200 do not intersect . using @xmath199 and @xmath207 we will construct a broken geodesic segment @xmath201 with endpoints on @xmath65 which intersects @xmath200 , see figure [ fig - ref4 ] . + and @xmath202 in the proof of the fundamental lemma [ fundamentallemma ] . ] + we choose @xmath208 such that on @xmath209 the set of intersections @xmath210{$|$}\makebox[0pt]{$\cap$}\;\;}\eta \alpha({\mathbb{r}})$ ] lies between the finite sets @xmath211{$|$}\makebox[0pt]{$\cap$}\;\;}\eta \alpha({\mathbb{r}})$ ] and @xmath212{$|$}\makebox[0pt]{$\cap$}\;\;}\eta \alpha({\mathbb{r}})$ ] .
+ let @xmath213)$ ] be the smallest segment of @xmath207 such that @xmath214 ) \cap m_+ \not= \emptyset \not= \eta \alpha([t_1,t_2 ] ) \cap m_-.\ ] ] let @xmath215 be the union of the bounded connected components of @xmath216 ) \cup \tau^n c_1([0 , a ] ) \cup \eta \alpha([t_1,t_2 ] ) \right).\ ] ] we consider the broken geodesic @xmath217)$ ] endowed with a new parametrization consisting of @xmath213)$ ] and segments of @xmath218 and @xmath199 , such that it separates @xmath219 from @xmath220 . by construction @xmath221)$ ] and @xmath222 ) = \eta^{l+1 } c_2([0 , b])$ ] intersect .
let @xmath223 be large enough such that @xmath224 for fixed @xmath225 and a finite sequence of @xmath226 elements @xmath227 with @xmath228 we consider the set @xmath229 for each @xmath230 let @xmath231 be a smooth curve such that @xmath232 , see figure [ fig - ref5 ] .
[ c]@xmath198 [ c]@xmath233 , @xmath202 , and @xmath234 in the proof of the fundamental lemma [ fundamentallemma].,title="fig : " ] furthermore , we can choose these curves such that the length of the segment of @xmath235 joining @xmath236 and @xmath237 is less than @xmath238 and larger than @xmath239 for some universal constants @xmath240 .
consider for each @xmath225 the cylinder @xmath241 where @xmath242 is the subgroup of @xmath243 generated by @xmath244 .
in particular , the projection of @xmath197 onto @xmath245 defines an annulus . by assumption the curves @xmath246 project to closed curves
@xmath247 , where @xmath248 is the projection of @xmath249 onto @xmath245 . by construction
the interior angles at the vertices @xmath250 of the connected component of @xmath251 containing @xmath246 are all less than @xmath252 . applying the curve shortening flow to the smooth curve @xmath246 we obtain smooth curves @xmath253 which stay in @xmath245 and do not intersect @xmath254 .
this follows from theorem [ 2seg ] and the fact that due to their homotopy class @xmath255 never become singular .
since the curves @xmath235 are not contractible , m. a. grayson s theorem [ go ] implies that the curvature of @xmath255 converges to zero .
this yields the existence of a closed geodesic on @xmath245 .
+ let @xmath256 be the bounded connected component of @xmath257 for @xmath258 .
let @xmath259 be a geodesic ball with radius @xmath79 . obviously , by construction the curves @xmath235 do not intersect @xmath260 for @xmath261 for all sequences @xmath230 and @xmath255 will not intersect @xmath260 for all @xmath261 and @xmath262 by the properties of the curve shortening flow .
+ by this construction for fixed @xmath226 and different sequences @xmath230 we get @xmath263 different closed geodesics on @xmath245 of length between @xmath264 and @xmath265 .
we project the constructed closed geodesics onto @xmath266 . identifying @xmath65 and @xmath198 we get a torus @xmath267 on which we still have at least @xmath268 different closed geodesics . by construction the initial conditions of these closed geodesics form a @xmath269-separated set which grows exponentially for @xmath270 . by definition [ topent ]
this implies positive topological entropy for the riemannian metric on @xmath267 .
as @xmath267 is a finite cover of @xmath0 also the riemannian metric @xmath38 on @xmath0 has positive topological entropy . for the proof of the fundamental lemma it is not necessary that @xmath65 is minimal .
it suffices the condition @xmath271 for the special @xmath272 used in the proof .
[ closed ] let @xmath2 be a riemannian torus .
the existence of a contractible closed geodesic on @xmath0 implies the existence of non - primitive axes for all translation elements @xmath56 . considering an arbitrary translation element and replacing in the proof of the fundamental lemma [ fundamentallemma ] the segments @xmath201 and @xmath202 by a closed geodesic we can use a similar construction , as presented starting in ( [ m2 ] ) , to produce a non - primitive axis
using the recurrence of the geodesic flow on @xmath0 we obtain the following refinement of the fundamental lemma .
[ 1 g ] let @xmath2 be a riemannian torus .
let @xmath11 be a geodesic and @xmath65 a minimal axis such that @xmath273 for @xmath274 .
if there exists a translation element @xmath56 with @xmath275 and @xmath276{$|$}\makebox[0pt]{$\cap$}\;\;}c ( [ a , b])\not= \emptyset$ ] , then @xmath38 has positive topological entropy .
let @xmath11 be a geodesic and @xmath65 a minimal axis such that @xmath273 for @xmath274 and @xmath56 the translation element with @xmath275 and @xmath276{$|$}\makebox[0pt]{$\cap$}\;\;}c ( [ a , b])\not= \emptyset$ ] .
we will now prove that this implies positive topological entropy .
+ consider the two halfplanes given by the connected components of @xmath277 and denote by @xmath215 the one containing @xmath278 and by @xmath279 the other one . by assumption there
exists @xmath280 $ ] such that @xmath281 .
+ we can assume that @xmath282 does not intersect @xmath283 since this would immediately imply positive topological entropy by the fundamental lemma [ fundamentallemma ] .
we can also assume that @xmath12 is recurrent , since the recurrent geodesics are dense .
in particular , there exists an increasing sequence @xmath161 and a sequence @xmath162 of translates such that @xmath284 for @xmath153 .
but then for @xmath164 large enough @xmath285)$ ] must intersect the translates @xmath286 and @xmath287
. then , @xmath285)$ ] fulfills the assumption of the fundamental lemma [ fundamentallemma ] and hence , @xmath38 has positive topological entropy and there exist non - primitive axes of @xmath56 .
[ 2 g ] let @xmath2 be a riemannian torus .
let @xmath288 be a geodesic ray on the universal covering @xmath1 which intersects a minimal axis @xmath65 and disjoint translate infinitely often .
then @xmath38 has positive topological entropy .
[ osc ] let @xmath38 be a riemannian metric on @xmath0 .
then the existence of an oscillating geodesic ray @xmath12 on @xmath1 implies positive topological entropy for the metric @xmath38 .
due to the definition of an oscillating geodesic ray there exist sequences @xmath289 tending to infinity and a compact set @xmath290 such that @xmath291 and @xmath292 for @xmath153 . w.l.o.g .
we can assume that @xmath293 for some @xmath294 .
let @xmath182 be a minimal axis such that @xmath295 .
furthermore , we can choose @xmath65 and a disjoint translate @xmath296 in such a way that @xmath297 and @xmath298 are contained in the two different halfplanes of @xmath299 for sufficiently large @xmath300 .
this implies that the geodesic ray @xmath288 will intersect @xmath65 and @xmath296 infinitely often . by corollary
[ 2 g ] the metric @xmath38 has positive topological entropy .
in the proof of theorem i we will need the following property of minimal geodesics : as already mentioned , the study of minimal geodesics on @xmath0 goes back to h. m. morse @xcite , g. a. hedlund @xcite , and v. bangert @xcite .
central results are that for each @xmath301 there exists a minimal geodesic @xmath11 with asymptotic direction @xmath302 and that furthermore there exists a constant @xmath303 , such that for each minimal geodesic @xmath20 there exists a euclidean line @xmath304 , and for each euclidean line @xmath304 there exists a minimal geodesic @xmath12 such that @xmath305 furthermore , we can assume that @xmath306 is larger than the diameter of the fundamental domain .
as shown by v. bangert @xcite , for irrational rotation numbers the set of minimal geodesics with this rotation number is totally ordered , i.e. , all these minimal geodesics have pairwise no intersections with each other . for the set of minimal geodesics with a fixed rational rotation number the subset of axes
is ordered .
two minimal axes with the same rotation number bounding a strip containing no further minimal axes are called neighboring minimals .
combining corollary [ bounded2 ] , lemma [ osc ] , and theorem [ no - intersections ] we conclude that in the case of vanishing topological entropy all geodesic rays are escaping and have no self - intersections on the universal covering @xmath1 .
+ if for @xmath21 the asymptotic direction does not exist , there exist two accumulation points @xmath307 for the quotient @xmath308 as @xmath110 tends to @xmath309 .
choose a rational direction @xmath294 which lies in the connected component of @xmath310 , which is not larger than a half - circle .
let @xmath65 be a minimal axis of the primitive translation element @xmath56 with @xmath311 .
then the geodesic @xmath312 must intersect @xmath278 and a given disjoint translate @xmath313 with @xmath314 \not= [ \tau]$ ] ( which is a minimal axis of @xmath56 as well ) infinitely often .
but then , by corollary [ 2 g ] the metric @xmath38 has positive topological entropy in contradiction to the assumption .
+ assume that @xmath315 .
choose @xmath294 rational such that @xmath316 both lie in one connected component of @xmath317 .
let @xmath65 be an axis with @xmath318 which intersects @xmath12 .
then @xmath65 divides @xmath1 in two halfplanes .
there is precisely one of those halfplanes denoted by @xmath215 for which there exists @xmath319 such that @xmath320 for @xmath321 .
choose a disjoint translate @xmath322 which is contained in @xmath215 as well . then there exist @xmath323 such @xmath324 and @xmath325 { \;\;\makebox[0pt]{$|$}\makebox[0pt]{$\cap$}\;\;}\alpha({\mathbb{r } } ) \not= \emptyset$ ] which contradicts theorem [ 1 g ] .
+ we will now show that there exists a euclidean strip @xmath42 with direction @xmath39 such that @xmath326 .
we already know that the asymptotic directions exist and that @xmath41 .
assume that @xmath12 is not contained in a euclidean strip .
consider an arbitrary minimal geodesic with asymptotic direction @xmath39 .
as minimal geodesics are accompanied by euclidean lines with the same direction and as @xmath12 is not bounded by any strip with direction @xmath39 , c also intersects an infinite number of translates of this geodesic .
consider @xmath303 as introduced in the previous remark .
choose a translation element @xmath56 , a time @xmath319 and a minimal geodesic @xmath327 with asymptotic direction @xmath39 such that @xmath328 $ ] crosses @xmath327 and @xmath329 , @xmath330 and @xmath331 lie in different halfplanes of @xmath332 and such that there fits a euclidean strip @xmath333 with direction @xmath39 and width @xmath334 between @xmath327 and @xmath329 , see figure [ fig - ref6 ] .
[ r]@xmath331 then there exists a euclidean strip @xmath197 with width @xmath335 and a rational asymptotic direction @xmath336 near @xmath39 on @xmath337 but @xmath338 such that @xmath330 and @xmath331 lie in different halfplanes of @xmath339 . as @xmath197 has width @xmath335 , by the previous remark
there exists a minimal axis @xmath65 and a disjoint translate @xmath207 with asymptotic direction @xmath318 and @xmath340 .
furthermore , the minimal axes @xmath341 and @xmath342 intersect @xmath21 . since @xmath343 , the ray @xmath21 has to intersect @xmath341 and @xmath342 a second time which implies positive topological entropy by theorem [ 1 g ] .
now we like to prove regularity properties of the rotation number .
we begin with the following lemma which holds for all riemannian metrics .
[ surj ] for all @xmath301 and @xmath344 there exists @xmath345 such that the corresponding geodesic @xmath4 fulfills @xmath346 , @xmath347 and for the lift of @xmath3 on the universal covering it holds @xmath348 .
let @xmath349 .
then , for fixed @xmath344 the map @xmath350 is surjective .
we fix @xmath301 and @xmath344 . by the remark at the beginning of this section
there exists a minimal geodesic @xmath351 with rotation number @xmath302 .
we assume that @xmath352 , otherwise we set @xmath353 after a reparameterization such that @xmath354 and so we choose @xmath355 .
+ we consider minimal geodesic segments @xmath356 connecting @xmath357 and @xmath358 on the universal covering such that @xmath359 with @xmath360 . as @xmath361 is compact there exists a sequence @xmath104 with @xmath152 such that @xmath362 converges to @xmath174 . by its minimality the segment @xmath356
intersects @xmath363 only once in @xmath358 .
+ let @xmath364 be a translate of @xmath351 such that @xmath357 lies in the bounded strip @xmath197 between @xmath363 and @xmath365 . repeating the minimality arguments we conclude that @xmath356 and @xmath365 have no intersections for all @xmath110 . by the continuous dependence on the initial conditions also the limit geodesic ray @xmath21 with @xmath366 and @xmath367 will not intersect @xmath365 and @xmath363 .
+ hence , the geodesic ray @xmath21 lies in the geodesic strip @xmath197 with rotation number @xmath302 .
this implies the existence of the rotation number and even that @xmath368 .
we extend @xmath21 on @xmath0 to the geodesic @xmath69 with @xmath354 and @xmath369 .
the choice @xmath370 fulfills the required properties . by lemma [ surj ]
the rotation number is surjective .
we have to show its continuity .
as the topological entropy vanishes , by theorem i the asymptotic directions @xmath371 and @xmath372 exist and it holds @xmath373 for all @xmath374 . here , we think of the asymptotic directions as a function on the unit tangent bundle of @xmath1 .
assume that @xmath375 is not continuous at @xmath376 .
let @xmath12 be the geodesic with the initial condition @xmath377 and let @xmath378 . as @xmath375 is not continuous at @xmath379 , there exists a sequence @xmath380 such that @xmath381 does not converge to @xmath336 . since @xmath337 is compact there exists a subsequence @xmath382 with @xmath383 .
we will again denote this subsequence by @xmath384 and the geodesics corresponding to @xmath384 by @xmath171 , i.e. @xmath385 .
+ consider the two euclidean rays @xmath386 and @xmath387 given by @xmath388 and @xmath389 .
choose a minimal axis @xmath390 and a disjoint translate @xmath391 such that the two halfplanes @xmath392 and @xmath393 which are two of the three connected components of @xmath394 have the following properties : first , there exists an @xmath79 such that @xmath395 .
second , the sets @xmath396 and @xmath397 are unbounded , see figure [ fig - ref7 ] .
[ c]@xmath398 and @xmath171 in the proof of theorem ii.,title="fig : " ] since @xmath378 and @xmath395 , the continuous dependence on the initial conditions implies the existence of @xmath399 such that for all @xmath400 we have @xmath401 and @xmath402 for a suitable @xmath403 .
moreover , @xmath404 as @xmath153 yields the existence of @xmath405 such that the set @xmath406 is unbounded for each @xmath407 .
in particular , for a given @xmath407 we will find @xmath408 such that @xmath409 .
such a geodesic @xmath171 crosses the pair of minimal axes @xmath351 and @xmath410 at least twice which implies by theorem [ 1 g ] positive topological entropy in contradiction to the assumption in theorem ii .
hence , the asymptotic direction is continuous and its continuity implies the continuity of the rotation number .
in this section we show that vanishing topological entropy implies strong restrictions of the intersections of geodesics on the universal cover with their translates . as @xmath411 , by theorem i the asymptotic direction @xmath412 exists for all @xmath12 and for each @xmath12 there exists a euclidean strip @xmath42 bounding @xmath12 .
assume there exists a geodesic ray @xmath413 with @xmath414 .
consider a translation element @xmath56 with @xmath415{$|$}\makebox[0pt]{$\cap$}\;\;}\tau c\}= \infty$ ] and @xmath416 ,
e.g. it does not leave @xmath42 invariant .
let @xmath417 be large enough such that the distance between @xmath42 and @xmath418 is larger than @xmath334 with @xmath303 introduced in the remark at the beginning of the previous section .
we denote the euclidean strip between @xmath42 and @xmath419 by @xmath420 . for each @xmath421
we denote the connected component of @xmath422 which contains @xmath423 by @xmath424 and by @xmath425 the other one .
+ as @xmath326 it follows that @xmath426 .
consider @xmath427 .
as @xmath415{$|$}\makebox[0pt]{$\cap$}\;\;}\tau c\}= \infty$ ] there exist @xmath428 with @xmath429 such that @xmath72 passes in @xmath130 from @xmath430 to @xmath431 .
analogously there exist @xmath432 with @xmath433 such that @xmath434 passes in @xmath435 from @xmath436 to @xmath437 . by this construction
we get a broken geodesic segment @xmath438:=c[t_0,t_1 ] \cup \tau c[\tilde t_1 , t_2 ] \cup \tau^2 c[\tilde t_2 , t_3]\cup \dots
\cup \tau^{k-1 } c[\tilde t_{k-1 } , t_{k}]\ ] ] connecting @xmath12 and @xmath439 .
consider @xmath440 and analogously to the previous construction there exist @xmath441 with @xmath442 such that @xmath443 passes in @xmath444 from @xmath445 to @xmath446 .
analogously we get the broken geodesic segment @xmath447:=\tau^k c[s_0,s_1 ] \cup \tau^{k-1 } c[\tilde s_1 , s_2 ] \cup \tau^{k-2 } c[\tilde s_2 , s_3]\cup \dots \cup \tau c[\tilde s_{k-1 } , s_{k}].\ ] ] gluing these two geodesic segments we obtain the broken geodesic segment @xmath448 ) \cup
\tau^k c([\tilde t_{k } , s_0 ] ) \cup \beta_1([s_0 , s_{k}])\ ] ] which connects @xmath42 , @xmath418 , and @xmath42 .
we call the unbounded connected component of @xmath449 which is not a euclidean halfplane the exterior of @xmath450 . by construction
the exterior angles of @xmath450 are smaller than @xmath252 .
analogously we construct a broken geodesic @xmath451 connecting @xmath418 , @xmath42 , and @xmath418 such that the exterior angles of @xmath452 are smaller than @xmath252 .
consider two minimal axes @xmath453 and @xmath454 of @xmath56 bounding a geodesic strip @xmath455 such that @xmath456 . as
the distance of @xmath42 and @xmath418 is larger than @xmath334 consider three ordered translates of minimal axes denoted by @xmath457 , @xmath458 , and @xmath459 with a rational asymptotic direction near @xmath39 such that @xmath460 for all @xmath461 .
then , there exist subsegments @xmath199 and @xmath462 of @xmath463 and @xmath451 with endpoints on @xmath458 intersecting @xmath457 or @xmath459 , respectively .
these broken geodesic segments @xmath199 and @xmath462 fulfill by construction the assumptions of the segments in the fundamental lemma [ fundamentallemma ] .
[ c]@xmath42 [ c]@xmath419 and @xmath451 in the proof of theorem iii.,title="fig : " ] hence , @xmath464 in contradiction to the assumption .
furthermore , by construction we conclude , that geodesics @xmath12 with irrational rotation number intersect their translates @xmath72 only a finite number of times .
geodesics @xmath12 with rational rotation number intersect their translates @xmath72 an infinite number of times at most for @xmath56 with @xmath73 .
in order to prove theorem iv we first prove the following lemma due to v. bangert which is of independent interest . [ lemmaflat ]
let @xmath465 be a riemannian torus .
suppose there exists a primitive translation element @xmath56 such that its minimal axes do not foliate @xmath53
. then there exists @xmath67 and a non - primitive axis of @xmath66 .
if the minimal axes of @xmath56 do not foliate @xmath53 , then there exist two neighboring minimal axes @xmath466 of @xmath56 .
consider the cylinder @xmath467 and the projections @xmath468 of @xmath469 and @xmath470 which are closed geodesics of equal length @xmath471 . by assumption
all closed curves between @xmath199 and @xmath462 in the same homotopy class have length strictly larger than @xmath471 .
choose locally convex neighborhoods @xmath266 and @xmath472 about @xmath199 and @xmath462 .
( take for instance minimal geodesic loops homotopic to @xmath199 and @xmath462 , respectively , close to @xmath199 and @xmath462 .
) there exists @xmath79 such that for all closed curves @xmath351 homotopic to @xmath199 with @xmath473 we obtain that @xmath474 using the methods in @xcite , pp .
87/88 we obtain a constant @xmath475 such that for all @xmath476 there exists a closed geodesic @xmath477 between @xmath199 and @xmath462 with the following properties : choose @xmath481 large enough such that @xmath482 .
then there does not exist a curve @xmath483 homotopic to @xmath484 \to c$ ] whose @xmath164-th iterate is equal to @xmath477 . otherwise using property ( ii ) , @xmath485 but by ( @xmath486 ) this
would imply that @xmath483 and hence @xmath477 is contained in @xmath487 which contradicts property ( iii ) .
hence , @xmath477 is a non - primitive axis of @xmath488 .
obviously the flatness of the metric on @xmath0 implies that no periodic geodesic ( and in fact no geodesic ) intersects its translates .
conversely , assume that no axis intersects its translates
. then lemma [ lemmaflat ] implies that for each primitive translation element @xmath56 the associated minimal axes foliate @xmath53 .
since all axes are minimal the result of n. innami in @xcite implies that @xmath2 does not have conjugate points .
then by e. hopf s theorem @xcite the metric @xmath38 is flat . _
_ the authors like to thank sigurd angenent for explaining to us the essential features of the curve shortening flow which is a crucial technique in this paper .
we are also grateful to victor bangert for explaining to us his variational methods which led to theorem iv .
99 s. b. angenent , _
monotone recurrence relations , their birkhoff orbits and topological entropy _ , ergod . th . and dynam .
sys . , * 10 * ( 1990 ) , 15 - 41 .
s. b. angenent , _ parabolic equations for curves on surfaces , part ii .
intersections , blow - up , and generalized solutions _ , annals of mathematics , * 133 * ( 1991 ) , 171 - 215 .
s. b. angenent , _ self - intersecting geodesics and entropy of the geodesic flow _ ,
preprint 2005 , ` http://www.math.wisc.edu/\simangenent/preprints.html ` s. aubry , p. y. le daeron , _ the discrete frenkel - kontorova model and its generalizations _ , physica , * 8d * ( 1983 ) , 381 - 422
. v. bangert , _ closed geodesics on complete surfaces _ , math .
* 251 * ( 1980 ) , 83 - 96 .
v. bangert , _ geodesics and totally convex sets on surfaces _ , invent . math . * 63 * ( 1981 ) , 507 - 517 . v. bangert , _ mather sets for twist maps and geodesics on tori _ , dynamics reported * 1 * ( 1988 ) , 1 - 56 .
m. l. bialy and l. v. polterovich , _ geodesic flows on the two - dimensional torus and phase transitions commensurability - noncommensurability _ , english translation : functional anal .
appl . * 20 * ( 1986 ) no . 4 , 260 - 266 . s. v. bolotin and p. h. rabinowitz , _ some geometrical conditions for the existence of chaotic geodesics on a torus _ , ergod .
dynam . sys .
* 22 * ( 2002 ) , 1407 - 1428 .
r. bowen , _
entropy for group endomorphisms and homogeneous spaces _ ,
trans . of am .
soc . * 153 * ( 1971 ) , 401 - 414 .
r. bowen , _ entropy - expansive maps _ , trans . of am .
* 164 * ( 1972 ) , 323 - 331 .
j. denvir and r. s. mackay , _ consequences of contractible geodesics on surfaces _ , trans .
* 350 * ( 1998 ) , no .
11 , 4553 - 4568 .
e. glasmachers , _ characterization of riemannian metrics on @xmath0 with and without positive topological entropy _ ,
thesis 2007 , ` http://www-brs.ub.rub.de/netahtml/hss/diss/glasmacherseva/ ` m. a. grayson , _ shortening embedded curves _ , ann . of math .
( 2 ) * 129 * ( 1989 ) , no . 1 , 71 - 111 . g. a. hedlund , _ geodesics on a two - dimensional riemannian manifold with periodic coefficients _ , ann . of math .
* 33 * ( 1932 ) , 719 - 739 .
e. hopf , _ closed surfaces without conjugate points _ , proc .
sci . * 34 * ( 1948 ) , 47 - 51 .
n. innami , _ families of geodesics which distinguish flat tori _ , math .
j. okayama univ .
* 28 * ( 1986 ) , 207 - 217 .
a. katok , b. hasselblatt , _ introduction to the modern theory of dynamical systems _ , cambridge university press ( 1995 ) .
a. katok , _ lyapunov exponents , entropy and periodic orbits for diffeomorphisms _ , inst .
hautes tudes sci .
publ . math .
* 51 * ( 1980 ) , 137 - 173 .
j. n. mather , _ existence of quasi - periodic orbits for twist homeomorphisms of the annulus _ , topology , * 21 * ( 1982 ) , 457 - 67
. h. m. morse , _ a fundamental class of geodesics on any closed surface of genus greater than one _ , trans .
* 26 * ( 1924 ) , 25 - 60 .
m. wojtkowski , _ oscillating geodesics on two - dimensional manifolds _
, asterisque * 51 * ( 1978 ) , 443 - 356 .
p. walters , _ an introduction to ergodic theory _
, graduate texts in mathematics , springer - verlag , new york , berlin , heidelberg ( 1982 ) . | in the present work we consider the behavior of the geodesic flow on the unit tangent bundle of the 2-torus @xmath0 for an arbitrary riemannian metric . a natural non - negative quantity which measures
the complexity of the geodesic flow is the topological entropy . in particular
, positive topological entropy implies chaotic behavior on an invariant set in the phase space of positive hausdorff - dimension ( horseshoe ) .
we show that in the case of zero topological entropy the flow has properties similar to integrable systems .
in particular there exists a non - trivial continuous constant of motion which measures the direction of geodesics lifted onto the universal covering @xmath1 .
furthermore , those geodesics travel in strips bounded by euclidean lines
. moreover we derive necessary and sufficient conditions for vanishing topological entropy involving intersection properties of single geodesics on @xmath0 . |
the search for an understanding of an energy source great enough to explain the gamma - ray burst ( grb ) phenomena has attracted much attention from the astrophysical community since its discovery .
one of the more recent candidates for the grb model is based on the fire - ball model.@xcite the proposal for the central engine for such bursts concerns @xmath3 production and their subsequent scattering in the environs of a dense accretion disk / low - mass black hole system @xcite .
such a system may be formed by the merger of a black hole and a neutron star , the merger of two neutron stars , by a collapsar , or by a supernova @xcite .
the astrophysical details of the geometry or environment of such systems are currently hidden from us both observationally and computationally , although this situation may change in the near future with the development of gravity wave observatories and more sophisticated general relativistic astrophysical simulation capabilities . for the mean time in this paper
we focus our attention on a rather simplified model for the scattering of neutrinos and anti - neutrinos in the vicinity of a low - mass rotating black hole .
our goal is to extend the recent work of asano & fukuyama ( af ) @xcite and salmonson and wilson @xcite ( sw ) by providing the first general - relativistically covariant off - axis calculation of the energy momentum deposition rate ( mdr ) .
we focus our attention , as did the previous authors , on the general relativistic enhancement or degradation of the deposited mdr . since we now intend to calculate the off - axis contributions , the direction of the energy - momentum 4-vector
can no longer be assumed , as opposed to the along axis case , where symmetry forces the net energy upward along the symmetry axis .
an accurate treatment of off - axis contributions requires a full 4-dimensional spacetime calculation , as opposed to the analytic and semi - analytic calculations performed by af and sw .
therefore , we derive the full energy - momentum ( hereafter referred to as `` momenergy'')@xcite deposited in the vicinity of the black hole / accretion disk system . the calculations are based on the basic principles as discussed in misner , thorne and wheeler.@xcite we consider the direction of the momenergy in the frame of an observer at a given point above the disk .
our aim in this work is not solely to study and draw conclusions on the relativistic case of neutrino and anti - neutrino collisions above the accretion disk for grb models , but also to develop a method to provide a more in - depth understanding of the scattering mechanisms in the environs of small accreting black hole systems . for the above purpose
, we consider an idealized situation which includes gravitational redshift , the bending of neutrino trajectories , and the redshift due to accretion disk rotation .
this paper is organized as follows : in sec .
[ sec1 ] and sec .
[ sec2 ] we define the appropriate global spacetime metric and establish the frame of the observer .
the derivation of the momenergy deposition rate ( mdr ) is discussed in sec .
[ sec3 ] . sec .
[ sec4 ] deals with the computation of the mdr in an observer s frame with regards to evaluating the number densities and the generation of the 4-momentum from the @xmath4-@xmath5 scattering . in sec .
[ sec5 ] we give an estimate of the cumulative effect of the scattering on the energy deposition at a radius of @xmath6 from the black hole . finally , in sec . [ sec6 ]
we draw some conclusions , discuss the applicability of our results and methods on a larger scale , and make recommendations for future work in this area .
we use the boyer - lindquist ( bl ) spacetime metric in our neutrino ray tracing code . @xmath7 where @xmath8 with @xmath9 .
the momenergy deposition rate ( mdr ) is computed in the frame of an observer resting with respect to the global coordinates @xmath10 at a point with coordinates @xmath11 .
spacetime is stationary which implies that spatial pictures are sufficient .
also , @xmath12 can be fixed ( say @xmath13 ) ; however , the observational effects are nonetheless ordinarily only representable in 3-dimensions because of frame dragging .
the frame of the observer is defined by four orthonormal basis vectors ( @xmath14 ) .
we define these basis vectors such that ( 1 ) the basis is a lorentz frame ( @xmath15 ) , and ( 2 ) the 4-velocity , or temporal basis vector of the observer ( @xmath16 ) is parallel to the time component of the bl coordinate frame , @xmath17 . in the global coordinate frame
@xmath18 @xmath19 this leads to @xmath20 likewise @xmath21 vector @xmath22 is determined by the conditions @xmath23 , @xmath24 , @xmath25 , which leads to @xmath26
the mdr in the frame of the observer is given by @xmath27 where @xmath28 , @xmath29 are number densities in phase space ( lorentz invariant ) , and @xmath30 is the rest frame cross section .
the expression in curly brackets is lorentz invariant and can be computed as @xmath31 where @xmath32 @xmath33 in eq .
( [ dw ] ) , @xmath34 is the weinberg angle , @xmath35 and the plus sign is used for @xmath36 pairs , while minus is used for @xmath37 and @xmath38 pairs . in order to perform integration in spherical coordinates
, we use expressions @xmath39 and @xmath40 , where @xmath41 is the unit vector in the direction of @xmath42 and @xmath43 is the solid angle element . equation ( [ mdr ] ) then reduces to @xmath44
the number densities @xmath28 , @xmath29 are lorentz invariant and conserved along the @xmath4 , @xmath5 world lines ( null geodesics ) . if @xmath45 is the neutrino ( anti - neutrino ) 4momentum in the observer frame ( @xmath46 ) , and @xmath47 is the 4momentum of the same neutrino in the comoving frame of the disk at the point of emission ( @xmath48 ) then @xmath28 is computed as @xmath49 where @xmath50 is the specific intensity of radiation at a given frequency @xmath4 .
note that the ratio @xmath51 is lorentz invariant by liouville s theorem .
the value of @xmath52 in this equation is determined by the 4momentum of the neutrino parallel translated to the disk along the world line of the neutrino @xmath53 and on the 4velocity of the particle in the disk emitting the neutrino @xmath54 .
the latter is computed as @xmath55 where@xcite @xmath56 for the direct circular orbit of the emitting particle .
the normalization condition @xmath57 written as @xmath58 allows the evaluation of @xmath59 as @xmath60^{-\frac{1}{2}}\ ] ] the value of @xmath61 is given by @xmath62 the steps for evaluating @xmath63 are as follows : 1 .
@xmath64 in the frame of observer ( supplied by the integrating program ) 2 .
@xmath65 transformation to global coordinates 3 . tracing geodesics starting at @xmath66 tangent to @xmath42 to the disk .
pick the geodesic parameter @xmath67 such that @xmath68 , integrate back to @xmath69 ( @xmath70 at the observer , @xmath71 at the disk ) .
this procedure yields @xmath72 and @xmath73 4 . computing @xmath74 5 . computing @xmath75 6 .
computing @xmath76 using equation ( [ fnu ] ) assuming the emission rates of the neutrinos and the anti - neutrinos at the disk are identical , and assuming the neutrinos are emitted isotropically as a planck black - body spectrum characterized by the temperature @xmath77 in the frame comoving with the disk material , then the spectrum of neutrinos that our observer sees is also a planck spectrum .
however , the temperature characterizing our observer s spectrum is scaled by the frequency shift of the neutrinos . @xmath78 following the derivation of salmonson and wilson@xcite , the mdr integral ( eq.[mdr1 ] ) can be transformed into a 4-dimensional integral by analytically integrating over @xmath79 and @xmath80 .
@xmath81 here , the temperature @xmath82 is the temperature of the black body spectrum of the neutrinos as measured by the observer @xmath83 .
it will ordinarily depend on both the sky angles ( @xmath84 , @xmath85 ) we integrate over . in a similar way @xmath86 .
using eqn .
[ tscale ] we can rewrite the momenergy deposition rate in terms of the disk temperature .
@xmath87 where @xmath88 and @xmath89 are the bl coordinate location of the observer , the integral function is driven by the redshift of the neutrinos and anti - neutrinos arriving at the observer , @xmath90 and the constant @xmath91 depends strongly on the temperature of the disk as measured by the keplerian observer .
@xmath92 in order to calculate the generation of 4-momentum from the @xmath4@xmath93 scattering , we must first image the accretion disk in the observer s reference frame .
we do this by using a full 3-dimensional ray - tracing code for a kerr spacetime . the input needed by the ray - tracing code in order to image a given pixel of the disk is the location of the observer ( @xmath94 ) and the components of the 4-momentum ( @xmath95 ) in boyer - lindquist coordinates .
@xmath96 we can express the 4-momentum of the neutrino in the observer s frame by using the two spherical sky angles ( @xmath84 and @xmath85 ) shown in fig .
[ skyangles ] . @xmath97 the magnitude of the 3-momentum ( @xmath98 ) , and the time component of the 4-momentum , were obtained by assigning an energy to the neutrino in the observers frame ( @xmath99 ) and using the normalization condition , @xmath100 .
we can recover these components in the boyer - lindquist coordinates by inverting eqns.([e0h],[erh],[eth],[eph ] ) .
@xmath101 to reconstruct the image of the entire disk at @xmath102 , we integrate over all values of the sky angles ( @xmath103 and @xmath104 ) .
we use the ray - tracing code@xcite to calculate the 4-momentum of the neutrino ( @xmath105)when it hits the disk . in fig .
[ diskimage ] we provide an illustrative example of the imaging of an accretion disk extending from @xmath106 ( the inner - most stable orbit of a kerr black hole of mass @xmath107 and specific angular momentum @xmath108 ) out to @xmath109 . here
the observer @xmath102 is located at bl coordinates @xmath110 and @xmath111 .
the image of the disk is color coded as to the energy of the neutrinos reaching the observer .
this image is used to calculate the mdr ( eqn .
[ mdr2 ] ) .
we also provide a convergence plot for the mdr in fig .
[ convmdr ] .
up to this point we have calculated the total momenergy density created in a given observer s orthonormal frame per unit 3-volume per unit proper time .
these results are posed properly in general relativity as they represent a meaningful local observation .
however , for the grb application we address in this paper it would be useful to determine the electron distribution function for an observer at large radii ( e.g. @xmath112 ) from the accretion disk .
this location would approach the inner envelope of a hypothetical stellar shell surrounding the black - hole / accretion disk system .
at such a location we could construct a source term for the fluid at the inner envelope of the star that would provide the energy for a jet which would ultimately punch through the outer atmosphere of the star a jet that would be the generator of a fireball - type solution .
unfortunately , the construction of such a source is a computationally difficult problem even under the ideal assumptions that the electrons created evolve as a collisionless fluid and are highly energetic ( @xmath113 ) .
if we want to know how many electrons with momentum @xmath114 arrive at an observer located at spacetime event @xmath115 , then we would need to construct the geodesic into the past tangent to @xmath114 at @xmath115 and sum up the electron creation functions along and tangent to that geodesic.@xcite this would require modifying the calculation of the mdr ( eq . [ mdr ] ) to calculate the probability distribution functions for the creation of electrons at each point . we would then need to construct and store these electron creation functions on a dense grid in the @xmath116-@xmath115 plane perpendicular to the disk and integrate these creation distribution functions over the 3-parameter ( 2 sky angles and electron energy defining @xmath114 ) family of geodesics of the observer at @xmath115 . alternatively , we could solve a 3-dimensional collisionless boltzmann equation for the kerr spacetime geometry.@xcite unfortunately , both of these approaches , while correct in a gr sense , are computationally demanding in the extreme . while there is no technical reason that would prevent us from carrying out these calculations ,
practical considerations prevent us from completing them in the near future .
consequently , we elect to pursue a simpler computational strategy for the estimation of the grb jet source term . the difficulties with a general covariant treatment originate with the kerr black hole and not with the schwarzschild solutions . in particular , we currently calculate the total 4-momentum deposited per unit 4-volume in the observer s orthonormal frame .
any attempt to assign a 4-volume to the observer in bl coordinates ( or any coordinate ) would break general covariance .
in other words , the spatial slices of the bl metric are not orthogonal to the time lines ( @xmath117 ) , and thus the 3-space of our observers is not integrable .
nevertheless , if we take the total mdr calculated in the last section and transform this to the bl coordinates and multiply by the `` effective bl spatial 3-volume '' of our observer , we can obtain an approximation to a 4-momentum per proper time of electrons created at that location .
this is a coordinate - dependent quantity which should be adequate for a qualitative understanding as long as the observers are not too close the the horizon of a rotating black hole .
moreover , electron - positron production close to the horizon should not be expected to contribute significantly to the results at large radii , as the electron - positron production for observers close to the horizon will be dominated by in - scatter of the electrons and positrons into the black hole . given
coordinate dependent mdr quantities , we then parallel transport each of these 4-dimensional momenergy vectors parallel to themselves out to the inner shell of the collapsar ( say at @xmath112 ) .
we then determine the time dilation and bin the results as a function of the poloidal angle . in this way we can estimate the mdr from neutrino scattering as a function of the angular momentum of the black hole .
this treatment is obviously non - covariant .
however , as the astrophysical uncertainties of the environs of such a black hole / accretion disk system appear to be , at present , far greater than the issues with relativity we raise here ; we feel that our violation of general covariance to simplify the computational requirements may be justified in order to provide at least a qualitative understanding of the off - axis energy - deposition at the inner envelope .
we acknowledge that for a more complete characterization of the source term , a covariant treatment should ultimately be undertaken . in the meantime
, we feel that the results presented herein provide a useful qualitative estimate .
we have provided a pictorial representation of the total mdr calculated for a field of observers in the @xmath116-@xmath115 plane perpendicular to the accretion disk in fig [ pictures ] .
here we have placed @xmath118 observers in this plane and calculated the mdr for each using eq .
( [ mdr2 ] ) .
the observers were evenly spaced in 100 equal bl increments ( @xmath119 ) in radius from @xmath120 out to @xmath121 , and equally spaced in 100 bl poloidal increments ( @xmath122 ) from @xmath123 to @xmath124 .
the mdr densities were transformed to bl coordinates .
the 3-volume assigned to each observer is , as mentioned above , coordinate dependent ( @xmath125 ) the last column of the figure illustrates the structure of the mdr generated above the surface of the accretion disk . in fig .
[ angprofiles ] we integrated the bl mdr 4-momentum out along their geodesics until they either hit the black hole or they reach @xmath112 .
we do this for five black hole / disk systems .
the five systems correspond to specific angular momentums with the values , @xmath126 . for each value of @xmath2
the bl time component of the mdr in the bl frame reaching @xmath112 was binned into 20 bins ranging from @xmath123 to @xmath124 .
the relative magnitude of these curves increase with increasing values of the specific angular momentum of the central black hole .
we have investigated the origin of this increase and find that the source of the increase is the region of the accretion disk near the inner - most stable orbit .
this is illustrated in ( fig .
[ erd_on_a ] ) where we compare the angular mdr profile for an accretion disk extending from @xmath127 out to @xmath128 with ( 1 ) a schwarzschild black hole and ( 2 ) and extreme kerr black hole .
the magnitudes are approximately equal and only a slight tightening of the angle of the peak mdr is apparent for the extreme kerr black hole system . we demonstrate convergence of our results with regard to the number of observers we use in the @xmath116-@xmath115 plane by comparison of a high and lower resolution run ( fig .
[ highlow ] ) . the higher resolution , which we use in our calculations in this manuscript , is a little smoother than the lower resolution run ; however , the both compare well with each other . the relevant quantity reported by af was the total mdr at a distant observer . if we integrate each of the five mdr curves shown in fig .
[ angprofiles ] over @xmath129 at @xmath112 we will obtain the integrated mdr into the jet as shown in fig . [ totaledr(a ) ] .
it will be concentrated along the axis and peak along a cone of opening angle @xmath1 .
this opening angle is due to the tilt of the mdr near the surface of the disk .
the tilt is in the direction of the disk s rotation .
we find a linear dependence for this integrated mdr as a function of the specific angular momentum of the black hole .
this is an extension of the earlier results because we are including the relatively large off - axis source term the mdr created just above the surface of the disk via @xmath4 @xmath130 scattering .
this disk mdr produces a 10 to 20 fold enhancement to the total mdr energy deposited at @xmath112 over and above the on - axis mdr , even for a schwarzschild black hole .
such a disk - generated mdr was also found in the non - gravitational models of af.@xcite the dependence of this disk - driven source of momenergy on the specific angular momentum of the black hole is roughly linear .
we predict a twofold increase in the mdr for an extreme kerr black hole relative to a schwarzschild hole .
in addition , in ( fig .
[ angprofiles ] ) we only see a slight tightening of the cone of the momenergy deposition with increasing @xmath2 .
it would be interesting to observe the jet dynamics through the outer envelope of the star with sources of this kind .
the goal of the research in the grb central engine modeling community over the last few years was to look for a needed enhancement of a factor of 10 in the mdr .
previous work by af and sw looked for such an enhancement over the non - gravitating models by examining the gr effects on neutrino pair scattering along the symmetry axis .
they did not find this enhancement ; however , in this paper we identify the off - axis disk - driven mdr component that yields such an increase over corresponding on - axis mdr values .
furthermore , this disk - driven momenergy occurs in the non - rotating black holes as well as the rotating black holes , and is even present in the non - physical models with no gravity effects ( e.g. fig 4 of ref .
we therefore have strong indication that the the current fireball central engine model discussed here is viable .
of course , more detailed astrophysical simulations are needed which include the complex environs of the black hole disk system , geometry and transient behavior of the disk and black hole , and a simulation of the induced plasma transport of the scattering out to the fireball region . since this disk - driven mdr source is relatively large , and
since the baryon pollution problem@xcite remains an open issue in this field,@xcite we argue that it may be premature to discount this off - axis source .
we should reserve such a decision until more detailed gr , hydrodynamic , mhd and plasma transport calculations have been done . as we outlined in the paper , we have extended the earlier work of af to calculate the off - axis contribution to the mdr . when using the assumptions made by these authors , our results are consistent their work ; however , we find the generation of a substantial mdr above the disk , which , when added to the on - axis calculations of these authors produce a substantial enhancement to the deposition of energy - momentum .
in particular , we find an approximately linear dependence of the total mdr at large @xmath131 as a function of specific angular momentum of the black hole .
@xmath132 here we did a least - squares fit to the numerical results in fig .
[ totaledr(a ) ] , multiplied by @xmath91 times the cube of the solar mass ( @xmath133 ) .
the bulk of the mdr at large @xmath131 is deposited in a conic region centered on the symmetry axis of the system with an opening angle of approximately @xmath1 .
although we provide a rigorous general relativistic calculation of the mdr at an arbitrary observer , computational constraints required us to make some approximations regarding the transport of the scattered energy out to the inner envelope ( @xmath112 ) of the star . assuming ( 1 ) the electrons produced have substantially higher energy than their rest mass , ( 2 ) that the electrons propagate away in a collisionless fashion and ( 3 ) that the observationally - relevant electron creation distribution functions are peaked around the total mdr calculated , then our results should be rigorously correct for a schwarzschild black hole , and qualitatively correct for spinning black holes .
we would not be surprised if a general - relativistically rigorous calculations would yield different quantitative results for the extreme kerr black hole close to the horizon . however , one of the main results of this paper is the identification of the off - axis disk - driven mdr as providing an enhancement to the energy deposition rate rather than the gr - driven enhancements .
we outlined in the last section an approach to calculate the electron distribution function at an observer @xmath134 at large @xmath131 in a relativistically covariant way .
this involved integrating the electron - creation distribution functions along geodesics into the past which originated at @xmath134 .
one step in this direction would be to extend the work here and calculate the electron creation distributions in the frame of our observers as opposed to the total mdr .
this is not difficult in principle but is computationally demanding as it would require evaluating a 6-dimensional scattering integral as opposed to the 4-dimensional integral we evaluated numerically in sec .
[ sec4 ] , eqn .
the utility of such a calculation would be to evaluate the point along the axis on the border between out - scatter and in - scatter ( i.e. when the total mdr was zero ) .
would the electron creation function at and about this event yield a non - negligible contribution to the electron distribution function at large @xmath131 ? on a more speculative note ,
we also are interested in calculating secondary scattering of the electrons and positrons within the ergosphere of the black hole .
we have found evidence of significant in - scatter into this region .
will the penrose process produce some observable high energy events from this region that can give us more clues to the inner workings of the central engine of these low mass black hole / accretion disk systems ?
one question seems of paramount importance in this field : _ what observational features of a grb can give clues as to the inner - workings of the central engine?_@xcite surely future x - ray , @xmath135-ray and gravity wave observatories complemented with theory can soon unveil these cosmic explosions .
we wish to thank doug eardley , adrian gentle , bob geroch , daniel holz and rosalba perna for stimulating conversations on this subject , and chris fryer for suggesting this problem .
we thank the institute for theoretical physics at uc santa barbara for providing a supportive and stimulating research environment to complete this research .
this project was supported by los alamos under the ldrd / er program and in part by the national science foundation under grant no .
phy99 - 07949 .
s. e. woosley , gamma - ray bursts : the central engine , in * gamma - ray bursts in the afterglow era * , _ proceedings of the international workshop held in rome , cnr headquarters , 17 - 20 october , 2000_. eds .
e. costa , f. frontera , and j. hjorth ( berlin heidelberg : springer , 2001 , p. 257 ) . | the search for an understanding of an energy source great enough to explain the gamma - ray burst ( grb ) phenomena has attracted much attention from the astrophysical community since its discovery . in this paper
we extend the work of k. asano and t. fukuyama , and j. d. salmonson and j. r. wilson , and analyze the off - axis contributions to the energy - momentum deposition rate ( mdr ) from the @xmath0 collisions above a rotating black hole / thin accretion disk system .
our calculations are performed by imaging the accretion disk at a specified observer using the full geodesic equations , and calculating the cumulative mdr from the scattering of all pairs of neutrinos and anti - neutrinos arriving at the observer .
our results shed light on the beaming efficiency of grb models of this kind .
although we confirm asano and fukuyama s conjecture as to the constancy of the beaming for small angles away from the axis ; nevertheless , we find the dominant contribution to the mdr comes from near the surface of the disk with a tilt of approximately @xmath1 in the direction of the disk s rotation .
we find that the mdr at large radii is directed outward in a conic section centered around the symmetry axis and is larger , by a factor of 10 to 20 , than the on - axis values . by including this off - axis disk source , we find a linear dependence of the mdr on the black hole angular momentum ( @xmath2 ) .
in addition , we find that scattering is directed back onto the black hole in regions just above the horizon of the black hole . this gravitational `` in scatter '' may provide an observable high energy signature of the central engine , or at least another channel for accretion . |
[ lemma_relating_the_three ] let @xmath86 be a real @xmath87 matrix of orthogonal projection ( i.e. let @xmath86 be idempotent and symmetric : @xmath88 and @xmath89 , where @xmath90 denotes transposition ) .
then for any two @xmath91-dimensional real vectors @xmath92 and @xmath93 , @xmath94 + 2 |\!\cos[\theta_{\vec{\imath},\,\vec{a}}]|\frac{\|\hat{q}\vec{\imath}\|}{\|\vec{\imath}\| } \right\ } { \ ; } , \label{upper_bound_abstract } \ ] ] where @xmath95 is the angle between two vectors @xmath96 and @xmath97 , i.e.@xmath98 $ ] and @xmath99 , and @xmath100 is the norm of a vector @xmath101 .
let @xmath102 be the orthogonal projection complementary to @xmath86 . similarly to @xmath86 , it is idempotent and symmetric .
any vector @xmath101 can be decomposed onto the sum @xmath103 , with @xmath104 . as a consequence , @xmath105 .
the proof goes as follows : @xmath106 + 2 |\!\cos[\theta_{\vec{\imath},\,\vec{a}}]|\frac{\|\hat{q}\vec{\imath}\|}{\|\vec{\imath}\| } \right\ } { \;}. \nonumber \ ] ] [ lemma_on_two_reals ] for any two real numbers @xmath110 and @xmath111 , @xmath112 \ge \\ \|\vec{a}_{\perp}\|^2 \sin^2[\theta_{\vec{\imath}_{\perp},\ , \vec{a}_{\perp } } ] { \ ; } , \label{lemma_2 } \ ] ] where @xmath113 , @xmath114 are real @xmath91-dimensional vectors , @xmath115 is a real unit @xmath91-dimensional vector , @xmath116 , and @xmath117 . since neither the left- nor the right - hand - side of the inequality ( [ lemma_2 ] ) depends on the norm of @xmath93 , we may assume , without loss of generality , that @xmath93 is a unit norm vector : @xmath118 , where @xmath119 .
it can be further decomposed as @xmath120 , where @xmath121 .
the proof is as follows : @xmath122 \stackrel{\vec{\imath } = \vec{e}_{\vec{\imath}}}{= } \nonumber \\ & \stackrel{\vec{\imath } = \vec{e}_{\vec{\imath}}}{= } & \|\vec{a}_{\perp } + \alpha\vec{u}\|^2 \sin^2[\theta_{\vec{e}_{\vec{\imath}},\ , \vec{a}_{\perp } + \alpha\vec{u } } ] \nonumber \\ & = & \|\vec{a}_{\perp } + \alpha\vec{u}\|^2 - ( \vec{e}_{\vec{\imath } } \cdot ( \vec{a}_{\perp } + \alpha\vec{u } ) ) \nonumber \\ & = & \|\vec{a}_{\perp}\|^2 - ( \vec{e}_{\vec{\imath } } \cdot \vec{a}_{\perp})^2 + ( 1-(\vec{e}_{\vec{\imath } } \cdot \vec{u})^2 ) \alpha^2 \nonumber \\ & & \mbox{}\hspace{10em}\mbox { } - 2 ( \vec{e}_{\vec{\imath } } \cdot \vec{a}_{\perp } ) ( \vec{e}_{\vec{\imath } } \cdot \vec{u } ) \alpha \nonumber \\ & \ge & \|\vec{a}_{\perp}\|^2 - ( \vec{e}_{\vec{\imath } } \cdot \vec{a}_{\perp})^2 \nonumber \\ & & \mbox{}\hspace{0em}\mbox { } + \min_{\alpha}\large [ \underbrace{(1-(\vec{e}_{\vec{\imath } } \cdot \vec{u})^2)}_{\ge 0 } \alpha^2 - 2 ( \vec{e}_{\vec{\imath } } \cdot \vec{a}_{\perp } ) ( \vec{e}_{\vec{\imath } } \cdot \vec{u } ) \alpha \large ] \nonumber \\ & = & \|\vec{a}_{\perp}\|^2 - ( \vec{e}_{\vec{\imath } } \cdot \vec{a}_{\perp})^2 - \frac{(\vec{e}_{\vec{\imath } } \cdot \vec{a}_{\perp})^2 ( \vec{e}_{\vec{\imath } } \cdot \vec{u})^2}{1-(\vec{e}_{\vec{\imath } } \cdot \vec{u})^2 } \nonumber \\ & = & \|\vec{a}_{\perp}\|^2 - \frac{(\vec{e}_{\vec{\imath } } \cdot \vec{a}_{\perp})^2}{1-(\vec{e}_{\vec{\imath } } \cdot \vec{u})^2 } \nonumber \\ & = & \|\vec{a}_{\perp}\|^2 - \frac{\cos^2[\eta]((\vec{e}_{\vec{\imath}})_{\perp } \cdot \vec{a}_{\perp})^2}{1-\sin^2[\eta ] } \nonumber \\ & & \;\ ; \mbox{(using $ \vec{e}_{\vec{\imath } } = \cos(\eta ) ( \vec{e}_{\vec{\imath}})_{\perp } + \sin(\eta ) \vec{u}$ ) } \nonumber \\ & = & \|\vec{a}_{\perp}\|^2 - ( ( \vec{e}_{\vec{\imath}})_{\perp } \cdot \vec{a}_{\perp})^2 \nonumber \\ & = & \|\vec{a}_{\perp}\|^2 \sin^2[\theta_{\vec{\imath}_{\perp},\ , \vec{a}_{\perp } } ] { \;}. \;\ ; \mbox{(using $ ( \vec{e}_{\vec{\imath}})_{\perp } = \vec{\imath}_{\perp}$ ) } \nonumber \ ] ] [ [ accuracy - of - a - generalized - gibbs - ensemble .- a - single - additional - integral - of - motion - and - a - single - observable - of - interest . ] ] 2 .
accuracy of a generalized gibbs ensemble .
a single additional integral of motion and a single observable of interest .
imagine that we are interested in predicting the infinite time average of the quantum - mechanical mean of an observable @xmath10 , given the initial state .
according to the standard microcanonical scenario , the energy scale is divided onto narrow intervals , @xmath123 $ ] .
the only information about the initial state we are given is which energy interval , @xmath124 , the quantum expectation value of the energy belongs to .
the microanonical prediction for the time average of the observable in the subsequent evolution is @xmath125 = \operatorname{mean}_{{\textrm{\upshape mc}}}[a ] \nonumber \\ & & \qquad\qquad\qquad\qquad\qquad\qquad \equiv \frac { \sum_{\alpha \in d_{i^{\star}}}\ , { \langle}\alpha\vert \hat{a } \vert\alpha{\rangle } } { \sum_{\alpha \in d_{i^{\star}}}\ , 1 } { \ ; } , \nonumber \ ] ] where @xmath126 is the interval of the eigenstate indices @xmath127 populated by the eigenstates whose energy belongs to the interval @xmath123 $ ] : @xmath128 { \;}. \label{defining_d_i } \ ] ] now , assume that in the initial state , the system s energy is measured exactly , yielding a value @xmath129 , but the only information we get is , again , the interval @xmath130 this energy belongs to .
assuming that within the interval @xmath130 each eigenstate can appear with equal probability , the mean square error of the microcanonical prediction for the quantum - mechanical mean of @xmath10 is @xmath131 = \frac { \sum_{\alpha \in d_{i^{\star } } } ( { \langle}\alpha\vert \hat{a } \vert\alpha{\rangle}- \operatorname{mean}_{{\textrm{\upshape mc}}}[a])^2 } { \sum_{\alpha \in d_{i^{\star}}}\ , 1 } { \;}. \nonumber \ ] ] the vanishing of this variance in the thermodynamic limit is the essence of the eigenstate thermalization hypothesis @xcite .
imagine now that we are given an extra piece of information : via a second measurement in the initial state , we are allowed to place another integral of motion @xmath132an observable with all off - diagonal matrix elements equal to zero to an interval @xmath133 , one of a set of intervals @xmath134 $ ] .
an ensemble of states with @xmath135 $ ] and @xmath136 $ ] , distributed with equal probability , forms a minimal version of the generalized gibbs ensemble @xcite .
its prediction for the quantum mean of the observable @xmath10 reads : @xmath125 = \operatorname{mean}_{{\textrm{\upshape gge}}\,\vert\,j^{\star}}[a ] \nonumber \\ & & \qquad\qquad\qquad\qquad\qquad\qquad \equiv \frac { \sum_{\alpha \in d_{i^{\star } } \cap s_{j^{\star } } } \ , { \langle}\alpha\vert \hat{a } \vert\alpha{\rangle } } { \sum_{\alpha \in d_{i^{\star } } \cap s_{j^{\star } } } \ , 1 } { \ ; } , \nonumber \ ] ] where @xmath137 { \;}. \nonumber \ ] ] here and below , @xmath138 . in predicting the infinite time average of an observable @xmath10 , the mean square error ( mse ) of the prediction of generalized gibbs ensemble based on an integral of motion @xmath132 , @xmath139 ] \equiv
\operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \nonumber \\ & & \qquad = \frac { \sum_{j } \sum_{\alpha \in d_{i^{\star } } \cap s_{j } } ( { \langle}\alpha\vert \hat{a } \vert\alpha{\rangle}- \operatorname{mean}_{{\textrm{\upshape gge}}\,\vert\,j}[a])^2 } { \sum_{\alpha \in d_{i^{\star}}}\ , 1 } { \ ; } , \nonumber \ ] ] ( where it is assumed that the initial states are still uniformly distributed inside @xmath140 $ ] ) is bounded from the above as @xmath141 ]
\equiv \operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \ ; \leq \\
\operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \ , \left(\sin^2[\theta_{\hat{i}_{{\textrm{tl}}},\ , \hat{\hat{p}}_{h } \hat{a}_{{\textrm{tl } } } } ] + |\cos[\theta_{\hat{i}_{{\textrm{tl}}},\ , \hat{\hat{p}}_{h } \hat{a}_{{\textrm{tl}}}}]| \ , \underbrace { { \cal o}\left [ \frac{\delta i}{\sqrt{\operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{i } \vert\alpha{\rangle } ] } } \right ] } _ { \ll 1 } \right ) \label{upper_bound } { \ ; } , \ ] ] where @xmath142 is the hilbert - schmidt angle between the observables @xmath143 and @xmath144 , a traceless version of a given observable , @xmath62 , is defined as @xmath145/n_{{\textrm{\upshape mc}}}$ ] , @xmath72 is a `` super - operator '' that removes the off - diagonal ( with respect to the basis of the eigenstates of @xmath67 ) matrix elements , and @xmath73 is the maximal width of a microcanonical window for the additional integral of motion .
the estimate in eq . ( [ upper_bound ] ) shows that the biggest increase in the predictive power of the generalized gibbs ensemble is delivered by the integrals of motion at a small hilbert - schmidt angle to the observable of interest . to the dominant order in the size of the microcanonical boxes for the integral of motion @xmath146 ,
the bound does not depend on the details of the partition of the axis of the integral of motion @xmath75 .
remark that if the bound ( [ upper_bound ] ) is used to assess the accuracy in a quench from a linear superposition of the eigenstates , the variance of the integral of motion in the initial state must be smaller than @xmath146 . the dictionary used to relate lemma [ lemma_relating_the_three ] to the bound ( [ upper_bound ] ) is as follows : @xmath147 \ , \vert \ , \alpha \in d_{i^{\star } } \ }
\nonumber \\ \hat{q}\vec{\imath } & \to & \{{\langle}\alpha\vert \hat{i } \vert\alpha{\rangle}- \operatorname{mean}_{{\textrm{\upshape gge}}\,\vert\,j(\alpha)}[i ] \ , \vert \ , \alpha \in d_{i^{\star } } \ } \nonumber \\ \|\vec{a}\|^2 & \to & \operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] + \operatorname{mean}_{{\textrm{\upshape mc}}}[a]^2 \nonumber
\\ \|\vec{\imath}\|^2 & \to & \operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{i } \vert\alpha{\rangle } ] + \operatorname{mean}_{{\textrm{\upshape mc}}}[i]^2 \nonumber \\ \|\hat{q}\vec{a}\|^2 & \to & \operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \nonumber \\ \|\hat{q}\vec{\imath}\|^2 & \to & \operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{i } \vert\alpha{\rangle } ] \nonumber { \;}. \ ] ] lemma [ lemma_on_two_reals ] shows that the first term of the bound in eq .
( [ upper_bound_abstract ] ) ( which is the dominant term in our context ) can be further improved by removing a component along a particular unit vector @xmath115 . in our case
, the role of the vector @xmath115 will be played by a hilbert - schmidt - normalized identity operator .
the following dictionary completes the proof of eq .
( [ upper_bound ] ) : @xmath148 \nonumber \\ \|\vec{\imath}_{\perp}\|^2 & \to & \operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{i } \vert\alpha{\rangle } ] \nonumber \\ \|\hat{q}\vec{a}_{\perp}\|^2 & \to & \operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \nonumber \\ \|\hat{q}\vec{\imath}_{\perp}\|^2 & \to & \operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{i } \vert\alpha{\rangle } ] \nonumber { \;}.\end{aligned}\ ] ] [ [ several - additional - integrals - of - motion - and - a - single - observable - of - interest . ] ] 3 .
several additional integrals of motion and a single observable of interest .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ one can further introduce another integral of motion , @xmath149 , orthogonal to the first one : @xmath150 . introducing a fictitious observable with matrix elements @xmath151
$ ] , the bound ( [ upper_bound ] ) can be generalized to @xmath152 ] \equiv \operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \lesssim & & \nonumber \\ & & \mbox{}\hspace{-21em } \operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \times \nonumber \\ & & \mbox{}\hspace{-18em } \left ( 1 - \cos^2[\theta_{\hat{i}_{{\textrm{tl}}},\ , \hat{\hat{p}}_{h } \hat{a}_{{\textrm{tl } } } } ] - \cos^2[\theta_{(\hat{i}_{2})_{{\textrm{tl}}},\ , \hat{\hat{p}}_{h } \hat{a}_{{\textrm{tl } } } } ] \right ) { \;}. \nonumber \ ] ] furthermore , by induction , one can extend this bound to any number @xmath153 of ( mutually orthogonal ) integrals of motion : @xmath152 ] \equiv \operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \lesssim & & \nonumber \\ & & \mbox{}\hspace{-20em}\mbox { } \operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \left ( 1 - \sum_{m=1}^{n_{i}}\cos^2[\theta_{(\hat{i}_{m})_{{\textrm{tl}}},\ , \hat{\hat{p}}_{h } \hat{a}_{{\textrm{tl } } } } ] \right ) { \;}. \nonumber \ ] ] for future applications , we will quote two other expressions for this bound : @xmath152 ] \equiv \operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \lesssim & & \label{bound_100 } \\ & & \mbox{}\hspace{-20em}\mbox { } \operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] - \operatorname{var}_{{\textrm{\upshape mc}}}[a ] \sum_{m=1}^{n_{i}}\cos^2[\theta_{(\hat{i}_{m})_{{\textrm{tl}}},\ , \hat{a}_{{\textrm{tl } } } } ] { \ ; } , \nonumber \ ] ] where @xmath154 \equiv \operatorname{mean}_{{\textrm{\upshape mc}}}[a^2]-\operatorname{mean}_{{\textrm{\upshape mc}}}[a]^2 $ ] ; and @xmath152 ] \equiv \operatorname{var}_{{\textrm{\upshape gge}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \lesssim & & \label{bound_110 } \\ & & \mbox{}\hspace{-21em}\mbox { } \operatorname{var}_{{\textrm{\upshape mc}}}[a ] \cos^2[\theta_{{\cal l}_{{\textrm{tl,\,d-}}\hat{h}}/\operatorname{span}[\{(\hat{i}_{m})_{{\textrm{tl } } } \;|\ ; m=1,\,\dots,\,n_{i } \}],\ , \hat{a}_{{\textrm{tl } } } } ] { \ ; } , \nonumber \ ] ] where @xmath155 is the orthogonal complement of a subspace @xmath156 of a vector space @xmath63 with respect to an inner product @xmath157 ( the frobenius or hilbert - schmidt product @xmath158 in our case ) .
[ [ several - additional - integrals - of - motion - and - several - observables - of - interest . ] ] 4 .
several additional integrals of motion and several observables of interest .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ now , assume that we have several observables of interest ( spanning a linear space @xmath159 ) whose long - term behavior we want to be able to predict .
assume further that we are given several integrals of motion ( spanning a space @xmath160 ) we are allowed to use as thermodynamical predictors .
next , we are going to form linear spaces @xmath161 and @xmath160 spanned by the traceless versions of the same observables .
now , let us introduce orthonormal bases for both spaces : @xmath162 in particular , this implies that @xmath163 = \operatorname{var}_{{\textrm{\upshape mc}}}[i_{m } ] = 1 $ ] for all @xmath164 and @xmath16 .
let us now form the following matrix : @xmath165 \cos[\theta_{(\hat{a}_{q})_{{\textrm{tl}}},\ , ( \hat{i}_{m'})_{{\textrm{tl } } } } ] \nonumber \\ & & m,\,m'=1,\,2,\,\ldots,\,n_{i } { \ ; } , \nonumber \ ] ] cf .
[ bound_100 ] .
it can be shown that the eigenvectors of @xmath166 , @xmath167 , when ordered in the descending order of their eigenvalues @xmath168 form a sequence such that the first @xmath169 members constitute the most optimal generalized gibbs ensemble involving @xmath169 integrals , in the sense that it minimizes , on average , the error in the prediction of the ensemble .
the requests for linear combinations of the observables of interest are supposed to be distributed according to a spherically symmetric probability distribution , with respect to the hilbert - schmidt measure .
we show examples of a relevance sequence in figs . 3a and b. in the examples considered in fig . 3 , the space @xmath160 was represented by the whole space of traceless integrals of motion @xmath47 .
the number of the most optimal ( relevant ) integrals of motion was 16 for both fig .
3c and fig .
the classical mechanics example considered in the letter consists of a two - dimensional particle of mass @xmath16 , moving in a rectangular billiard with periodic boundary conditions ( thus topologically equivalent to a torus ) . as an integrability - breaking perturbation
, we use a `` truncated '' @xmath170-function : a potential that consists of the first @xmath171 ( including the zeroth one ) spatial harmonics of a @xmath170-function ( see fig .
[ olshanii_figsd1 ] ) .
* the integrability - breaking barrier for the classical system considered:*. the plot shows the `` truncated '' @xmath170-function potential used as an integrability - breaking perturbation acting on a particle in a two - dimensional square billiard . ]
the resulting hamiltonian is @xmath172}{(2m+1)\sin[\pi x / l ] } \frac{\sin[2\pi ( m+1/2)y / l]}{(2m+1)\sin[\pi
y / l ] } { \;},\end{gathered}\ ] ] where @xmath173 in what follows , we will be using a system of units where @xmath174 .
the hamiltonian then becomes @xmath175}{(2m+1)\sin[\theta_{x}/2 ] } \frac{\sin[(m+1/2 ) \theta_{y}]}{(2m+1)\sin[\theta_{y}/2 ] } { \ ; } , \ ] ] where , in this system of units , we have @xmath176 and @xmath177 ( @xmath178 ) ; also , @xmath179 ; @xmath180 .
we used @xmath181 for all data points . for any strength of the perturbation @xmath182 , the initial conditions were drawn from a microcanonical ensemble bounded by the equi - energy surfaces ( specific for a given @xmath182 ) in such a way that the phase - space volume below the lower surface was @xmath7 , and in between the lower and the upper was @xmath6 , regardless of the perturbation strength .
a quantum - mechanical analogue of such an ensemble would have the same lower and upper quantum indices of the microcanonical window for all realizations . to give the reader an idea of the energy scale involved
, we will note that for a zero barrier height , the mean energy in the ensemble is @xmath20 . for the test observable @xmath5 , we used the difference between the kinetic energies in the @xmath0- and @xmath109-directions : @xmath183 ( which is @xmath184 in our system of units ) .
we studied both the ensemble variance of the temporal mean , @xmath185 \equiv \frac { \int_{{\textrm{\upshape mc } } } d\theta_{x}^{t=0 } d\theta_{y}^{t=0 } di_{x}^{t=0 } di_{y}^{t=0 } \left\ { \left ( \lim_{t_{\textrm{max}\to\infty}}\frac{1}{t_{\textrm { max}}}\int_{t=0}^{t_{\textrm{max } } } \,dt\,a(t ) \right ) - \operatorname{mean}_{{\textrm{\upshape mc}}}[a ] \right\}^2 } { \int_{{\textrm{\upshape mc } } } d\theta_{x } d\theta_{y } di_{x } di_{y}\ ; 1 } { \ ; } , \ ] ] @xmath186 \equiv \frac { \int_{{\textrm{\upshape mc } } } d\theta_{x}^{t=0 } d\theta_{y}^{t=0 } di_{x}^{t=0 } di_{y}^{t=0 } \lim_{t_{\textrm { max}\to\infty}}\frac{1}{t_{\textrm { max } } } \int_{t=0}^{t_{\textrm{max}}}\,dt\ , \left\ { a(t ) - \left ( \lim_{t_{\textrm { max}\to\infty}}\frac{1}{t_{\textrm { max}}}\int_{t'=0}^{t_{\textrm{max}\to\infty}}\,dt'\ , a(t ' ) \right ) \right\}^2 } { \int_{{\textrm{\upshape mc } } } d\theta_{x } d\theta_{y } di_{x } di_{y}\ ; 1 } { \ ; } , \nonumber \ ] ] where the ensemble average is @xmath187 \equiv \frac { \int_{{\textrm{\upshape mc } } } d\theta_{x } d\theta_{y } di_{x } di_{y}\ , a } { \int_{{\textrm{\upshape mc } } } d\theta_{x } d\theta_{y } di_{x } di_{y}\ , 1 } { \ ; } , \nonumber \ ] ] @xmath188 is an integral over the microcanonical volume describe above , and @xmath189 the time dependence for the observable @xmath5 along a trajectory that starts at @xmath190 .
[ [ the - hamilitonian - for - one - dimensional - hard - core - bosons - perturbed - by - soft - core - interactions . ] ] 6 .
the hamilitonian for one - dimensional hard - core bosons perturbed by soft - core interactions .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the quantum hamiltonian used in this letter reads @xmath191 where the commutation relations for the hard - core boson creation and annihilation operators obey @xmath192 the soft - core interaction potential we used was a rectangular potential of a four - site range : @xmath193 with @xmath194 ( see fig .
[ olshanii_figsd2 ] ) . * the lattice configuration for the quantum system considered*. one - dimensional hard - core bosons on a lattice with @xmath23 sites .
the number of atoms was @xmath195 for the system of fig .
1b and @xmath196 for the system of fig . 3 .
open boundary conditions were imposed on the former system and periodic ones on the latter .
the integrability - breaking perturbation was an added two - body interaction : a constant potential energy @xmath24 for any two atoms separated by four sites or less . ] for the calculations resulting in fig .
1b , the open boundary conditions were used : the first sum in eq .
( [ h_qm ] ) was extended to a range between @xmath197 and @xmath198 , where @xmath23 was the length of the lattice .
both sums in the second double sum were fixed between @xmath199 and @xmath200 . the calculations that led to fig .
3 were performed using periodic boundary conditions . there , the first sum involved in eq .
( [ h_qm ] ) covers all sites , from @xmath197 to @xmath201 . in the second sum
, the index @xmath202 covers the same range , from @xmath203 to @xmath204 ; the second index , @xmath205 , ranges from @xmath206 to @xmath207 , so that the soft - core potential respects the periodic boundary conditions .
also , in both cases , a weak random on - site perturbation was used , @xmath208 where @xmath209 is a set of independent random variables uniformly distributed between @xmath28 and @xmath29 , and the strength of the potential was fixed to @xmath210 .
the strenghth @xmath211 has been chosen in such a way that it remains weak enough not to alter any of the macroscopic properties , but strong enough to lift all possible degeneracies and , in the periodic case , to relax the selection rules associated with the translational invariance . in fig .
1b we analyze the properties of two integrals of motion , which are , in the case of a lattice with open boundary conditions , related to the counterparts of the fourth and sixth moments of the momentum distribution of the underlying free fermions .
when expressed through the bosonic creation and annihilation operators , the functionals of the fermionic momentum distribution quadratic in the fermionic representation become complicated many - body observables , such as @xmath212 the second term inside the sum ( the four - body one ) is absent in the fermionic representation .
the last term , present in both representations , is responsible for the finite size effects originating , in turn , from the open boundary conditions .
we also studied a `` generic '' integral of motion : @xmath213 , where the states @xmath214 are the eigenstates of the hamiltonian in eq .
( [ h_qm ] ) in the absence of the integrability - breaking perturbation , and the independent random coefficients @xmath215 were uniformly distributed between @xmath28 and @xmath29 . in fig . 3
, we are comparing three thermodynamic ensembles : the microcanonical ensemble ( which is based on the energy alone ) , @xmath216 { \ ; } , \nonumber \ ] ] the conventional generalized gibbs ensemble @xcite ( which is based on the occupation numbers of all free - fermionic one - body orbitals , or , in the case of periodic boundary conditions , all moments of the fermionic momentum distribution ) , @xmath217 { \ ; } , \nonumber \ ] ] and the optimized generalized gibbs ensemble ( which is designed to maximize the quality of prediction for one - body observables ) , @xmath218 { \;}. \nonumber \ ] ] only the first few most optimal integrals of motion generated by the latter space were used .
here @xmath40 are the free - fermionic annihilation operators , related to their bosonic counterparts via a jordan - wigner map , @xmath219 . here and below
, we are assuming that the number of sites @xmath17 is even .
@xmath221 the author thanks marcos rigol , david weiss , vanja dunjko , alessandro silva , and bala sundaram for comments .
this work was supported by the us national science foundation grant no .
phy-1019197 , the office of naval research grants no .
n00014 - 09 - 1 - 0502 and n00014 - 12 - 1 - 0400 , and a grant from the _ institut francilien de recherche sur les atomes froids _ ( ifraf ) .
support from the _ laboratoire de physique des lasers _ of paris 13 university is also appreciated .
@xmath221 supplementary information is available in the online version of the paper .
reprints and permissions information is available online at www.nature.com/reprints . correspondence and requests for materials should be addressed to m.o . | the concept of ergodicity the convergence of the temporal averages of observables to their ensemble averages is the cornerstone of thermodynamics .
the transition from a predictable , integrable behavior to ergodicity is one of the most difficult physical phenomena to treat ; the celebrated kam theorem is the prime example .
this letter is founded on the observation that for many classical and quantum observables , the sum of the ensemble _ variance _ of the temporal average and the ensemble average of temporal _ variance _ remains constant across the integrability - ergodicity transition .
we show that this property induces a particular geometry of quantum observables
frobenius ( also known as hilbert - schmidt ) one that naturally encodes all the phenomena associated with the emergence of ergodicity : the eigenstate thermalization effect @xcite , the decrease in the inverse participation ratio @xcite , and the disappearance of the integrals of motion . as an application
, we use this geometry to solve a known problem of optimization of the set of conserved quantities regardless of whether it comes from symmetries or from finite - size effects to be incorporated in an extended thermodynamical theory of integrable , near - integrable , or mesoscopic systems @xcite . consider a ball in a classical rectangular billiards with periodic boundary conditions , first without and then with a strong localized obstacle inside .
( when present , the obstacle makes the system ergodic . ) in both cases , consider two kinds of problems : 1 . given a specific initial velocity of the ball , what are the subsequent temporal fluctuations in ( say ) the @xmath0-component of the velocity ?
2 . draw a bunch of initial velocities from a thermal distribution ; for each , compute the infinite time average of the @xmath0-component of the velocity ; what is the variance among these infinite time averages ? if the obstacle is absent , both components of the velocity vector will remain equal to their initial values forever , and so the temporal fluctuations are zero ; as far as the second question , the variance of the infinite time averages is equal to that in the thermal distribution , and thus is large . on the other hand , if the obstacle is present , the situation is reversed ( this will be so provided the ball keeps hitting the obstacle , which will happen provided the ratio of the velocity components , @xmath1 , is incommensurate with the ratio of the billiard lengths , @xmath2 ; and this will hold with probability one as long as the initial velocities as sampled from some continuous probability distribution ) : now the temporal fluctuations in the velocity component are large no matter what the initial velocity ; at the same time , the infinite time average of the velocity component is the same no matter what the initial velocity ( namely , it is zero ) , and so the variance of the infinite time averages is zero .
moreover , as we will show , the shot - to - shot variance in the ( exactly integrable ) case of an empty billiards equals the temporal variance in the ergodic case .
given some imagination , one may suspect a `` conservation law '' acting across the transition from an integrable to an ergodic system as one increases the strength of the integrability - breaking perturbation . indeed , fig .
[ olshanii_fig1]a shows that in the case of square billiard perturbed by a soft localized potential barrier in the middle @xcite , the sum of the microcanonical variance of the time average , @xmath3 $ ] , and the shot - to - shot average of the temporal variance , @xmath4 $ ] , with the observable @xmath5 being the difference between the horizontal and vertical kinetic energies , remains the same for the heights of the barrier less than or comparable to the kinetic energy .
for all points , the statistical ensembles used were microcanonical ensembles with the same phase - space volume ( @xmath6 ) covered and with the same phase - space volume ( @xmath7 ) occupied by the phase - space points with energies below the lower - energy boundary of the microcanonical window .
such a set of microcanonical ensembles is the closest classical analogue of a quantum set that uses the same window of quantum state _
indices _ for all perturbation strengths used .
let us now try to translate the conjecture expressed in the first paragraph of this letter to quantum language . naively , one may try to replace the ensemble of classical initial conditions by an ensemble of random superpositions of quantum eigenstates .
however , consider the limiting case of an integrable system .
there , a single given initial state will already cover a variety of generally unrelated sets of integrals of motion .
to the contrary , in the classical case , a given initial point corresponds to a single set .
accordingly , we suggest an ensemble of randomly chosen eigenstates of the hamiltonian as the ensemble of initial conditions . in particular , the ensemble variance of the temporal means will be translated to the quantum language as @xmath8\big{|}_{\textrm{qm } } \equiv \operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \\
\qquad = ( n_{{\textrm{\upshape mc}}})^{-1 } \sum_{\alpha \in [ { \textrm{\upshape mc } } ] } ( { \langle}\alpha\vert \hat{a } \vert\alpha{\rangle}- { \langle}a { \rangle})^2 { \ ; } , \label{var_mc_mean_t } \ ] ] where @xmath9 \equiv ( n_{{\textrm{\upshape mc}}})^{-1 } \sum_{\alpha \in { \textrm{\upshape mc } } } { \langle}\alpha\vert \hat{a } \vert\alpha{\rangle}$ ] is the ensemble mean of the observable @xmath10 , @xmath11 } \ldots \equiv \sum_{\alpha \in [ \alpha_{min},\,\alpha_{max } ] } \ldots$ ] is a sum over a microcanonical window bounded by some energies @xmath12 and @xmath13 , and @xmath14 is the energy spectrum of the system .
the vanishing of the fluctuations ( [ var_mc_mean_t ] ) in the thermodynamic limit the so - called eigenstate thermalization effect is sufficient @xcite for the emergence of ergodicity .
our letter , in part , aims to devise a scale for these fluctuations that determines if a given observable is closer , by behavior , to an integral of motion or to a thermalizable observable .
the question of a proper quantum analogue of the temporal fluctuations is both more involved and better studied .
a complication arises from the fact that a quantum state is altered after each measurement .
however , consider the following procedure : for every realization of the initial state of interest , the observable is measured only once , and then the initial state is prepared again . for every instant of time of interest , the observable is measured several times , and then , another instant of time is addressed .
it has been argued@xcite that this * two variances*. the units are such that @xmath15 , where @xmath16 is the particle mass , and @xmath17 is the side of the billiard . *
a * , ensemble mean of the temporal variance vs. the ensemble variance of the temporal mean of the difference between the @xmath18 and @xmath19kinetic energies for rectangular billiard with periodic boundary conditions perturbed by a soft - core barrier . at zero barrier hight ,
the mean energy over the ensemble is @xmath20 .
the points plotted are labeled by the ratio , @xmath21 , of the barrier hight to the mean energy of the ensemble ( that includes the energy of the barrier ) .
* b * , ensemble mean of the quantum variance vs. the ensemble variance of the quantum expectation values for three integrals of motion of a system of hard core bosons , where the integrals of motion are partially destroyed by adding a soft - core two - body repulsive potential . for this figure : the number of particles is @xmath22 , the number of lattice sites is @xmath23 , and we impose open boundary conditions .
the soft - core interaction potential has a constant hight @xmath24 at distances of four sites or less , and it is zero otherwise .
@xmath25 and @xmath26 ( see the text for expressions ) are the integrals of motion related to the forth and sixth moments of the momentum distribution of the underlying free fermions .
@xmath27 is an artificial integral of motion represented by a diagonal in the basis of the unperturbed eigenstates matrix with random entries , uniformly distributed between @xmath28 and @xmath29 .
for this observable , the square cosine of the angle between a point on the figure and the horizontal axis equals ( up to small corrections of the order of @xmath30 ) the inverse participation ratio @xmath31 ( see eq . ( [ eta ] ) ) .
observe also that the behavior of the two other observables is qualitatively simlar . ]
procedure is indeed the most suitable quantum counterpart of the classical temporal fluctuations along a trajectory .
furthermore , it has been suggested that in case of an ergodic motion , these fluctuations are nothing else but the _ thermal _ fluctuations in the system@xcite . according to this scenario , the quantum uncertainty in the results of the measurements is the way a quantum system emulates the classical instability with respect to the initial conditions : in both cases , the outcome of a single measurement is irreproducible , fundamentally so in the quantum case , and operationally so in the classical one .
accordingly , we define the quantum analogue of the ensemble mean of the temporal variance as @xmath32\big{|}_{\textrm{qm } } \\ \equiv \operatorname{mean}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{a}^2 \vert\alpha{\rangle}- { \langle}\alpha\vert \hat{a } \vert\alpha{\rangle}^2 ]
\\ = ( n_{{\textrm{\upshape mc}}})^{-1 } \sum_{\alpha \in [ { \textrm{\upshape mc } } ] } { \langle}\alpha\vert \hat{a}^2 \vert\alpha{\rangle}- { \langle}\alpha\vert \hat{a } \vert\alpha{\rangle}^2 { \;}. \label{mean_mc_var_t } \ ] ] consider now an integrable system , with hamiltonian @xmath33 perturbed by a non - integrable perturbation @xmath34 .
the full hamiltonian reads @xmath35 , where the parameter @xmath36 determines the degree by which the integrability is broken .
let the states @xmath37 be the eigenstates of the full hamiltonian .
the conjecture in the first paragraph of the letter then reads : @xmath8\big{|}_{\textrm{qm } } + \operatorname{mean}_{{\textrm{\upshape mc}}}[\operatorname{var}_{t}[a]]\big{|}_{\textrm{qm } } \\
= \text{a constant independent of $ g$ } { \;}. \label{conjecture } \ ] ] the data presented in fig .
[ olshanii_fig1]b test the conjecture ( [ conjecture ] ) using the example of a one - dimensional gas of lattice hard - core bosons perturbed by an added two - body soft - core repulsive interaction . in one dimension ,
both continuous - space and lattice hard - core bosons are known to be integrable : in both cases , there is a map ( girardeau s map@xcite , and the jordan - wigner transformation , respectively ) that connects the eigenstates of the system to the eigenstates of a free fermi gas .
the integrals of motion are thus represented by the occupation numbers of the eigenstates of the one body hamiltonian for the particles of the underlying free fermi gas .
we analyze the decay of the fourth and sixth integrals of motion , @xmath38 @xmath39 where @xmath40 is the @xmath41-th site annihilation free - fermionic operator ( see also the supplementary discussion ) , as we increase the strength of the non - integrable perturbation .
one can see that while the quantum ( the analogue of thermal ) fluctuations gradually increase , the deviations from the ergodicity decrease .
however , in accordance with the conjecture ( [ conjecture ] ) , the sum of the two variances remains approximately constant .
the square of the radius of the circles corresponds to the ensemble variance of the observable , over a series of _ single _ measurements with no subsequent quantum or temporal averaging on a randomly chosen eigenstate .
let us now reveal the intuition behind the conjecture ( [ conjecture ] ) .
the left - hand - side of the relationship ( [ conjecture ] ) can be written , by rearranging the terms in eqs .
[ var_mc_mean_t ] and [ mean_mc_var_t ] , as the _ ensemble variance _ of the observable @xmath10 : @xmath8\big{|}_{\textrm{qm } } + \operatorname{mean}_{{\textrm{\upshape mc}}}[\operatorname{var}_{t}[a]]\big{|}_{\textrm{qm } } \\ = \operatorname{mean}_{{\textrm{\upshape mc}}}[a^2]-\operatorname{mean}_{{\textrm{\upshape mc}}}[a]^2 { \;}. \label{ensemble_variance } \ ] ] in turn , the ensemble variance on the right - hand - side of the above equation is a function of two ensemble means ; and the key realization is that these ensemble means remain constant as the coupling constant @xmath36 of the integrability - breaking perturbation is increased from zero up to a value where the system first becomes ergodic .
indeed remark that typically , ergodicity emerges for an interaction strength much weaker than the one required to alter the thermal expectations of observables .
for example , even though the van der waals interactions between the molecules constituting air do not modify the maxwell distribution , the interactions are strong enough to lead to a maxwell distribution from any initial state .
mathematically , the general criterion for the independence of the ensemble means on @xmath36 may be stated as follows . for every @xmath36
, there is a characteristic energy interval @xmath42 such that the integrability - breaking perturbation appreciably couples only those eigenstates of the integrable hamiltonian whose energy difference is less than @xmath42 .
our criterion is that @xmath42 be much smaller than the energy width of the microcanonical window used to define the ensemble means .
the reason is as follows : consider the eigenstates of the unperturbed system .
the perturbation results in their mutual coupling ; but if the criterion is fulfilled , then we may neglect the coupling of states within the window to states outside the window .
but if we do that , then we may , in fact , truncate our hilbert space to just the states inside the window ( let us call the corresponding hamiltonian a ` truncated hamiltonian ' ) .
indeed , let us first truncate , and then turn on the perturbation . in that case ,
the eigenstates of the perturbed truncated hamiltonian are related to those of the unperturbed truncated hamiltonian by a unitary transformation . and now note that the two ensemble means are in fact traces , and thus do not change under unitary transformation of the basis ; thus they do not depend on @xmath36 .
observe now that in the right hand side of the relationship in eq .
( [ ensemble_variance ] ) , we find a quadratic polynomial , built out of the matrix elements of the observable @xmath10 , that is approximately invariant with respect to changes in the parameter @xmath36 ; in contrast , on the left - hand side is a sum of two quadratic polynomials that both vary with @xmath36 .
this realization inspires one to seek a geometric meaning of the relationship in eq .
( [ ensemble_variance ] ) , where the right hand side is the square norm of an unknown vector that is linearly related to the operator @xmath10 , and the left hand side is a decomposition of this square norm over some complementary subspaces ( that change with the perturbation strength @xmath36 ) of the linear space the vector belongs to . from what follows , we will see that this is almost what happens , with the exception of the @xmath43 ^ 2 $ ] term that belongs rather to the left - hand side .
* the hilbert - schmidt geometry and quantum integrability - ergodicity transition*. the @xmath44-dimensional space of quantum operators acting in an @xmath45-dimensional hilbert space is divided onto a sum of three subspaces : a one - dimensional space spanned by the identity operator , an @xmath46-dimensional space , @xmath47 , of traceless diagonal in the basis of the eigenstates of the hamiltonian in question real matrices , and an @xmath48-dimensional space of the purely off - diagonal in the above sense
hermitian matrices , @xmath49 .
if the coupling constant @xmath36 in the system hamiltonian @xmath35 changes , the @xmath47 and @xmath49 undergo a rotation , while the identity axis , the projection to which is proportional to the microcanonical average , remains the same .
however , the hilbert - schmidt vector corresponding to a given observable of interest remains fixed along with the identity axis ( and thus also along with the microcanonical average ) .
observables with a large projection onto @xmath47 correspond to quasi - conserved , non - ergodic quantities . in the opposite case ,
a large projection onto the @xmath49 space signifies a thermalizable observable , whose infinite time average coincides with its thermal expectation value .
ergodicity is reached when @xmath49 aligns with the traceless versions of all empirically relevant observables
e.g . with all one- , two- , and three - body obseravbles in a many - body system . ] the inner product that generates the anticipated geometric structure is the frobenius , or hilbert - schmidt , inner product .
the frobenius or hilbert - schmidt ( hs ) inner product between two matrices reads : @xmath50 { \;}. \label{hs } \ ] ] quantum observables , represented by hermitian matrices , form a linear space over the field of real numbers ; the product in eq .
( [ hs ] ) induces a real - valued inner product on this space .
observe also that this product is invariant under unitary transformations @xmath51 , where @xmath52 is a unitary matrix .
thus the unitary transformations form a subgroup of all possible linear transformations that preserve the product .
the hilbert space where we are going to deploy the hs structure is a microcanonical window of eigenstates , of size @xmath53 .
let us introduce an @xmath54-dimensional linear space , over the field of real numbers , of all hermitian operators ( generally represented by complex matrices ) acting within an @xmath53-dimensional microcanonical window .
the hs product in eq .
( [ hs ] ) constitutes an inner product on this space .
this space can be conveniently decomposed into a direct sum of three pairwise orthogonal subspaces .
the first one is a one - dimensional space , @xmath55 , spanned by the identity operator .
the second , @xmath47 , is the space of all traceless _ integrals of motion _ ; i.e. the space of traceless operators that are purely diagonal in the basis of the eigenstates @xmath56 .
its dimension is @xmath57 .
the third one , @xmath49 , of dimension @xmath58 , is the space of all purely off - diagonal operators in the same basis .
now , eq . ( [ ensemble_variance ] ) can be rewritten as @xmath59 = \operatorname{mean}_{{\textrm{\upshape mc}}}[a]^2 + \operatorname{var}_{{\textrm{\upshape mc}}}[\operatorname{mean}_{t}[a]]\big{|}_{\textrm{qm } } + \operatorname{mean}_{{\textrm{\upshape mc}}}[\operatorname{var}_{t}[a]]\big{|}_{\textrm{qm } } $ ] and interpreted as a relationship between the norm of the operator @xmath10 and its projections to the members of a complete set of orthogonal subspaces .
the hs angle between the traceless version of a given observable , @xmath60/n_{{\textrm{\upshape mc}}}$ ] , and the space of the traceless integrals of motion constitutes a useful measure that defines where the observable lies on the `` integral of motion '' vs.``thermalizable observable '' scale .
indeed , the tangent square of this angle is just the ratio between ( the quantum analogues of ) the temporal fluctuations and the time - average vs. ensemble average discrepancy : @xmath61 = \frac{\operatorname{mean}_{{\textrm{\upshape mc}}}[\operatorname{var}_{t}[a]]\big{|}_{\textrm { qm}}}{\operatorname{var}_{{\textrm{\upshape mc}}}[\operatorname{mean}_{t}[a]]\big{|}_{\textrm{qm } } } \label{integrability_angle } \ ] ] ( see fig .
[ olshanii_fig2 ] for an illustration ) . here and below
, the angle between an hs vector @xmath62 and an hs hyperplane @xmath63 is defined through @xmath64 , where @xmath65 is any orthonormalized basis set in @xmath63 .
if eigenstate thermalization @xcite holds , the angle ( [ integrability_angle ] ) approaches @xmath66 .
the inverse participation ratio @xmath31 , a measure closely related to the transition to thermal behavior @xcite , also acquires a clear geometric meaning .
it defines the average angle between the projector to an eigenstate of the integrable system @xmath33 and the space of the integrals of motion of the new system @xmath67 : @xmath68 } |{\langle}\alpha_{0 } \vert \alpha { \rangle}|^4 \\
= \overline { \cos^2[\theta_{{\cal l}_{{\textrm{d-}}\hat{h } } , \ , ( \vert \alpha_{0 } { \rangle}{\langle}\alpha_{0}\vert ) } ] } ^{\alpha_{0 } } { \ ; } , \qquad\mbox { } \label{eta } \ ] ] where @xmath69 is the space of all diagonal operators , _ including _ the identity .
let us now turn to a particular application of our theory : namely , to the problem of optimization of the set of conserved quantities to be used to enhance the predictive power of thermodynamics , extended to include additional conserved quantities ; both integrable systems and the systems in between integrable and completely thermalizable will be considered . since the first demonstration of such enhancement , on the example of one - dimensional hard - core bosons @xcite , the question of which set of conserved quantities should be used in a general case has remained largely open . in the hard - core boson case and in subsequent realizations
@xcite , there existed a straightforward map between the integrable system of interest and a hidden underlying system of free particles .
conventionally , the occupation numbers of the one - body eigenstates of the latter were assumed to constrain the relaxation dynamics the most .
these were subsequently included in the gibbs exponent producing the so - called generalized gibbs ensemble ( gge)@xcite .
the above choice of the conserved quantities is indeed the most natural one .
however , there exist several strong incentives to formalize the choice of the conserved quantities for gge : ( a ) there are numerical indications that in case of a disorder - induced _ localization _ , the one - body free - fermionic occupations do not improve the predictive power of thermodynamics at all @xcite ; ( b ) not every integrable system can be mapped to free particles ; and ( c@xmath70 in the intermediate systems , between integrable and ergodic , the variations of the expectation values of observables from one eigenstate to another are , for all practical purposes , indistinct from the ones generated by well defined integrals of motion .
it is therefore desirable to devise a thermodynamic recipe that does not rely on a priori chosen constants of motion .
ideally , one should not even assume that constants of motion exist .
the recipe we suggest is presented below .
to begin , one identifies a linear subspace of observables whose relaxation the generalized ensemble aims to describe .
next , one chooses a diagonal traceless observable that minimizes the hilbert - schmidt angle between the space of observables of interest and itself .
next , one repeats the procedure in the space of diagonal traceless observables orthogonal to the first chosen .
the procedure is repeated recursively ( every time in the space orthogonal to all the integrals of motion previously chosen ) till the desired predictive power is reached .
this procedure is based on the following exact result and its corollaries , where the hs structure naturally emerges : @xmath71 \ ; \leq \operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{a } \vert\alpha{\rangle } ] \ , \left(\sin^2[\theta_{\hat{i}_{{\textrm{tl}}},\ , \hat{\hat{p}}_{h } \hat{a}_{{\textrm{tl } } } } ] + |\cos[\theta_{\hat{i}_{{\textrm{tl}}},\ , \hat{\hat{p}}_{h } \hat{a}_{{\textrm{tl}}}}]| \ , \underbrace { { \cal o}\left [ \frac{\delta i}{\sqrt{\operatorname{var}_{{\textrm{\upshape mc}}}[{\langle}\alpha\vert \hat{i } \vert\alpha{\rangle } ] } } \right ] } _ { \ll 1 } \right ) \label{upper_bound_000 } { \ ; } , \ ] ] where @xmath72 is a `` super - operator '' that removes the off - diagonal ( with respect to the basis of the eigenstates of @xmath67 ) matrix elements , @xmath73 is the maximal width of the microcanonical window for the additional integral of motion , and @xmath74\}$ ] is a set of intervals tiling the axis of the integral of motion @xmath75 .
( see the extensive discussion in supplementary material @xcite . ) here and below , @xmath76 $ ] and @xmath77 $ ] define the mean square error of the ( microcanonical version of ) gge and of the microcanonical ensemble proper .
figure [ olshanii_fig3 ] shows the result of an application of this procedure to another system of one - dimensional hard - core bosons , smaller than before so that finite - size effects are enhanced .
the space of observables of interest was formed by all distinct matrix elements of the one - body density matrix .
we analyze the momentum distribution , considering both the integrable case ( figs .
[ olshanii_fig3]a , c ) and the case where the integrability is broken by a soft - core finite - range repulsive potential ( figs .
[ olshanii_fig3]b , d ) .
we compare three thermodynamical ensembles .
the first one is the traditional microcanonical ensemble .
the second is a generalized microcanonical ensemble that fixes not only the value of the energy but generally all one - body occupation numbers of the underlying free - fermionic system ( or , in the nonintegrable case , their time - averaged values ) .
the third is an ensemble based on the first few most relevant integrals of motion chosen using the hilbert - schmidt optimization procedure .
fig .
[ olshanii_fig3]c demonstrates that in the integrable case , the accuracy of the second and the third ensembles are comparable while both greatly exceed the one of the conventional thermal ensemble .
this indicates that the free - fermionic integrals of motion are indeed the optimal predictors of the after - relaxation state of the system in this case .
however , the free - fermionic one - body occupations are not available at all in the perturbed case ( see fig . [
olshanii_fig3]d ) . nevertheless , the optimized generalized microcanonical ensemble remains well defined , and it fully retains its predictive power .
remark that for the strength of perturbation chosen , the system remains relatively close to integrable , showing a high inverse participation ratio of @xmath78 . finally , in fig . [ olshanii_fig4 ]
we consider a quench from the ground state in the integrable regime to a strongly perturbed regime ( @xmath79 ) . here
we directly compare the infinite time average with the predictions of both the microcanonical and the optimal gge ensembles .
the predictive power of the latter is indeed higher than of the former .
two distinct sources of deviation from the thermal behavior are traditionally identified .
the first , the mathematically elegant one , is associated with either nontrivial symmetries or with the bethe ansatz @xcite .
the second source , the empirically important , stems from the deviations from the eigenstate thermalization @xcite in finite systems . in small systems , such as the nano - opto - mechanical resonators
@xcite the two become practically indistinct , and no obvious candidates for the relevant conserved quantities are any longer visible : here , our theory offers a unified approach , based on a `` blind '' optimization of the predictive power of thermodynamics .
* predictive power of two extensions of the microcanonical ensemble*. we use the example of a system of one - dimensional hard - core bosons . for this figure : the number of particles is @xmath80 , we use periodic boundary conditions , and the rest of the parameters is the same as for fig .
[ olshanii_fig1]b . *
a * , an upper bound @xcite on the mean - square error of the predictions of the generalized gibbs ensemble , optimal with respect to all one - body observables ( the optimal gge ) , as a function of the allowed number of the additional integrals of motion involved .
the error of the standard microcanonical ( mc ) ensemble is used as a reference .
we consider the integrable case , @xmath81 .
the result is averaged over all one - body observables . *
b * , the same as for * a * , but away from integrability , with @xmath82 .
* c * , an upper bound on the mean - square error of the predictions of the optimal gge , for the momentum distribution , in the integrable case ( @xmath81 ) .
the number of the integrals of motion is fixed to the number of lattice sites , @xmath23 .
the results for the microcanonical ensemble , and for the conventional free - fermionic generalized gibbs ensemble ( gge)@xcite are shown for comparison .
the ensemble variance of the observables is used as a reference .
* d * , the same as for * c * , but with @xmath82 . note that in the nonintegrable case , no free theory is available for comparison . ]
* momentum distribution after a quench from the ground state of another hamiltonian*. the initial state is the ground state of a hard - core boson hamiltonian for @xmath22 atoms on @xmath23 sites , with periodic boundary conditions . at @xmath83 , a soft - core repulsion of strength @xmath84
is turned on .
the microcanonical ensemble is represented by @xmath85 lowest eigenstates .
the gge ensemble incorporates the values of 8 most optimal integrals of motion .
each integral of motion ( except the first , which was strongly correlated with the energy ) was fixed to a window around its initial state value .
the half - width of each window was 10% of the corresponding microcanonical standard deviation . ] |
we study a class of piecewise - deterministic markov processes ( pdmps ) which we call semiflows with jumps . as defined in @xcite a pdmp without active boundaries
is determined by three local characteristics @xmath1 , where @xmath2 is a semiflow describing the deterministic parts of the process , @xmath3 is the intensity of a jump from @xmath4 , and @xmath5 is the distribution of the state reached by that jump .
the problem of existence of invariant measures for markov processes is of fundamental importance in many applications of stochastic processes @xcite .
we consider semiflows that arise as solutions of ordinary differential equations @xmath6 where @xmath7 is a ( locally ) lipschitz continuous mapping .
we assume that @xmath8 is a borel subset of @xmath9 such that for each @xmath10 the solution @xmath11 of with initial condition @xmath12 exists and that @xmath13 for all @xmath14 .
we denote this solution @xmath15 .
then the mapping @xmath16 is borel measurable and satisfies @xmath17 , @xmath18 for @xmath19 , @xmath20 . as
concern jumps we consider a family of measurable transformations @xmath21 , @xmath22 , where @xmath23 is a metric space which carries a borel measure @xmath24 , and a family of measurable functions @xmath25 , @xmath22 , satisfying @xmath26 so that the stochastic kernel @xmath27 is of the form @xmath28 for @xmath29 , where @xmath30 be the borel @xmath31-algebra of subsets of @xmath8 .
this roughly means that if the value of the process is @xmath4 then we jump to the point @xmath32 with probability @xmath33 .
the following standing assumptions will be made .
the intensity function @xmath34 is continuous and @xmath35 the mappings @xmath36 and @xmath37 are measurable so that the stochastic kernel in is well defined .
we assume also that each mapping @xmath38 as well as each @xmath21 is nonsingular with respect to a reference measure @xmath39 on @xmath8 . recall that a measurable transformation @xmath40 is called _ nonsingular _ with respect to @xmath39 if the measure @xmath41 is absolutely continuous with respect to @xmath39 , i.e. , @xmath42 whenever @xmath43 .
let us briefly describe the construction of the pdmp @xmath44 with characteristics @xmath1 ( see e.g. @xcite for details ) . define the function @xmath45 and note that the assumptions imposed on @xmath34 imply that @xmath46 is a distribution function of a positive and finite random variable for every @xmath19 .
let @xmath47 and let @xmath48 be an @xmath8-valued random variable .
for each @xmath49 we can choose the @xmath50th _ jump time _
@xmath51 as a positive random variable satisfying @xmath52 and we define @xmath53 where the @xmath50th _ post - jump position _
@xmath54 is an @xmath8-valued random variable such that @xmath55 and @xmath56 . in this way
, the trajectory of the process is defined for all @xmath57 and @xmath58 is called the explosion time . to define the process for all times , we set @xmath59 for @xmath60 , where @xmath61 is some extra state representing a cemetery point for the process .
the pdmp @xmath44 is called the _ minimal _ pdmp corresponding to @xmath1 .
it is said to be _ non - explosive _ if @xmath62 for @xmath39-almost every ( @xmath39-a.e . ) @xmath19 , where @xmath63 is the distribution of the process starting at @xmath64 .
we denote by @xmath65 the expectation operator with respect to @xmath63 .
our main result is the following .
[ t : cont ] assume that the chain @xmath66 has only one invariant probability measure @xmath67 absolutely continuous with respect to @xmath39 .
if the density @xmath68 is strictly positive a.e . then the process @xmath44 is non - explosive and it can have at most one invariant probability measure absolutely continuous with respect to @xmath39 . moreover , if @xmath69 then the process @xmath70 has a unique invariant density and it is strictly positive a.e .
the problem of existence and uniqueness of an invariant probability measure for the process @xmath70 with comparison to the similar problem for the chain @xmath66 was studied in @xcite in the context of general pdmps with boundaries and under some technical assumptions .
we also refer the reader to @xcite for the study of equivalence between stability properties of continuous time processes and yet another discrete time processes associated with them . here
we concentrate on the existence of absolutely continuous invariant measures and we make use of the results from @xcite .
that is why we need to assume that the semiflow @xmath71 satisfies @xmath72 for all @xmath14 ( this implies that there are no active boundaries ) and that the stochastic kernel @xmath27 describing jumps gives rise to a transition operator @xmath73 on @xmath0 ( see ) so that we can use ( * ? ? ?
* theorem 5.2 ) .
in particular , the kernel @xmath27 as in has the required property and covers many interesting examples .
however , any refinements entail considerable mathematical difficulties and are currently under research .
we study the continuous time process with the help of a strongly continuous semigroup of positive contraction operators @xmath74 ( _ substochastic semigroup _ ) on the @xmath0 space of functions integrable with respect to the measure @xmath39 .
the semigroup can be obtained from the kato
voigt perturbation theorem for substochastic semigroups on @xmath0-spaces and this functional analytic framework is recalled in section [ ssec : evol ] as theorem [ thm : kato ] . using results from @xcite ,
this gives that the chain @xmath66 has the property that there exists a unique linear operator @xmath75 ( _ stochastic operator _ ) on @xmath0 which satisfies : if the distribution of the random variable @xmath76 has a density @xmath77 , i.e. , @xmath78 then @xmath79 has a density @xmath80 .
hence , the density @xmath81 in theorem [ t : cont ] is invariant for the operator @xmath75 .
sufficient conditions for the existence of only one invariant density for stochastic operators are described in section [ sec : ssg ] and are based on @xcite .
section [ ssec : evol ] presents relationships between invariant densities for the semigroup @xmath82 and for the operator @xmath75 . here
the most important results are obtained in theorems [ thm : exunifp ] and [ thm : exunifpi ] and give corollary [ c : unique ] which is our main tool in the proof of theorem [ t : cont ] .
theorems [ thm : exunifp ] and [ thm : exunifpi ] together with corollaries [ c : exunifp ] and [ c : exunifpi ] should be compared with ( * ? ? ?
* theorems 1 and 2 ) and ( * ? ? ?
* theorem 5 ) .
however , we need not to assume that the process is non - explosive and we look for absolutely continuous subinvariant measures . moreover , in @xcite a perturbed substochastic semigroup is obtained with the help of desch s theorem @xcite , which in our setting becomes a particular case of theorem [ thm : kato ] .
if for some @xmath83 and for @xmath4 from a set of positive lebesgue measure the absolutely continuous part in the lebesgue decomposition of the measure @xmath84 is nontrivial , then the semigroup @xmath74 is partially integral as in @xcite .
this allows us to combine theorem [ t : cont ] with ( * ? ? ?
* theorem 2 ) , recalled in section [ sec : ssg ] as theorem [ thm : pr00 ] , to obtain asymptotic stability of the semigroup @xmath74 , i.e. , the density of @xmath85 converges to the invariant density in @xmath0 irrespective of the density of @xmath76 . in that case condition appears to be not only sufficient but also necessary for the existence of an invariant density for the process , see corollary [ cor : as_stab ] .
in section [ sec : pdmp ] we provide sufficient conditions for existence of a unique invariant density for the markov chain @xmath66 in terms of the local characteristics of the semiflow with jumps .
we also show that dynamical systems with random switching evolving in @xmath86 with a finite set @xmath87 , as in @xcite , can be studied with our methods .
section [ sec : example ] contains a detailed study of a two dimensional model of gene expression with bursting illustrating applicability of our results .
our framework can be used to analyze biological processes described by pdmps , see e.g. @xcite for gene regulatory dynamics with bursting and @xcite for dynamics with switching .
let @xmath88 be a @xmath31-finite measure space and @xmath89 be the space of integrable functions .
we denote by @xmath90 the set of all _ densities _ on @xmath8 , i.e. @xmath91 and @xmath92 is the norm in @xmath0 . a linear operator @xmath93 such that @xmath94 is called _ stochastic _ or _ markov _
it is called _ substochastic _ if @xmath73 is a positive contraction , i.e. , @xmath95 and @xmath96 for all @xmath97 .
if @xmath40 is nonsingular then there exists a unique stochastic operator @xmath98 satisfying @xmath99 for all @xmath100 and @xmath101 . the operator @xmath102 is usually called @xcite the _ frobenius - perron _ operator corresponding to @xmath103 . in particular , if @xmath40 is one - to - one and nonsingular with respect to @xmath39 , then @xmath104 where @xmath105 is the radon - nikodym derivative of the measure @xmath41 with respect to @xmath39 .
let @xmath106 $ ] be a _
stochastic transition kernel _
, i.e. , @xmath5 is a probability measure for each @xmath19 and the function @xmath107 is measurable for each @xmath100 , and let @xmath73 be a stochastic operator on @xmath108 .
if @xmath109 for all @xmath110 , then @xmath73 is called the _ transition _ operator corresponding to @xmath27 . a stochastic operator @xmath73 on @xmath108 is called _ partially integral _ or _ partially kernel _ if there exists a measurable function @xmath111 such that @xmath112 for @xmath39-a.e .
@xmath19 and for every density @xmath77 .
we can extend a substochastic operator @xmath73 beyond the space @xmath108 in the following way . if @xmath113 , @xmath114 , @xmath115 , then the pointwise almost everywhere limit of @xmath116 exists and will be denoted by @xmath117 .
for @xmath118 we define @xmath119 ( note that @xmath120 is independent of the particular approximating sequence @xmath116 and that @xmath120 may be infinite . )
moreover , if @xmath73 is the transition operator corresponding to @xmath27 then holds for all measurable nonnegative @xmath77 .
a nonnegative measurable @xmath81 is said to be _ subinvariant ( invariant ) _ for a substochastic operator @xmath73 if @xmath121 ( @xmath122 ) .
note that if @xmath81 is a subinvariant density for a stochastic operator @xmath73 then @xmath81 is invariant for @xmath73 .
a substochastic operator @xmath73 is called _ mean ergodic _ if @xmath123 if a substochastic operator has a subinvariant density @xmath81 with @xmath124 a.e .
, then it is mean ergodic ( see e.g. ( * ? ? ?
* lemma 1.1 and theorem 1.1 ) ) .
we say that a stochastic operator is _ uniquely mean ergodic _ if there is an invariant density @xmath81 such that @xmath125 in particular , if @xmath73 has a unique invariant density @xmath81 and @xmath124 a.e .
then @xmath73 is uniquely mean ergodic ( see e.g. ( * ? ? ?
* theorem 5.2.2 ) ) . moreover
, an operator with this property can not have a non - integrable subinvariant function as the following result shows .
for any measurable @xmath77 the _ support _ of @xmath77 is defined up to sets of measure @xmath39 zero by @xmath126 [ p : exsub ] suppose that a stochastic operator @xmath73 is uniquely mean ergodic with an invariant density @xmath81 . if @xmath127 is subinvariant for @xmath73 and @xmath128 , then @xmath129 .
it is a direct consequence of and the fact that the measure @xmath39 is @xmath31-finite . to prove that an operator has a unique strictly positive invariant density we use the approach from @xcite . a stochastic operator @xmath73 is called _ sweeping _ with respect to a set @xmath100
if @xmath130 from lemma 2 and theorem 2 of @xcite we obtain the following result [ t : inv ] let @xmath8 be a metric space and @xmath131 be the @xmath31-algebra of borel subsets of @xmath8 .
suppose that @xmath73 is the transition operator corresponding to the stochastic kernel @xmath27 satisfying the following conditions 1 .
[ cond : comms ] there is no @xmath73-_absorbing sets _
, i.e. , there does not exist a set @xmath100 such that @xmath132 , @xmath133 and @xmath134 for @xmath39-a.e .
@xmath19 , 2 .
[ cond : funs ] for every @xmath10 there exist @xmath135 , a nonnegative measurable function @xmath136 satisfying @xmath137 , and a positive integer @xmath50 such that @xmath138 for @xmath39-a.e .
@xmath19 and all @xmath139 , where @xmath140 is the ball with center at @xmath141 and radius @xmath142 . then either @xmath73 is sweeping with respect to compact sets or @xmath73 has an invariant density @xmath81 . in the latter case
, @xmath81 is unique and @xmath124 a.e . in order to exclude sweeping
we can use a foster
lyapunov drift condition @xcite .
for the proof of the following see e.g. @xcite .
[ l : sweeping ] let @xmath73 be the transition operator corresponding to a stochastic transition kernel @xmath143 .
assume that the following condition holds 1 .
there exist a set @xmath144 , two positive constants @xmath145 , @xmath146 , and a nonnegative measurable function @xmath147 satisfying @xmath148 and @xmath149 then @xmath150 for all @xmath101 such that @xmath151 .
in particular , @xmath73 is not sweeping with respect to the set @xmath144 .
we conclude this section with the notion of stochastic semigroups and a general result from @xcite concerning possible asymptotic behavior of such semigroups .
a family of substochastic ( stochastic ) operators @xmath74 on @xmath0 which is a @xmath152-_semigroup _ , i.e. , 1 .
@xmath153 ( the identity operator ) ; 2 .
@xmath154 for every @xmath155 ; 3 . for each @xmath156
the mapping @xmath157 is continuous : for each @xmath158 @xmath159 is called a _
substochastic ( stochastic ) semigroup_. a nonnegative measurable @xmath81 is said to be _ subinvariant ( invariant ) _ for the semigroup @xmath74 if it is subinvariant ( invariant ) for each operator @xmath160 .
a stochastic semigroup @xmath74 is called _ asymptotically stable _ if it has an invariant density @xmath81 such that @xmath161 and _ partially integral _ if , for some @xmath162 , the operator @xmath163 is partially integral . [
thm : pr00 ] let @xmath74 be a partially integral stochastic semigroup .
assume that the semigroup @xmath164 has only one invariant density @xmath81 . if @xmath124 a.e .
then the semigroup @xmath74 is asymptotically stable .
note that if the semigroup @xmath74 is asymptotically stable then , for each @xmath162 , the operator @xmath163 is uniquely mean ergodic .
thus , proposition [ p : exsub ] gives the following .
[ p : exsubct ] suppose that a stochastic semigroup @xmath164 is asymptotically stable with an invariant density @xmath81 .
if @xmath127 is subinvariant for @xmath74 and @xmath165 , then @xmath129 .
in this section we study the problem of existence of invariant densities for substochastic semigroups on @xmath0 .
we first recall some notation and a generalization of kato s perturbation theorem @xcite .
let @xmath166 be a substochastic semigroup on @xmath108 .
the infinitesimal _ generator _ of @xmath167 is by definition the operator @xmath168 with domain @xmath169 defined as @xmath170 the operator @xmath168 is closed with @xmath171 dense in @xmath0 . if for some real @xmath172 the operator @xmath173 is one - to - one , onto , and @xmath174 is a bounded linear operator , then @xmath172 is said to belong to the resolvent set @xmath175 and @xmath176 is called the resolvent at @xmath172 of @xmath168 .
if @xmath168 is the generator of the substochastic semigroup @xmath167 then @xmath177 and we have the integral representation @xmath178 the operator @xmath179 is substochastic and @xmath180 for @xmath181 , @xmath182 . we assume throughout this section that @xmath73 is a stochastic operator on @xmath108 , @xmath183 is a measurable function , and that @xmath167 is a substochastic semigroup with generator @xmath184 such that @xmath185 for @xmath186 , where @xmath187 our starting point is the following generation result @xcite for the operator @xmath188 [ thm : kato ] there exists a substochastic semigroup @xmath74 on @xmath108 such that the generator @xmath189 of @xmath74 is an _ extension _ of the operator in , i.e. , @xmath190 the generator @xmath191 of @xmath74 is characterized by @xmath192 and the semigroup @xmath74 is _ minimal _ ,
i.e. , if @xmath193 is another semigroup with generator which is an extension of @xmath194 then @xmath195 for all @xmath182 . moreover , the following are equivalent : 1 .
@xmath74 is a stochastic semigroup .
the generator @xmath191 is the closure of the operator @xmath194 .
[ i : pert ] there is @xmath196 , @xmath197 a.e .
such that for some @xmath198 @xmath199 note that ( see e.g. @xcite ) the generator of @xmath74 is the operator @xmath194 if and only if for some @xmath200 @xmath201 in particular , if @xmath34 is bounded then this condition holds . we also need the substochastic operator @xmath202 defined by @xmath203 it follows from ( *
* theorem 3.6 ) that @xmath75 is stochastic if and only if the semigroup @xmath204 generated by @xmath168 is _ strongly stable _ , i.e. , @xmath205 moreover , if @xmath75 is mean ergodic then the minimal semigroup @xmath74 from theorem [ thm : kato ] is stochastic .
we study relationships between invariant densities of the operator @xmath75 defined by and invariant densities of the minimal semigroup @xmath74 .
our first main result in this section is the following .
[ thm : exunifp ] suppose that the operator @xmath75 has a subinvariant density @xmath81 and let @xmath206 then @xmath207 is subinvariant for the semigroup @xmath82 . in particular ,
if @xmath208 and the semigroup @xmath74 is stochastic , then it has an invariant density .
let @xmath209 for @xmath198 .
since @xmath210 is the resolvent of a substochastic semigroup , we have @xmath211 , @xmath212 , and @xmath207 is nontrivial . from it
follows that @xmath213 .
we have @xmath214 and @xmath215 for @xmath216 .
hence @xmath217 for every @xmath156 , which implies that @xmath218 for all @xmath198 .
the semigroup @xmath219 has the generator @xmath220 , thus @xmath221 for all @xmath222 and @xmath223 . since @xmath224 for every @xmath225
, we conclude that @xmath226 for all @xmath227 and @xmath83 .
consequently , @xmath228 and taking pointwise limits of both sides when @xmath229 and then @xmath230 shows that @xmath207 is subinvariant for @xmath160 . finally , if @xmath160 is stochastic and @xmath208 then @xmath231 and @xmath232 is an invariant density for @xmath160 .
we now give a useful observation .
[ c : pos ] if the operator @xmath75 has a subinvariant density @xmath81 and @xmath124 a.e .
, then the semigroup @xmath74 is stochastic and @xmath207 as defined in satisfies @xmath233 a.e . since @xmath234 and @xmath124 a.e , the operator @xmath75 is mean ergodic .
thus @xmath74 is stochastic .
we have @xmath235 for @xmath198 .
since @xmath210 is a positive bounded operator with dense range , we get @xmath236 a.e .
[ r : pos ] note that if @xmath74 has an invariant density @xmath237 with @xmath238 a.e .
then @xmath74 is stochastic . to see this we check
that condition of theorem [ thm : kato ] holds . by ( * ? ? ?
* remark 3.3 ) , we obtain that @xmath239 for any @xmath97 . on the other hand
, we have @xmath240 , which shows that there is @xmath241 , @xmath238 a.e . , satisfying . the assumption in theorem [ thm : exunifp ] that the subinvariant function @xmath81 is integrable is essential , as the following example shows ( * ? ? ?
* example 4.3 ) .
let @xmath8 be the set of integers and let @xmath39 be the counting measure on @xmath242 so that @xmath243 .
consider @xmath244 where @xmath34 is a positive function such that @xmath245 the semigroup generated by @xmath244 , @xmath246 , being of the form @xmath247 has the resolvent operator @xmath248 , @xmath198 .
let @xmath73 be the frobenius - perron operator corresponding to @xmath249 so that @xmath250 .
we have @xmath251 and @xmath252 for @xmath253 .
thus @xmath254 and @xmath255 .
since the operator @xmath256 with the maximal domain @xmath257 is an extension of the generator @xmath191 of the semigroup @xmath74 ( see e.g. ( * ? ? ?
* theorem 1.1 ) ) , we have @xmath258 and @xmath259 .
it follows from ( * ? ? ?
* example 4.3 ) that @xmath74 is not stochastic .
thus @xmath260 , because otherwise @xmath207 is a strictly positive invariant density for the semigroup @xmath74 , implying that @xmath74 is stochastic , by remark [ r : pos ] .
we next also discuss the problem of integrability of @xmath207 given by .
[ c : boundedphi ] let @xmath207 be defined as in . if @xmath261 then @xmath208 .
in particular , if the function @xmath34 is bounded away from @xmath262 then @xmath208 . if @xmath261 , then @xmath263 is a bounded operator and @xmath264 , which implies that @xmath208 .
suppose now that there is a positive constant @xmath265 such that @xmath266 .
it follows from that @xmath267 for all @xmath268 .
thus the operator @xmath269 is the generator of a substochastic semigroup @xmath270 ( see e.g. ( * ? ? ?
* lemma 4.3 ) ) . on the other hand @xmath271 for every @xmath83 , which shows that @xmath272 for all @xmath156 and @xmath83 .
hence , @xmath273 . the generator @xmath168 might not have a bounded inverse operator , but if the semigroup @xmath166 is strongly stable , then @xmath168 has always a densely defined inverse operator .
we next recall its definition and properties .
let the operator @xmath274 be defined by @xmath275 the mean ergodic theorem for semigroups ( * ? ? ?
* chapter viii.4)(see also ( * ? ? ?
* theorem 12 ) ) together with additivity of the norm in @xmath0 and the characterization ( * ? ? ?
* theorem 3.1 ) of the range of the generator of a substochastic semigroup gives the following .
[ l : dyn ] let @xmath276 be defined by
. then @xmath277 , @xmath278 for @xmath279 , and @xmath280 where @xmath281 is the range of the operator @xmath168
. moreover , if the semigroup @xmath166 is strongly stable then @xmath282 is dense , @xmath283 , @xmath284 for @xmath285 , and @xmath286 we can now prove the following simple fact . [
c : exunifp ] let @xmath276 be defined by .
suppose that @xmath75 is stochastic .
then @xmath75 is the unique bounded extension of the densely defined operator @xmath287 .
moreover , if @xmath81 is an invariant density for @xmath75 then @xmath288 if and only if @xmath289 , in which case @xmath290 and @xmath291 .
we have @xmath277 and @xmath292 .
let @xmath293 . from it follows that @xmath294 since @xmath278 , we obtain that @xmath295 .
the multiplication operator @xmath296 defined by @xmath297 for @xmath298 is closed . since @xmath299 and @xmath300
, we obtain that @xmath301 . hence , @xmath302 and the result follows from proposition [ l : dyn ] .
we next prove a partial converse of theorem [ thm : exunifp ] .
[ thm : exunifpi ] suppose that the semigroup @xmath74 has a subinvariant density @xmath303
. then @xmath304 a.e . and @xmath305 is subinvariant for the operator @xmath75
. moreover , if @xmath306 then @xmath307 .
let @xmath198 be fixed and let @xmath308 .
since @xmath309 for every @xmath83 , we obtain that @xmath310 .
thus @xmath311 .
define @xmath312 from it follows that @xmath313 we have @xmath314 , @xmath315 , and @xmath316 a.e .
( see e.g. ( * ? ? ?
* lemma 6.17 ) ) .
moreover , @xmath317 , @xmath315 , and @xmath318 .
thus , we obtain that @xmath319 which gives @xmath320 consequently , @xmath304 a.e .
since @xmath179 is substochastic , the operator @xmath210 can be extended to the space of nonnegative measurable functions by setting @xmath321 which implies that @xmath322 since @xmath323 , we conclude that @xmath324 which gives @xmath325 and completes the proof of the first part .
suppose now that @xmath326 .
this implies that @xmath327 and that @xmath156 , by .
hence , @xmath328 .
[ c : exunifpi ] suppose that the semigroup @xmath74 has an invariant density @xmath127 .
then @xmath305 is subinvariant for the operator @xmath75 .
moreover , if @xmath326 and @xmath75 is stochastic , then @xmath329 and @xmath330 is an invariant density for @xmath75 .
recall that @xmath127 is a fixed point of each operator @xmath160 if and only if @xmath331 .
thus , @xmath303 and @xmath332 . from theorem
[ thm : exunifpi ] it follows that @xmath333 , thus @xmath334 .
suppose that @xmath335 .
then @xmath336 , which implies that @xmath337 .
since the operator @xmath75 is stochastic , condition holds .
recall that @xmath168 is the generator of the semigroup @xmath166 .
thus @xmath338 and we infer that @xmath339 , which contradicts the fact that @xmath340 and completes the proof that @xmath81 is a density . because @xmath75 is stochastic , the subinvariant @xmath81 is invariant .
we establish the following useful result when combined with theorem [ thm : pr00 ] .
[ c : unique ] assume that the operator @xmath75 is stochastic and uniquely mean ergodic with an invariant density @xmath81 .
then the semigroup @xmath74 is stochastic , it can have at most one invariant density , and @xmath326 for any invariant density @xmath127 . moreover , if @xmath341 , where @xmath342 is as in , then @xmath343 is the unique invariant density for the semigroup @xmath74 . from theorem [ thm : exunifpi ]
it follows that if @xmath77 is an invariant density for @xmath74 then @xmath344 a.e . and @xmath345 .
we have @xmath346 , by proposition [ p : exsub ] , implying that @xmath347 .
hence , @xmath348 and @xmath349 is an invariant density for @xmath75 , by corollary [ c : exunifpi ] .
suppose now that the semigroup @xmath74 has two invariant densities @xmath350 .
we have @xmath351 and @xmath215 for @xmath285 .
since @xmath81 is the unique invariant density for the operator @xmath75 , we obtain that @xmath352 which implies that @xmath353 the operator @xmath75 is stochastic thus @xmath338 by .
consequently @xmath354 and @xmath355 , because @xmath356 .
the last part follows from theorem [ thm : exunifp ] .
observe that if the function @xmath34 is bounded then the assumption that @xmath75 is mean ergodic is not needed in corollary [ c : unique ] , since then automatically the semigroup is stochastic and @xmath346 for every @xmath357 .
instead we can only assume that @xmath75 has a unique invariant density @xmath81 .
before we give the proof of theorem [ t : cont ] , we recall the relation established in ( * ? ? ? * section 5.2 ) between minimal pdmps and the minimal semigroups .
let @xmath44 be the minimal pdmp on @xmath8 with characteristics @xmath1 and let @xmath39 be a @xmath31-finite measure on @xmath131 .
we assume that @xmath358 is the transition operator corresponding to @xmath27 and that the semigroup @xmath359 , with generator @xmath184 satisfying , is such that @xmath360 for all @xmath14 , @xmath196 , @xmath361 . observe that if @xmath34 satisfies condition then the semigroup @xmath166 is strongly stable .
the semigroup @xmath74 will be referred to as the _ minimal semigroup on @xmath108 corresponding to _ @xmath1 .
the following result combines theorem 5.2 and corollary 5.3 from @xcite .
[ thm : exist ] let @xmath362 be the sequence of jump times and @xmath363 be the explosion time for @xmath44 . then the following hold : 1 .
the operator @xmath75 as defined in is the transition operator corresponding to the discrete - time markov process @xmath66 with stochastic kernel @xmath364 2 . for any @xmath365 , a density @xmath77 , and
@xmath83 @xmath366 3 .
[ cond : nexplosive ] the semigroup @xmath367 is stochastic if and only if @xmath368 in that case if the distribution of @xmath76 has a density @xmath369 then @xmath85 has the density @xmath370 for all @xmath83 .
theorem [ t : cont ] is a direct consequence of the following result .
observe also that it follows from condition of theorem [ thm : exist ] that the process @xmath371 is non - explosive .
[ t : contsemigroup ] let @xmath75 be the transition operator corresponding to the stochastic kernel given by .
suppose that @xmath75 has a unique invariant density @xmath81 and that @xmath124 a.e
. then the minimal semigroup @xmath372 corresponding to @xmath1 is stochastic and it can have at most one invariant density . moreover , if condition holds , then the semigroup @xmath372 has a unique invariant density and it is strictly positive a.e . since the stochastic operator @xmath75 has a unique invariant density @xmath81 and @xmath124 a.e .
, @xmath75 is uniquely mean ergodic .
thus the first assertion follows from corollary [ c : unique ] . if , moreover , condition holds then @xmath341 , where @xmath342 is defined by , since @xmath373 in that case @xmath374 is the unique invariant density for @xmath367 .
we conclude this section with the following characterization of asymptotic behavior of the minimal semigroup .
[ cor : as_stab ] assume that the minimal semigroup @xmath367 is partially integral .
suppose that @xmath75 has a unique invariant density @xmath81 and that @xmath124 a.e .
then @xmath367 is asymptotically stable if and only if condition holds .
the semigroup @xmath367 is stochastic .
if condition holds then theorems [ t : cont ] and [ thm : pr00 ] imply asymptotic stability . to get the converse
we show that we can apply corollary [ p : exsubct ] to @xmath375 . since @xmath73 is the transition operator corresponding to @xmath27
, we obtain , by approximation , equation , and fubini s theorem , @xmath376 for all @xmath365 and @xmath101 .
substituting @xmath377 and @xmath378 gives @xmath379 which implies that @xmath380 for @xmath39-a.e .
hence @xmath381 . from corollary
[ c : unique ] it follows that @xmath326 for any invariant density @xmath127 for the semigroup @xmath367 , which , by corollary [ c : exunifpi ] , implies that @xmath382 . from theorem [ thm : exunifp ] it follows that @xmath290 is subinvarint for the semigroup @xmath367 .
consequently , @xmath383 and if the semigroup is asymptotically stable then corollary [ p : exsubct ] implies that @xmath341 giving condition .
let the standing hypothesis from introduction hold and let @xmath384 , where @xmath39 is the lebesgue measure on @xmath9 .
the transition operator @xmath73 corresponding to @xmath27 , as in , is of the form @xmath385 where @xmath386 is the frobenius - perron operator for @xmath387 .
the stochastic kernel @xmath388 in is given by @xmath389 for @xmath390 and can be represented as @xmath391 where @xmath392 for all @xmath393 , @xmath19 .
the transition operator @xmath75 on @xmath0 corresponding to @xmath388 becomes @xmath394 given @xmath395 and @xmath396 we denote by @xmath397 the sequence @xmath398 .
we define inductively transformations @xmath399 for @xmath49 , by setting @xmath400 and nonnegative functions @xmath401 by @xmath402 consequently , the @xmath50th iterate stochastic kernel @xmath403 is of the form @xmath404 where @xmath405 denotes the product of the measure @xmath24 on @xmath406 . in the rest of this section
we assume that both mappings @xmath36 and @xmath37 are continuous as well as the intensity function @xmath34 .
furthermore , for every @xmath19 and @xmath407 let the transformation @xmath408 be continuously differentiable and let @xmath409 denote its derivative .
[ t : meyn ] let @xmath10 .
assume that there exists @xmath410 such that @xmath411 and the rank of @xmath412 is equal to @xmath413 .
then there exist a constant @xmath414 and open sets @xmath415 , @xmath416 containing @xmath417 and @xmath418 , respectively , such that for all @xmath419 and @xmath19 @xmath420 we adapt the proof of lemma 6.3 in @xcite to our situation .
if the rank of @xmath421 is equal to @xmath413 , then we can choose @xmath413 variables @xmath422 from @xmath423 in such a way that the derivative of the transformation @xmath424 is invertible .
in that case , we write @xmath425 and we take @xmath426 as the remaining coordinates of @xmath427 , so that , up to the order of coordinates , we denote @xmath427 by @xmath428 . we also write @xmath429 for @xmath430 . by assumption
, there exists @xmath431 such that @xmath432 and the rank of @xmath433 is equal to @xmath413 for @xmath434 , @xmath435 , @xmath436 so , in what follows , we identify every @xmath427 with this particular choice of coordinates @xmath437 and @xmath426 . since the rank is a lower semicontinuous function , the rank of @xmath438 is equal to @xmath413 in a neighborhood of @xmath439 , @xmath440 , @xmath441 , @xmath141 .
for @xmath428 we define the mapping @xmath442 by the formula @xmath443 consequently , the determinant of @xmath444 $ ] is nonzero in a neighborhood of @xmath439 , @xmath440 , @xmath441 , @xmath141
. we can rewrite @xmath403 in the form @xmath445 for all @xmath19 and @xmath419 . using continuity ,
we can find a positive constant @xmath446 and open sets @xmath447 , @xmath448 , @xmath449 and @xmath450 such that @xmath451|^{-1}\geq c$ ] for @xmath452 , @xmath453 , @xmath454 , @xmath455 .
we write @xmath456 to indicate that the point @xmath457 belongs to @xmath456 . moreover , for @xmath458 we can find an open set @xmath459 such that @xmath460 . hence , for all @xmath452 and for every set @xmath419
we have @xmath461\right| du dv \nu^n(dw).\ ] ] substituting @xmath462 and @xmath463 we obtain @xmath464 by the choice of the set @xmath465 we get @xmath466 where @xmath467 and @xmath468 is the @xmath469 dimensional lebesgue measure of the set @xmath470 when @xmath471 , and it is @xmath472 , otherwise . to apply lemma [ t : meyn ] we have to calculate the rank of @xmath412 , which is the most difficult part .
we next describe two possibilities how to make these calculations easier .
[ rem : small_s ] using the continuity of derivatives with respect to @xmath473 and taking the limit when each @xmath474 goes to zero from the right , the limit of the derivative @xmath412 becomes of the form @xmath475,\ ] ] where @xmath476 and @xmath477 for @xmath478 is given inductively by @xmath479 .
since the transformations @xmath480 , @xmath481 , and the mapping @xmath482 are explicitely defined , the rank of the matrix in can be obtained much easier then the rank of @xmath412 . moreover
, lower semicontinuity of the rank allows us to find @xmath427 with positive coordinates .
[ rem : theta ] suppose that @xmath23 is an open subset of @xmath483 for some positive @xmath484 and @xmath24 is the lebesgue measure .
assume also that transformations @xmath485 are continuously differentiable .
then , for a given @xmath19 we can consider the derivative of the transformation @xmath486 , which can be written as @xmath487.\ ] ] lemma [ t : meyn ] remains true under the assumption that the rank of the matrix @xmath488 , instead of @xmath409 , is equal to @xmath413 . as in @xcite , we can introduce the notation @xmath489_{\substack{y = t_{(\theta^n , s^n)}(x)\;\;\\\theta=\theta_{n+1 } , s = s_{n+1 } } } , \\ \psi_{n } & : = \psi_{n}(x,(\theta^{n+1},s^{n+1 } ) ) = \left[\frac{\partial t_{(\theta , s)}(y)}{\partial ( \theta , s)}\right]_{\substack{y = t_{(\theta^n , s^n)}(x)\;\;\\\theta=\theta_{n+1 } , s = s_{n+1 } } } , \end{split}\ ] ] where the derivatives are evaluated at @xmath490 and for @xmath491 . here
@xmath492 for @xmath493 .
then the matrix @xmath488 can be rewritten in the form @xmath494.\ ] ] now we provide sufficient conditions for which the assumptions of theorem [ t : inv ] are satisfied for the transition operator @xmath75 corresponding to @xmath388 as defined in . for each @xmath19
we define the set @xmath495 [ c : stcom ] assume that @xmath496 for every @xmath19 .
suppose also that there is no @xmath75-absorbing sets . then either @xmath75 is sweeping with respect to compact subsets of @xmath8 or @xmath75 has a unique invariant density @xmath81 . in the latter case , @xmath124 a.e .
[ rem : k - abs ] observe that if there is a non - trivial @xmath75-absorbing set , then there is a non - trivial set @xmath497 such that @xmath498 this may be rewritten as @xmath499 where @xmath500 and @xmath501 once we know that a unique invariant density exists for the operator @xmath75 , we can use corollary [ cor : as_stab ] to prove asymptotic stability of the semigroup @xmath372 .
we need to check that the semigroup @xmath372 is partially integral .
our next result gives a simple condition for that .
[ t : meyncont ] let @xmath10 , @xmath83 and @xmath502 .
define @xmath503 and assume that there exists @xmath504 such that @xmath411 and the rank of @xmath505 is equal to @xmath413 .
then there exist a constant @xmath414 and open sets @xmath415 , @xmath416 containing @xmath417 and @xmath506 , respectively , such that for all @xmath419 and @xmath19 @xmath507 in particular , the semigroup @xmath372 is partially integral .
observe that if @xmath4 is such that @xmath508 , then @xmath509 thus , to check whether condition is satisfied , it is sufficient to prove that @xmath510 since we have @xmath511 where @xmath512 is a positive continuous function defined by @xmath513 for @xmath19 , @xmath14 , we can obtain in an analogous way as in the proof of lemma [ t : meyn ] .
as in remarks [ rem : small_s ] and [ rem : theta ] , we can simplify the calculation of the rank of @xmath505 .
[ rem : cont ] analogously to remark [ rem : small_s ] , the limit of the derivative @xmath505 when @xmath514 go to zero , is of the form @xmath515,\ ] ] where @xmath476 and @xmath479 for @xmath478 .
a similar approach to check this `` rank condition '' is used in ( * ? ? ?
* proposition 3.1 ) and @xcite as well as in @xcite and @xcite .
in the case when @xmath23 is an open subset of @xmath483 and we can take derivative with respect to @xmath481 we have @xmath516,\ ] ] for @xmath19 . using the notation as in and defining additionally the derivatives @xmath517_{\substack{s = t - s(n)\,\quad\\ y = t_{(\theta^n , s^n)}(x ) } }
= \left[0|-g(t_{(\theta^n , s^n)}(x))\right],\\ & \upsilon_{x , n}:=\upsilon_{x , n}(x,(\theta^{n},s^{n}))= \left[\frac{\partial\pi_{s}y}{\partial y}\right]_{\substack{s = t - s(n)\,\quad\\ y = t_{(\theta^n , s^n)}(x ) } } , \end{split}\ ] ] we have @xmath518.\ ] ] we will show how our results can be applied in one particular example in the next section .
we conclude this section with the idea how to write dynamical systems with random switching as studied in @xcite , in our framework . given a finite or countable set @xmath87 , consider a family of locally lipschitz functions @xmath519 , @xmath520 , and the differential equation @xmath521 we assume that there exists a set @xmath522 such that for every @xmath523 and @xmath524 the solution @xmath11 of @xmath525 with initial condition @xmath12 exists and that @xmath526 for all @xmath527 .
we denote this solution by @xmath528 .
then , the general solution of the system may be written in the form @xmath529 this gives one semiflow on @xmath530 which is generated by the differential equation @xmath531 where the function @xmath532 is of the form @xmath533 let @xmath39 be the product of the lebesgue measure @xmath534 on @xmath9 and the counting measure @xmath24 on @xmath535 .
we define the transformation @xmath536 , @xmath537 , by @xmath538 each transformation is nonsingular with respect to @xmath39 since @xmath539 we assume that @xmath540 , @xmath541 , are nonnegative continuous functions satisfying @xmath542 for all @xmath520 , @xmath543
. then we can define the intensity function @xmath34 by @xmath544 and the densities @xmath545 , @xmath537 , by @xmath546 and @xmath547 as a particular example of dynamical systems with random switching , one can consider a standard birth - death process by taking @xmath548 , @xmath549 and @xmath550 for @xmath551 or @xmath552 . then @xmath553 .
according to , we can write explicitly formulas for the density @xmath554 and for the transformation @xmath555 for each @xmath50 we get a general form of @xmath556 for @xmath557 and @xmath423 , which is @xmath558 this may be rewritten as @xmath559 where @xmath560 using this notation we adjust the definition of the set in as follows @xmath561 for such semiflow with jumps , we can modify the proof of lemma [ t : meyn ] , to get the next result for the corresponding operator @xmath75 .
[ c : jumps ] assume that @xmath562 for every @xmath563 .
suppose also that there is no @xmath75-absorbing sets .
then either @xmath75 is sweeping with respect to compact subsets of @xmath8 or @xmath75 has a unique invariant density @xmath81 . in the latter case , @xmath124 a.e .
in particular , if @xmath564 is compact , then @xmath75 has a unique invariant density . to verify whether the rank of @xmath565 is equal to @xmath413
, we may use either remark [ rem : small_s ] or lie brackets as in ( * ? ? ?
* theorem 3 ) , ( * ? ? ?
* theorem 4.4 ) .
it is worth to mention that in @xcite it is assumed that the set @xmath564 is compact .
in this section we study a particular example of a two dimensional pdmp @xmath566 with values in @xmath567 . we let @xmath568 and @xmath569 denote the concentrations of mrna and protein respectively .
we assume that the protein molecules undergo degradation at rate @xmath570 and that the translation of proteins from mrna is at rate @xmath571 .
the mrna molecules undergo degradation at rate @xmath572 that is interrupted at random times @xmath573 when new molecules are being produced with intensity @xmath34 depending at least on the current level @xmath569 of proteins . at each @xmath574 a random amount @xmath575 of mrna molecules
is produced , which is independent of everything else and distributed according to a density @xmath576 .
therefore , @xmath577 and the transformation @xmath480 is given by the formula @xmath578 hence , the jump kernel is of the form @xmath579 so that the transition operator @xmath73 is as follows @xmath580 the semiflow is defined by the solutions of the system of equations @xmath581 and it can be expressed by the formula @xmath582 where @xmath583 if @xmath584 then we have @xmath72 for all @xmath14 and the transformation @xmath585 is of the form @xmath586 the assumption @xmath584 is biologically reasonable , see e.g. @xcite and references therein , were it was recalled that a fast process of mrna degradation has been observed in bacterias , i.e. _ e. coli_. the production of mrna molecules can be described by exponential density with mean @xmath587 @xmath588 while the intensity @xmath34 is a hill function depending only on the second coordinate , @xmath589 where @xmath590 and @xmath591 are constants . if @xmath592 we assume , additionally , that @xmath593 and @xmath594 .
we show that the minimal semigroup @xmath372 is asymptotically stable . taking @xmath595 with @xmath24 being the lebesgue measure on @xmath596 , we can express the stochastic kernel @xmath388 as in . with the help of corollary [ c : stcom ]
we prove that the transition operator @xmath75 corresponding to @xmath388 has a unique invariant density , which is strictly positive a.e .
first , we need to check the assumptions of corollary [ c : stcom ] .
the function @xmath597 defined as in is strictly positive for all @xmath19 and @xmath598 , since both @xmath34 and @xmath576 are strictly positive . taking into account remark [ rem : theta ] , we consider the derivative @xmath488 instead of @xmath409 .
we have @xmath599,\quad \psi_{k}=\left [ \begin{array}{cc } 1 , & \multirow{2}{*}{$g(\pi_{s_{k+1 } } t_{(\theta^{k},s^{k})}(x))$ } \\ 0 , & \end{array } \right],\ ] ] where @xmath600 for arbitrary @xmath601 we can calculate @xmath602 = \left [ \begin{array}{cc } 1 , & -\gamma_1 x_1 e^{-\gamma_1 s_1}\\ 0 , & -\gamma_2 x_2 e^{-\gamma_2 s_1 } + x_1\frac{\beta_2}{\gamma_1-\gamma_2}(\gamma_1 e^{-\gamma_1 s_1 } - \gamma_2 e^{-\gamma_2 s_1 } ) \end{array } \right].\ ] ] the rank of @xmath603 is equal to @xmath604 if and only if @xmath605 if this condition does not hold we need to consider @xmath606 .
we have @xmath607 \\ = \left [ \begin{array}{cc|cc } e^{-\gamma_1 s_2 } , & e^{-\gamma_1 s_2}g_1(\pi_{s_1}x ) & 1 , & g_1(\pi_{s_2 } t_{(\theta^1,s^1)}(x))\\ \vartheta(s_2 ) , & \vartheta(s_2)g_1(\pi_{s_1}x ) + e^{-\gamma_2 s_2}g_2(\pi_{s_1}x ) & 0 , & g_1(\pi_{s_2 } t_{(\theta^1,s^1)}(x ) ) \end{array } \right]\ ] ] and , looking at the first and the third column , we see that the rank of @xmath608 is equal to @xmath604 .
this implies that @xmath496 for every @xmath19 .
we now show that there is no @xmath75-absorbing sets . by remark [ rem : k - abs ] it is enough to show that @xmath609 for @xmath39-a.e .
@xmath19 . assume first that the point @xmath610 is such that @xmath611 .
then its trajectory has the shape shown in figure [ fig : semiflow1 ] .
then the grey area covers the set @xmath612 and we see that consecutive iterates give the rest .
suppose now that @xmath613 .
then the set @xmath612 is as in figure [ fig : semiflow2 ] .
corollary [ c : stcom ] implies that either @xmath75 is sweeping with respect to compact sets or @xmath75 has a unique invariant density @xmath81 . to exclude sweeping , we use proposition [ l : sweeping ] for the operator @xmath75 and we take @xmath614 we have @xmath615
since @xmath616 has the distribution function as in , we obtain @xmath617 hence , we get @xmath618 where @xmath619 notice that @xmath620 is bounded from above by a constant and that @xmath621 tends to @xmath622 as @xmath623 for every @xmath624 . since the function @xmath34 has a positive lower bound @xmath265 , we obtain @xmath625 from fatou s lemma it follows that @xmath626 the function in is continuous , thus bounded on compact sets .
consequently , implies that condition is satisfied and completes the proof that @xmath75 has a unique invariant density .
now we look at the process @xmath627 .
the matrices @xmath628 and @xmath629 from remark [ rem : cont ] are of the form @xmath630,\quad \upsilon_{x , n}=\left[\begin{array}{cc } e^{-\gamma_1(t - s(n ) ) } , & 0 \\ \vartheta(t - s(n ) ) , & e^{-\gamma_2(t - s(n ) ) } \end{array}\right].\ ] ] hence @xmath631 can be expressed by @xmath632\\ & = \left[\begin{array}{cc|cc } e^{-\gamma_1(t - s_1 ) } , & * & e^{-\gamma_1(t - s(2 ) ) } , & * \\ e^{-\gamma_1s_2}\vartheta(t - s(2))+e^{-\gamma_2(t - s(2))}\vartheta(s_2 ) , & * & \vartheta(t - s(2 ) ) , & * \end{array}\right ] , \end{split}\ ] ] where the first and the third column are linearly independent and the remaining columns are not important for the calculation .
it is worth to notice that we need to use instead of the matrix in since its every two columns are linearly dependent .
this proves that lemma [ t : meyncont ] holds , in other words , the semigroup @xmath367 corresponding to the process @xmath371 is partially integral .
we conclude from corollary [ cor : as_stab ] that the semigroup @xmath74 is asymptotically stable .
10 [ 1]`#1 ` [ 2][]#2 l. arlotti , a perturbation theorem for positive contraction semigroups on @xmath633-spaces with applications to transport equations and kolmogorov s differential equations , acta appl . math . 23
( 1991 ) 129144 .
m. mokhtar - kharroubi , on strong convergence to ergodic projection for perturbed substochastic semigroups , in : semigroups of operators - theory and applications , springer proceedings in mathematics & statistics , vol .
113 , springer , new york , 2015 , 89103 .
r. rudnicki , m. tyran - kamiska , piecewise deterministic markov processes in biological models , in : semigroups of operators - theory and applications , springer proceedings in mathematics & statistics , vol .
113 , springer , new york , 2015 , 235255 .
a. tomski , the dynamics of enzyme inhibition controlled by piece - wise deterministic markov process , in : semigroups of operators - theory and applications , springer proceedings in mathematics & statistics , vol .
113 , springer , new york , 2015 , 299316 . | the problem of existence and uniqueness of absolutely continuous invariant measures for a class of piecewise deterministic markov processes is investigated using the theory of substochastic semigroups obtained through the kato
voigt perturbation theorem on the @xmath0-space .
we provide a new criterion for the existence of a strictly positive and unique invariant density for such processes .
the long time qualitative behavior of the corresponding semigroups is also considered . to illustrate our general results we give a detailed study of a two dimensional model of gene expression with bursting .
piecewise deterministic markov process , stochastic semigroup , invariant density , dynamical systems with switching , gene expression models 47d06,60j25,60j99,92c40 |
the combination of unitary techniques together with chiral dynamics of meson and baryon systems already used in @xcite and @xcite to study low - energy @xmath11 scattering , has reached a certain maturity , it has been made more systematic and it has also been applied to the study of a large variety of physical processes @xcite .
the idea behind this work is the use of the unitarity constraints in coupled channels , which provide the imaginary part of the inverse of the scattering @xmath12 matrix .
from there , by means of dispersion relations , or equivalent techniques , one can construct the whole @xmath12 matrix . in the strangeness @xmath13
sector the n / d , or dispersion relation method has been applied in @xcite with the result that the @xmath12 matrix in coupled channels can be written as @xmath14^{-1}\ , v \label{eq : bs1}\ ] ] with @xmath15 obtained from the lowest order chiral lagrangian for meson baryon interaction and @xmath16 a properly regularized loop function of the meson baryon propagators of the intermediate states .
the function @xmath15 is given by @xmath17 with @xmath18 coefficients evaluated in @xcite and the @xmath16 function given by @xmath19 \right\ } \
, \label{eq : gpropdr}\end{aligned}\ ] ] where @xmath20 is the scale of regularization and @xmath21 is a subtraction constant which is chosen by fits to the data .
( [ eq : bs1 ] ) is nothing but the bethe salpeter equation where the kernel @xmath15 is given by the lowest order chiral amplitude .
in @xcite it was also found that the subtraction constant is easily related to the cut off employed in @xcite , and hence the equivalence of the n / d method of @xcite and the bethe salpeter equation of @xcite was established .
yet , the n / d method with the dimensionally regularized @xmath16 function is preferable to extend the approach to higher energies .
the structure of the @xmath12 matrix in eq .
( [ eq : bs1 ] ) allows for the existence of poles which correspond to resonances that we call dynamically generated , since they appear thanks to the multiple scattering implicit in the bethe salpeter equation , with the lowest order amplitudes playing the role of the potential . in this sense , both in @xcite and @xcite , as well as in @xcite , one finds very clearly the @xmath22 resonance ( see fig .
1 ) , which has been claimed for long to be a quasibound state of @xmath23 .
more novel have been the findings of @xcite where the same approach of @xcite , with the subtraction constants evaluated from the unique cut off in all channels of @xcite , has lead to two new resonances , which we identify with the @xmath1 and the @xmath2 . in the amplitudes in fig . 2
one can see how the @xmath1 shows up both in the real and imaginary parts of the amplitudes .
one of the interesting findings concerning the @xmath1 state is that it couples very strongly to the @xmath24 states , and quite weakly to the other channels , which allows one to identify the @xmath1 with a quasibound state of @xmath24 .
even more interesting are the findings of @xcite , where using also subtraction constants of the natural size found in the study of the @xmath13 sector , one finds a resonance for @xmath25 which could be identified in principle with the @xmath4 or @xmath26 @xmath27 resonances .
however , a detailed study of the partial decay widths of the found resonance ( which are obtained from the residues of the transition amplitudes at the pole in the second riemann sheet of the complex plane ) , clearly shows that this resonance is at odds with the properties of the @xmath26 , and by elimination can only correspond to the @xmath4 state , hence allowing one to theoretically determine the spin and parity for this resonance as @xmath28 , which is unknown in the particle data book .
at the same time one makes predictions for the partial decay rates of this resonance which are also unknown so far .
one of the results in the study of @xcite is the good description of the low energy cross sections of @xmath29 to different channels . the success of this theory has allowed us to revise the problem of the scattering length of @xmath7 , which has been approached before @xcite . using the fixed centre approximation to the faddeev equations , proved to be excellent for this problem in @xcite ,
we have studied this scattering length in @xcite .
the problem is subtle since there is no convergence of the expansion of the multiple scattering series implicit in the faddeev equations .
second , the results are quite sensitive to the input used for the @xmath23 amplitudes . the chiral approach has allowed one to readdress this latter issue from a new perspective in which special attention is given to questions of unitarity , analyticity , and resonance properties , which make the new amplitudes quite accurate in principle .
the results obtained in our approach are @xmath30 fm , which differ appreciably from the results quoted as best in ref .
@xcite in their multichannel approach of @xmath31 fm .
this quantity is one of the important quantities to be measured in the near future at frascati in the dear experiment @xcite and should put further constraints on the theoretical approaches , and , in particular , a challenge to the chiral unitary theory which makes these novel predictions .
another of the applications of this chiral unitary approach in the strange sector is the evaluation of the @xmath8 selfenergy in the nuclear medium .
pauli exclusion effects in the intermediate states were taken into account in @xcite , which lead to a shift of the @xmath22 resonance to higher energies and a considerable attraction on the kaon . a further selfconsistent treatment of the problem , including the calculated kaon selfenergy in the intermediate meson baryon loop functions , @xcite , moved the resonance back and produced a moderate attraction .
further work in @xcite taking into account the renormalization of the intermediate pions and also the baryon selfenergies , produced new modifications leading to wider @xmath8 spectral functions and still a moderate @xmath8 attraction .
this attraction is only of the order of @xmath32 mev at @xmath33 and threshold , and moderately dependent on the energy .
it is surprising that this potential turns out to be so small when it was claimed in the past to be of the order of 200 mev to be able to interpret the data on @xmath8 atoms .
yet , it was proved in @xcite that with this potential one could obtain a good description of @xmath8 atoms , something that has been corroborated later on in @xcite and @xcite .
this potential also leads to very deeply bound @xmath8 states , with about 30 - 40 mev binding , but with very large widths of the order of 80 - 100 mev . with a potential of this type
there are no hopes that one can obtain narrow deeply bound @xmath8 states as claimed in @xcite .
the reason for the claim in that latter work is that a large binding energy is obtained , of the order of those used in the absence of sefconsistency in previous works , and also that many decay channels , which would be still open at these energies ( like @xmath34 ) , and which are evaluated in @xcite , are omitted in @xcite .
another outcome of these calculations is that kaon condensation in neutron stars is far more unlikely than has been assumed in the past .
the weak decay of @xmath9 hypernuclei and particularly , the neutron to proton induced decay ratio , @xmath10 , has been a battlefield for years , with simple models based on just one pion exchange providing values around 0.1 for this ratio , while experimentally one had values around unity or bigger , albeit with large errors .
a thorough recent study of this process , including kaon exchange and correlated and uncorrelated two pion exchange , was done in @xcite , where good results for the total rates were found for different nuclei and a ratio @xmath10 of the order of 0.54 was found for all nuclei .
the contribution of the correlated two pion exchange was done following the lines of the work for @xmath35 exchange in the @xmath36 interaction of @xcite , where no @xmath35 was put but was simulated by the interaction of two pions in the scalar isoscalar sector .
similar results for @xmath10 have also been obtained more recently in @xcite .
the still puzzling thing is that , in spite of the apparently quite thorough work done in @xcite , there seems to be still discrepancies with the latest and most accurate analysis from the experimental point of view of @xcite , which provides values bigger than one for the @xmath10 ratio , with errors small enough to make the results incompatible with those of @xcite .
the puzzle has been solved recently .
the proton spectra of @xcite were analyzed using the code of @xcite that takes into account the final state interaction of the nucleons after the @xmath9 decay .
this code contained one fatal error and has been corrected recently @xcite , leading to quite different spectra than those obtained in the original work .
the experimental results of @xcite have also been corrected to the light of the new results in @xcite , with the outcome that the new values for @xmath10 are of the order of 0.5 - 0.6 @xcite , which are in perfect agreement with the theoretical results in @xcite .
a short overview of different problems related to the field of few body physics in strong and weak interactions have been made , with the common denominator that they could be faced with fresh light from the new perspective provided by the unitary extensions of chiral perturbation theory .
the short overview here has served to show the potential of this approach to face different problems at intermediate energies , allowing one to establish connections which could not be done prior to the introduction of these techniques .
unitarized chiral perturbation theory is thus proving to be a a quite useful tool to address hadron and nuclear physics at intermediate energies , providing both a systematic working technique , as well as some novel understanding on the nature of those mesonic and hadronic resonances which could not be accommodated in the traditional quark models .
99 n. kaiser , p. b. siegel and w. weise , nucl .
a594 ( 1995 ) 325 n. kaiser , t. waas and w. weise , nucl .
a612 ( 1997 ) 297 j.a .
oller , e. oset and a. ramos , prog .
45 ( 2000 ) 157 j.a .
oller and u.g .
meissner , phys .
b500 ( 2001 ) 263 e. oset and a. ramos , nucl .
phys . a635 ( 1998 ) 99 e. oset , a. ramos and c. bennhold , phys .
b537 , 99 ( 2002 ) a. ramos , e. oset and c. bennhold , nucl - th/0204044 , phys .
lett . in print
g. toker , a. gal and j. m. eisenberg , nucl .
phys . a362 ( 1982 ) 405 .
m. torres , r. h. dalitz and a. deloff , phys .
b174 ( 1986 ) 213 .
a. deloff , phys .
c61 ( 2000 ) 024004 .
kamalov , e. oset , a. ramos , nucl .
a690 ( 2001 ) 494 c. guaraldo et al . ,
a623 ( 1997 ) 311c .
v. koch , phys .
b337 ( 1994 ) 7 t. waas , n. kaiser and w. weise , phys .
b365 ( 1996 ) 12 m. lutz , phys .
lett . b426 ( 1998 ) 12 a. ramos and e. oset , nucl .
a671 ( 2000 ) 481 s. hirenzaki , y. okumura , h. toki , e. oset and a. ramos , phys .
c61 ( 2000 ) 055205 a. baca , c. garcia - recio and j. nieves , nucl .
a673 ( 2000 ) 335 a. cieply , e. friedman , a. gal and j. mares , nucl .
a696 ( 2001)173 y. akaishi and t. yamazaki , phys .
c65 ( 2002 ) 044005 e. oset , h. toki , m. mizobe and t.t .
takahashi , prog .
103 ( 2000 ) 351 d. jido , e. oset and j.e .
palomar , nucl .
a694 ( 2001 ) 525 a. parreno and a. ramos , phys .
c65 ( 2002 ) 015204 o. hashimoto et al .
88 ( 2002 ) 042503 a. ramos , m.j .
vicente - vacas and e. oset , phys .
c55 ( 1997 ) 735 , erratum - ibid.c66 ( 2002 ) 039903 h. outa , talk at panic2002 conference , osaka , october 2002 . | in this talk a brief review of several problems involving systems with strangeness is made . in the first place one
shows how the @xmath0 , @xmath1 and @xmath2 states , for @xmath3 , and the @xmath4 for @xmath5 are generated dynamically in the context of unitarized chiral perturbation theory .
the results for the @xmath6 interaction are then used to evaluate the @xmath7 scattering length .
results obtained for the kaon selfenergy in a nuclear medium within this approach , with application to @xmath8 atoms , are also mentioned .
finally a few words are said about recent developments in the weak decay of @xmath9 hypernuclei and the puzzle of the @xmath10 ratio . |
many works in signal / image processing are concerned with data restoration problems . for such problems ,
the original data @xmath0 is degraded by a stable convolutive operator @xmath1 and by a non - necessarily additive noise .
denotes the space of discrete - time real - valued signals defined on @xmath2 having a finite energy . ]
the resulting observation model can be written as @xmath3 where @xmath4 denotes the noise effect and @xmath5 is some related parameter ( for example , @xmath6 may represent the variance for gaussian noise or the scaling parameter for poisson noise ) . in this context , our objective is to recover a signal @xmath7 , the closest possible to @xmath8 , from the observation vector @xmath9 assumed to belong to @xmath10 and available prior information ( sparsity , positivity , ) . in early works ,
this problem was solved , mainly for gaussian noise , by using wiener filtering , or equivalently quadratic regularization techniques .
later , multiresolution analyses were used for denoising by applying a thresholding to the generated coefficients @xcite .
then , in order to improve the denoising performance , redundant frame representations were substituted for wavelet bases @xcite . in @xcite
, authors considered convex optimization techniques to jointly address the effects of a noise and of a linear degradation within a convex variational framework . when the noise is gaussian , the forward - backward ( fb ) algorithm @xcite ( also known as thresholded landweber algorithm when the regularization term is an @xmath11-norm @xcite ) and its extensions @xcite can be employed in the context of wavelet basis decompositions and its use
can be extended to arbitrary frame representations @xcite . however , in the context of a non - additive noise such as a poisson noise or a laplace noise , fb algorithm is no longer applicable due to the non - lipschitz differentiability of the data fidelity term .
other convex optimization algorithms must be employed such as the douglas - rachford ( dr ) algorithm @xcite , the parallel proximal algorithm ( ppxa ) @xcite or the alternating direction method of multipliers ( admm ) @xcite .
these algorithms belong to the class of proximal algorithms and , for tractability issues , they often require to use tight frame representation for which closed forms of the involved proximity operators can be derived @xcite .
the goal of this paper is to propose a way to relax the tight frame requirement by considering an appropriate class of frame representations . in the following ,
we consider two general convex minimization problems , which are useful to solve frame - based restoration problems formulated under a synthesis form ( sf ) or an analysis form ( af ) .
the sf can be expressed as : @xmath12 and the af is : @xmath13 @xmath14 ( resp .
@xmath15 ) denotes the frame analysis ( resp .
synthesis ) operator . for every @xmath16 , @xmath17-\infty,+\infty\right]}}$ ] is a convex , lower semicontinuous , and proper function , @xmath18 is a stable convolutive operator , and for every @xmath19 , @xmath20-\infty,+\infty\right]}}$ ] is a convex , lower semicontinuous , and proper function . in several works , sf has been preferred since af appears to be more difficult to solve numerically @xcite . in the proposed framework ,
both approaches have a similar complexity .
this paper is organized as follows : in section [ sec : frames ] , the class of frames considered in this work is defined and their connections with filter bank structures is emphasized . in section [ sec : prox_algo ] , we show how these ( non necessarily tight ) frames can be combined with parallel proximal algorithms in order to solve problems and .
the proposed approach is also applicable to related augmented lagrangian approaches .
finally , restoration results are provided in section [ sec : results ] for scenarios involving poisson noise or laplace noise by using dual - tree transforms ( dtt ) and filter bank representations .
* notation : * throughout this paper , @xmath21 designates the class of lower semicontinuous convex functions @xmath22 defined on a real hilbert space @xmath23 and taking their values in @xmath24-\infty,+\infty\right]}}$ ] , which are proper in the sense that their domain @xmath25 is nonempty .
if @xmath26 has a unique minimizer , it is denoted by @xmath27 .
the interior of a set @xmath28 is denoted by @xmath29 .
physical properties of the target signal @xmath30 , such as sparsity or spatial regularity , may be suitably expressed in terms of its coefficients @xmath31 where @xmath32 and @xmath33 denotes a dictionary of signals in @xmath10 .
such a dictionary constitutes a frame if there exist two constants @xmath34 and @xmath35 in @xmath240,+\infty\right[}}$ ] such that @xmath36 the associated frame operator is the injective linear operator defined as @xmath37 the adjoint of which is the surjective linear operator given by @xmath38 when @xmath39 , an orthonormal basis is obtained .
further constructions as well as a detailed account of frame theory in hilbert spaces can be found in @xcite .
a tight frame is such that , for some @xmath400,+\infty\right[}}$ ] , @xmath41 where @xmath42 denotes the identity operator . in condition
, this means that @xmath43 .
a simple example of a tight frame is the union of @xmath44 orthonormal bases .
other examples of tight frames can be found in @xcite .
we consider a linear operator @xmath45 which is basically obtained by cascading a non necessarily maximally decimated filter bank and a semi - orthogonal transform .
a linear operator @xmath46 , with @xmath47 and @xmath48 , is said to be semi - orthogonal if there exists @xmath490,+\infty\right[}}$ ] such that @xmath50 .
recall that an analysis filter bank can be put under its polyphase form by performing a polyphase decomposition @xmath51 followed by a real mimo ( multi - input multi - output ) filtering @xmath52 @xcite : + the polyphase decomposition @xmath51 is an operator from @xmath10 to @xmath53 with @xmath54 such that , for every @xmath55 , @xmath56 where @xmath57 is the @xmath58-th polyphase component of order @xmath59 of the signal @xmath60 .
the adjoint operator of @xmath51 is given by @xmath61 where , for every @xmath62 and @xmath63 , @xmath64 .
so , @xmath65 allows us to concatenate @xmath59 square summable sequences into a single one .
it can be noticed that @xmath66 and @xmath67 , which means that @xmath51 is an isometry and @xmath68 .
+ the mimo filter @xmath52 is defined as where , for every @xmath69 and @xmath70 , @xmath71 is a siso ( single - input single - output ) stable filter .
hence , the impulse response of this filter belongs to @xmath72 and its frequency response @xmath73 is a continuous function .
in addition , it is assumed that @xmath52 is left invertible , that is : for every @xmath74 $ ] , the rank of the matrix @xmath75_{1\le i \le n,1\le j \le d}$ ] is equal to @xmath59 .
the adjoint operator of @xmath52 is the @xmath76 mimo filter given by where , for every @xmath69 and @xmath70 , @xmath77 is the siso filter with complex conjugate frequency response @xmath78 .
+ we have then the following result ( the proof is provided in appendix [ ap : f ] ) : [ prop : f ] the operator @xmath79 is a frame operator with frame constants @xmath80@xmath81 } \sigma_{\rm min}(\nu)$ ] and @xmath82 } \sigma_{\rm max}(\nu)$ ] , where , for every @xmath83 , @xmath840,+\infty\right[}}$ ] and @xmath850,+\infty\right[}}$ ] are the minimum and maximum eigenvalues of @xmath86 .
corresponds to the transconjugate of @xmath87 . ] in addition , we have : @xmath88 the resulting frame is not necessarily tight
. however , when @xmath89)$ ] @xmath90 , the frame is tight .
this includes paraunitary systems @xmath52 as particular cases when @xmath91 .
below , we provide examples of popular transforms in signal processing which belong to the class of considered frame representations . in order to obtain low redundancy representations , frames such as the 2d @xmath92-band dtt
have been proposed @xcite . the real ( resp .
complex ) dtt consists of performing @xmath93 ( resp .
@xmath94 ) @xmath92-band orthonormal wavelet decompositions in parallel where , for every @xmath95 , @xmath96 denotes the @xmath97-th orthogonal wavelet transform .
an orthogonal combination of the subbands modeled by @xmath98 is applied to ensure directionality properties . @xmath99 and @xmath100 are related to @xmath101 by the following relation : since @xmath102 and , for every @xmath69 , @xmath103 , @xmath101 is an orthogonal transform . in the general form of dtt
, each orthonormal wavelet decomposition is preceded by a prefiltering stage related to the discretization process .
more precisely , it takes the form : the variables @xmath104 and @xmath59 as defined before are thus equal to @xmath105 and @xmath106 , respectively .
the left invertibility condition here reduces to the fact that the frequency response @xmath107 does not vanish .
note that , due to the presence of prefilters , the _ discrete _ dtt is not a tight frame in general .
furthermore , an extended class of dtts can be obtained by using tight frame overcomplete representations @xmath108 having the same frame constant @xmath109 ( e.g. redundant wavelet representations derived from orthogonal filters after appropriate renormalization ) , in which case @xmath101 as defined by is a semi - orthogonal transform .
analysis filter banks with perfect reconstruction correspond to the case when @xmath110 and @xmath111 .
@xmath59 is the decimation factor and @xmath105 is the number of channels .
the redundancy introduced by such a filter bank structure is @xmath112 .
more details about filter bank design can be found in @xcite . lapped transforms @xcite
can also be implemented with filter bank structures .
as already mentioned , filter banks do not constitute tight frames in general . note that if @xmath113 , a fully undecimated filter bank is obtained .
tight frame representations have been widely used for data recovery by using convex optimization methods . in the next section ,
we recall some convex optimization tools and show the relevance of the considered class of frames in recent optimization approaches .
a number of algorithms such as the alternating split bregman algorithm @xcite , augmented lagrangian techniques @xcite and parallel proximal methods @xcite have been recently proposed to address possibly nonsmooth convex optimization problems encountered in the solution of restoration problems . here
, we will focus on this specific class of algorithms which have proven to be useful for solving problems like and when a tight frame representation is employed .
a common feature of the aforementioned algorithms is that they require a large - size linear system to be solved at each iteration .
when non tight frame representations are used , the computational cost of the associated inversion may become prohibitive .
we will see however that for the class of frames introduced in section [ sec : frames ] , this inversion can be performed in an efficient manner .
this fact will be demonstrated by considering an extension of the parallel proximal algorithm ( ppxa ) , hereafter designated as ppxa+ . as shown in @xcite , ppxa+
is closely related to the alternating direction method of multipliers and its parallel extensions @xcite .
we recall that the proximity operator @xcite of a function @xmath114 is defined as @xmath115 it can be observed that when @xmath22 is the indicator function @xmath116 of a nonempty closed convex subset @xmath28 of @xmath23 , i.e. it takes on the value @xmath117 in @xmath28 and @xmath118 in @xmath119 , @xmath120 reduces to the projector @xmath121 onto @xmath28 .
other examples of proximity operators corresponding to potential functions of standard log - concave univariate probability densities have been listed in @xcite .
the proximity operators employed in the experimental part of this paper , are recalled below .
( soft - thresholding rule ) + [ ex : gg ] let @xmath5 , and set @xmath122 . then , for every @xmath123 , @xmath124 .
( poisson minus log - likelihood function ) @xcite + [ ex : gamd ] let @xmath5 , @xmath125 , and set then , for every @xmath123 , @xmath126 one of the difficulties in the resolution of problems like or is that they involve linear operators .
unfortunately , the proximity operator of the composition of a linear operator and a convex function takes a closed form expression only under restrictive assumptions as stated below : @xcite + [ p : linprox ] let @xmath127 be a real hilbert space , @xmath128 and let @xmath129 denote a bounded linear operator .
suppose that @xmath130 , for some @xmath1310,+\infty\right[}}$ ] .
then , @xmath132 and @xmath133 .
when dealing with linear operators @xmath134 such that @xmath135 for any @xmath136 , proximal algorithms requiring the inversion of a linear operator at each iteration can however be designed .
for example , algorithm [ algo : sa ] ( resp .
algorithm [ algo : aa ] ) can be applied to problem ( resp .
problem ) .
( in these algorithms , the sequences @xmath137 and @xmath138 model possible numerical errors in the computation of the proximity operators at iteration @xmath139 .
) initialization + @xmath1400,+\infty\right[}}^r,(\kappa_s)_{1\le s \le s}\in { \ensuremath{\left]0,+\infty\right[}}^s ; \ ; ( v_{r,0})_{1\leq r\leq r } \in \big(\boldsymbol{\ell}^2({\ensuremath{\mathbb z}})\big)^r,(w_{s,0})_{1\leq s\leq s } \in
\big(\boldsymbol{\ell}^2({\ensuremath{\mathbb z}})\big)^s \\
x_0 = \arg \min_{u\in \boldsymbol{\ell}^2({\ensuremath{\mathbb z } } ) } \sum_{r=1}^r \eta_r { \vert l_r f^ { * } u - v_{r,0}\vert}^2 + \sum_{s=1}^s \kappa_s { \vert u - w_{s,0}\vert}^2\\ \end{array } \right.$ ] for @xmath141 + @xmath1420,2[\\ c_{\ell } = \arg \min_{u\in \boldsymbol{\ell}^2({\ensuremath{\mathbb z } } ) } \sum_{r=1}^r \eta_r { \vert l_r f^ { * } u - p_{r,\ell}\vert}^2 + \sum_{s=1}^s \kappa_s { \vert u - r_{s,\ell}\vert}^2\\ \mbox{for } \;\ ; r=1,\ldots , r \qquad\qquad
v_{r,\ell+1 } = v_{r,\ell } + \lambda_\ell \big ( l_r f^ { * } ( 2c_{\ell } - x_{\ell } ) - p_{r,\ell}\big)\\ \mbox{for } \;\ ; s=1,\ldots , s \qquad\qquad
w_{s,\ell+1 } = w_{s,\ell } + \lambda_\ell \big ( 2c_{\ell } - x_{\ell } - r_{s,\ell}\big)\\ x_{\ell+1 } = x_{\ell } + \lambda_{\ell}(c_{\ell } - x_{\ell})\\ \end{array } \right.$ ] initialization + @xmath1400,+\infty\right[}}^r,(\kappa_s)_{1\le s \le s}\in { \ensuremath{\left]0,+\infty\right[}}^s;\ ; ( v_{r,0})_{1\leq r\leq r } \in \big(\boldsymbol{\ell}^2({\ensuremath{\mathbb z}})\big)^r,(w_{s,0})_{1\leq s\leq s } \in
\big(\boldsymbol{\ell}^2({\ensuremath{\mathbb z}})\big)^s \\
y_0 = \arg \min_{u\in \boldsymbol{\ell}^2({\ensuremath{\mathbb z } } ) } \sum_{r=1}^r \eta_r { \vert l_r u - v_{r,0}\vert}^2 + \sum_{s=1}^s \kappa_s { \vert f u - w_{s,0}\vert}^2\\ \end{array } \right.$ ] for @xmath141 + @xmath1420,2[\\ c_{\ell } = \arg \min_{u\in \boldsymbol{\ell}^2({\ensuremath{\mathbb z } } ) } \sum_{r=1}^r \eta_r { \vert l_r u - p_{r,\ell}\vert}^2 + \sum_{s=1}^s \kappa_s { \vert f u - r_{s,\ell}\vert}^2\\ \mbox{for } \;\ ; r=1,\ldots , r \qquad\qquad v_{r,\ell+1 } = v_{r,\ell } + \lambda_{\ell } \big ( l_r ( 2c_{\ell } - y_{\ell } ) - p_{r,\ell}\big)\\ \mbox{for } \;\ ; s=1,\ldots , s \qquad\qquad
w_{s,\ell+1 } = w_{s,\ell } + \lambda_{\ell } \big ( f ( 2c_{\ell } - y_{\ell } ) - r_{s,\ell}\big)\\ y_{\ell+1 } = y_{\ell } + \lambda_{\ell}(c_{\ell } - y_{\ell})\\ \end{array } \right.$ ] the convergence of the sequence @xmath143 ( resp .
@xmath144 ) generated by algorithm [ algo : sa ] ( resp .
algorithm [ algo : aa ] ) to an optimal solution of the related optimization problem is guaranteed under the following technical assumptions ( see @xcite for more details ) : 1 . 2 .
@xmath145 3 .
@xmath146 ) .
4 . there exists @xmath1470,2[$ ] such that @xmath148 .
@xmath149 @xmath150 and @xmath151 @xmath152 .
note that quadratic minimizations need to be performed in the initialization step and in the computation of the intermediate variable @xmath153 at iteration @xmath139 .
these amount to inverting a large - size linear operator which is untractable for arbitrary frames .
we will now show that the class of frame operators @xmath45 introduced in section [ ssec : ntf ] allows us to overcome this difficulty . to do so , recall that the convolutive operators @xmath154 can be put under a polyphase form @xcite : @xmath155 where the polyphase decomposition operator @xmath51 has been defined in section [ ssec : ntf ] and @xmath156 $ ] is a miso ( multi - input single - output ) filter ( for every @xmath63 , @xmath157 is a siso filter ) .
now , invoking proposition [ prop : f ] and making use of sherman - morrison - woodbury identity yield : + for sf ( algorithm [ algo : sa ] ) @xmath158 for af ( algorithm [ algo : aa ] ) @xmath159 where @xmath160 .
note that the idea of using the woodbury matrix identity to handle the inversion for sf was proposed in @xcite in the context of tight frames .
the inversions in ( resp . ) can be performed by noticing that , as @xmath161 and @xmath52 are multivariate filters with frequency responses @xmath162 and @xmath163 , @xmath164 ( resp .
@xmath165 ) is a mimo filter with frequency response : for every @xmath74 $ ] , @xmath166 hence , by resorting to fast fourier transform implementations , the problem reduces to the inversion of @xmath167 matrices .
we apply the proposed optimization method to the restoration of images degraded by a blur and a poisson ( resp .
laplace ) noise . for simplicity sake
, the proposed formulation has been presented in the 1d case , but its 2d extension is straightforward .
the scaling factor of the poisson ( resp .
laplace ) noise is @xmath168 ( resp .
@xmath169 ) .
the data fidelity term is related to the poisson ( resp .
laplace ) minus log - likelihood , which corresponds to the potential function @xmath22 defined by ( resp . by example [ ex : gg ] ) . in the considered examples ,
the regularization term simply is an @xmath11-norm modeling the sparsity of @xmath8 through a frame representation .
two different frames are considered : a ( complex ) dual - tree transform , presented in section [ sss : dtt ] , and an eigenfilter bank decomposition @xcite , which is a special case of the subband structure presented in section [ sss : fb ] .
we constrain the data values to belong to @xmath170 $ ] , so defining a closed convex constraint set @xmath28 .
the considered sf ( resp .
af ) problem is a particular case of problem ( resp . )
where @xmath171 , @xmath172 , @xmath173 , @xmath174 , and @xmath175 with @xmath176 .
the last function corresponds to the regularization term operating in the frame domain . in the considered problems , @xmath177 and @xmath178
the proximity operators associated to @xmath179 , @xmath180 , and @xmath181 are derived from example [ ex : gamd ] , the projection onto @xmath28 , and example [ ex : gg ] . in our simulations , the parameter @xmath182 is empirically chosen to maximize the signal - to - noise - ratio ( snr ) .
in general , the value of this parameter is not the same for fa and fs .
note that an alternative to the proposed approach consists of resorting to primal - dual algorithms @xcite .
these algorithms are appealing as they do not require any operator inversion and they can thus be employed with arbitrary frames .
however , after appropriate choices for the weights @xmath183 , @xmath184 , and the relaxation parameter @xmath185 , the proposed approach appeared to be faster .
for example , similar frequency domain implementations of the monotone + skew forward backward forward ( m+sfbf ) algorithm @xcite were observed to be about twenty times slower than the proposed method in term of iterations and computation time .
figure [ fig : tf_ntf ] shows the restoration results for a cropped version of the `` barbara '' image in the presence of poisson noise and a uniform blur of size @xmath186 .
we adopt a sf criterion and we consider a tight version ( i.e. , for every @xmath187 , @xmath188 ) as well as a non - tight version of complex dtt . the complex dtt @xcite is computed using symlets of length 6 over 3 resolution levels . in order to efficiently perform the inversions in and , fast discrete fourier diagonalization techniques have been employed .
the use of the non - tight complex dtt including prefilters allows us to improve the quality of the results both visually and in terms of snr and ssim @xcite .
figure [ fig : sa_aa ] displays a second restoration example for a cropped version of the `` straw '' image in the presence of laplace noise and a uniform blur of size @xmath189 .
af results are presented by using a dtt and an eigenfilter bank ( @xmath190 and @xmath191 ) computed from the degraded image .
this formulation leads to better results than those obtained with sf .
significant gains in favour of the eigenfilter bank can be observed .
[ cols="^,^,^,^ " , ] | a fruitful approach for solving signal deconvolution problems consists of resorting to a frame - based convex variational formulation . in this context ,
parallel proximal algorithms and related alternating direction methods of multipliers have become popular optimization techniques to approximate iteratively the desired solution . until now , in most of these methods , either lipschitz differentiability properties or tight frame representations were assumed . in this paper , it is shown that it is possible to relax these assumptions by considering a class of non necessarily tight frame representations , thus offering the possibility of addressing a broader class of signal restoration problems . in particular , it is possible to use non necessarily maximally decimated filter banks with perfect reconstruction , which are common tools in digital signal processing .
the proposed approach allows us to solve both frame analysis and frame synthesis problems for various noise distributions . in our simulations , it is applied to the deconvolution of data corrupted with poisson noise or laplacian noise by using ( non - tight ) discrete dual - tree wavelet representations and filter bank structures . |
a famous chemist p. flory @xcite proposed to consider self - avoiding ( _ i.e. _ visiting every vertex at most once ) walks on a lattice as a model for polymer chains .
self - avoiding walks turned out to be a very interesting object , leading to rich mathematical theories and challenging questions , see @xcite .
denote by @xmath2 the number of @xmath3-step self - avoiding walks on the hexagonal lattice @xmath4 started from some fixed vertex , _
e.g. _ the origin .
elementary bounds on @xmath2 ( for instance @xmath5 ) guarantee that @xmath2 grows exponentially fast .
since a @xmath6-step self - avoiding walk can be uniquely cut into a @xmath3-step self - avoiding walk and a parallel translation of a @xmath7-step self - avoiding walk , we infer that @xmath8 from which it follows that there exists @xmath9 such that @xmath10 the positive real number @xmath11 is called the _ connective constant _ of the hexagonal lattice . using coulomb gas formalism , b. nienhuis @xcite proposed physical arguments for @xmath11 to have the value @xmath12 .
we rigorously prove this statement . while our methods are different from those applied by nienhuis , they are similarly motivated by considerations of vertex operators in the @xmath13 model .
our methods do not directly apply to the square lattice , for which the value of the connective constant is different and currently unknown . [
theorem ] for the hexagonal lattice , @xmath14 it will be convenient to consider walks between _ mid - edges _ of @xmath4 , _
i.e. _ centers of edges of @xmath4 ( the set of mid - edges will be denoted by @xmath15 )
. we will write @xmath16 if a walk @xmath17 starts at @xmath18 and ends at some mid - edge of @xmath19 . in the case
@xmath20 , we simply write @xmath21 .
the _ length _ @xmath22 of the walk is the number of vertices visited by @xmath17 .
we will work with the partition function @xmath23.\ ] ] this sum does not depend on the choice of @xmath18 , and is increasing in @xmath24 . establishing the identity
@xmath25 is equivalent to showing that @xmath26 for @xmath27 and @xmath28 for @xmath29 . to this end , we analyze walks restricted to bounded domains and weighted depending on their winding .
the modified sum can be defined as a _
parafermionic observable _ arising from a disorder operator .
such observables exist for other models , see @xcite .
the paper is organized as follows . in section 2 ,
the parafermionic observable is introduced and its key property is derived .
section 3 contains the proof of theorem [ theorem ] .
section 4 discusses conformal invariance conjectures for self - avoiding walks . to simplify formul , below we set @xmath30 and @xmath31
a ( hexagonal lattice ) _ domain _ @xmath32 is a union of all mid - edges emanating from a given collection of vertices @xmath33 ( see fig . [
fig : sawpicture ] ) : a mid - edge @xmath34 belongs to @xmath35 if at least one end - point of its associated edge is in @xmath35 , it belongs to @xmath36 if only one of them is in @xmath35 .
we further assume @xmath35 to be simply connected , _ i.e. _ having a connected complement . with boundary mid - edges labeled by small black squares , and vertices of @xmath33 labeled by circles .
* right . *
winding of a curve @xmath17 . ] for a self - avoiding walk @xmath17 between mid - edges @xmath18 and @xmath37 ( not necessarily the start and the end ) , we define its _ winding @xmath38 _ as the total rotation of the direction in radians when @xmath17 is traversed from @xmath18 to @xmath37 , see fig . [
fig : sawpicture ] .
our main tool is given by the following the _ parafermionic observable _ for @xmath39 , @xmath40 , is defined by @xmath41 [ lem : relation ] if @xmath42 and @xmath43 , then @xmath44 satisfies the following relation for every vertex @xmath45 : @xmath46 where @xmath47 are the mid - edges of the three edges adjacent to @xmath48 .
note that with @xmath49 , the complex weight @xmath50 can be interpreted as a product of terms @xmath51 or @xmath52 per left or right turn of @xmath17 drawn from @xmath18 to @xmath34 , with @xmath53 we start by choosing notation so that @xmath54 , @xmath55 and @xmath56 follow counter - clockwise around @xmath48 .
note that the left - hand side of can be expanded into the sum of contributions @xmath57 of all possible walks @xmath17 finishing at @xmath58 or @xmath56 .
for instance , if a walk ends at the mid - edge @xmath54 , its contribution will be given by @xmath59 one can partition the set of walks @xmath17 finishing at @xmath58 or @xmath56 into pairs and triplets of walks in the following way , see fig [ fig : pairs ] : * if a walk @xmath60 visits all three mid - edges @xmath47 , it means that the edges belonging to @xmath60 form a disjoint self - avoiding path plus ( up to a half - edge ) a self - avoiding loop from @xmath48 to @xmath48 .
one can associate to @xmath60 the walk passing through the same edges , but exploring the loop from @xmath48 to @xmath48 in the other direction .
hence , walks visiting the three mid - edges can be grouped in pairs . *
if a walk @xmath60 visits only one mid - edge , it can be associated to two walks @xmath61 and @xmath62 that visit exactly two mid - edges by prolonging the walk one step further ( there are two possible choices ) .
the reverse is true : a walk visiting exactly two mid - edges is naturally associated to a walk visiting only one mid - edge by erasing the last step . hence , walks visiting one or two mid - edges can be grouped in triplets .
if one can prove that the sum of contributions to of each pair or triplet vanishes , then their total sum is zero , and holds .
let @xmath60 and @xmath61 be two associated walks as in the first case . without loss of generality , we may assume that @xmath60 ends at @xmath55 and @xmath61 ends at @xmath56 .
note that @xmath60 and @xmath61 coincide up to the mid - edge @xmath54 , and then follow an almost complete loop in two opposite directions .
it follows that @xmath63 in order to evaluate the winding of @xmath60 between @xmath54 and @xmath55 above , we used the fact that @xmath18 is on the boundary and @xmath35 is simply connected .
we conclude that @xmath64 where the last equality holds since @xmath65 by our choice of @xmath66 .
let @xmath67 be three walks matched as in the second case .
without loss of generality , we assume that @xmath60 ends at @xmath54 and that @xmath61 and @xmath62 extend @xmath60 to @xmath55 and @xmath56 respectively .
as before , we easily find that @xmath68 plugging these values into the respective contributions , we obtain @xmath69 above is the _ only _ place where we use that @xmath24 takes its critical value , i.e. @xmath70 .
the claim of the lemma follows readily by summing over all pairs and triplets . and differing by rearranged connections at @xmath48 .
* right : * a triplet of walks , one visiting one mid - edge , the two others visiting two mid - edges , and obtained by prolonging the first one through @xmath48 . ]
[ rem : cr ] coefficients in are three cube roots of unity multiplied by @xmath71 , so its left - hand side can be seen as a discrete @xmath72-integral along an elementary contour on the dual lattice .
the fact that the integral of the parafermionic observable along discrete contours vanishes suggests that it is discrete holomorphic and that self - avoiding walks have a conformally invariant scaling limit , see section 4 .
[ [ counting - argument - in - a - strip - domain .
] ] counting argument in a strip domain .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we consider a vertical strip domain @xmath73 composed of @xmath74 strips of hexagons , and its finite version @xmath75 cut at heights @xmath76 at angles @xmath77 , see fig .
[ fig : domain ] .
namely , position a hexagonal lattice @xmath4 of meshsize 1 in @xmath78 so that there exists a horizontal edge @xmath79 with mid - edge @xmath18 being 0
. then @xmath80 denote by @xmath81 the left boundary of @xmath82 , by @xmath83 the right one .
symbols @xmath84 and @xmath85 denote the top and bottom boundaries of @xmath75 .
introduce the following ( positive ) partition functions : @xmath86 in the next lemma , we deduce from relation a global identity without the complex weights . and
boundary intervals @xmath81 , @xmath83 , @xmath84 and @xmath85 . ] for critical @xmath42 , the following identity holds @xmath87 with positive coefficients @xmath88 and @xmath89 . sum the relation over all vertices in @xmath90 .
values at interior mid - edges disappear and we arrive at the identity @xmath91 the symmetry of our domain implies that @xmath92 , where @xmath93 denotes the complex conjugate of @xmath24 . observe that the winding of any self - avoiding walk from @xmath18 to the bottom part of @xmath81 is @xmath94 while the winding to the top part is @xmath95
thus @xmath96 above
we have used the fact that the only walk from @xmath18 to @xmath18 is a trivial one of length @xmath97 , and so @xmath98 .
similarly , the winding from @xmath18 to any half - edge in @xmath83 ( resp . @xmath84 and
@xmath85 ) is 0 ( resp . @xmath99 and @xmath100 )
, therefore @xmath101 the lemma follows readily by plugging the last three formul into .
observe that sequences @xmath102 and @xmath103 are increasing in @xmath104 and are bounded for @xmath105 thanks to and their monotonicity in @xmath24 .
thus they have limits @xmath106 identity then implies that @xmath107 decreases and converges to a limit @xmath108 . passing to a limit in
, we arrive at @xmath109 we start by proving that @xmath110 , and hence @xmath111 .
suppose that for some @xmath74 , @xmath112 . as noted before , @xmath113 decreases in @xmath104 and so @xmath114 which completes the proof .
assuming on the contrary that @xmath115 for all @xmath74 , we simplify to @xmath116 observe that a walk @xmath117 entering into the count of @xmath118 and not into @xmath119 has to visit some vertex adjacent to the right edge of @xmath120 . cutting @xmath117 at the first such point ( and adding half - edges to the two halves ) , we uniquely decompose it into two walks crossing @xmath120 ( these walks are usually called bridges ) , which together are one step longer than @xmath117 .
we conclude that @xmath121 combining for two consecutive values of @xmath74 with , we can write @xmath122 and so @xmath123 it follows easily by induction , that @xmath124}~/~{t}\ ] ] for every @xmath125 , and therefore @xmath126 this completes the proof of the estimate @xmath127 .
it remains to prove the opposite inequality @xmath128 . to estimate the partition function from above
, we will decompose self - avoiding walks into bridges .
bridge _ of width @xmath74 is a self - avoiding walk in @xmath73 from one side to the opposite side , defined up to vertical translation .
the partition function of bridges of width @xmath74 is @xmath129 , which is at most @xmath130 by . noting that a bridge of width @xmath74 has length at least @xmath74
, we obtain for @xmath131 @xmath132 thus , for @xmath131 , the series @xmath133 converges and so does the product @xmath134 .
let us assume for the moment the following fact : _ any self - avoiding walk can be canonically decomposed into a sequence of bridges of widths @xmath135 and @xmath136 , and , if one fixes the starting mid - edge and the first vertex visited , the decomposition uniquely determines the walk_. such decomposition was first introduced by hammersley and welsh in @xcite ( for a modern treatment , see section 3.1 of @xcite ) . applying the decomposition to walks starting at @xmath18 ( the first visited vertex is 0 or -1 )
, we can estimate @xmath137 the factor 2 is due to the fact that there are two possibilities for the first vertex once we fix the starting mid - edge .
therefore , @xmath28 whenever @xmath131 and @xmath138 . to complete the proof of the theorem it only remains to prove that such a decomposition into bridges does exist . once again , this fact is well - known @xcite , but we include the proof for completeness . .
the first bridge corresponds to the maximal bridge containing the origin .
note that the decomposition contains one bridge of width 0 .
* right : * the reverse procedure .
if the starting mid - edge and the first vertex are fixed , the decomposition is unambiguous . ]
first assume that @xmath139 is a half - plane self - avoiding walk , meaning that the start of @xmath139 has extremal real part : we prove by induction on the width @xmath140 that the walk admits a canonical decomposition into bridges of widths @xmath136 .
without loss of generality , we assume that the start has minimal real part
. out of the vertices having the maximal real part , choose the one visited last , say after @xmath3 steps .
the @xmath3 first vertices of the walk form a bridge @xmath141 of width @xmath140 , which is the first bridge of our decomposition when prolonged to the mid - edge on the right of the last vertex .
we forget about the @xmath142-th vertex , since there is no ambiguity in its position .
the consequent steps form a half - plane walk @xmath143 of width @xmath144 . using the induction hypothesis , we know that @xmath143 admits a decomposition into bridges of widths @xmath145 .
the decomposition of @xmath139 is created by adding @xmath141 before the decomposition of @xmath143 .
if the walk is a reverse half - plane self - avoiding walk , meaning that the end has extremal real part , we set the decomposition to be the decomposition of the reverse walk in the reverse order .
if @xmath17 is a self - avoiding walk in the plane , one can cut the trajectory into two pieces @xmath60 and @xmath61 : the vertices of @xmath17 up to the first vertex of maximal real part , and the remaining vertices .
the decomposition of @xmath17 is given by the decomposition of @xmath60 ( with widths @xmath146 ) plus the decomposition of @xmath61 ( with widths @xmath147 ) .
once the starting mid - edge and the first vertex are given , it is easy to check that the decomposition uniquely determines the walk by exhibiting the reverse procedure , see fig . [
fig : decomposition ] for the case of half - plane walks .
[ remark ] the proof provides bounds for the number of bridges from @xmath18 to the right side of the strip of width @xmath74 , namely , @xmath148 in paragraphs 3.3.3 and 3.4.3 of @xcite , precise behaviors are conjectured for the number of self - avoiding walks between two points on the boundary of a domain , which yields the following ( conjectured ) estimate : @xmath149 where @xmath15 is the boundary derivative of the poisson kernel . integrating with respect to @xmath150 , we obtain that @xmath151 should decay as @xmath152 when @xmath74 goes to infinity .
similar estimates are conjectured for walks in @xmath73 from @xmath97 to @xmath153 .
in @xcite , nienhuis proposed a more precise asymptotical behavior for the number of self - avoiding walks : @xmath154 with @xmath155 . here
the symbol @xmath156 means that the ratio of two sides is of the order @xmath157 , or perhaps even tends to a constant .
moreover , nienhuis gave arguments in support of flory s prediction that the mean - square displacement @xmath158 satisfies @xmath159 with @xmath160 . despite the precision of the predictions and , the best rigorously known bounds are very far apart and almost 50 years old ( see @xcite for an exposition ) .
the derivation of these exponents seems to be one of the most challenging problems in probability .
it was shown by g. lawler , o. schramm and w. werner in @xcite that @xmath117 and @xmath161 could be computed if the self - avoiding walk would posses a conformally invariant scaling limit .
more precisely , let @xmath162 be a simply connected domain in the complex plane @xmath163 with two points @xmath18 and @xmath37 on the boundary . for @xmath164 , we consider the discrete approximation given by the largest finite domain @xmath165 of @xmath166 included in @xmath35 , and @xmath167 and @xmath168 to be the vertices of @xmath165 closest to @xmath18 and @xmath37 respectively .
a probability measure @xmath169 is defined on the set of self - avoiding trajectories @xmath117 between @xmath167 and @xmath168 that remain in @xmath165 by assigning to @xmath117 a weight proportional to @xmath170 .
we obtain a random curve denoted @xmath171 .
conjectured conformal invariance of self - avoiding walks can be stated as follows , see @xcite : [ sle ] let @xmath35 be a simply connected domain ( not equal to @xmath163 ) with two distinct points @xmath18 , @xmath37 on its boundary . for @xmath42 ,
the law of @xmath171 in @xmath172 converges when @xmath173 to the ( chordal ) schramm - loewner evolution with parameter @xmath174 in @xmath35 from @xmath18 to @xmath37 .
thus it would suffice to show that a normalized version of @xmath175 has a conformally invariant scaling limit , which can be achieved by showing that it is holomorphic and has prescribed boundary values . as discussed in @xcite , the winding of an interface leading to a boundary edge @xmath34
is uniquely determined , and coincides with the winding of the boundary itself .
thus one can say that @xmath175 satisfies a discrete version of the following _ riemann boundary value problem _ ( a homogeneous version of the riemann - hilbert - privalov bvp ) : @xmath176 with a singularity at @xmath18 .
note that the problem above has conformally covariant solutions ( as @xmath177-forms ) , and so is well defined even in domains with fractal boundaries .
as noted in remark [ rem : cr ] , relation amounts to saying that discrete contour integrals of @xmath175 vanish .
so any ( subsequential ) scaling limit of @xmath175 would have to be holomorphic .
unfortunately , relation alone , is unsufficient to deduce the existence of such a limit , unlike in the ising case @xcite .
the reason is that for a domain with @xmath178 edges , imposes @xmath179 relations ( one per vertice ) for @xmath178 values of @xmath175 , making it impossible to reconstruct @xmath175 from its boundary values .
so @xmath175 is not exactly holomorphic , it can be rather thought of as a divergence - free vector field , which seems to have non - trivial curl .
however , we expect that in the limit the curl vanishes , which is equivalent to @xmath180 having the same limit regardless of the orientation of the edge @xmath34 .
let @xmath35 be a simply connected domain ( not equal to @xmath163 ) , let @xmath40 , and let @xmath18 , @xmath37 be two distinct points on the boundary of @xmath35 .
we assume that the boundary of @xmath35 is smooth near @xmath37 . for @xmath164 , let @xmath175 be the holomorphic observable in the domain @xmath181 approximating @xmath182 , and let @xmath183 be the closest point in @xmath165 to @xmath34 .
then the right - hand side of is well - defined , since the conformal map @xmath187 is unique up to multiplication by a real factor .
proving this conjecture would be a major step toward conjecture [ sle ] and the derivation of critical exponents .
the authors would like to thank g. slade for useful comments on the manuscript , and g. lawler for suggesting remark [ remark ] .
this research was supported by the eu marie - curie rtn cody , the erc ag confra , as well as by the swiss fns .
the second author was partially supported by the chebyshev laboratory ( department of mathematics and mechanics , st .- petersburg state university ) under rf governement grant 11.g34.31.0026 | we provide the first mathematical proof that the connective constant of the hexagonal lattice is equal to @xmath0 .
this value has been derived non rigorously by b. nienhuis in 1982 , using coulomb gas approach from theoretical physics .
our proof uses a parafermionic observable for the self avoiding walk , which satisfies a half of the discrete cauchy - riemann relations .
establishing the other half of the relations ( which conjecturally holds in the scaling limit ) would also imply convergence of the self - avoiding walk to sle(@xmath1 ) . |
although a spatially extended field may be smooth , and contain no infinities , it may nonetheless have point singularities associated with its topology .
for example , a 2-dimensional landscape , specified by a real function , has _
critical points _
( stationary points ) , where the gradient of the field vanishes and the gradient direction can not be defined ; in the neighbourhood of such a point , the gradient direction changes arbitrarily quickly through all of its values .
such points are characterised by a _
topological charge _ , a signed integer which is determined by the local geometry of the singularity ; in the case of critical points , the number is the ( signed ) number of rotations of the gradient vector around the critical point ; it is @xmath1 for maxima and minima , and @xmath2 for saddles .
the topological charge is also called the _ poincar index _ ( or _ hopf index _ ) ( milnor 1965 ) , and is defined as a signed integer at the point zeros of @xmath0-dimensional real vector fields in @xmath0 dimensions . although the number of zeros may change as a field evolves , the total topological charge is constant ; it is a topological invariant . here
, i discuss the statistical properties of various zeros / singularities , in fields which are specified by gaussian random functions . in this case , only structurally stable zeros of charge @xmath3 occur .
specifically , i shall discuss how densities and two point charge correlation functions of the distributions of signed singularities may be calculated under a rather general scheme , and then use this scheme to calculate the topological charge correlation functions of three types of singularity in isotropic random fields : zeros of @xmath0-dimensional vectors in @xmath0-dimensional space ( rederiving a result originally due to halperin ( 1981 ) ) ; critical points of random scalar fields in two dimensions ; and umbilic points of random scalar fields in two dimensions .
the charge correlation functions for each of these types of topological singularity are found to be different in each case , and dependent on the underlying correlation function of the gaussian field .
each function is found to satisfy a screening relation associated with ionic liquids .
topological zeros are very important in many areas of physics and mathematics : in addition to critical points ( gradient zeros ) which have obvious importance , the zeros of 2-dimensional complex scalar fields ( phase singularities , wave dislocations , vortices , realised as 2-dimensional vectors ) are also of great interest , especially where the field is a quantum wavefunction or an optical field ( e.g. nye and berry 1974 , berry and dennis 2000 , dennis 2001b ) , or an order parameter ( mermin 1979 ) .
only point zeros are considered here - the dimensionality @xmath0 of the vector field , whose zeros provide the singularity , is assumed equal to the dimensionality of the space . the first systematic study of the statistical geometry of random real scalar fields in two dimensions was by longuet - higgins ( 1957a , b , 1958 ) , who generalised one - dimensional methods of rice ( 1954 ) to calculate , amongst other things , the densities of critical points and probability density function of the gaussian curvature of the function .
halperin ( 1981 ) derived the @xmath0-dimensional vector correlation function , whose proof was supplied by liu and mazenko ( 1992 ) , and has recently been recast in the language of riemannian geometry by foltin ( 2003a ) .
various correlation functions in two dimensions , including those for phase singularities and critical points , were investigated numerically by freund and wilkinson ( 1998 ) .
planar phase singularity correlations ( including density correlations ) were investigated by berry and dennis ( 2000 ) .
the topological singularities of interest are the zeros of an @xmath0-dimensional vector field @xmath4 the field is a smooth function defined on an @xmath0-dimensional euclidean space , with points labelled by the vector @xmath5 in random fields , the only statistically significant zeros are those of first order , whose jacobian determinant @xmath6 is nonzero ( where @xmath7 ) . the topological charge of such zeros is given by @xmath8 when the field is random , the density @xmath9 of zeros at a position @xmath10 is @xmath11 where @xmath12 denotes averaging over the statistical ensemble , and the zeros are identified by the @xmath0-dimensional dirac @xmath13-function .
the modulus of the jacobian is included to ensure that each zero has the correct statistical weight , so @xmath9 has the units of density .
this expression confirms that degenerate zeros are so rare they have no statistical significance .
the average topological charge @xmath14 at @xmath10 is expressed as the density ( [ eq : densdef ] ) , but each singularity is weighted by its charge , that is , the sign of the jacobian .
the charge correlation functions @xmath15 with which this paper is concerned are the generalisation of the average charge density to that at two points @xmath16 normalised by the density : @xmath17 in the following section , the scheme for calculating topological charge correlation functions ( [ eq : gdef ] ) in general gaussian random fields is explained .
section [ sec : fields ] then provides background to gaussian fields which are statistically stationary and isotropic . the scheme is then applied to zeros of @xmath0-dimensional vector fields ( section [ sec : vec ] ) , critical points of 2-dimensional scalar fields ( section [ sec : crit ] ) , and umbilic points of the same fields ( section [ sec : umbilic ] ) .
the phenomenon of topological charge screening in these three cases is discussed in section [ sec : screening ] .
as this paper was being completed , i became aware of the work of foltin ( 2003b ) , who performs similar calculations for critical points by a different , possibly simpler method .
an ensemble of scalar functions @xmath18 in @xmath0-dimensional space is said to be a _
gaussian random function _
if the probability distribution of the function at each point of space is given by a gaussian distribution , so the function at each point defines a gaussian random variable ( adler 1981 ) .
the only restrictions on gaussian random functions made in this section are that they be centred ( the average @xmath19 ) and that first derivatives exist .
our starting point is the well - known expression for the probability density function for a set of @xmath20 independent gaussian random functions @xmath21 @xmath22 where @xmath23 is the _ correlation matrix _ , with components defined @xmath24 it is now possible to present the general scheme for calculating 2-point topological charge correlation functions of zeros of gaussian random functions , expressed in ( [ eq : gdef ] ) .
the topological charges considered are all zeros of a real vector gaussian random function @xmath25 of dimension @xmath26 and the correlation function is the average of the product of the local density at points labelled @xmath27 quantities evaluated at these places are denoted with the appropriate subscript @xmath28 or @xmath29 the variables that appear in the average ( [ eq : gdef ] ) are therefore the components of @xmath30 and their derivatives @xmath31 that appear in the jacobians @xmath32 there are a total of @xmath33 different derivatives @xmath34 appearing in each jacobian , and in general @xmath35 for example , in the case of critical points , @xmath36 , and @xmath37 ( the terms appearing in the jacobian are @xmath38 ) .
the calculation therefore involves an average of the @xmath39-dimensional gaussian random vector @xmath40 the correlation matrix @xmath23 for this @xmath41 is defined as in ( [ eq : cormatdef ] ) , and averages are evaluated according to the probability density function ( [ eq : gausspdf1 ] ) .
the correlation function in ( [ eq : gdef ] ) is therefore @xmath42 where @xmath43 are the appropriate densities of zeros ( [ eq : densdef ] ) .
the jacobians @xmath44 can be quite complicated ; each is a multilinear function , involving a sum of @xmath0-fold products of @xmath45 ( for @xmath28 ) or @xmath46 ( for @xmath47 ) .
the @xmath48-dimensional lower right submatrix of @xmath23 ( i.e. the averages dependent only on @xmath49 ) is denoted by @xmath50 the complementary @xmath51-dimensional upper left submatrix of @xmath52 is denoted by @xmath53 that is , @xmath54 where the other blocks marked @xmath55 are not denoted by special symbols . in [ sec : jacobi ] , jacobi s determinant theorem is used to show that @xmath56 the integrals in @xmath57 only involving @xmath13-functions , are performed , leaving the @xmath58 derivative terms @xmath59 to integrate ; let @xmath60 then @xmath61 upon fourier transforming the gaussian in @xmath62 with the @xmath58-dimensional fourier vector variable @xmath63 the jacobian terms @xmath64 ( depending on @xmath65 ) and @xmath66 ( depending on @xmath67 ) may be replaced with partial derivative terms in @xmath68 @xmath69 where @xmath70 is used to denote partial derivatives in @xmath58-dimensional @xmath71-space .
the jacobians , with this replacement , have now become differential operators , denoted @xmath72 therefore @xmath73 where , in the second line , the integral over @xmath62 is realised as the fourier transform of the @xmath13-function of @xmath74 and ( [ eq : jacobi ] ) has been used in the prefactor .
the expression is then integrated by parts ( so the @xmath75 operators act on the exponential term rather than the @xmath13-function ) , and then integrated in @xmath63 the final expression is @xmath76 where @xmath77_{\bi{t } = \mathbf{0}}. \label{eq : ddef}\ ] ] the charge correlation function therefore only depends on @xmath78 and components of the inverse reduced inverse correlation matrix @xmath79 this final step , evaluating @xmath80 in ( [ eq : ddef ] ) is the most complicated part of the calculation , and depends on the precise form of the jacobian determinant @xmath81 each summand in the operator @xmath82 is a @xmath83-fold derivative over some index set @xmath84 and is comprised of @xmath0 terms from @xmath85 ( for @xmath28 ) , and @xmath0 terms from @xmath86 ( for @xmath47 ) , either set possibly including repetitions .
the result of this particular operation is @xmath87_{\bi{t } = \mathbf{0 } } = \sum^{{\rm all\ , pairings}}_{p } \prod_{(\tau_{\lambda},\tau_{\lambda ' } ) \in p } \xi_{\tau_{\lambda } \tau_{\lambda ' } } , \label{eq : dexp}\ ] ] where the sum on the right hand side is over all pairings @xmath88 of indices in @xmath89 there are @xmath90 such pairings , each involving @xmath0 sets of pairs .
the @xmath0-fold product on the right hand side is over all components of @xmath91 with indices given by the appropriate pair .
the @xmath92 themselves are found by further application of jacobi s determinant theorem in [ sec : jacobi ] , and are expressed in terms of minors of @xmath93 in ( [ eq : xiij ] ) .
the correlation functions calculated explicitly in this paper are not too complicated , either because @xmath0 is small ( only two - dimensional fields are considered in sections [ sec : crit ] , [ sec : umbilic ] ) , or @xmath91 is sparse , as in the case of random vectors ( section [ sec : vec ] ) .
the density @xmath94 of zeros appears in ( [ eq : gabfin ] ) ; in general , this may be difficult to calculate , due to the modulus sign in ( [ eq : densdef ] ) .
for random @xmath0-dimensional vector fields , the main part of the density calculation is in [ sec : dveccalc ] . for critical and umbilic points ,
these densities were calculated by longuet - higgins ( 1957a , b ) and berry and hannay ( 1977 ) .
the topological charge correlation function in ( [ eq : gabfin ] ) is extremely general , applying to any centred differentiable gaussian random vector field . in this section , and for the remainder of the article , attention will be restricted to stationary isotropic random fields . for these fields ,
all averages are ( statistically ) translation and rotation invariant .
they are conveniently given by a fourier representation @xmath95 where @xmath96 are now the fourier variable vectors ( wavevectors ) .
the components @xmath97 of random vectors are specified by independent identically distributed realisations of ( [ eq : ffourier ] ) .
the amplitude @xmath98 only depends on the magnitude @xmath99 and the phase @xmath100 is uniformly random - each ensemble member is therefore labelled by the choice of @xmath100 for each @xmath101 it also may represent the spatial part of a real linear homogeneous nondispersive wavefield , for which the representation ( [ eq : ffourier ] ) is particularly evocative .
the infinite @xmath96 set is assumed sufficiently dense that they may be represented as an integral , and @xmath102 where @xmath103 is the power spectrum of the field ; by the wiener - khinchine theorem ( feller 1950 ) , @xmath103 is the @xmath0-dimensional fourier transform of the _ field correlation function _
@xmath104 where @xmath105 and @xmath106 normalised such that @xmath107 the only condition on @xmath108 is that it is symmetric and has positive fourier transform .
averages of derivatives of @xmath18 may be represented as moments of @xmath109 ( as by longuet - higgins 1957a , b , berry and hannay 1977 , berry and dennis 2000 ) , or equivalently derivatives of @xmath108 as is done here .
coordinates are chosen where @xmath110 since the fields are isotropic , the results are not affected by this choice .
the correlations computed in sections [ sec : crit ] , [ sec : umbilic ] are in two dimensions ; in this case , direction 1 is denoted by @xmath111 direction 2 by @xmath112 representing derivatives by subscripts , the correlations of first derivatives of @xmath18 are found to be ( berry and dennis 2000 ) @xmath113_{r_1 = r , r_2,\dots = 0 } = -c ' \nonumber \\
\fl f & \equiv \langle f_{a,1 } f_{b,1 } \rangle = -\langle f_{a,11 } f_b
\rangle = -\langle f_a f_{b,11 } \rangle & = -\left[\partial_1 ^ 2 c\right]_{r_1 = r , r_2,\dots = 0}= -c '' , \nonumber \\ \fl h & \equiv \langle f_{a , j } f_{b , j } \rangle = -\langle f_{a , jj } f_b \rangle = -\langle f_a f_{b , jj } \rangle & = -\left[\partial_2 ^ 2 c\right]_{r_1 = r , r_2,\dots = 0 } = -c'/r , \quad j\neq 1 .
\label{eq:1stcorrs}\end{aligned}\ ] ] averages involving @xmath18 and its first derivatives other than those in ( [ eq : cdef ] ) , ( [ eq:1stcorrs ] ) are zero .
the averages are equal to appropriate derivatives of @xmath114 and then setting @xmath115 ( as in ( [ eq : rfix ] ) ) , and derivatives in @xmath116 of odd order vanish . the functions in ( [ eq:1stcorrs ] ) , when @xmath117 ( denoted by subscript 0 ) , are @xmath118 it is easily verified that @xmath119 since @xmath108 has a positive fourier transform .
the correlation matrices ( [ eq : cormatcrit ] ) , ( [ eq : cormatumb ] ) required to calculate the topological charge correlation functions of critical points and umbilic points , involve higher derivatives of @xmath120 given in ( [ eq:2ndcorrs ] ) , ( [ eq:3rdcorrs ] ) . as the separation between @xmath28 and @xmath47 increases , the correlation between @xmath121 and @xmath122 decreases and @xmath123 this decay is slowest in the case where all the @xmath96 in ( [ eq : ffourier ] ) have the same length @xmath124 ( set to 1 for convenience ) and the power spectrum @xmath125 for any @xmath126 the corresponding correlation function is a bessel function times
a factor of @xmath127 with oscillatory decay that falls off like @xmath128 of particular interest is the @xmath129 case , for which @xmath130 the spectrum in this case was called the ring spectrum by longuet - higgins ( 1957a , b ) and berry and dennis ( 2000 ) , and is conjectured to model the high eigenfunctions in quantum chaotic billiards ( berry 1978 , 2002 ) .
in the present section , we shall consider @xmath0-dimensional gaussian random vector fields @xmath131 in @xmath0 dimensions whose cartesian components are independent and identically distributed gaussian random fields ( the derivatives of the components are also assumed completely independent ) .
the jacobian @xmath132 whose sign determines the topological charges of the zeros , is the determinant of the matrix of first derivatives ( [ eq : jdef ] ) .
we begin by calculating the density ( [ eq : densdef ] ) of zeros of random vectors .
this was calculated for @xmath133 by halperin ( 1981 ) and liu and mazenko ( 1992 ) , ( the @xmath134 case was previously found by rice ( 1954 ) , and @xmath36 by berry ( 1978 ) ) . for general @xmath26 the density @xmath135 ( [ eq : densdef ] )
is expressed as a probability integral with density function ( [ eq : gausspdf1 ] ) and correlations given by ( [ eq:1stcorrs]),([eq:1stcorr0 ] ) .
therefore @xmath136 where in the final line the @xmath13-functions in the @xmath97 were integrated , and the @xmath137 were rescaled ( each by @xmath138 ) to be dimensionless .
the remaining integral is solved in [ sec : dveccalc ] , giving the density of zeros in @xmath0 dimensions @xmath139 in this expression , @xmath140 is the surface area of the unit @xmath141-sphere in @xmath0 dimensions , given by ( [ eq : spherearea ] ) . as is common in such problems in statistical geometry ,
the result is a spectral quantity ( @xmath142 ) multiplied by a geometric factor .
the scheme of the previous section may now be applied to calculate the topological charge correlation function for zeros in these gaussian random vector fields .
implementation of the scheme is facilitated by the fact that the components @xmath97 are completely independent , and @xmath143 the submatrix @xmath144 of the correlation matrix @xmath145 only depends on the correlations of the components of the vectors @xmath146 from section [ sec : fields ] , the only correlations that do not vanish are @xmath147 and @xmath148 for @xmath149 it is easy to see that @xmath150 from ( [ eq : dexp ] ) , the other necessary ingredient of the correlation function scheme is the components of the matrix @xmath151 defined in ( [ eq : sigkxidef ] ) .
the elements of @xmath91 are labelled by the multiindex of the components @xmath152 using the correlations ( [ eq:1stcorrs ] ) , ( [ eq:1stcorr0 ] ) and the arguments of [ sec : jacobi ] , particularly equation ( [ eq : xiij ] ) , the only nonvanishing elements are @xmath153 the problem remains to use these components and ( [ eq : dexp ] ) to evaluate ( [ eq : ddef ] ) . by ( [ eq : jdef ] ) ,
( [ eq : ftu ] ) , @xmath154 ( similarly for @xmath155 ) .
each of the @xmath156 summands in this determinant is an @xmath0-fold product @xmath157 where @xmath158 is a permutation of @xmath159 therefore @xmath160 from ( [ eq : dexp ] ) , the result of one of these summands acting on @xmath161 and setting @xmath162 is nonzero if there is a pairing of these multiindices where the corresponding elements of @xmath91 are nonzero . from ( [ eq : xivec ] ) , this is only the case when the permutations @xmath163 are the same .
thus , from ( [ eq : dexp ] ) and ( [ eq : xivec ] ) , @xmath164 this , together with ( [ eq : vecdetk ] ) , can now be put into ( [ eq : gabfin ] ) to give @xmath165 where the function @xmath166 is defined @xmath167 this decays to 0 as @xmath168 and when @xmath169 @xmath170 ( the quantity @xmath171 is always negative , since @xmath108 is the fourier transform of the positive power spectrum @xmath109 ) .
the topological charge correlation function of zeros of @xmath0-dimensional gaussian random vector fields @xmath172 was first calculated by halperin ( 1981 ) , in a form equivalent to ( [ eq : vecgab ] ) , but without proof .
this function was also derived by different means by liu and mazenko ( 1992 ) , and in the @xmath134 case by rice ( 1954 ) , and the @xmath36 case by berry and dennis ( 2000 ) and foltin ( 2003a ) .
@xmath172 is plotted in figure [ fig : g2 ] for two choices of the field correlation function @xmath108 for @xmath173 when @xmath174 @xmath175 is oscillatory ; when @xmath176 it increases monotonically to zero as @xmath177 plotted for two choices of @xmath178 ( a ) @xmath179 ( b ) @xmath180,width=453 ]
in this section , the scheme of section [ sec : scheme ] is used to calculate the topological charge correlation function of critical points of isotropic gaussian random functions in the plane , that is the poincar index correlation function of isotropic random surfaces .
the gaussian random function examined shall be written @xmath181 ( where @xmath182 ) , which will be assumed stationary and isotropic , so the expressions in section [ sec : fields ] may be used . in particular , @xmath18 and its first derivatives
have the correlations ( [ eq:1stcorrs ] ) , ( [ eq:1stcorr0 ] ) , as well as further correlations involving second derivatives , described below . as in the previous section , for convenience in calculations ,
the two points @xmath28 and @xmath47 are separated only in the @xmath183 coordinate .
at a critical point , the gradient @xmath184 is zero .
the critical point jacobian @xmath185 whose sign defines the topological charge ( poincar index ) is the hessian determinant @xmath186 this is the gaussian curvature of the surface . unlike a general 2-dimensional random vector field ( as in the previous section ) ,
the gradient field @xmath187 is irrotational , which gives rise to relationships and correlations between the derivatives of the components ( e.g. @xmath188 , whereas before , @xmath189 in general ) .
this makes the explicit computation of the topological charge correlation function more difficult than in the previous section , even in the 2-dimensional case that is considered here ; the scheme of section [ sec : scheme ] applies to gradient zeros in fields of any dimension .
the statistical properties of critical points of gaussian random fields in 2 dimensions were considered by longuet - higgins ( 1957a , b ) ; he found that the density @xmath190 of critical points is ( longuet - higgins 1957b equations ( 71 ) , ( 78 ) ) : @xmath191 ( @xmath192 denotes the fourth derivative of @xmath120 evaluated at @xmath193 ) .
the density of saddles equals the density of extrema ( maxima and minima ) , and the density of maxima equals the density of minima .
the probability density function of the gaussian curvature @xmath194 , despite its asymmetry ( longuet - higgins 1958 equation ( 7.14 ) , dennis 2002 equation ( 57 ) ) , has zero first moment , implying the average topological charge @xmath195 is zero , as expected .
the topological charge correlation function is again calculated using the scheme of section [ sec : scheme ] , particularly ( [ eq : gabfin ] ) .
therefore , the vector @xmath41 ( [ eq : bigudef ] ) of gaussian random variables , in a convenient ordering , is @xmath196 with correlation matrix ( c.f .
( [ eq : cormatdef ] ) ) @xmath197 where the correlations between elements of @xmath198 are computed to be @xmath199_{x = r , y=0 } & = c^{(3 ) } \nonumber \\
i & \equiv \left[\partial_x^2 \partial_y c\right]_{x = r , y=0 } & = ( r c '' - c')/r^2 \nonumber \\
l & \equiv \left[\partial_x^2 \partial_y^2 c\right]_{x = r , y=0 } & = ( r^2 c^{(3 ) } - 2 r c '' + 2 c')/r^3 \nonumber \\ m & \equiv \left[\partial_x^4 c\right]_{x = r , y=0 } & = c^{(4)}\nonumber \\ n & \equiv \left[\partial_y^4 c\right]_{x = r , y=0 } & = 3(r c '' - c')/r^3 \nonumber \\ g_0 & = i_0 = 0 , & m_0 = n_0 = 3l_0 = c^{(4)}_0 .
\label{eq:2ndcorrs}\end{aligned}\ ] ] the last line gives the special value of these correlations when @xmath200 the matrix @xmath201 is the @xmath202 lower right submatrix of @xmath203 and has determinant @xmath204 the pair of differential jacobian operators are , from ( [ eq : hessdef ] ) , @xmath205 using ( [ eq : dexp ] ) , the result of these operators acting is on @xmath206 and setting @xmath162 ( c.f . ( [ eq : gabfin]),([eq : ddef ] ) ) is @xmath207 where the necessary entries of @xmath208 are found using jacobi s determinant theorem in ( [ eq : xiij ] ) ; as an example , @xmath209 the topological charge correlation function @xmath210 for critical points is obtained by substituting ( [ eq : critdetk ] ) , ( [ eq : critdxexp ] ) ( with all terms like ( [ eq : critxiexp ] ) found using ( [ eq : xiij ] ) into ( [ eq : gabfin ] ) ) .
this expression is complicated and not very illuminating , and is not given here . upon substituting ( [ eq:1stcorrs ] ) ,
( [ eq:1stcorr0 ] ) , ( [ eq:2ndcorrs ] ) in , one finds that @xmath211 can be written as a perfect derivative ( c.f .
( [ eq : vecgab ] ) ) , @xmath212 where @xmath213 \nonumber \\ \lo = \frac{1}{r ( c '' - c'/r ) } \frac{\rmd}{\rmd r } \frac{(c '' - c'/r)^3}{\sqrt{(c_0''^2 - c''^2)(c_0''^2 - c'^2/r^2)}}. \label{eq : hcrit}\end{aligned}\ ] ] this process of finding @xmath214 is long and tedious , and details are omitted here . it is easy to see that @xmath215 as @xmath216 it is straightforward , by taylor expanding derivatives of @xmath120 to show that @xmath217 the critical point topological charge correlation function for two particular field correlation functions is shown in figure [ fig : gc ] . as with the 2-dimensional vector case plotted in figure [ fig : g2 ] ,
the properties of the correlation function ( oscillatory , exponential decay , etc ) are similar to that of the underlying field correlation function @xmath104 on which the correlation function depends .
features of interest in these plots include the sharp initial minimum of ( a ) , and the fact that the monotonic decay in ( b ) is from above , not below ( in contrast to its counterpart in figure [ fig : g2 ] .
nevertheless , the form of @xmath214 is significantly more complicated than @xmath218 especially when @xmath173 plotted for two choices of @xmath178 ( a ) @xmath179 ( b ) @xmath180,width=453 ]
less well - known than critical points , umbilic points are geometric point singularity features associated with the second derivative of @xmath18 - namely where hessian matrix of second derivatives @xmath219 is degenerate ( berry and hannay 1977 , porteous 2001 , hilbert and cohn - vossen 1952 ) .
geometrically , the principal axes of gaussian curvature are not defined at these points .
the eigenvalues of the hessian coincide when @xmath220 umbilic points are therefore zeros of the 2-dimensional vector field @xmath221 the factor of half in the first term ensures that @xmath222 is statistically rotation invariant .
an umbilic point has an index , determined geometrically by the sense of rotation of the principal axes of curvature around the umbilic point , and the index is generically @xmath223 ( berry and hannay 1977 ) ; only the sign of the index is of interest here , and this is determined by the appropriate jacobian @xmath224 on @xmath222 , @xmath225 depending on the third partial derivatives of @xmath226 the calculation of the topological charge correlation function for umbilic points can proceed according to the scheme of section [ sec : scheme ] , in a similar way to the corresponding calculation for critical points .
umbilic points for isotropic random functions was considered by berry and hannay ; they found the density @xmath227 to be ( berry and hannay 1977 equation ( 34 ) ) @xmath228 ( @xmath229 is the sixth derivative of @xmath108 at 0 ) and that the average index @xmath230 is zero ( the separate densities are @xmath231 for stars , @xmath232 for monstars , and @xmath233 for lemons ) . in the present work , the distinction between monstars and lemons , which both have positive index , is not used .
the ordering of the vector of gaussian random functions @xmath234 is chosen @xmath235 the correlation matrix ( [ eq : cormatdef ] ) is @xmath236 where @xmath237 and @xmath238_{x = r , y=0 } & = - c^{(5 ) } , \nonumber \\ \fl q & \equiv -\left[\partial_x^3 \partial_y^2 c\right]_{x = r , y=0 } & = -(r^3 c^{(4)}-3r^2 c^{(3)}+ 6r c '' - 6 c')/r^4 , \nonumber \\ \fl r & \equiv -\left[\partial_x \partial_y^4 c\right]_{x = r , y=0 } & = -3(r^2 c^{(3 ) } - 3
r c '' + 3 c')/r^4 , \nonumber \\
\fl s & \equiv-\left[\partial_x^6 c\right]_{x = r , y=0 } & = - c^{(6 ) } , \nonumber \\ \fl t & \equiv -\left[\partial_x^4 \partial_y^2 c\right]_{x = r , y=0 } & = -(r^4 c^{(5 ) } - 4 r^3 c^{(4 ) } + 12 r^2 c^{(3 ) } - 24 r c '' + 24 c')/r^5 , \nonumber \\ \fl u & \equiv -\left[\partial_x^2 \partial_y^4 c\right]_{x = r , y=0 } & = -3(r^3 c^{(4 ) } - 5 r^2 c^{(3 ) } + 12 r c '' - 12 c')/r^5 , \nonumber \\ \fl v & \equiv -\left[\partial_y^6 c\right]_{x = r , y=0 } & = -15(r^2 c^{(3 ) } - 3
r c '' + 3 c')/r^5 , \nonumber \\
\fl p_0 & = q_0 = r_0 = 0 , & s_0 = 5t_0 = 5u_0 = v_0 = -c^{(6)}_0 . \label{eq:3rdcorrs}\end{aligned}\ ] ] the matrix @xmath239 defined using ( [ eq : sigkxidef ] ) , has determinant @xmath240 the result of the jacobian derivative operators ( [ eq : ddef ] ) gives @xmath241 the necessary entries of @xmath242 are found using ( [ eq : xiij ] ) .
the resulting expression for @xmath243 and therefore for @xmath244 is very complicated , but may be reduced to the following form : @xmath245 where @xmath246 it can be shown that @xmath247 and @xmath248 as @xmath249 @xmath250 is plotted in figure [ fig : gu ] for @xmath251 and @xmath252 the most obvious feature of these two plots , compared to figures [ fig : g2 ] , [ fig : gc ] , is that they have a negative maximum , a property that seems to be general for @xmath244 although this has not been proved .
it is unclear what the physical significance of this kink should be ; mathematically , it probably arises from interference between the two summands in @xmath253 in ( [ eq : humb ] ) .
plotted for two choices of @xmath178 ( a ) @xmath179 ( b ) @xmath180,width=453 ]
three particular charge correlation functions have been derived exactly ( equations ( [ eq : vecgab ] ) , ( [ eq : hndef ] ) , ( [ eq : critgab ] ) , ( [ eq : hcrit ] ) , ( [ eq : gumb]),([eq : humb ] ) ) . in each case , the charge correlation function has the form @xmath254 where @xmath0 is the number of dimensions , @xmath94 is the density of zeros and @xmath255 is a function such that @xmath256 the total charge density around a given ( say positive ) topological charge is therefore @xmath257 where the hypersphere area @xmath140 appears in the first line from integration in polar coordinates .
it implies that the distribution of topological charges is such that every topological charge tends to be surrounded by charges of the opposite sign , such that the topological charge is ` screened ' at large distances .
this fact was noticed in the random vector case by halperin ( 1981 ) and liu and mazenko ( 1992 ) ( although the general zero density @xmath135 had not been determined explicitly ) and is independent of the field correlation function @xmath258 the derivation here shows that this is a more general phenomenon , possibly a universal feature of topological charge correlations for gaussian random fields .
it should be noted that ( [ eq : screening1 ] ) is not necessarily satisfied for an arbitrary distribution of signed points ; for instance , @xmath259 always for poisson points , for which there is no screening . @xmath13-function correlations at the origin are ignored in the following .
an analogy may be drawn from the theory of ionic liquids ( hansen and mcdonald 1986 ) ; in a fluid or plasma , consisting of two species identical except for opposite ( coulomb ) charges , the following stillinger - lovett sum rules ( stillinger and lovett 1968a , b ) are found to hold : @xmath260 here , @xmath15 is the charge - charge correlation function , @xmath261 is a characteristic screening length dependent on temperature and @xmath26 and @xmath262 is a constant dependent on dimensionality .
these rules are discussed for @xmath263 by hansen and mcdonald ( 1986 ) , stillinger and lovett 1968a , b , and @xmath36 by jancovici 1987 , jancovici 1994 .
the screening relation ( [ eq : screening1 ] ) is equivalent to the first stillinger - lovett sum rule ( [ eq : slsumrule1 ] ) , which is derived using the statistical mechanics of pairwise , coulomb interacting fluids .
it is unclear whether the fact that topological and coulombic charges screen in the same way is coincidence , or evidence of some deeper connection between the two statistical theories .
it is natural to ask whether the topological charge correlation functions satisfy the second sum rule , which ( upon integrating the left hand side of ( [ eq : slsumrule2 ] ) by parts ) , depends on the integral of @xmath264 for @xmath265 it was found ( berry and dennis 2000 ) that for certain choices of @xmath120 this integral may diverge .
the slowest decay comes when @xmath266 and by ( [ eq : hndef ] ) , @xmath267 giving a logarithmic divergence for the second moment . for comparison ,
the critical and umbilic functions @xmath268 ( equations ( [ eq : hcrit ] ) , ( [ eq : humb ] ) respectively ) both give , for the same choice of @xmath120 @xmath269 implying that the second sum rule is satisfied generally for critical and umbilic points , although , as for the cases of random vectors where the integral converges , the screening length @xmath270 defined in analogy to ( [ eq : slsumrule2 ] ) , depends on the choice of @xmath258 for random vectors with @xmath126 the second moment of @xmath15 always converges , because of the higher power of @xmath271 appearing in ( [ eq : hndef ] ) ( also , the decay of @xmath108 will be faster , as discussed at the end of section [ sec : fields ] ) .
comparison may be drawn to the electrostatic analogy in random matrix theory , particularly in the case of the so - called ginibre ensemble of @xmath272 matrices whose entries are independent , identically distributed circular gaussian random variables ( ginibre 1965 ) .
the eigenvalues of these matrices are found to have exactly the same statistical behaviour as a 1-component 2-dimensional coulomb gas of @xmath20 charges in a harmonic oscillator potential , and the 2-point density correlation function screens against a uniform background ( i.e. @xmath273 ) , and have finite second moment .
certain random polynomial analogues have zeros that can also be expressed as 2-dimensional coulomb gases with additional interactions ( hannay 1998 , forrester and honner 1999 ) .
the eigenvalues of random matrices ( which may be expressed as the zeros of the characteristic polynomial ) and zeros of random polynomials are all of the same sign , since they are zeros of complex analytic functions , and the density correlation functions of zeros are unique ( there is no analogue of @xmath108 )
. there may be a danger in taking the analogy with fluids too far ; for instance , the oscillations of the functions in figures [ fig : g2],[fig : gc],[fig : gu]a are reminiscent of those of charged fluids ( e.g. hansen and mcdonald 1986 ) . however , the physical causes for these oscillations are very different . in fluids ,
the oscillations usually arise from packing considerations ( the ions themselves are of finite size , fixing the lengthscale , although in plasmas they are usually represented as point charges ( baus and hansen 1980 ) ) .
topological charges , on the other hand , are points , and the oscillations in these figures originate from the oscillations in the underlying field correlation function @xmath104 which is @xmath274 in this case .
although halperin ( 1981 ) discusses the similarity between the short - range behaviour of the 2-dimensional vector correlation function ( [ eq : vecgab ] ) , ( [ eq : hndef ] ) and kosterlitz - thouless theory , the present situation is more general , both in that the ( possibly long - range ) screening is exact , and that the results hold for any reasonable field correlation function @xmath275
using a general scheme for calculating topological charge correlation functions , three particular correlation functions were found explicitly , and were found always to satisfy a screening relation ( [ eq : screening1 ] ) .
the scheme of section [ sec : scheme ] used to calculate the charge correlation functions is very general , and can be generalised to calculate the charge - charge correlation function between two different sets of topological charges - for instance with a critical point at @xmath276 and an umbilic point at @xmath29 although not done so here , the scheme may be applied to anisotropic fields .
it is rather more difficult to calculate the density correlation function for topological charges ( the analogue of ( [ eq : gdef ] ) where the moduli of the jacobians are taken ) .
it was calculated for 1-dimensional vectors ( i.e. random functions in 1 dimension ) by rice ( 1954 ) , and for 2-dimensional vectors ( realised as complex scalars ) by berry and dennis ( 2000 ) , saichev ( 2001 ) .
however , it has not been possible to generalise such methods to the case of critical points . also , numerical evidence ( freund and wilkinson 1998 ) suggests that the correlation function of extrema signed by the sign of their laplacian ( @xmath1 for minima , @xmath2 for maxima ) also satisfies the screening relation ( [ eq : slsumrule1 ] ) .
all such functions would be needed to calculate the partial correlation functions between the species ( e.g. maxima with maxima , maxima with saddles , maxima with minima , etc ) , which would give a complete statistical picture .
the presence of boundaries in the random function will affect the statistical properties of topological charges ( e.g. berry ( 2002 ) for nodal points in the plane ) , and it is possible that there may be some further analogy with the physics of interfaces of coulomb fluids .
the scheme of section [ sec : scheme ] ought to be adaptable to calculate charge correlation functions in this case
. only zeros of fields linear in gaussian random functions have been considered here , although the density of others may be calculated , for instance , in addition to nodal points , a 2-dimensional gaussian random complex scalar has critical points of its modulus squared ( weinrib and halperin 1982 ) and its argument ( dennis 2001a ) .
the scheme employed here can not be used to calculate correlation functions for these , although numerical evidence ( i freund , personal communication ) suggests that the critical points of argument ( together with the nodal points ) do screen , therefore adding to the cases shown here .
it is tempting to conjecture that topological charge screening may be a universal phenomenon in gaussian random fields .
i would like to thank michael berry and robert evans for useful discussions , john hannay for discussions leading to the arguments in appendix b , and isaac freund for correspondence .
this research was supported by the leverhulme trust .
let @xmath277 be a square matrix .
the _ minor _
@xmath278 is the determinant of the @xmath279 submatrix of @xmath277 with rows @xmath280 and columns @xmath281 @xmath282 shall be used to denote the complementary minor , that is , the minor of the submatrix of @xmath277 with rows @xmath280 and columns @xmath283 excluded .
then jacobi s determinant theorem ( jeffreys and jeffreys 1956 , page 135 ) states @xmath284 applying this to @xmath23 in ( [ eq : sigkxidef ] ) , and choosing @xmath285 as the submatrix whose determinant is the minor of @xmath286 @xmath287 from which ( [ eq : jacobi ] ) follows directly .
jacobi s theorem can also be used to find the elements @xmath92 of the inverse reduced inverse matrix @xmath91 in ( [ eq : sigkxidef ] ) , needed for ( [ eq : dexp ] ) . in this case ,
( [ eq : jacobith ] ) is applied twice , once on the matrix @xmath288 and once on @xmath289 therefore @xmath290 where in the last line @xmath55 represents the terms @xmath291 where @xmath292 thus the @xmath92 appearing in the expression for @xmath80 in ( [ eq : dexp ] ) , is the determinant of the @xmath293 submatrix comprised of the @xmath294th row and @xmath295th column of @xmath296 and the submatrix @xmath297
in order to integrate ( [ eq : dveccalc ] ) , the following must be integrated @xmath298 this is , mathematically , the average hypervolume of an @xmath0-dimensional parallelepiped specified by gaussian random vectors @xmath299 these gaussian random vectors are identically distributed isotropically in @xmath0-dimensional space .
the hypervolume is nonzero exactly when the set of @xmath0 vectors is linearly independent .
this hypervolume may be found explicitly in a manner reminiscent of the gram - schmidt orthogonalization procedure for vectors .
geometrically , @xmath300 a given factor in this product is therefore the average length of the gaussian random vector @xmath301 in the orthogonal complement of a @xmath302-dimensional subspace @xmath303 now , since the vector @xmath301 is isotropic , it may be represented identically in any choice of orthonormal basis of @xmath0-dimensional space ; in particular , its first @xmath304 components @xmath305 may be chosen to be in @xmath306 ( as in the gram - schmidt procedure ) .
the total contribution of @xmath301 to the integral in ( [ eq : dveccalc ] ) involves the average length of the vector made up of the other components @xmath307 where @xmath308 this is @xmath309^{j-1 } \int_{\mathcal{s}_{k-1 } } \rmd^{k-1}\omega_{k-1 } \int_0^{\infty } \rmd \rho \ ,
\rho^k \exp(-\rho^2/2 ) \label{eq : hyperint}\end{aligned}\ ] ] where , in the second line , the first @xmath310 components have been pulled out as trivial gaussians , integrating to @xmath311 the remaining @xmath312 integrals are the average length of a gaussian random vector in @xmath312-dimensional space .
this integral has been converted to polar coordinates , with @xmath313 the solid angle infinitesimal on the unit @xmath314-sphere @xmath315 and @xmath316 is the radius .
it is well known that the surface area @xmath317 of the unit @xmath314-hypersphere is @xmath318 the @xmath316 integral in ( [ eq : hyperint ] ) is @xmath319 therefore , the numerical part of ( [ eq : dveccalc ] ) is @xmath320 times the product in ( [ eq : parallelgeom ] ) , with each term in the product , now labelled by @xmath321 given by the expression ( [ eq : hyperint ] ) .
therefore @xmath322 this value agrees with that stated by halperin ( 1981 ) , liu and mazenko ( 1992 ) of @xmath323 ( @xmath324 ) , @xmath325 ( @xmath326 ) and @xmath327 ( @xmath263 ) . | 2-point topological charge correlation functions of several types of geometric singularity in gaussian random fields are calculated explicitly , using a general scheme : zeros of @xmath0-dimensional random vectors , signed by the sign of their jacobian determinant ; critical points ( gradient zeros ) of real scalars in two dimensions signed by the hessian ; and umbilic points of real scalars in two dimensions , signed by their index .
the functions in each case depend on the underlying spatial correlation function of the field .
these topological charge correlation functions are found to obey the first stillinger - lovett sum rule for ionic fluids . |
geometrically frustrated magnets give rise to many interesting phenomena originating from the macroscopic ground state degeneracy .
@xcite spin liquid ground state is expected in strongly frustrated quantum kagom and pyrochlore lattice heisenberg antiferromagnets .
@xcite although quantum triangular lattice heisenberg antiferromagnet shows 120@xmath3 long - range magnetic order with 3 sublattices , @xcite additional multiple - spin interactions can lead to spin liquid state .
@xcite triangular lattice heisenberg antiferromagnets with classical spins also show interesting phenomena related to the chirality degree of freedom . a magnetic transition driven by the binding - unbinding of the @xmath4 vortices
is predicted to occur at a temperature , where the vortex correlation length diverges but the spin correlation length remains finite .
@xcite spin - lattice coupling sometimes plays an important role in selecting the ground state in the frustrated magnets .
even if there is no orbital degree of freedom , a small amount of structural distortion is sufficient to lift the ground state degeneracy and stabilize a long - range magnetic order .
for example , in the cr - based spinels , consisting of the pyrochlore lattice , the magnetic order is observed far below the curie - weiss temperature @xmath5 due to spin - lattice coupling .
@xcite furthermore , a half - magnetization plateau is stabilized in a wide range of magnetic field by the spin - lattice coupling .
@xcite similar effect is also predicted for the triangular and kagom lattice antiferromagnets .
@xcite ag@xmath6o@xmath0 ( @xmath7=cr , mn , and ni ) consists of triangular lattice planes of @xmath7o@xmath0 , which are well separated by the metallic ag@xmath0 layers .
@xcite ag@xmath0nio@xmath0 ( ni@xmath8 , @xmath9 low spin @xmath1=1/2 ) and ag@xmath0mno@xmath0 ( mn@xmath8 , @xmath10 high spin @xmath1=2 ) , which have orbital degree of freedom , were studied previously .
@xcite both compounds show jahn - teller distortions , which affect the magnetic ground state . in ag@xmath0nio@xmath0 , a long - range magnetic order with @xmath11=(@xmath12 , @xmath12 , 0)@xmath13 occurs below nel temperature @xmath2=56 k. in ag@xmath0mno@xmath0 , a glassy state with a short - range magnetic order with @xmath11=(@xmath14 , @xmath14 , 0)@xmath15 is observed as the ground state .
cro@xmath0 in trigonal ( @xmath16 ) phase above @xmath2 ( = 24 k ) .
2@xmath172@xmath172 unit cells are shown .
note that there is only one cro@xmath0 triangular lattice plane in a single unit cell .
( b ) and ( c ) show schematic structures of the triangular lattice plane of the cr@xmath8 spins in trigonal ( @xmath18 ) and monoclinic phases ( @xmath19 ) , respectively .
magnetic interactions ( @xmath20 , @xmath21 , and @xmath22 ) are shown in ( b).,width=309 ] ag@xmath0cro@xmath0 has a trigonal structure ( @xmath16 ) with @xmath23=2.9298 and @xmath24=8.6637 at @xmath25=200 k , as shown in fig .
[ structure](a ) , and consists of regular triangular lattices of the magnetic cr@xmath8 ( nominally @xmath26 ions without orbital degree of freedom .
therefore , ag@xmath0cro@xmath0 is a good candidate for a model compound of a regular triangular lattice heisenberg antiferromagnet .
the specific heat shows a sharp peak at @xmath25=24 k. @xcite the magnetic susceptibility shows that @xmath5 = @xmath2797 k and the effective moment @xmath28 = 3.55 @xmath29 , consistent with 3.87 @xmath29 expected for @xmath1=3/2 .
@xcite these results indicate that an antiferromagnetic long - range order develops below @xmath2=24 k. in addition , the magnetic susceptibility shows weak ferromagnetic component below @xmath2 .
we performed neutron diffraction experiments on a powder sample of ag@xmath0cro@xmath0 .
it was found that ag@xmath0cro@xmath0 shows a partially disordered ( pd ) state with 5 sublattices and a structural distortion simultaneously below @xmath2=24 k , indicating a spin - lattice coupling to stabilize the nel order .
we also measured the magnetic diffuse scattering and found that the material does not prefer the 5-sublattice ( 5sl ) structure but 4-sublattice ( 4sl ) structure above @xmath2 .
the structural distortion is considered to modify further - neighbor interactions so that the 5sl structure becomes stable . as far as we know , ag@xmath0cro@xmath0 is the first quasi - two - dimensional ( 2d ) triangular lattice antiferromagnet that shows the long - range pd state .
a powder sample of ag@xmath0cro@xmath0 was prepared by the solid state reaction technique with stoichiometric mixture of ag , ag@xmath0o , and cr@xmath0o@xmath30 powder in high pressure .
@xcite the powder sample that weighs @xmath312.5 g was used .
the neutron powder diffraction experiments were carried out on a neutron powder diffractometer hb-2a , installed at hfir in oak ridge national laboratory ( ornl ) .
we utilized two wavelengths ( @xmath32 ) 1.5374 and 2.410 for structural analysis and high - resolution measurement , respectively .
magnetic diffuse scattering was measured on a thermal triple - axis neutron spectrometer hb-1 , installed at hfir in ornl .
the horizontal collimator sequence was 48-80-s-80-120. the fixed incident neutron energy was 13.5 mev with the energy resolution ( @xmath33 ) of 1.2 mev .
contamination from higher - order beams was effectively eliminated using pg filters .
cro@xmath0 at 4 and 45 k measured with @xmath32=2.410 .
the data at low scattering angles show that some magnetic bragg peaks develop below @xmath34 ( a ) and those at high scattering angles show that a nuclear bragg peak splits below @xmath34 ( b ) .
the inset shows the temperature dependence of the magnetic intensity at 19.8@xmath35 and nuclear intensity at 117.9@xmath35 , measured with increasing temperature .
no hysteresis was observed .
t " and m " denote that the indices are for nuclear reflections in trigonal and monoclinic phases , respectively .
m " denotes that the indices are for magnetic reflections .
the magnetic reflections are indexed on the basis of the trigonal structure.,width=302 ] figure [ profile ] shows the neutron powder diffraction patterns in ag@xmath0cro@xmath0 at 4 and 45 k , measured with a high resolution mode ( @xmath32=2.410 ) .
the magnetic bragg reflections , observed below @xmath2 , are shown in fig .
[ profile](a ) . the magnetic reflections with @xmath36@xmath36@xmath37 and @xmath38@xmath36@xmath37 , where @xmath37=0 and 1 , were observed , indicating that the magnetic structure has a 5sl state in the triangular plane and also the unit cell along the @xmath24 axis is the same as the chemical one so that the magnetic arrangement is ferromagnetic along the @xmath24 axis .
figure [ profile](b ) shows that a nuclear bragg reflection 110 splits into two below @xmath2 .
the inset in fig .
[ profile](b ) shows the temperature dependence of @xmath36@xmath360 magnetic and 110 nuclear bragg peak intensities .
the magnetic intensity abruptly develops below @xmath2 , whereas the nuclear intensity drops , indicating that the magnetic and structural phase transitions occurs simultaneously at @xmath2 .
this result indicates that the spin - lattice coupling releases the highly frustrated state and gives rise to a nel state .
= 1.5374 at @xmath25= 4 k , where nuclear and magnetic bragg peaks coexist.,width=321 ] figure [ rietveld ] shows the neutron powder diffraction pattern at @xmath25=4 k in ag@xmath0cro@xmath0 , measured with @xmath32=1.5374 .
we performed a rietveld refinement , using the fullprof package .
@xcite in the refinement , monoclinic structure with @xmath39 symmetry , which is a subgroup of @xmath16 , is assumed .
the schematic structures for the high - temperature trigonal and low - temperature monoclinic phases are shown in fig .
[ structure](b ) and [ structure](c ) , respectively . in the monoclinic phase , the triangular lattice is slightly contracted along the @xmath40 axis .
the structural parameters determined by the refinement are shown in table i. for the magnetic structure , a pd magnetic structure with 5 sublattices , shown in fig .
[ structure2 ] , is assumed . in this model spin arrangement
is up - down - up - down - disordered- ..... along the @xmath40 axis .
the spins are assumed to point along the @xmath24 axis with a ferromagnetic spin arrangement along this axis .
the weak ferromagnetic component observed in the magnetization measurement is consistent with the pd model since it probably originates from the small ferromagnetic moments induced at the disordered sites in magnetic field .
both the nuclear and magnetic bragg reflections are fitted with the models reasonably well .
the bragg @xmath41-factors for crystal and magnetic structures are @xmath42=6.66% and @xmath43=11.5% , respectively .
the ordered magnetic moment at the ordered spin sites was fitted to be 2.9(1 ) @xmath29 , which almost corresponds to the full moment for @xmath1=3/2 .
@xmath25=45 k , trigonal ( @xmath44@xmath45@xmath461 ) + .structural parameters of ag@xmath0cro@xmath0 determined by the rietveld refinements at @xmath25=4 and 45 k. [ cols="^,^,^,^,^,^",options="header " , ] + @xmath23=5.07976(16 ) , @xmath40=2.916985(9 ) , @xmath24=8.66164(2 ) , @xmath47=90.0723(3)@xmath35 + @xmath48=2.9(1 ) @xmath29 + @xmath42=6.66% , @xmath43=11.5% the structural distortion , which is accompanied by the magnetic transition , is unexpected because the cr@xmath8 ions have no orbital degree of freedom .
this behavior is very similar to that observed in the highly - frustrated cr - based spinel @xmath49cr@xmath0o@xmath50 ( @xmath49=zn , cd , and hg ) , in which the spin - lattice coupling stabilizes the magnetic structure both in ambient and high magnetic field .
@xcite it is also reported that a triangular lattice antiferromagnet cufeo@xmath0 , in which fe@xmath8 ion with @xmath1=5/2 has no orbital degree of freedom , shows a very similar behavior .
@xcite in cufeo@xmath0 , a monoclinic lattice distortion elongates the triangular lattice along the @xmath40 axis and the spin structure is a 4sl state ( up - down - up - down- ..... along the @xmath40 axis ) .
since the antiferromagnetic coupling between nearest - neighbor ( nn ) fe@xmath8 spins is mediated by the fe - o - fe superexchange interaction , the lattice elongation along the @xmath40 axis increases the bond angle of the fe - o - fe coupling so that the antiferromagnetic coupling along the @xmath40 axis is enhanced . as a result ,
the 4sl spin structure is stabilized . in ag@xmath0cro@xmath0 ,
one side of the triangle along the @xmath40 axis is contracted by 0.4% at @xmath25=4 k , as shown in fig .
[ structure](c ) . the antiferromagnetic coupling between nn cr@xmath8 spins is mediated by direct cr - cr exchange interaction .
therefore , the contraction along the @xmath40 axis enhances the antiferromagnetic coupling along the @xmath40 axis . as shown in fig .
[ structure2 ] , the 5sl structure can be stabilized by enhancing the antiferromagnetic coupling along the @xmath40 axis .
it is very interesting that the triangular lattice is elastic and can expand or contract to stabilize long - range magnetic order .
the spin - lattice coupling is a common feature in triangular lattice antiferromagnets in which further - neighbor interactions give rise to multiple - sublattice states . .
the ordered spins point along the @xmath24 axis almost perpendicular to the triangular plane .
the dashed lines represent the magnetic unit cell.,width=302 ]
what is the origin of the pd state with 5sl in ag@xmath0cro@xmath0 ?
further - neighbor interactions should be relevant to such an unusual magnetic state .
performed monte carlo calculations in the triangular lattice ising magnet with @xmath20 , @xmath21 , and @xmath22 , shown in fig .
[ structure](b ) .
@xcite the 5sl state is not stable at @xmath25=0 k but can appear at finite temperature when @xmath21 and @xmath22 are antiferromagnetic and relatively large , although the region where the phase exists is very narrow . the 4- or 8-sublattice
state becomes more stable at low temperatures .
it was also reported that the heisenberg model for nearest - neighbor spins including a spin - lattice coupling explains the collinear 4- and 8-sublattice states .
@xcite as a result of the spin - lattice coupling , the effective spin hamiltonian becomes an ising model with nearest- , second- , and third - nearest neighbor coupling , @xmath51 , @xmath52 , and @xmath52 , respectively , where @xmath24 is a spin - lattice coupling . in cufeo@xmath0
large @xmath21(@xmath310.52@xmath20 ) and @xmath22(@xmath310.70@xmath20 ) were actually observed .
@xcite this is because the overlap between fe and o orbitals is relatively large .
however , in ag@xmath0cro@xmath0 , in which @xmath53 orbitals of the cr@xmath8 ion are not occupied , the overlap of the cr and o orbitals is small .
therefore , @xmath21 and @xmath22 , corresponding to cr - o - o - cr super - super exchange interactions , are not expected to be so large compared to those in cufeo@xmath0 .
it can be deduced that the non - negligible @xmath21 and @xmath22 originate from the rkky interaction mediated by the metallic ag@xmath0 layers .
the interlayer couplings , which give rise to the long - range magnetic order , is also considered to originate from the rkky interaction .
an abrupt drop of the resistivity around @xmath2 @xcite indicates a strong coupling between the magnetic cro@xmath0 and metallic ag@xmath0 layers .
theoretical studies are desirable to clarify the interesting coupling between the two layers .
further inelastic neutron scattering study is also needed to determine the further - neighbor interactions .
the pd state has been reported in the ising spin chains , which form the triangular lattice , such as cscobr@xmath30 , @xcite cscocl@xmath30 , @xcite and rbcobr@xmath30 @xcite which are antiferromagnetic chain compounds and ca@xmath30corho@xmath54 that is a ferromagnetic chain compound . @xcite these compounds show pd state with 3 sublattices .
in particular , ca@xmath30corho@xmath54 shows a pd spin structure , in which the spin arrangement along the @xmath24 axis is ferromagnetic , as in ag@xmath0cro@xmath0 . in ca@xmath30corho@xmath54 ,
the pd state appears at @xmath55=90 k and the disordered spins randomly freeze at @xmath56=30 k. the magnetic susceptibility shows a weak ferromagnetic component with a large hysteresis between measurements with field - cool ( fc ) and zero - field - cool ( zfc ) processes below @xmath57 . on the other hand , the magnetic susceptibility in ag@xmath0cro@xmath0 shows a weak ferromagnetic component with a tiny hysteresis below @xmath2 down to 2 k , @xcite indicating that the pd state persists at 2 k. it is probable that the disordered spins fluctuate even at low temperatures due to strong frustration and high two - dimensionality .
as far as we know , this is the first observation of the long - range pd state in quasi-2d triangular lattice antiferromagnet . the spin - lattice coupling is considered to stabilize the pd state with 5sl , which is expected to appear only at finite temperatures in a limited range of @xmath21 and @xmath22 and has never been observed experimentally .
= 26 k above @xmath2 . the scattering at @xmath25=100 k was subtracted .
the negative intensity at low @xmath58 probably originates from the paramagnetic scattering that increases at low @xmath58 at high temperatures .
the dotted , solid , and broken curves are the results of model calculations for 3- , 4- , and 5-sublattice states with a 2d spin correlation length @xmath59=5 , respectively.,width=302 ] we found the interesting spin - lattice coupling that stabilizes the pd state with 5 sublattices . as the next step , it is interesting to clarify whether the 5sl state is preferred above @xmath2 .
we studied how the spin correlation develops with decreasing temperature .
figure [ diffuse ] shows a neutron diffuse scattering spectrum , observed at @xmath25=26 k , which is 2 k above @xmath2 .
the spectrum measured at @xmath25=100 k is subtracted as a background . a broad peak centered around 1.2 @xmath60 is observed .
the broad peak was simulated using the powder averaged 2d squared lorentzian .
magnetic form factor was also included in the calculation .
since the peak width is much larger than the instrumental resolution ( @xmath310.04 @xmath60 ) , the resolution correction was not performed . the solid curve , which is the result of a model calculation for 4sl state with a 2d spin correlation length @xmath59=5 , reproduces the observed data reasonably well . considering the paramagnetic contribution ,
which is oversubtracted , the agreement would become improved .
the model calculations with 3-sublattice ( 3sl ) and 5sl states do not describe the observed data .
this result suggests that the spin system prefers 4sl state just above @xmath2 .
therefore , the further - neighbor interactions are considered to be already dominant above the structural transition temperature . in the @xmath20-@xmath21-@xmath22 model ,
the energies per spin for the 4sl and 5sl phases are evaluated to be @xmath61 and @xmath62 , respectively , where @xmath63 is negative for antiferromagnetic interaction . here ,
@xmath64 and @xmath65 represent nearest- and second - nearest - neighbor couplings between the triangular lattice layers , respectively .
our results indicate that @xmath66 above @xmath2 .
therefore , the relation of @xmath670 is expected .
the structural transition , which gives rise to an anisotropy in @xmath20 , occurs at @xmath2 .
we discuss the magnetic structure using averaged @xmath63 s , since there is no theoretical calculation for the @xmath20-@xmath21-@xmath22 model in the distorted lattice . from the formulas of @xmath68 and @xmath69 , @xmath20 , @xmath21 , and @xmath65
are considered to be influential to select one from 4sl and 5sl states .
as mentioned above , @xmath20 enhanced along the @xmath40 axis stabilizes both 4sl and 5sl states . in order to stabilize the 5sl state more than the 4sl state
, @xmath70@xmath21/@xmath20@xmath70 should increase .
@xcite since @xmath21 perhaps originates from the rkky interaction , the abrupt decrease in resistivity around @xmath2 @xcite is considered to change the rkky interaction and enhance @xmath21 .
an increase of @xmath65 also stabilizes the 5sl state .
it is noted that the pd state is stable only at finite temperatures .
other collinear states such as 8-sublattice phase should become stable at @xmath25=0 k. @xcite although a pd state with 9-sublattices is predicted between the 4sl and 5sl phases , it was not observed experimentally .
we have studied spin correlations and spin - lattice coupling in a triangular lattice antiferromagnet ag@xmath0cro@xmath0 . with decreasing temperature , a short - ranged 4sl spin state first develops .
at @xmath2 , spin - lattice coupling gives rise to a long - range pd state with 5 sublattices , which is expected to appear only in limited ranges of @xmath21 , @xmath22 , and temperature . since this is the first observation of the long - range pd phase in quasi-2d triangular lattice magnet , theoretical progress in this field is highly desirable .
we would like to thank profs .
y. maeno and y. motome for stimulating discussions .
the work at ornl was sponsored by the scientific user facilities division and materials sciences and engineering division , office of basic energy sciences , u. s. department of energy .
see , for example , _ frustrated spin systems _ , edited by h. t. diep ( world scientific , singapore , 2004 ) .
r. moessner and j. t. chalker , phys .
rev . lett . * 80 * , 2929 ( 1998 ) ; phys .
b * 58 * , 12049 ( 1998 ) .
b. canals and c. lacroix , phys .
lett . * 80 * , 2933 ( 1998 ) ; phys .
b * 61 * , 1149 ( 2000 ) .
s. yan , d. a. huse , and s. r. white , science * 332 * , 1173 ( 2011 ) .
b. bernu , p. lecheminant , c. lhuillier , and l. pierre , phys . rev .
b * 50 * , 10048 ( 1994 ) . w. liming , g. misguich , p. sindzingre , and c. lhuillier , phys . rev .
b. * 62 * , 6372 ( 2000 ) .
h. kawamura and s. miyashita , j. phys .
* 53 * , 9 ( 1984 ) .
h. kawamura , a. yamamoto , and t. okubo , j. phys .
. jpn . * 79 * , 023701 ( 2010 ) . s .-
lee , c. broholm , t. h. kim , w. ratcliff , and s .- w .
cheong , phys . rev
. lett . * 84 * , 3718 ( 2000 ) .
h. ueda , h. aruga - katori , h. mitamura , t. goto , and h. takagi , phys .
* 94 * , 047202 ( 2005 ) .
k. penc , n. shannon , and h. shiba , phys .
* 93 * , 197203 ( 2004 ) .
f. wang and a. vishwanath , phys .
lett . * 100 * , 077201 ( 2008 ) .
h. yoshida , y. muraoka , t. srgel , m. jansen , and z. hiroi , phys .
b * 73 * , 020408(r ) ( 2006 ) .
h. yoshida , s. ahlert , m. jansen , y. okamoto , j. yamaura , and z. hiroi , j. phys .
. jpn . * 77 * , 074719 ( 2008 ) .
h. nozaki , j. sugiyama , m. janoschek , b. roessli , v. pomjakushin , l. keller , h. yoshida , and z. hiroi , j. phys . : condens .
matter * 20 * , 104236 ( 2008 ) .
s. ji , e. j. kan , m .- h .
whangbo , j .- h .
kim , y. qiu , m. matsuda , h. yoshida , z. hiroi , m. a. green , t. ziman , and s .- h .
lee , phys .
b * 81 * , 094421 ( 2010 ) h. yoshida , e. takayama - muromachi , and m. isobe , j. phys . soc .
jpn * 80 * , 123703 ( 2011 ) .
j. rodriguez - carvajal , physica b * 192 * , 55 ( 1993 ) .
f. ye , y. ren , q. huang , j. a. fernandez - baca , p. dai , j. w. lynn , and t. kimura , phys .
b * 73 * , 220404(r ) ( 2006 ) .
n. terada , s. mitsuda , h. ohsumi , and k. tajima , j. phys .
soc . jpn .
* 75 * , 023602 ( 2006 ) .
m. mekata , n. yaguchi , t. takagi , t. sugino , s. mitsuda , h. yoshizawa , n. hosoito , and t. shinjo , j. phys .
62 * , 4474 ( 1993).t .
takagi and m. mekata , j. phys .
* 64 * , 4609 ( 1995 ) .
r. s. fishman , f. ye , j. a. fernandez - baca , and j. t. haraldsen , and t. kimura , phys .
b * 78 * , 140407(r ) ( 2008 ) .
w. b. yelon , d. e. cox , and m. eibschtz , phys . rev .
b * 12 * , 5007 ( 1975 ) .
m. mekata and k. adachi , j. phys .
soc . jpn . *
44 * , 806 ( 1978 ) .
y. nishiwaki , t. nakamura , a. oosawa , k. kakurai , n. todoroki , n. igawa , y. ishii , and t. kato , j. phys .
. jpn . * 77 * , 104703 ( 2008 ) .
s. niitaka , h. kageyama , k. yoshimura , k. kosuge , s. kawano , n. aso , a. mitsuda , h. mitamura , and t. goto , j. phys .
. jpn . * 70 * , 1222 ( 2001 ) . | ag@xmath0cro@xmath0 is an @xmath1=3/2 frustrated triangular lattice antiferromagnet without orbital degree of freedom . with decreasing temperature
, a 4-sublatice spin state develops .
however , a long - range partially disordered state with 5 sublattices abruptly appears at @xmath2=24 k , accompanied by a structural distortion , and persists at least down to 2 k. the spin - lattice coupling stabilizes the anomalous state , which is expected to appear only in limited ranges of further - neighbor interactions and temperature .
it was found that the spin - lattice coupling is a common feature in triangular lattice antiferromagnets with multiple - sublattice spin states , since the triangular lattice is elastic . |
to prepare two identically , polarization squeezed modes we use a well known technique like in @xcite .
each of these modes is generated by launching two orthogonally polarized femtosecond pulses ( @xmath76200fs ) with balanced powers onto the two birefringent axes of a polarization maintaining fiber ( fs - pm-7811 , thorlabs , 13 m ) .
the pump source is a soliton - laser emitting light at a center wavelength of 1559 nm and a repetition rate of 80mhz . by exploiting the optical kerr effect of the fibers , the orthogonally polarized pulses are individually quadrature squeezed and
subsequently temporally overlapped with a relative phase of @xmath38 , resulting in a circular polarized light beam . the relative phase is actively controlled using an interferometric birefringence compensator including a piezoelectric transducer and a locking loop based on a 0.1% tap - off signal after the fiber . in terms of stokes observables ( see @xcite ) this results in states with zero mean values of @xmath30 and @xmath31 , but a bright @xmath77 component .
these states exhibit polarization squeezing at a particular angle in the @xmath30-@xmath31-plane .
we implement the entanglement distribution protocol using optical modes which are systems in infinitely - dimensional hilbert state space . an @xmath78-mode system can be conveniently characterized by the quadrature operators @xmath79
, @xmath80 satisfying the canonical commutation rules @xmath81=i\delta_{jk}$ ] which can be expressed in the compact form as @xmath82={i}{\omega_{n}}_{jk}.\ ] ] here we have introduced the vector of quadratures @xmath83 and @xmath84 is the symplectic matrix .
the present protocol relies on gaussian quantum states . as any standard gaussian distribution , a gaussian state @xmath85
is fully characterized by the vector of its first moments @xmath86 and by the covariance matrix @xmath41 with elements @xmath87,\end{aligned}\ ] ] where @xmath88 is the anticommutator .
a real symmetric positive - definite @xmath89 matrix @xmath41 describes a covariance matrix of a physical quantum state if and only if it satisfies the condition @xcite : @xmath90 the separability of gaussian states can be tested using the positive partial transpose ( ppt ) criterion .
a single mode @xmath91 is separable from the remaining @xmath92 modes if and only if the gaussian state @xmath85 has a positive partial transposition @xmath93 with respect to the mode @xmath91 @xcite . on the level of the covariance matrices ,
the partial transposition is represented by a matrix @xmath94 , where @xmath95 is the diagonal pauli @xmath96-matrix of mode @xmath91 and @xmath97 is the @xmath98 identity matrix .
the matrix @xmath99 corresponding to a partially transposed state @xmath93 reads @xmath100 . in terms of the covariance matrix
, one can then express the ppt criterion in the following form .
a mode @xmath91 is separable from the remaining @xmath92 modes if and only if @xcite @xmath101 the ppt criterion ( [ heisenbergtj ] ) is a sufficient condition for separability only under the assumption of gaussianity . in our experiment , however , non - gaussian states can be generated for which this criterion represents only a necessary condition for separability .
therefore it can fail in detecting entanglement .
there are two sources of imperfections in our experimental set up that are potential sources of non - gaussianity .
these are phase fluctuations and the modulation of the initial squeezed states before the first beam splitter .
they are discussed in the following sections .
the experiment includes an interference of the modes @xmath0 and @xmath2 on a beam splitter , which is the first beam splitter @xmath102 in the protocol .
imperfect phase locking at this beam splitter might cause a phase drift resulting in a non - gaussian character of the state @xmath36 after the beam splitter .
the phase fluctuations can be modelled by a random phase shift of mode @xmath0 before the beam splitter described by a gaussian distribution @xmath103 with zero mean and variance @xmath104 . denoting the operator corresponding to a beam splitter transformation as @xmath105 and the phase shift @xmath106 on mode @xmath0 as @xmath107
, the state @xmath36 can be linked to the state @xmath108 before the onset of phase fluctuations as @xmath109 hence we can express the measured covariance matrix @xmath43 given in eq .
( 3 ) of the main letter , and the vector of the first moments @xmath110 of the state @xmath36 in terms of the covariance matrix @xmath111 and the vector of the first moments @xmath112 of the input state @xmath108 . for this
it is convenient to define matrices @xmath113 and @xmath114 of the first moments with elements @xmath115 and @xmath116 , @xmath117 . using eq .
( [ rhoacprimed ] ) and after some algebra , one gets the transformation rule for the matrix of the first moments in the form @xmath118 , where @xmath119 describes the beam splitter on the level of covariance matrices .
@xmath120 is a diagonal matrix .
similarly we get the covariance matrix @xmath121 where @xmath122 is the @xmath98 zero matrix and @xmath123 here the matrix @xmath0 is the @xmath98 matrix with elements @xmath124 , @xmath125 , @xmath126 is the @xmath98 matrix with elements @xmath127 , @xmath125 , and @xmath128 is defined in eq .
( [ omega ] ) . similar to ref .
@xcite we can now invert the relation ( [ gammaacprimed ] ) and express the input covariance matrix @xmath111 via the output covariance matrix @xmath43 and the first moments after the beam splitter @xmath102 as @xmath129 where @xmath130 the @xmath98 matrices @xmath131 and @xmath132 possess the elements @xmath133 and @xmath134 , @xmath125 .
our estimate for the variance of the phase fluctuations is @xmath135 and the vector @xmath110 of the measured mean values of the state @xmath36 reads @xmath136 by substituting these experimental values for @xmath137 and @xmath110 in eq .
( [ gammaac ] ) and using the beam splitter with the measured transmissivity @xmath138 we get a legitimate covariance matrix @xmath111 before the phase fluctuations as can be easily verified by checking the condition ( [ heisenberg ] ) .
provided that the state with the covariance matrix @xmath111 is classical it can be expressed as a convex mixture of products of coherent states .
gaussian distributed phase fluctuations and a beam splitter preserve the structure of the state , hence the state after the first beam splitter can not be entangled .
the covariance matrix @xmath111 determines a physical gaussian quantum state .
moreover , the covariance matrix possesses all eigenvalues greater than one and therefore the state is not squeezed @xcite which is in a full agreement with the fact that modulations of modes @xmath0 and @xmath2 completely destroy the squeezing .
it then follows that this gaussian state is classical and it therefore transforms to a separable state after the first beam splitter .
the inversion ( [ gammaac ] ) thus allows us to associate a gaussian state before the phase fluctuations with the covariance matrix @xmath43 measured after the first beam splitter .
the separability properties of the state after the beam splitter can then be determined from the non - classicality properties of this gaussian state .
we have paid great attention on the modulations on modes @xmath0 and @xmath2 to preserve gaussian character of the state @xmath36 .
our success can be visually inspected at the examples in fig .
[ fig : hists ] , which illustrates that both the modulation and the subsequent gaussian mixing faithfully samples the required gaussian shape . of the gaussian mixed state ( blue ) and three exemplary individual modes .
* this figure illustrates the preparation of the gaussian mixed state via post processing .
exemplarily , three of the 6400 displaced individual modes are visualized by their histograms ( in green , black and red colour ) .
the normalization is chosen such that they can be depicted in the same plot as the histogram of the mixed state ( blue ) , which is normalized to its maximum value . by merging the data for all individual modes using a weighting with a two dimensional gaussian distribution
, the mixed state is achieved .
its gaussianity is visualized by the gaussian fit ( red curve ) . ] besides this raw visual check we have also tested quantitatively gaussianity of the involved states by measuring higher - order moments of the stokes measurements on modes @xmath9 and @xmath7 .
specifically , we have focused on the determination of the shape measures called skewness @xmath139 and kurtosis @xmath140 defined for a random variable @xmath18 as the following third and fourth standardized moments @xmath141 where @xmath142 is the @xmath143th central moment , @xmath144 is the mean value and @xmath145 is the standard deviation .
skewness characterizes the orientation and the amount of skew of a given distribution and therefore informs us about its asymmetry in the horizontal direction .
gaussian distributions possess skewness of zero .
the exemplary values of skewness for various measurement settings are summarized in the table [ stable ] .
.kurtosis @xmath140 for stokes measurements on modes @xmath9 and @xmath7 in different measurement directions
. [ cols="^,^,^",options="header " , ] [ ktable ] the skewness can vanish also for the other symmetrical distributions , which may , however , differ from a gaussian distribution in the peak profile and the weight of tails .
these differences can be captured by the kurtosis which is equal to 3 for gaussian distributions .
the exemplary values of kurtosis for various measurement settings are summarized in the table [ ktable ] .
the tables reveal that the measured probability distributions satisfy within the experimental error the necessary gaussianity conditions @xmath146 and @xmath147 .
more sophisticated normality tests can be performed , which is beyond the scope of the present manuscript .
by dividing our dataset in 10 equal in size parts we can estimate the statistical errors of our measured covariance matrices @xmath43 and @xmath39 given in eqs .
( 3 ) and ( 5 ) of the main letter .
we calculate the covariance matrix for each part and use the standard deviation as error estimation .
the covariance matrix @xmath43 including the statistical error turns out to be 99 e. schrdinger , naturwissenschaften * 23 * , 807 ( 1935 ) .
a. einstein , b. podolsky , and n. rosen , phys . rev . * 47 * , 777 ( 1935 ) .
m. d. reid , phys .
a * 40 * , 913 ( 1989 ) .
h. m. wiseman , s. j. jones , and a. c. doherty , phys .
. lett . * 98 * , 140402 ( 2007 ) .
j. s. bell , physics ( long island city , n.y . )
* 1 * , 195 ( 1965 ) . n. j. cerf and c. adami , phys . rev . lett . * 79 * , 5194 ( 1997 ) .
m. horodecki , j. oppenheim , and a. winter , nature ( london ) * 436 * , 673 ( 2005 ) . c. h. bennett , g. brassard , c. crpeau , r. jozsa , a. peres , and w. k. wootters , phys .
lett . * 70 * , 1895 ( 1993 ) .
m. ukowski , a. zeilinger , m. a. horne , and a. k. ekert , phys .
lett . * 71 * , 4287 ( 1993 ) .
briegel , w. dr , j. i. cirac , and p. zoller , phys .
lett . * 81 * , 5932 ( 1998 ) .
a. k. ekert , phys .
67 * , 661 ( 1991 ) . c. h. bennett and s. j. wiesner , phys .
lett . * 69 * , 2881 ( 1992 ) .
r. raussendorf and h .- j .
briegel , phys .
lett . * 86 * , 5188 ( 2001 ) .
pan , d. bouwmeester , h. weinfurter , and a. zeilinger , phys .
80 * , 3891 ( 1998 ) .
t. s. cubitt , f. verstraete , w. dur and j. i. cirac , phys .
* 91 * , 037902 ( 2003 ) .
a. streltsov , h. kampermann , and d. bru , phys .
lett . * 108 * , 250501 ( 2012 ) .
t. k. chuan , j. maillard , k. modi , t. paterek , m. paternostro , and m. piani , phys .
lett . * 109 * , 070501 ( 2012 ) .
a. kay , phys .
lett . * 109 * , 080503 ( 2012 ) .
l. mita , jr . and n. korolkova , phys . rev . a * 80 * , 032310 ( 2009 )
. l. mita , jr . and n. korolkova , phys .
a * 86 * , 040305 ( 2012 ) .
a. peres , phys .
lett . * 77 * , 1413 ( 1996 ) .
m. horodecki , p. horodecki , and r. horodecki , phys .
a * 223 * , 1 ( 1996 ) .
r. f. werner and m. m. wolf , phys .
lett . * 86 * , 3658 ( 2001 ) . v. giovannetti , s. mancini , d. vitali , and p. tombesi , phys . rev .
a * 67 * , 022320 ( 2003 ) .
r. dong , j. heersink , j .-
yoshikawa , o. glckl , u. l. andersen , and g. leuchs , new j. phys .
* 9 * , 410 ( 2007 ) .
j. heersink , v. josse , g. leuchs , and u. l. andersen , opt . lett .
* 30 * , 1192 ( 2005 ) .
n. korolkova , g. leuchs , r. loudon , t. c. ralph , and c. silberhorn , phys .
a * 65 * , 052306 ( 2002 ) .
g. leuchs , t. c. ralph , c. silberhorn , and n. korolkova , j. mod.opt . * 46 * , 1927 ( 1999 ) . c. silberhorn , p. k. lam , o. wei , f. knig , n. korolkova , and g. leuchs , phys
lett . * 86 * 4267 ( 2001 ) .
f. benatti and r. floreanini , j. phys .
a * 39 * , 2689 ( 2006 ) .
d. mogilevtsev , t. tyc and n. korolkova , phys .
a * 79 * , 053832 ( 2009 ) .
g. adesso and a. datta , phys . rev
. lett . * 105 * , 030501 ( 2010 ) . c. e. vollmer , d. schulze , t. eberle , v. hndchen , j. fiurek , and r. schnabel , arxiv:1303.1082 [ accepted by phys .
a. fedrizzi , m. zuppardo , g. g. gillett , m. a. broome , m. de almeida , m. paternostro , a. g. white , and t. paterek , arxiv:1303.4634 [ accepted by phys .
lett . ] ( 2013 ) .
r. simon , n. mukunda , and b. dutta , phys .
a * 94 * , 1567 ( 1994 ) .
r. simon , and p. horodecki , phys .
lett . * 84 * , 2726 ( 2000 ) . | we experimentally demonstrate a protocol for entanglement distribution by a separable quantum system . in our experiment , two spatially separated modes of an electromagnetic field
get entangled by local operations , classical communication , and transmission of a correlated but separable mode between them .
this highlights the utility of quantum correlations beyond entanglement for the establishment of a fundamental quantum information resource and verifies that its distribution by a dual classical and separable quantum communication is possible . like a silver thread
, quantum entanglement @xcite runs through the foundations and breakthrough applications of quantum information theory .
it can not arise from local operations and classical communication ( locc ) and therefore represents a more intimate relationship among physical systems than we may encounter in the classical world .
the `` nonlocal '' character of entanglement manifests itself through a number of counterintuitive phenomena encompassing the einstein - podolsky - rosen paradox @xcite , steering @xcite , bell nonlocality @xcite , or negativity of entropy @xcite .
furthermore , it extends our abilities to process information . here
, entanglement is used as a resource which needs to be shared between remote parties . however , entanglement is not the only manifestation of quantum correlations .
notably , separable quantum states can also be used as a shared resource for quantum communication .
the experiment presented in this letter highlights the quantumness of correlations in separable mixed states and the role of classical information in quantum communication by demonstrating entanglement distribution using merely a separable ancilla mode .
the role of entanglement in quantum information is nowadays vividly demonstrated in a number of experiments .
a pair of entangled quantum systems shared by two observers enables us to teleport @xcite quantum states between them with a fidelity beyond the boundary set by classical physics . concatenated teleportations @xcite can further span entanglement over large distances @xcite which can be subsequently used for secure communication @xcite .
an _ a priori _ shared entanglement also allows us to double the rate at which information can be sent through a quantum channel @xcite or one can fuse bipartite entanglement into larger entangled cluster states that are `` hardware '' for quantum computing @xcite .
the common feature of all entangling methods used so far is that entanglement is either produced by some global operation on the systems that are to be entangled or it results from a direct transmission of entanglement ( possibly mediated by a third system ) between the systems .
even entanglement swapping @xcite , capable of establishing entanglement between the systems that do not have a common past , is not an exception to the rule because also here entanglement is directly transmitted between the participants .
however , quantum mechanics admits conceptually different means of establishing entanglement which are free of transmission of entanglement .
remarkably , the creation of entanglement between two observers can be disassembled into local operations and the communication of a _
separable _ quantum system between them @xcite .
the impossibility of entanglement creation by locc is not violated because communication of a quantum system is involved .
the corresponding protocol exists only in a mixed - state scenario and obviously utilizes fewer quantum resources in comparison with the previous cases because communication of only a discordant @xcite separable quantum system is required . in this letter
, we experimentally demonstrate the entanglement distribution by a separable ancilla @xcite with gaussian states of light modes @xcite .
the protocol aims at entangling mode @xmath0 which is in possession of a sender alice , with mode @xmath1 held by a distant receiver bob by local operations and transmission of a separable mediating mode @xmath2 from alice to bob .
this requires the parties to prepare their initial modes @xmath0 , @xmath1 , and @xmath2 in a specific correlated but fully separable gaussian state .
once the resource state @xmath3 is established , no further classical communication is needed to accomplish the protocol . to emphasize this
, we attribute the state preparation process to a separate party , david .
note that this resource state preparation is performed by locc only .
no global quantum operation with respect to david s separated boxes is executed at the initial stage , and no entanglement is present . , a position squeezed vacuum mode @xmath2 , and a vacuum mode @xmath1 .
he then applies random displacements ( green boxes ) of the @xmath4 quadrature ( horizontal arrow ) and the @xmath5 quadrature ( vertical arrow ) as in eq .
( [ displacements ] ) , which are correlated via a classical communication channel ( green line ) .
david passes modes @xmath0 and @xmath2 to alice and mode @xmath1 to bob .
alice superimposes modes @xmath0 and @xmath2 on a balanced beam splitter bs@xmath6 and communicates the separable output mode @xmath7 to bob ( red line connecting alice and bob ) .
bob superimposes the received mode @xmath7 with his mode @xmath1 on another balanced beam splitter bs@xmath8 , which establishes entanglement between the output modes @xmath9 and @xmath10 ( black lemniscata ) .
note the position of the displacement on mode @xmath1 . in the original protocol ,
the displacement is performed before bs@xmath8 , which is depicted by the corresponding box with a dashed green line .
equivalently , this displacement on mode @xmath1 can be performed after bs@xmath8 ( dashed arrow indicates the respective relocation of displacement ) on mode @xmath10 , and even _ a posteriori _ after the measurement of mode @xmath10 . ]
[ [ protocol . ] ] protocol .
+ + + + + + + + + the protocol @xcite depicted in fig .
[ fig1 ] consists of three steps . initially , a distributor david prepares modes @xmath0 and @xmath2 in momentum squeezed and position squeezed vacuum states , respectively , with quadratures @xmath11 and @xmath12 , whereas mode @xmath1 is in a vacuum state with quadratures @xmath13 and @xmath14 . here , @xmath15 is the squeezing parameter and the superscript @xmath16 denotes the vacuum quadratures .
david then exposes all the modes to suitably tailored local correlated displacements @xcite : @xmath17 the uncorrelated classical displacements @xmath18 and @xmath19 obey a zero mean gaussian distribution with the same variance @xmath20 . the state has been prepared by locc across @xmath21 splitting and hence is fully separable . in the second step ,
david passes modes @xmath0 and @xmath2 of the resource state to alice and mode @xmath1 to bob .
alice superimposes modes @xmath0 and @xmath2 on a balanced beam splitter bs@xmath6 , whose output modes are denoted by @xmath9 and @xmath7 .
the beam splitter bs@xmath6 can not create entanglement with mode @xmath1 .
hence the state is separable with respect to @xmath22 splitting .
moreover , the state also fulfils the positive partial transpose ( ppt ) criterion @xcite with respect to mode @xmath7 and hence is also separable across @xmath23 splitting @xcite , as required ( see appendix ) . in the final step
, alice sends mode @xmath7 to bob who superimposes it with his mode @xmath1 on another balanced beam splitter bs@xmath8 .
the presence of the entanglement between modes @xmath9 and @xmath10 is confirmed by the sufficient condition for entanglement @xcite @xmath24 where @xmath25 is a variable gain factor . minimizing the left - hand side of ineq .
( [ duan ] ) with respect to @xmath25 , we get fulfilment of the criterion for any @xmath26 , which confirms successful entanglement distribution . [ [ experiment . ] ] experiment .
+ + + + + + + + + + + the experimental realization is divided into three steps : state preparation , measurement , and data processing .
the corresponding setup is depicted in fig .
[ fig : setuppaper ] . from now on , we will work with polarization variables described by stokes observables ( see , e.g. , refs .
@xcite ) instead of quadratures .
we choose the state of polarization such that mean values of @xmath27 and @xmath28 equal zero while @xmath29 .
this configuration allows us to identify the `` dark '' @xmath30-@xmath31 plane with the quadrature phase space .
@xmath32 in this plane correspond to @xmath33 renormalized with respect to @xmath34 and can be associated with the effective quadratures @xmath35 .
we use the modified version of the protocol indicated in fig .
[ fig1 ] by the dashed arrow showing the alternative position of displacement in mode @xmath1 : the random displacement applied by david can be performed after the beam splitter interaction of @xmath1 and @xmath7 , even _ a posteriori _ after the measurement of mode @xmath10 .
this is technically more convenient and emphasizes that the classical information is sufficient for the entanglement recovery after the interaction of mode @xmath0 with mode @xmath2 and mode @xmath1 with mode @xmath7 .
david prepares two identically polarization squeezed modes @xcite and adds noise in the form of random displacements to the squeezed observables .
the technical details on the generation of these modes can be found in the appendix .
the modulation patterns applied to modes @xmath0 and @xmath2 to implement the random displacements are realized using electro - optical modulators ( eoms ) and are chosen such that the two - mode state @xmath36 is separable . by applying a sinusoidal voltage @xmath37 , the birefringence of the eoms changes at a frequency of 18.2mhz . in this way
, the state is modulated along the direction of its squeezed observable .
two such identically prepared modes @xmath0 and @xmath2 are interfered on a balanced beam splitter ( bs@xmath6 ) with a fixed relative phase of @xmath38 by controlling the optical path length of one mode with a piezoelectric transducer and a locking loop .
this results in equal intensities of both output modes . in the final step
, bob mixes the ancilla mode @xmath7 with a vacuum mode @xmath1 on another balanced beam splitter and performs a measurement on the transmitted mode @xmath10 . and
@xmath2 ) is modulated using eoms and sinusoidal voltages from a function generator ( dotted lines ) .
the hwps before the eoms are used to adjust the direction of modulation to the squeezed stokes variable , whereas the qwps compensate for the stationary birefringence of the eoms .
such prepared modes interfere with a relative phase of @xmath38 on a balanced beam splitter bs@xmath6 . in the last step of the protocol
, the mode @xmath7 interferes with the vacuum mode @xmath1 on a second balanced beam splitter bs@xmath8 .
measurement process : a rotatable hwp , followed by a ws and a pair of detectors , from which the difference signal is taken , allows us to measure all possible stokes observables in the @xmath30-@xmath31 plane . to determine the two - mode covariance matrix @xmath39 , all necessary combinations of stokes observables are measured . removing the second beam splitter of the state preparation allows us to measure the covariance matrix of the two - mode state @xmath40 .
data acquisition to achieve displacements of the modes in the @xmath30-@xmath31 plane we electronically mix the stokes signals with a phase matched electrical local oscillator and sample them by an analog - to - digital converter . ] the states involved are gaussian quantum states and , hence , are completely characterized by their first moments and the covariance matrix @xmath41 comprising all second moments ( see appendix ) . to study the correlations between modes @xmath9 and @xmath7 after bs@xmath6 , multiple pairs of stokes observables ( @xmath42 ) are measured .
the covariance matrix @xmath43 is obtained by measuring five pairs of observables : @xmath44 , @xmath45 , @xmath46 , @xmath47 , and @xmath48 , which determine all of its 10 independent elements . here
, @xmath49 is the angle in the @xmath30-@xmath31 plane between @xmath50 and @xmath51 . for the measurements of the different stokes observables , we use two stokes measurement setups , each comprising a rotatable half - wave plate , a wollaston prism , and two balanced detectors .
the difference signal of one pair of detectors gives one stokes observable @xmath52 in the @xmath30-@xmath31 plane , depending on the orientation of the half - wave plate .
the signals are electrically down - mixed using an electric local oscillator at 18.2mhz , which is in phase with the modulation used in the state preparation step . with this detection scheme
, the modulation translates to a displacement of the states in the @xmath30-@xmath31 plane .
the difference signal is low pass filtered ( 1.9mhz ) , amplified , and then digitized using an analog - to - digital converter card ( gage compuscope 1610 ) at a sampling rate of @xmath53 samples / s .
after the measurement process , we digitally low pass filter the data by an average filter with a window of 10 samples . because of the ergodicity of the problem , we are able to create a gaussian mixed state computationally from the data acquired as described above . by applying @xmath54 different modulation depths @xmath55 to each of the eoms we acquire a set of @xmath56 different modes . from this set of modes ,
we take various amounts of samples , weighted by a two - dimensional gaussian distribution .
the covariance matrix @xmath43 for the two - mode state after bs@xmath6 has been measured to be @xmath57 the estimation of the statistical errors of this covariance matrix @xmath43 can be found in the appendix . a necessary and sufficient condition for the separability of a gaussian state @xmath58 of two modes @xmath59 and @xmath60 with the covariance matrix @xmath61
is given by the ppt criterion @xmath62 where @xmath63 is the matrix corresponding to the partial transpose of the state @xmath58 with respect to the mode @xmath60 ( see appendix ) .
effects that could possibly lead to some non - gaussianity of the utilized states are also discussed in detail in the appendix .
the state described by @xmath43 fulfils the condition ( [ pptcondition ] ) as the eigenvalues ( 39.84 , 28.47 , 13.85 , and 9.371 ) of @xmath64 are positive ; hence , mode @xmath7 remains separable after bs@xmath6 . the measured two - mode covariance matrix of the output state @xmath39 is given by @xmath65 the statistical error of this measured covariance matrix is given in the appendix .
the separability is proven by the ppt criterion ( eigenvalues 28.24 , 21.79 , 8.646 , and 5.756 ) .
the postprocessing for the recovery of the entanglement is performed on the measured raw data of mode @xmath10 .
therefore , the displacement of the individual modes caused by the two modulators is calibrated . by means of this calibration ,
suitable displacements are applied digitally . the classical noise inherent in mode
@xmath10 is completely removed .
a part of the classical noise associated with @xmath66 is subtracted from @xmath67 , while the same fraction of the noise in @xmath68 is added to @xmath69 . in this way
, the noise partially cancels out in the calculation of the separability criterion ( [ duan ] ) and allows us to reveal the entanglement .
we chose the fraction as in eq .
( [ displacements ] ) , which is compatible with the separability of the transmitted mode @xmath7 from the subsystem @xmath70 in the scenario with modulation on mode @xmath1 before the beam splitter bs@xmath8 . only as bob receives the classical information about the modulation on the initial modes @xmath0 and @xmath2 from david is he able to recover the entanglement between @xmath9 and @xmath10
. bob verifies that the product entanglement criterion ( [ duan ] ) is fulfilled , as illustrated in fig .
[ fig : duan ] . that proves the emergence of entanglement .
the used gain factor @xmath25 considers the slightly different detector response and the intentional loss of 50% at bob s beam splitter . the clearest confirmation of entanglement @xmath71 is shown for @xmath72 ( fig . [ fig : duan ] ) .
this is the only step of the protocol , where entanglement emerges , thus demonstrating the remarkable possibility to entangle remote parties alice and bob solely by sending a separable auxiliary mode @xmath7 . and @xmath10 .
the experimental values for the criterion ( [ duan ] ) are depicted in dependence of the gain factor @xmath25 . because of the attenuation of mode @xmath1 by 50% , a gain factor of about 0.5 yields a value smaller than 1 , i.e. , below the limit for entanglement ( solid red line ) .
the inset zooms into the interesting section around the minimum .
the depicted estimated errors are so small because of the large amount of data taken . ]
[ [ discussion ] ] discussion + + + + + + + + + + the performance of the protocol can be explained using the structure of the displacements ( [ displacements ] ) .
entanglement distribution without sending entanglement highlights vividly the important role played by classical information in quantum information protocols .
classical information lies in our knowledge about all the correlated displacement involved .
this allows the communicating parties ( or david on their behalf ) to adjust the displacements locally to recover through clever noise addition quantum resources initially present in the input quantum squeezed states .
mode @xmath7 transmitted from alice to bob carries on top of the sub - shot - noise quadrature of the input squeezed state the displacement noise which is anticorrelated with the displacement noise of bob s mode .
therefore , when the modes are interfered on bob s beam splitter , this noise partially cancels out in the output mode @xmath10 when the light quadratures of both modes add .
moreover , the residual noise in bob s position ( momentum ) quadrature is correlated ( anticorrelated ) with the displacement noise in alice s position ( momentum ) quadrature in mode @xmath9 , again initially squeezed .
because of this , the product of variances in criterion ( [ duan ] ) drops below the value for separable states , and thus entanglement between alice and bob s modes emerges .
the difference between the theoretically proposed protocol @xcite and the experimental demonstration reported in this letter lies merely in the way classical information is used . in the original protocol , the classical information is retained by david and he is responsible for clever tailoring of correlated noise .
bob evokes the required noise cancellation by carrying out the final part of the global operation via superimposing his mode with the ancilla on bs@xmath8 . in the experimentally implemented protocol , david shares part of his information with bob , giving bob a possibility to get entanglement _ a posteriori _ , by using his part of the classical information after the quantum operation is carried out .
thus entanglement distribution in our case is truly performed via a dual classical and quantum channel , via classical information exchange in combination with the transmission of separable quantum states .
there are other interesting aspects to this protocol , which may open new , promising avenues for research .
noise introduced into the initial states by displacements contains specific classical correlations . on a more fundamental level
, these displacements can be seen as correlated dissipation ( including mode @xmath2 into the `` environment '' ) .
it is already known that dissipation to a common reservoir can even lead to the creation of entanglement @xcite .
our scheme can be viewed as another manifestation of a positive role dissipation may play in quantum protocols .
the presence of correlated noise results in nonzero gaussian discord at all stages of the protocol , a more general form of quantum correlations , which are beyond entanglement @xcite .
the role of discord in entanglement distribution has recently been discussed theoretically @xcite .
the requirements devised there are reflected in the particular separability properties of our global state after the interaction of modes @xmath0 and @xmath2 on alice s beam splitter .
the state @xmath73 contains discord and entanglement across @xmath74 splitting and is separable and discordant across @xmath75 splitting as required by the protocol .
our work thus illustrates an interplay of entanglement and other quantum correlations , such as correlations described by discord , across different partitions of a multipartite quantum system .
l. m. acknowledges project no .
p205/12/0694 of gar .
n. k. is grateful for the support provided by the a. von humboldt foundation .
the project was supported by the bmbf grant `` quorep '' and by the fp7 project qessence .
we thank christoffer wittmann and christian gabriel for fruitful discussions . c. p. and v. c. contributed equally to this work . +
[ [ note - added . ] ] note added .
+ + + + + + + + + + + recently , an experiment has been presented in ref .
@xcite , which is based on a similar protocol .
the main difference consists in the fact that it starts with entanglement which is hidden and recovered with thermal states . for this implementation
no knowledge about classical information has to be communicated to bob , besides the used thermal state .
by contrast the setup presented in this work exhibits entanglement only at the last step of the protocol .
thus both works give good insights on different aspects of the theoretically proposed protocol @xcite .
another independent demonstration of a similar protocol based on discrete variables was recently presented in ref @xcite . |
the general theory of relativity conveyed an understanding of phenomena such as the distortion of time - space by a gravitational or acceleration field@xcite .
however , the manner in which the curvature of space - time , that is the riemannian space , affects the electronic properties of condensed matters systems on a microscopic scale is largely unknown and due to its experimental accessibility of great interest@xcite .
riemannian geometric effects in a quantum system , which can either be free or constrained , stem from the dependance of the kinetic term on the metric of the embedding space or the metric of the sub - manifold onto which the quantum system is constrained by a confining potential ( rigid chemical bond ; electrostatic attraction ) .
the problem of constraining particle motion to a curved sub - manifold embedded in a euclidean space @xmath0 can be resolved in one of two alternative ways : i. ) in the _ intrinsic _ quantization approach , the motion is constrained to the curved sub - manifold in the first place .
a hamiltonian is constructed from generalized coordinates and momenta intrinsic to the sub - manifold and the system is quantized canonically . as a result ,
the embedding space is inaccessible and the quantum system depends only on the geometry intrinsic to the sub - manifold / hyper - plane@xcite ; ii . ) in the _ confining _ potential approach , a free in the embedding space quantum particle is subjected by a normal to the sub - manifold force that in effect confines the dynamics onto it .
the effective hamiltonian depends on the intrinsic geometry and on the way this sub - manifold is immersed in the embedding space . on one hand ,
the intrinsic quantization procedure is plagued with ordering ambiguities that allow for multiple consistent quantization procedures different by a term proportional to the curvature of the sub - manifold .
on the other hand , the confining potential procedure leads to a unique effective hamiltonian that depends on the constraint . in real microscopic quantum systems ,
constrained motion is a result of a strong confining force ( electrostatic ; rigid chemical bonds , ect . ) .
therefore , confining potential formalism seems a physically more realistic approach to constraints .
the non - relativistic quantum mechanics in a three dimensional hyper - plane of the four dimensional curved space - time can be treated in a well established manner@xcite . in this case
the embedding space - time is non - euclidean but equipped with a metric . this four - dimensional metric is related to the matter distribution by the einstein equation .
suppose the four dimensional space - time @xmath1 is topologically the product @xmath2 where @xmath3 represents a space - like three dimensional hyper - plane .
we can then foliate @xmath1 by a one parameter family of imbeddings given by the map @xmath4 @xmath5 such that @xmath6 that is @xmath7 is the image of the map @xmath8 in @xmath1 for a fixed time @xmath9 we assume that the leaves @xmath7 are space - like with respect to the metric in @xmath10 as a result , there exists in @xmath1 a time - like field normal to the leaves @xmath11 therefore there is a notion of future and past .
this time evolution vector field @xmath12 satisfies @xmath13 so that local coordinates @xmath14 ( satisfying @xmath15 for @xmath16 ) can be introduced . in effect , the space - time is splittable into 3 + 1 dimensions and the induced riemannian metric @xmath17 onto the three dimensional @xmath7 can be used to write the laplace - beltrami operator @xmath18 , which is the kinetic energy term in the schrdinger equation for the subjected to the geometric field quantum particle or quantum condensate @xmath19 the emergence of the geometric field from the kinetic term can be made clearer in the vicinity of the origin where the following taylor expansion of the induced metric in normal coordinates applies : @xmath20 @xcite and expanding the square root of the determinant of the metric yields : @xmath21 using a standard re - normalisation of the wave - function @xmath22 and keeping the lowest order terms ( the only relevant for the quantum dynamics ) in the taylor expansion we get for the kinetic term in the schrdinger equation @xmath23 here @xmath24 is the laplacian on flat space . adding an additional potential @xmath25 that may act in the system we convey the complete symbolic equation with which we will further work with @xmath26 here @xmath27 where @xmath28 is the three - dimensional ricci scalar curvature and @xmath29 is a numeric coefficient .
the emergence of a geometric potential from the kinetic term is obvious .
such a term is a standard coupling term between curvature and a quantum field in quantum field theory in curved space - time .
note , the particular form of the geometric potential may vary and in the case of a constraining potential approach takes the expressions : i. ) @xmath30 where @xmath31 is the principle curvature of a space curve embedded in @xmath32 ; ii . )
@xmath33 where @xmath34 for @xmath35 are the principle curvatures of a surface embedded in @xmath32 ; iii . ) @xmath36,$ ] where @xmath34 for @xmath37 are the principle curvatures of a three dimensional manifold embedded in @xmath38 . when electric field ( defined with the potential @xmath39 ) and magnetic field , defined through the vector potential @xmath40 are present the schrdinger equation takes the following form @xmath41
suppose , that we deal with a cooper pair condensate inside a superconductor .
the schrdinger equation for the cooper pair will be the above equation ( [ schrodinger&r ] ) with @xmath42 that is twice the charge of the electron .
this equation will describe the state of the entire condensate .
therefore , we may write @xmath43 where @xmath44 is the the charge density of the condensate and @xmath45 its phase . upon substitution of this form of the wave - function into ( [ schrodinger&r ] )
we can separate the real and imaginary part of the equation to arrive at slightly modified standard result : @xmath46 here @xmath47 is the current density , which in the case of a superconducting condensate stands also for the probability current .
the generalised momentum is contained in the expression for@xmath48 @xmath49 therefore the current density is just the velocity of the superconducting current times the charge density . taking the gradient of the whole equation ( [ eq : theta_t ] ) and expressing @xmath50 from ( [ eq : rho_t ] ) ( akin to @xcite )
we obtain the modified version of the hydrodynamic interpretation of the quantum condensate dynamics : @xmath51,\\ \nabla \times \vec{v}=-\frac{q}{m}\vec{b}\end{aligned}\ ] ] where @xmath52 is the lorentz force acting on the charged cooper pairs .
these two equations are the equations of motion of the superconducting cooper pair fluid in the presence of an induced from the embedding space - time curvature , which in this paper ofter refer to geometric field .
note , the geometric field enters the gradient of the mystical quantum mechanical potential , recognised by bohm as a unique interaction with the @xmath53field itself@xcite . as a result ,
the geometric field and the @xmath53field ( in view of bohmian quantum mechanics ) have similar and competing action .
next we recall the london equations for the quantum superconducting current density@xcite @xmath54 where @xmath55 ( see eq.(21.20 ) in @xcite ) for brevity we will call the introduced quantity @xmath56 which can be either of scalar or tensorial character , the polarisation operator . a correct microscopic theory of superconductivity can produce an expression for it in therms of the energy gap and critical temperature@xcite .
next , we divide both sides of ( [ eq : london ] ) by the current density @xmath57 and then differentiate with respect to time @xmath58 only to equate the r.h.s . of ( [ eq : dv / dt ] ) with the r.h.s of ( [ eq : london / rho ] ) @xmath59=\frac{d } { d t } m \frac{\hat{\pi } \vec{a}}{\rho}.\ ] ]
in the case when the superconducting state is robust , we may assume that @xmath60 the current density in the superconductor is approximately constant as well as the polarisation operator ( no internal changes in the microscopic mechanism ) .
the ideal diamagnetism of the superconducting state reduces the lorentz force to its electrostatic part , which is non - vanishing only in the case when a josephson junction is present ( two separated conducting domains at different electrostatic potentials ) . finally , the above simplifications yield @xmath61= - q\vec{e } + m \frac{\hat{\pi } } { \rho } \frac{d \vec{a}}{d t}.\ ] ] which is simply an expression for the conservation of energy .
let us take a line integral of the above along an open path from point a to point b @xmath62 next we introduce the geometric field energy @xmath63 next recall that @xmath64 that is @xmath65 is the electromotive potential difference between the two points and @xmath66 finally ( [ eq : nabla_r_2 ] ) can be re - written using ( [ eq : rho_t ] ) as @xmath67 \\ \nonumber & & -\frac{m}{q}\frac{\hat{\pi } } { \rho } \left [ \mathcal{e}_{ind}(b ) - \mathcal{e}_{ind}(a ) \right]+
\frac{m}{q } \frac{\hat{\pi } } { \rho } q \int_a^b ( \vec{v}.\nabla \vec{a } \ , ) .
d\vec{l}\end{aligned}\ ] ] now , suppose the copper pair charge velocity is constant at the two adjacent points , than the last integral quantity measures the difference in the interaction energy @xmath68 between the cooper pairs and the vector potential at the two points : @xmath69 introducing the electrostatic energy @xmath70 we can put ( [ eq : nabla_r_2 ] ) in its final form @xmath71 \\ & & \qquad = w_g(a ) - w_{stat}(a ) + \frac{m}{q}\frac{\hat{\pi } } { \rho } \left [ \mathcal{e}_{ind}(a ) - w_{int}(a ) \right].\end{aligned}\ ] ] as a result of introducing the scalar polarisation operator from ( [ eq : london ] ) and @xcite , the following conserved quantity at each material point of the superconductor emerges : @xmath72 here @xmath73 is the electrical energy of the cooper pairs .
in the previous section we have seen that the geometric field is equivalent to an electric field in the superconductor ( [ eq : nabla_r ] ) which statement is analogous to the law of conservation of energy ( [ eq : conservation ] ) .
therefore , provided the superconducting element is homogeneous , we can expect @xmath74 and as a result of the perfect conductor aspect of the superconducting state , we can also expect a redistribution of the shifted by the geometric field charges inside the superconductor in order to maintain the superconductor at a constant potential
. a completely different behaviour can be expected at the josephson junction .
we will discuss two cases of the junction , one between superconducting sides made from the same superconductor ( symmetric ) and one between two different superconductors ( asymmetric ) .
clearly , in the symmetric junction , the interaction energy @xmath75 will be the same on both sides .
however the electrostatic potential on the two sides can be different and the voltage drop @xmath39 can be equated to the geometric potential , that is the curvature scalar r itself . in effect , the difference in the geometric field between the two sides @xmath76 can produce a voltage drop at the junction ( direct effect ) or the voltage drop across the junction can produce curvature difference ( reversed effect ) : @xmath77 in this case , we may regard the josephson junction as a curvature - to - voltage converter with the following ratio ( @xmath78 is the free electron mass ) : @xmath79 \approx 6.3 \times 10^{20 } [ { \rm m}^{-2}].\end{aligned}\ ] ] note , the only difference the asymmetric junction can introduce is the difference in the interaction energy @xmath68 between the supercurrent and the electromagnetic field at the two sides of the junction .
suppose the supercurrent flows in the junction at vanishing potential difference @xmath80 , then the interaction energy gradient can produce rippling in the geometric field according to : @xmath81 the interaction energy ( [ eq : interaction ] ) is a function of the supercurrent drift velocity and one may view the asymmetric effect as produced by a sharp change in momentum , which converts into rippling of the geometric field .
we may expect a back - reaction on the entire junction as well .
the kinetic energy of the bulk material can certainly be included in ( [ eq : conservation ] ) .
the back reaction will increase with the increase in the difference between the interaction energy of the supercurrent with the electromagnetic field on both sides .
an argument that the josephson junction can act as a reversible curvature - to - voltage converter was presented in the previous section . in effect , the argument is prone to experimental testing and now we will discuss how and to what extent .
note , the conversion factor ( [ eq : conservation ] ) points to the impossibility to observe the travelling ripples in space - time , that is gravitational waves , with a josephson junction . along the span of the junction ( few angstrom [ @xmath82 )
the expected difference in the induced scalar curvature is very small @xmath83[m@xmath84 , therefore according to ( [ eq : conservation ] ) we may not hope for potential difference greater than @xmath85[v ] which is unmeasurable .
we are in a position to answer the questions from the introduction .
we are unable to measure the geometric field at the josephson junction .
however , since we expect that the effect is reversible , we may attempt to create a geometric field at the junction by an electric discharge between its sides . the greater the potential difference that can be created between the sides , the greater the geometric field that could be created .
we are unaware of the dynamics of the created geometric field and do not have any governing equations at the present stage of discussion on its propagation .
nevertheless , we can propose two detection methods which in theory can confirm the proposed effect .
the first approach stems from the reversibility argument .
suppose we have two junctions in close proximity .
we have no firm reason to choose a particular set - up , but choose to discuss the idea of the experiment with the two junctions placed along a line in such a way that the plane of the junctions ( the insulating layer ) is normal to the imaginary line connecting them .
both junctions should be magnetically and electrically shielded from each other and the surrounding environment
. one of them will serve as an emitter and the other as detector .
a high - voltage discharge should be conducted in the emitter and an induced voltage drop should be recorded at the detector .
provided such an electric potential difference is recorded via proper coincidence scheme ( an additional detector junction involved ) , we may confirm that the geometric effect at the josephson junction is a physical reality .
the second approach involves the ability of the geometric field to impart motion to objects with inertia .
the geometric field while a measure of the curvature of space - time is an acceleration field as well , or better induces force ( in the lab frame ) on a free object of inertia via the newton s second law @xmath86 where @xmath87 is the imparted acceleration .
we can give a rough estimate of the imparted acceleration in order to come up with an experimental procedure to verify the effect .
there are two possibilities to arrive at an estimate .
the first approach involves the use of the gauss s law for gravitation @xmath88 and the @xmath89 component of einstein s field equations for a perfect fluid ( where @xmath90 ) : @xmath91 here @xmath92 is the velocity of light in vacuum , @xmath93 is newton s gravitation constant and @xmath94 is the matter density .
next we make use of an exact result valid for a 3 + 1 decomposition of space - time:@xmath95 where @xmath28 is the scalar ricci curvature of the three - dimensional hyper - surface@xcite . combining these relations we end up with @xmath96 that is equivalent to @xmath97 in the case when the imparted acceleration is in one dimension along the span of @xmath98 interestingly
, a similar result is obtained if one takes up the geodesic deviation equation @xmath99 where @xmath100 is the 4-velocity of an object travelling along the geodesic and @xmath101 is the deviation vector .
note , the riemann tensor enters the relation directly .
the geodesic deviation measures the acceleration with which two neighbouring geodesics deviate from each other in the curved geometry .
suppose @xmath100 is a unit vector in the time direction and @xmath102 where @xmath103 is a constant , then the geodesic deviation equation in one dimension is approximately @xmath104 upon a substitution of the riemann tensor component with the ricci scalar curvature .
most importantly , the two expressions agree in the order of magnitude and the expected acceleration a geometric field pulse with a magnitude of @xmath105 $ ] and a width of the size of the josephson junction @xmath106 $ ] can impart on an object along its path is enormous @xmath107 the dislocation such a pulse can cause is proportional to the time of its duration squared .
however , we have no estimate of this quantity , but given the large value for the acceleration even a femtosecond pulse can lead to substantial dislocation of the order of meters .
we also have no estimate of the spread with distance of this geometric field pulse and believe the assumed value for the scalar curvature and imparted to a material body acceleration to be largely exaggerated .
now suppose we conduct a high - voltage discharge in a josephson junction and try to measure the dislocation of a mirror , mechanically shielded from the junction .
such a dislocation will be induced by the emitted at the junction geometric field .
the position of the mirror with respect to the junction is unknown , therefore few geometric set - ups should be tried .
next , in order to increase the sensitivity of the experiment , we suggest the inclusion of the detection mirror as a primary or secondary mirror in a michelson interferometer . provided a dislocation in the mirror
is recorded we may confirm the generation of a geometric field in a josephson junction .
in conclusion we would like to point out the origin of the geometric potential from the kinetic term of a constrained to a curved three - dimensional hyper - plane of space - time quantum mechanical condensate ( superconductor ) .
this potential makes its way into the hydrodynamic interpretation of the schrdinger equation and enters it on an equal footing with the bohm s `` quantum potential . ''
when external electromagnetic field is included in the dynamics and a suitable simplifications applied , one is able to derive an obvious energy conservation relation at every material point in the superconductor .
this conservation relation includes a geometric field part associated with the curvature of the hyper - plane .
it turns out that at a tunnelling junction ( josephson junction ) the energy conservation relation implies the possibility to transform electric energy into geometric energy , that is create curvature in the hyper - plane and vice versa . in effect , it turns out that the josephson junction can act as a voltage - to - curvature converter .
experimental procedures are discussed in hope the present study invites experimental effort to verify the effect .
77 in fact , riemann used the expansion of a metric in normal coordinates to originally define the curvature tensor . in normal coordinates there is no necessity to distinguish between co- and contra- variant indices .
see u. muller , c. schubert and a.e.m .
van de ven , gen . rel . and grav .
, * 31 * , 1759 ( 1999 ) ; g. herglotz , math . ann . * 93 * 46 ( 1925 ) . | a geometric potential from the kinetic term of a constrained to a curved hyper - plane of space - time quantum superconducting condensate is derived .
an energy conservation relation involving the geometric field at every material point in the superconductor is demonstrated . at a josephson junction
the energy conservation relation implies the possibility to transform electric energy into geometric field energy , that is curvature of space - time .
experimental procedures to verify that the josephson junction can act as a voltage - to - curvature converter are discussed .
quantum fields in curved spacetime , tunneling phenomena ( josephson effects ) , superconductor response to electromagnetic fields , josephson devices _ pacs _ : 04.62.+v , 03.65.-w , 74.50.+r , 74.25.n- , 85.25.cp the success of the laser interferometer gravitational wave observatory ( ligo ) and its sister collaboration , virgo@xcite , in observing the geometric field s ripples in space - time has in addition to opening a new experimental method of astrophysical observation , proved a practical scheme for measuring the space - time metric , that is ligo had seen the death spiral of a pair of black holes , through the dislocation and the associated change in optical ( and physical ) path travelled by light beams in the arms of an interferometer as the gravitational perturbation travels between the reflection points . in theory ,
the geometry of space - time , that is the components of the riemann curvature tensor can be reconstructed by taking measurements of the deviation of two adjacent light paths ( geodesics)@xcite .
having in mind this spectacular success in experimental astrophysics , we pose the following questions : _ i. ) are there other experimentally relevant strategies for detecting space - time geometry ?
ii . ) is there other extremely sensitive measurement technique besides two light beam interference that can potentially sense the geometric field ? _
the present paper aims at answering both questions and is organised as follows .
the first section discusses the appearance of a geometric potential term from the kinetic energy term in the schrdinger equation . in order to improve on the detection possibilities we assume this constrained to a hyper - plane quantum dynamics to concern a superconducting condensate .
the second section focuses on the hydrodynamic interpretation of the governing quantum equation and reveals that the geometric field enters it on an equal footing with the `` quantum potential '' ( in bohm s views ) thus making its way as a real force moving the condensate superfluid .
the third section contains the proof that the emergent geometric field enters an energy conservation relation valid at each point of the superconducting condensate .
based on this conservation relation applied to a josephson junction , an experimentally verifiable voltage - to - curvature conversion effect is proposed in the fourth section .
the fifth section discusses possible experimental methodologies to test the reality of the effect . |
* fig.1 * + rheed patterns of the fe bottom layer for ( a ) carbon free fe and ( b ) fe / fe - c along the [ 110 ] crystallographic direction .
additional pattern for fe / fe - c surface demonstrate the @xmath5 reconstruction related to carbon . * fig.2 * + color online : ( a ) dynamic conductivities in p ( top panel ) and ap ( bottom panel ) magnetization states at 300k ( open circles ) and 10k ( full circles ) .
top panel inset : tmr curves at 300k ( red open circles ) and 4k ( black full circles ) .
( b ) bulk band structure diagram of bcc fe . * fig.3 * + color online : ( a ) dynamic conductivities at 10k ( top panel ) and related tmr(v)(bottom panel ) . ( b ) shot noise measurements in p and ap states measured at 10k in bias when the electrons are injected from the top toward the bottom mtj electrode ( negative voltage in fig . 2 and 3.a ) . | low temperature ( 10k ) high voltage bias dynamic conductivity ( up to 2.7v ) and shot noise ( up to 1v ) were studied in epitaxial fe(100)/fe - c / mgo(100)/fe(100 ) magnetic tunnel junctions , as a function of the magnetic state .
the junctions show large tunnel magnetoresistance ( 185% at 300k and 330% at 4k ) .
multiple sign inversion of the magnetoresistance is observed for bias polarity when the electrons scan the electronic structure of the bottom fe - c interface .
the shot - noise shows a poissonian character .
this demonstrates a pure spin dependent direct tunneling mechanism and validates the high structural quality of the mgo barrier .
magnetic tunnel junctions ( mtjs)@xcite are nowadays one of the most active areas of material science and spintronics .
recent theoretical predictions@xcite and experimental demonstration@xcite of coherent spin - dependent tunneling in single crystal fe(100)/mgo(100)/fe(100 ) mtjs revolutionized this area providing new ways to create devices with room temperature tunneling magneto - resistance ( rt - tmr ) exceeding 100% .
the large tmr at low bias is mostly due to fully spin polarized @xmath0 bulk electron states in fe(001 ) , reflected for antiparrallel ferromagnetic electrodes configuration ( ap ) or well transmitted for the parallel ( p ) state@xcite .
however , the tunneling mechanism gets more complex when taking into account the electronic structure of the interfaces@xcite and when biasing the junction .
therefore , for finite bias polarities the antiparallel conductance may exceed the parallel one , resulting in tmr suppression@xcite or its sign reversal@xcite . by engineering the chemical and electronic structure of the fe / mgo interface ,
the voltage variation of the tmr in amplitude and sign can be skilfully manipulated .
it has been recently demonstrated that the carbon - doping of the bottom fe / mgo interface leads to strongly asymmetric tmr vs bias , providing a root for creation of high - output voltage device applications@xcite .
our letter presents a first study of dynamical conductance and tmr in a large bias window , up to 2.7 v , for fe(100)/fe - c / mgo(100)/fe(100 ) mtjs . the shot noise analysis in different magnetization configuration
is performed at voltages up to 1v .
the experiments are done at room temperature ( 300k ) and low temperature ( 4k-10k ) .
the measured tmr ratio increases from 185% at 300k to 330% at 4k , mostly due to the strong temperature variation of the tunnel conductivity in the antiparallel ( ap ) state .
moreover , our tunneling spectroscopy experiments show a clear maximum in the ap conductivity for a finite bias and a multiple tmr sign inversion .
these experiments demonstrate the role of the minority spin fe interface resonance state ( irs ) to the tunneling .
furthermore , in both parallel ( p ) and antiparallel magnetization configuration , the shot noise measurements demonstrate an uncorrelated direct tunneling mechanism across the mgo barrier .
the shot noise analysis and the large breakdown voltage of the junctions ( up to 3v ) demonstrates the high quality of our mgo barriers ( i.e. absence of defects such as oxygen vacancies ) .
our epitaxial fe(45nm)/ mgo(3nm)/ fe(10nm)/ co(20nm)/ pd(10nm)/ au(10 nm ) samples were grown by molecular beam epitaxy ( mbe ) on mgo(100 ) substrates under uhv condition ( @xmath1 mbar base pressure ) .
different coercive field of the mtj electrodes is obtained by hardening the top bcc fe electrode with hcp co epitaxially grown with in - plane c axis .
prior to deposition , the substrate is annealed at @xmath2 , then the layers are grown at room temperature . for flattening ,
the fe electrodes are annealed to @xmath3 ( bottom fe ) and @xmath4 ( top fe ) . following the growth procedure@xcite ,
two different samples can be grown : samples with clean fe / mgo bottom interfaces and samples with carbon doping at bottom fe / mgo interface ( fe / fe - c / mgo ) .
the reflection high - energy electron diffraction ( rheed ) analysis performed on each layer of the mtj stack allows a direct control of the epitaxial growth and the high crystalline quality of the epitaxial layers . within the pseudomorphical growth regime ,
the two - dimensional layer - by - layer growth of the mgo barrier is monitored by the intensity oscillations in the rheed patterns .
compared to clean samples , in the samples with carbon the bottom fe(001 ) electrode presents a @xmath5 surface reconstruction , as shown by the rheed picture ( fig .
1 ) . after the growth of the multilayer stack , mtjs with micrometric lateral size have been patterned using standard optical lithography/ ion etching process .
all the mtjs studied here contain carbon doped fe / mgo interface .
they have shown a large voltage stability , up to 3 volts .
dynamic conductance @xmath6 and shot - noise bias dependence have been studied using four - probe method with a set - up allowing to vary the temperature between 2 and 300k , equipped with preamplifiers situated on top of the cryostat .
two different techniques were employed to measure dynamic conductance in p or ap states , providing nearly identical results .
the first one uses a current - voltage ( @xmath7 ) converter and a voltage amplifier . in this case , the mtj is biased using a constant _ dc _ voltage with superimposed low amplitude sinusoidal wave ( @xmath8 ) .
the voltage drop on the junctions and the current were obtained by using an analogue - digital converter ( adc ) and a lock - in amplifier , providing the dynamic conductance .
the second technique , mainly employed at high bias , uses square current wave of current superimposed on _ dc _ current .
shot noise measurements were done using a cross - correlation technique .
more details of experimental setup were published elsewhere@xcite . at 300k ,
the fe / fe - c / mgo(3nm)/fe / co mtjs show @xmath9 product values ( rt ) ranging from 0.42 to 0.48 m@xmath10.@xmath11@xmath12 .
the inset in the top panel of fig .
2 shows typical tmr curves measured at 10mv either at 300k and at 4k .
the large tmr ratio of 185% at 300k indicates the high quality of the mtjs .
interestingly , the low temperature tmr ( @xmath13 330% ) notably exceed previously reported ( 250% ) maximum values of zero - bias tmr in epitaxial fe(100)/mgo / fe mtjs with undopped fe / mgo interfaces@xcite .
the temperature variation of the tmr is understood from the dynamic conductivity experiments @xmath14 shown in fig .
2(a ) which plots @xmath6 at 300 and 10k within a voltage range of 0.8v .
firstly , asymmetric g(v ) characteristics in positive and negative voltage demonstrate different electronic structure of the top and bottom electrodes and fe / mgo interfaces@xcite .
secondly , we remark significantly different temperature variation of conductivity in p and ap magnetization configurations . in the ap configuration ( fig .
2(b ) bottom panel ) , we remark almost no temperature dependent shape variation , except the enhancement of low bias anomaly at 10k .
however , we notice a strong reduction of @xmath15 by 50% at low temperature . on the other hand , a net temperature dependent shape variation between 300k and 10k ( fig .
2(a ) top panel ) is clearly seen for @xmath16 .
interestingly , the zero bias @xmath17 is mostly constant with temperature ( only 2% variation ) .
additional local minima appear at 10k for both positive and negative finite bias voltage . at low temperature ,
all studied mtjs reveal novel p - state low - bias conductance oscillations with about 4 minima ( top panel of fig .
2 ) .
we note that low - bias conductivity minima in the p state have been already observed in carbon free samples even at 300k .
however , we always measured only two local conductance minima@xcite .
these minima were explained by the @xmath18 majority electron contribution to the total conductivity at low voltage ( @xmath19 which is the top of the majority @xmath18 band ( fig .
2(b ) ) .
the origin of low temperature @xmath16 minima observed in fe / fe - c / mgo / fe mtjs opens interesting theoretical perspectives .
these calculations should investigate in detail the effect of the realistic electronic structure of the fe / fe - c / mgo ( i.e. effects of fe - c bounding ) on the tunneling , in the low bias regime .
figure 3(a ) presents high bias conductance for voltages up to 2.7v , measured at 10k .
the influence of joule heating ( few k ) on the @xmath20 is neglected due to the rather weak observed low temperature temperature dependence of both @xmath17 and @xmath21 .
interestingly , while @xmath22 is rather symmetric , in negative voltage when the electrons tunnel into the bottom fe - c / mgo electrode , the @xmath15 shows a strong asymmetric local maximum superimposed on roughly parabolic background .
this local resonant increase of the @xmath21 ( @xmath23 ) in a narrow@xcite energy window will lead to the lower voltage sign reversal of the tmr ( fig .
3(a ) , bottom panel).similar to scanning tunneling spectroscopy experiments@xcite , and as we already previously shown@xcite , the resonant enhancement of @xmath21 is attributed to the contribution to the tunneling of the fe minority interfacial resonance ( irs ) .
however , we only observed this phenomena in carbon free fe / mgo / fe samples with thinner mgo barrier , where the fe irs still significantly contribute to the tunneling@xcite . in the samples studied here , having carbon at the fe - c / mgo interface , an important effect of the @xmath21 resonant activation by irs
is observed even for 3 nm thick mgo barriers . to elucidate this interesting property , theoretical investigation of two effects
is in progress : ( i ) the effect of fe - c - mgo bounding on the minority spin fe(001 ) irs ( i.e. shift in energy , dispersion in k ) ; ( ii ) the carbon induced periodical perturbation of the potential at the bottom fe / mgo interface ( i.e. @xmath5 reconstruction , fig .
1 ) induces scattering events which change k - vector .
this has direct consequences on the total conductivity . in positive bias ,
when electrons are injected toward the top electrode , the low bias tmr changes the sign above 1.5v .
this is determined by the @xmath21 strong enhancement when , in the ap configuration the injected @xmath0 electrons from the bottom fe electrode arrive as hot electrons in the top electrode and find an equivalent symmetry in the minority band . in negative voltage , when electrons tunnel into the bottom fe - c / mgo electrode , similar contribution of the minority @xmath0 symmetry to the conductivity is expected .
however , the tmr second sign reversal seems to appear at much higher voltages , above 2.5v ( fig .
3(a ) , bottom panel ) .
one possible reason would be the reduction of the hot electrons thermalization length in the bottom electrode . the effect of the irs at fe - c / mgo interface on this phenomena requires further theoretical investigation .
figure 3(b ) presents shot noise measurements carried out at t = 10k on fe / fe - c / mgo / fe mtjs , with bias direction corresponding to injection of electrons from the top to the bottom ( carbon doped ) fe / mgo interface . for comparison ,
the solid curves shows the theoretical expectation for the shot noise , for electron tunneling having poissonian character : @xmath24 with @xmath25 the dynamic conductivity ( fig .
3(a ) ) and @xmath26 the applied current . within the error bars ,
showing dispersion of the shot noise white spectrum in the khz range , the experimental data clearly indicate the absence of electron correlations and/or sequential tunneling phenomena .
this proves that both p and ap spin dependent conductance s and the shot noise are due to direct tunneling between electron bands , as expected for the coherent tunneling@xcite .
the absence of resonant assisted tunneling in the shot noise demonstrates the high quality of our epitaxial mgo barriers ( i.e. the absence of oxygen vacancies ) .
this high quality is furthermore confirmed by the large breakdown voltage of the mtjs ( up to 3v ) .
the authors thank g. lengaigne for technological process of mtjs and r.villar for discussions and continious support .
20 j.s .
moodera , l.r .
kinder , t.m .
wong , r. meservey , phys.rev.lett .
* 74 * , 3273 ( 1995 ) .
t. miyazaki , n. tezuka , j. magn .
magn . mat .
* 139 * , l231 ( 1995 ) .
w.h.butler , x.g.zang , t.c.schulthess , j.m.maclaren , phys . rev.b**63 * * , 054416 ( 2001 ) .
j.mathon , a.umerski , phys.rev.b**63 * * , 220403r ( 2001 ) .
m. bowen , v. cros , f. petroff , a. fert , c. martnez boubeta , j.l .
costa - krmer , j.v .
anguita , a. cebollada , f. briones , j.m .
de teresa , l. morelln , m.r .
ibarra , f. gell , f. peir , a.cornet , appl.phys.lett .
* 79 * , 1655 ( 2001 ) .
j.faure-vincent , c.tiusan , e.jouguelet , f.canet , m.sajieddin , c. bellouard , e.popova , m.hehn , f.montaigne , a. schuhl , appl.phys.lett . *
82 * 4507 ( 2003 ) .
s.s.p .
parkin , c. kaiser , a. panchula , p.m. rice , b. hughes , m.samant , s.h.yang , nature mat . * 3 * , 862 ( 2004 ) .
s. yuasa , t. nagahama , a. fukushima , y. suzuki , k. ando , nature mat . * 3 * , 868 ( 2004 ) .
c.tiusan , m.sicot , m.hehn , c.bellouard , s.andrieu , f.montaigne , a.schuhl , appl.phys.lett . *
88 * , 62512 ( 2006 ) .
c.tiusan , j.faure-vincent , c.bellouard , m.hehn , e.jouguelet , a.schuhl , phys.rev.lett . *
93 * , 106602 ( 2004 ) .
r.guerrero , f.g.aliev , r.villar , j.hauch , m.fraune , g.gntherodt , k.rott , h.brckl , g. reiss , appl.phys.lett . * 87 * , 042501 ( 2005 ) .
r.guerrero , f.g.aliev , y.tserkovnyak , t.s.santos , j.s.moodera , phys.rev.lett . *
97 * , 0266602 ( 2006 ) .
c.tiusan , f.greullet , m.hehn , f.montaigne , s.andrieu , a.schuhl , j. phys .
: cond.mat .
* 19 * , 165201 ( 2007 ) .
j.a.stroscio , d.t.pierce , a.davies , r.j.celotta , m.weinert , phys.rev.lett . * * 7**5 ( 16 ) , 2960 ( 1995 ) .
the interfacial resonance state of fe has a @xmath27 orbital character and belongs to the @xmath0 symmetry .
however , due to the strong localization of the @xmath27 band , this state decays rapidly with the mgo thickness .
therefore , for large mgo thickness its contribution to the total ap conductivity is expected to be small .
the spatial variation of the realistic electronic structure of the fe - c / mgo interface ( related to realistic morphology ) will lead to a distribution in energy of the interfacial resonance .
ya.m .
blanter , m. bttiker , phys . rep .
* 336 * , 1 ( 2000 ) . |
there are some ways to define the frlicher nijenhuis bracket in non - commutative differential geometry .
the frlicher nijenhuis bracket on the algebra of universal differential forms of a non - commutative algebra , is presented in @xcite , the frlicher nijenhuis bracket in several kinds of differential graded algebras are defined in @xcite and the frlicher nijenhuis bracket on colour commutative algebras is defined in @xcite . but this notion is not defined on @xmath0-algebras in the context of non - commutative geometry . in this paper
we introduce the frlicher nijenhuis bracket on a @xmath0-algebra @xmath1 using the algebra of universal differential forms @xmath2 .
a @xmath0-algebra @xmath1 over the field @xmath3 ( @xmath4 or @xmath5 ) is a @xmath6-graded algebra ( @xmath6 is a commutative group ) together with a twisted cocycle these algebras were defined for the first time in the paper @xcite and are generalizations of usual algebras ( the case when @xmath6 is trivial ) and of @xmath7 ( @xmath8)-superalgebras ( the case when @xmath6 is @xmath7 resp .
@xmath8 ) .
our construction of the frlicher nijenhuis bracket for @xmath0-algebras , in this paper , is a generalization of this bracket from @xcite . in 2
we present a class of non - commutative algebras which are @xmath0-algebras , derivations and bimodules . in
3 we define the algebra of ( non - commutative ) universal differential forms @xmath9 of a @xmath0-algebra @xmath1 . in
4 we present the frlicher nijenhuis calculus on @xmath1 , the nijenhuis algebra of @xmath1 , and the frlicher nijenhuis bracket on @xmath1 .
we also show the naturality of the frlicher nijenhuis bracket .
in this section we present a class of non - commutative algebras that are @xmath0-algebras . for more details see @xcite .
let @xmath6 be an abelian group , additively written , and let @xmath1 be a @xmath6-graded algebra .
this implies that the vector space @xmath1 has a @xmath6-grading @xmath11 , and that @xmath12 @xmath13 .
the @xmath6-degree of a ( non - zero ) homogeneous element @xmath14 of @xmath1 is denoted as @xmath15 .
futhermore let be a map which satisfies @xmath16 & \rho ( a+b , c ) = \rho ( a , c)\rho ( b , c ) , \quad a , b , c\in g.\end{aligned}\ ] ] this implies @xmath17 , @xmath18 and @xmath19 for all @xmath20 , @xmath21 .
we define for homogeneous elements @xmath14 and @xmath22 in @xmath1 an expression , which is @xmath0-commutator of @xmath14 and @xmath22 as @xmath23_{\rho } = fg-\rho ( \left| f\right| \left| g\right|)gf.\ ] ] this expression as it stands make sense only for homogeneous elements @xmath14 and @xmath22 , but can be extended linearly to general elements
. a @xmath6-graded algebra @xmath1 with a given cocycle @xmath0 will be called @xmath0-commutative if @xmath24 for all homogeneous elements @xmath14 and @xmath22 in @xmath1 . @xmath25 1 .
any usual ( commutative ) algebra is a @xmath0-algebra with the trivial group @xmath26 2 .
let @xmath27 ( @xmath28 ) be the group and the cocycle @xmath29 , for any @xmath30 .
in this case any @xmath0-(commutative ) algebra is a super(commutative ) algebra .
the @xmath31-dimensional quantum hyperplane @xcite @xmath32 , is the algebra generated by the unit element and @xmath31 linearly independent elements @xmath33 satisfying the relations : @xmath34 for some fixed @xmath35 @xmath36 .
@xmath32 is a @xmath37-graded algebra , i.e. , @xmath38 with @xmath39 the one - dimensional subspace spanned by products @xmath40 .
the @xmath37-degree of these elements is denoted by @xmath41 define the function as @xmath42 with @xmath43 for @xmath44 , 0 for @xmath45 and @xmath46 for @xmath47 .
it is obvious that @xmath32 is a @xmath0-commutative algebra .
the algebra of matrix @xmath48 @xcite is @xmath0-commutative as follows : + let @xmath49 @xmath50 , where @xmath51 , @xmath52 .
then @xmath53 and @xmath54 is generated by the set @xmath55 .
it is easy to see that @xmath56 and @xmath57 for any @xmath58 .
let @xmath59 , @xmath60 and @xmath61 . if we denote @xmath62 then @xmath63 , for any @xmath64 , @xmath65 .
it is obvious that the map , @xmath66 is a cocycle and that @xmath67 is a @xmath0-commutative algebra .
let @xmath68 be an element of the group @xmath6 .
a @xmath0-derivation @xmath69 of @xmath1 , of degree @xmath68 is a bilinear map of @xmath6-degree @xmath70 i.e. , such that one has for all elements @xmath71 and @xmath72 , @xmath73 without any difficulties it can be obtained that if algebra @xmath1 is @xmath0-commutative , @xmath74 and @xmath69 is a @xmath0-derivation of degree @xmath68 , then @xmath75 is a @xmath0-derivation of degree @xmath76 and the @xmath6-degree @xmath77 i.e. @xmath78 and we say that is a @xmath0-derivation if it has degree equal to @xmath6-degree @xmath70 i.e. and @xmath79 for any @xmath74 and @xmath72 .
it is known @xcite that the @xmath0-commutator of two @xmath0-derivations is again a @xmath0-derivation and the linear space of all @xmath0-derivations is a @xmath0-lie algebra , denoted by @xmath0-der@xmath1 .
one verifies immediately that for such an algebra @xmath1 , @xmath0-der@xmath1 is not only a @xmath0-lie algebra but also a left @xmath1-module with the action of @xmath1 on @xmath0-der@xmath1 defined by @xmath80 let @xmath81 be a @xmath6-graded left module over a @xmath0-commutative algebra @xmath1 , with the usual properties , in particular @xmath82 for @xmath83 . then @xmath81 is also a right @xmath1-module with the right action on @xmath81 defined by @xmath84 in fact @xmath81 is a bimodule over @xmath1 , i.e. @xmath85 let @xmath81 and @xmath31 be two @xmath6-graded bimodules over the @xmath0-algebra @xmath1 .
let be an @xmath1-bimodule homomorphism of degree @xmath86 if such that @xmath87 and @xmath88 for any @xmath89 and @xmath90 .
we denote by hom@xmath91 the space of @xmath1-bimodule homomorphisms of degree @xmath68 and by hom@xmath92hom@xmath93 the space of all @xmath1-bimodule homomorphisms .
@xmath1 is a @xmath0-algebra as in the previous section .
we denote by @xmath94 the space generated by the elements : _ adb _ of @xmath6-degree @xmath95 with the usual relations : @xmath96 where 1 is the unit of the algebra @xmath1 .
if we denote by @xmath97 then @xmath98 is an @xmath1-bimodule and satisfies the following theorem of universality . for any @xmath1-bimodule @xmath81 and for any derivation of degree @xmath70
there is an @xmath1-bimodule homomorphism of degree @xmath70 @xmath99hom@xmath100 @xmath101 such that @xmath102 .
the homomorphism is uniquely determined and the corresponding @xmath103 establishes an isomorphism between @xmath0-@xmath104 and hom@xmath105 .
we define the map by @xmath106 which transform the usual leibniz rule for the operator @xmath107 into the @xmath0-leibniz rule for the deriva - tion @xmath69.@xmath108 starting from the @xmath1-bimodule @xmath109 and the @xmath0-algebra @xmath110 we build up _ the algebra of differential forms over _ @xmath1
. this algebra will be a new @xmath111-algebra @xmath112 graded by the group @xmath113 and generated by elements @xmath114 of degree @xmath115 and their differentials @xmath116 of degree @xmath117 .
we will also require the universal derivation which can be extended to a @xmath111-derivation of the algebra @xmath9 of degree @xmath118 in such a way that @xmath119 and @xmath120 denote by @xmath121 the product of forms @xmath122 @xmath123 in the algebra @xmath124 .
then @xmath125 and @xmath126 hence @xmath127 from these relations it follows that @xmath128 where is the group homomorphism @xmath129 . from the properties of the cocycle @xmath0 , @xmath130 for any @xmath131 and @xmath132
. @xmath25 let @xmath1 be a @xmath0-algebra with the cocycle @xmath0 .
then any cocycle @xmath111 on the group @xmath133 with the conditions @xmath134 and @xmath135 are given by @xmath136 for some homomorphism we will denote below @xmath137 or simply @xmath138 the @xmath133-graded algebra of forms with the cocycle @xmath111 and the derivation @xmath139 of degree ( 1 , 0 ) . therefore for any @xmath0-algebra @xmath1 , a group homomorphism and an element @xmath86 , we have the complex : @xmath140 .
the cohomology of this complex term @xmath141 is denoted by @xmath142 and will be called as the _ de rham cohomology of the _ @xmath0-__algebra__@xmath1 .
@xmath25 let be a homomorphism of degree @xmath86 between the @xmath6-graded @xmath0-algebras .
there is a natural homomorphism which in degree @xmath143 is and has the @xmath144-degree @xmath145 given by @xmath146
here we present the frlicher nijenhuis calculus over the algebra of forms defined in the previous section .
denote by der@xmath147 the space of derivations of degree @xmath148 i.e. an element @xmath149 der@xmath150 satisfies the relations : 1 .
@xmath151 is linear , 2 .
the @xmath144-degree of @xmath151 is @xmath152 , and 3 .
@xmath153 for any @xmath154 .
the space @xmath111-der@xmath155der@xmath156 is a @xmath111-lie algebra with the bracket @xmath157=d_1\circ d_2-\overline{\rho } ( \left| d_1\right|,\left| d_2\right|)d_2\circ d_1 $ ] .
let us denote by hom@xmath158-der @xmath159 the isomorphism from theorem 1 .
we also denote by @xmath160 hom@xmath161 the space of fields of the algebra @xmath1 .
then -der@xmath162 is an isomorphism of vector @xmath6-graded spaces .
the space of @xmath0-derivations @xmath0-der@xmath159 is a lie @xmath0-algebra with the @xmath0-bracket @xmath163 $ ] , and so we have an induced @xmath0-lie bracket on @xmath164 which is given by @xmath165)=[\mathcal{l}_{x},\mathcal{l}_{y}]= \mathcal{l}_{x}\mathcal { l}_{y}-\rho ( \left\vert x\right\vert,\left\vert y\right\vert)\mathcal{l } _ { y}\mathcal{l}_{x } \label{12}\ ] ] and will be referred to as the @xmath0-lie bracket of fields .
each field @xmath166 is by definition an @xmath1-bimodule homomorphism @xmath167 and it prolongs uniquely to a graded @xmath168-derivation of degree @xmath169 by@xmath170 j_x(\omega ) & = x(\omega ) \quad \hbox { for } \
\omega \in \omega ^1(a)\end{aligned}\ ] ] and @xmath171 & \quad\ = \sum_{i=1}^{k-1 } \overline{\rho } \left ( \left(-1 , |x|\right ) , \left(i - 1 , \sum_{j=1}^{i-1 } |\omega_i|
\right ) \right ) \omega _ 1\wedge\cdots \wedge \omega_{i-1 } \wedge x(\omega _ i)\\[.2pc ] & \qquad\ \
\times \omega _ { i+1}\wedge\cdots \wedge \omega _ k + \overline{\rho } \left(\left(-1 , |x|\right ) , \left(k - 1 , \sum_{j=1}^{k-1 } & \qquad\ \ \times \omega _ 1\wedge\cdots \wedge \omega _ { k-1}x(\omega _ k)\end{aligned}\ ] ] for any @xmath172 . the @xmath111-derivation @xmath173 is called the contraction operator of thefield @xmath69 .
this is an easy computation.@xmath108 with some abuse of notation we also write @xmath174 for @xmath175 and @xmath176hom@xmath177 .
a @xmath111-derivation @xmath149 der@xmath178 is called @xmath179 if @xmath180 . then @xmath181 and @xmath182 for any @xmath89 and @xmath183 .
it results that @xmath151 is an @xmath1-bimodule homomorphism .
we denote by hom@xmath184 the space of @xmath1-bimodule homomorphisms from @xmath185 to @xmath186 of degree @xmath148 .
then an algebraic derivation @xmath151 of degree @xmath148 is from hom@xmath187 .
we denote by @xmath111-der@xmath188 the space of all @xmath111-algebraic derivations of degree @xmath148 from @xmath9 .
since @xmath151 is a @xmath111-derivation , @xmath151 has the following expression on the product of 1-forms @xmath189 : @xmath190 and the derivation @xmath151 is uniquely determined by its restriction on @xmath191 , @xmath192 we write @xmath193 to express this dependence .
note that @xmath194 for @xmath195 .
next we will use the following notations : @xmath196 \omega _ { * } ^1 = \omega _ { * } ^1(a ) & = \mathop{\oplus}\limits_{k\geq 0,\alpha \in g } \omega _ { ( k,\alpha)}^1(a).\end{aligned}\ ] ] elements of the space @xmath197 will be called _ field_-_valued _
@xmath148-_forms_. the map -der@xmath198 , @xmath199 defined by @xmath200 & \quad\ = \sum_{i=1}^{k } \overline{\rho } \left((k+1,\alpha ) , \left(i -1 , \sum_{j=1}^{i-1 } |\omega _ { i}| \right ) \right)\\ & \qquad \ \
\times \omega _ { 1 } \wedge\cdots \wedge \omega _ { i-1}\wedge j_{k}(\omega _ { i})\wedge\cdots \wedge \omega
_ { k}\end{aligned}\ ] ] is an isomorphism and satisfies the following properties : 1 .
2 . @xmath202 for any @xmath203 .
3 . @xmath204 and @xmath205 for any @xmath206 .
the module of @xmath111-algebraic derivations is obviously closed with respect to the @xmath111-commutator of derivations . therefore we get a @xmath111-lie algebra structure on @xmath207 which is called the _ nijenhuis algebra _ of the @xmath0-algebra @xmath1 and its bracket is the @xmath111-_nijenhuis _
_ bracket_. by definition , the nijenhuis bracket of the elements @xmath208 hom @xmath209 and @xmath210 hom@xmath211 is given by the formula @xmath212^{\delta } = j_{k}\circ l-\overline{\rho } ( ( k,\alpha ) , ( l,\beta))j_{l}\circ k\ ] ] or @xmath212^{\delta } ( \omega)=j_{k}(l(\omega))-(-1)^{kl}\varphi ^{-l}(\alpha)\varphi ^{k}(\alpha)\rho ( \alpha,\beta)j_{l}k(\omega ) \label{14}\ ] ] for all @xmath213 the exterior derivative @xmath107 is an element of @xmath111-der@xmath214 . in the view of the formula
@xmath215 $ ] for fields @xmath69 we define @xmath216 the lie derivation @xmath217-der@xmath218 by @xmath219 $ ] .
then the mapping -der @xmath220 is injective by the universal property of @xmath221 , since @xmath222 for @xmath223 . for any @xmath111-derivation @xmath224-der@xmath225 , there are unique homomorphisms @xmath226 and @xmath227 such that @xmath228 we have @xmath229 if and only if @xmath230=0 $ ] .
@xmath151 is algebraic if and only if @xmath231 .
the map is a @xmath0-derivation of degree @xmath68 so has the form @xmath232 for an unique @xmath233 .
the defining equation for @xmath234 is @xmath235 for @xmath236 .
thus @xmath237 is an algebraic derivation , so @xmath238 for an unique @xmath227 . by the jacobi identity , we have @xmath239=[[j_k , d],d]+\overline{\rho } ( ( k,\alpha),(1,0))[d,[j_k , d]]\ ] ] so @xmath240=0 $ ] .
it follows that @xmath230=[j_l , d]=\mathcal{l}_l$ ] and using the injectivity of @xmath241 results that @xmath229.@xmath108 let @xmath242 and @xmath243 .
definition of the @xmath111-lie derivation results in @xmath244,d]=0 $ ] and using the previous theorem results that is a unique element which is denoted by @xmath245\in \omega_{(k+l,\alpha + \beta)}^{1}(a)$ ] such that @xmath246=\mathcal{l}_{[k , l]}\ ] ] and this element will be the denoted by the abstract frlicher nijenhuis of @xmath234 and @xmath247 . the space @xmath248 with the usual grading and the frlicher
nijenhuis is a @xmath133-graded lie algebra .
-der @xmath220 is an injective homomorphism of @xmath133-graded lie algebras . for fields in hom@xmath249
the frlicher nijenhuis coincides with the bracket defined in @xmath250 .
let be an homomorphism of degree 0 between the @xmath6-graded @xmath0-algebras @xmath1 and @xmath251 .
two forms @xmath252 hom@xmath253 and @xmath254 hom@xmath255 @xmath256 , @xmath257 are @xmath14-_related _ or @xmath14-_dependent _ if we have @xmath258 where is the homomorphism from ( [ 11 ] ) induced by @xmath14 . @xmath25 1 . if @xmath234 and @xmath259 are @xmath14-related as above then @xmath260 .
2 . if @xmath261 , then @xmath234 and @xmath259 are @xmath14-related , where @xmath262 is the space of exact @xmath263-forms .
if @xmath264 and @xmath265 are @xmath14-related for @xmath266 then @xmath267 and @xmath268 are @xmath14-related and also @xmath269^\delta , [ k_1^{\prime},k_2^{\prime}]^\delta$ ] are @xmath14-related .
if @xmath234 and @xmath259 are @xmath14-related then @xmath270 . 5 .
if @xmath271 then @xmath234 and @xmath259 are @xmath14-related .
if @xmath272 and @xmath273 are @xmath14-related for @xmath266 then their frlicher nijenhuis brackets @xmath274 $ ] and @xmath275 $ ] are also @xmath14-related .
the author wishes to express his thanks to prof .
piti for many valuable remarks and for a very fruitful and exciting collaboration . | in this paper we introduce non - commutative fields and forms on a new kind of non - commutative algebras : @xmath0-algebras .
we also define the frlicher nijenhuis bracket in the non - commutative geometry on @xmath0-algebras .
= mtmib at 10.4pt |
the main goals of reactor neutrino experiments are to detect @xmath3 oscillation and precisely measure the mixing angle of neutrino oscillation @xmath4 .
the experiment is designed to detect reactor @xmath5 s via the inverse @xmath6-decay reaction @xmath7 .
the signature is a delayed coincidence between @xmath8 and the neutron captured signals . in the paper , only three important sources of backgrounds are taken into account and they are the uncorrelated background from natural radioactivity and the correlated backgrounds from fast neutrons and @xmath0he/@xmath1li .
the backgrounds like the neutrino events consist of two signals , a fast signal and a delay signal .
it is vital to separate neutrino events from backgrounds accurately in the reactor neutrino experiments .
the selection of the neutrino events based on the cuts is a methods that the event space is divided into two regions by a hyper - cuboid based on the cuts , and the events inside the hyper - cuboid , called the signal region , are regarded as neutrino events and the events outside the hyper - cuboid are regarded as backgrounds . in fact , the backgrounds in the signal region could nt be rejected by the method .
the bayesian neural networks ( bnn)@xcite is an algorithm of the neural networks trained by bayesian statistics . it is not only a non - linear function as neural networks , but also controls model complexity . so its flexibility makes it possible to discover more general relationships in data than the traditional statistical methods and its preferring simple models make it possible to solve the over - fitting problem better than the general neural networks@xcite .
bnn has been used to particle identification and event reconstruction in the experiments of the high energy physics , such as ref.@xcite . in this paper
, bnn will be applied to discriminate the neutrino events from the background events in the signal region in the reactor neutrino experiments .
the idea of bayesian neural networks is to regard the process of training a neural network as a bayesian inference .
bayes theorem is used to assign a posterior density to each point , @xmath9 , in the parameter space of the neural networks .
each point @xmath9 denotes a neural network . in the method of the bayesian neural network ,
one performs a weighted average over all points in the parameter space of the neural network , that is , all neural networks .
the methods make use of training data @xmath10 , where t@xmath11 is the known label associated with data @xmath12 .
@xmath13 , if there are @xmath14 classes in the problems of classification ; @xmath12 has @xmath15 components if there are @xmath15 factors on which the classification is influenced .
that is the set of data @xmath16which corresponds to the set of target @xmath17.the posterior density assigned to the point @xmath9 , that is , to a neural network , is given by bayes theorem @xmath18 where data @xmath19 do not depend on @xmath9 , so @xmath20 .
we need the likelihood @xmath21 and the prior density @xmath22 , in order to assign the posterior density @xmath23to a neural network defined by the point @xmath9 .
@xmath24 is called evidence and plays the role of a normalizing constant , so we ignore the evidence . that is , @xmath25 we consider a class of neural networks defined by the function @xmath26}\ ] ] where @xmath27 .
the neural networks have @xmath15 inputs , a single hidden layer of @xmath28 hidden nodes and a single output . in the particular bayesian neural networks described here , each neural network has the same structure .
the parameter @xmath29 and @xmath30 are called the weights and @xmath31 and @xmath32 are called the biases . both sets of parameters
are generally referred to collectively as the weights of the bayesian neural networks , @xmath9 .
@xmath33 is the probability that an event , @xmath34 , belongs to the signal .
so the likelihood of @xmath35 training events is @xmath36 where it has been assumed that the events are independent with each other .
we get the likelihood , meanwhile we need the prior to compute the posterior density . but the choice of prior is not obvious
. however , experience suggests a reasonable class is the priors of gaussian class centered at zero , which prefers smaller rather than larger weights , because smaller weights yield smoother fits to data . in the paper , a gaussian prior
is specified for each weight using the bayesian neural networks package of radford neal .
however , the variance for weights belonging to a given group(either input - to - hidden weights(@xmath29 ) , hidden -biases(@xmath31 ) , hidden - to - output weights(@xmath30 ) or output - biases(@xmath32 ) ) is chosen to be the same : @xmath37 , @xmath38 , @xmath39 , @xmath40 , respectively .
however , since we do nt know , a priori , what these variances should be , their values are allowed to vary over a large range , while favoring small variances .
this is done by assigning each variance a gamma prior @xmath41 where @xmath42 , and with the mean @xmath43 and shape parameter @xmath44 set to some fixed plausible values .
the gamma prior is referred to as a hyperprior and the parameter of the hyperprior is called a hyperparameter .
then , the posterior density , @xmath23 , is gotten according to eqs .
( 2),(5 ) and the prior of gaussian distribution . given an event with data @xmath45 , an estimate of the probability that it belongs to the signal
is given by the weighted average @xmath46 currently , the only way to perform the high dimensional integral in eq .
( 7 ) is to sample the density @xmath23 with the markov chain marlo carlo ( mcmc ) method@xcite . in the mcmc method ,
one steps through the @xmath9 parameter space in such a way that points are visited with a probability proportional to the posterior density , @xmath23 .
points where @xmath23 is large will be visited more often than points where @xmath23 is small .
( 7 ) approximates the integral using the average @xmath47 where @xmath48 is the number of points @xmath9 sampled from @xmath23 .
each point @xmath9 corresponds to a different neural network with the same structure .
so the average is an average over neural networks , and the probability of the data @xmath45 belongs to the signal .
the average is closer to the real value of @xmath49 , when @xmath48 is sufficiently large .
in the paper , a toy detector is designed to simulate central detectors in the reactor neutrino experiments , such as daya bay experiment@xcite and double chooz experiment@xcite , with cern geant4 package@xcite .
the toy detector consists of three regions , and they are the gd - doped liquid scintillator(gd - ls from now on ) , the normal liquid scintillator(ls from now on ) and the oil buffer , respectively . the toy detector of cylindrical shape like the detector modules of daya bay experiment and double chooz experiment is designed in the paper . the diameter of the gd - ls region is 2.4 meter , and its height is 2.6 meter . the thickness of the ls region is 0.35 meter , and the thickness of the oil part is 0.40 meter . in the paper , the gd - ls and ls are the same as the scintillator adopted by the proposal of the chooz experiment@xcite .
the 8-inch photomultiplier tubes ( pmt from now on ) are mounted on the inside the oil region of the detector .
a total of 366 pmts are arranged in 8 rings of 30 pmts on the lateral surface of the oil region , and in 5 rings of 24 , 18 , 12 , 6 , 3 pmts on the top and bottom caps .
the response of the neutrino and background events deposited in the toy detector is simulated with geant4 .
although the physical properties of the scintillator and the oil ( their optical attenuation length , refractive index and so on ) are wave - length dependent , only averages@xcite ( such as the optical attenuation length of gd - ls with a uniform value is 8 meter and the one of ls is 20 meter ) are used in the detector simulation .
the program could nt simulate the real detector response , but this wo nt affect the result of the comparison between bnn and the method based on the cuts . according to the anti - neutrino interaction in the detector of the reactor neutrino experiments@xcite , the neutrino events are uniformly generated throughout gd - ls region ( see fig .
the uncorrelated background events are generated in such a way that the fast signal energies are generated on the base of the energy distribute of the natural radioactivity in the proposal of the day bay experiment@xcite , the energies for the neutron events of the single signal are regarded as the delay signal energies , the delay times are uniformly generated from 2 @xmath43s to 100 @xmath43s and the positions of the fast signal and the delay signal are uniformly generated throughout gd - ls region .
the fast neutron events are uniformly generated throughout gd - ls region and their energy are uniformly generated from 0 mev to 50 mev , therein the events of two signals are regarded as the fast neutron backgrounds .
since the behavior of @xmath0he/@xmath1li decay in the detector could nt be simulated by the geant4 package , @xmath0he/@xmath1li events are generated in such a way that the fast signal energies are generated on the base of the energy distribute of @xmath0he/@xmath1li in the proposal of the day bay experiment@xcite , and the other physical quantities are from fast neutron events in the paper .
the task of the event reconstruction in the reactor neutrino experiments is to reconstruct the energy and the vertex of a signal .
the maximum likelihood method ( mld ) is a standard algorithm of the event reconstruction in the reactor neutrino experiments .
the likelihood is defined as the joint poisson probability of observing a measured distribution of photoelectrons over the all pmts for given ( @xmath50 ) coordinates in the detector .
the ref.@xcite for the work of the chooz experiment shows the method of the reconstruction in detail . in the paper ,
the event reconstruction with the mld are performed in the similar way with the chooz experiment@xcite , but the detector is different from the detector of the chooz experiment , so compared to ref.@xcite , there are some different points in the paper : \(1 ) the detector in the paper consists of three regions , so the path length from a signal vertex to the pmts consist of three parts , and they are the path length in gd - ls region , the one in ls region , and the one in oil region , respectively .
\(2 ) considered that not all pmts in the detector can receive photoelectrons when a electron is deposited in the detector , the @xmath51 equation is modified in the paper and different from the one in the chooz experiment , that is , @xmath52 , where @xmath53 is the number of photoelectrons received by the j - th pmt and @xmath54 is the expected one for the j - th pmt@xcite .
\(3 ) @xmath55 and the coordinates of the charge center of gravity for the all visible photoelectrons from a signal are regarded as the starting values for the fit parameters(@xmath50 ) , where @xmath56 is the total numbers of the visible photoelectrons from a signal and @xmath57 is the proportionality constant of the energy @xmath58 , that is , @xmath59 .
@xmath57 is obtained through fitting @xmath56 s of the 1 mev electron events , and is @xmath60 in the paper .
the fast and delay signals of a event in the toy detector are reconstructed using mld , respectively .
the selections of neutrino events are as follows : \(1 ) positron energy : 1.3 mev < @xmath61 < 8 mev ; \(2 ) neutron energy : 6 mev < @xmath62 < 10 mev ; \(3 ) neutron delay : 2 @xmath43s < @xmath63@xmath64 < 100 @xmath43s ; \(4 ) relative positron - neutron distance : @xmath65 < 100 cm . a hyper - cuboid in the event space
is defined by the selection , and the inside is the signal region and the outside is the background region .
39000 events of neutrino are generated in the signal region .
11000 events each of uncorrelated background , fast neutrons and @xmath0he/@xmath1li are generated in the signal region , respectively .
the energies of the fast signal and the delay signal(@xmath61 , @xmath62 ) , the delay time of the delay signal(@xmath63@xmath64 ) and the distance between the fast signal and the delay signal ( @xmath65 ) are used as inputs to all neural networks , which have the same structure . in the paper ,
all the networks have the input layer of four inputs , the single hidden layer of nine nodes and the output layer of a single output which is just the probability that an event belongs to the neutrino event . a markov chain of neural networks is generated using the bayesian neural networks package of radford neal , with a training sample consisting of the neutrino events and the backgrounds .
one thousand iterations , of twenty mcmc steps each , are used .
the neural network parameters are stored after each iteration , since the correlation between adjacent steps is very high .
that is , the points in neural network parameter space are saved to lessen the correlation after twenty steps here .
it is also necessary to discard the initial part of the markov chain because the correlation between the initial point of the chain and the points of the part is very high .
the initial three hundred iterations are discarded here .
3000 events each of the neutrino and the three backgrounds are used to test the identification capability of the trained bnn . in the paper , the bnns are trained by the different training samples , which consist of the neutrino events and three backgrounds at different rates , since the different identification efficiencies are obtained with those bnns .
the results of the identification with those bnns are listed in tab . 1 .
1 , the neutrino discrimination increases from 82.6% to 91.2% with the increase of the neutrino rate from one second to nine fourteenth in the training sample using bnn in signal region . however , the background discriminations decrease with the decrease of the background rate in the training sample . the uncorrelated background discrimination decrease from 88.2% to 73.6% with the decrease of its rate from one sixth to one fourteenth in the training sample .
the fast neutron background discrimination decreases from 48.5% to 37.6% with the decreases its rate from one sixth to one seventh in the training sample .
the @xmath0he/@xmath1li background discrimination decreases from 51.8% to 39.9% with the decrease of its rate from one sixth to one seventh in the training sample . as a result ,
the most neutrino events and uncorrelated background events in the signal region can be identified with bnn , and the part events each of the fast neutron and @xmath0he/@xmath1li backgrounds in the signal region can be identified with bnn . the different signal to noise ratios in signal region
are obtained with bnns trained by the training samples consisting of neutrino events and background events at different rates in the reactor neutrino experiments . in a word ,
the signal to noise ratio in signal region can be enhanced with bnn in the reactor neutrino experiments .
this work is supported by the national natural science foundation of china ( nsfc ) under the contract no .
10605014 .
p. c. bhat and h. b. prosper _ beyesian neural networks_. in : l. lyons and m. k. unel ed .
_ proceedings of statistical problems in particle physics , astrophysics and cosmology , oxford ,
uk 12 - 15 , september 2005_. london : imperial college press .
151 - 154 .the different identification efficiencies are obtained with the bnns trained by the different training samples , which consist of the neutrino and three backgrounds at different rates .
the term after @xmath66 is the statistical error of the identification efficiencies .
the 3000 events each of the uncorrelated background , fast neutron and @xmath0he/@xmath1li are regarded as the test sample . [ cols="^,^,^,^,^,^ " , ] | a toy detector has been designed to simulate central detectors in reactor neutrino experiments in the paper .
the samples of neutrino events and three major backgrounds from the monte - carlo simulation of the toy detector are generated in the signal region .
the bayesian neural networks(bnn ) are applied to separate neutrino events from backgrounds in reactor neutrino experiments . as a result
, the most neutrino events and uncorrelated background events in the signal region can be identified with bnn , and the part events each of the fast neutron and @xmath0he/@xmath1li backgrounds in the signal region can be identified with bnn .
then , the signal to noise ratio in the signal region is enhanced with bnn .
the neutrino discrimination increases with the increase of the neutrino rate in the training sample .
however , the background discriminations decrease with the decrease of the background rate in the training sample .
@xmath2department of physics , nankai university , tianjin 300071 , people s republic of china bayesian neural networks , neutrino oscillation , identification pacs numbers : 07.05.mh , 29.85.fj , 14.60.pq |
only 50% of the baryons consistent with the cosmic microwave background radiation ( cmbr ) and big bang nucleosynthesis ( bbn ) observations have been detected observationally @xcite ; the validation of the standard cosmological model relies on our ability to detect the missing baryons observationally @xcite . using data from the canada france hawaii lensing survey and the planck satellite , recent studies , that use cross - correlation of the thermal sunyaev - zeldovich ( tsz ) effect and weak gravitational lensing ,
have shown that up to @xmath7 of the baryons may reside outside the virial radius of the halos @xcite .
the cosmological simulations too suggest that majority of the intergalactic medium ( igm ) are in the form of warm - hot intergalactic medium ( whim ) within a temperature range @xmath8 @xcite .
it is also believed that whims reside in moderately overdense structure such as filaments . however , being collisionally ionized these baryons do not leave any footprints in lyman-@xmath9 absorption systems .
the emission from whims in either uv photons or x - rays is too weak to be detected given the sensitivity of current instruments .
their detection is also not feasible in x - ray given the low level of emission from whim .
however the baryons in the cosmic web have sufficient velocity and column density to produce a detectable cmb secondary anisotropy effect known as the kinetic sunyaev - zeldovich ( ksz ) effect @xcite .
cmb secondary anisotropies arise at all angular scales ; the largest secondary anisotropy at the arcminute scale is due to the thermal sunyaev - zeldovich ( tsz ) effect .
the tsz effect is caused by the thermal motion of electrons mainly from hot ionized gas in galaxy clusters where as the _ kinetic _ sunyaev - zeldovich ( ksz ) effect is attributed to the bulk motion of electrons in an ionized medium @xcite .
the tsz can be separated from cmb maps using spectral information . along with lensing of cmb ,
the ksz is the most dominant secondary contribution at arcminute scale after the removal of tsz , as primary cmb is sub - dominant due to the silk - damping @xcite .
the _ thermal _ and _ kinetic _ sunyaev - zeldovich effects ( tsz and ksz respectively ) are both promising probes of the ionized fractions of the baryons .
majority of tsz effect is caused by electrons in virialized collapsed objects @xcite .
the overdensity in these regions is considerably high , @xmath10 .
the missing baryons however are more likely to reside in filaments with moderate overdensity and lower temperature and hence unlikely to be probed efficiently by tsz .
interestingly , the ksz is linked to the peculiar velocity which is sensitive to large scale potential and less likely to be sensitive to individual collapsed objects .
it is also less sensitive to the thermal state and can probe contributions from individual baryonic components in a relatively unbiased way @xcite . for previous studies of ksz effect from whims
see e.g. @xcite . the methods for extracting the ksz which we present here are free from systematics that one might encounter when trying to detect it using the cmb two - point auto correlation function or equivalently the power - spectral analysis .
the ksz signal is not only extremely weak ; it also lacks any spectral feature @xcite and is overwhelmed by the primary cmb below @xmath11 .
the ksz effect dominate at the damping tail of the primary beyond @xmath12 only at @xmath13 ghz where the dominant _ secondary _ contribution from the non - relativistic tsz effect vanishes .
contamination from dusty star forming galaxies can also further complicate detection of ksz effect @xcite .
use of non - gaussianity has been suggested as possible method to separate ksz from confusion due to lensing of cmb @xcite .
being a cmb secondary anisotropy , the ksz effect includes projected contributions from various independent components and lacks any redshift information .
it is a projection of electron momentum along the line of sight .
lack of redshift information means it is impossible to separate individual contribution such as from a _ patchy _ reionization history , which is expected to contribute significantly and thus can in principle bias any estimates of missing baryons .
the tight cross - correlation of tsz and galaxy densities has been probed in the literature in detail @xcite .
however such a cross - correlation vanishes due to the vectorial ( spin-1 ) nature of the ksz . to circumvent this problem various options have currently been tried , e.g. using higher order cumulant correlators @xcite or tomographic cross - correlation with galaxy redshift surveys @xcite .
cross - correlation with other tracers of low redshift large scale structure e.g. intensity mapping from future 21 cm surveys has also been considered @xcite . the recently proposed technique dubbed ksz _ tomography _
@xcite , involves reconstructing the galaxy peculiar velocity field from spectroscopic redshift surveys and including galaxy number density and other redshift dependent pre - factors .
notice that bulk flows on scales of @xmath14 and the pairwise velocity dispersion have been observed @xcite .
the reconstructed momentum field has very similar directional dependence as the true ksz effect .
thus the cross - correlation of these two can be used as an indicator for the ksz contribution to cmb maps . with redshift space binning it is also possible to study evolution of ksz contribution as a function of redshift .
this method can be useful in separating out not only the primary cmb from ksz but also various contribution such as the contribution from patchy reionization .
a high value for the signal to noise @xmath15 was estimated using planck and the bigboss survey .
notice that similar tomographic reconstruction was also proposed for tsz @xcite as was recently extended to 3d recently @xcite . in recent years
there has been significant progress in extracting survey information using three - dimensional spherical fourier - bessel ( sfb ) decomposition of galaxy surveys which goes beyond a tomographic treatment @xcite .
the formalism involves expanding the galaxy density field in a sfb expansion .
the formalism was found extremely useful in analysing weak lensing data in 3d @xcite . going beyond the tomographic cross - correlation in this paper we use a 3d treatment to reconstruct the ksz from galaxy surveys .
this is in line with the reconstruction technique developed for tsz by @xcite using a 3d sfb decomposition .
spectroscopic galaxy redshift surveys , such as lamost , boss , bigboss , ska , euclid and jdem / adept will map the galaxy distribution in 3d .
these surveys will have overlap in terms of observed regions of the sky with ongoing cmb surveys , e.g. , act , spt and planck .
it is possible to recover the velocity fields from such surveys and construct an estimator in 3d that can be correlated with cmb maps to reconstruct the distribution of ionized gas whose bulk velocity is causing the ksz . the primary aim of this paper is to disentangle the low - redshift ( late - time ) contribution to ksz effect and that from a rather poorly understood reionization history .
this will allow us to to use future ksz surveys in constraining the interaction of dark energy ( de ) and dark matter ( dm ) using the ksz effect as such a non - gravitational coupling in the dark sector can significantly affect the expansion history of the universe as well as growth of structure @xcite .
indeed various modified gravity theories that affect the growth rate and power spectrum of perturbation can also be constrained using ksz effect .
the results presented here will be relevant to extending studies of ksz in constarining radial inhomegeneites in lemaitre - tolman bondi cosmologies @xcite or dark flow @xcite .
the current upper limit on the ksz power spectrum from the atacama cosmology telescope ( act ) is @xmath16 at @xmath17 .
the upper limits from the south pole telescope ( spt ) is @xmath18 at @xmath17 .
most recently @xcite reported @xmath19 ( @xmath20 ) .
these upper limits depend on the reionization history , modelling of cosmic infra - red background , tsz contribution and separation of the ksz effect into homogeneous and patchy components . using arcminute resolution maps from act and boss statistical evidence was presented for motion of galaxy clusters through their ksz signature @xcite .
the foreground - cleaned maps from planck nominal mission data were used to achieve a @xmath21 detections of the ksz signal .
this was achieved by estimating the pairwise momentum of the ksz temperature fluctuations at the positions of the cgc ( central galaxy catalogue ) samples extracted from sloan digital sky survey ( dr7 ) data @xcite . in the same study ,
data from wmap i.e. wmap-9yr w band data a detection of @xmath22 was also reported .
the best constraints on reionization history are likely to result from redshifted 21-cm experiments with low - frequency radio interferometers gmrt@xcite , lofar@xcite , mwa@xcite and paper@xcite
. the near infra - red background measurements with ciber and akari will also provide valuable clues to reionization history .
in addition , ground - based observations of high - redshift qso , galaxies and grbs too provide important clues to reionization history @xcite .
the plan of this paper is as follows .
the [ sec : not ] is devoted to introducing our notation . in [ sec:3dest ]
we motivate a 3d analysis for ksz reconstruction . in [ sec:3dps ]
derive the expressions for 3d power spectrum . in [ sec : tomo ] we present how to construct the tomographic and 2d maps and power spectrum from 3d information .
the section [ sec : s2n ] gives analytical expression for s / n of 3d reconstruction .
the section [ sec : ps ] is devoted to the discussion of the two power - spectra @xmath23 and @xmath24 which are used as an input for the ksz power spectra computation .
the [ sec : results ] is reserved for discussion of our results .
finally , the conclusions are presented in [ sec : conclu ] .
we have included the discussions of issues related to the partial sky coverage in appendix-[sec : mask ] .
in this section we introduce our cosmological notations for the ksz effect that will be used later .
we will use the following line element : & & ds^2 = -c^2 dt^2 + a^2(r)(dr^2 + d^2_a(r)(^2d^2 + d^2 ) ) throughout @xmath25 will denote speed of light and will be set to unity . in our notation , @xmath26 is the comoving angular diameter distance at a ( comoving ) radial distance @xmath27 and can be expressed in terms of the curvature density parameter @xmath28 . here
@xmath29 is the hubble constant ; @xmath30 is the current non - relativistic matter density in units of critical density and @xmath31 is current contribution of cosmological constant to the critical density .
the angular diameter distance will be denoted as @xmath32 as a function of the radial comoving distance @xmath33 at a redshift @xmath34 : & & d_a ( r ) = _ h ; _ h = ch^-1_0 ; where @xmath35 means @xmath36 if @xmath37 or @xmath38 if @xmath39 ; if @xmath40 , then @xmath41 . the radial comoving distance @xmath33 from a source at redshift @xmath34 to an observer at @xmath42 is given by : & & r(z ) = _ h^z_0 ; ( z ) = = .
[ eqn : ez ] we will denote the linear growth factor as @xmath43 defined such that the fourier transform of the over - density field grows as @xmath44 is given by : & & d_+(z ) = _
z^ dz^ ( 1+z^ ) ^-3/ _ 0^ dz^ ( 1+z^ ) ^-3 .
the particular cosmology that we will adopt for numerical study is specified by the following parameter values ( to be introduced later ) : @xmath45 .
the cosmic ionized transverse momentum can be studied as a temperature fluctuation in the cmb maps .
the line - of - sight ( los ) component of the momentum of electrons in the ionized igm introduces temperature fluctuation via the doppler effect also known as ksz .
the contribution from the longitudinal component suffers cancellation due to contribution from successive troughs and crests @xcite .
the measured ksz power spectrum is the sum over contribution from the epoch of reionization ( eor ) and the post - reionization or late - time ksz contribution .
inhomogeneity in the ionization fraction during eor gives a large boost to the ksz power spectrum .
the physics that determines the _ patchy _ or inhomogeneous reionization is complex .
analytical @xcite , semi - analytical @xcite and numerical n - body simulations coupled with radiative transfer @xcite are often used to understand various aspects .
the contribution from _ late - time _
ksz is relatively easier to model and will be the main focus of this paper .
inhomogeneity in electron density , defined by the density contrast @xmath46 and peculiar velocity @xmath47 as well as in ionization fraction @xmath48 can all lead to secondary fluctuation in the cmb temperature @xmath49 .
the specific ionized momentum field @xmath50 at @xmath51 is given by @xcite : & & * q*(*r * ) _ ( r ) * v*(*r * ) ( 1 + _ ( * r * ) ) [ eq : momentum ] the ionized fraction @xmath48 is defined in terms of electron density @xmath52 and number density of hydrogen(h ) and helium(he ) atoms denoted as @xmath53 and @xmath54 respectively i.e. @xmath55 . for post
reionization or late time ksz computation we will set @xmath56 . if we assume that hydrogen reionization was finished at @xmath57 and heii reionization occurred instantaneously at @xmath58 we will have @xmath59 for @xmath60 and @xmath56 for @xmath61 .
our reconstruction technique is based on local large scale tracers which typically cover the redshift range @xmath61 which justifies our choice of @xmath56 .
modelling of the post - reionization signal is simpler because the igm is fully ionized and @xmath48 does not fluctuate to a good approximation : @xmath62 .
we thus need to model the electron density @xmath63 and velocity @xmath64 fluctuation of gas .
the velocity filed is purely longitudinal in the quasi - linear regime . in general
the baryonic density contrast @xmath46 can be different from the dark matter density especially on scales smaller than the jeans - length scale .
indeed the shock - heated gas will be less clustered than the underlying dark - matter due to baryonic pressure .
this effect can have modest impact on ksz power spectrum at @xmath17 .
star formation converts gas into stars thus reducing the ksz effect further ; gas - cooling and star formation can reduce the ksz effect up to @xmath65 .
these effects can only be dealt with using numerical simulations . for this paper
we will assume that the baryons trace the dark matter on the scales of interest which are much larger than the jeans length scale .
the ksz is given by the transverse momentum field , hence is a vector mode which can be described also also as a spin-1 field . & & ( * * ) = _ 0^z_lss dr ( r ) ( * r * ) ; ( * r*)= * * c ; + & & ( r ) = -_t _ _ , 0
c e^-(r ) a^2 .
the projected momentum @xmath66 is the 3d momentum projected along the line of sight direction at a distance @xmath67 . here
@xmath68 is the unit vector along the line of sight direction , @xmath27 is the comoving distance along the line of sight direction .
we also note that the optical depth @xmath69 due to thompson scattering can be expressed as : & & d= c n_(z ) _ t dt dzdz .
where @xmath69 is the optical depth to thompson scattering integrated from @xmath70 to surface of last scattering at @xmath71 ; @xmath72 is the thompson scattering cross - section for electrons and @xmath73 . the visibility function @xmath74 is defined as : & & g(t ) -= c_t n_(t ) .
[ eq : g ] in this paper , we propose to construct a 3d estimator @xmath75 for ksz induced temperature fluctuation in cmb maps .
the corresponding projected @xmath76 and tomographic estimators @xmath77 can be computed using suitable line of sight integration .
we construct @xmath78 from 3d spectroscopic surveys : & & ( * r * ) = ( r ) ( * r * ) ( * * ) = _ 0^z_lss dr ( * r * ) ; _ i ( * * ) = _ z_i^z_i+1 dr ( * r * ) .
the weighted momentum field @xmath50 is constructed to have the same directional dependence as the ksz effect and thus should cross - correlate with the cmb maps and help us to extract the ksz component of cmb .
similar correlations have been studied before in tomographic bins @xcite where the radial integral was restricted to a particular bin .
our approach is to go beyond the tomographic description and constructs the ksz effect in 3d .
the cross - correlation technique has several advantages over the auto - correlation technique . due to the use of 3d template that involves matching characteristic directional dependence , it is possible to avoid contamination from tsz , dusty star forming galaxies , as well as from the primary cmb
this is rather important given the weak nature of the ksz signal and overwhelming systematic error .
the redshift information in our estimator also allows separating out the ksz contributions from patchy reionization and late time ksz effect .
typically the high number density of galaxy ensures a higher signal to noise ratio for such an estimator .
similar 3d techniques was be used to reconstruct the thermal sz effect @xcite .
spherical coordinates are a natural choice for the analysis of cosmological data sets as , by an appropriate choice of coordinates , we can place an observer at the origin of the analysis .
future large scale surveys will provide an unprecedented level of detail by generating extended sky maps with large radial coverage .
we therefore require a simultaneous treatment of the extended radial coverage and the spherical sky geometry .
for this problem the spherical fourier - bessel ( sfb ) expansion is a natural basis for the analysis of cosmological random fields .
the sfb formalism has seen an increased use in the literature over the past few years , for example it has been used in : weak lensing studies @xcite , redshift space distortions @xcite , relativistic effects @xcite and studies of baryon acoustic oscillations @xcite .
the forward and inverse sfb decomposition of the 3d field @xmath79 is defined as : & & _
m(k ) = d^3*r * k j_(kr ) y^*_m ( * * ) ( * r * ) ; [ eq : forward ] + & & ( * r * ) = kdk _ m _
m(k ) j_(kr ) y^*_m ( * * ) ; ( r , * * ) ; [ eq : backward ] + & & * * =( , , ) .
the basis function for the sfb expansion is an eigen function of the laplacian operator @xmath80 in spherical co - ordinates .
here , @xmath81 denotes the radial wave number and @xmath82 denotes the azimuthal wave numbers ; @xmath83 are the spherical harmonics and @xmath84 are the spherical bessel functions . in general
the continuous integral over @xmath27 can be a replaced with a discrete sum over galaxy position i.e. effectively replacing the galaxy density with a sum over dirac s delta functions @xmath85 that are non - zero at galaxy position @xmath86 . throughout results from an all - sky experiment will be presented . for flat - sky results relevant for small patch of the sky
the spherical harmonics are replaced by exponential function .
to relate the radial distance @xmath27 with the observed redshift of galaxies @xmath34 a background cosmology needs to be assumed .
the expressions derived above is technically valid for a flat universe . to relax this assumption we have to replace the spherical bessel functions with hyperspherical bessel functions .
to analyse realistic data weights are introduced .
such arbitrary masks will be incorporated using a 3d pseudo-@xmath87 approach .
our aim is to relate the 3d harmonics of @xmath88 with the underlying perturbations in specific ionized momentum field @xmath89 defined in eq.([eq : momentum ] ) . to this end
, we will fourier decompose the momentum vector @xmath50 whose fourier components are @xmath90 : & & ( ) = d^3*r*(*r * ) e^i*k*. [ eq : four_q ] we will separately deal with the contribution to ksz from longitudinal @xmath91 and transverse @xmath92 components of @xmath93 also known as the divergence free ( curl ) component of @xmath94 i.e. @xmath95 and curl free ( divergence ) components .
they are defined using the following expressions : & & _ ( ) = ( ) - _ ( ) ; + & & _ ( ) = * * _ k[()_k ] ; = ( k , * * _ k ) .
the two power spectra @xmath96 and @xmath97 for @xmath98 and @xmath99 are defined as follows : & & _ ( * k * ; r)_(*k* ; r)= ( 2)^3 p_q_(k;r , r)_3d(*k*-*k* ) ; + & & _ ( * k * ; r ) _ ( * k* ; r ) = ( 2)^3 p_q_(k;r , r)_3d(*k*-*k* ) . [ eq : qdef ] the expressions for @xmath100 and @xmath101 in terms of the matter power spectrum @xmath102 will be presented in [ sec : ps ] .
we will next deal with the contribution to ksz from the longitudinal and transverse components independently .
first we express the @xmath103 in terms of the fourier components of @xmath66 i.e. @xmath104 .
following the same convention for fourier transform as described in eq.([eq : four_q ] ) : & & ( * * ) = dr ( r ) ( * k*,*r * ) e^-i .
we will write @xmath105 where @xmath106 depends only on @xmath107 and @xmath108 on @xmath109 .
contribution to the total @xmath103 can now be broken into two parts namely @xmath110 and @xmath111 with @xmath112 .
& & ^ ( * * ) dr ^ ( ) ; ^()(r ) _ ( * k*,r ) e^-i*k * ; _ ( * k*,r ) = ; + & & ^(**)=dr ^ ( ) ; ^ ( ) ( r ) _ ( * k*,r)e^-i*k * ; _ ( * k*,r)=. [ eq : theta_divide ] we have defined @xmath113 $ ] . throughout we will denote unit vector in real space as @xmath114 and as @xmath115 in the fourier domain . using a 3d spherical harmonic decomposition of @xmath116 and @xmath117 , denoted as @xmath118 and @xmath119 respectively ,
corresponding projected ( 2d ) harmonics @xmath120 and @xmath121 are obtained by spherical harmonics decomposition of the respective 2d analogs i.e @xmath122 and @xmath123 .
we will construct two power 3d spectra @xmath124 and @xmath125 similarly to @xmath126 and @xmath127 in 2d .
we will also consider cross - correlation of our 3d harmonics with the corresponding 2d harmonics to construct @xmath128 and @xmath129 .
we will express these power spectra in terms of @xmath101 and @xmath100 . in order to compute the contribution from the _ transverse _ component we will use the definition of @xmath132 from eq.([eq : theta_divide ] ) to express the power spectrum @xmath126 in terms of @xmath23 : & & _ ( * r * ) _ ( * k*,*r * ) ; + & & ^(*k*,*r*)= ( _ q-)_(*k * , r ) e^-i*k * .
next , we use the following definitions for @xmath133 : & & y_1,-1 ( * * ) = 1 2 e^-i ; y_1,1 ( * * ) = -1 2 e^i ; [ eq : sph ] to express the trigonometric function in terms of lower order spherical harmonics @xmath134 : & & = [ y_1,-1(**)- y_1,1 ( * * ) ] .
[ eq : cossin ] we will also be using the rayleigh expansion of plane wave in terms of spherical wave : & & e^-i*k * = 4_m ( -i)^j_(kr)y_m(**)y_m(**_k ) .
[ eq : ray ] choosing the unit vector @xmath135 along the @xmath34 direction , @xmath136 we can replace @xmath137 with @xmath138 . using expressions from eq.([eq : cossin ] ) we can write : ^(kz;*r * ) & = & 4 [ y_1,-1(**)- y_1,1(**)]_(*k * , r ) _ m(-i)^^ j_(kr)y_ m ( * * ) y_ m(**_k ) + & = & [ y_1,-1(**)- y_1,1 ( * * ) ] _ ( * k * , r ) _ ( -i)^^ j_(kr ) y_ 0 ( * * ) .
we will express the overlap - integrals involving three spherical harmonics in terms of @xmath139 symbols : ^_m(kz , r ) & = & _ ( kz , r ) _
(-i)^^ j_(kr ) _ d y_0 ( * * ) [ y_1,-1(**)- y_1,1 ( ) ] y^*_m ( * * ) ; + & = & _ ( kz , r ) _
( -i)^^ i_,,1 j_(kr ) _ ( c c c & & + 1 + 0 & 0 & 0 ) + & & .
[ eq : theta_perp ] the @xmath139 symbols are defined through the _ gaunt _ integral : & & d * * y__1 m_1 ( * * ) y__2 m_2 ( * * ) y__3 m_3 ( * * ) = i__1,_2,_3 ( c c c _ 1 & _ 2 & _ 3 + 0 & 0 & 0 ) ( c c c _ 1 & _ 2 & _ 3 + m_1 & m_2 & m_3 ) ; + & & i__1,_2,_3=
. for various symmetry properties of @xmath139 symbols see e.g. @xcite .
notice , only @xmath140 multipoles contribute as the @xmath139 symbols need to satisfy the condition @xmath141 .
the first term in the eq.([eq : theta_perp ] ) contributes for @xmath142 and the second term contributes for @xmath143 .
we will use the following identities in our derivation to evaluate the special cases of the 3j symbols that appear : ( c c c _ 1 & _ 2 & _ 3 + 0 & 0 & 0 ) & = & ( -1)^ ; _ 1+_2+_3 = 2 + & = & 0 ; _
1+_2+_3 = 2 + 1 .
[ eq:3j1 ] ( c c c _ 1 & _ 2 & + _ 1 & -_1 & -m ) = ( -1)^-_1+_2+m ^1/2 .
[ eq:3j2 ] * ( a ) * modes with we will consider the @xmath142 term first : ^_,m=-1(kz ; r ) & = & - _ ( kz;r ) _
i_,,1 j_(kr ) ( c c c & & + 1 + 0 & 0 & 0 ) ( c c c & & + 1 + 0 & + 1 & -1 ) .
the 3j symbols get contribution from @xmath144 and @xmath145 due to the _ triangular _ inequality : ^_,m=-1(kz;r ) & = & - _ ( kz;r ) .
we use the identities in eq.([eq:3j1 ] ) and eq.([eq:3j2 ] ) to derive the following expressions : & & ( c c c + 1 & & + 1 + 0 & 0 & 0 ) = ( -1)^+1 ; ( c c c + 1 & & + 1 + 0 & + 1 & -1 ) = ( -1)^-1 .
[ eq : new3j1 ] we also use @xmath146 we eventually get : & & _ m=-1^(kz ; r ) = ( -i)^+1 _ ( kz , r ) ; _ ( + 1 ) . where we have also used the following relation for the spherical bessel functions : & & j_+1(x)+ j_-1(x ) = _ x j_(x ) .
* ( b ) * modes with this component can also be computed in an analogous manner : & & ^_l , m=1(kz ; r ) = _ ( k z , r ) _
i_1 j_(kr ) ( c c c & & + 1 + 0 & 0 & 0 ) ( c c c & & + 1 + 0 & + 1 & -1 ) .
[ eq : theta_perp ] following the same steps and using the following expressions for the @xmath139 symbols : & & ( c c c -1 & & + 1 + 0 & 0 & 0 ) = ( -1)^ ; ( c c c -1 & & + 1 + 0 & + 1 & -1 ) = ( -1)^-1 ; [ eq : new3j2 ] it can be shown that @xmath147 . * ( c ) * combining modes next , to rotate @xmath148 defined for @xmath81 along the @xmath34-axis to an arbitrary observer s orientation @xmath119 , we use the expression @xmath149 , where @xmath150 is the wigner @xmath150-matrix , we can write eq.([eq : theta_perp ] ) in following form : & & ^_m(k ) = ( -i)^+1 _ m d^_m,1(**_k^ ) dr ( r)k j_(kr ) j_(k^ r ) k^r_(k^z;r ) .
[ eq : theta_elm_perp ] * ( d ) * constructing the power spectra finally the 3d sfb covariance @xmath151 can be expressed in terms of @xmath23 as : ^_(k_1,k_2 ) & & _
m^(k_1)_m^*(k_2 ) + & = & 2 _ dk k^2 dr_1 ( r_1 ) k_1 j_(k_1r_1 ) j_(kr_1 ) k r_1 dr_2 ( r_2 ) k_2 j_(k_2r_2 ) p_q(k ; r_1,r_2 ) .
[ eq : full_perp ] to perform the @xmath152 integral in eq.([eq : theta_elm_perp ] ) we use the following expression for the orthogonality of wigner @xmath150-matrices : & & d**_k d^_m_1,m_2(**_k ) d^_m_1,m_2(**_k ) = 4_^k_^k_m_1m_1^k_m_2m_2. [ eq : d_matrix ] we note here that even for all - sky coverage the covariance matrix above can be ill - conditioned , reflecting the very nature of the ksz effect which inherently non - linear even at linear regime due to coupling of modes resulting from density and velocity field .
thus a practical implementation may require binning to improve the condition number .
the expression presented in eq.([eq : full_perp ] ) can be evaluated directly . in case of
more conservative estimates one can directly evaluate using numerical simulations .
nevertheless we also quote results that use the following approximation @xmath153 @xcite .
to reduce the dimensionality of the integration in eq.([eq : full_perp ] ) .
& & c^_(k_1,k_2 ) k^2 dk ^2_(k , k_1 ) ^2_(k , k_2 ) ; [ eq : perp_cl1 ] + & & i^2_(k , k ) dr ( r ) k j_(kr ) .
[ eq : perp_cl2 ] the kernel @xmath154 defined above encapsulates the information regarding the mode - mode coupling .
finally we recover the projected ( 2d ) power spectrum for the ksz effect is given by @xcite : & & ^_= 1 2 ^2(r ) p_q(k= r ; r ) .
[ eq:2dcls ] the derivation of eq.([eq:2dcls ] ) depends on the use of limber approximation stated below for an arbitrary function @xmath155 : & & k^2 dk f(k ) j_(kr)j_(kr^ ) f(k = r)_d(r - r^ ) ; which generalises the orthogonality property of the spherical bessel functions : & & k^2 dk j_(kr)j_(kr^ ) = 2 r^2_d(r - r^ ) .
cross - correlating the cmb maps @xmath156 against reconstructed 3d estimators : & & ^_(k ) _ m_m^*(k)= 2 _ dk^^2 dr_1r_1 ^ 2 ( r_1 ) dr_2r_2 ^ 2 ( r_2 ) k j_(kr_2 ) p_q(k^ ; r_1,r_2 ) .
[ eq : full_recon_perp_est ] notice that the estimator @xmath75 is constructed using 3d distribution of galaxies and the recovered 3d velocity distribution .
the weight @xmath157 included in the construction depends on a prior choice of cosmology .
the sfb expansion decomposes the reconstructed ksz estimator to mode specified by @xmath158 .
the cross - correlation of this estimator with cmb maps @xmath103 thus extracts only ksz contribution as a function of radial harmonics @xmath81 and radial harmonics @xmath159 .
this provides an unique method to recover radial information about evolution of ksz effect .
the construction has other advantages .
it avoids any contamination from other cmb secondary anisotropies such as lensing or tsz , or even from primary cmb and thus reduces the statistical error .
given that the ksz signal is weak and foreground contamination ( as well as confusion from dusty star forming galaxies ) is high at angular scales where ksz peaks the method presented here offers an unique opportunity .
the estimator in eq .
( [ eq : full_recon_perp ] ) involves a 3d integral .
we can use the same factorization of unequal time correlator introduced above to reduce it s dimensionality : & & ^_(k ) dk^2 ^2_(k , k^)^_(k^ ) ; [ eq : factor_recon_perp1 ] + & & i^_(k ) dr r_1 ^ 2 ( r ) .
[ eq : factor_recon_perp2 ] eq .
( [ eq : full_recon_perp_est ] ) and its simplified version eq .
( [ eq : factor_recon_perp1])-eq .
( [ eq : factor_recon_perp2 ] ) constitute some of the main results of this paper . *
( e ) * reconstructing the velocity field : the construction of the estimator introduced above for obtaining a 3d galaxy distribution from a 3d survey is just a first step .
the galaxy density field @xmath160 is used as a proxy for the underlying dark matter distribution @xmath161 .
it is possible to incorporate redshift- and scale - dependent bias but at large scales where we will use our estimator , a simple linear bias @xmath162 should be sufficient .
construction of a ksz estimator should include appropriate weighting using the recovered velocity field @xmath64 .
this is required to avoid the cancellation of cross - correlation of density field and an estimator for ksz due to the spinorial nature of the ksz field .
the reconstruction of 3d velocity field uses the linearised continuity equation .
& & + = 0 . in fourier space
it takes the following form : & & ( * k * ) = -if(z ) h(z ) ( * k * ) k ; f(z ) .
[ eq : lin_den_vel ] the estimator thus constructed using linear theory will be insensitive to multi - streaming resulting from shell - crossing that introduces vorticity . however , non - linearities at the perturbative level can be included in a more sophisticated treatment which we ignore here . the linear redshift - space distortion ( the kaiser effect )
is ignored here but can be incorporated in our treatment @xcite . a prior on background cosmology
is required for specification of @xmath163 and @xmath164 . the contribution from @xmath166
can be used as a check on systematics .
the computation of the _ longitudinal _ contribution or @xmath125 in terms of @xmath24 follows similar steps . & & ^(*r * ) = ( r ) _ ( * k*,*r * ) ; + & & ^(*k * ; * r * ) = _ ( * k * ; * r*)e^-i * k * using the fact : & & = y_1,0 ( * * ) ; [ eq : rj ] and taking advantage of rayleigh expansion eq .
( [ eq : ray ] ) we can write : & & ^(*k * ; * r * ) = 4 _ ( * k * , r ) _ m
( -1)^ j_(k r ) y_m(**)y_m(**_k ) .
next , aligning the @xmath167 vector along the @xmath34-axis ( @xmath168 ) we use the identity @xmath169 : & & ^(kz ; * r * ) = 4 _ ( kz , r)y_10 ( ) _ j_(kr)y_0 ( ) .
we perform a spherical harmonic transform of @xmath106 .
we note that the only non - zero harmonic coefficients come from . using the expression @xmath170 & & ^ _ ,
m=0(kz ) = 4 _ ( kz;r ) _ ( -i)^ i_1 j_ ( kr ) ( c c c & & + 1 + 0 & 0 & 0 ) ^2 . as before , due to the triangular inequality , only terms that contribute are @xmath171 .
& & ^ _ , m=0(kz ; r)= 4 _ ( kz ; r ) . using the expression in eq .
( [ eq : new3j1 ] ) and eq .
( [ eq : new3j2 ] ) for the @xmath139 symbols that appear in the above expression we can write : & & ^ _ , m=0(kz ; r ) = - ( -i)^+1 _ ( kz ; r ) .
next we use the following recurrence relation for @xmath172 to simplify further : & & j_-1(x ) - ( + 1)j_+1(x ) = _ . here
the prime denotes a derivative w.r.t . the argument . using this expression we can write : & & ^ _ , m=0(kz)= _ ( kz;r ) . for higher
@xmath159 the contribution from @xmath173 and @xmath174 nearly cancels each other thus making @xmath166 sub dominant .
= 16 cm = 6 cm performing the rotation using wigner @xmath175 - matrices as before to a general orientation we have : & & _
m^(k ) = ( -i)^+1 _ m d^_m,0(**_k^ ) k j_(kr ) j^_(k^r ) _
( kz , r ) . finally using the orthogonality relation for wigner d - matrices given in eq .
( [ eq : d_matrix ] ) , and the definition @xmath24 in eq .
( [ eq : qdef ] ) we arrive at the following expression : ^_(k_1,k_2 ) & & _
m^(k_1)_m^*(k_2 ) + & = & 2 _ dk k^2 dr_1r^2 ( r_1 ) k_1 j_(k_1r_1 ) j^_(kr_1 ) dr_2r^2 ( r_2 ) k_2 j_(k_2r_2 ) p_q(k ; r_1,r_2 ) .
[ eq : full_para ] consequently , we can use the approximation @xmath176 to factorize the integrals : & & c^_(k_1,k_2 ) k^2 dk^2_(k , k_1)^2_(k , k_2);[eq : para_cl1 ] + & & i^2_(k , k ) dr r^2 ( r ) k j_(kr)r^2 j^_(k^r ) .
[ eq : para_cl2 ] cross - correlating the 2d cmb maps against reconstructed 3d estimators : & & ^_(k ) _ m_m^*(k)= 2 _ dk^^2 dr_1r_1 ^ 2 ( r_1 ) dr_2r_2 ^ 2 ( r_2 ) k j_(kr_2 ) p_q(k^ ; r_1,r_2 ) .
[ eq : full_recon_perp ] the corresponding expressions using the factorization are as follows : & & ^_(k ) dk^^2 ^2_(k , k^)^_(k^ ) ; + [ eq:2dcls_para ] & & i^_(k ) dr r^2 ( r ) .
[ eq:2dkernel_para ] [ sec : results ] various approximations are typically used to collapse a full 3d analysis to projection ( 2d ) or , in other words to perform tomography ( see [ sec : tomo ] ) . however
, these methods invariably involve binning in redshift interval with a consequent loss of the information content of the data .
the 3d analysis also gives the freedom to treat the angular and radial modes separately .
projection is achieved using limber approximation invariably links the radial and angular modes . in projected or tomographic survey
its very difficult to disentangle individual scales .
treatment of individual scales independently is important for keeping a handle on the effect of non - linearity due to gravitational evolution .
a 3d analysis treats each galaxy individually and thus can be use to control the systematics at a level of individual galaxies , which is very important for ksz reconstruction .
any contaminating region can always be masked out .
a pseudo-@xmath87 ( pcl ) based approach will be developed that can be used to recover the power spectra in the presence mask ( see appendix-[sec : mask ] ) .
= 16 cm = 6 cm
in order to connect with the tomographic reconstruction of ksz effect by @xcite , in this section we will show how to reconstruct the tomographic power spectra from the 3d power spectra we have developed .
the tomographic estimator @xmath177 from the @xmath178-th bin as introduced in eq .
( [ eq : theta_divide ] ) and its harmonic coefficients @xmath179 can be expressed in terms of their 3d counterparts : & & ^(i ) = ^r_i+1_r_i dr w(r ) ^(*r * ) ; + & & ^(i)_m = d**y_m ( * * ) _ ^m^ dk k j_(kr ) y_^ m^ ( * * ) ^_^ m^(k ) . using the orthogonality property of spherical harmonics
we can express the @xmath179 in terms of @xmath119 using a kernel @xmath180 .
the @xmath178-th bin is defined between the radial distance @xmath181 and @xmath182 .
& & ^(i)_m = dk j^(i)_(k)^_m(k ) ; t^(i)_(k)= k^r_i+1_r_i dr j_(kr ) w(r ) .
the tomographic power - spectra for the @xmath178-th bin is denoted as @xmath183 .
similarly the covariance between @xmath178-th and @xmath184-th bin is denoted as @xmath185 .
& & ^(ij ) _
^(i)_m^(j)*_m = dk j^(i)_(k ) dk j^(j)_(k ) c_^(k , k ) ; [ eq : cov ] + & & ^(i)_^(i)_m^*_m= dk j^(i)_(k ) c_^(k ) .
similar expressions were derived by @xcite in the context of 3d weak lensing .
an exactly similar relation can be obtained for @xmath186 and @xmath187 . to construct the 2d estimator given in eq .
( [ eq:2dcls ] ) we need to use only one bin and use the limber approximation with @xmath188 .
the tomography essentially probes discrete sets of physical wave numbers .
the position of the tomographic bins constraints the radial wave numbers probed as a function of angular harmonics .
the advantage of 3d is being able to ensure the integrated analysis of the entire set of wave - numbers probed by a survey .
indeed , tomography allows for straightforward interpretation of the results particularly redshift dependence of the ksz signal .
however , using limber approximation as outlined above can always be used to collapse the 3d information to tomographic bins .
a simplistic expression for the signal - to - noise @xmath189 for reconstruction of the individual modes described by radial and angular wave numbers @xmath190 respectively is given by : ( s n)^_(k ) = f_sky .
[ eq : s2n ] expressions for @xmath130 is derived in eq .
( [ eq : perp_cl1 ] ) and @xmath131 in eq .
( [ eq : factor_recon_perp2 ] ) .
the cmb power spectrum @xmath191 should also include the tsz power spectrum at high-@xmath159 for @xmath192 estimation .
we assume the noise of estimation to be independent of cmb noise ; @xmath193 denotes the fraction of sky covered .
a similar expression for @xmath166 can be obtained by replacing @xmath131 and @xmath131 with @xmath166 and @xmath166 .
= 16 cm = 6 cm
the computation of the ksz power spectra depends on modelling @xmath23 .
early analytical results were based on the linear perturbation theory and ignored non - linearity in density and velocity field @xcite . the ksz effect in linear regime
is also known as the ostriker - vishniac ( ov ) effect @xcite . in @xcite the linear theory prediction for the density power spectrum
was replaced with non - linear prescription for density power spectrum by @xcite .
the velocity power spectrum was from linear theory prediction ( see e.g. @xcite and @xcite for a detailed discussion of this issue ) . in @xcite
the non - linear power spectrum used was that from halo - model prescription .
we outline the major step in the derivation to underline the approximations used in various steps . in terms of density @xmath196 and velocity @xmath197 components we have : & & _ ( * k * ) = _ k ( k)(|*k - k*| ) ; + & & _ ( * k * ) = ( _ k - _ k ) ( k)(|*k - k*| ) ; = * * _ k .
the corresponding 3d power spectra are given by : & & p_q(k , r ) _ ( * k*)^*_(*k * ) = ; [ eq : perp ] + & & p_q(k , r)_(*k*)^*_(*k * ) = .
[ eq : para ] these expressions were computed by neglecting the connected fourth order moments which were found to be negligible .
we will use the following expression from @xcite ( see also @xcite ) in our derivation of the 3d ksz power spectrum : & & p_q(k , r ) = a^2 f^2 p_(|k - k|)p_(k ) k(k-2k)(1-^2 ) k^2(k^2+k^2 -2 k k ) .
[ kperp ] + & & p_q(k , r ) = a^2 f^2 p_(|k - k|)p_(k ) k(k -2 k^2 + k ) k^2(k^2+k^2 -2 k
k ) ; f .
[ kpara ] the above expressions were derived under the assumption @xmath198 ( see e.g. * ? ? ?
* and references therein ) .
this approximation is a necessary ingredient as the density - velocity cross - correlation is poorly understood . in the linear theory
the the velocity and cross - correlation power spectra are expressed as follows : & & p_vv(k ) = ( f ak ) ^2 p_(k ) ; p_v(k ) = ( f ak ) p_(k ) .
[ eq : power_vv ] to evaluate the ksz power spectra defined in eq .
( [ eq : factor_recon_perp1 ] ) and eq .
( [ eq:2dcls_para ] ) we use these expressions for @xmath199 and @xmath200 .
following @xcite we use the halo model to compute the density power spectra and linear theory to compute the velocity power spectrum ( see @xcite and @xcite for related discussions ) . in the high -k limit eq .
( [ eq : perp])-eq .
( [ eq : para ] ) is approximated as : & & p_q_(k ) = 2 3 p_(k)v^2_rms(k ) ; p_q_(k)= 2p_q_(k ) ; v^2_rms(k ) _ k^ k d^3k^ ( 2)^3 p_vv(k^ ) . here
@xmath201 is the velocity dispersion . in a recent study
the effect of the four - point correlation function , which we have ignored in our study has been outlined by @xcite .
they pointed out that in the non - linear regime at @xmath202 the contribution from the four - point correlation function ( connected - part ) can reach as high as @xmath203 .
in figure [ fig : pofk ] we show the input power - spectra @xmath204 ( right - panel ) @xmath205 ( left - panel ) and @xmath206 ( middle - panel ) used for our calculation .
the solid lines represent non - linear power - spectra and the dashed lines linear power - spectra .
the results are for @xmath70 . in general , to relate the baryonic power spectrum and the dark matter power spectrum a window function is used .
we have not included any such filtering in our computation to keep the analysis transparent , however such modification is straightforward .
notice that the treatment we followed involves using the linear relationship in eq .
( [ eq : lin_den_vel ] ) between density and velocity , which breaks down at high @xmath81 .
accurate modelling in that regime can only be obtained using numerical simulations .
nevertheless , we believe the qualitative nature of our results will remain unchanged in the range of @xmath190 considered here .
the approximate form of the power spectrum we have chosen that allows us to separate two internal integrals and reduces computation time significantly .
indeed it can be argued that presence of spherical bessel functions introduces a cut - off that suppresses long range ( low-@xmath159 ) modes thus justifying such a factorization . in figure
[ fig : cls2d ] we show the computed @xmath131 and @xmath166 as a function of @xmath81 for @xmath207 , @xmath208 and @xmath209 .
these spectra were defined in eq .
( [ eq : perp_cl1 ] ) and eq .
( [ eq : factor_recon_perp1 ] ) respectively .
the spectra correspond to a maximum redshift of reconstruction of @xmath210 .
different panels correspond to harmonics @xmath211 and @xmath212 from left to right .
the peak of the spectra shifts to higher @xmath81 with increasing @xmath159 . due to
the presence of the spherical bessel functions , which become increasingly sharp - peaked , the 3d cross - spectra effectively collapse to the projected ones at higher @xmath159s , thus virtually linking the radial wave - number @xmath81 and the angular wave - number @xmath159 .
the range of @xmath81 that contributes to a given @xmath159 becomes more sharply defined .
the power spectrum @xmath131 increasingly dominates over @xmath166 with increasing @xmath159 . in figure
[ fig : cls3d ] the diagonal entries for 3d covariance @xmath213 ( solid - curves ) and @xmath214 ( dashed - curves ) are shown as a function of wave number @xmath215 .
these covariances are defined in eq .
( [ eq : perp_cl1 ] ) and eq .
( [ eq : para_cl1 ] ) respectively .
these plots show diagonal cuts in the @xmath216 plane for various @xmath159 values through the 3d covariance matrices .
the effect of the bessel function is to introduce a sharp cut - off on the long wave length contributions with @xmath217 , where @xmath218 is the extent of the survey .
as the @xmath159 values increases the diagonal entries terms of the covariance do not become significant until @xmath219 .
the kernels @xmath220 and @xmath221 which are used to construct the cross - spectra @xmath166 ( dashed - lines ) and @xmath131 ( solid - lines ) as a function of wavelength are plotted in figure [ fig : kernel2d ] .
the corresponding kernels @xmath222 and @xmath223 are shown in figure [ fig : kernel3d ] .
= 16 cm = 6 cm in figure [ fig : s2n ] we display the result of @xmath224 computed using eq .
( [ eq : s2n ] ) .
we show results for @xmath211 and @xmath212 as a function of the radial wave number @xmath81 .
the reconstruction is assumed to be done up to a redshift of @xmath210 .
we ignore the residuals from component separation .
we also ignore the shot - noise due to finite number density of galaxies .
however , the s /
n of maps from galaxy surveys are much higher compared to their cmb counterpart and are not likely affect the s / n of the reconstruction .
higher multipoles have better @xmath189 of reconstruction primarily due to the @xmath225 prefactor .
however , the available range of radial wavenumber @xmath81 decreases with increasing @xmath159 . at higher @xmath81
the linear reconstruction of velocity from density field may not be accurate .
increasing the survey depth will clearly shift the peak of the reconstructed @xmath226 to lower radial wave - numbers @xmath81 , but , number density of galxies used for reconstruction will decrease thus decreasing the s / n .
maps from future 21 cm surveys or other tracers of large scale structure can be used to extend the reconstruction technique proposed here to moderate redshifts @xmath61 .
the ksz signal is overwhelmed by various contaminations and lacks any redshift information due to the projection effect .
moreover , the contribution to ksz can be from the reionization epoch as well as from post - reionization era .
the contribution from reionization depends on accurate modelling of reionization which can in principle be inhomogeneous ( patchy ) and thus difficult to model . with the recent detection of ksz effect
there is a greater effort in the community to understand it than was previously . in this paper
we focus on estimation of post - reionization late - time ksz effect .
this contribution is relatively easier to model as the physics involve is simpler compared to reionization and may provide us clues about our local universe and when subtracted from the total ksz can help constrain the contribution originating from reionization . indeed from cmb measurements alone , we only expect to detect ksz from the auto power - spectra of temperature fluctuations .
however , ksz is likely to dominate the power spectrum at @xmath227 where the primary cmb is damped due to silk - damping and contribution from tsz shows a null at @xmath13 ghz .
the method proposed here can probe ksz below @xmath228 as the method proposed here is independent of contamination from primary cmb or other secondary anisotropies .
the method proposed here uses spectroscopic galaxy surveys to reconstruct peculiar velocity field .
our method weighs the peculiar velocity field with galaxy distribution .
the resulting momentum when projected along the line of sight direction can act as an estimator for ksz effect in 3d .
our analysis extends previous analysis performed using tomographic bins @xcite and allows maximal exploitation of the available data with greater degree of freedom .
we have constructed two different 3d estimators for ksz power spectrum @xmath229 in eq .
( [ eq : para_cl2 ] ) and @xmath230 in eq .
( [ eq : perp_cl2 ] ) that can be used to extract information .
this will eliminate any contamination from primary cmb which remains one of the dominant source especially at low @xmath159 .
the tsz contribution to cmb sky is typically removed by using frequency information .
the lensing of cmb and ksz lacks any frequency information . in the past
, several methods have been proposed to separate such contribution , including the use of higher - order information or the non - gaussianity of the signal .
most of the cmb secondary anisotropies originate in the relatively low redshift universe and have a characteristic redshift evolution . using this fact we propose a cross - correlation based estimator to extract ksz effect from cmb maps .
the proposal is based on using the spectroscopic redshift surveys to construct an estimator for the peculiar velocity and density . with
suitable weighting these reconstructed fields can next be used to construct an estimator for ksz that takes into account the spin-1 nature of ksz . when cross - correlated with the cmb sky it will only cross - correlate with the ksz component thus avoiding usual contamination from other secondaries .
there are many advantages to the technique developed here .
it separates the late time , @xmath231 contribution to ksz effect from the poorly understood _ patchy reionization _ scenarios .
the ksz is effect is known to be a potentially powerful probe of the baryons - the estimators presented here can exploit this to probe directly the 3d distribution of the missing baryons in the low redshift universe .
the galaxy surveys typically have relatively higher @xmath232 and are thus expected to allow such reconstruction . for tomographic reconstruction using planck as target cmb survey and boss as a spectroscopic galaxy survey a @xmath232 of @xmath233 was shown to be achievable .
it is expected that sz survey e.g. spt and galaxy surveys such as adept , euclid and ska would work still better .
this will thus provide measurement of the power - spectra over a range of scales ( @xmath159 and @xmath81 ) with a reasonable accuracy .
the detection of baryons using the methods presented here using ksz signature in cmb maps does not rely on detection of hot gas or metals ; it directly detects the free electrons in the igm .
the line - of - sight velocity is uncorrelated with many standard systematics that plague usual power spectrum analysis applied to cmb data , e.g. it is insensitive to the unsubtracted tsz , galactic foregrounds and detector noise .
= 5 cm = 5 cm we have ignored the redshift space distortion in our formulation of an estimator for the ksz . the redshift space distortion can be effectively accounted for in the our 3d formulation at least at scales where linear kaiser effect remains valid . at smaller scales
higher order contributions will have to be included .
however at even smaller scales the finger - of - god effect caused by small scale random motion becomes important especially at low redshift where the perturbative results break down .
indeed our modelling of galaxy bias which we take as a linear and deterministic can also be stochastic ( see e.g. * ? ? ? * and references therein ) .
such issues can only be dealt with through full numerical simulations .
the 3d reconstruction technique developed here is more fundamental than the tomographic methods presented previously . indeed
the tomographic or the projected power spectra can both be constructed from the 3d power spectra presented here .
the 3d sfb transforms developed here generalises similar expansion schemes used in other areas in cosmology including e.g. studies of weak lensing @xcite and redshift space distortions @xcite .
we have already emphasised the power of ksz to map out the baryonic web .
constraints on the dark sector , i.e. decaying dark matter or dark matter - dark energy interactions has recently attracted a lot of attention .
they can be probed by tsz or ksz effects ( e.g. * ? ? ?
the power spectra @xmath23 and @xmath24 depend on the underlying matter power spectra @xmath234 as well as on the growth rate of perturbations , @xmath235 , see eq .
( [ kperp])-eq .
( [ kpara ] ) .
clearly in many modified gravity theories such as the @xmath236 theories , the dilaton theories and the k - mouflage models the power spectrum as well as the function @xmath164 are very different from their @xmath237cdm counterparts .
future surveys with accurate determination of the ksz power spectrum may be able to constrain such theories .
such issues in the context of 3d ksz that we have studied here will be presented elsewhere .
dm , iti , kld and pc acknowledge support from the science and technology facilities council ( grant number st / l000652/1 ) .
the power spectrum @xmath238 and @xmath239 used in this work was computed by g. w. pettinari .
we acknowledge useful discussion with h. park and p. zhang .
dm also acknowledges results obtained by g. pratten at the early stage of this work which helped obtaining some of the results presented here .
dm benefited from discussions with a. heavens and p. valageas .
the power - spectra we have constructed so far assume a full sky coverage . in general surveys will not cover the entire sky .
however , once survey configuration is specified by angular mask and selection function it is possible to use a pseudo-@xmath240 ( pcl ) formalism in 3d to relate the observed @xmath87s with the underlying 3d @xmath87s . for the analytical derivation
we will assume that the 3d mask @xmath241 can be separated angular mask @xmath242 and the radial selection @xmath243 can be treated separately . denoting the masked field by @xmath244 which is recovered from a survey and the underlying by @xmath245 we can write : & & ^s(*r * ) = w_3d(*r*)^(*r * ) .
we will assume a separable mask @xmath246 .
this will allow us to deal with the angular and radial part separately making analytical treatment possible .
the harmonic coefficients of the masked field and the underlying field are related by the following expression : & & ^s_m(k ) = _ ^m^ dk^ ^_^ m^(k^ ) w^r_^(k , k^ ) |w^a_m ^ m^. we will apply the same mask to the cmb @xmath247 .
the corresponding angular harmonics denoted respectively as @xmath248 and @xmath121 can be related as & & ^s_m = _
^m^ as @xmath249 and @xmath250 depends on the @xmath251 and @xmath252 respectively and can be expressed as follows : & & w^r_^(k , k^ ) = 2 k k^ dr r^2 w_r^2(r ) j_(kr)j_^(k^r ) ; [ eq : window_radial ] + & & |w^a_^ = d * * w ( * * ) y_m ( * * ) y_^ m^ ( * * ) .
[ eq : window_angular ] the pseudo-@xmath87s for the masked field @xmath253 are denoted as @xmath254 and @xmath255 .
the pseudo @xmath87s are linear combinations of the underlying @xmath87s .
& & c^s_(k ) ^s_m(k)^s_m = _
^ w^a_^dk^ w^r_^(k , k^ ) c^_^(k^ ) ; [ eq : cross ] + & & c^s_(k_1,k_2 ) ^s_m(k)^s_m(k ) = _ ^ w^a_^dk^_1dk^_2
w^r_^(k_1,k^_1 ) w^r_^(k_2,k^_2 ) c^_^(k^_1,k^_2 ) [ eq : auto ] .
the linear transform relating the two power - spectra e.g. @xmath254 and @xmath229 , or , equivalently @xmath255 and @xmath256 , are encapsulated in a mode - mixing matrix . however , due to our assumption of factorizability of the mask we can write the mixing matrix as product of two different mode - mixing matrices .
the radial and angular mode - mixing matrices denoted as @xmath249 and @xmath257 depends on the @xmath251 and @xmath252 respectively and can be expressed as follows : & & w^a_^ = _ l w_l i^2_l^ 2 + 1 ( ) ^2 ; w_l = 1 2l+1_m w_lmw^*_lm .
[ eq : coupling ] for nearly all - sky coverage the mixing matrices can be inverted to recover the underlying @xmath87s .
however , binning may be required for partial sky coverage .
the power - spectra @xmath166 and @xmath165 can be dealt with in an exactly similar manner .
the mixing matrices remain unchanged .
a few comments are in order .
the pcl based sub - optimal estimators for projected or 2d data sets were developed by @xcite in the context of cmb power spectrum estimation .
it has also been generalised for computation of cross - spectra involving external data sets @xcite .
more recently , it has been extended for the 3d data by @xcite and was successfully used for analysing 3d weak lensing power spectrum using cfhtlens data @xcite . though pcl based estimators remain sub - optimal , suitable use of weights can improve their performance . the results presented here are generalisation of the one presented in @xcite ( where the effect of radial selection function was not included ; i.e. @xmath258 . ) and will have general applicability in cross - correlating arbitrary projected field ( 3d ) against a 3d cosmological tracer .
similar results can be derived for higher - order spin fields by suitably modifying @xmath259 .
* generalisation to arbitrary spin : * for the sake of completeness we also provide the expression for a generic 3d field with an _ arbitrary _ spin @xmath260 @xmath9 .
we would like to cross - correlate it with another arbitrary projected field with spin @xmath261 say @xmath262 . in this case
the basis of expansion involves spinorial harmonics @xmath263 .
the spin - harmonics are defined in terms of @xmath264 matrices as @xmath265 @xcite .
the forward and backward transformations are obtained by replacing @xmath266 in eq .
( [ eq : forward])-eq .
( [ eq : backward ] ) with spin - weighted spherical harmonics or spinorial harmonics @xmath267 : & & _
s^s_m(k ) = _ ^m^ dk^ _ ^ m^(k^ ) w^r_^(k , k^ ) _
s|w^a_m ^ m^ ; + & & _
s^s_m = _ ^m^ _ s|w^a_m
^ m^_^ m^. we follow the same notation as before but we assume the 3d field @xmath268 is a spinorial field of spin - weight @xmath260 and @xmath269 is a projected filed of spin @xmath261 .
the harmonic components of these fields are now denoted as @xmath270 and @xmath271 .
the suffix @xmath272 denotes quantities defined in partial sky .
the radial component of the mixing matrix given in eq .
( [ eq : window_radial ] ) remains unchanged .
however , the angular mode - mixing matrix depends on the spin of the specific field eq .
( [ eq : window_angular ] ) . & & _
s|w^a_^ = d * * w(**)_sy_m(**)_sy^*_^ m^ ( * * ) . the corresponding auto- and cross - spectrum
are defined as follows : & & c^s_(k ) ^s_m(k)^s*_m = _ ^ _ ss^w^a_^dk^ w^r_^(k , k^ ) c_^(k^ ) ; + & & c^s_(k_1,k_2 ) _
m(k)^*_m(k ) = _ ^ _ ss^w^a_^dk^_1dk^_2
w^r_^(k_1,k^_1 ) w^r_^(k_2,k^_2 ) c_^(k^_1,k^_2 ) .
these expressions generalises the ones obtained in eq .
( [ eq : cross]-[eq : auto ] )
. the coupling matrix is denoted by @xmath273 which is a generalisation of the coupling matrix given in eq .
( [ eq : coupling ] ) : & & _ ss^w^a_^ = _
l w_l i^2_l^ 2 + 1 ( ) ; w_l = 1 2l+1_m w_lmw^*_lm .
previous results correspond to the specific case when we set @xmath274 .
one immediate practical use of the above expressions could be joint analysis of cmb lensing and 3d galaxy lensing @xcite . | we propose a novel technique to separate the late - time , post - reionization component of the kinetic sunyaev - zeldovich ( ksz ) effect from the contribution to it from a ( poorly understood and probably patchy ) reionization history .
the ksz effect is one of the most promising probe of the _ missing baryons _ in the universe .
we study the possibility of reconstructing it in three dimensions ( 3d ) , using future spectroscopic surveys such as the euclid survey .
unlike the thermal sunyaev - zeldovich ( tsz ) effect , direct cross - correlation of ksz effect with projected galaxy distribution suffers from cancellation due to the vector ( or spin-@xmath0 ) nature of the ksz effect . by reconstructing a 3d template from galaxy density and peculiar velocity fields from spectroscopic surveys we cross -
correlate the estimator against cmb maps .
the resulting cross - correlation can help us to map out the ksz contribution to cmb in 3d as a function of redshift thereby extending previous results which use tomographic reconstruction .
this allows the separation of the late time effect from the contribution owing to reionization . by construction
, it avoids contamination from foregrounds , primary cmb , tsz effect as well as from star forming galaxies .
due to a high number density of galaxies the signal - to - noise ( s / n ) for such cross - correlational studies are higher , compared to the studies involving cmb power spectrum analysis . using a spherical bessel - fourier ( sfb ) transform we introduce a pair of 3d power - spectra : @xmath1 and @xmath2 that can be used for this purpose .
we find that in a future spectroscopic survey with near all - sky coverage and a survey depth of @xmath3 , reconstruction of @xmath2 can be achieved in a few radial wave bands @xmath4 with a s / n of upto @xmath5 for angular harmonics in the range @xmath6 .
: cosmology kinetic sunyaev zeldovich surveys methods : analytical , statistical , numerical |
the evolution of life shows an overall trend towards an increase in size and complexity@xcite .
one of the determining major innovations that have allowed biological systems to achieve a high degree of complexity has been the evolution of multicellularity and the emergence of supra - cellular hierarchies beyond single - cell organization@xcite . together with multicellularity , mechanisms to maintain stable phenotypes that underly consistent division of labor had to be developed@xcite .
the study of the origins of form have a long tradition in biology@xcite .
initiated by turing@xcite and rashevsky@xcite , numerous attempts to formalize a mathematical description of pattern formation have been made . as a result , spatial instabilities were proposed as a powerful rationale for the creation of spatial order , out of random fluctuations , around a homogeneous state in reaction - diffusion systems@xcite .
the main feature of reaction - diffusion systems is the presence of diffusion - driven instabilities under certain parametric conditions , by which small perturbations in the system are amplified , leading to ordered spatial patterns .
this family of models has been systematically studied @xcite and provides the basis for several natural mechanisms of pattern formation @xcite .
the structures generated by these processes have a characteristic scale whose wavelength depends on the model parameters . along with this class of pattern - forming mechanisms , another possible class of models capable of organizing structures in space is based on cell - cell differential adhesion@xcite .
such a mechanism explains the spatial re - arrangement of different cells belonging to disrupted tissues when mixed together _ in vitro_@xcite . after a transient , clusters involving cells of the same class
are often observed as spatially segregated from other cell types by means of the formation of well defined boundaries or layers . in this case , the underlying mechanism explaining the origin of patterns is that of energy - minimization dynamics , similar to the one used in physics for strongly interacting particle systems .
both reported mechanisms are crucial in the formation of natural self - organized structures in developing embryos @xcite , and have been connected to the early forms of multicellularity @xcite . in this paper
we focus our attention on the early stages of the transition towards multicellularity , where the explicit connection between fitness , function and structure has been particularly difficult to elucidate and , thus , is commonly overlooked . to assess whether the structural organization of multicellular assemblies is related to differential fitness , we have developed an embodied computational model where turing - like structures appear , stemming from differential adhesion and stochastic phenotypic switching .
fitness is intrinsically obtained by the introduction of a limiting nutrient and the production of a toxic waste byproduct , which respectively increase cell reproduction or death .
one of the two cellular states is able to process waste at the cost of reduced proliferation .
we observe that different parameter sets produce different spatial patterns , and that spatial organization can have a role in increasing fitness . finally , we discuss the implications of such results for the transition from unicellular to multicellular organisms and for the evolution of complexity .
[ corerules ] is minimized .
( b ) the final global configuration is a direct consequence of the micro rules imposed by the adhesion matrix .
( c ) markovian process modelling the cell state transitions used in this paper .
this very simple approach can aptly describe persister cell dynamics and phase variation phenomena @xcite . ]
our model considers a population of cells living on a two - dimensional square lattice ( fig .
1 ) , along with empty medium , and following the rules of a cellular potts model as described by steinberg @xcite ( see also@xcite ) . within this framework ,
cells are discrete entities that occupy single lattice positions , have an associated state ( @xmath0 ) and move across the lattice trying to minimize their energetic potential .
two states correspond to cellular phenotypes , namely white cells ( @xmath1 ) and black cells ( @xmath2 ) , while state @xmath3 represents empty space .
in this paper we build and analyse two different expansions of the basic potts model : a _ hybrid differential adhesion - stochastic phenotipic switching _
( da - sps ) model and an _ ecology and competition _ ( ec ) model . in the da - sps model ,
cells are sorted by differential adhesion and can reversibly switch their phenotypes . in the ec model
we include a simple metabolism by adding nutrient and toxic waste , whose concentrations drive cell proliferation and death . in the following sections we explain how cellular adhesion , phenotypic switching and metabolism are implemented in our models .
the cell sorting process fundamentally occurs due to the differences in adhesion energy between states . following steinberg s differential adhesion hypothesis ( dah )
we assume that the adhesion kinetics are driven by the minimization of adhesion energy between lattice sites , being cells more or less prone to remain together , and avoid or maximize contact with the external medium @xcite .
the strength of interactions among different states can be defined by means of an adhesion matrix @xmath4 : @xmath5 each term @xmath6 in this matrix describes how favourable the pairwise interaction between two states is .
the matrix is symmetric , i.e. @xmath7 , and has @xmath8 always . to avoid confusion
, we will use the notation @xmath0 when we refer to a given state , and the notation @xmath9 to indicate a state occupying a given lattice site @xmath10 .
[ differences ] the underlying idea here is that cells will tend to move whenever this allows the system to reach a lower energy . it can be shown that the energy function @xmath11 in a given position @xmath10 can be defined as follows : @xmath12 where @xmath13 is the set defined by the eight nearest neighbours of a cell in position @xmath10 ( moore s neighbourhood ) , each of which occupies a position @xmath14 , and has a defined state @xmath15 . to calculate the probability that the cell in ( @xmath16 ) will swap with a randomly chosen neighbour
, we calculate the energy function when no swap occurs .
this energy function , named @xmath17 , consists of two terms , one involving the cell in its original position and its neighbourhood set @xmath13 , and another involving the cell s neighbour , located in @xmath18 , with its neighbourhood @xmath19 .
we then virtually swap the positions of the cell with its neighbour in @xmath18 , and calculate the energy function when swap occurs ( @xmath20 ) .
the energy difference is then defined as : @xmath21 when the difference is negative , a decrease in the global energy occurs and the states will swap position . instead , when @xmath22 , the larger the difference the less the swap is likely to happen , with a probability following the boltzmann distribution . if we indicate @xmath23 as the probability that our cell moves from @xmath10 to @xmath14 , it can be shown that : @xmath24 where the parameter @xmath25 is a noise factor acting as a ` temperature ' , essentially tuning the degree of determinism of our system .
the boltzmann factor @xmath26 acts in such a way that if @xmath27 , the probability of swapping is @xmath28 .
note that cell - cell ( and cell - medium ) interactions are local ( fig .
1 ) , meaning that a cell in ( @xmath16 ) interacts only with the set @xmath13 of its eight nearest neighbours . depending on the form of adhesion matrix , different patterns can be observed in a cell sorting system with two cell types .
unless otherwise indicated , in our simulations we will apply a symmetric adhesion matrix ( see fig .
sm-1 ) , where cells tend to attach preferentially to other cells in the same state , and secondarily to cells of the opposite state , while attachment with empty space is not favoured .
the values of the adhesion matrix determine the structure of patterns formed by the cell sorting algorithm ( fig .
2 ) , which can be perturbed by the effects of phenotypic switching . cells can perform reversible transitions between their states @xmath1 and @xmath2 , similarly to phase variation and persistence in natural bacterial populations @xcite .
switching is regulated by transition probabilities @xmath29 and @xmath30 , @xmath31 where @xmath32 is a fixed scaling factor , introduced to regulate the relative speed between adhesion kinetics and phenotypic switching . by simply adding sps to a classical da model , cell sorting properties can change drastically for some adhesion matrices ( see fig .
it is worth mentioning that in sps the transitions between states are not dependent on any molecular cue nor any cellular memory beyond their current state .
cellular metabolism is defined by two simple pathways : the ability of both @xmath1(white ) and @xmath2(black ) phenotypes to transform nutrient @xmath33 ( constantly added to the lattice ) into cellular energy @xmath34 and waste byproduct @xmath35 : @xmath36 and the unique ability of @xmath2 cells to degrade waste : @xmath37 the @xmath2 cells can allocate resources for waste degradation , at the cost of reduced energy production and therefore proliferation , following a linear trade - off ( @xmath38 ) consistent with a maximum metabolic load and shared resources for protein synthesis .
the temporal dynamics of metabolites @xmath39 , @xmath40 , @xmath41 for a given position @xmath10 in the lattice are described by : @xmath42 @xmath43 @xmath44 here , @xmath45 is the rate of input of the nutrient resource , and @xmath46 and @xmath47 correspond to the diffusion rate of nutrient and waste respectively .
it is worth noting that , being @xmath34 an intracellular metabolite , it does not diffuse through the lattice .
variables @xmath48 , @xmath49 and @xmath50 correspond to the exponential decay parameter of each of the metabolites , while @xmath51 defines the maximal absorption rate .
the trade - off parameter @xmath52 adjusts the proportion of nutrient allotted to energy production or waste degradation in @xmath2 cells .
taking into account spatial dynamics of metabolites , both kinds of cells can die either due to local excess of toxic waste or due to lack of internal energy : @xmath53 where @xmath54 indicates that the process equally affects both types of cells .
@xmath55 is the maximum value of toxic waste a cell can sustain , @xmath56 is the minimum value of inner energy needed for survival , and @xmath57 defines the inner energy threshold needed for a cell to divide , provided an empty position exists in its neighbourhood : @xmath58 mother and daughter cells have the same properties .
energy is equally split between the two cells after division .
[ meanfield ] , @xmath30 ) phase space in the computational mean - field model , for @xmath59 .
( sm-4 ) to see how the separation slope varies at different values of @xmath52 . ]
to better understand the general properties of the model , we developed a mean - field ( @xmath60 ) approach to the set of odes that constitute the metabolism of cells in our system .
as a starting point we use the waste differential equation ( 9 ) . in a well mixed scenario there is no spatial structure and all variables are homogeneous .
hence , the diffusion term and the position subindices cease to be of relevance .
therefore , at the steady state we get : @xmath61 = 0 , \nonumber\end{aligned}\ ] ] where @xmath62 and @xmath63 are the equilibrium concentrations of nutrient and waste respectively , and @xmath64 is the probability that a cell in the system has state @xmath0 . here
we separate the analysis into two solutions : @xmath65 \end{array } \right.\end{aligned}\ ] ] @xmath63 is equal to zero when @xmath66 ( trivial unstable solution ) or when the second nullcline is met . to develop the mathematical treatment we assume that the two populations of cell states @xmath1 and @xmath2 are at equilibrium .
the ratio between populations can be then deduced from the persister cell population dynamics : @xmath67 @xmath68 the three terms on the right - hand side of these equations represent reproduction , stochastic switching and death , respectively . at the steady state , given that : @xmath69 the relation between the two populations is : @xmath70 since we have @xmath71 , the expected probabilities for each population at equilibrium are : @xmath72 being cell death a threshold function , all cells will die if @xmath73 , and in the opposite case no cell will die . therefore , whether the population will reach full occupation or not can be determined by incorporating @xmath55 into eq .
( 1 ) and transforming the equation into an inequality : @xmath74 } \end{aligned}\ ] ] this expression defines the region of the parameter space in which , even at maximum population , @xmath75 and cells do not die .
reordering the terms , we obtain : @xmath76.\end{aligned}\ ] ] this inequality defines the boundary dividing the inhabited from the uninhabited region in the ( @xmath77 ) phase space . in fig .
( sm-4 ) the boundary for different values of @xmath52 is shown .
if waste degradation performed by @xmath2 cells is far greater than the passive decay term ( @xmath78 ) , then the denominator of eq .
( 13 ) becomes : @xmath79 hence eq .
( 12 ) gets simplified to : @xmath80 and the inequality in eq .
( 14 ) becomes simpler : @xmath81.\end{aligned}\ ] ] the concentration of nutrient at equilibrium is given by : @xmath82
equations ( 14 ) and ( 16 ) show that the boundary that in the mf model separates the inhabited from the uninhabited domain depends on @xmath29 , @xmath30 , @xmath52 and on the other parameters of the model appearing in the equations . as @xmath55 , @xmath62 , @xmath83 and @xmath51 have non - zero , positive values , and @xmath29 , @xmath30 and @xmath52 are constrained to the range @xmath84 $ ] , it can be shown that this boundary , if represented in the ( @xmath85 ) phase space , has a positive linear slope for @xmath86 , which decreases non - linearly as @xmath52 increases , as shown in fig .
( sm-4 ) .
we tested this mf prediction against the actual simulations of the ec model , finding also in the latter a boundary , now defining a sharp transition between full occupation and sparse structures .
the slope is very similar in the ec and mf models ( fig .
6 ) , indicating that the latter properly captures some features of the ec model .
equations ( 14 ) and ( 16 ) also show that there exists a critical @xmath87 beyond which the slope is negative : @xmath88 under such conditions no pair @xmath89 in @xmath84 $ ] can make the inequality true , resulting in a system dominated by death processes . in the simplified scenario described by inequality
( 16 ) , @xmath87 is always in the range @xmath84 $ ] , but if we consider the non - simplified eq .
( 14 ) , then @xmath87 can assume a value out of the @xmath84 $ ] range , when @xmath90 .
[ patterns ] to @xmath91 , @xmath92 , @xmath93 , with a symmetric adhesion matrix and after @xmath94 iterations .
different arrangements -from spots to stripes to mazes- of both cell types are attained , reminiscent of fixed wavelength structures .
( b ) average reachable fraction ( @xmath95 ) of cells of a particular type ( white and black circles represent @xmath1 and @xmath2 cells respectively ) , fixed @xmath96 .
( c ) average domain count ( @xmath97 ) -defined as groups of contiguous cells with the same state- for each pair ( @xmath29 , @xmath30 ) . ]
[ kappa ] in the da - sps expansion of potts model , cells can perform phenotypic switching through a transition rule regulated by transition probabilities @xmath29 and @xmath30 , and their movement is driven by differential adhesion and the tendency to minimize interaction energy between cells .
parameter @xmath32 is introduced as a scaling factor which regulates the relative speed at which cell sorting by adhesion and phenotypic switching occur . using this model
, we assess the role in pattern formation of the individual transition probabilities ( @xmath29 , @xmath30 ) for a fixed @xmath32 .
this analysis reveals that , in spite of model simplicity , @xmath1 and @xmath2 cells are able to self - organize in space in periodic structures .
these patterns can range from spots to stripes to mazes , depending on the relative values of @xmath29 and @xmath30 ( fig .
4a ) . to characterize the phases of the morphospace we applied a standard percolation algorithm to the final macro - state of each simulation . figure ( 4b ) shows that the average reachable fraction of each of the two states @xmath1 and @xmath2 displays a sharp transition .
this defines three clear regimes : non - percolating ( spots ) , percolating ( fully connected maze ) and transition regime , where spots become stripes of increasing length and are marginally able to extend to other domains .
interestingly , this transition occurs for a different value of the control parameter with respect to non - correlated percolation studies @xcite ( @xmath98 instead of @xmath99 ) . for this value
the structures of both cell states percolate , giving rise to the labyrinth phase .
although percolation analysis shows a sharp transition , the number of domains -i.e .
clusters of lattice sites with same state- varies smoothly over the phase space ( fig .
another interesting feature that can be observed in fig .
( 4a , 4c ) is that even if the ratio @xmath100 remains constant , lower values of the transition parameters generate bigger structures with fewer domains . since in our model
@xmath32 is a scaling factor for both @xmath29 and @xmath30 , we set out to quantify the effect of this parameter in the pattern formation process .
( 5a ) shows the qualitative impact of the tuning of @xmath32 in a full - occupation da - sps model . at low @xmath101 values
, sps occurs at a slower pace than the cell sorting process , which is therefore able to properly separate cells in two phases . instead , for high @xmath101 values
, sps occurs at a faster pace than the cell sorting process , which brings about an almost random distribution of states .
[ pqvary ] following the same strategy as before , we ran a set of simulations with periodic boundary conditions , transition probabilities @xmath102 , and different values of @xmath32 .
we then applied a standard fourier transform analysis in order to confirm the existence of dominant frequencies in the spatial distribution of lattice states after @xmath103 iterations .
( 5b ) displays the spatial frequency contribution in a radially averaged power spectrum ( raps ) .
the presence of a single peak in the raps indicates that the periodic structures are built by a single dominant frequency without any specific orientation in the spatial domain , i.e. we obtain fixed wavelength structures reminiscent of turing patterns .
furthermore , the peak position and width are subject to the particular value of @xmath32 , specifically the wavelength of the pattern decreases as @xmath101 increases . the particular mathematical relation between these two variables is displayed in fig .
( 5c ) . [ adhstruct ] in the _ ecology and competition _
( ec ) model cells are not only subject to adhesion processes and phenotypic switching , but can absorb the substrate which is constantly produced all over the lattice , and transform it into energy only ( @xmath1 cells ) or energy and waste ( @xmath2 cells ) .
we compare the ec model with the _ mean - field _
( mf ) model , to understand which properties can be predicted from the latter and which ones instead emerge from the complexity introduced by spatial organization and inhomogeneities in the levels of waste and substrate . we run a set of simulations , each with a different pair of @xmath29 and @xmath30 values , using periodic boundary conditions and fixing @xmath59 and @xmath104 .
the @xmath32 parameter is fixed at @xmath105 , in such a way that adhesion processes occur @xmath106 times faster than phenotypic switching . in the initial time step , @xmath107 of lattice
is occupied by randomly distributed @xmath1 and @xmath2 cells , being the ratio between the two phenotypes already set at the equilibrium value following ( eq .
@xmath108 , @xmath109 ) , depending on @xmath29 and @xmath30 values .
the results in fig .
( 6 ) and in fig .
( sm-2 ) show a clear correspondence between the analytical mf result and the simulated mf , indicating that the model we implemented properly reproduces the expected theoretical results . in the ec model , cells
can also occupy the region of the phase space which was left empty in the mf model .
in fact , while in the mf model simulation all cells die instantaneously when the level of waste reaches @xmath55 ( the maximum level of @xmath35 a cell can sustain ) , in the ec model the death of a few cells can reduce the pressure on the system and allow population survival also in those parametrically unfavoured habitats where waste concentration can locally exceed @xmath55 .
the slope in the ( @xmath85 ) phase space separating inhabited from uninhabited region in the mf model varies depending on the value of @xmath52 and of other relevant parameters of the model ( fig .
sm-4 ) . for @xmath110 ,
black and white cells are indistinguishable , and their behaviour is independent from the value of @xmath29 and @xmath30 , as can be calculated in the mf analytical model . as @xmath52 increases , @xmath2 cells can degrade waste better , but are also less able to elaborate nutrient and can die from lack of energy , unless @xmath29 and @xmath30 are set in such a way that there is a high probability for the @xmath2 cell to switch to a @xmath1 cell before inner @xmath34 value gets below @xmath56 ( the minimum value of @xmath34 a cell needs for survival ) . at steady state
different structures emerge , depending on @xmath29 , @xmath30 and @xmath52 values .
the results for @xmath59 described in this paragraph are shown in fig .
( 6 ) and in fig .
( sm-2 ) . for high values of @xmath29 and @xmath30 , @xmath1 and @xmath2 cells
form turing patterns in a percolating structure . for high values of @xmath29 and low values of @xmath30 ,
@xmath2 cells dominate ( note that for @xmath59 they are still able to degrade w so as not to reach @xmath55 ) . for lower values of @xmath29 and higher values of @xmath30 ,
@xmath1 cells substitute @xmath2 cells less rapidly , @xmath55 can be reached more easily and cells start to die .
lastly , we want to assess whether our ec model , integrating da , sps and metabolism in a habitat with nutrient and toxic waste , can present situations in which structural organization can affect the fitness of cells in the habitat .
to do so , we launch a set of simulations in which cells with different adhesion matrices compete for reproduction . at the beginning of the simulation @xmath111 of the lattice
is occupied by cells , which are divided in equal populations differing only by type of adhesion matrix , which can be _ aggregate _ , _ trabecular _ , _ symmetric _ , _ null _ , _ onions _ , _ sponge _ or _ unicellular _ ( see fig .
the adhesion matrix type is transmitted by each cell to its offspring .
cells are randomly distributed over the lattice regardless of the population they belong to .
it is important to stress that in terms of preferential attachment , cells only sense the states @xmath1 and @xmath2 of neighbouring cells , independently from population type . for each set of parameters @xmath52 , @xmath29 and @xmath30 in the phase space ,
we assess which type of adhesion matrix brings the related population to maximum fitness , by comparing the level of occupation of the lattice for each population . in fig .
( 7a ) we represent cells of state @xmath1 or @xmath2 independently from their population , while in fig .
( 7b ) we show the same cells differentiated by population . at the bottom ( fig .
7c , 7d ) we represent the change of occupation level in time for each population . in the region of the phase space which is fully inhabited in the mf model
, cells can survive independently from their adhesion matrix values and never die .
for this reason the populations related to the various adhesion matrices are equally numbered in this area -with slight differences due only to growth speed before the lattice becomes saturated- and randomly distributed over space , with no emerging structure . however , in the domain of the phase space which was uninhabited in the mf , @xmath29 and @xmath30 have such values that do not guarantee survival of cells .
since cell death may occur in the ec model for this region of the phase space , here the values of the adhesion matrix do make a difference and some species get selected over others .
in particular , the population with ` trabecular ' adhesion matrix prevails . moreover , we can observe that in this area cells organize in a maze structure , exhibiting division of labour within the same population .
lastly , in the frontier between the two zones various populations can coexist , with a prevalence for ` onion ' adhesion matrix at high values of @xmath30 . in fig .
( sm-3 ) we show the relative occupation levels of each population at varying @xmath29 and @xmath30 .
in this paper we have shown a novel way of constructing periodic arrangements of cell types in the form of a hybrid differential adhesion and stochastic switching process .
this mechanism does not rely on differential diffusion ( normally found between the activator and inhibitor species in canonical turing - type systems ) yet it can create the same kind of structures in a predictable , scalable way .
the key ingredients proposed at this level include the differential adhesion hypotheses stemming from steinberg s work ( that considers the minimization dynamics associated with a set of interacting spins or adhesion strengths ) and genetic switches following markovian stochastic dynamics , which are the source of cell diversity and the basis of some adaptive responses displayed by microbial populations @xcite .
the switching dynamics can modify the types of patterns expected from the purely energy - driven scenario , thus indicating that potential forms of phenotypic change can lead to additional richness of pattern forming rules .
a range of spatially ordered structures is obtained displaying characteristic length scales .
being both key ingredients present in extant organisms , we consider that this simple mechanism might have been used originally ( and might be reproduced in the future by synthetic means ) to create regular structures in aggregates and colonies . in relation to pattern formation dynamics
, our hybrid adhesion model offers an alternative way to generate turing patterns , which were up to now directly related with turing s rd mechanisms mediated by a diffusible molecule , or with apparently unrelated but mathematically equivalent systems such as direct contact - mediated regulation by means of which cells are affecting each other s internal rates of reactions @xcite .
in fact , differently from what was proposed by babloyantz , in our hybrid da - sps model the molecules on the surface of one cell do not affect the rates of reactions in its neighbours : the phenotypic switching process occurs in any cell independently from its past and from its neighbour s state , and it is not influenced by the values of the adhesion matrix .
the second relevant aspect considered is how these forming structures might be of benefit to a developing cooperative population in presence of nutrient resources and toxic agents .
to do so we developed an ecology and competition model where a minimal metabolism enables positive or negative interactions between cells .
cells can cooperate by metabolizing waste byproducts , yet they will suffer from decreased growth rates at higher population densities due to substrate attrition . in the ec model
further pattern - forming processes can be predicted . to further asses how structural organization can affect the fitness of cells in the habitat
, we studied how cells with different adhesion matrices compete for reproduction .
interestingly when many populations differing in terms of adhesion properties compete , in the region of the phase space with strong selective pressures only one of the populations survives ( fig . 7 ) .
the selected specie consistently develops a periodic multicellular structure which is superior to both the unicellular and the unstructured multicellular one , suggesting that higher order properties might be of relevance to the establishment of functionality and cooperation .
this simple competition model shows how minimal interaction properties pervading the metabolism of multiple species might come to play a central role in forcing the transition to collective fitness and behaviour , and sets the groundwork for explicitly evolutionary automata @xcite , where cells can optimize several genotype dimensions in order to attain more resources .
we thank the members of the complex systems lab for useful discussions , and amads pags for useful hints on code debugging .
this work has been supported by the botn foundation by banco santander through its santander universities global division , a mineco fellowship and by the santa fe institute .
carroll , s.b .
2001 . _ chance and necessity : the evolution of morphological complexity and diversity_. nature 409,1102 - 1109 nedelcu , a.m. and ruiz - trillo , i. 2015 ( eds . )
_ evolutionary transitions to multicellular life : principles and mechanisms_. springer - verlag , london .
economou , a. d. , ohazama , a. , porntaveetus , t. , sharpe , p. t. , kondo , s. , basson , m. a. and green , j. b. 2012 .
_ periodic stripe formation by a turing mechanism operating at growth zones in the mammalian palate_. nature genetics , 44(3 ) , 348 - 351 .
newman , stuart a. , and ramray bhat .
_ dynamical patterning modules : a `` pattern language '' for development and evolution of multicellular form_. international journal of developmental biology 53.5 ( 2009 ) : 693 .
steinberg , m s. 1975 .
_ adhesion - guided multicellular assembly : a commentary upon the postulates , real and imagined , of the differential adhesion hypothesis , with special attention to computer simulations of cell sorting_. _ j. theor .
_ 55 ( 2 ) : 431 - 43 .
duran - nebreda , s. , bonforti , a. , montaez , r. , valverde , s. , and sol , r. 2015 .
_ emergence of proto - organisms from bistable stochastic differentiation and adhesion_. _ arxiv preprint _ arxiv:1511.02079 . | spatial self - organization emerges in distributed systems exhibiting local interactions when nonlinearities and the appropriate propagation of signals are at work .
these kinds of phenomena can be modeled with different frameworks , typically cellular automata or reaction - diffusion systems .
a different class of dynamical processes involves the correlated movement of agents over space , which can be mediated through chemotactic movement or minimization of cell - cell interaction energy .
a classic example of the latter is given by the formation of spatially segregated assemblies when cells display differential adhesion .
here we consider a new class of dynamical models , involving cell adhesion among two stochastically exchangeable cell states as a minimal model capable of exhibiting well - defined , ordered spatial patterns .
our results suggest that a whole space of pattern - forming rules is hosted by the combination of physical differential adhesion and the value of probabilities modulating cell phenotypic switching , showing that turing - like patterns can be obtained without resorting to reaction - diffusion processes . if the model is expanded allowing cells to proliferate and die in an environment where diffusible nutrient and toxic waste are at play , different phases are observed , characterized by regularly spaced patterns .
the analysis of the parameter space reveals that certain phases reach higher population levels than other modes of organization . a detailed exploration of the mean - field theory is also presented .
finally we let populations of cells with different adhesion matrices compete for reproduction , showing that , in our model , structural organization can improve the fitness of a given cell population .
the implications of these results for ecological and evolutionary models of pattern formation and the emergence of multicellularity are outlined . |
in recent years , approximate characterizations of capacity regions of multi - user systems have gained more and more attention , with one of the most prominent examples being the characterization of the capacity of the gaussian interference channel to within one bit / s / hz in @xcite .
one of the tools that arised in the context of capacity approximations and has been shown to be useful in many cases is the _ linear deterministic model _ introduced in @xcite @xcite . here
, the channel is modeled as a deterministic mapping that operates on bit vectors and mimics the effect the physical channel and interfering signals have on the binary expansion of the transmitted symbols . basically , the effect of the channel is to erase a certain number of ingoing bits , while superposition of signals is given by the modulo addition . even though this model deemphasizes the effect of thermal noise , it is able to capture some important basic features of wireless systems , namely the superposition and broadcast properties of electromagnetic wave propagation . hence , in multi - user systems where interference is one of the most important limiting factors on system performance , the model can also be useful to devise effective coding and interference mitigation techniques .
there are many examples where a linear deterministic analysis can be successfully carried over to coding schemes for the physical ( gaussian ) models or be used for approximative capacity or ( generalized ) degrees of freedom determination , see for example @xcite-@xcite .
* contributions . * from a practical viewpoint , _ cellular systems _ are of major interest . generally , a cellular system consists of a set of base stations each communicating with a distinct set of ( mobile ) users .
effective coding and interference mitigation schemes are still an active area of research .
approximative models such as the linear deterministic approach might help to gain more insight into these problems . in @xcite , the capacity of a basic cellular setup , a multiple access interfering with a point to point link , has been determined for the linear deterministic model in the case of _ symmetric weak interference_. in this paper , we extend these results to the case of arbitrarily strong interference with respect to the achievable sum rate .
furthermore , we use these results to lower bound the generalized degrees of freedom for the corresponding gaussian channel
. * organization . *
the paper is organized as follows : section [ sec : systemmodel ] introduces the system model . in section [ sec :
outerbounds ] , we derive outer bounds on the achievable sum rate . in section [ sec : achievability ] , we construct coding schemes that achieve these outer bounds , thereby characterizing the sum capacity of the system . from these results , a lower bound on the the generalized degrees of freedom for the gaussian case
is derived in section [ sec : gendof ] .
finally , section [ sec : conclusions ] concludes the paper .
* notation .
* throughout the paper , @xmath0 denotes the binary finite field , for which addition is written as @xmath1 , which is addition modulo 2 . for two matrices @xmath2 and @xmath3 , we denote by @xmath4 \in \mathbb{f}_2^{n_a + n_b \times m}$ ] the matrix that is obtained by stacking @xmath5 over @xmath6 .
moreover , for a sequence of matrices @xmath7 , we write @xmath8 for the stacked matrix , i.e. @xmath9 $ ]
. similarly , @xmath10 $ ] stands for placing @xmath5 next to @xmath6 . for a matrix @xmath11 , for @xmath12 the sub - matrix @xmath13 \in \mathbb{f}_2^{j - i+1 \times m}$ ]
is given by taking only the rows @xmath14 to @xmath15 of the matrix @xmath5 .
for @xmath16 , the positive part is denoted by @xmath17 .
finally , @xmath18 and @xmath19 denote integer division and the modulo operation , respectively , where we use the convention @xmath20 .
the system we consider here represents a basic version of the uplink of cellular system and consists of three transmitters ( mobile users ) @xmath21 and @xmath22 and two receivers ( base stations ) @xmath23 .
the system is modeled using the _ linear deterministic model _ @xcite . here
, the input symbol at transmitter @xmath24 is given by a bit vector @xmath25 and the output bit vectors @xmath26 at @xmath27 are deterministic functions of the inputs : defining the shift matrix @xmath28 by @xmath29 the input / output equations of the system are given by rcl y_1 & = & s^q - n_11 x_1 s^q - n_12x_2 s^q - n_13x_3 , + y_2 & = & s^q - n_21 x_1 s^q - n_22x_2 s^q - n_23x_3 . here
, @xmath30 is chosen arbitrarily such that @xmath31 .
note that @xmath32 gives the number of bits that can be passed over the channel between @xmath33 and @xmath34 , i.e. @xmath32 represent channel gains .
there are three messages to be transmitted in the system : @xmath35 denotes the message from transmitter @xmath33 to the intended receiver @xmath34 .
the definitions of ( block ) codes , error probability , achievable rates and the capacity region are according to the standard information - theoretic definitions .
for the remainder of the paper , the transmission rate corresponding to message @xmath36 is represented by @xmath37 , the rate corresponding to @xmath38 by @xmath39 and the rate for @xmath40 by @xmath41 .
the sum rate is written as @xmath42 . in the following ,
we assume without loss of generality that @xmath43 and write @xmath44 . also , we let @xmath45 . in order to keep the presentation clear and reduce the number of cases to be distinguished , we make some further assumptions on the channel gains : we let @xmath46 and @xmath47 .
we remark that these restrictions may seem unrealistic , but can easily be removed and the techniques applied in the following extend to more general cases as well .
the corresponding ( real ) gaussian channel is defined by output symbols rcl [ eq : gaussianchannel ] y_1 & = & h_1 x_1 + h_2x_2 + h_ix_3 + z_1 , + y_2 & = & h_i x_1 + h_ix_2 + h_1 x_3 + z_2 with ( non - varying ) channel coefficients @xmath48 , input symbols @xmath49 subject to power constraints @xmath50 \leq p_i$ ] and additive white gaussian noise @xmath51 .
we remark that a related model with identical channel gains in the two - user cell ( @xmath52 ) is studied in @xcite . in this work , a capacity outer bound
is derived , which is achievable in some cases . for the _ weak interference _ case , for which @xmath53 , the capacity region for the deterministic model has been characterized in @xcite . here
, the maximum sum rate is given by @xmath54 where the function @xmath55 is defined in ( [ eq : phi1 ] ) below .
in the following section , we derive outer bounds on the achievable sum rate for scenarios with arbitrarily strong interference .
before stating the outer bounds , we introduce some definitions . throughout the following , we let @xmath56 , @xmath57 and @xmath58 .
furthermore , for @xmath59 , we define @xmath60 for @xmath61 and @xmath62 .
moreover , we define the functions @xmath63 as ( note that @xmath64 is considered as an even number ) @xmath65 @xmath66 moreover , let @xmath67 . then , we have the following outer bound on the achievable sum rate : [ prop : outerbounds ] for @xmath68
, it holds that @xmath69 \\ 2n_i + \varphi_2(n_2-n_i,\delta ) , & \alpha \in \left(\frac{1}{2},\overline{\alpha}\right ) \\
2n_i + \varphi_2(2n_1- 3n_i,\delta ) , & \alpha \in \left[\overline{\alpha},\frac{2}{3}\right ) \\
\text{min}\left(2n_1,\text{max}(n_1,n_i ) + ( n_1-n_i)^{+}\right ) , & \alpha \in \left[\frac{2}{3},\infty\right)\\ \end{cases}\ ] ] for @xmath70 , l r _ ( ( n_1,n_i ) + ( n_1-n_i)^+ , . + .
2 ( n_i,(n_1-n_i)^+ ) ) .
we remark that the bounds for @xmath70 and @xmath71 coincide with the sum capacity for the interference channel formed by @xmath72 and @xmath73 ( also see section [ sec : achievability ] ) . in order to prove the outer bounds on the sum rate , we first apply fano s inequality :
for each triple of achievable rates @xmath74 , fano s inequality implies that there exists a sequence @xmath75 with @xmath76 for @xmath77 and a sequence of joint factorized distributions @xmath78 such that for all @xmath79 @xmath80 and @xmath81 . ) ; the zero parts due to the channel shifts are not displayed .
the upper figure represents the case @xmath82 and the lower figure the case @xmath83 ( for @xmath68 ) . ]
the upper bound obtained in this way is bounded further in different ways for the different channel gain regimes determined by @xmath84 and @xmath85 .
* i ) the case @xmath68 : * we start with the case @xmath86 . here
, the bound is obtained by letting a genie provide @xmath87 and @xmath88 to receiver @xmath89 , which results in l [ eq : alphagreater23bound ] ( r_1 + r_2 + r_3)n -_nn + i(x_1^n , x_2^n;y_1^n ) + i(x_3^n;y_2^n , x_1^n , x_2^n ) + = h(y_1^n ) + h(x_3^n ) - h(y_1^n|x_1^n , x_2^n ) + n ( 2n_1,(n_1,n_i ) + ( n_1-n_i)^+ ) . for @xmath90 ,
we write @xmath91 $ ] , @xmath92 $ ] , @xmath93 $ ] , @xmath94,x_2^d = x_2^n[\sigma+1:n_1-n_i]$ ] , @xmath95 $ ] and @xmath96 $ ] ( c.f .
figure [ fig : geniediagram ] ) .
then , we have l ( r_1 + r_2 + r_3)n -_nn + i(x_1^n , x_2^n;y_1^n , g_1 ) + i(x_3^n;y_2^n , g_2 ) + = h(g_1 ) + h(y_1^n|g_1 ) - h(y_1^n|g_1,x^n_1,x^n_2 ) + h(g_2 ) + - h(g_2|x^n_3)+ h(y^n_2|g_2 ) - h(y^n_2|g_2,x^n_3 ) + = h(g_1 ) + h(y^n_1|g_1 ) - h ( ) + h(g_2 ) + - h(x_1^t x_2^t ) + h(y^n_2|g_2 ) - h(y^n_2|g_2,x^n_3 ) + 2 n_i n+ h(g_1 ) - h(x_1^d x_2^d ) , where we used @xmath97 , @xmath98 , @xmath99 and @xmath100 .
now @xmath101 which for @xmath102 follows from applying lemma [ lem : entropybound1 ] given below .
a slight modification of lemma [ lem : entropybound1 ] ( which is not given here due to the lack of space ) shows the other case .
note that @xmath103 corresponds to @xmath83 and @xmath102 to @xmath82 .
the bound for @xmath104 is proved by a similar argument , which we sketch only here : @xmath105 and @xmath106 are taken as @xmath107 $ ] and @xmath92 $ ] , respectively and by an appropriate adjustment of lemma [ lem : entropybound1 ] , the bound follows .
we remark that for @xmath108 , the outer bound also follows from the results in @xcite .
* ii ) the case @xmath70 : * the bound @xmath109 follows easily by providing the genie information @xmath110 $ ] and @xmath111 $ ] to receiver @xmath112 and @xmath89 , respectively .
the other bound @xmath113 has already been shown above in ( [ eq : alphagreater23bound ] ) .
[ lem : entropybound1 ] let @xmath114 , @xmath115 be independent random matrices with @xmath116 and @xmath117 , b '' = b[\delta+1:n+\delta]$ ] .
then , it holds that @xmath118 we let @xmath119 and introduce the following labels for blocks of rows of the matrices @xmath120 and @xmath121 : @xmath122 , b ' = [ ( b_k)_{k=0}^{l-1};q'_{b } ] , b '' = [ ( b_k)_{k=1}^{l};q''_{b}]$ ] , where @xmath123 and @xmath124 .
first consider the case that @xmath125 is even .
then we have l h(a b ) - h(a b ) + m+
h + - h + m+ h + - h + = m+ h + - h + m + mq + h + - h + m + mq = m _ 2(n , ) , where the last inequality is due to the independence of @xmath5 and @xmath6 . a similar line of argument can be applied for the case that @xmath125 is odd .
in this section , we describe how the outer bounds given in proposition [ prop : outerbounds ] can be achieved .
it turns out that it suffices to restrict to _ linear coding _ over a single symbol period . for this ,
each transmitter @xmath24 chooses a precoding matrix @xmath126 and transmits the signal @xmath127 , where @xmath128 contains the data . in general , writing @xmath129 , @xmath130 , @xmath131 , @xmath132 , @xmath133 and @xmath134 , it is easy to see that the following rates are achievable under linear precoding : rcl r_1 & = & ( [ a b c ] ) - ( [ b c ] ) , + r_2 & = & ( [ a b c ] ) - ( [ a c ] ) , + r_3 & = & ( [ d e f ] ) - ( [ d e ] ) . in the following , we sketch how to construct precoding matrices @xmath135 that achieve the outer bounds .
the construction of the precoding matrices again is different for different channel parameter configurations . for the cases @xmath70 and @xmath136 ,
the maximum sum - rate can be achieved by keeping transmitter @xmath137 silent and applying the sum - rate optimal interference channel code for the interference channel formed by @xmath72 and @xmath138 .
the results from @xcite show that in this way , the bounds can be achieved ( by linear coding ) . for @xmath139 $ ]
, an optimal construction follows from the results in @xcite , and the extension to the case @xmath140 $ ] is straightforward . here
, the maximum sum rate can be achieved by orthogonal coding , where for each transmitter , a set of bit levels to be used for data transmission is specified such that at the intended receiver , there is no overlap of these levels with levels used by any other transmitter .
the assignment can be interpreted as _ interference alignment _
, where the bit levels are chosen such that the interference caused by @xmath141 and @xmath137 aligns at @xmath89 as much as possible in the levels that are unused by @xmath142 . for the remaining case @xmath90 , the maximum sum rate can not be obtained by orthogonal coding ; instead , _ coding across levels _ is necessary ( as is for the interference channel in a certain interference range @xcite ) .
however , interference alignment still plays a key role . to describe the construction
, we let @xmath143 in the following . for the construction ,
three cases have to be distinguished : a ) @xmath144 , b ) @xmath145 and @xmath83 and c ) @xmath145 and @xmath82 .
due to space limitations , we only consider case c ) here ; the constructions for a ) and b ) are similar . for @xmath146 ,
we let @xmath147 and for an element @xmath148 , @xmath149 denotes the position of the element in the sequence of increasingly ordered elements in @xmath150 . then ,
precoding matrices achieving the outer bound can be constructed as follows : rcl [ eq : optimalcodinrange4 ] v_1(k , k ) & = & 1 , 1 k + v_1(n_1-n_i+k , k ) & = & 1 , 1 k + v_1(+k,+ i^_(k ) ) & = & 1 , k ^ _
+ v_1(n_i + + k,+|^_|+k ) & = & 1 , 1 k n_2-n_i + v_2(k , k ) & = & 1 , 1 k + v_2(n_1-n_i+k , k ) & = & 1 , 1 k + v_2(+ k,+ i^_- ( k ) ) & = & 1 , k ^_- + v_3(k , k ) & = & 1 , 1 k + v_3(n_i + k,+ i^_+ ( k ) ) & = & 1 , k ^_+ + v_3(2(n_1-n_i)+k,+ |^_+ |+ k ) & = & 1 , 1 k an example for this coding scheme is given in figure [ fig : achievabilityexample ] for @xmath151 and @xmath152 .
we summarize this section in the following [ thm : sumcapacity ] the sum capacity for the linear deterministic model equals the expressions given in proposition [ prop : outerbounds ] . ) for @xmath153 , achieving the outer bound @xmath154 ( with @xmath155 ) . here , the bars represent the bit vectors as seen at the two receivers and @xmath156 and @xmath157 are the data bits for @xmath158 and @xmath142 , respectively . ]
in this section , we briefly turn to the gaussian channel given in ( [ eq : gaussianchannel ] ) .
it can equivalently be written as rcl y_1 & = & x_1 + x_2 + x_3 + z_1 , + y_2 & = & x_1 + x_2 + x_3 + z_2 with @xmath159 and with power constraints @xmath160 .
then , the _ generalized degrees of freedom _ @xcite are defined as @xmath161 where @xmath162 is the sum capacity of the channel .
this measure represents a high snr description of the system , where the channel gains are kept in constant relation ( determined by @xmath163 and @xmath164 ) in the db scale , which allows a more detailed asymptotic description of the system as opposed to the degrees of freedom characterization ( also see e.g. @xcite ) . .
] . ]
the technique given in @xcite @xcite ( and also applied in e.g. @xcite ) allows to transfer an achievable scheme for the deterministic model to a coding scheme for the gaussian channel .
the basic idea is to inscribe the code obtained for the linear deterministic model into a @xmath165-ary expansion of the transmit signals , asymptotically mimicking the behavior of the deterministic channel .
this technique can also be applied for the channel at hand . omitting the details , we just remark that here we use the fact that for channel gain approximation sequences @xmath166 and @xmath167 for @xmath168 ( where @xmath169,@xmath170 , @xmath171 , we have that @xmath172 for @xmath168 .
similarly , @xmath173 and @xmath174 . using the sum capacity results from theorem [ thm : sumcapacity ]
, one obtains the following lower bound on the generalized degrees of freedom : [ prop : gendof ] for @xmath175 , it holds that @xmath176 \\ 2a + \varphi_2(b - a,1-b ) , & a \in \left(\frac{1}{2},\overline{a}\right ) \\
2a + \varphi_2(2 - 3a,1-b ) , & a \in \left[\overline{a},\frac{2}{3}\right ) \\
\text{min}\left(2,\text{max}(1,a ) + ( 1-a)^{+}\right ) , & a \in \left[\frac{2}{3},\infty\right).\\ \end{cases}\ ] ] for @xmath177 , @xmath178 figure [ fig : gendof ] displays the lower bound on @xmath179 in the range @xmath180 , b \in [ 0.8,1]$ ] . for @xmath181 ,
the achievable generalized degrees of freedom are shown in figure [ fig : wcurve ] for different @xmath163 values , together with the generalized degrees of freedom for the interference channel consisting of only @xmath72 and @xmath73 .
note that the latter one represents the well - known _ w curve _ @xcite of the generalized degrees of freedom for the interference channel . for
@xmath182 , the channel gain difference in the two - user cell can be exploited for interference alignment , pushing the achievable generalized degrees of freedom higher than the w curve , whereas for @xmath183 , the lower bound can be achieved by coding only for the interference channel consisting of @xmath72 and @xmath73 .
in this paper , we studied the linear deterministic model for a cellular - type channel where a two user multiple access channel mutually interferes with a point to point link . under certain symmetry assumptions on the channel gains
, we derived the sum capacity and the corresponding transmission schemes , which use interference alignment and linear coding across bit levels .
while for a large parameter range , the sum capacity is identical to the sum capacity of the interference channel obtained by silencing the weaker user in the two - user cell ( multiple access channel ) , for a certain parameter range , the channel gain difference of in the two - user cell allows to get a higher sum rate using interference alignment .
finally , from these results , a lower bound on the generalized degrees of freedom for the gaussian channel was given , increasing the w curve for the interference channel in a certain interference range .
although we have considered a restricted setup in this paper , we believe that the achievability and converse arguments used in this paper give valuable insights for the consideration of more general systems . future work will study extensions to the more general cases with additional users .
another interesting direction for future investigations is to further explore the connections to the gaussian equivalent of the channel , specifically concerning outer bounds on the generalized degrees of freedom of the system and approximate capacity characterizations .
v. cadambe , s. jafar , and s. shamai , `` interference alignment on the deterministic channel and application to fully connected gaussian interference networks , '' _ ieee trans .
inform . theory _
55 , no . 1 ,
pp . 269274 , 2009 .
c. huang , v. cadambe , and s. jafar , `` interference alignment and the generalized degrees of freedom of the x channel , '' in _ proc .
information theory ( isit ) _ ,
seoul , korea , june / july 2009 , pp . 19291933 .
j. bhler and g. wunder , `` on interference alignment and the deterministic capacity for cellular channels with weak symmetric cross links , '' in _ proc .
symp . on information theory ( isit )
_ , saint petersburg , russia , 2011 , accepted for publication , available at arxiv:1104.0136 . | in this paper , we use the linear deterministic approximation model to study a two user multiple access channel mutually interfering with a point to point link , which represents a basic setup of a cellular system .
we derive outer bounds on the achievable sum rate and construct coding schemes achieving the outer bounds . for a large parameter range , the sum capacity is identical to the sum capacity of the interference channel obtained by silencing the weaker user in the multiple access channel . for other interference configurations
, the sum rate can be increased using interference alignment , which exploits the channel gain difference of the users in the multiple access channel . from these results , lower bounds on the generalized degrees of freedom for the gaussian counterpart are derived . |
this paper is the second in a series of articles @xcite on the statistics of vacua in string / m theory and associated effective supergravity theories .
mathematically , vacua are critical points @xmath19 of a holomorphic section @xmath20 of a line bundle @xmath21 over a complex manifold relative to a connection @xmath22 which we always choose to be the chern connection @xmath2 of a hermitian metric @xmath5 on @xmath23 .
equivalently , they are critical points of the norm @xmath24 of @xmath25 relative to @xmath5 .
our motivation to study critical points of holomorphic sections in this metric sense comes initially from physics , where critical points of this kind model extremal black holes in addition to string / m vacua ( cf .
but we also find the statistics of critical points to have an independent geometric interest . a basic statistical quantity
, the average number of critical points of a random holomorphic section , defines a new geometric invariant of a positive hermitian holomorphic line bundle , and in this paper we show that its asymptotic minima are given by calabi extremal metrics .
the physical setting for vacua of string / m theory ( and for extremal black holes ) is a holomorphic line bundle over the moduli space of complex structures on a polarized calabi - yau manifold .
supersymmetric vacua are critical points of a holomorphic section ( known as a superpotential ) relative to the weil - petersson metric .
the program of studying statistics of critical points of gaussian random holomorphic sections was proposed by the first author @xcite as a means of dealing with the large number of string / m vacua .
there exists at this time no reasonable selection principle to decide which superpotential nor which of its critical points gives the vacuum state which correctly describes our universe in string / m theory , so it makes sense to study the statistics of vacua of random superpotentials . in our first paper @xcite
, we gave explicit formulas for the expected distribution of critical points of sections of general holomorphic line bundles over any complex manifold , including those which arise in physics .
the formulas , recalled in [ background ] ( cf .
theorem [ kncrit2 ] ) , involve complicated complex symmetric matrix integrals , and it is difficult to see how the expected distribution of critical points depends on the metric @xmath5 .
the purpose of this article is therefore to study a purely geometric simplification of the physical problem where @xmath26 is a positive hermitian line bundle over a ( usually compact ) khler manifold @xmath3 and where the gaussian measure on @xmath27 is derived from the inner product induced by @xmath5 .
our aim is to understand the metric dependence of the statistics of the random critical point set @xmath28 of a gaussian random section of @xmath27 .
we note that @xmath29 where @xmath30 is the zero set of @xmath25 . from the probabilistic viewpoint ,
the critical points of random holomorphic sections relative to the chern connection @xmath2 of a fixed hermitian metric on @xmath23 determine a _ point process _ on @xmath3 , that is , a measure on the configuration space of finite subsets of @xmath3 , which gives the probability density of a finite subset being the critical point set of a holomorphic section .
the process is determined by its @xmath31-point correlation functions @xmath32 , which give the probability density of critical points occurring simultaneously at the points @xmath33 .
in this article , we focus on the @xmath34-point correlation , namely the expected distribution of critical points @xmath35\,d{\gamma}_h(s)\ ] ] where @xmath36 is the dirac point mass at @xmath37 , and where @xmath38 is the gaussian measure probability @xmath39 on @xmath27 induced by @xmath5 and the khler form @xmath40 ( see [ background ] ) .
we showed in @xcite that @xmath41 is a smooth form on @xmath3 if @xmath23 is sufficiently positive .
in particular , we are interested in the expected ( average ) number of critical points @xmath42 of a random section , a purely geometric invariant of a positive hermitian holomorphic line bundle @xmath43
. the methods of this paper also give results on general @xmath31-point correlations between critical points and their scaling asymptotics in the sense of @xcite , but we do not carry out the analysis here . since it is a crucial point ,
let us explain why the number @xmath44 is a ( non - constant ) random variable on @xmath45 , unlike the number of zeros of @xmath12 independent sections which is a topological invariant of @xmath23 . as indicated above , connection critical points are the same as critical points of @xmath46 for which @xmath47 , or equivalently as critical points of @xmath48 ( see @xcite for the simple proof ) .
hence , there are critical points of each morse index @xmath49 ( see @xcite ) , and only the alternating sum of the number of critical points of each index is a topological invariant .
another way to understand the metric dependence of the number of critical points is to write the covariant derivative in a local frame @xmath50 as @xmath51 where we locally express a section as @xmath52 , and @xmath53 .
hence , the critical point equation @xmath54 in the local frame fails to be holomorphic when the connection form is only smooth .
although @xmath55 and @xmath56 depend on @xmath5 , it is not clear at the outset whether @xmath4 is a topological invariant or to what degree it depends on the metric @xmath5 . to investigate the metric dependence of @xmath57 and @xmath4 we consider their asymptotic behavior as we take powers @xmath58 of @xmath23 . as in @xcite
, it is natural to expect that the density and number of critical points will have asymptotic expansions which reveal their metric dependence .
we therefore let @xmath59 denote the expected distribution of critical points of random holomorphic sections @xmath60 with respect to the chern connection and hermitian gaussian measure induced by @xmath5 , as given by in [ background ] .
we also let @xmath61 denote the expected number of critical points .
the covariant derivative associated to @xmath62 has the semi - classical form @xmath63 we let @xmath64 denote the curvature form of @xmath65 .
our first result is a complete asymptotic expansion for the expected distribution of critical points for powers @xmath66 in terms of curvature invariants of the khler metric @xmath67 : [ univdist ] for any positive hermitian line bundle @xmath68 over any compact khler manifold with @xmath69 , the expected critical point distribution of holomorphic sections of @xmath58 relative to the hermitian gaussian measure induced by @xmath5 and @xmath70 has an asymptotic expansion of the form @xmath71 where the coefficients @xmath72 are curvature invariants of order @xmath73 of @xmath74 . in particular
, @xmath75 is the universal constant @xmath76 @xmath77 , where @xmath14 is the scalar curvature of @xmath70 and @xmath78 is a universal constant , and @xmath79 is a quadratic curvature polynomial . here , @xmath80 is the space of @xmath81 complex symmetric matrices .
it follows that the density of critical points is asymptotically uniform relative to the curvature volume form with a universal asymptotic density @xmath82 .
the values of the constant @xmath75 for low dimensions are : @xmath83 ( see appendix 1 . ) with only minor changes in the proofs , our methods give refinements of the asymptotic results which take the morse indices of the critical points into account . by the morse index @xmath84 of a critical point ,
we mean its morse index as a critical point of @xmath85 ; it is well known that @xmath86 for positive line bundles @xcite .
thus we let @xmath87 denote the expected distribution of critical points of @xmath88 of morse index @xmath84 , and we let @xmath89 denote the expected number of these critical points .
thus we have @xmath90 we obtain a similar asymptotic expansion for the distribution of critical points of given morse index : [ univmorse ] let @xmath91 be as above
. then the expected density of critical points of morse index @xmath84 of random sections in @xmath92 is given by @xmath93 where the @xmath94 are curvature invariants of order @xmath73 of @xmath74 .
in particular , @xmath95 is given by the integral in with the domain of integration @xmath96 replaced by @xmath97 with @xmath98 .
the coefficient integrals are very complicated to evaluate except in dimension one , where we obtain the following formula : [ critrs ] let @xmath65 be a positive line bundle on a compact complex curve @xmath99 of genus @xmath100
. then the expected number of saddle points and of local maxima , respectively , of random holomorphic sections of @xmath58 are given by : @xmath101 where @xmath102 , and @xmath14 is the gaussian curvature of the metric @xmath70 . hence , the expected total number of critical points in dimension 1 is @xmath103 in dimension 1 , the coefficients in theorem [ univmorse ] are : @xmath104 in appendix 1 , we provide a table of numerical values of @xmath105 ( for @xmath106 ) , which we computed using maple . in [ s - exact ] , we study the case where @xmath107 and @xmath65 is the hyperplane section bundle @xmath108 with the fubini - study metric .
we show that in this case the expected number @xmath89 of critical points of morse index @xmath84 of sections in @xmath109 is a rational function of @xmath8 , and we provide an exact formula ( proposition [ numcrit ] ) for this expected number . in appendix 1 , we use proposition [ numcrit ] and a computer assisted computation to give explicit formulas for @xmath89 in dimensions @xmath110 . for example , in dimension 1 , the expected total number of critical points , as computed in @xcite , is @xmath111 and in dimension 2 , the expected total number is @xmath112 thus we gain a quantitative sense of how many more critical points there are in the metric sense in comparison with the classical critical point equation @xmath113 . in the case of @xmath114 , whose sections are polynomials of degree @xmath8 , we may view this classical critical point equation as a connection critical point equation by viewing the derivative @xmath115 as a flat meromorphic connection with pole at @xmath116 .
alternately , it is the chern connection of a singular hermitian metric . the critical point equation being purely holomorphic , the number of critical points of a generic section is a constant @xmath117 .
all critical points relative to this connection are saddle points . by comparison ,
sections @xmath118 have , on average , an additional @xmath119 local maxima and @xmath120 additional saddles relative to the fubini - study connection .
the study of critical points relative to meromorphic connections ( known as minkowski vacua ) is simpler than that relative to chern connections and will be explored further in a subsequent work . as a corollary of theorem [ univmorse ]
, we find the rather surprising fact that the asymptotics of the expected number of critical points is a topological invariant of the bundle @xmath21 to two orders in @xmath8 .
[ en ] let @xmath121 be a positive holomorphic line bundle on a compact khler manifold , with @xmath122 . then the expected number of critical points of morse index @xmath84 ( @xmath86 ) of random sections in @xmath92 has the asymptotic expansion @xmath123n^m + \left[\frac{\pi^m{\beta}_{1q}}{(m-1)!}\ , c_1(m)\cdot c_1(l)^{m-1}\right]n^{m-1 } \nonumber \\&+ & \left[{\beta}_{2q}\int_m\rho^2d{{\operatorname{vol}}}_h + { \beta}'_{2q}\ , c_1(m)^2\cdot c_1(l)^{m-2 } + { \beta}''_{2q}\ , c_2(m)\cdot c_1(l)^{m-2}\right ] n^{m-2}+\cdots\,,\nonumber\end{aligned}\ ] ] where @xmath124 are universal constants depending only on the dimension @xmath12 .
we recall that the alternating sum of the @xmath89 is topological : @xmath125 and hence by corollary [ en ] , @xmath126 since each @xmath127 is strictly positive and their sum equals @xmath75 , it follows from that @xmath128 corollary [ en ] shows that the metric dependence of @xmath129 is a ` lower - order ' effect , and hence renews the question to what degree @xmath130 , or at least the asymptotic expansion of @xmath131 , is a topological invariant .
we see from corollary [ en ] that the expansion is not topological provided that the constant @xmath132 does not vanish .
computations in dimensions @xmath133 show that @xmath134 is positive in these dimensions ( see corollary [ less4 ] ) , and we conjecture that @xmath135 for all @xmath12 . as we now explain , this conjecture is suggested by a connection between extremals of @xmath136 and calabi extremal metrics @xcite .
this connection involves a notion of asymptotic minimality of @xmath137 . to introduce it ,
we revisit the motivating problem of determining how @xmath4 varies as @xmath5 varies over hermitian metrics on @xmath23 .
one could consider all hermitian metrics on @xmath23 , but we focus on the smaller class of positively curved hermitian metrics , @xmath138 if we fix one such metric @xmath139 , the others may be expressed as @xmath140 with @xmath141 it is reasonable to conjecture that @xmath142 is unbounded as @xmath143 varies over @xmath144 , since by ( [ cpe ] ) the number of critical points of a section should be ` large ' if the ` degree ' of the connection form @xmath145 is ` large ' .
here , @xmath146 in a local frame . on the other hand , @xmath4 is bounded below by @xmath147 , and it is plausible that it has a minimum
. it would be interesting to determine this minimal metric ( assuming one exists and is unique ) , but it is difficult to solve the critical point equation @xmath148 .
we therefore consider the simpler asymptotic problem .
since the first two leading coefficients in the expansion of @xmath137 are topological , we try to find metrics for which the first non - topological term is critical .
the first non - topological term is a multiple of the calabi functional @xmath149 where @xmath150 is the scalar curvature of the khler metric @xmath122 , and @xmath151 .
thus the problem of finding metrics which are critical for the metric invariant @xmath137 is closely related to the problem of finding critical points of calabi s functional . in keeping with our intuition that @xmath137 should have a minimizer , we note that critical points of calabi s functional are necessarily minima ( cf .
@xcite ) .
existence of critical metrics is one of the fundamental problems in complex geometry , and we refer to @xcite for background .
it was suggested by s .-
yau @xcite that existence of a canonical metric should be related to the stability of @xmath3 .
one class of canonical metrics are hermitian metrics @xmath5 for which @xmath152 is a khler metric of constant scalar curvature , i.e. for which @xmath150 is constant . by a theorem due to s. donaldson ( * ? ? ?
* cor . 5 )
, there exists at most one khler metric of constant scalar curvature in the cohomology class of @xmath153 .
hence if there exists such a metric of constant scalar curvature , there exists a unique hermitian metric minimizing calabi s functional .
this leads us to make the following definition : let @xmath21 be an ample holomorphic line bundle over a compact khler manifold . for @xmath154 and @xmath155
, we say that @xmath156 ( resp .
@xmath137 ) is _ asymptotically minimal _ if for all @xmath157 in @xmath158 , there exists @xmath159 such that @xmath160 assuming @xmath161 has a hermitian metric @xmath5 minimizing calabi s functional , we see from corollary [ en ] that @xmath156 ( resp .
@xmath137 ) is asymptotically minimal as long as @xmath162 ( resp .
@xmath163 ) .
since we believe this to be the case for all dimensions , we state the following conjecture .
[ conj ] let @xmath164 , @xmath86 .
then the following are equivalent : * @xmath137 is asymptotically minimal , * @xmath156 is asymptotically minimal ( and hence @xmath165 is asymptotically minimal @xmath166 ) , * @xmath5 minimizes calabi s functional . in lemma [ k0 ] , we show that @xmath167 where @xmath168 it is unfortunately difficult to determine from this formula whether @xmath169 is non - zero and ( if so ) what sign it has . in [ s - coeff ] , we transform to a rather complicated , but more elementary , integral ( lemma [ li4 ] ) , which can be evaluated by a routine ( but long ) calculation . for low dimensions , we perform this calculation using maple to obtain : [ asympmin ] the constants @xmath169 are positive for @xmath170 , and hence conjecture [ conj ] is true for @xmath171 . the calculation in dimension @xmath172 was done by b. baugher @xcite .
in particular , we have : [ less4 ] suppose that @xmath171 and that @xmath23 possesses a metric @xmath5 for which the scalar curvature of @xmath173 is constant . then @xmath5 is the unique metric on @xmath23 such that @xmath137 is asymptotically minimal.[calabi ] thus , for instance , the fubini - study metric @xmath5 on the hyperplane section bundle @xmath174 is the unique metric on @xmath108 such that @xmath137 is asymptotically minimal , at least for @xmath170 .
additional evidence that @xmath175 was found by b. baugher @xcite . by studying patterns in computer - assisted calculations of the terms arising from the computation of the integral in ( [ k0intro ] ) , baugher discovered an identity for @xmath176 for dimensions @xmath17 which he conjectures is valid in all dimensions .
baugher showed that this identity implies that @xmath177 in all dimensions , and hence baugher s conjecture implies conjecture [ conj ] .
baugher s conjecture and identity are stated in appendix 2 .
we close the introduction with some comments on the organization of the paper .
the proof of theorems [ univdist ] and [ univmorse ] is based on the tian - yau - zelditch asymptotic expansion of the szeg kernel @xmath178 @xcite and on formulas from @xcite for the density of critical points .
we then need to evaluate the coefficients explicitly to obtain concrete results linking geometry to numbers of critical points .
once we know the leading coefficient is universal , we may calculate it for @xmath179 and this is done in [ s - exact ] and appendix 1 .
unfortunately , the fubini - study metric is not useful for finding the sign of @xmath180 since it is impossible to separate out the topologically invariant terms from the calabi functional for this metric .
hence in [ s - coeff ] , we analyze instead the case of @xmath181 for @xmath182 an elliptic curve , where the relevant topological terms vanish .
this leads to an explicit integral which we analyze by a variant of the itzykson - zuber formula in random matrix integrals .
we thank x. x. chen and z. lu for discussions of calabi s functional and references to the literature .
let @xmath183 be a hermitian holomorphic line bundle over a complex manifold @xmath3 , and let @xmath184 be its chern connection , i.e. the unique connection of type @xmath185 on @xmath23 compatible with both the metric and complex structure of @xmath23 .
thus , it satisfies @xmath186 for any holomorphic section @xmath25 where @xmath187 is the splitting of the connection into its @xmath188 , resp.@xmath189 parts .
it follows that @xmath190 we denote by @xmath191 the curvature of @xmath5 and @xmath192 we now introduce the gaussian measures @xmath39 , called _
hermitian gaussian measures _ in @xcite which we use exclusively in this paper .
they are determined by the inner product @xmath193 on @xmath27 , where @xmath194 . by definition ,
@xmath195 where @xmath196 is lebesgue measure and @xmath197 is an orthonormal basis for @xmath198 relative to @xmath199 .
[ defk ] the _ expected distribution of critical points _ of @xmath200 with respect to @xmath39 is defined by @xmath201\,d{\gamma}_h(s)\ ] ] where @xmath36 is the dirac point mass at @xmath37 ; i.e. , @xmath202\ , d\gamma_{h}(s)\ , , \qquad { \varphi}\in { \mathcal{c}}^\infty(m)\,.\ ] ] the density of @xmath57 with respect to @xmath203 is denoted @xmath204 ; i.e. , @xmath205 let @xmath206 be a hermitian holomorphic line bundle on an @xmath12-dimensional khler manifold .
we say that @xmath207 has the @xmath208-jet spanning property if all possible values and derivatives of order @xmath209 are attained by the global sections @xmath20 at every point of @xmath3 .
in @xcite we showed that if @xmath207 has the @xmath208-jet spanning property , then @xmath57 is absolutely continuous with respect to @xmath203 , and we obtained an integral formula for @xmath210 in terms of the szeg kernel @xmath211 of @xmath212 with respect to @xmath5 . to describe this formula ,
we choose normal coordinates about @xmath213 and define the following matrices : @xmath214 \,,\label{b}\\[8pt ] c(z_0)&=&\left [ \begin{array}{cc } \big(\tau_{jq}\tau_{j'q'}\ , \nabla_{z_q}\nabla_{z_j}\nabla_{\bar w_{q'}}\nabla_{\bar w_{j'}}\pi\big ) & \big(\tau_{jq}\ , \nabla_{z_q}\nabla_{z_j}\pi\big ) \\[8pt ] \big(\tau_{j'q'}\,\nabla_{\bar w_{q'}}\nabla_{\bar w_{j'}}\pi\big ) & \pi\end{array}\right ] \,,\label{c}\\ & & \tau_{jq}=\sqrt 2 \ \ \mbox { if } \
j < q\;,\quad \tau_{jj}=1\;,\label{tau}\\ & & \qquad \qquad 1\le j\le m\,,\ 1\le j\le q\le m\,,\ 1\le j'\le q'\le m\,,\nonumber\end{aligned}\ ] ] where the szeg kernel @xmath215 and its derivatives are evaluated at @xmath216 ( see [ s - szego ] and ) . in the above matrices , @xmath217 index the rows , and @xmath218 index the columns .
note that @xmath219 are @xmath220 matrices , respectively , where @xmath221 we then let @xmath222 the matrices @xmath219 give the second moments of the joint probability distribution of the random variables @xmath223 and @xmath224 on @xmath198 .
@xcite [ kncrit2 ] let @xmath43 denote a positive holomorphic line bundle with the 2-jet spanning property .
give @xmath3 the volume form @xmath225 induced from the curvature of @xmath23 .
then the expected density of critical points relative to @xmath203 is given by @xmath226 here , @xmath227 is a complex symmetric matrix , @xmath228 and @xmath229 denote lebesgue measure , and @xmath230 is the hermitian operator on the complex vector space @xmath231 described as follows : let @xmath232 , be the basis for @xmath233 given by @xmath234 i.e. , for @xmath235 , @xmath236 is the matrix with @xmath237 in the @xmath238 and @xmath239 places and 0 elsewhere , while @xmath240 is the matrix with 1 in the @xmath241 place and 0 elsewhere .
we note that @xmath242 is an orthonormal basis ( over @xmath243 ) for @xmath80 with respect to the hilbert - schmidt hermitian inner product @xmath244 for @xmath245 , we have @xmath246 where @xmath247 is given by . lebesgue measure @xmath228 ( with respect to the hilbert - schmidt norm )
is given by @xmath248 writing @xmath249\big(\lambda_{0}^{j'q'}\big ) & \lambda_0 ^ 0 \end{array}\right]\;,\ ] ] we then define @xmath250 to study the asymptotics , we consider powers @xmath58 and we use the following result .
[ kncrit ] with the same notation and assumptions as above , the density of the expected distribution @xmath251 of critical points of random sections @xmath252 relative to @xmath203 is given by @xmath253 where @xmath254 @xmath255\,,\label{an}\\ b_n(z_0)&= & \big[\big(\tau_{j'q'}\,\nabla_{z_j}\nabla_{\bar w_{q'}}\nabla_{\bar w_{j ' } } \pi_n\big ) \quad \big(n\,\nabla_{z_j}\pi_n\big ) \big ] \,,\label{bn}\\[8pt ] c_n(z_0)&=&\left [ \begin{array}{cc } \big(\tau_{jq}\tau_{j'q'}\ , \nabla_{z_q}\nabla_{z_j}\nabla_{\bar w_{q'}}\nabla_{\bar w_{j'}}\pi_n\big ) & \big(\tau_{jq}n\ , \nabla_{z_q}\nabla_{z_j}\pi_n\big ) \\[8pt ] \big(\tau_{j'q'}n\,\nabla_{\bar w_{q'}}\nabla_{\bar w_{j'}}\pi_n\big ) & n^2\,\pi_n\end{array}\right ] \,,\label{cn}\\ & & \tau_{jq}=\sqrt 2 \ \ \mbox { if } \
j < q\;,\quad \tau_{jj}=1\;,\nonumber\\ & & \qquad \qquad 1\le j\le m\,,\ 1\le j\le q\le m\,,\ 1\le j'\le q'\le m\,,\nonumber\end{aligned}\ ] ] where @xmath256 and its derivatives are evaluated at @xmath216 .
rescale @xmath257 .
then the curvature of @xmath58 is given by @xmath258 so that the @xmath259 are normal coordinates ( at a point @xmath260 ) for the curvature of @xmath58 .
apply theorem [ kncrit2 ] , using the coordinates @xmath261 to obtain @xmath262 . since @xmath263 and the transformation @xmath264 introduces a factor @xmath265 , we let @xmath266 to obtain the desired formula .
we also have a formula for the density of critical points of specific morse indices : [ morse ] under the above assumptions , the density relative to @xmath203 of the expected distribution @xmath267 of critical points of morse index @xmath84 of @xmath268 for random sections @xmath252 is given by @xmath269 where @xmath270 the case @xmath271 is given as theorem 6 in @xcite .
the general case follows immediately by rescaling as in the proof of corollary [ kncrit ] .
recall that the index of a nonsingular hermitian matrix is the number of its negative eigenvalues , and the morse index of a nondegenerate critical point of a real - valued function is the index of its ( real ) hessian . as in our previous work ,
it is useful to lift the analysis on positive line bundles @xmath21 to the associated principal @xmath272 bundle @xmath273 .
sections then become scalar functions and it is simpler to formulate various asymptotic properties for powers @xmath58 @xcite .
the same analysis is also useful for general line bundles although the asymptotic results no longer hold . given a holomorphic line bundle @xmath23 and a hermitian metric @xmath5 on @xmath23 , we obtain a hermitian metric @xmath274 on the dual line bundle @xmath275 and we define the associated circle bundle by @xmath276 .
thus , @xmath277 is the boundary of the disc bundle @xmath278 , where @xmath279 .
when @xmath280 is a positive line bundle , the disc bundle @xmath281 is strictly pseudoconvex in @xmath275 , hence @xmath277 inherits the structure of a strictly pseudoconvex cr manifold .
when @xmath23 is negative , as is the case for the line bundles relevant to string theory , @xmath277 is pseudo - concave .
we endow @xmath277 with the contact form @xmath282 and the associated volume form @xmath283 we define the hardy space @xmath284 of square - integrable cr functions on @xmath277 , i.e. , functions that are annihilated by the cauchy - riemann operator @xmath285 and are @xmath286 with respect to the inner product @xmath287 we let @xmath288 ( @xmath289 ) denote the @xmath272 action on @xmath277 and denote its infinitesimal generator by @xmath290 .
the @xmath272 action on @xmath277 commutes with @xmath291 ; hence @xmath292 where @xmath293 .
a section @xmath294 of @xmath58 determines an equivariant function @xmath295 on @xmath275 by the rule @xmath296 where @xmath297 .
we henceforth restrict @xmath298 to @xmath277 and then the equivariance property takes the form @xmath299 .
the map @xmath300 is a unitary equivalence between @xmath301 and @xmath302 .
we let @xmath50 be a nonvanishing local section , or local frame , of @xmath23 .
as above , we write @xmath303 thus , a positive line bundle @xmath23 induces the khler form @xmath304 with khler potential @xmath305 .
the szeg kernel @xmath306 is the kernel of the orthogonal projection @xmath307 ; it is defined by @xmath308 let @xmath309 be an orthonormal basis for @xmath92 . then @xmath310 is an orthonormal basis of @xmath311 , and the szeg kernel can be written in the form @xmath312 it is the lift of the section @xmath313 where @xmath314 we let @xmath315 denote the coordinates of the point @xmath316 .
the equivariant lift of a section @xmath317 is given explicitly by @xmath318 } f(z)\;.\ ] ] the szeg kernel is then given by @xmath319}f_n(z,\bar w)\;.\ ] ] we denote by @xmath320 and obtain a splitting @xmath321 into horizontal and vertical spaces .
the chern connection @xmath322 on @xmath58 then lifts to @xmath277 as the horizontal derivative @xmath323 , i.e. @xmath324 to describe the connection explicitly , we choose local holomorphic coordinates @xmath325 in @xmath3 , and we write @xmath326 in particular , @xmath327 , which vanishes when the section @xmath294 is holomorphic , or equivalently , when @xmath328 . for a section @xmath329 of @xmath58 , we have @xmath330 we also write @xmath331 where @xmath323 is the horizontal derivative on @xmath332 .
( we used the fact that the horizontal derivatives of @xmath256 with respect to the @xmath333 and @xmath334 variables vanish . ) by , we have @xmath335 } \left ( \frac{{\partial}}{{\partial}z_j } -n \frac{{\partial}k } { { \partial}z_j}(z,\bar z)\right)f_n(z,\bar w)\;,\label{nablaz}\\ \nabla_{\bar w_j}\pi_n & = & e^{n\left[-{{\frac{1}{2}}}k(z,\bar z)-{{\frac{1}{2}}}k(w,\bar w ) + i(\theta-{\varphi})\right ] } \left ( \frac{{\partial}}{{\partial}\bar w_j } -n \frac{{\partial}k } { { \partial}\bar w_j}(w,\bar w)\right)f_n(z,\bar w)\;.\label{nablaw}\end{aligned}\ ] ]
the integrals in theorems [ kncrit2][morse ] are difficult to evaluate because of the absolute value sign , which prevents a direct application of wick methods . to compute the densities
, we shall replace our integral by another one which can be evaluated by residue calculus in certain cases .
this new integral is given by the following lemma : [ izint ] let @xmath336 be a positive definite hermitian operator on @xmath337 . then @xmath338 } \
, dg\ , d \xi\ , d\lambda,\end{aligned}\ ] ] where * @xmath339 , * @xmath340 is unit mass haar measure on @xmath341 , * @xmath342 is the hermitian operator on @xmath343 given by @xmath344 * @xmath14 is the representation of @xmath341 on @xmath343 given by @xmath345 the integrand is analytic in @xmath346 but rather complicated .
its principal features are : * @xmath347 are homogeneous polynomials of degree @xmath348 , and @xmath349 is homogeneous of degree @xmath12 ; * @xmath350 $ ] is a ( family of ) polynomial(s ) in @xmath351 of degree @xmath352 with no real zeros @xmath353 .
but the polynomials are not elliptic ( or even hypo - elliptic ) , that is , do not satisfy @xmath354 we may drop the second term @xmath355 and find that the growth at infinity is that of @xmath356 $ ] . since @xmath357 is a diagonal matrix
as described in theorem [ i ] , its determinant is a product of linear polynomials in @xmath351 , and hence vanishes along a union of real hyperplanes .
* the ratio @xmath358 } $ ] is thus a rational function in @xmath351 which is a ` symbol ' of order @xmath359 , i.e. each @xmath351-derivative decays to one extra order .
repeated partial integrations in @xmath360 using @xmath361 e^{i \langle \lambda , \xi \rangle } = e^{i \langle \lambda , \xi \rangle}$ ] simultaneously lowers the order in both @xmath351 and @xmath362 by two and renders the @xmath363 integral absolutely convergent without the gaussian factor .
the proof of lemma [ izint ] is given in
[ simplify ] below . as a consequence
, we have the following alternative formula for the expected critical point density : [ i ] under the hypotheses of theorem [ kncrit2 ] and notation of lemma [ izint ] , the density of the expected distribution of critical points of sections of @xmath92 is also given by : @xmath364 } \ , dg\ , d \xi\ , d\lambda,\ ] ] where @xmath365 the formula follows by combining corollary [ kncrit ] and lemma [ izint ] . in [ s - exact ] we shall use theorem [ i ] to calculate the density of critical points for random sections @xmath366 of the @xmath8-th power of the hyperplane bundle . in this case
the @xmath341 integral drops out , and the integral can be evaluated as an iterated integral without the gaussian factor @xmath367 .
we also have an alternative formula for the morse index densities , which follows by a similar argument ( given in
[ iq - proof ] ) : [ iq ] under the above assumptions , the density of the expected distribution of critical points of morse index q of @xmath268 is also given by : @xmath368 } \ , dg\ , d \xi\ , d\lambda,\ ] ] where @xmath369 we write @xmath370 here , @xmath371 ( previously denoted by @xmath372 ) is a complex @xmath81 symmetric matrix , so @xmath373 .
the proof is basically to rewrite using the itzykson - zuber integral and gaussian integration .
we first observe that @xmath374 where @xmath375 is the absolutely convergent integral , @xmath376 here , @xmath377 is the space of @xmath81 hermitian matrices .
absolute convergence is guaranteed by the gaussian factors in each variable @xmath378 .
if the @xmath379 integral is done first , we obtain a dual gaussian which converges ( in the sense of tempered distributions ) to the delta function @xmath380 as @xmath381 .
then , as @xmath382 , the @xmath383 integral then evaluates the integrand at @xmath384 and we retrieve the original integral @xmath385 .
we next conjugate @xmath386 in ( [ ical1a ] ) to a diagonal matrix @xmath387 with @xmath388 by an element @xmath389 . recalling that @xmath390 ( see for example ( * ? ? ? * ( 1.9 ) ) ) , we then obtain @xmath391 since the factor @xmath392 is invariant under the conjugation @xmath393 with @xmath394 , we apply the same identity in the @xmath395 variable . we write @xmath396 where @xmath397 is diagonal
this replaces @xmath379 by @xmath398 .
the inner product is bi - invariant so we may transfer the conjugation to @xmath399 .
we thus obtain : @xmath400 \nonumber \\ & & \times\ \delta(\lambda)^2 \delta(\xi)^2\,d\xi\ , d{\lambda}\ , dh\ , dx \ , dg \,dh\;.\label{ical2}\end{aligned}\ ] ] next we recognize the integral @xmath401 as the well - known itzykson - zuber - harish - chandra integral @xcite ( cf . ,
@xcite ) : @xmath402_{j , k}}{\delta(\lambda ) \delta(\xi)}\;.\ ] ] we note that both numerator and denominator are anti - symmetric in @xmath403 and @xmath404 under permutation , so that the ratio is well - defined
. we substitute into and expand @xmath405_{jk } = \sum_{\sigma \in s_m } ( -1)^\sigma\ ; e^{i\langle \xi , \sigma(\lambda)\rangle } , \ ] ] obtaining a sum of @xmath406 integrals .
however , by making the change of variables @xmath407 and noting that @xmath408 , we see that these integrals are equal , and then becomes @xmath409 further we observe that the @xmath410 integral is a gaussian integral .
we simplify the phase by noting that @xmath411 where @xmath342 and @xmath412 are as in the statement of the theorem .
thus , @xmath413 \,dh\ , dx \hspace{-3in}\nonumber\\ & = & \frac { 1}{\det{\lambda}\,\det\left[i \rho(g)^ * { \widehat}d(\xi ) \rho(g)+{\lambda}{^{-1}}\right]}\nonumber\\ & = & \frac { 1}{\det\left[i \rho(g)^ * { \widehat}d(\xi ) \rho(g){\lambda}+i\right]}\nonumber\\ & = & \frac { 1}{\det\left[i { \widehat}d(\xi ) \rho(g){\lambda}\rho(g)^ * + i\right]}\;. \label{hxint}\end{aligned}\ ] ] substituting into , we obtain the desired formula . the proof is similar to that of theorem [ i ] above , with the modifications provided by theorem [ morse ] .
the proof of lemma [ izint ] shows that @xmath414 } \ , dg\ ,
d \xi\ , d\lambda,\end{aligned}\ ] ] where @xmath415 denotes the set of points in @xmath416 with exactly @xmath417 coordinates positive .
since the integrand on the right is invariant under identical simultaneous permutations of the @xmath403 and the @xmath404 , it follows that the integral equals @xmath406 times the corresponding integral over @xmath418 .
the desired formula then follows from theorem [ morse ] .
to illustrate our results for fixed @xmath8 , we compute the density @xmath420 of the expected distribution of critical points of morse index @xmath84 of @xmath88 for random sections @xmath421 , where @xmath62 is the fubini - study metric on @xmath422 . here ,
the probability measure on @xmath109 is the gaussian measure induced from @xmath62 and the volume form @xmath423 on @xmath419 .
since this hermitian metric and gaussian measure are invariant under the @xmath424 action on @xmath419 , the density is independent of the point @xmath425 , and hence the expected number of critical points of morse index @xmath84 is given by @xmath426 these numbers turn out to be rational functions of @xmath8 given by the following integral formula : [ numcrit ] the expected number of critical points of morse index @xmath84 for random sections @xmath421 is given by @xmath427 \displaystyle 1 & \mbox{for } \ q = m\end{array}\right\},\end{aligned}\ ] ] for @xmath428 , where @xmath429 is as in theorem [ iq ] .
thus , the expected number @xmath430 of critical points of minimum morse index is of the form @xmath431 , for some constant @xmath432 .
it is also easy to see from the form of the above integral that the expected number @xmath433 of critical points of each morse index @xmath84 is a rational function of @xmath8 . in appendix 1
, we give explicit formulas for @xmath433 in low dimensions obtained by using maple to evaluate the integral in proposition [ numcrit ] .
the following formula derived from theorem [ iq ] is the starting point for our proof of proposition [ numcrit ] .
[ kcritpm ] the expected critical point density for random sections @xmath421 of morse index @xmath84 is given by @xmath434 where @xmath435 and @xmath429 are as in theorems [ i ] and [ iq ] .
since the critical point density @xmath436 is constant , it suffices to compute it at @xmath437 , using the local frame @xmath50 corresponding to the homogeneous ( linear ) polynomial @xmath260 .
we recall that the szeg kernel is given by @xmath438 ( see , for example , @xcite . ) since the formula in theorem [ kncrit2 ] is invariant when the szeg kernel is multiplied by a constant , we can replace the above by the _ normalized szeg kernel _ @xmath439 in our computation .
we notice that @xmath440 @xmath441 hence when computing the ( normalized ) matrices @xmath442 , we can take the usual derivatives of @xmath443 . indeed , we have @xmath444 it follows that @xmath445 where @xmath446 is the identity matrix of rank @xmath447 .
the stated formula now follows from theorem [ iq ] by observing that @xmath448 , and @xmath449=(- i)^{\frac{m^2+m+2}2 } \left(n^2\sum\xi_j+i\right)\prod_{1\le j
\le k\le m } \{i - n(n-1)(\xi_j+\xi_k)\}\;.\ ] ] to complete the proof of proposition [ numcrit ] , we must evaluate the @xmath450 integral in lemma [ kcritpm ] .
we begin by writing @xmath451 to simplify the constant factors , we make the redefinitions @xmath452 and @xmath453 , after which holds with @xmath454 where @xmath455 recalling the definition of @xmath435 in the statement of theorem [ i ] , we therefore have : @xmath456 where @xmath457 is given by . proposition [ numcrit ] now follows from , , and the following lemma with @xmath458 , @xmath459 .
[ xiint ] let @xmath460 and let @xmath461 .
then for @xmath462 we have @xmath463 \displaystyle i^{m^2 - 1}\,\frac{\pi^m}{c } & \mbox{for } \
p = m \end{array}\right.\ , \ ] ] where @xmath464 is given by .
we note that @xmath464 , resp .
@xmath465 , is positive if @xmath12 is odd , resp .
even , and thus by monotone convergence we can set @xmath466 in .
we let @xmath467 we note that @xmath468 is a tempered distribution ( in @xmath469 ) . furthermore
, the map @xmath470 is a continuous map from @xmath471 to the tempered distributions . here , @xmath472 is the change in the contour @xmath416 obtained by translating @xmath473 in each factor .
hence @xmath474 we now use to evaluate @xmath475 by iterated residues .
we first suppose that @xmath476 , and we start by doing the integral over @xmath477 . since the @xmath477 integral is absolutely convergent when @xmath478 , we can set @xmath478 and do the integral by residues . if @xmath476 we close the contour in the upper half plane , and pick up poles at @xmath479 , and at @xmath480 for @xmath481 .
the pole at @xmath482 is below the contour . the residue of @xmath483 at the pole @xmath479 is @xmath484 the residue at the pole @xmath485 is @xmath486 } 2t_2(t_2+t_3)\cdots(t_2+t_m)\delta(t_2,\dots , t_m ) } { ( t_3+\cdots + t_m+ci)\,2t_2(-t_2+t_3)\cdots(-t_2+t_m)\prod_{2\le j \le k\le m } ( t_j+t_k)}\\ & & = \ \frac{\pm e^{i(\lambda_2-\lambda_1)t_2}\,e^{-{\varepsilon}_2|t_2|^2}}{2t_2}\ , { \mathcal{i}}({\lambda}_3,\dots,{\lambda}_m , t_3,\dots , t_m;c)\;.\end{aligned}\ ] ] when we then do the @xmath487 integral and
let @xmath488 , we get zero .
indeed , @xmath489 since @xmath490 and the pole at @xmath491 is above the contour .
similarly , when we compute the residue of the pole @xmath480 , @xmath492 , and then perform the @xmath493 integration , we also get zero .
hence we can ignore the residues of the poles @xmath480 .
applying recursively , the integral with @xmath476 can be reduced to the case with all @xmath469 s negative : @xmath494 we now treat the case @xmath495 ( i.e. , @xmath496 ) .
this time , we do the @xmath497 contour integral first
. we close it in the lower half plane , picking up the residue at @xmath498 .
this residue is @xmath499 ( to simplify the discussion , we set @xmath500 , and regard the integrals as distributions , as above . ) next we perform the @xmath477 integration . since @xmath501 is the most negative eigenvalue , we close the contour in the upper half plane .
the terms in the denominator with @xmath502 all give poles in the lower half plane , so can be ignored . and , the poles @xmath480 will be ignorable , by the same type of reasoning we saw earlier .
indeed , after computing the residue at @xmath480 we find that @xmath493 appears in the exponent as @xmath503 with @xmath504 and the only factor of the denominator with a zero below the contour is @xmath505 ; but this factor also appears in the numerator and hence the @xmath493 integral gives zero .
this leaves the residue at all @xmath506 with @xmath507 .
the residue at @xmath479 of ( [ res2 ] ) is @xmath508 continuing recursively , for the case @xmath495 , we obtain ( remembering that the @xmath497 pole below the contour contributes negatively ) : @xmath509 combining ( with @xmath12 replaced by @xmath510 ) and , we obtain the formula of the proposition . in this article
, we are studying the @xmath8-dependence of the critical point density and number of critical points , but the growth rate of @xmath511 ( and the growth rate of the total number of critical points ) as the dimension @xmath512 is of considerable interest as well . in the vacuum statistics problem in string / m theory ,
one is interested in the total number of critical points of certain holomorphic sections ( known as flux superpotentials ) of a line bundle @xmath513 over a moduli space @xmath514 of complex structures on a calabi - yau fourfold @xmath515 where @xmath277 is a calabi - yau threefold and @xmath516 is an elliptic curve .
the relevant dimension is @xmath517 where @xmath518 is the third betti number of @xmath277 .
as discussed in some detail in 7.3 of @xcite , vacuum statistics involves integrals like ( [ b0 ] ) over certain subspaces of @xmath519 .
the growth rate in @xmath518 of the number of vacua is important to obtain an order of magnitude of string vacua , since @xmath518 is rather large for typical calabi - yau threefolds . in @xcite , we state a conjectural formula for the growth in @xmath518 of such integrals which is based on the explicit calculations in this article for @xmath179 for fixed @xmath8 .
we recall that by proposition [ numcrit ] , the expected number of critical points on @xmath419 of minimum morse index is of the form @xmath431 .
our maple computations given in appendix 1 show that @xmath520 , and furthermore that the leading terms of the expansion are monotonically decreasing in the morse index @xmath84 , for each @xmath106 ; in particular , on average there are more critical points on @xmath419 of morse index @xmath12 than there are of any index @xmath521 ( at least for @xmath106 ) . recall that by our universality results ( theorems [ univdist][univmorse ] and corollary [ en ] ) , the leading coefficient in the @xmath8-expansion of @xmath522 is the universal coefficient @xmath523 .
thus we make the following conjecture based on our maple computations : let @xmath524 denote the leading coefficient in the expansion of @xmath89 from corollary [ en ] , and let @xmath525 : @xmath526 then @xmath527 and hence the expected total number of critical points @xmath528 equivalently , @xmath529 , where @xmath530 is the number of simultaneous zeros of @xmath12 generic sections of @xmath58 .
this conjecture implies that the expected number of critical points of morse index @xmath12 grows exponentially in the dimension , a growth rate consistent with quite analogous estimates of vacua in string / m theory @xcite and of metastable states of spin glasses @xcite .
in this section , we compute the asymptotics of the expected density and number of critical points of sections of powers @xmath58 of a positive holomorphic line bundle . in particular , we prove theorems [ univdist ] , [ critrs ] , and [ univmorse ] as well as corollary [ en ] . we begin with some further background on the szeg kernel .
we first use the asymptotic expansion of the szeg kernel to show that @xmath531 has an expansion of the type given in theorems [ univdist][univmorse ] .
it is evident from theorem [ kncrit2 ] and theorem [ morse ] , respectively , and from formulas ( [ lambdan])([cn ] ) for @xmath532 and @xmath533 that the asymptotics of the critical - point densities @xmath534 and @xmath535 , respectively , can be determined by canonical algebraic operations on the asymptotics of the following derivatives of the szeg kernel ( @xmath536 ) * @xmath537 ; * @xmath538 * @xmath539 and @xmath540 * @xmath541 ; * @xmath542 ( here we write @xmath543 . ) we can obtain their asymptotics by differentiating the following tian - yau - zelditch expansion : @xcite [ tyz ] let @xmath544 be a positive hermitian holomorphic line bundle over a compact complex manifold @xmath3 of dimension @xmath12 with khler form @xmath545 . then there is a complete asymptotic expansion : @xmath546\,,\ ] ] for certain smooth coefficients @xmath547 . to apply to the differentiated szeg kernel
, we use . by a change of frame in @xmath23 , we can assume that @xmath305 and its holomorphic derivatives up to any fixed order , as well as the anti - holomorphic derivatives , vanish at @xmath260
. writing @xmath548 , we then have : @xmath549\right|_{z_0}\,,\nonumber\\ & & \nabla_{z_j}\nabla_{\bar w_{j'}}\pi_n(z_0,z_0 ) = \frac{{\partial}^2f_n}{{\partial}z_j{\partial}\bar w_{j'}}(z_0,\bar z_0 ) = { \partial}_j\bar{\partial}_{j'}f_n(z,\bar z)|_{z_0 } = { \partial}_j\bar{\partial}_{j ' } \left.\left[e^{nk(z)}\pi_n(z , z)\right ] \right|_{z_0}\,,\nonumber\\[-5pt ] & & \qquad\vdots\label{derivatives}\\ & & \nabla_{z_j}\nabla_{z_q}\nabla_{\bar w_{j'}}\nabla_{\bar w_{q'}}\pi_n(z_0,z_0 ) = { \partial}_j{\partial}_q\bar{\partial}_{j ' } \bar{\partial}_{q ' } \left.\left[e^{nk(z)}\pi_n(z , z)\right ] \right|_{z_0}\,.\nonumber\end{aligned}\ ] ] here , we used the fact that @xmath550 is holomorphic in @xmath37 and anti - holomorphic in @xmath551 .
( in these expressions , we have no @xmath552 or @xmath553 derivatives of @xmath554 . )
it follows by substituting into that the components of @xmath555 and @xmath556 have asymptotic expansions in powers of @xmath8 , and hence by theorem [ kncrit2 ] , resp .
theorem [ morse ] , that @xmath531 , resp.@xmath420 , does .
next we study the coefficients @xmath557 of the expansion of @xmath531 . integrating the density of critical points , we find that the expected total number of critical points has the expansion @xmath558 the leading order term is universal
. we will use theorem [ kncrit2 ] and the following result of z. lu @xcite to calculate the coefficients in these expansions : [ zlu ] @xcite with the notation as in theorem [ tyz ] , each coefficient @xmath547 is a polynomial of the curvature and its covariant derivatives at @xmath559 . in particular , @xmath560 where @xmath561 and @xmath14 denotes the curvature tensor , the ricci curvature and the scalar curvature of @xmath70 , respectively , and @xmath562 denotes the laplace operator of @xmath563 .
we now calculate @xmath532 and @xmath533 to two orders .
the key point is to calculate the mixed derivatives of @xmath256 on the diagonal .
it is convenient to do the calculation in khler normal coordinates about a point @xmath260 in @xmath3 .
it is well known that in terms of khler normal coordinates @xmath564 , the khler potential @xmath305 has the expansion : @xmath565 ( in general , @xmath305 contains a pluriharmonic term @xmath566 , but a change of frame for @xmath23 eliminates that term up to fourth order . ) we further use the notation @xmath567 .
we first claim that @xmath568 indeed , by , @xmath569 & = & e^{nk } \left[nk_j(1+a_1n{^{-1}}+a_2n^{-2 } ) + { \partial}_ja_1n{^{-1}}+{\partial}_ja_2n^{-2}+\cdots\right]\\ { \partial}_j{\bar\partial}_{j'}\left[e^{nk(z)}\pi_n(z , z)\right ] & = & e^{nk}\big[n^2k_jk_{\bar j'}(1+a_1n{^{-1}}+a_2n^{-2 } ) + k_{\bar j'}{\partial}_ja_1 + k_{\bar j'}{\partial}_ja_2n^{-1}\\ & & + nk_{j\bar j'}(1+a_1n{^{-1}}+a_2n^{-2 } ) + k_j{\bar\partial}_{j'}a_1 + k_{j}{\bar\partial}_{j'}a_2n^{-1 } \\&&+ { \partial}_j{\bar\partial}_{j ' } a_1 n{^{-1}}\cdots\big].\end{aligned}\ ] ] evaluating at @xmath260 using , we then obtain .
we now compute the expansion of @xmath556 . continuing
the above computation , @xmath570 & = & e^{nk}\big[n^2(k_{j\bar q'}k_{\bar j ' } + k_{j\bar j'}k_{\bar q'})(1+a_1n{^{-1}}+a_2n^{-2})\\ & & + k_{\bar j'}{\bar\partial}_{q'}a_1 + k_{\bar q'}{\partial}_j{\bar\partial}_{j'}a_1 + k_{j\bar j'}{\bar\partial}_{q'}a_1 + k_{j\bar q'}{\bar\partial}_{j'}a_1 \\ & & + nk_{j\bar j ' \bar q'}(1+a_1n{^{-1}})\cdots\big ] + \mbox{unimportant terms}.\end{aligned}\ ] ] ( the ` unimportant terms ' are those which vanish at @xmath260 and whose holomorphic derivatives also vanish at @xmath260 . ) we have @xmath571\nonumber\\ & = & \big [ \big({\delta}_{jj'}{\bar\partial}_{q'}a_1 + { \delta}_{jq'}{\bar\partial}_{j'}a_1\big)\quad \big({\delta}_{j}a_1\big ) \big ] \ +
\ o(n{^{-1}})\;.\label{bexp}\end{aligned}\ ] ] differentiating again and evaluating at @xmath260 using , we obtain @xmath572\big|_{z_0 } & = & \big[n^2({\delta}_{jj'}{\delta}_{qq'}+{\delta}_{jq'}{\delta}_{qj'})(1+a_1n{^{-1}}+a_2n^{-2})\\ & & + { \delta}_{jj'}{\partial}_q{\bar\partial}_{q'}a_1+{\delta}_{qq'}{\partial}_j{\bar\partial}_{j'}a_1 + { \delta}_{jq'}{\partial}_q{\bar\partial}_{j'}a_1 + { \delta}_{qj'}{\partial}_q{\bar\partial}_{j'}a_1 \\ & & + nk_{j\bar j ' q \bar q'}(1+a_1n{^{-1}})\cdots\big]\big|_{z_0 } \;.\end{aligned}\ ] ] noting that @xmath573 , and recalling that @xmath574 , where @xmath575 \big(n\nabla_{\bar w_{q'}}\nabla_{\bar w_{j'}}\pi_n \big ) & n^2\pi_n \end{array}\right]\,,\ ] ] we obtain @xmath576 with @xmath577 @xmath578 @xmath579 where @xmath446 is the identity operator on @xmath80 , and @xmath580 to prove theorem [ univdist ] , we want the asymptotics of @xmath581 making the change of variables @xmath582 , the integral is transformed to @xmath583 where @xmath584 next we observe that @xmath585 + \cdots\ ] ] hence @xmath586 h , h \right\rangle } \\ & & \\ & = & e^{- \langle h , h \rangle } \{1 + \frac{1}{n } \langle \lambda_{-1 } h , h \rangle \\ \\ & + & \frac{1}{n^2 } [ \langle \lambda_{-2 } h , h \rangle + \frac{1}{2 } \langle \lambda_{-1 } h , h \rangle^2 - \langle \lambda^{2}_{-1 } h , h \rangle ] \}.
\end{array}\ ] ] furthermore @xmath587n^{-2 } \cdots\,,\ ] ] and similarly for @xmath588 .
altogether , we obtain : @xmath589 \right . \\ \\ & + & \frac{1}{n^2 } \left[{{\frac{1}{2}}}{{\operatorname{tr}}}({\lambda}_{-1}^2)-\left . { { \operatorname{tr}}}{\lambda}_{-2 } + { { \frac{1}{2}}}{{\operatorname{tr}}}(a_{-1}^2)-{{\operatorname{tr}}}a_{-2}+{{\frac{1}{2}}}({{\operatorname{tr}}}a_{-1}+{{\operatorname{tr}}}{\lambda}_{-1})^2 \right]\right\ } \\ & & \\ & \times & \int_{{{\operatorname{sym}}}(m,{{\mathbb c } } ) \times { { \mathbb c } } } \left|\det(2h'h'{}^*-|x|^2i)\right|e^{- \langle h , h \rangle } \{1 + \frac{1}{n } \langle \lambda_{-1 } h , h \rangle \\ \\ & + & \frac{1}{n^2 } [ \langle \lambda_{-2 } h , h \rangle + \frac{1}{2 } \langle \lambda_{-1 } h , h \rangle^2 - \langle \lambda^{2}_{-1 } h , h \rangle ] \}\,dh'\,dx\ , .
\end{array}\ ] ] expanding , we obtain @xmath590 b_0&=&\int d\mu\,,\\[6pt ] b_1 & = & \int\left[\langle \lambda_{-1 } h , h \rangle- { { \operatorname{tr}}}a_{-1 } - { { \operatorname{tr}}}\lambda_{-1}\right ] d\mu\ , , \\[6pt ] b_2 & = & \int\left [ { { \frac{1}{2}}}{{\operatorname{tr}}}({\lambda}_{-1}^2)- { { \operatorname{tr}}}{\lambda}_{-2 } + { { \frac{1}{2}}}{{\operatorname{tr}}}(a_{-1}^2)-{{\operatorname{tr}}}a_{-2}+{{\frac{1}{2}}}({{\operatorname{tr}}}a_{-1}+{{\operatorname{tr}}}{\lambda}_{-1})^2 \right.\\[6pt ] & & \left . -({{\operatorname{tr}}}a_{-1 } + { { \operatorname{tr}}}\lambda_{-1 } ) \langle \lambda_{-1 } h , h \rangle
+ \langle ( \lambda_{-2}-\lambda^{2}_{-1 } ) h , h \rangle + \frac{1}{2 } \langle \lambda_{-1 } h , h \rangle^2 \right]d\mu\ , , \end{array}\ ] ] where @xmath591 recalling and , we see that @xmath592 is of the form @xmath593 where @xmath594 is universal .
since @xmath592 is also invariant under the unitary group , we must have @xmath595 where @xmath78 is a universal constant ( depending only on the dimension @xmath12 of @xmath3 ) .
similarly , @xmath596 is of the form @xmath597 where @xmath598 is a universal quadratic form in the curvature tensor @xmath599 .
but @xmath596 is also @xmath341-invariant and hence is a curvature invariant ( of order 2 ) .
thus , latexmath:[\[\label{kappa - two}b_2 = { \gamma}_0 \delta\rho+{\gamma}_1\rho^2 + { \gamma}_2|r|^2+{\gamma}_3 depending only on @xmath12 , which completes the proof of theorem [ univdist ] .
the proof of theorem [ univmorse ] is exactly as above , except we integrate over @xmath601 instead of @xmath602 in the computation of the expansion of @xmath436 .
the coefficient @xmath603 is of the form @xmath604 by the above argument ( or by the fact that @xmath14 is the only curvature invariant of order 1 ) .
furthermore , it is well known ( see , e.g. , @xcite ) that for any khler metric @xmath605 on @xmath3 , we have @xmath606 \wedge { \omega}^{m-2}\;,\ ] ] where @xmath607
. therefore @xmath608 where we now write @xmath609 for the @xmath73-th chern form of the khler metric @xmath610 . integrating and , noting that @xmath611
, we obtain the asymptotic expansion of corollary [ en ] with @xmath612 . on a compact riemann surface @xmath99 of genus @xmath100 , corollary [ en ] says that @xmath89 has a universal expansion of the form @xmath613 for @xmath614 .
there are several ways to compute the constants .
a quick way to find @xmath615 , is to consider the case of @xmath616 with the fubini - study metric on @xmath617 . by an elementary computation in @xcite ( or by [ s - exact ] ) , we showed that for this case @xmath618 therefore , @xmath619 to find @xmath620 , we note that @xmath621 , where @xmath622 is the fubini - study khler form on @xmath616 . furthermore , since @xmath623 , we have @xmath624 .
( this can be checked directly as follows : the khler potential @xmath625 , where @xmath37 is the affine coordinate , and hence by , @xmath626 . )
hence , @xmath627 , and therefore @xmath628 which completes the proof of theorem [ critrs ] .
we have already shown that * @xmath629 is topological ; * @xmath630 is the sum of a topological term plus a positive multiple of @xmath631 . to complete the proof of theorem [ asympmin ] and
show that the metric with asymptotically minimal @xmath89 ( and @xmath9 ) is the one for which @xmath70 has minimal @xmath286 norm of the scalar curvature , we must show that the @xmath134 are positive . the proof consists of finding an integral formula for @xmath134 and then transforming it to one that is amenable to ( computer ) evaluation .
we first summarize the key results : [ k0 ] in all dimensions , @xmath632 where @xmath633 and @xmath634 is given by .
after a sequence of manipulations as in the proof of lemma [ izint ] , the integral ( [ i0 ] ) will be rewritten in the following form : [ li4 ] @xmath635 where @xmath636\frac 1{\left(1 - \frac i2 \sum_j\xi_j\right)}+ \frac 2{(m+1)(m+3)\left(1 - \frac i2 \sum_j\xi_j\right)^2 } } { \left(1 - \frac i2 \sum_j\xi_j\right)\prod_{j\le k}\left [ 1 + \frac i2 ( \xi_j+\xi_k)\right ] } \;.\ ] ] here , @xmath387 is the diagonal matrix with diagonal entries @xmath388 , @xmath637 is the vandermonde determinant and @xmath638for ( hermitian ) @xmath639 matrices @xmath386 .
the iterated @xmath640 integrals are defined in the distribution sense .
the final step is the evaluation of @xmath176 .
having simplified the integral as far as we could , we complete the computation for the cases @xmath641 using maple and find that it is positive for these cases .
the resulting values of the constants @xmath169 , @xmath641 , are given in
[ s - beta2 ] .
we use the case of @xmath181 where @xmath182 is an elliptic curve , and @xmath23 is the product of degree 1 line bundles on the factors ( with the fubini - study metric on @xmath642 and the flat metric on @xmath182 ) . ( the manifold @xmath3 is a homogeneous space with respect to @xmath643 , so the critical point density is invariant and hence constant . ) since @xmath644
, it follows from that the coefficient @xmath645 of the expansion @xmath646 is given by @xmath647 , and hence @xmath648 the szeg kernel for @xmath649 is the product of the szeg kernels on @xmath616 and @xmath650 . since the universal cover of @xmath650 is @xmath651 , the szeg kernel on @xmath650 is given by the heisenberg szeg kernel on @xmath651 ( see @xcite ) modulo an @xmath652 term , and we have : @xmath653 as in [ s - exact ] , we consider the normalized szeg kernel @xmath654 we have : @xmath655 it suffices to compute the density at @xmath656 . from the above , we have : @xmath657 recalling , we then have : @xmath658 \ = \ 0\,,\label{bn1 } \\[8pt ] { \widetilde}c_n(0)&= & \left [ \begin{array}{cc}\displaystyle \left ( \tau_{jq}\tau_{j'q'}\,\frac{\partial^4 { \widetilde}\pi_n } { \partial z_q \partial z_j \partial \bar{w}_{q'}\partial \bar{w}_{j ' } } ( 0,0 ) \right ) & \displaystyle \left ( \tau_{jq}n\,\frac{\partial^2 { \widetilde}\pi_n } { \partial z_j \partial z_q}(0,0 ) \right ) \\[8pt ] \displaystyle\left(\tau_{j'q'}n\ , \frac{\partial^2 { \widetilde}\pi_n}{\partial \bar{w}_{q'}\partial \bar{w}_{j ' } } ( 0,0 ) \right ) & n^2\,{\widetilde}\pi_n(0,0 ) \end{array } \right ] \,,\label{cn1}\\[8pt ] & & \qquad\qquad \qquad 1\le j\le m\ , , 1\le j\le q\le m\ , , 1\le j'\le q'\le m\,.\nonumber\end{aligned}\ ] ] it follows that @xmath659 i.e. the diagonal matrix with diagonal entries @xmath660 , @xmath661 repeated @xmath662 times , @xmath663 .
we want to compute @xmath664 making the change of variables @xmath665 , the integral is transformed to @xmath666 where @xmath667 therefore @xmath668d{\widetilde}\mu + o\left ( \frac1{n^3}\right)\,,\end{aligned}\ ] ] where @xmath669 therefore @xmath670 and the desired formula then follows from . as an intermediate step between lemmas [ k0 ] and [ li4 ] ,
we prove : [ ical2 ] @xmath671 where @xmath672 for ( hermitian ) @xmath639 matrices @xmath386 . since the change of variables @xmath673 ( @xmath674 )
is unitary on @xmath80 ( with respect to the hilbert - schmidt inner product ) , we can make this change of variables in , and then integrate over @xmath674 to obtain @xmath675 we now evaluate the integral @xmath676 [ g24 ] for @xmath677 , @xmath678 to prove the claim , we write @xmath679 so that @xmath680 , and we replace @xmath681 with @xmath682 , where @xmath683 is haar probability measure on @xmath684 . next we recall that if @xmath417 is a homogeneous polynomial of degree @xmath685 on @xmath686 , @xmath687 we easily see using wick s formula that @xmath688 .
although the above approach can also be used to verify , we find it easier to use invariant theory , since the integral in is a @xmath341-invariant function of @xmath689 , under the @xmath341 action @xmath673 . indeed , it is a @xmath341-invariant hermitian inner product on @xmath690 .
the action of @xmath341 on symmetric complex matrices defines a representation equivalent to @xmath691 where @xmath692 is the defining representation of @xmath693 .
it is well known from schur - weyl duality that @xmath694 is irreducible .
we then consider the @xmath341 representation @xmath695 with the diagonal action .
henceforth we put @xmath696.\ ] ] we then regard @xmath697 as the value on @xmath698 of the quadratic form @xmath699 this defines the hermitian inner product @xmath700 we next recall that @xmath701 decomposes into a direct sum of two @xmath341 irreducibles , one corresponding to the young diagram @xmath702 with @xmath34 row of four boxes and one corresponding to the diagram @xmath703 with @xmath208 rows each with two boxes .
see for instance proposition 1 of @xcite .
the young projectors are respectively , @xmath704 for @xmath703 the @xmath705 permutes @xmath706 since an irreducible @xmath341 representation has ( up to scalar multiples ) a unique invariant inner product , it follows that @xmath707 where @xmath708 are the invariant inner products @xmath709 for the irreducibles corresponding to the young diagrams @xmath710 as above .
we now calculate these inner products on @xmath698 .
we have @xmath711 it is easy to see that each of these expressions is a linear combination of the two quadratic forms @xmath712 \circ [ h^ * \otimes h^*]\},\;\;\ ; h\otimes h \mapsto [ tr h \circ h^*]^2.\ ] ] hence @xmath713 to determine the constants @xmath714 , it suffices to consider the case where @xmath371 is diagonal .
let @xmath715 denote the eigenvalues of @xmath371 .
then by wick s formula we obtain @xmath716 formula now follows from with @xmath717 .
having proved the claim , the formula stated in lemma [ ical2 ] now follows from and lemma [ g24 ] .
we proceed exactly as in the proof of lemma [ izint ] .
we rewrite the integral as @xmath718 where @xmath719 recall that @xmath720 . as in
[ simplify ] , we note that absolute convergence is guaranteed by the gaussian factors in each variable @xmath721 . evaluating @xmath722 first
, we obtain a dual gaussian , which approximates the delta function @xmath380 . as @xmath381 , the @xmath383 integral then yields the integrand at @xmath723 ; then letting @xmath724 we obtain the original integral stated in lemma [ ical2 ] .
continuing as in [ simplify ] , we conjugate @xmath386 to a diagonal matrix @xmath387 with @xmath388 by an element @xmath725 and we replace @xmath726 with @xmath727 . recalling , we obtain : @xmath728}\ , dh \,dx \,d{\lambda}\,d \xi\,dh\;.\end{aligned}\ ] ] again using applied this time to @xmath395 , we obtain : @xmath729 we then transfer the conjugation by @xmath100 to the right side of the @xmath730 in the first exponent and make the change of variables @xmath731 to eliminate @xmath100 from the integrand : @xmath732 next we substitute the itzykson - zuber - harish - chandra integral formula into the above and expand @xmath405_{jk } = \sum_{\sigma \in s_m } ( -1)^\sigma\ ; e^{i\langle \xi , \sigma(\lambda)\rangle } , \ ] ] obtaining a sum of @xmath406 integrals .
however , by making the change of variables @xmath407 and noting that @xmath733 , we see as before that the integrals of these terms are equal , and so we obtain @xmath734 where @xmath735 the phase @xmath736\nonumber \\&= & -\left[\sum_{j\le k}\left ( 1 + \frac i2 ( \xi_j+\xi_k)\right)|{\widehat}h_{jk}|^2 + \left(1 - \frac i2 \sum_j\xi_j\right)|x|^2\right]\;. \label{phase}\end{aligned}\ ] ] thus , @xmath737 where , @xmath738 to evaluate the @xmath229 integral , we first expand the amplitude : @xmath739|x|^2\\&&+ \frac 1{(m+1)(m+3 ) } |x|^4 \;,\end{aligned}\ ] ] and then integrate to obtain . to evaluate @xmath740 ,
we first observe as in [ s - exact ] that the map @xmath741 is a continuous map from @xmath471 to the tempered distributions .
hence by and , we have : @xmath742 letting @xmath743 sequentially , we obtain the formula of lemma [ li4 ] .
we use the integral formula of lemma [ li4 ] to compute the constants @xmath169 .
the @xmath403 integrals can be evaluated using residues as in [ residues ] ; the resulting @xmath469 integrand is a polynomial function of the @xmath744 and @xmath745 .
the integrals were evaluated in dimensions @xmath110 using maple 10 . in dimension 1 ,
we reproduce the result from @xcite : @xmath746 in dimension 2 , we have : @xmath747 in dimension 3 , we have : @xmath748 in dimension 4 , we have : @xmath749 \displaystyle { \beta}_{27}(4)=\frac{2^{15}\cdot 937}{3 ^ 8\cdot 5 ^ 6\cdot 7\cdot\pi^4 } \;,\quad { \beta}_{28}(4)=\frac{267828299}{2 ^ 6\cdot 3 ^ 8\cdot 5 ^ 6\cdot 7\cdot\pi^4 } \;.\end{array}\ ] ] this completes the proof of theorem [ asympmin ] for @xmath641 .
b. baugher @xcite used mathematica to compute the constants @xmath750 , and found that they are positive as well .
the values of the coefficient @xmath11 for the expected total number of critical points are : @xmath751\displaystyle { \beta}_2(3)= \frac{104}{729\pi^3 } = \frac{2 ^ 3\cdot 13}{3 ^ 6\cdot\pi^3}\;,\quad { \beta}_2(4)=\frac { 17152}{45927\pi^4}= \frac{2 ^ 8\cdot 67}{3 ^ 8\cdot 7\cdot\pi^4}\;.\end{array}\ ] ]
we give in this appendix the precise formulas for the expected numbers @xmath752 of critical points of morse index @xmath84 of random sections of @xmath753 for @xmath641 , as well as the leading coefficients in dimensions @xmath754 .
these formulas were obtained by evaluating the integral in proposition [ numcrit ] using maple 10 . for @xmath755 , we reproduce the result from @xcite : @xmath756 for @xmath757
, we obtain : @xmath758 hence , the expected total number of critical points is : @xmath759 to check the computation , we note that @xmath760 in the case @xmath761 , we obtain : @xmath762 { \mathcal n}^{{\operatorname { crit}}}_{n,5}({{{\mathbb c}}{{\mathbb p}}}^3)= { \frac { 256\ , \left ( n-1 \right ) ^{4}{n}^{5 } } { \left ( 5\,n-2 \right ) \left ( 3\,n-2 \right ) ^{5}}}\ ; , & { \mathcal n}^{{\operatorname { crit}}}_{n,6}({{{\mathbb c}}{{\mathbb p}}}^3)= { \frac { { n}^{9 } \left(451\,{n}^{4}-1248\,{n } ^{3 } + 1280\,{n}^{2 } -576\,n+96\right ) } { \left ( 2\,n-1 \right ) ^{4 } \left ( 3\,n-2 \right ) ^{5 } \left ( 5\,n-2 \right ) } } \;.\end{array}\ ] ] the expected total number of critical points is : @xmath763 to check the computation : @xmath764 for @xmath765 , we obtain : @xmath766 { \mathcal n}^{{\operatorname { crit}}}_{n,6}({{{\mathbb c}}{{\mathbb p}}}^4)= { \frac { \left ( n-1 \right ) ^{5}{n}^{5 } \left ( 396227\,{n}^{7}-1078546\,{n}^{6 } + 1261212\,{n}^{5 } -821326\,{n}^{4 } + 321695\,{n}^{3 } -75780\,{n}^{2}+9940\,n-560\right ) } { \left ( 5\,n-2 \right ) ^{5 } \left ( 2\,n-1 \right ) ^{7 } \left ( 3\,n-1 \right ) } } \;,\\[8pt ] { \mathcal n}^{{\operatorname { crit}}}_{n,7}({{{\mathbb c}}{{\mathbb p}}}^4)= { \frac { 4096\,\left ( n-1 \right ) ^{5}{n}^{9 } \left ( 109\,{n}^{2}-102 \,n+24 \right ) } { \left ( 5\,n-2 \right ) ^{5 } \left ( 3\,n-2 \right ) ^{7 } } } \;,\\[8pt ] { \mathcal n}^{{\operatorname { crit}}}_{n,8}({{{\mathbb c}}{{\mathbb p}}}^4)={\frac { { \alpha } } { \left ( 5\,n-2 \right ) ^{5 } \left ( 3\,n-2 \right ) ^{7 } \left ( 2\,n-1 \right ) ^{7 } \left ( 3\,n-1 \right ) } } \;,\\[8pt ] \scriptstyle \qquad { \alpha}= { n}^{14 } ( 14251551\,{n}^{10}-86984891\,{n}^{9}+237134546\,{n } ^{8 } - 380216704\,{n}^{7}+397067360\,{n}^{6}-282219280\,{n}^{5}+138269792\,{n}^{4}\\ \scriptstyle\qquad\qquad -46114432\,{n}^{3}+ 10020608\,{n}^{2}-1281280\,n+73216)\;. \end{array}\ ] ] the expected total number of critical points is : @xmath767 again , to check the computation : @xmath768 formulas for dimensions 5 and 6 were also obtained using maple , but they are too lengthy to be informative . instead , we give below a table of approximate numerical values of the leading coefficients @xmath769 of @xmath770 , for @xmath771 , for @xmath106 : @xmath772 & 0 & 1 & 2 & 3 & 4 & 5 & 6 & total + @xmath755 & 1.33333 & 0.33333 & & 1.66667 + @xmath757 & 1.5 & 0.59259 & 0.09259 & & 2.18519 + @xmath761 & 1.6 & 0.78750 & 0.21070 & 0.02320 & & 2.62140 + @xmath765 & 1.66667 & 0.93696 & 0.33019 & 0.06533 & 0.00543 & & 3.00457 + @xmath773 & 1.71429 & 1.05448 & 0.44235 & 0.11939 & 0.01844 & 0.00121 & & 3.35015 + @xmath774 & 1.75 & 1.14903 & 0.54457&0.17979 & 0.03884 & 0.00486 & 0.00026 & 3.66734 +
in this appendix , we describe baugher s conjectured identity and its implications for the positivity of @xmath176 . after calculating the integral of each of the individual terms in lemma [ li4 ] using mathematica 5
, baugher made the following conjecture @xcite : p. bleher , b. shiffman and s. zelditch , universality and scaling of zeros on symplectic manifolds .
_ random matrix models and their applications _ , edited by p. bleher and a.r .
its , msri publications 40 , cambridge univ . press , 2001 , pp . 31 - 70 .
e. calabi , extremal khler metrics .
_ seminar on differential geometry .
papers presented at seminars held during the academic year 19791980 _ , edited by s .-
annals of mathematics studies , 102 , princeton university press , princeton , n.j , 1982 , pp . 259290 .
yau , problem section .
_ seminar on differential geometry .
papers presented at seminars held during the academic year 19791980 _ , edited by s .-
annals of mathematics studies , 102 , princeton university press , princeton , n.j , 1982 , pp . 669706 .
yau , open problems in geometry .
_ differential geometry : partial differential equations on manifolds .
proceedings of the ams summer research institute on differential geometry ( los angeles , 1990 ) _ , edited by r. greene and s .- t .
proceedings of symposia in pure mathematics , 54 , part 1 .
american mathematical society , providence , ri , 1993 , pp . 128 . | motivated by the vacuum selection problem of string / m theory , we study a new geometric invariant of a positive hermitian line bundle @xmath0 over a compact khler manifold : the expected distribution of critical points of a gaussian random holomorphic section @xmath1 with respect to the chern connection @xmath2 . it is a measure on @xmath3 whose total mass is the average number @xmath4 of critical points of a random holomorphic section .
we are interested in the metric dependence of @xmath4 , especially metrics @xmath5 which minimize @xmath4 .
we concentrate on the asymptotic minimization problem for the sequence of tensor powers @xmath6 of the line bundle and their critical point densities @xmath7 .
we prove that @xmath7 has a complete asymptotic expansion in @xmath8 whose coefficients are curvature invariants of @xmath5 .
the first two terms in the expansion of @xmath9 are topological invariants of @xmath10 .
the third term is a topological invariant plus a constant @xmath11 ( depending only on the dimension @xmath12 of @xmath3 ) times the calabi functional @xmath13 , where @xmath14 is the scalar curvature of the khler metric @xmath15 .
we give an integral formula for @xmath11 and show , by a computer assisted calculation , that @xmath16 for @xmath17 , hence that @xmath9 is asymptotically minimized by the calabi extremal metric ( when one exists ) .
we conjecture that @xmath18 in all dimensions , i.e. the calabi extremal metric is always the asymptotic minimizer . |
the wannier functions ( wfs ) play a key role in the solid - state physics , as they make a direct connection between the reciprocal @xmath4-momentum space and the real space representations for the quantum - mechanical operators , which allows us to formulate the problem of electronic - structure calculations for the infinite periodical systems in a basis of localized ( wannier ) orbitals.@xcite the construction of wfs using first - principles electronic structure calculations has attracted a considerable attention recently.@xcite the reason is mainly twofold : * wfs appear to be extremely useful for the visualization and interpretation of the behavior of the local quantities in the real space .
for example , in the electronic structure calculations based on the plane waves , such a construction can add many new functionalities , which were initially a privilege of the methods working in the basis of localized atomic orbitals.@xcite * another important application of wfs is related with the construction of the so - called `` _ _ ab initio _
_ models for the strongly - correlated materials '' , which are designed to deal with the low - energy properties of these systems and go beyond the conventional ( and oversimplified in the case of the strongly - correlated systems ) local - density approximation ( lda ) . in this case , the wannier orbitals form the basis of the physical hilbert space , which is then used for the construction of the model ( typically , hubbard - type ) hamiltonians.@xcite the choice of the wannier function is not unique , and their extension in the real space can vary depending on the procedure and the parameters of calculations . in order to illustrate the importance of this problem
, one may remind to the reader that one extreme example of wfs are the bloch states , which are totally delocalized in the real space .
of course , the physics should not depend on whether the wfs are localized or not , because all representations provide the complete sets of the basis functions ( for the low - energy part of the spectrum ) , which are totally equivalent from the physical point of view .
however , from the viewpoint of practical applications , it is more convenient to work with the localized orbitals .
for example , in the process of construction of the hubbard hamiltonian one would always like to keep only the site - diagonal elements of coulomb and exchange interactions and neglect all intersite interactions .
this is only possible when the wannier basis is sufficiently well localized in the real space , so that the overlap between the functions centered at different lattice sites is minimal .
the localization of the basis functions is also important in applications dealing with the response of the electronic system onto a perturbation confined within a single atomic site , like the orbital magnetization in solids.@xcite since the idea of construction of `` _ _ ab initio _ _ models for the strongly - correlated systems '' is still relatively new , there is no clear consensus among researchers working in this area about the origin of apparent differences , accuracy , and the scopes of applicability of their results.@xcite we believe that in such a situation it is very important to make a direct comparison between different methods of calculation of wfs , and prove several theorems regarding both the formulation of the methods and the choice of the computational parameters , which directly control the localization of wfs .
particularly , we will prove rigorously that the downfolding method , considered in refs .
, , and , can be reformulated as a projector - operator method , employed in refs . , , and ( and , as an initial stage , in ref . ) .
this can be done by a simple ( scissor - operator - like ) change of the full hamiltonian of the system in the high - energy part of the spectrum , which does not contribute to the low - energy part .
this procedure has been already implemented in refs . , , and .
therefore , two methods are simply equivalent , and the main difference comes from the form of trial orbitals , which are used to generate the wannier function in the projector - operator method and , directly , the effective low - energy hamiltonian in the downfolding method . concerning the last part of the problem , we will argue that an optimal choice for the trial wavefunctions can be based on the maximization of the site - diagonal part of the density matrix , as it was originally proposed in ref . .
we will also consider an analogy with the search of the ground state of the many - electron systems .
we assume that there is certain set of orthonormalized atomic - like orbitals @xmath5 centered at the atomic sites @xmath6 and specified by the orbital indices @xmath7 .
the basis is complete and sufficient to reproduce the one - electron band structure of a considered compound in the valent part of the spectrum .
the corresponding hamiltonian matrix is denoted by @xmath0 .
the concrete examples of such bases could be the orthonormalized atomic orbitals or the muffin - tin orbitals.@xcite we further assume that the basis orbitals are maximally localized ( in accordance with a certain criterion of maximal localization , the precise form of which is not important here),@xcite in the sense that any linear combination of @xmath5 will be either less localized or have the same degree of the localization as @xmath5 .
then , we immediately recognize that the basis functions @xmath5 , selected in such a way , satisfy to all criteria of wfs , and can be regarded as one of possible representations of wfs for the full hamiltonian @xmath0 .
indeed , the basis functions @xmath5 are localized , orthonormalized , and ( due to the property of completeness of the basis set ) any eigenvector of @xmath0 in the valent part of the spectrum can be expressed as a linear combination of @xmath5 .
this is a natural result and advantage of using the localized ( or atomic - like ) basis . in the plane - wave basis
, the localized wfs can be constructed from certain number of eigenvectors of @xmath0 in the valent part of the spectrum , for example , by minimizing the square of the position operator @xmath8.@xcite however , we would like emphasize that this is nothing but an elegant way of constructing a compact atomic basis from the extended plane waves , a step which becomes rather unnecessary if one works from the very beginning in an appropriate basis of atomic orbitals .
the precise criterion of the maximal localization ( for example , the minimization of @xmath8 ) is not really important at this stage , because this is merely a mathematical construction and depending on the considered physical property one can introduce different criteria of the `` maximal localization ''
. however , what we typically need in the process of construction of the model hamiltonians for the strongly - correlated systems is different .
for example , for the solution of the many - electron problem it is practically impossible to operate with a large number of wfs of the full hamiltonian @xmath0 .
instead , one would like to concentrate on the evolution of a small number of bands , located near the fermi level , and construct the wfs only for this group of bands .
a concrete example is the @xmath1 bands in many transition - metal oxides , which are typically well separated from the rest of the spectrum.@xcite of course , the wfs for the @xmath1 bands should be orthogonal to the rest of the eigenstates of the full hamiltonian @xmath0 .
this causes an additional complication and the basis functions @xmath5 , though can be regarded as wfs for the full hamiltonian @xmath0 , are no longer the wfs for a limited subspace of eigenstates of @xmath0 , restricted by the @xmath1 bands .
at present , there are two methods , which are commonly used for the construction of wfs for a restricted number of states of the full hamiltonian @xmath0 : the projector - operator method ( for example , employed in refs . , , , and ) and the downfolding method ( employed in refs . , , and ) . in the projector - operator method ,
each ( nonorthonormalized ) wannier function is generated by projecting a trial basis function @xmath9 centered at the site @xmath10 onto a chosen subset of bands ( for instance , the @xmath1 bands , located near the fermi level ) : @xmath11 where @xmath12 is the projector - operator onto the @xmath1 bands , @xmath13 denotes the eigenstate of @xmath0 , and @xmath14 is a joint index combining the band index and the position of the momentum @xmath4 in the first brillouin zone .
the effective tight - binding hamiltonian @xmath15@xmath16@xmath17 in the region of @xmath1-bands is constructed from the matrix elements of @xmath0 in the basis of these wannier states : @xmath18 the orbitals @xmath19 shall be further orthonormalized : @xmath20_{{\bf r}'{\bf r}}^{t't } , \label{eqn : orthonormalization}\ ] ] where @xmath21@xmath16@xmath22 is the overlap matrix : @xmath23 then , the effective tight - binding hamiltonian @xmath24 in the basis of orthonormalized wannier orbitals @xmath25 takes the following form : @xmath26 @xmath24 is typically regarded as the kinetic - energy part of an effective hubbard - type model in the projector - operator method.@xcite for the subsequent discussions , it will be also convenient to write the overlap matrix @xmath21 in the form : @xmath21@xmath16@xmath27 , where @xmath27 is the block of matrix elements of the projector operator ( [ eqn : projector ] ) in the basis of trial orbitals @xmath28 .
since @xmath29 commutes with @xmath0 , the projector - operator method guarantees that @xmath24 has the same set of eigenvalues in the region of @xmath1-bands as the original hamiltonian @xmath0 .
the conventional downfolding method also implies that all atomic basis functions can be divided into two parts : @xmath30@xmath16@xmath31@xmath32@xmath33 , so that the low - energy part of the spectrum in the direct proximity to the fermi level is mainly composed of the @xmath31-states , while @xmath33 is the rest of the basis states , which mainly contribute to the high - energy part .
then , each eigenstate @xmath34 of the full hamiltonian @xmath0 can be presented identically as the sum @xmath35@xmath16@xmath36@xmath37@xmath38 , where @xmath36 and @xmath38 are expanded over the basis states of correspondingly `` @xmath39''- and `` @xmath40''-type , and the matrix equation for @xmath34 takes the form : @xmath41 where @xmath42 are the blocks of matrix elements of @xmath0 in the basis of `` @xmath39''(``@xmath40'')-states .
then , @xmath43 can be expressed from eq .
( [ eqn : seceq2 ] ) as @xmath44 and substituted into eq .
( [ eqn : seceq1 ] ) .
this yields an effective @xmath45-dependent hamiltonian , which acts formally only on @xmath46 : @xmath47 however , @xmath46 is only a part of the eigenvector , which is not orthonormalized .
therefore , @xmath48 should be additionally transformed to an orthonormal basis : @xmath49 this transformation is specified by the overlap matrix @xmath50 which is obtained after the substitution of @xmath43 , given by eq .
( [ eqn : relimination ] ) , into the normalization condition for the full eigenvector @xmath34 : @xmath51@xmath37@xmath52@xmath16@xmath53 .
@xmath54 , which is typically calculated in the center of gravity of the low - energy bands ( @xmath55 ) constitutes the kinetic - energy part of the effective hubbard - type model in the downfolding method.@xcite although the downfolding method does not explicitly require the construction of wfs , they are certainly implied also in this approach and , at least , can be formally reconstructed from @xmath54.@xcite the conventional downfolding method is exact .
however , this property is enforced by the @xmath45-dependence of @xmath56 , which is hardly useful from the practical point of view .
formally , for each @xmath13 , @xmath45 in eq .
( [ eqn : tb ] ) should coincide with the eigenvalue of @xmath0 corresponding to this @xmath13 .
moreover , @xmath57 retains an excessive information about @xmath0 , and the full spectrum of eigenvalues and eigenfunctions of the original hamiltonian @xmath0 can be formally derived from @xmath57 .
however , typically we do not need such a redundant information and would like to use @xmath56 only in order to describe a small group of electronic states located near the fermi level , and do it in the most exact form .
therefore , what we want to do next is to show that in a restricted subspace of eigenstates @xmath58 of the original hamiltonian @xmath0 , the downfolding method can be reformulated as a projector - operator method and retains all attractive features of the latter : namely , it becomes exact and does not depend on @xmath45 .
the trick ( which was actually implemented in refs . and ) is to replace the original hamiltonian @xmath0 in the downfolding method by certain modified hamiltonian @xmath59 , which has the same set of eigenvalues @xmath60 and eigenfunctions @xmath61 in the region of @xmath1-bands .
the rest of the eigenstates is not important for our purposes and can be shifted from the valent part of the spectrum to the region of infinite energies , specified by the parameter @xmath62 .
hence , the modified hamiltonian @xmath59 can be taken in the form : @xmath63 where @xmath64@xmath16@xmath65@xmath66@xmath29 is the projector operator to the subspace orthogonal to the @xmath1-bands . according to the choice of the @xmath31- and @xmath33-basis functions in the downfolding method ,
the latter mainly contribute to the high - energy part of the spectrum , and the overlap between @xmath13 and any linear combination of @xmath33 should be small for the low - energy bands .
therefore , all eigenvalues of @xmath67 are of the order of @xmath62 and located in the high - energy part of the spectrum .
then , it is intuitively clear that , for @xmath62@xmath68@xmath69 , the @xmath45-dependence in eq .
( [ eqn : relimination ] ) can be neglected .
therefore , the method should be exact and not depend on @xmath45 . in order to prove it rigorously ,
it is convenient to use the idempotency of the projector operator : @xmath70@xmath16@xmath64 .
this yields the following identities for the matrix elements @xmath71 in the basis of `` @xmath39''(``@xmath40'')-states : @xmath72 @xmath73 and @xmath74 then , using eqs .
( [ eqn : palgebra1 ] ) , ( [ eqn : palgebra2 ] ) , and ( [ eqn : palgebra3 ] ) , one can prove the following identities , which are valid for the modified hamiltonian ( [ eqn : modifiedh ] ) in the limit @xmath62@xmath68@xmath69 : 1 .
the overlap matrix ( [ eqn : overlapdf ] ) becomes @xmath75 which is the inverse overlap matrix of the projector - operator method ( [ eqn : overlapp ] ) ; 2 .
therefore , all terms proportional to @xmath62 in the hamiltonian ( [ eqn : heff ] ) are exactly cancelled out ; 3 . the @xmath45-dependent part of the hamiltonian ( [ eqn : heff ] ) can be transformed to the form : @xmath66@xmath77 ; 4 .
the remaining part of the hamiltonian ( [ eqn : heff ] ) , which depends neither on @xmath62 nor on @xmath45 takes the following form : @xmath78@xmath66@xmath79@xmath66@xmath80@xmath37@xmath81 , where @xmath82 stand for the matrix elements of the operator @xmath83 in the basis of `` @xmath39''(``@xmath40'')-states . moreover , it is easy to see that @xmath78 coincides with the hamiltonian matrix ( [ eqn : hamiltonianp ] ) of the projector - operator method prior the orthonormalization . then , by replacing one of the projector operators @xmath29 in @xmath84 by @xmath85 and writing explicitly the product of two operators in the basis of `` @xmath39''(``@xmath40'')-states , one obtains the following identities : @xmath86 @xmath87 and @xmath88 using these properties , it is straightforward to verify that the hamiltonian matrix ( [ eqn : heff ] ) of the downfolding method takes the following form : @xmath89 then , the orthonormalization transformation ( [ eqn : tb ] ) yields the following tight - binding hamiltonian : @xmath90 which is totally equivalent to the tight - binding hamiltonian ( [ eqn : tbp ] ) obtained in the projector - operator method .
thus , we have proven that by introducing the modified hamiltonian of the form ( [ eqn : modifiedh ] ) , the downfolding method can be naturally reformulated as the projector - operator method
. therefore , as long as we are interested in the low - energy properties of the system , these two methods are equivalent . here
, one comment is in order . in our proof , we have used the fact that the overlap between @xmath33 orbitals and the eigenstates @xmath61 of the hamiltonian @xmath0 should be small for the low - energy bands .
however , this implies a specific choice for the orbitals @xmath33 ( and also for @xmath31 ) , for which the equivalence between the downfolding and the projector - operator methods actually takes place .
if this property is not satisfied , and the overlap between @xmath33 and @xmath61 is large , the downfolding method will eventually breaks down , as we will clearly see it in our calculations for v@xmath2o@xmath3 below .
the equivalence between downfolding and projector - operator methods appears to be more generic and has a direct implication to the theory of many - electron systems .
indeed , any many - electron state @xmath91 can be expanded over the basis of slater determinants @xmath92 , which play the same role as atomic orbitals in the one - electron version of the downfolding method .
suppose that the many - electron ground state is nondegenerate and one can identify a single slater determinant ( @xmath93 ) , which has the largest weight in the many - electron wavefunction @xmath94 corresponding to the ground state . in practice , such @xmath95 can be obtained in the frameworks of one - electron hartree - fock ( or any other ) theory .
then , each @xmath91 can be formally presented in the form @xmath96@xmath16@xmath97@xmath37@xmath98 , where @xmath97 is the part proportional to @xmath99 and @xmath98 is expanded over the rest of the basis of slater determinants .
@xmath100 can be further eliminated by applying the downfolding method to the true many - electron hamiltonian @xmath101 and its eigenfunctions @xmath102 . by doing so , the solution of the many - electron problem can be reformulated in terms of an @xmath45-dependent self - energy @xmath103,@xcite in a restricted hilbert space of many - electron states , which is formally spanned by only one slater determinant @xmath104 .
nevertheless , such a self - energy contains all information about the ground state as well as the excited states of the many - electron system , which is formally incorporated into the @xmath45-dependence of @xmath105 .
however , the analysis presented in our work clearly shows that if we are interested only in the ground - state properties of the many - electron system , the problem can be reformulated in terms of the projection onto the true ground state @xmath106@xmath16@xmath107 . in this case , the slater determinant @xmath99 plays the role of the trial wavefunction , which makes a formal correspondence between the true many - electron ground state @xmath94 and an auxiliary state of independent quasiparticles , represented by @xmath99 .
as long as we are concerned with the ground - state properties , such a formulation is exact and does not depend on @xmath45 .
of course , the direct application of the projector - operator method to the many - electron problem is hardly meaningful from the practical point of view , because it implies the knowledge of the true many - electron ground state @xmath94 at the first place .
however , this simple example , suggested by the projector - operator method , clearly shows that , in principle , the procedure of obtaining the total energy of a many - electron system can be formulated : * in a restricted hilbert space , which is spanned by only one slater determinant ; * in terms of an @xmath45-independent self - energy ( or an effective potential ) .
this is a simple but exact property of the total energy of a many - electron system , which does not relies on any approximations .
it reminiscences some basic theorems of the density - functional theory ( dft),@xcite and provides some insight into why the same many - electron problem can be formulated in two seemingly different ways , one of which is dft and the other one is the self - energy approach .
now let us come back to the construction of wfs and specify the choice of the trial orbitals @xmath31 .
as it was already pointed out in the introduction , the basis functions @xmath5 can be regarded as wfs of the full hamiltonian @xmath0 .
each basis function is localized around a central atomic site and satisfy certain criterion of the `` maximal localization '' such that any linear combination of @xmath5 will be `` less localized '' or at least have the same degree of the localization as the original basis set @xmath5 . however
, this is not necessarily true if one wants to construct the wfs only for some part of the electronic structure , specified by certain ( restricted ) set of eigenstates @xmath61 of the hamiltonian @xmath0 . due to the additional orthogonality condition to other bands , such a wannier function
will inevitably be a linear combination of @xmath5 and , hence , a less localized function in comparison with the trial atomic orbital @xmath108 .
nevertheless , one can formulate the problem in a slightly different way and ask which atomic orbital centered at the single atomic site is the best representation for the wannier orbital . therefore , we search a new set of orthonormalized trial orbitals in the form : @xmath109 and find the coefficients @xmath110 from the condition that maximizes the projections @xmath111 \rangle$ ] , where @xmath112\rangle$ ] is the nonorthonormalized wannier function constructed from @xmath113 using the projector - operator method : @xmath112\rangle$]@xmath16@xmath114 .
it will automatically guarantee that @xmath113 is the best single - orbital representation for the wannier function in the projector - operator method among the trial orbitals of the form ( [ eqn : newtrial ] ) . by substituting @xmath112\rangle$ ] into expression for the projection @xmath111 \rangle$ ]
, one can easily find that the problem is reduced to the maximization of the functional @xmath115 with respect to the coefficient @xmath110 , where the lagrange multipliers @xmath116 enforce the orthonormalization condition for the new trial orbitals ( [ eqn : newtrial ] ) .
then , the maximization of @xmath117 is equivalent to the diagonalization of @xmath118@xmath16@xmath119 , which is nothing but the site - diagonal part of the density matrix constructed for the @xmath1 bands in the basis of atomic orbitals @xmath120 . after the diagonalization
, we should simply pick up @xmath121 eigenstates @xmath122 , corresponding to the @xmath121 maximal eigenvalues @xmath116 , where @xmath121 is the number of wannier orbitals centered at the atomic site @xmath10.@xcite these @xmath122 will automatically maximize @xmath117 .
this procedure has been proposed in ref . without proof .
then , some intuitive arguments in support of the diagonalization of the density matrix have been given in ref . .
here , we have argued that it can be derived on the basis of a rigorous variational principle .
the simplest model , which nicely illustrates the main idea of the previous section is the one - dimensional chain of atoms @xmath123 ( though the spacial dimensionality is not really important here because a similar conclusion can be obtained in two and three dimensions ) .
we assume that the atomic basis at each site consists of two orthonormal orbitals , say @xmath124 and @xmath125 .
the full hamiltonian @xmath0 is specified by two parameters : the splitting @xmath126 between atomic levels @xmath53 and @xmath127 , and the nearest - neighbor transfer integral @xmath39 ( @xmath128@xmath129 ) , which is assumed to be the same for all pairs of orbitals , @xmath130 the hamiltonian can be easily diagonalized in the reciprocal ( @xmath131 ) space .
this yields two eigenstates : @xmath132 and @xmath133 .
our goal is to construct the wfs for the lowest eigenstate @xmath132 , which can be presented as @xmath134 where @xmath135 and @xmath136 are the fourier transforms of @xmath137 and @xmath138 , respectively , and @xmath139 is nonnegative ( that is defined by a proper choice of the phase of @xmath132 ) .
if @xmath126 is large , this eigenstate will be composed mainly by the atomic orbitals of the first type , i.e. @xmath137 , which correspond to the largest diagonal matrix elements of the density matrix .
therefore , according to the procedure considered in the previous section , in order to obtain the maximally localized representation for the wfs , @xmath137 should be used as the trial orbitals .
now , imagine that we do not know this property and continue to use for the trial orbital @xmath140 a linear combination of @xmath124 and @xmath125 : @xmath141 then , what will happen
? it is straightforward to verify that the wannier function @xmath142 ( in the reciprocal space ) obtained after the projection of @xmath124 onto @xmath143 and the orthonormalization ( [ eqn : orthonormalization ] ) takes the following form @xmath144 | \psi_k^- \rangle,\ ] ] where @xmath145 $ ] stands for the sign of the argument .
this means that @xmath142 differs from @xmath132 by a phase , which is controlled by @xmath146 and depends on @xmath131 . for small @xmath146 , @xmath147@xmath66@xmath148$]@xmath16@xmath53 for all @xmath131
therefore , @xmath142 is a smooth function in the reciprocal space and its fourier image is well localized in the real space ( fig .
[ fig.wfmodel ] ) .
however , when the angle @xmath146 increases ( and starting from certain critical value of @xmath146 ) , @xmath147@xmath66@xmath148 $ ] becomes a _ discontinuous function _ of @xmath131 .
the position of this discontinuity depends both on @xmath146 and on the ratio @xmath149 .
certainly , the discontinuity will affect the spread of wfs and make them much less localized in the real space .
thus , the model nicely illustrates the connection of the localization of the wfs in the real space with the problem of phase of the bloch waves in the reciprocal space .
the degree of localization of the wfs in the real space as a function of @xmath146 is explained in fig .
[ fig.localizationmodel ] . for @xmath146@xmath16@xmath129 ,
when the trial orbital corresponds to the largest eigenvalue of the local density matrix , the wannier function spreads only over the central and nearest - neighbor atomic sites ( which for @xmath149@xmath16@xmath66@xmath150 accumulate @xmath151% of the total weight of the wannier function ) . when the angle @xmath146 increases , the weight of the wannier function transfers ( abruptly , starting from certain value of @xmath146 ) from the central part to more remote atomic sites .
hence , the wannier function becomes less localized .
this is clearly manifested in the extremely slow convergence of the total weight of the wannier function in the real space ( for example , for @xmath149@xmath16@xmath66@xmath150 and @xmath146@xmath16@xmath152 , @xmath151% of the total weight can be regained only after including about 200 coordination spheres around the central site ) .
let us discuss the behavior of the model hamiltonian for the low - energy band in the wannier basis .
the kinetic - energy part of the model hamiltonian ( [ eqn : tbp ] ) is a quadratic function of @xmath153 .
moreover , since @xmath0 is periodic in the real space , its matrix elements will be diagonal with respect to the momentum @xmath131 in the reciprocal space : @xmath154@xmath16@xmath155 .
therefore , the phase @xmath147@xmath66@xmath148 $ ] does not contribute to @xmath154 , and the kinetic - energy part of the model hamiltonian _ will not depend on the form of the trial orbitals_. however , for the coulomb ( and exchange ) matrix elements , the situation can be different .
although the bare coulomb interaction is also periodic in the real space , this is a two - particle interaction , and its matrix elements in the reciprocal space will contain a combination of the four wannier orbitals : @xmath156.@xcite in this case there will be no phase cancelation , and the matrix elements of the coulomb interaction will generally depend on the form of trial orbitals .
then , one may ask which representation is better ?
of course , all are equivalent and the main question here is whether we would like employ some additional approximations or not .
a typical approximation is to retain only the site - diagonal part of coulomb ( and exchange ) matrix elements in the real space,@xcite which may be justified only in the basis of localized wfs ( i.e. , for @xmath146@xmath16@xmath129 in the considered example ) .
then , it is reasonable to expect the matrix elements between different lattice sites to be smaller in comparison with the site - diagonal part .
however , the same approximation is no longer valid for @xmath146@xmath16@xmath152 , where intersite interactions can be of the same order of magnitude as the site - diagonal ones and , therefore , can not be dropped .
now let us illustrate how the same procedure works for more realistic systems .
for these purposes we pick up rather typical ( and , in some sense , canonical ) example of v@xmath2o@xmath3 ( the space group is @xmath157 , no . 167 in the international tables ) .
the sketch of the crystal structure as well as the positions of the main bands of v@xmath2o@xmath3 in lda are summarized in fig .
[ fig.v2o3strucdos ] ( more detailed information about the lattice parameters and the basis functions used the lmto calculations can be found in ref . ) .
since each v is surrounded by six oxygen sites , which form a distorted octahedron , the v(@xmath158 ) states are split into the v(@xmath158-@xmath1 ) and v(@xmath158-@xmath159 ) bands , so that the former ones are located near in the low - energy part of the spectrum and crossed by the fermi level within lda .
the trigonal distortion of the vo@xmath160 octahedra will further split the v(@xmath158-@xmath1 ) levels into one - dimensional @xmath161 and two - dimensional @xmath162 representations .
since the @xmath159 and @xmath162 orbitals belong to the same representation of the point symmetry group , they will mix .
the magnitude of this mixing is no longer controlled by the symmetry of the system , and can be different for different quantities ( depending on how these quantities are constructed from the wavefunctions of v@xmath2o@xmath3 ) . in the following , we will call as `` the @xmath162 orbitals associated with a site @xmath10 '' two eigenstates of the @xmath163 symmetry obtained from the diagonalization of the density matrix @xmath118@xmath16@xmath119 , constructed from the @xmath1 bands , and corresponding to two largest eigenvalues @xmath116 of @xmath118 .
two other eigenstates of the @xmath163 symmetry for which @xmath116 are considerably smaller will be called as the @xmath159 orbitals ( in realistic calculations for v@xmath2o@xmath3 , @xmath164 is about @xmath165 and @xmath166 for the @xmath163 orbitals of the @xmath167 and @xmath168 type , respectively ) .
thus , the situation is similar to the toy model considered in sec .
[ sec : toy ] .
then , we take trial orbitals of the @xmath163 symmetry in the form @xmath169 and construct wfs for the @xmath1 bands . in this expressions , @xmath170 and @xmath171
were obtained from the diagonalization of the local density matrix , and @xmath146 controls the mixing of the @xmath163 orbitals of the @xmath167 and @xmath168 types . on the contrary , the trial orbital of the @xmath161 symmetry
is uniquely determined by the symmetry of the system .
then , according to the arguments of sec . [
sec : trial ] , @xmath146@xmath16@xmath129 should pick up the most localized representation for wfs .
this effect is clearly seen in our calculations , though there is an important differences from the model considered in sec .
[ sec : toy ] . note that in the multi - orbital case , the form of wfs in the reciprocal ( @xmath4 ) space is controlled by a _ matrix _ @xmath172@xmath16@xmath173 ( rather than a single phase factor ) , which generates a new set of wfs after the transformation:@xcite @xmath174 in such a case , the change of @xmath146 will generally induce the change of the whole matrix @xmath172 [ for example , through the additional orthonormalization transformation ( [ eqn : orthonormalization ] ) ] , which makes some difference from the model analysis presented in sec .
[ sec : trial ] .
for instance , there will be no cancelation of @xmath172 in the expression for the matrix elements of @xmath0 in the wannier basis .
therefore , such matrix elements will generally depend on the representation @xmath146 .
nevertheless , we have also observed an abrupt change of wfs in the region of large @xmath146 , in a close analogy with results of the model analysis in fig .
[ fig.localizationmodel ] .
first , let us discuss the results obtained using the downfolding method for the modified hamiltonian ( [ eqn : modifiedh ] ) , which are shown in fig .
[ fig.v2o3localization ] . in practical calculations we set @xmath62@xmath16@xmath175 ry .
then , the downfolding method remains stable only up to @xmath146@xmath16@xmath176 . for larger @xmath146
, some eigenvalues of @xmath67 can fall into the low - energy part of the spectrum , and the whole procedure becomes meaningless .
the wfs have been reconstructed from the tight - binding hamiltonian ( [ eqn : tbp ] ) using the method proposed in ref . , which is stable up to @xmath146@xmath16@xmath177 .
as the angle @xmath146 increases , the weight of the wannier function of the @xmath163 symmetry at the central v site decreases from @xmath178 to @xmath179 ( i.e. , by 5% ) , and the expectation value of @xmath180 , calculated on these wfs , increases from @xmath181 to @xmath182 @xmath183 ( i.e. , by nearly 30% ) .
hence , the wfs become less localized .
is it a big effect or not ?
in fact , everything depends on the considered property .
for example , by assuming for a while that all weight of the wannier function is uniformly distributed over the sphere of the radius @xmath184,@xcite we can estimate ( very roughly ) the change of the bare coulomb interaction as 12% , which is an appreciable value .
it is interesting to note that with the increase of @xmath146 , the degree of localization of the wfs of the @xmath161 symmetry also decreases , though the effect is considerably smaller .
this is not surprising , and the @xmath161 functions should inevitably change because of the additional orthogonalization of them to the @xmath163 orbitals . note that although the @xmath185 and @xmath186 orbitals are orthonormal at the same v site , the matrix elements @xmath187 between different atomic sites @xmath10 and @xmath188 are generally nonzero .
therefore , the wfs of the @xmath161 and @xmath163 symmetry should be additionally orthonormalized , and this transformation depends on the value of @xmath146 .
the same effects leads to the existence of nonvanishing transfer integrals between @xmath161 and @xmath163 orbitals , which are defined as the matrix elements of the tight - binding hamiltonian @xmath189 for @xmath10@xmath190@xmath188 .
it is also instructive to consider other possibilities for trial orbitals , which are found from other principles , and not directly related with the diagonalization of the density matrix .
one possible choice is the @xmath163 orbitals corresponding to the ideal trigonal environment .
for the site 1 depicted in fig .
[ fig.v2o3strucdos ] such orbitals have the following form:@xcite @xmath191@xmath37@xmath192 and @xmath193@xmath66@xmath194@xmath66@xmath195 .
another possibility is the so - called `` crystal - field orbitals '' obtained from the diagonalization of the site - diagonal part of a more general @xmath196@xmath197@xmath196 tight - binding hamiltonian , constructed in the basis of all v(@xmath158 ) orbitals.@xcite both constructions yield less localized wfs of the @xmath163 symmetry in comparison with the ones obtained from the diagonalization of the local density matrix .
nevertheless , the difference is very small ( for example , @xmath8@xmath16 @xmath198 and 2.470 @xmath183 for the ideal trigonal orbitals and the crystal - field orbitals , respectively ) , suggesting that in the case of v@xmath2o@xmath3 , a good starting point for the construction of wfs can be deduced from some other ( for example , geometrical ) consideration .
unfortunately , this rule can not be applied equally well for other compounds .
for example , one clear exception is the distorted perovskite oxides , where the parameters of the tight - binding hamiltonian in the real space are extremely sensitive to the choice of the local coordinate frame , which is not unique and , therefore , the construction of the trial orbitals by means of the diagonalization of the local density matrix seems to be the most reliable approach.@xcite in the multi - orbital case , the degree of localization of the wfs is related with the distribution of the transfer integrals , and it is reasonable to expect that less localized wfs will produce longer - range interactions in the real space .
this is clearly seen in the behavior of averaged transfer integrals in fig .
[ fig.v2o3localization ] .
if for @xmath146@xmath16@xmath199 all nonvanishing transfer integrals are confined within the radius @xmath200@xmath201@xmath202 (including 16 coordination spheres of the v atoms ) , those for @xmath146@xmath16@xmath176 have a much longer tail spreading up to @xmath200@xmath16@xmath203 .
at the same time , the energy dispersion obtained after the diagonalization of the tight - binding hamiltonian in the reciprocal space , @xmath204@xmath16@xmath205 , is practically identical for @xmath146@xmath16 @xmath129 and @xmath176 , and well reproduce the behavior of the original lmto bands ( fig .
[ fig.v2o3ek ] ) .
finally , let us discuss the results of the projector - operator method .
they are summarized in fig .
[ fig.v2o3po ] , showing the weight of wfs at the central v site as a function of @xmath146 and the convergence of the total weight of wfs in the real space for two types of trial orbitals , which are again specified by the parameter @xmath146 . when the angle @xmath146 is not particularly large so that the downfolding method is applicable , these methods are identical , as it follows from the rigorous proof in sec .
[ sec : downfoldingasprojector ] .
for example , the behavior of the central weight in fig .
[ fig.v2o3po ] is practically identical ( within the numerical accuracy , which is caused by slightly different details of calculations ) to that obtained for the downfolding method ( fig .
[ fig.v2o3localization ] ) .
however , the projector - operator method , which can be supplemented practically with any type of the trial orbitals , allows us to consider more extreme scenarios .
for example , what may happen if our guess about the form of trial orbitals was totally wrong ?
this situation is nicely illustrated by results of our calculations for @xmath146@xmath16@xmath152 ( i.e. , by choosing the trial @xmath163 orbital in an orthogonal subspace to that , which maximizes local part of the density matrix ) . in this case
the wannier function simply `` blows up '' , and its total weight does not fully converge even within @xmath200@xmath206@xmath207 around the central v site.@xcite thus , the situation is very similar to the model analysis in sec .
[ sec : toy ] and apparently related with the discontinuity of the transformation matrix ( [ eqn : transformationmatrix ] ) in the reciprocal space .
we have considered the problem of construction of the wannier functions in the first - principles electronic structure calculations starting from localized atomic orbitals .
there is a number of methods , which are used for these purposes .
we have argued that some of them can be unified and formulated in an absolutely exact and identical way .
we have demonstrated this idea for two quite a popular nowadays branches of methods : the so - called projector - operator method ( considered , for example , in refs . , , and ) and the downfolding method ( considered in refs . , , and ) .
we believe that such an analysis will help to resolve a number of controversies existing in this area .
of course , identical methods should produce identical results , and we have shown that the root of the differences between different applications is related with the choice of the trial orbitals , which are used in order to generate the wfs in the projector - operator method or the basis functions for the construction of the effective tight - binding hamiltonian in the downfolding method .
we believe that it would be very important in the nearest future to make a similar analysis and generalizations for the order-@xmath208 muffin - tin orbital ( nmto ) method.@xcite however , it is beyond our present abilities .
the trial orbitals effectively control the localization of wfs . to this end , we have argued that an optimal choice for the trial orbitals can be obtained by maximizing the site - diagonal part of the density matrix , constructed from the bands of interest .
our numerical calculations show that such a construction already provide a good degree of localization for wfs . of course
, depending on the purposes , one can further optimize the wfs ( for example , by minimizing @xmath8 or any other quantity).@xcite however , as a starting point in this procedure one can always use the wfs generated by minimizing the site - diagonal part of the density matrix .
this is simple and very efficient procedure .
we have also shown that some deviations from this principle and selection of the trial orbitals in an arbitrary form may lead to a substantial delocalization of wfs in the real space , which is related with discontinuity of their phase in the reciprocal space .
the work of i. v. solovyev has been partially supported by grant - in - aids for scientific research in priority area `` anomalous quantum materials '' from the ministry of education , culture , sports , science and technology of japan .
z. v. pchelkina and v. i. anisimov acknowledge the support by russian foundation for basic research under the grants rffi-04 - 02 - 16096 and rffi-06 - 02 - 81017 .
z. v. pchelkina also acknowledges the support from the research council of president of the russian federation ( grant no .
2006.2 ) , russian science support foundation , dynasty foundation , and international center for fundamental physics in moscow .
v. i. anisimov , d. e. kondakov , a. v. kozhevnikov , i. a. nekrasov , z. v. pchelkina , j. w. allen , s .- k .
kim , p. metcalf , s. suga , a. sekiyama , g. keller , i. leonov , x. ren , and d. vollhardt , phys .
b * 71 * , 125119 ( 2005 ) . note that the matrices @xmath209 and @xmath210 have the same set of eigenvectors .
moreover , if @xmath210 is positively defined ( that is specified by the choice of the phase factor in the definition @xmath209@xmath16@xmath210@xmath197@xmath210 ) , these eigenvectors will be arranged in the same order ( i.e. , of decreasing eigenvalues ) .
therefore , the additional orthonormalization of wfs ( [ eqn : orthonormalization ] ) should not change our conclusion .
note that further increase of the radius does not seem to make sense , because our calculations are always performed in the finite mesh of @xmath4-points ( in our case , generated using @xmath207@xmath211@xmath207@xmath211@xmath207 divisions of the reciprocal translation vectors ) .
therefore , for large distances in the real space we inevitably face the problem of accuracy of integration in the reciprocal space . | the problem of construction of the wannier functions ( wfs ) in a restricted hilbert space of eigenstates of the one - electron hamiltonian @xmath0 ( forming the so - called low - energy part of the spectrum ) can be formulated in several different ways .
one possibility is to use the projector - operator techniques , which pick up a set of trial atomic orbitals and project them onto the given hilbert space .
another possibility is to employ the downfolding method , which eliminates the high - energy part of the spectrum and incorporates all related to it properties into the energy - dependence of an effective hamiltonian .
we show that by modifying the high - energy part of the spectrum of the original hamiltonian @xmath0 , which is rather irrelevant to the construction of wfs in the low - energy part of the spectrum , these two methods can be formulated in an absolutely exact and identical form , so that the main difference between them is reduced to the choice of the trial orbitals . concerning the latter part of the problem
, we argue that an optimal choice for trial orbitals can be based on the maximization of the site - diagonal part of the density matrix .
this idea is illustrated for a simple toy model , consisting of only two bands , as well as for a more realistic example of @xmath1 bands in v@xmath2o@xmath3 . using the model analysis
, we explicitly show that a bad choice of trial orbitals can be linked to the _ discontinuity _ of phase of the bloch waves in the reciprocal space , which leads to the delocalization of wfs in the real space .
nevertheless , such a discontinuity does not necessary contribute to the matrix elements of @xmath0 in the wannier basis .
similar tendencies are seen in realistic calculations for v@xmath2o@xmath3 , though with some variations caused by the multi - orbital effects .
an analogy with the search of the ground state of a many - electron system is also discussed . |
in anticipation of the new observational window afforded by interferometric gravitational - wave observatories now under construction , there has been considerable interest in the rates of formation and merger of binary systems containing at least one neutron star . for double neutron - star ( ns - ns ) systems , estimates of these rates depend on the assumed `` detectable lifetimes '' of the radio pulsars in known close ns - ns binaries .
the detectable lifetime of a known ns - ns binary is , presumably , the sum of the post - accretion age of the observed radio pulsar and the remaining timespan during which it will continue to be observable . for the latter ,
some past authors have used the time to coalescence of the system , while others have argued that luminosity evolution limits the detectable lifetime . for age estimates , past
work has relied exclusively on the `` characteristic '' spin - down timescale @xmath9 , but @xmath10 is an accurate chronological age only if the present - day spin period @xmath6 is much greater than the pulsar s period @xmath11 at `` birth '' ( i.e. , at the end of the accretion phase for recycled pulsars ) , and for dipolar radiation braking with a constant magnetic field .
since @xmath11 may not be small compared to @xmath6 , the spindown timescale is generally an overestimate of the time since spinup ; consequently , birth and merger rates for ns - ns systems based on @xmath10 may be underestimates .
we argue that a neutron star s rotational history since the cessation of spin - up may be used to better estimate its true age . for each known system , we check such age estimates against kinematic constraints derived from motion in the galactic gravitational potential ( sections 2 and 3 ) . in section [ br.sec ] , we use our results to estimate the detectable lifetimes , and hence birthrates , of the known ns - ns binaries , revisiting the assumptions underlying earlier work .
in addition , we show that there are kinematic solutions for both b1534 + 12 and b1913 + 16 for which they are young ( ages @xmath12 myr ) ; we describe how these ages , if characteristic of binary neutron stars , might alter estimates of their birth and merger rates .
finally , we show in section [ accret.sec ] that the phenomena of spiral - in and merger of ns - ns systems are reflected in the @xmath6-@xmath7 diagram , and argue that details of the spin - up process for other binary pulsars similarly leave behind important imprints .
through accretion of mass and angular momentum from an evolving companion star , a neutron star can be spun up to an equilibrium period approximated by the kepler orbital period at the alfvn radius ( ghosh & lamb 1992 ) , @xmath13 where @xmath14 is the neutron star s magnetic moment expressed in units of @xmath15 g@xmath16 , @xmath17 is the opacity of the accreting material in units of 0.4 @xmath18g@xmath1 ( the thomson opacity for ionized hydrogen ) , @xmath19 is the stellar radius in units of @xmath20 cm , @xmath21 is the neutron star s mass in units of 1.4 @xmath22 , and @xmath23 is the mass accretion rate in units of the eddington rate ( accretion onto a sphere is assumed ) .
the quantity @xmath24 is of order unity , where @xmath25 encapsulates uncertainty in the alfvn radius and @xmath26 is the fastness parameter . in the magnetic dipole model , @xmath27 , where @xmath28 is the stellar moment of inertia and @xmath29 is a geometrical factor , which then implies the `` spin - up line '' relationship between @xmath6 and @xmath7 for the endpoint of spin - up by accretion , @xmath30 where @xmath6 is in seconds , and @xmath31 g@xmath18 .
= -1 while the scaling @xmath32 in equation ( [ peqline.eq ] ) is well - founded for neutron stars with predominantly dipolar magnetic fields , there are numerous sources of uncertainty in the coefficient @xmath33 : the parameters @xmath34 , @xmath35 , and @xmath36 are not well known , and sub - eddington accretion , @xmath37 , is likely ( e.g. , if inflow onto the neutron star is restricted to the polar caps ) . also , the preferred spin - up scenario for ns - ns systems , accretion from the helium core of a companion stripped of its hydrogen envelope , suggests an opacity @xmath38 .
we adopt a `` fiducial '' spin - up relation with coefficient @xmath39 s@xmath40 and perform calculations for a range of values .
if the evolutionary timescale of the donor star is short , the accretion phase may not last long enough to spin the neutron star up to equilibrium .
the time to reach equilibrium is @xmath41 ; if mass accretion lasts a time @xmath42 yr @xmath43 , the spin period at the end of the accretion phase is @xmath44 , which implies the relationship @xmath45 where we have used @xmath46 ms .
equation ( [ p-8.eq ] ) is extremely sensitive to all of the parameters , and represents a family of @xmath47 spin - up lines in the @xmath6-@xmath7 diagram ; the line for @xmath48 and @xmath49 yr is plotted in fig . 1 .
the intersection of this alternate spin - up line with the equilibrium @xmath50 line defines a minimum post - accretion period ( corresponding to @xmath51 ) that depends on the total mass accreted , but is insensitive to other parameters ( see also kulkarni & phinney 1994 ) . if we include the effects of accretion - induced field decay following shibazaki ( 1989 ) , we find no qualitative change in the form of the non - equilibrium spin - up line . the time to spin down from birth period @xmath11 to present - day period @xmath6 , with constant braking index @xmath52 , is @xmath53,\ ] ] where the characteristic age @xmath54 $ ] approaches the true age only in the limit @xmath55 . in the @xmath6-@xmath7 diagram ,
intersection of the spin - down trajectory @xmath56 with an assumed spin - up line provides a recycled pulsar s initial period which , through equation ( [ age.eq ] ) yields the epoch since the end of spin - up .
we apply this technique to estimate the ages of the known ns - ns systems ; should accretion in a given system have halted before the equilibrium spin period was reached , or if the accretion rate was sub - eddington , such an estimate becomes a strong upper bound on the true age . for the fiducial spin - up line and @xmath57 , spin - down trajectories for the known ns - ns systems are shown in fig . 1 , the @xmath6-@xmath7 plane for recycled pulsars .
the validity of post - accretion spin - down ages may be questioned if there is no single spin - up line as given by equation ( [ peqline.eq ] ) .
evidence that accretion consistent with equation ( [ peqline.eq ] ) is responsible for b1913 + 16-like systems ( see section [ accret.sec ] ) is provided by examination of the @xmath6-@xmath7 diagram : all three of the known systems lie near the spin - up line .
the pulsars in these binaries must have originated above and to the left of their present positions in this diagram , and in any case age estimates become increasingly insensitive to large values of the coefficient @xmath33 ( cf .
fig . 2 ) .
more importantly , @xmath33 can not be made arbitrarily big , as there are no known recycled pulsars above the spin - up line , excepting two pulsars in globular clusters ( section [ rega.sec ] ) .
we therefore adopt the value @xmath58
s@xmath40 as a reasonable maximum to constrain ages at the upper end .
we compute spin - down ages for several values of the braking index , but for @xmath59 the slope of the spin - up line differs from @xmath50 in a way that is difficult to quantify without a model for the braking torque produced by the magnetic field .
we have applied the model of melatos ( 1997 ) , which successfully explains the measured braking indices @xmath60 of three young pulsars as a result of their magnetospheric structure , to objects near the spin - up line , and find that they should have @xmath61 , where @xmath62 is initially very small and decreases further as objects spin down .
this result supports age estimates derived for spin - up lines @xmath63 and spin - down with @xmath57 . if magnetic fields of recycled pulsars evolve on timescales comparable to @xmath64 , however , @xmath65 is possible ( for specific models , see e.g. ruderman , zhu & chen 1998 ) ; in such models , the spin - down time is altered , but not necessarily the equilibrium spin - up relation .
-1 a radio pulsar s distance @xmath66 from the galactic plane together with its velocity derived from proper motion measurements can be used to estimate its age through integration of the pulsar s trajectory in the galactic gravitational potential .
sources of uncertainty in addition to measurement errors in distance and proper motion are the unknown line - of - sight velocities and non - zero @xmath66-heights at `` birth '' ( the end of the spin - up phase for recycled pulsars ) .
we calculate gravitational restoring forces towards the disk at the present - day galactocentric radii of b1534 + 12 and b1913 + 16 using the potential of paczyski ( 1990 ) .
useful kinematic age constraints can be derived for b1913 + 16 because it is at low galactic latitude and its distance and proper motion are well determined .
kinematic and spin - down age estimates are compared in fig .
2 . trajectories in the top panel are constrained to cross the system s current @xmath66-height moving upward at the correct present - day velocity @xmath67 ; they imply that the system is either leaving the galactic plane for the first time ( @xmath68 myr ) , or has completed at least one - half cycle of its oscillation in @xmath66 . in the latter case , i.e. , for ages greater than a few tens of myr , the system is likely to have sampled the galactic potential over a large range of galactocentric radii @xmath69 , so that the assumption of a constant @xmath69 breaks down this will stretch or compress the plotted trajectories in a way that depends on the actual path taken by the system through the galaxy , which is unknown .
the unknown line - of - sight velocity @xmath70 also modifies these trajectories , but at the system s low galactic latitude , @xmath71 , only a very large @xmath70 would be significant : to change @xmath67 by 10% , approximately the difference between tracks _ ( a ) _ and _ ( b ) _ in fig . 2 , requires @xmath72 km - s@xmath1 . the characteristic age of b1913 + 16 , assuming a @xmath7 corrected for galactic acceleration ( damour & taylor 1991 ) , is @xmath73 myr , while the combined kinematic and spin - evolution constraints yield two possible solutions for the age : a young solution , @xmath74 myr , which is largely insensitive to uncertainties in the assumed distance , present - day @xmath67 , and braking index , and an old solution , 6080 myr , where the upper bound assumes @xmath57 and @xmath75 s@xmath40 .
it is noteworthy that the old solutions provide an upper bound on the braking index , @xmath76 , unless @xmath33 is very large .
the characteristic age of psr b1534 + 12 is 252 myr , assuming the acceleration correction for @xmath7 given by stairs et al .
the pulsar s distance from the spin - up line ( fig .
3 ) implies @xmath77 myr , for @xmath57 and @xmath78 s@xmath40 .
the two estimates differ little because the condition @xmath55 is nearly met if the pulsar was born on the spin - up line .
based only on proper motion and distance , psr b1534 + 12 appears to be moving _ toward _ the galactic plane with transverse velocity @xmath79 km - s@xmath1 from its current position at @xmath80 pc .
because the system is at galactic latitude @xmath81 , a modest radial velocity could reverse the sign of this apparent motion . without more information about the present - day @xmath67 the system s kinematic history is largely unconstrained . we have nevertheless derived the relevant motion in the galactic potential under various assumptions . in general , the minimum @xmath67 required at birth for the system to just reach its present @xmath66-height from a starting point @xmath82 pc is @xmath83 km - s@xmath1 .
if we constrain the age to be roughly 200 myr , allowed solutions span a range of initial velocities and maximum @xmath66-heights depending on the number of oscillation periods that have elapsed since the system s birth , but likely solutions require a birth @xmath67 between 100 and 200 km - s@xmath1 and @xmath84 of 2.04.5 kpc .
notably , the solution in which the pulsar has completed just less than half of its oscillation cycle and is returning to the plane for the first time requires a birth velocity @xmath85 km - s@xmath1 and @xmath86 kpc , which seem unlikely .
if the age is in fact as large as we believe , the system has probably crossed the plane more than once . on the other hand
, the kinematic age can be made arbitrarily small if the radial velocity is assumed to be large and the system is leaving the plane for the first time : an age @xmath12 myr requires a @xmath67 at birth of 90 km - s@xmath1 and a present - day @xmath87 km - s@xmath1 .
the characteristic age of b2127 + 11c is @xmath88 myr ; the apparent @xmath7 of this pulsar is likely contaminated by the system s unknown acceleration in the gravitational potential of m15 ( but phinney 1992 estimates the contamination to be at most 2% ) . although the evolutionary history of b2127 + 11c is murky , the pulsar s spin - down track in @xmath6 and @xmath7 probably originated somewhere near the conventional spin - up line
we estimate the time since spin - up for this pulsar to be @xmath89 myr ( fig . 3 ) .
for a given pulsar in a ns - ns binary , the birthrate of similar objects may be estimated , following phinney ( 1991 ) , using @xmath90 , where @xmath91 is the total galactic volume in which ns - ns binaries are found , @xmath92 is the luminosity- and density - weighted volume that pulsar surveys have sampled for objects with identical orbital period , and @xmath93 is an appropriate source lifetime .
the total birthrate follows by summing over objects , @xmath94 .
a more rigorous treatment , based on a likelihood function , uses all information in pulsar surveys , including non - detections in much of the survey volume as well as detected pulsars .
although the scarcity of known binary pulsars would frustrate any attempt to constrain meaningfully the many important parameters that characterize the galactic population of ns - ns binaries , we can nevertheless use the formalism to understand the proper interpretation of the parameters that enter into any birthrate estimate .
we find that a maximum likelihood estimate for the binary neutron star birthrate @xmath95 is @xmath96 where the number of detected binary pulsars is @xmath97 and @xmath98 is the joint expectation value of the fraction of all existing binary pulsars detected times source lifetime , summed over surveys .
the denominator involves an integral of the number density of binary pulsar systems ( normalized to one ) times their average lifetime convolved with observational selection functions that limit the spatial volume and range of lifetimes amenable to detection . at a more fundamental level ,
the necessary integral involves physical distributions of the positions , center of mass speeds , orbital elements and luminosities of binary pulsars at birth , and also requires physical models for the evolution of these distributions along with detailed understanding of the phenomenological limitations of searches for such systems .
ideally , the physical distributions involved in computing @xmath98 should be determined from the available data , but , since very few ns - ns binaries are known , we , like others before us , are forced to appeal to prior knowledge or theoretical expectations about the binary populations instead .
moreover , we approximate the expectation value for each survey by @xmath99 where @xmath100 is the fraction of all ns - ns binaries we expect to be observable , and @xmath101 is the detectable lifetime of such systems ( see , e.g. , van den heuvel & lorimer )
. we can also approximate @xmath102 where @xmath103 is a measure of how representative the detected binary pulsars are of the overall galactic population .
if the ns - ns binaries found so far are typical of the overall population then @xmath104 ; if not , then @xmath3 could be small ( e.g. , a substantial fraction of binary pulsars could have orbital periods short enough that detection is unlikely ; johnston & kulkarni 1991 , bloom , sigurdsson & pols 1998 ) . for @xmath105 ,
we find a birthrate estimate from a single detected binary pulsar similar to phinney s ( 1991 ) @xmath106 , except that it is now clear that ( i ) there is an underlying assumption that the detected system is typical of the population and ( ii ) the appropriate lifetime to use is the expected time over which the pulsar should be detectable . finally , in the absence of detailed knowledge of the spatial distribution of ns - ns binaries , we shall approximate ( following the language of cordes & chernoff 1997 ) the `` detection volume '' @xmath92 by the `` sampled volume '' @xmath107 , which is the sensitivity - limited region searched by pulsar surveys .
this approximation is conservative in that @xmath107 overestimates @xmath92 ( see below ) .
apart from @xmath3 , a major uncertainty in any birthrate estimate is the value of @xmath101 .
below , we estimate this parameter by computing the expected detectable lifetimes for b1913 + 16 and b1534 + 12 separately .
how to translate these values into an estimate of @xmath101 is somewhat subjective , as it depends on one s prior notions of which , if either , of them is `` more typical . ''
this subjectivity pervades previous estimates of @xmath95 as well as our own : _ all _ are proportional to nonrigorous estimates of @xmath108 . as should be evident from equation ( [ eq : brest ] ) , evaluating @xmath95 as the sum of birthrate estimates for individual binary pulsar detections represents a particular choice for @xmath108 . in a manner similar to phinney ( 1991 ) , the birth and merger rates of ns - ns systems have been estimated by narayan , piran & shemi ( 1991 ) and curran & lorimer ( 1995 ) under , effectively , different assumptions for @xmath91 and @xmath107 . in each case , the lifetimes of individual systems were taken to be the sum of the pulsar s characteristic age and the binary s coalescence time .
curran & lorimer s ( 1995 ; hereafter cl95 ) estimate has twice been revised .
van den heuvel & lorimer ( 1996 ; hereafter vl96 ) argued that the effective `` detectable lifetime '' ( rather than the full lifetime culminating in merger ) should be used in birthrate calculations ; they estimated radio luminosity decay timescales for each pulsar and proposed a merger rate nearly three times that of cl95 .
stairs et al .
( 1998 ; hereafter sac+98 ) determined that the distance to psr b1534 + 12 is larger than previously believed , implying that @xmath107 had been underestimated by cl95 .
their correction reduced van den heuvel & lorimer s ( 1996 ) estimate by a factor of 2.5 , bringing it effectively back to cl95 s value .
these revisions are summarized in table 1 . [ cols="<,^,^,^,^,^ " , ] @xmath109this entry summarizes our estimate @xmath110 , equation ( [ br1.eq ] ) , which is based on different assumptions for computing galactic scale factors than the first four entries ( see text ) .
we first report here our own revision consistent with the methods and assumptions of cl95 .
we then evaluate the assumptions entering into previous @xmath91 and @xmath107 calculations and report a birthrate reflecting recently completed radio pulsar surveys . as argued by van den heuvel & lorimer ( 1996 ) , and as emphasized above , we need the detectable lifetime @xmath111 of the sources to estimate their birthrate .
vl96 considered radio luminosity decay and showed that both b1534 + 12 and b1913 + 16 will likely become undetectable before they merge with their companions .
however , detection is biased against finding recycled pulsars in short - period binaries ; in present surveys , neither pulsar might have been found if their binary periods were @xmath112 times their present values .
the gravitational radiation lifetimes of b1534 + 12 and b1913 + 16 are @xmath113 yrs and @xmath114 yrs , respectively . for b1534 + 12
, the orbital period will halve in about @xmath115 yrs and for b1913 + 16 , in about @xmath116 yrs . the latter result implies that the detectable lifetime of b1913 + 16 is limited by orbital selection rather than by luminosity decay .
as shown in table 1 , then , our revision to cl95 s estimate consists of adopting ( 1 ) the stairs et al .
( 1998 ) distance to b1534 + 12 , ( 2 ) our preferred age estimates for b1913 + 16 , 65 myr , and b1534 + 12 , 200 myr , and ( 3 ) retaining van den heuvel & lorimer s ( 1996 ) luminosity decay for b1534 + 12 , but assuming b1913 + 16 will remain detectable for 180 myr as a result of observational selection due to the rapidly decaying orbit determine the typical detectable lifetime . ] .
we thus derive a birthrate , uncorrected for beaming or a luminosity function to account for dim objects , of @xmath117 yr@xmath1 by summing the individual rates of the two objects , a 23% increase over the latest revision .
we note that cl95 suggest a factor of 10 increase to account for low - luminosity objects , and a further factor of three increase for beaming .
values of @xmath91 used previously range from @xmath118900 kpc@xmath119 ; the lowest @xmath91 ( cl95 ) attributes a population i radial distribution and a scale height @xmath120 kpc to ns - ns binaries .
the kinematic `` ages '' of binary pulsars are directly related to their scale heights .
previously , a range @xmath1211 kpc has been used to estimate binary pulsar birthrates , but this does not typify the @xmath66 distribution of old ns which , for isolated pulsars , includes a large fraction that oscillate to @xmath122 kpc .
we argue that the space velocities of ns - ns binaries are sufficently large ( up to @xmath123 km - s@xmath1 ) that both the radial and @xmath66 distributions should be extended , to roughly a 10 kpc characteristic radius and @xmath124 kpc .
this yields @xmath125 kpc@xmath119 .
estimates of @xmath107 are highly sensitive to modeling of selection effects in surveys that have or could have found the ns - ns binaries .
the volume sampled ( per unit solid angle ) in a survey is ( e.g. , cordes & chernoff 1997 ) [ vmax1.eq ] = dl_p ( l_p ) _ 0^
dd d^2 , where @xmath126 is the maximum distance sampled , @xmath127 is the minimum detectable flux density , and @xmath128 is the luminosity function ( with unit area ) of the `` pseudo luminosity '' @xmath129 , where @xmath130 is the 400 mhz flux density tabulated in taylor , manchester & lyne ( 1993 ) .
equation ( [ vmax1.eq ] ) provides a good estimate of the detection volume @xmath92 when the scale height @xmath131 of the ns - ns population is high enough , that is , when even the brightest pulsars in the population can not be detected beyond a vertical distance @xmath131 above the location of the solar system in the galactic disk ; for smaller @xmath131 , equation ( [ vmax1.eq ] ) overestimates @xmath92 .
previous work indicates @xmath132 over several orders of magnitude of @xmath133 , for both isolated pulsars ( lyne et al .
1998 ) and millisecond pulsars ( cordes & chernoff 1997 ) .
the lower cutoff is taken to be @xmath134 mjykpc@xmath135 .
for the upper cutoff , we use the value for b1913 + 16 , @xmath136 mjykpc@xmath135 , as its pulse shape suggests that the radio beam points directly at us , yielding a large , perhaps maximal luminosity for its @xmath6 and @xmath7
. this luminosity function does not account for pulsars beamed away from us completely , so our rates still must be adjusted upward for beaming .
the minimum detectable flux density , @xmath127 , is generally a strong function of @xmath6 , @xmath137 , @xmath138 , and direction , as well as of instrumental parameters . for long spin- and orbital - period pulsars
, @xmath127 asymptotes to a constant for low @xmath137 . using the asymptotic @xmath139 mjy representative of recent surveys at arecibo , we obtain [ vmax.eq ]
< s_min^-3/2l_p_1l_p_2 ^ 1/2 10 ^ -2.4kpc^3deg^-2 s_min,0.8 ^ -3/2 l_p_1,1 l_p_2,200 ^ 1/2 , where @xmath140 mjy , @xmath141 mjykpc@xmath135 , and @xmath142 mjykpc@xmath135 .
arecibo surveys have covered @xmath143 deg@xmath135 ( camilo 1998 ) in recent years ; surveys at parkes and jodrell have covered more solid angle but with @xmath144 mjy . together ,
these imply @xmath145 kpc@xmath119 . we stress that this value may be a significant overestimate , for two reasons .
first , the asymptotic flux @xmath127 is not appropriate for rapidly rotating , highly dispersed pulsars , especially at low galactic latitudes .
second , the true low - end cutoff of the radio luminosity distribution could occur at luminosities much lower than 1 mjykpc@xmath135lyne et al .
( 1998 ) show that the luminosity function for non - recycled pulsars begins to turn over for @xmath146 mjykpc@xmath135 , but that there is no evidence for a turnover at 1 mjykpc@xmath135 for spun - up objects .
using our estimates of @xmath91 and @xmath107 , and expressing @xmath101 in units of the mean of the detectable lifetimes of b1913 + 16 and b1534 + 12 , we obtain a rate @xmath147 where @xmath2 is the beaming fraction , under the assumption that these binaries are typical of the galactic population of ns - ns systems .
this result is nearly a factor of 7 smaller than our revision of the cl95 estimate corrected for their suggested factor of 10 contribution from low - luminosity objects , @xmath148 yr@xmath1 .
we attribute this difference to the increased galactic volume recently surveyed for rapidly rotating pulsars as represented here by @xmath107 , and our inclusion of the radio luminosity function in estimating @xmath107 instead of the somewhat arbitrary multiplying factor of 10 of cl95 . aside from the beaming factor @xmath2 , equation ( [ br1.eq ] ) is a _ lower bound _ to the actual birthrate for three reasons .
first , the birthrate function for binary neutron stars could be skewed toward short orbital periods ( bloom , sigurdsson & pols 1998 ) where detection becomes unlikely ( johnston & kulkarni 1991 ) . if this is true , then @xmath3 is small , and the birthrate is larger than @xmath149 by a factor of @xmath150 ( see eqs . [
[ eq : brest ] ] and [ [ eq : fest ] ] ) .
second , @xmath107 overestimates @xmath92 : for high - sensitivity surveys such as those carried out at arecibo , sky coverage at high latitudes samples a large volume at @xmath66-heights beyond the likely scale height of ns - ns binaries .
third , the parameters entering into our estimate of @xmath107 , equation ( [ vmax.eq ] ) , are sufficiently uncertain that the sampled volume itself is likely overestimated .
we have probably overestimated @xmath101 as well . taken together , these considerations suggest that @xmath149 may underestimate the true rate by an order of magnitude or more .
previous estimates of the binary pulsar birthrate , including our @xmath149 , presume that their lifetimes are limited by either spiral - in or spin - down .
ideally , one should dispose with this prior assumption and determine @xmath101 from the data , if possible .
we note that kinematical considerations imply that b1913 + 16 may be @xmath12 myr old , and a similarly young age can not be ruled out for b1534 + 12 .
if a significant fraction of recycled pulsars discovered in future surveys prove to be young ( age @xmath12 myr ) , but in binary systems with long gravitational radiation and spindown lifetimes ( @xmath1181000 myr ) , then we would be forced to reduce our estimate of the detectable lifetime @xmath93 .
( this would also imply that something other than gravitational radiation or spindown regulates the timescale over which binary pulsars are observable . )
such a short lifetime would require a large birthrate , @xmath151 yr@xmath1 , for such systems , which could dominate the rate of neutron star mergers if they are common .
as long as there is not a statistically significant population of anomalously young binary pulsars , this large birthrate is not required .
given that only a few pulsars are known to exist in tight binaries , however , we can not rule out the existence of short - lived binary pulsars that are born frequently . both of our birthrate estimates @xmath149 and @xmath152 can easily take on values larger than @xmath153 yr@xmath1 when provision is made for beaming and the large uncertainties in @xmath91 , @xmath107 , and @xmath92 .
this possibility prompts us to re - examine the upper bound of @xmath154 yr@xmath1 proposed by bailes ( 1996 ) based on the absence of any known `` normal '' ( i.e. , non - recycled , long - period , high magnetic - field ) pulsars in a close ns - ns binary .
we begin by defining the following quantities : * @xmath95 , the birthrate of ns - ns binaries that will merge in a hubble time , * @xmath155 , the detectable lifetime of the non - recycled neutron star in such a binary , * @xmath156 , the fraction surveyed to date of the galactic volume occupied by such binaries ( @xmath157 ) , * @xmath158 , the fraction of such binaries that are detectable despite orbital smearing of the _ non - recycled _
pulsar s pulsations , * @xmath159 , the birthrate of ( normal ) isolated pulsars , * @xmath160 , the detectable lifetime of a normal pulsar , * @xmath161 , the fraction surveyed to date of the galactic volume occcupied by normal pulsars , * @xmath162 , the number of non - recycled pulsars known to date .
as no normal pulsars have been found in ns - ns systems expected to merge in a hubble time , days and will not coalesce in less than a hubble time .
] we have n t_nb f_vb f_orb f_b < 1 .
similarly , for the normal , isolated pulsar population , n_n t_n f_vn f_b = n_n . taking the ratio , and assuming @xmath163 , we get [ bailes.eq ] n < ( ) n_n .
the quantity in parentheses , which was not considered by bailes ( 1996 ) explicitly , can be less than or greater than one , but is probably close to unity .
we believe the scale height of binaries is somewhat smaller than that of the isolated normal pulsars , so @xmath164 is likely ( i.e. , a larger fraction of the galactic volume of binaries has been searched ) . because of orbital acceleration , @xmath158 is less than one , but represents a less severe selection effect for slow pulsars than for recycled objects .
bailes ( 1996 ) assumed @xmath165 , and adopted the largest allowed birthrate , @xmath166 yr@xmath1 , from the analysis by lorimer et al .
( 1993 ) of the population of normal , isolated pulsars .
a similar recent effort by lyne et al . ( 1998 ) based on a larger sample of objects finds a maximum birthrate of one every 60 yrs .
furthermore , the number of cataloged normal pulsars with measured spin - down rates , i.e. , for which non - membership in a ns - ns binary has been firmly established , is 501 . assuming unity for the factor in parentheses in equation ( [ bailes.eq ] ) , an upper bound on the birthrate of close ns - ns binaries is then @xmath167 yr@xmath1 , independent of beaming .
this rate should be corrected upward by a statistical `` poisson '' factor and by the factor @xmath158 ; we therefore believe that a robust upper limit is n < 10 ^ -4yr^-1 .
we point out , however , that the rate @xmath152 need not be subject to this bound , as the assumption @xmath163 might not hold if the detectable lifetimes of ns - ns binaries were somehow not governed by the spin - down or gravitational radiation timescales . finally , we refer , for completeness , to the recent work of rosswog et al .
( 1998 ) who find that a ns - ns merger rate of @xmath168@xmath169 yr@xmath1 could acount for the apparent deficit of @xmath170-process nuclei .
surveys over the last 16 years have been sensitive to objects with @xmath171 ms to varying degrees .
certain portions of the @xmath6-@xmath7 diagram have become populated as a result , while others have remained empty .
we argue that three of these unpopulated regions , labelled , , and in fig . 1 ,
provide insights into the evolution of binary systems containing neutron stars .
in particular , there is a dearth of pulsars in region where we would expect to find inspiralling ns - ns binaries .
we assess the significance of this and other deficits below .
there are no known selection effects to prevent detection of pulsars in region : objects with spin - down rates higher than the low - field millisecond pulsars would likely be brighter and so more easily detectable .
still , no recycled pulsars in the galactic disk are known above the spin - up line .
we interpret this as a qualitative validation of the ghosh - lamb spin - up line .
we discount pulsars in globular clusters , here and in the discussions to follow , because of their uncertain spin - down rates and complicated evolutionary histories .
region is below and to the right of the known ns - ns systems , for @xmath172 ( fig . 1 ) .
a deficit of known pulsars here can occur for three reasons , all of which likely play some role : 1 .
ns - ns binaries born near the spin - up line will coalesce before reaching region . in fig . 1
, the spin - down tracks plotted for the three ns - ns systems end at lower right at the expected coalescence time for each system .
only b1534 + 12 penetrates into region before merging .
2 . for systems like b1534 + 12
, inspiral results in short orbital periods , and high orbital velocities , within region , and such pulsars have been selected against .
3 . as the pulsars in these systems age ,
their radio luminosities probably decay , making them more difficult to detect ( van den heuvel & lorimer 1996 ) .
evidently , any systems in are selected against in pulsar surveys , or few live long enough to reach or spend sufficient time in the region .
apparently , no new systems like b1913 + 16 are being formed far from the spin - up line .
the other known pulsars in the vicinity of region are consistent with this hypothesis . the 0.8 @xmath22 white - dwarf companion to psr b0655 + 64 , the `` failed ns - ns '' system , has cooling age @xmath173 gyr ( hansen & phinney 1998 ) , which suggests the pulsar emerged from the accretion phase near the spin - up line with @xmath174 ms . also , j2235 + 1506 and two objects with @xmath175 ms are solitary and so
could not have escaped detection through merger or orbital acceleration .
the suggestion of camilo , nice & taylor ( 1993 ) that j2235 + 1506 is the result of a disrupted binary is consistent with this view .
the deficit of pulsars in region supports our use of the spin - up line as a zero - age starting point , and is consistent with the existence of a population of ns - ns systems with periods shorter than 8 hours which has remained undetected because of sensitivity limits and observational selection against binary pulsars with short orbital periods and massive companions .
pulsars should already be known in region , periods 530 ms and @xmath176 s - s@xmath1 , where they would have spin - down luminosities 1.52 orders of magnitude higher than , say , j0621 + 1002 .
we assert that the evolutionary path for producing radio pulsars in region is disallowed or rarely followed .
a clue to the absence of pulsars in may lie in the apparent bimodality in the spin parameters of the massive binaries ( fig . 1 ) .
the `` b1913 '' family ( b1913 + 16 , b1534 + 12 , b2127 + 11c , and b0655 + 64 ) consists of pulsars in short - period orbits with ns ( or white - dwarf ) companions ; the pulsars had @xmath177 ms at the end of the spin - up phase .
the `` j0621 '' family ( j0621 + 1002 , j1518 + 4904 , j2145@xmath1780750 , and j1022 + 1001 ) consists of pulsars in @xmath179-day orbits with massive white - dwarf ( or ns ) companions . the division between the two families is strikingly clean the systems with discrepant companion types ( b0655 + 64 and j1518 + 4904 ) nonetheless have orbital periods consistent with their respective family members .
the shorter average spin period of the j0621 family implies that spin - up in the wide - orbit systems must have been at least as effective as in the close ns - ns binaries , but produced final magnetic fields smaller by more than an order of magnitude .
these facts suggest that different accretion and spin - up processes were responsible : we believe that the b1913 family was spun up through disk accretion and the j0621 family predominantly during a wind - driven accretion phase ( e.g. , van den heuvel 1994 presents a spin - up scenario for j2145@xmath1780750 involving wind accretion ) .
it is generally believed that disk accretion drives monotonic spin - up ( until the spin period @xmath180 ) once the donor star overflows its roche lobe , whereas wind accretion produces a non - monotonic random walk in the angular momentum imparted to the ns . to achieve the same final spin period ,
then , much more mass must be transferred , at lower accretion rates , in wind accretion , resulting in smaller magnetic fields for the wide - orbit binaries ; this is consistent with viable models of field attenuation during accretion ( konar & bhattacharya 1998 ) .
( wijers 1998 refutes the hypothesis that field loss scales with accreted mass , but without considering non - monotonic spinup . )
although recent observations show that disk accretion is also not entirely monotonic ( bildsten 1997 ; nelson 1997 ) , and certain numerical simulations of wind accretion question the existence of angular momentum `` flip - flops '' ( e.g. , font & ibanez 1998 ) , frequent or deep torque reversals during spin - up in wide binaries can account for the two distinct families of massive binary pulsars without the need for abandoning field decay that scales with accreted mass .
given the proximity of region to the spin - up line and the implied high magnetic fields relative to j0621 + 1002 , our arguments suggest that the missing ns undergo accretion in close systems .
we speculate , then , that the initial conditions ( orbital separation and companion type ) needed to produce pulsars in region lead instead to accretion ending with collapse of the ns to a black hole or merger with its companion in a common - envelope phase . in detail , the outcome would depend on the maximum mass of neutron stars and the efficiency of the accretion ( see , e.g. , brown 1995 and fryer , benz , & herant 1996 ) . detecting pulsars in region
should also be made difficult by the same pulse - smearing effects that would prevent detection of pulsars like b1913 + 16 in binary orbits with periods smaller than about four hours ( johnston & kulkarni 1991 ; section 4 above ) .
the minimum detectable spin period for a given orbital period is @xmath181 , so it is unlikely that pulsars spinning faster than @xmath182 ms could be detected in binaries with @xmath183 hrs .
conversely , pulsars with @xmath182 ms could only be found in binaries with orbital periods longer than about 8 hours .
this means that recycled pulsars in binaries like those that contain b1534 + 12 , b1913 + 16 , and b2127 + 11c could remain undiscovered throughout region .
moreover , the absence of pulsars in somewhat longer period binaries ( e.g. @xmath1841 day ) in region suggests that there is a fairly sharp upper cutoff to the distribution of periods of close binary pulsar systems , the extended b1913 family .
this supports the idea that the much longer period binaries in the j0621 family have a completely different physical origin .
the explanation commonly put forth for millisecond pulsars with @xmath10 exceeding a hubble time is that these objects were born with spin periods close to their present values : spin - up endpoints are assumed to adhere to equation ( [ peqline.eq ] ) , but with very low accretion rates , @xmath185 ( e.g. , hansen & phinney 1998 ) .
we suggest two additional ways in which spin - up endpoints far from the canonical spin - up line may be produced : ( 1 ) for short - period binaries ( @xmath186 d ) , pulsars emerge on one of the @xmath187 spin - up lines if accretion ceases before the equilibrium period is reached ; and ( 2 ) for long - period binaries , wind accretion results in low - field radio pulsars with as yet undetermined initial periods .
the ages of such systems are therefore uncertain .
it would seem that near - eddington disk accretion given sufficient time to reach equilibrium rotation and ending in a pulsar is rare , as is indicated by the paucity of radio pulsars near the spin - up line for @xmath188 ms .
in addition , if prolonged mass transfer with minimal spin - up during the wind accretion phase is an important part of the evolution of wide - orbit millisecond pulsars , as we suggest , the masses of these objects may be considerably higher than previously believed .
future high - precision mass measurements for such systems may therefore provide important constraints on the maximum allowed neutron star mass .
by estimating the detectable lifetimes of psrs b1534 + 12 and b1913 + 16 , we have revised previous estimates of the galactic birthrate of close double neutron star binaries . in steady state , this would also be the coalescence rate for these relatively short - lived binaries .
we find , adopting the galactic `` scale factors '' of curran & lorimer ( 1995 ) , that the birthrate of such systems , uncorrected for a radio pulsar luminosity function and beaming , is @xmath189 yr@xmath1 . our own estimate of the birthrate , independent of curran & lorimer s ( 1995 ) methodology ,
is based on a maximum likelihood analysis that clarifies the assumptions entering into all birthrate estimates , including our own .
we find , given available knowledge of the ages , systemic velocities , and orbital properties of the two known close ns - ns systems in the galactic disk , along with an estimate of the galactic volume surveyed to date , that the birthrate ( and hence merger rate , for a steady - state distribution ) is @xmath190 yr@xmath1 , where @xmath2 is the beaming factor , and @xmath191 represents the degree to which the two known objects are typical of the entire population of ns - ns binaries .
important parameters entering into this estimate , e.g. , the lower cutoff of the luminosity function for recycled pulsars , are sufficiently uncertain , however , that the true rate is probably higher by an order of magnitude or more ; we therefore consider our result to be a realistic lower bound on the true rate .
furthermore , we find that kinematic and spin - down age constraints for psrs b1534 + 12 and b1913 + 16 allow both systems to be unexpectedly young if we dispose of prior prejudices about the detectable lifetimes of these systems , their birthrates could be substantially higher , of order @xmath192 yr@xmath1 . we have revised the upper bound on the ns - ns binary birthrate proposed by bailes ( 1996 ) based on the absence of any known non - recycled pulsar in a ns - ns binary , finding that the potentially large birthrates allowed by our analysis are consistent with a revised upper bound at @xmath5 yr@xmath1 .
we therefore bracket the merger rate of ns - ns binaries to the range @xmath0
yr@xmath193 yr@xmath1 . our considerations of @xmath194 for ns - ns binaries suggest that searches for such binaries out of the galactic plane will be more productive than those in the plane ; independent of how many are found , high - latitude searches will allow determination of the thickness of the population in @xmath66 . | from considerations of spin evolution and kinematics in the galactic potential , we argue that the pulsars b1913 + 16 , b1534 + 12 , and b2127 + 11c may be younger than previously assumed , and we find that a lower bound on the formation and merger rate of close double neutron star binaries is @xmath0 yr@xmath1 , where @xmath2 is the beaming fraction and @xmath3 accounts for the possibility that the known ns - ns binaries are atypical of the underlying population ( e.g. , if most such binaries are born with shorter orbital periods ) . if we assume no prior knowledge of the detectable lifetimes of such systems , the rate could be as large as @xmath4 yr@xmath1 . from both plausible bounds on @xmath2 and @xmath3 , and a revision of the independently derived limit proposed by bailes ( 1996 )
, we argue that a firm upper bound is @xmath5 yr@xmath1 .
we also present a unifying empirical overview of the spin - up of massive binary pulsars based on their distribution in spin period @xmath6 and spin - down rate @xmath7 , finding evidence for two distinct spin - up processes , one dominated by disk accretion , the other by wind accretion .
we argue that the positions of binary pulsars in the @xmath6-@xmath7 diagram can be understood if ( 1 ) there exists a galactic population of pulsars in double neutron star systems with combinations of spin and orbital periods that have prevented their detection in surveys to date ; ( 2 ) recycled pulsars in wide - orbit binaries are not born near the canonical spin - up line in the @xmath6-@xmath7 diagram because they were predominantly spun up through wind accretion ; and ( 3 ) there exists a disfavored evolutionary endpoint for radio pulsars with spin periods 530 ms and @xmath8 s - s@xmath1 . |
the extraction of ckm matrix elements from experimental data requires the calculation of qcd matrix elements of operators involving the bottom quark @xcite . for a number of years
the milc collaboration has been doing a systematic study of the @xmath0 decay constant @xcite .
the difficulty of simulating the bottom quark , with its mass larger than current feasible inverse lattice spacings , was overcome by interpolating between the results of simulations using wilson quarks with masses around the charm mass and those from static simulations ( infinite mass ) .
unfortunately , for some of the simulations the @xmath3 results were not useable .
these results came as a by - product of the hopping parameter expansion of the heavy quark , and for technical reasons had significant contamination by higher momentum intermediate states when the physical volume was large @xcite .
in addition , it is well known that the poor signal to noise ratio of the static simulations makes it important to use an efficient smearing method . to provide @xmath3 results on all lattices , and to reduce the errors even in the cases where the results were previously available
, we have started a set of static - light simulations using a variational smearing method .
the computational cost of our current implementation of the fft on parallel machines , prohibited the use of a more sophisticated smearing technique such as most @xcite , so we use the following basis of smearing functions : @xmath4 the parameters @xmath5 , @xmath6 , @xmath7 and @xmath8 were obtained from uncorrelated fits to the kentucky group s measured wave functions of static - light mesons ( at @xmath9 ) @xcite .
the @xmath5 , @xmath6 , @xmath7 and @xmath8 parameters were scaled to the appropriate value for a given @xmath10 , using the estimates of the lattice spacing .
because of uncertainties in the lattice spacing , and to have the flexibility to choose different smearing functions for each kappa value , two additional sets of the parameters were chosen .
thus a variational smearing matrix of order ten ( including the local operator ) is used in the simulations .
the static quark is smeared _ relative _ to the light quark using standard fft methods .
.simulation parameters [ cols="^,^,^,^,^",options="header " , ] chirally extrapolating the @xmath11 results to @xmath12 and converting the results to the one loop rg invariant @xmath13 parameter , we obtain the preliminary value of @xmath14 ( statistical error only ) .
this result is consistent with the values obtained from quenched simulations using wilson fermions @xcite @xcite .
to calculate the form factors of the semi - leptonic decays of the @xmath6 meson , we want to follow a strategy similar to the one used in the @xmath0 simulations , except that we will combine the results of heavy quark effective field theory ( hqet ) with the analogous propagating quark calculations , to allow the final results to be interpolated to the @xmath6 meson mass . as a `` warm up exercise ''
we studied the two point function of a hqet - light meson .
this exercise allows us to investigate the smearing of the quarks in the @xmath6 meson and to study the nonperturbative renormalization of the velocity ( a peculiarity of lattice hqet ) both of which are important prerequisites to the calculation of form factors .
we have implemented the hqet propagator equation introduced by mandula and ogilvie @xcite .
the two - point function for an hqet - light meson at finite residual momentum @xmath15 and bare velocity @xmath16 is @xmath17 when there is no excited state contamination , the correlator has the form @xmath18 where @xmath19 and @xmath20 are the smeared and local matrix elements .
the dispersion relation for an hqet - light meson is @xcite @xmath21 where @xmath22 is equal to the energy obtained in the static simulation . here
@xmath23 is the renormalized velocity . as a pilot study we generated 20 correlators at @xmath24 with a light
quark kappa value of 0.160 .
so far , a single exponential source , as in the static simulations , was used .
[ fig : hqet_ps ] shows the effective mass plots for a hqet - light meson at zero residual momentum with bare velocities only in the @xmath25 direction of 0.1 , 0.5 , and 0.8 ; the static effective mass plot is also included ( equivalent to zero velocity ) .
the single exponential produces almost usable plateaus for all three velocities the plateaus should improve when we use variational smearing . the signal to noise ratio does not decrease rapidly with increasing velocity .
the renormalization of the velocity is caused by the breaking of the lorentz symmetry by the lattice .
this renormalization is required in the calculation of form factors such as the isgur - wise function @xcite . our preferred way to extract the renormalization is to use the dispersion relation in eq ( [ eq : hqetdisp ] ) with the energy taken from simulations .
this approach seems to us @xcite @xcite to be simpler than the one used by mandula and ogilvie @xcite .
an estimate of the velocity renormalization can be obtained from the data in fig . [
fig : hqet_ps ] , using naive fits to the effective masses and using eq ( [ eq : hqetdisp ] ) with zero residual momentum . for the bare velocity of 0.5 ( 0.8 ) in the @xmath25 direction , the renormalized velocity was approximately eighty ( sixty five ) percent of the bare velocity . with higher statistics and better smearing a more sophisticated analysis will be done .
the perturbative calculation of mandula and ogilvie @xcite also gave a renormalized velocity smaller than the bare velocity .
this work is supported in part by the u.s .
department of energy and the nsf we would like to thank terry draper for discussions on hqet and @xmath2 , and joe christensen for providing us with the coefficients for the @xmath2 parameter calculation .
the runs are being done on the cornell theory center s sp2 , and on the 512 node paragon at the oak ridge center for computational science . ,
_ _ nucl .
. suppl . _ ) _ * 42 * , 388 ( 1995 ) ; ( milc collaboration ) c. bernard _
_ , _ _ nucl
. phys .
( _ proc .
_ ) _ * 47 * , 459 ( 1996 ) ; milc collaboration ( presented by c. bernard ) , these proceedings . | as part of a larger project to estimate the @xmath0 decay constant , we are recalculating @xmath1 using a variational smearing method in an effort to improve accuracy .
preliminary results for the static @xmath2 parameter and hqet two point functions are also presented . |
one of the important question in particle physics is the origin of neutrino masses and mixing . in type ii seesaw model @xcite , higgs triplet coupling to lepton doublet can generate neutrino mass terms once it develops non - trivial vacuum expectation value .
a distinct feature of this model is the presence of doubly charged higgs bosons which can give a very clean signal at collider experiments @xcite@xcite .
moreover it can also explain the deviation in higgs - to - diphoton rate as measured by cms and atlas @xcite . in this work @xcite
, we study a novel signature of same - sign tetra - leptons ( ss4l ) allowed in some parameter space of the type ii seesaw model that has not been studied so far .
when the mass splitting among triplet components is sizable and the doubly charged higgs boson is the lightest , the electroweak gauge interaction allows a fast decay of the neutral or singly charged component of the higgs triplet into the lighter singly or doubly charged component .
therefore , pair - produced higgs triplet components can end up with a pair of same - sign doubly charged higgs bosons leading to same - sign tetra - leptons if their leptonic yukawa coupling is larger than the ratio of the triplet and doublet higgs vacuum expectation values .
the ss4l final state , which is almost background free , provides an excellent new channel to test the model and probe sizes of the higgs triplet vacuum expectation value and the mass splitting among the higgs triplet components at the lhc .
in this model , the higgs sector of the standard model is extended with a @xmath0 @xmath1 scalar triplet @xmath2 in addition to a sm - higgs doublet @xmath3 .
the leptonic part of the lagrangian which generates neutrino masses is @xmath4 and the scalar potential is @xmath5 ^ 2+\lambda_3\mbox{det}(\delta^\dagger\delta ) + \lambda_4(\phi^\dagger\phi)\mbox{tr}(\delta^\dagger\delta)\\ & + & \lambda_5(\phi^\dagger\tau_i\phi)\mbox{tr}(\delta^\dagger\tau_i\delta ) + \left[\frac{1}{\sqrt{2}}\mu(\phi^ti\tau_2\delta\phi)+\mbox{h.c.}\right].\end{aligned}\ ] ] after the electroweak symmetry breaking with @xmath6 , the @xmath7 term in eq .
( [ pot ] ) gives rise to the vacuum expectation value of the triplet @xmath8 where @xmath9 with @xmath7 being real and positive .
there are seven physical massive scalar eigenstates denoted by @xmath10 , @xmath11 , @xmath12 , @xmath13 , @xmath14 . under the condition that @xmath15 where @xmath16 , the mixing among doublet and triplet components is too small and thus the first five states are mainly from the triplet and the last from the doublet .
we write the masses of the higgs bosons as @xmath17 the mass of @xmath14 is given by @xmath18 as usual .
the mass splitting among triplet scalars can be approximated as @xmath19 furthermore , the sign of the coupling @xmath20 determines two mass hierarchies among the triplet components : @xmath21 for @xmath22 ; or @xmath23 for @xmath24 . in this work ,
we focus on the scenario , where the doubly charged scalar @xmath25 is the lightest and thus it decays only to @xmath26 or @xmath27 , while @xmath28 ( @xmath11 ) decays mainly to @xmath29 ( @xmath30 ) unless the mass splitting @xmath31 is negligibly small . for more details ,
see , e.g. , ref .
@xcite .
the mass splitting @xmath32 between @xmath12 and @xmath13 plays a crucial role for ss4l signal .
the @xmath7 term in eq .
( [ pot ] ) , which is lepton number violating , generates not only the triplet vev @xmath33 but also @xmath34 : @xmath35 for a preferable choice of @xmath33 , @xmath32 can be comparable to the total decay rate of the neutral scalars , @xmath36 , enhancing the ss4l signal .
depending on the triplet mass splitting @xmath31 and the triplet vacuum expectation value @xmath33 , the triplet scalars have different decay properties .
when @xmath37 is the lightest , the possible decay channels for the triplet scalars are shown in table .
[ decay ] .
the decays of triplet scalars can be classified into three modes , a ) di - leptonic mode : which is proportional to yukawa coupling @xmath38 , b ) quark / di - boson mode : controlled by @xmath39 , and c ) gauge decay : due to @xmath40 gauge interaction which dominates if allowed kinematically .
.[decay]possible decay channels for the triplet higgs bosons for @xmath41 . [ cols="<,<,<,<",options="header " , ] in fig .
[ invmass ] , we plot the reconstructed @xmath25 mass from the sample of selected ss4l events for both nh and ih neutrino mass scenarios at lhc14 with 100 fb@xmath42 integrated luminosity .
the peaks in all plots correspond to @xmath43 decays while the broad part ( off - peak ) of distribution correspond to @xmath44 decays .
we point out that a remarkable signal of ss4l , @xmath45 , is allowed in the type ii seesaw mechanism .
such a signal at the lhc strongly depends on the mass splitting @xmath31 and the triplet vev @xmath46 when the doubly charged component @xmath25 is the lightest , larger @xmath31 allows more efficient gauge decay of the neutral component to the singly charged one and then to the doubly charged one .
thus , a pair production of the triplet components at colliders can end up with producing @xmath47 whose branching fraction to ss4l becomes larger for smaller @xmath33 .
another crucial ingredient for increasing the ss4l signal number is the @xmath12-@xmath13 mixing parameter @xmath48 which becomes smaller for smaller @xmath33 and larger @xmath31 .
therefore , there appear optimal values of the model parameters which maximize the same - sign tetra - lepton signal .
after studying such a behavior in the @xmath49 plane , we identified a benchmark point with @xmath50 gev and @xmath51 gev for maximized the signal numbers at the lhc . after making the typical selection cuts to identify four same - sign leptons
, we have shown that one can obtain sizable event numbers for @xmath52 gev with integrated luminosity of 100 fb@xmath42 at the lhc14 to reconstruct the doubly charged higgs mass in both cases of the normal and inverted neutrino mass hierarchy .
99 # 1#2#3phys .
* b#1 * ( # 2 ) # 3 # 1#2#3nucl .
* b#1 * ( # 2 ) # 3 # 1#2#3phys .
* d#1 * ( # 2 ) # 3 # 1#2#3phys .
# 1 * ( # 2 ) # 3 # 1#2#3mod . phys .
* a#1 * ( # 2 ) # 3 # 1#2#3phys . rep .
* # 1 * ( # 2 ) # 3 # 1#2#3science * # 1 * ( # 2 ) # 3 # 1#2#3astrophys .
j. * # 1 * ( # 2 ) # 3 # 1#2#3eur .
j. * c#1 * ( # 2 ) # 3 # 1#2#3jhep * # 1 * ( # 2 ) # 3 # 1#2#3prog .
# 1 * ( # 2 ) # 3 m. magg and c. wetterich , phys .
b * 94 * ( 1980 ) 61 ; t. p. cheng and l. -f .
li , phys .
d * 22 * ( 1980 ) 2860 ; j.f .
gunion , j. grifols , a. mendez , b. kayser and f. olness , ; r. vega and d. dicus , ; j.f .
gunion , r. vega and j. wudka , ; r. godbole , b. mukhopadhyaya and m. nowakowski , ; k. cheung , r. phillips and a. pilaftsis , ; k. huitu , j. maalampi , a. pietila and m. raidal , ; t.g .
rizzo , ; n. lepore , b. thorndyke , h. nadeau and d. london , ; b. dion _ et .
al . _ , ; a. datta and a. raychaudhuri , ; e. ma , m. raidal and u. sarkar , ; ; a. g. akeroyd and m. aoki , phys . rev .
d * 72 * ( 2005 ) 035011 ; t. han , b. mukhopadhyaya , z. si and k. wang , phys .
d * 76 * ( 2007 ) 075013 .
m. muhlleitner and m. spira , phys .
d * 68 * , 117701 ( 2003 ) [ hep - ph/0305288 ] .
s. kanemura and k. yagyu , phys .
d * 85 * , 115009 ( 2012 ) , a. g. akeroyd and s. moretti , phys .
d * 86 * , 035015 ( 2012 ) , e. j. chun , h. m. lee and p. sharma , jhep * 1211 * , 106 ( 2012 ) , e. j. chun and p. sharma . arxiv:1301.1437 [ hep - ph ] .
e. j. chun and p. sharma , jhep * 1208 * , 162 ( 2012 ) [ arxiv:1206.6278 [ hep - ph ] ] .
e. j. chun , k. y. lee and s. c. park , phys .
b * 566 * ( 2003 ) 142 [ hep - ph/0304069 ] .
a. g. akeroyd and h. sugiyama , phys . rev .
d * 84 * ( 2011 ) 035010 [ arxiv:1105.2209 [ hep - ph ] ] .
a. g. akeroyd , s. moretti and h. sugiyama , phys .
d * 85 * ( 2012 ) 055026 [ arxiv:1201.5047 [ hep - ph ] ] . | in type ii seesaw model of neutrino mass generation , we study a remarkable signal of same - sign tetra - lepton ( ss4l ) signal at the large hadron collider .
when doubly charged higgs is lightest , the heavier singly charged or neutral higgs boson produces a doubly charged higgs boson through its fast gauge decay .
this leads to a novel signature of same - sign tetra - leptons resulting from a pair production of same - sign doubly charged higgs bosons .
we study production cross section for the ss4l signal in parameter space of the mass splitting among triplet components and the triplet vacuum expectation value at the lhc . |
in an ideal nematic liquid crystal the molecules are on average aligned along one common direction @xmath0 @xcite .
there is usually some deformation of the alignment and the vector @xmath1 depends on the position @xmath2 in the liquid crystal .
the free - energy density due to the distortion of the vector @xmath1 is expressed in terms of the vector derivatives and the elastic constants @xmath3 @xcite @xmath4^{2 } + { \frac { 1}{2 } } k_{3 } [ \vec{n } \times ( \nabla \times \vec{n } ) ] ^{2},\ ] ] where @xmath5 , @xmath6 , and @xmath7 are splay , twist , and bend elastic constants , respectively . in a deformed uniaxial nematic liquid crystal
, there should appear in many cases a spontaneous dielectric polarization invented by meyer @xcite @xmath8 where @xmath9 and @xmath10 are the splay and the bend flexoelectric ( fe ) coefficients , respectively .
the appearance of the spontaneous polarization in liquid crystals as a result of orientational deformations is called the flexoelectric effect . the fe coefficients and the elastic constants are macroscopic parameters that are important in practical applications of liquid crystals .
thus , it is interesting to study the connection between microscopic and macroscopic parameters in order to fasilitate new material production . in this paper
we will focus on the fe effect .
a microscopic mechanism of the fe effect was proposed by meyer @xcite , who showed that it is a steric effect due to the asymmetry of the molecular shape .
but prost and marcerou @xcite noticed that the polarization in a deformed liquid crystal can be also produced as a result of a gradient in the average density of quadrupole molecule moments . in 1976
straley developed a microscopic theory @xcite by extending the onsager theory of orientational ordering in hard - rod gases .
other mean - field theory was given by derzhanski and petrov @xcite .
the density - functional theory was derived by singh and singh @xcite , were the dipole and the quadrupole contributions to the fe coefficients were taken into account .
somoza and tarazona @xcite tried to include the effects of the relaxation in the angular distribution function of the deformed nematic .
as far as computer simulations are concerned , we would like to mention the papers by stelzer _ et al .
_ @xcite and by billeter and pelcovits @xcite , where pear - shaped molecules were studied . in this paper
, we present two systems with wedge - shaped and banana - shaped molecules . in our calculations
, we use a simple potential energy of interactions which allows us to obtain the macroscpic parameters of the systems and to understand the fe effect origin in the deformed phases .
the microscopic free energy for the system can be derived in the thermodynamic limit ( @xmath11 , @xmath12 , @xmath13 ) from the born - bogoliubov - green - kirkwood - yvon ( bbgky ) hierarchy @xcite or as the cluster expansion for the uniaxial systems @xcite .
the total free energy @xmath14 consists of the entropy term and the interaction term , namely @xmath15 where @xmath16 - 1 \ } , \\
\beta f_{int } & = & - { \frac { 1}{2 } } \int { d\vec{r}_1}{dr_1}{d\vec{r}_2}{dr_2 } g(\vec{r}_1,r_1 ) g(\vec{r}_2,r_2 ) f_{12}.\end{aligned}\ ] ] here @xmath17 is the mayer function , @xmath18 is the potential energy of interactions , @xmath19 , @xmath20 , and @xmath21 is related to the ideal gas properties .
the normalization of the one - particle distribution function @xmath22 is @xmath23 the equilibrium distribution @xmath22 minimizing the free energy of eq .
( [ ftotal ] ) satisfies the equation @xmath24 - \int { d\vec{r}_2}{dr_2 } g(\vec{r}_2,r_2 ) f_{12 } = \mbox{const}.\ ] ] in the case of the homogeneous uniaxial nematic phase , the distribution function depends only on the molecule orientation @xmath25 .
we will further assume that @xmath22 depends on a single argument @xmath26 , where @xmath1 determines the phase orientation .
this is satisfied exactly for the molecules with @xmath27 symmetry but we expect that it is plausible for the banana - shaped molecules with @xmath28 symmetry . the molecule electric dipole moment is defined as @xmath29 where the unit vectors @xmath30 describe the molecule orientation . in the case of the wedge - shaped molecules ,
the fe coefficients have the form @xcite @xmath31 where @xmath32 determines the molecule @xmath27 symmetry axis . in the case of the banana - shaped molecules ,
the fe coefficients can be written in the approximated form provided that @xmath33 determines the molecule @xmath28 symmetry axis and @xmath32 determines the long molecule axis @xcite @xmath34 in both cases , the fe coefficients are proportional to the dipole moment component parallel to the molecule symmetry axis .
it results from the symmetries of the interactions and of the mayer function .
the elastic constants of the uniaxial nematic phase are @xcite @xmath35
the calculations were performed for the square - well potential energy of the form @xmath36 where @xmath37 is the distance between molecules , @xmath38 , and @xmath39 depends on the molecule orientations and on the vector @xmath40 . for @xmath39
, we used simple expressions with three parameters , @xmath41 defines the length scale , @xmath42 defines the nematic term , and @xmath43 defines the fe term . in the case of the wedge - shaped molecules @xmath39 has the form @xmath44 + \sigma_{2 } ( \vec{\delta } \cdot \vec{n}_1 - \vec{\delta } \cdot \vec{n}_2),\ ] ] whereas for the banana - shaped molecules @xmath45
+ \sigma_{2 } ( \vec{\delta } \cdot \vec{l}_1 - \vec{\delta } \cdot \vec{l}_2).\ ] ] we used the density @xmath46 , @xmath47 , @xmath48 .
the molecule volume @xmath49 was estimated from the mutually excluded volume .
the fe coefficients were expressed in @xmath50 , the elastic constants in @xmath51 , and the temperature in @xmath52 .
the potential energy of the form @xmath53 allows us to express the fe coefficients and the elastic constants as a finite series of the order parameters where the coefficients of the expansion can be calculated analitically . as a consequence
it is easy to change the set of parameters @xmath54 and to study different systems .
the temperature dependence of the order parameters and the elastic constants is similar to the known results @xcite , therefore the fe coefficients will be discussed only . in the system of the wedge - shaped molecules , on decreasing the temperature
we meet the first order transition from the isotropic to the uniaxial nematic phase at @xmath55 .
the temperature dependence of the fe coefficients is presented in fig .
the physical picture is shown in fig .
[ fig2 ] . on decreasing the temperature ,
the bend coefficient @xmath10 is almost constant while the splay coefficient @xmath9 increases monotonically .
the wedge - shaped molecules fit to the splayed structure and there is an excess of molecules pointing to the splay origin .
the induced polarization has the opposite direction . in the bend structure
the fe polarization points to the bend centre because the molecules are slightly turned in the opposite direction . in the system of the banana - shaped molecules , on decreasing the temperature
we meet the first order transition from the isotropic to the uniaxial nematic phase at @xmath56 .
the temperature dependence of the fe coefficients is presented in fig .
the physical picture is shown in fig .
the banana - shaped molecules fit to the bend structure and the fe polarization points to the bend centre . in the splayed structure
the molecules are slightly turned and the average polarization is opposite to the splay centre position .
in this paper , the statistical theory was used to study the dependence between the microscopic and macroscopic parameters of nematic liquid crystals .
the temperature dependence of the order parameters , the fe coefficients , and the elastic constants was obtained for the two systems . for the wedge - shaped molecules the bend coefficient @xmath10 is almost constant whereas the splay coefficient @xmath9 is changing monotonically . for the banana - shaped molecules the splay coefficient @xmath9 is almost constant whereas the bend coefficient @xmath10 is changing monotonically . in both cases
the fe coefficients are proportional to the molecule dipole moment component , parallel to the molecule symmetry axis .
this is the long axis of the wedge - shaped molecules and the short axis of the banana - shaped molecules .
it results from the symmetries of the interactions and of the mayer function .
in other theories , where the interactions are not explicitly used or they are more complicated , the dependence on the all dipole moment components is present .
the results are consistent with the microscopic pictures from the paper by meyer but two additional situations were taken into account , the splayed phase with banana - shaped molecules and the bend phase with wedge - shaped molecules . at present stage , the qualitative comparison between the theory and the experiment is very difficult .
the experimental data on the fe coefficients are still scarce and sometimes contradictory @xcite . on the other hand , a nontrivial dependence on the details of the chemical structure
was shown @xcite .
a small chemical modification of the molecule can generate significant change of the flexoelectricity .
thus , we would like to give only the estimates assuming typical values of the model parameters .
if we assume the molecular length @xmath57 , the interaction energy @xmath58 , and the electric dipole moment @xmath59 , then we can estimate the values of the fe coefficients @xmath60 , the elastic constants @xmath61 , and the temperature @xmath62 ( it is three / four times larger then the typical isotropic - nematic transition temperature ) . the theory should be improved in order to describe exactly real systems . for the distance @xmath39 , the general expansion proposed by blum and torruella @xcite can be applied in order to describe molecular shapes better .
the meyer function @xmath63 used in the low density limit can be replaced with the direct correlation function @xmath64 from the percus - yevick approximation . in many cases ,
liquid crystals are mixtures of different substances or different conformers and it should be taken into account . finally , it is possible to consider the fe effect in the biaxial nematic phase although this phase is still very rare .
this work was supported by the faculty of physics , astronomy and applied computer science , jagellonian university ( wrbw no .
the author is also grateful to j. spaek for his support .
a. g. petrov , _ measurements and interpretation of flexoelectricity _ , in _ physical properties of liquid crystals : nematics _ , pp .
251264 , edited by d. dunmur , a. fukuda , and g. luckhurst , inspec , the institution of electrical engineers , london , 2001 . the temperature dependence of the flexoelectric coefficients @xmath65 expressed in @xmath66 in the case of wedge - shaped molecules .
the temperature @xmath67 is expressed in @xmath52 .
the @xmath68 transition is at @xmath69 . ]
the homogeneous , the splay , and the bend phases of wedge - shaped molecules .
the solid lines denote the local direction @xmath1 .
the @xmath70 axis is the symmetry axis of the homogeneous nematic phase . in the splay phase
there is an excess of molecules pointing down to the splay origin . in the bend phase
there is an excess of molecules pointing to the left whereas the bend center is on the right . ] the temperature dependence of the flexoelectric coefficients @xmath65 expressed in @xmath71 in the case of banana - shaped molecules .
the temperature @xmath67 is expressed in @xmath52 .
the @xmath68 transition is at @xmath72 . ]
the homogeneous , the splay , and the bend phases of banana - shaped molecules .
the solid lines denote the local direction @xmath1 .
the @xmath70 axis is the symmetry axis of the homogeneous nematic phase . in the splay phase ,
molecules are slightly turned and the average polarization direction is opposite to the splay origin . in the bend phase , the molecule dipole moments point to the bend center on the right . ] | the statistical theory of the dipole flexoelectric ( fe ) polarization in liquid crystals is used to calculate the temperature dependence of order parameters , the elastic constants and the fe coefficients .
two systems with polar wedge - shaped and banana - shaped molecules are investigated . in both cases
the fe coefficients are proportional to the dipole moment component parallel to the molecule symmetry axis .
the origin of the fe effect and microscopic pictures of the distorted phases are discussed . *
keywords : * nematic liquid crystals , flexoelectric polarization , elastic constants , statistical theory . |
semileptonic decay of charm meson is important for studying strong and weak interactions .
it can be used to test techniques developed for solving perturbative and nonperturbative problems in quantum chromodynamics ( qcd ) , and to extract elements of cabibbo - kobayashi - maskawa ( ckm ) matrix .
semileptonic decay is simpler than hadronic decay of charm meson because leptons do not involve strong interaction .
the amplitude of semileptonic decay can be decomposed into several transition form factors due to lorentz property of the hadronic matrix element .
the form factors include all the nonperturbative effects .
several methods can be used to treat these problems , such as quark model , qcd sum rule and lattice qcd , among which , qcd sum rule and lattice qcd are based on the first principle of qcd .
the method of qcd sum rules @xcite has been widely used in hadronic physics since its establishment in the late 1970s . for semileptonic decays of charm
meson , @xmath9 was firstly studied in qcd sum rule method with three - point correlation function @xcite .
several years later , qcd sum rule method was extended to semileptonic decays of @xmath10 meson , @xmath11 @xcite and @xmath12 @xcite . in these works , form factors @xmath13 and @xmath14 are calculated at the point @xmath15 , where @xmath16 is the momentum transfer squared . for the whole physical region of @xmath17 ,
the form factors are either assumed to be pole dominance @xmath18 , or a linear approximation was used . in refs @xcite , @xmath19 , @xmath20 and @xmath21 , @xmath22 were studied , where qcd sum rule method was extended to a very large value of @xmath16 with a careful treatment of non - landau - type singularities .
@xmath23 decays to @xmath24 and @xmath25 final states were studied in @xcite . in this work
, we study @xmath5 in qcd sum rule method .
this decay mode has been measured in experiment long time before @xcite .
now it is necessary to analyze it theoretically .
we calculate up to contributions of operators of dimension 6 in the operator product expansion ( ope ) and keep the mass of s - quark . in our result ,
the large contributions come from unit operator @xmath26 ( result of perturbative diagram ) and condensate of operators of dimension 3 .
operator of dimension 5 gives smaller contribution .
the contributions of operators of dimension 6 are negligible . when calculating contribution of perturbative diagram and gluon condensate ( operator of dimension 4 ) , cutcosky s rule has been used .
therefore subtraction of continuum contribution is conveniently performed not only for perturbative diagram but also for contribution of gluon condensate .
after some long steps of calculation , we find the contributions of diagrams for the gluon condensate cancel each other , so there is no gluon - condensate contribution in @xmath0 transition .
this is our new finding .
there are two independent borel parameters @xmath27 and @xmath28 in manipulating three - point correlation functions . in general , to simplify the numerical analysis , a fixed ratio of @xmath29 was taken in recent references . in this work
, we make the numerical analysis in the whole region of independent @xmath27 and @xmath28 to select the stability window " , so it is different from taking a fixed ratio of these two borel parameters .
we calculate @xmath30 transition form factors @xmath1 , @xmath2 , @xmath3 , @xmath4 and the branching ratio of @xmath31 .
our result of the ratios of @xmath6 , @xmath7 , @xmath8 ( @xmath32 and @xmath33 denote decay width of @xmath34 to @xmath35 meson in longitudinal and transverse polarization , respectively ) , and the total branching fraction are in agreement with experimental data .
recently , just before this work is finished , we find that @xmath36 was also calculated in ref.@xcite . however , their analysis is very different from ours .
first , by carefully choosing the requirement that the double borel parameters @xmath27 and @xmath28 should not be too large for keeping the continuum contribution small , and at the same time , @xmath27 and @xmath28 should not be too small for keeping the truncated ope series effective , i.e. , keeping the contributions of higher dimension operators small , we get very different stability window " for the borel parameters .
second , our results of the transition form factors are different from theirs . especially for @xmath4 , they got negative value , however , we get positive . using their values of the form factors , although one can get the total branching ratio of @xmath5 to be compatible with experimental result , the ratio of @xmath8 will be too large .
but in our case , @xmath37 which is consistent with world average @xmath38 .
the paper is organized as follows . in section 2
, we briefly introduce the qcd sum rule method used in this work .
section 3 is the calculation .
section 4 is the numerical analysis and discussion .
section 5 is devoted to the summary .
to calculate the transition form factors of semileptonic @xmath23 meson decays , the standard procedure in qcd sum rule method is to consider the three - point correlation function defined as @xmath39 with the currents having the same quantum numbers as the relevant mesonic states under consideration , which are defined by : 1 ) the current of @xmath23 channel , @xmath40 ; 2 ) the current of weak transition : @xmath41 ; 3 ) the current of @xmath35 channel : @xmath42 . on one hand , inserting a complete set of intermediate hadronic states into the correlation function , and using the double dispersion relation , one can express the correlation function in terms of a set of hadronic states , @xmath43 with @xmath44 where @xmath45 and @xmath46 denote the complete set of hadronic states of @xmath35 and @xmath23 channels , respectively . @xmath47 and @xmath48 are the four - momentum of @xmath45 and @xmath46 states , @xmath49 , @xmath50 , and @xmath51 . integrate over @xmath52 and @xmath53 in eq .
( [ c2 ] ) , we can obtain @xmath54 separate the ground states of @xmath23 and @xmath35 channels apparently , the above equation becomes @xmath55 the weak transition matrix element @xmath30 can be decomposed as @xmath56 \varepsilon^*\cdot q \frac{a_2(q^2)}{m_{d_s}+m_{\phi } } \nonumber\\ & & -i\frac{2m_\phi \varepsilon^*\cdot q}{q^2}q_\mu a_0(q^2),\nonumber\end{aligned}\ ] ] where @xmath51 .
the vacuum - to - meson transition amplitudes can be parameterized through defining the corresponding decay constants , @xmath57 finally the correlation function can be expressed in terms of meson decay constants and @xmath58 transition matrix element , @xmath59 on the other hand , the correlation function of eq .
( [ correlator ] ) can be evaluated at negative values of @xmath60 and @xmath61 by the operator - product expansion in qcd , in which the time - ordered current operators in eq .
( [ correlator ] ) is expanded in terms of a series of local operators with increasing dimensions , @xmath62 where @xmath63 s are wilson coefficients , @xmath26 is the unit operator , @xmath64 is the local fermion field operator of light quarks , @xmath65 is gluon strength tensor , @xmath66 and @xmath67 are the matrices appearing in the procedure of calculating the wilson coefficients .
sandwich the left and right hand sides of eq .
( [ opef ] ) between two vacuum states , we get the correlation function in terms of wilson coefficients and condensates of local operators , @xmath68 for later convenience , we shall reexpress the above equation . in general
, it can be expressed in terms of six independent lorentz structures @xmath69 each @xmath70 includes perturbative and condensate contributions @xmath71 where @xmath72 , @xmath73 , @xmath74 are contributions of condensates of dimension 3 , 4 , 5 , 6 , @xmath73 in eq .
( [ conden ] ) . in next section
we can see that perturbative contribution and gluon condensate contribution can be finally written in the from of dispersion integration , @xmath75 we approximate the contribution of higher resonances and continuum states as integrations over some thresholds @xmath76 and @xmath77 in the above equations .
then equate the two representations of the correlation function in eq .
( [ c3 ] ) and ( [ c4 ] ) , we can get an equation for the form factors . to improve such equation ,
we make borel transformation over @xmath60 and @xmath61 in both sides , which can further suppress higher resonance contribution .
the definition of borel transformation to any function @xmath78 is @xmath79 some examples of borel transformation is given in the following , @xmath80 equating the two representations of the correlation function , subtracting the higher resonances and continuum contribution , and performing borel transformation in both variables @xmath60 and @xmath61 , we finally obtain the sum rules for the form factors , @xmath81 , \nonumber \end{aligned}\ ] ] where @xmath82 denotes borel transforming @xmath70 in both variables @xmath60 and @xmath61 , @xmath83 and @xmath84 are borel parameters .
because we have subtracted the higher resonance and continuum contribution , now the dispersion integration for perturbative and gluon condensate contribution should be performed under the threshold , @xmath85 in the next section , we will explain the technique of calculating the wilson coefficients and given the resulted form of the sum rules for the form factors .
in this work , we first calculate the wilson coefficients in the operator - product expansion @xcite , then extract the relevant terms @xmath70 s for the sum rules of the form factors in eq .
( [ formsr ] ) . we will not present the result of each wilson coefficient here because their forms are very tedious .
we only give the results of the form factors according to the contribution of each condensate .
the diagram for the perturbative contribution is depicted in fig.1 . the leading order in @xmath86 expansion
is considered here .
this contribution amounts to wilson coefficient @xmath87 in ope representation of the correlation function in eq .
( [ conden ] ) .
we can write down this amplitude ( see fig.1 ) , ( 150,150)(-15,-25.98 ) ( -15,-25.98)(0,0)3 ( 0,0)(40,69.28 ) ( 0,0)(40,69.28)(40,69.28)(80,0)(80,0)(0,0 ) ( 80,0)(95,-25.98)3(40,69.28)(40,94.28)14 ( 41,70)@xmath88 ( 80.5,0.5)@xmath89 ( -5.5,0.5)@xmath90 ( -24,-16)@xmath91 ( 95,-16)@xmath92 ( 10,34)@xmath93 ( 38,-10)@xmath94 ( 64,34)@xmath94 @xmath95 . \label{pert1}\end{aligned}\ ] ] the diagrams for the contributions of @xmath107 and @xmath108 are shown in fig .
2 . the contribution of fig .
2(b ) is zero after double borel transformation in both variables @xmath60 and @xmath61 because only one variable appears in the denominator @xmath109 .
so we will not consider fig .
2(b ) in the following . the contribution of fig .
2(a ) to the correlation function is ( 300,150)(-15,-25.98 ) ( -15,-25.98)(0,0)3 ( 0,0)(40,69.28 ) ( 0,0)(40,69.28)(40,69.28)(80,0)(80,0)(50,0)(30,0)(0,0 ) ( 80,0)(95,-25.98)3(40,69.28)(40,94.28)14(47,-2.7)@xmath110 ( 26.8,-2.7)@xmath110(41,70)@xmath88(80.5,0.5)@xmath89(-5.5,0.5)@xmath90 ( 185,-25.98)(200,0)3 ( 200,0)(240,69.28 ) ( 200,0)(240,69.28)(240,69.28)(255,43.3 ) ( 265,25.98)(280,0 ) ( 280,0)(200,0 ) ( 280,0)(295,-25.98)3(240,69.28)(240,94.28)14 ( 250.2,41.6)@xmath110(261,23.8)@xmath110 ( 241,70)@xmath88(280.5,0.5)@xmath89(194.5,0.5)@xmath90 ( 33,-40)@xmath111(233,-40)@xmath112 @xmath113 where @xmath114 and @xmath115 are the propagators of @xmath94 and @xmath93 quarks , respectively .
move the quark field operators @xmath116 and @xmath117 together , we get @xmath118_{\alpha\beta } , \label{pi2a}\ ] ] where @xmath119 and @xmath120 are dirac spinor indices .
the matrix element @xmath121 and @xmath122 using the technique explained in @xcite , @xmath123 , \label{qq } \end{aligned}\ ] ] @xmath124 and @xmath125 in the above are the color indices , @xmath126 is the quark mass , and the ellipsis stands for terms of higher orders in @xmath89 and @xmath90 expansion . from eq .
( [ qq ] ) we know that fig.2(a ) contributes to the coefficients of quark condensate @xmath127 , mixed quark - gluon condensate @xmath128 and the four - quark condensate @xmath129
. substitute eq .
( [ qq ] ) into ( [ pi2a ] ) and integrate over the coordinates @xmath89 and @xmath90 , we can obtain explicitly the coefficients of these condensates contributed by fig.2(a ) .
the diagrams for the contribution of bi - gluon operator are depicted in fig.3 .
they are calculated in the fixed - point gauge , in which the gauge fixing condition is taken to be @xmath131 @xcite .
then the external gauge field can be expressed directly in terms of the color field strength tensor @xcite , @xmath132 which is expanded to the first order to be , @xmath133 in the following calculation , it is convenient to transform @xmath134 to the momentum space , @xmath135 ( 300,200)(-15,-25.98 ) ( -11.25,100.515)(0,120)3 ( 0,120)(30,171.96 ) ( 0,120)(30,171.96)(30,171.96)(60,120)(60,120)(0,120 ) ( 60,120)(71.25,100.515)3(30,171.96)(30,190.71)14 ( 0,120)(30,40,60)22(0,120)(30,0,20)22 ( 19.88,137,25)@xmath110(23.5,129)@xmath110 ( 88.75,100.515)(100,120)3 ( 100,120)(130,171.96 ) ( 100,120)(130,171.96)(130,171.96)(160,120)(160,120)(100,120 ) ( 160,120)(171.25,100.515)3(130,171.96)(130,190.71)14 ( 130,171.96)(30,240,260)22(130,171.96)(30,280,300)22 ( 120.4,138.8)@xmath110(131.3,138.8)@xmath110 ( 188.75,100.515)(200,120)3 ( 200,120)(230,171.96 ) ( 200,120)(230,171.96)(230,171.96)(260,120)(260,120)(200,120 ) ( 260,120)(271.25,100.515)3(230,171.96)(230,190.71)14 ( 260,120)(30,160,180)22(260,120)(30,120,140)22 ( 232.3,136)@xmath110(228.125,128)@xmath110 ( -11.25,-19.485)(0,0)3 ( 0,0)(30,51.96 ) ( 0,0)(30,51.96)(30,51.96)(60,0)(60,0)(0,0 ) ( 60,0)(71.25,-19.485)3(30,51.96)(30,70.71)14 ( 11.25,19.485)(-1.74,26.985)22(18.75,32.475)(5.76,39.975)22 ( -5.625,23.5)@xmath110(1.65,36.5)@xmath110 ( 88.75,-19.485)(100,0)3 ( 100,0)(130,51.96 ) ( 100,0)(130,51.96)(130,51.96)(160,0)(160,0)(100,0 ) ( 160,0)(171.25,-19.485)3(130,51.96)(130,70.71)14 ( 122.5,0)(122.5,-15)22(137.5,0)(137.5,-15)22 ( 119.875,-16.5)@xmath110(135.25,-16.5)@xmath110 ( 188.75,-19.485)(200,0)3 ( 200,0)(230,51.96 ) ( 200,0)(230,51.96)(230,51.96)(260,0)(260,0)(200,0 ) ( 260,0)(271.25,-19.485)3(230,51.96)(230,70.71)14 ( 248.75,19.485)(261.74,26.985)22(241.25,32.475)(254.24,39.975)22 ( 256.5,24.75)@xmath110(250,37.5)@xmath110 ( 23,90)(a ) ( 125,90)(b ) ( 225,90)(c ) ( 23,-35)(d ) ( 125,-35)(e ) ( 225,-35)(f ) then the amplitude can be written down in the momentum space by following the standard feynman rule . again , as what we did in the previous subsections , we move the gluon strength tensor operator together : @xmath136 .
then using the following decomposition to obtain the bi - gluon condensate , @xmath137 in which @xmath138 is the abbreviation of @xmath139 . in the evaluation of the diagrams of fig.3
some types of loop integrals encountered are treated at first by derivatives with respect to the quark masses , then transform them to dispersion integrals by using cutkosky s rule and with help of @xmath26 , @xmath140 and @xmath141 functions given previously .
for instance , @xmath142 ^ 2[(k+p_1)^2-m_1 ^ 2]^2}\nonumber\\ & = & \frac{\partial}{\partial m_2
^ 2}\frac{\partial}{\partial m_1 ^ 2 } \int d^4 k \frac{k_\mu k_\nu}{(k^2-m^2)[(k+p_2)^2-m_2 ^ 2][(k+p_1)^2-m_1 ^ 2]}\nonumber\\ & = & -2\pi i\frac{\partial}{\partial m_2 ^ 2}\frac{\partial}{\partial m_1
^ 2 } \int ds_1 ds_2 \frac{i_{\mu\nu}}{(s_1-p_1 ^ 2)(s_2-p_2 ^ 2)},\end{aligned}\ ] ] and @xmath143 ^ 2[(k+p_1)^2-m_1 ^ 2]^2}\nonumber\\ & = & \frac{\partial}{\partial m_2 ^ 2}\frac{\partial}{\partial m_1 ^ 2 } \int d^4 k \frac{k_\mu k_\nu k\cdot p_2}{(k^2-m^2)^2[(k+p_2)^2-m_2 ^ 2][(k+p_1)^2-m_1 ^ 2]}\nonumber\\ & = & -2\pi i\frac{\partial}{\partial m^2}\frac{\partial } { \partial m_2 ^
2}\frac{\partial}{\partial m_1 ^ 2 } \int ds_1 ds_2 \frac{-\frac{1}{2}(s_2-m_2 ^ 2+m^2)i_{\mu\nu}}{(s_1-p_1 ^ 2)(s_2-p_2 ^ 2)},\end{aligned}\ ] ] where the term @xmath144 comes from the @xmath98 functions @xmath145 $ ] with the substitution @xmath106 when using the cutkosky s rule .
after some long steps of calculation , we finally find that the contributions of the diagrams ( a ) to ( f ) in fig.3 cancel each other .
therefore there are no contributions of gluon condensate in @xmath146 transition .
the diagrams for quark - gluon mixing and four - quark contributions are depicted in fig.4 and fig.5 , respectively .
the techniques are similar to that explained in previous subsections .
we only give some different points here .
( 400,100)(-15,-25.98 ) ( -11.25,-19.485)(0,0)3 ( 0,0)(30,51.96 ) ( 0,0)(30,51.96)(30,51.96)(60,0)(60,0)(37.5,0)(22.5,0)(0,0 ) ( 60,0)(71.25,-19.485)3(30,51.96)(30,70.71)14(35.25,-2.025)@xmath110 ( 19,-2.025)@xmath110 ( 0,0)(30,20,60)23(25.5,10)@xmath110 ( 88.75,-19.485)(100,0)3 ( 100,0)(130,51.96 ) ( 100,0)(130,51.96)(130,51.96)(160,0)(160,0)(137.5,0)(122.5,0)(100,0 )
( 160,0)(171.25,-19.485)3(130,51.96)(130,70.71)14(135,-2.025)@xmath110 ( 119,-2.025)@xmath110 ( 160,0)(30,120,160)23(127.5,8.5)@xmath110 ( 188.75,-19.485)(200,0)3 ( 200,0)(230,51.96 ) ( 200,0)(230,51.96)(230,51.96)(241.25,32.475 ) ( 248.75,19.485)(260,0 ) ( 260,0)(200,0 ) ( 260,0)(271.25,-19.485)3(230,51.96)(230,70.71)14 ( 236.5,31.2)@xmath110(245,17)@xmath110 ( 260,0)(30,140,180)23(233.5,17.5)@xmath110 ( 288.75,-19.485)(300,0)3 ( 300,0)(330,51.96 ) ( 300,0)(330,51.96)(330,51.96)(341.25,32.475 ) ( 348.75,19.485)(360,0 ) ( 360,0)(300,0 ) ( 360,0)(371.25,-19.485)3(330,51.96)(330,70.71)14 ( 336.5,31.2)@xmath110(345,17)@xmath110 ( 330,51.96)(30,240,280)23(332.25,19.5)@xmath110 ( 22,-28)(a ) ( 122,-28)(b ) ( 222,-28)(c ) ( 322,-28)(d ) ( 300,200)(-15,-25.98 ) ( -11.25,100.515)(0,120)3 ( 0,120)(30,171.96 ) ( 0,120)(30,171.96)(60,120)(37.5,120)(22.5,120)(0,120 ) ( 60,120)(71.25,100.515)3(30,171.96)(30,190.71)14 ( 33.3,117.975)@xmath110 ( 17.5,117.975)@xmath110 ( 30,171.96)(41.25,152.475 ) ( 48.75,139.485)(60,120 ) ( 36.5,151.2)@xmath110(45,137)@xmath110 ( 15,147.6)(48,120)1.68 ( 88.75,100.515)(100,120)3 ( 100,120)(130,171.96 ) ( 100,120)(130,171.96)(160,120)(137.5,120)(122.5,120)(100,120 ) ( 160,120)(171.25,100.515)3(130,171.96)(130,190.71)14 ( 133.3,117.975)@xmath110 ( 117.5,117.975)@xmath110 ( 130,171.96)(141.25,152.475 ) ( 148.75,139.485)(160,120 ) ( 136.5,151.2)@xmath110(145,137)@xmath110 ( 115,147.6)(154.375,128.8)1.68 ( 188.75,100.515)(200,120)3 ( 200,120)(230,171.96 ) ( 200,120)(230,171.96)(260,120)(237.5,120)(222.5,120)(200,120 ) ( 260,120)(271.25,100.515)3(230,171.96)(230,190.71)14 ( 233.3,117.975)@xmath110 ( 217.5,117.975)@xmath110 ( 230,171.96)(241.25,152.475 ) ( 248.75,139.485)(260,120 ) ( 236.5,151.2)@xmath110(245,137)@xmath110 ( 230,120)(15,0,180)1.68 ( -11.25,-19.485)(0,0)3 ( 0,0)(30,51.96 ) ( 0,0)(30,51.96)(60,0)(37.5,0)(22.5,0)(0,0 ) ( 60,0)(71.25,-19.485)3(30,51.96)(30,70.71)14 ( 33.3,-2.5)@xmath110 ( 18,-2.5)@xmath110 ( 30,51.96)(41.25,32.475 ) ( 48.75,19.485)(60,0 ) ( 36.5,31.2)@xmath110(45,17)@xmath110 ( 11.25,0)(54.375,9.675)1.68 ( 88.75,-19.485)(100,0)3 ( 100,0)(130,51.96 ) ( 100,0)(130,51.96)(160,0)(137.5,0)(122.5,0)(100,0 ) ( 160,0)(171.25,-19.485)3(130,51.96)(130,70.71)14 ( 133.3,-2.5)@xmath110 ( 117.5,-2.5)@xmath110 ( 130,51.96)(141.25,32.475 ) ( 148.75,19.485)(160,0 ) ( 136.5,31.2)@xmath110(145,17)@xmath110 ( 134.125,42.75)(145,0)1.68 ( 188.75,-19.485)(200,0)3 ( 200,0)(230,51.96 ) ( 200,0)(230,51.96)(260,0)(237.5,0)(222.5,0)(200,0 ) ( 260,0)(271.25,-19.485)3(230,51.96)(230,70.71)14 ( 233.3,-2.5)@xmath110 ( 217.5,-2.5)@xmath110 ( 230,51.96)(241.25,32.475 ) ( 248.75,19.485)(260,0 ) ( 236.5,31.2)@xmath110(245,17)@xmath110 ( 245,29.7)(15,126,293)1.68 ( 23,90)(a ) ( 125,90)(b ) ( 225,90)(c ) ( 23,-28)(d ) ( 125,-28)(e ) ( 225,-28)(f ) the vacuum average of non - local quark - gluon mixing operator @xmath147 is calculated to be @xmath148_{\beta\alpha}t^a_{ji},\end{aligned}\ ] ] where @xmath149 and @xmath150 are the abbreviations of @xmath151 and @xmath152 respectively .
@xmath153 is the strong coupling . because we calculate up to the condensate of dimension - six operators , the external gluon field @xmath154 in fig.4
should be expanded up to the second term , which will contribute a dimension - six operator , @xmath155 where @xmath156 is the covariant derivative in the adjoint representation , @xmath157 .
then another vacuum matrix element needed is @xcite @xmath158 we calculate these diagrams and find that the contributions of fig.4(c ) , ( d ) and fig.5(c ) , ( d ) vanish after double borel transformation in two variables @xmath60 and @xmath61 , because only one variable appearing in the denominator , for instance , @xmath159 .
the borel transformation in @xmath60 will kill such terms . following the above method , after some tedious algebraic derivation with the software mathematica
, we obtain the coefficients @xmath160 , @xmath161 , @xmath162 and @xmath163 needed in eq .
( [ formsr ] ) .
they are listed in appendix .
the borel parameters @xmath83 and @xmath84 are not physical parameters
. the physical result should not depend on them if the operator product expansion can be calculated up to infinite order .
however , ope has to be truncated to some finite orders in practice .
therefore , borel parameters have to be selected in some windows " to get the best stability of the physical results . the requirement to select the stable windows "
is : the borel parameters can not be too large , or , contributions of higher resonance and continuum states can not be effectively suppressed ; at the same time , they should not be too small , or , the truncated ope would fail because the series in ope generally depend on borel parameters in the denominator @xmath170 .
we find the optimal stability with the requirements shown in table [ t1 ] and the thresholds @xmath76 , @xmath77 in the ranges @xmath171 , @xmath172 .
the regions of borel parameters which satisfies the requirements of table [ t1 ] are shown in fig.[region ] in two - dimensional diagram of @xmath27 and @xmath28 .
we find good stability of the form factors within these regions .
.requirements to select borel parameters @xmath27 and @xmath28 for each form factors @xmath173 , @xmath174 , @xmath175 and @xmath176 [ cols="^,^,^,^ " , ] [ t4 ] we use the total decay width of @xmath23 meson @xmath177 @xcite to obtain the branching ratio of @xmath5 , our result is @xmath178 which is in good agreement with experimental data @xmath179 .
we calculate the transition form factors for @xmath58 transition in the region @xmath180 in qcd sum rule , where no non - landau - type singularity occurs . then fit the result from qcd sum rule in this region of momentum transfer , and extrapolate it to the whole physical region in the decay @xmath181 .
we treat the two borel parameters @xmath27 and @xmath28 as independent parameters , and select the allowed region for @xmath27 and @xmath28 by requiring that the higher resonance and continuum contributions in @xmath23 and @xmath35 channels are not large , at the same time requiring that the condensate of higher dimension operators do not contribute too much .
we find good stability for the transition form factors @xmath1 , @xmath2 , @xmath3 and @xmath4 in the relevant two - dimensional regions of @xmath27 and @xmath28 .
we obtain the results of the transition form factors @xmath1 , @xmath2 , @xmath3 and @xmath4 in these regions of @xmath27 and @xmath28 .
our result of the ratios of these form factors @xmath182 and @xmath183 are well consistent with experimental data .
we studied the process @xmath184 with the form factors calculated from qcd sum rule . for the transverse polarization state of the final @xmath35 meson ,
the rate of @xmath23 decaying to right - hand state is almost an order smaller than decaying to left - hand state .
the ratio of @xmath8 and the branching ratio of @xmath184 are in good agreement with the experimental data within the error bars of both the present experimental data and theoretical calculation .
borel transformed coefficients of perturbative and nonperturbative contributions to the transition form factors in eq .
( [ formsr ] ) are given here .
the contributions of condensate of dimension - six operator @xmath185 are numerically negligible , whereas their expressions are more tedious , therefore we do not present all of them here .
@xmath186 with @xmath187 & + s_1-s_2+q^2)m_c + m_s(-s_2 ^ 2 + 2m_s^2s_2+s_1s_2+q^2s_2+\lambda ) ] , \end{array } \eqno(a1)\ ] ] where @xmath188 .
the lower integration limit @xmath189 is determined by the condition that all internal quarks are on their mass shell @xcite , @xmath190 @xmath191 & + q^2)m_s^4m_2 ^ 2-(m_1 ^ 2+m_2 ^ 2)m_s^6)m_1 ^ 4+m_2 ^ 2(m_1 ^ 2+m_2 ^ 2)m_c^2m_s^2(3m_2 ^ 2\\[4 mm ] & -m_s^2)m_1 ^ 2+m_2 ^ 2m_cm_s(-3m_1 ^ 2m_2
^ 4 + ( m_1 ^ 2+m_2 ^ 2+q^2)m_s^2m_2 ^ 2\\[4 mm ] & -(m_1 ^ 2+m_2 ^ 2)m_s^4)m_1 ^ 2 -m_2 ^ 4(m_1 ^
2+m_2 ^ 2)m_c^3m_s^3 ] \times\langle \bar{s}s\rangle\ ; , \end{array } \eqno(a2)\ ] ] @xmath192 & -(3m_2 ^ 4 + 5m_1 ^ 2m_2 ^ 2)m_s^2)m_1 ^ 4-m_2 ^ 2(m_1 ^ 2+m_2 ^ 2)m_c^2(3m_2 ^ 2-m_s^2)m_1 ^ 2\\[4 mm ] & + m_2 ^ 2m_cm_s(2m_2 ^ 4 - 2m_1 ^ 2m_2 ^ 2+(m_1 ^ 2+m_2 ^ 2)m_s^2)m_1 ^ 2+m_2 ^ 4(m_1 ^ 2+m_2 ^ 2)m_c^3m_s\\[4 mm ] & + q^2(m_1 ^ 4m_2 ^ 2(3m_2 ^ 2-m_s^2)-m_1 ^ 2m_2 ^ 4m_cm_s)]\times g\langle \bar{s}\sigma tg s \rangle\ ; , \end{array } \eqno(a3)\ ] ] @xmath193 & + ( m_1 ^ 2+m_2 ^ 2)m_c^5m_s^3m_2 ^ 4+m_1 ^ 2(m_1 ^ 2+m_2 ^ 2)m_c^4m_s^4m_2 ^ 2\\[4 mm ] & + m_1 ^ 2q^4m_s^3(m_sm_1 ^ 2+m_2 ^ 2m_c)m_2 ^ 2+m_1 ^ 2m_c^3m_s^3(-13m_2 ^ 4 + 2m_1 ^ 2m_2 ^ 2\\[4 mm ] & + ( m_1 ^ 2+m_2 ^ 2)m_s^2)m_2 ^ 2+m_1 ^ 4m_c^2(-54(-1+e^{m_s^2/m_2 ^ 2})m_1 ^ 2m_2 ^ 6\\[4 mm ] & + 54m_1 ^ 2m_s^2m_2 ^ 4-(m_1 ^ 2 + 10m_2 ^ 2)m_s^4m_2 ^ 2+(m_1 ^ 2+m_2 ^ 2)m_s^6)\\[4 mm ] & + q^2(54(-1+e^{m_s^2/m_2 ^ 2})m_2 ^ 6m_1 ^ 6 - 54m_2 ^ 4m_s^2m_1 ^ 6-(m_1 ^ 2+m_2 ^ 2)m_s^6m_1 ^ 4\\[4 mm ] & -m_2 ^ 2(m_1 ^ 2+m_2 ^ 2)m_cm_s^5m_1 ^ 2+m_2 ^ 2(m_1 ^ 4 + 10m_2 ^ 2m_1 ^ 2-(2m_1 ^ 2\\[4 mm ] & + m_2 ^ 2)m_c^2)m_s^4m_1 ^ 2-m_2 ^ 2m_c(2m_1 ^ 4 - 13m_2
^ 2m_1 ^ 2+(2m_1 ^ 2\\[4 mm ] & + m_2 ^ 2)m_c^2)m_s^3 ) ] \times g^2\langle \bar{s}s\rangle^2\ ; . \end{array } \eqno(a4)\ ] ] @xmath194 with @xmath195 & + s_2+q^2)m_sm_c^4 + 2s_2(-4s_1 ^ 2 + 8s_2s_1 - 4s_2 ^ 2 + 2q^4 - 6(s_1-s_2)m_s^2\\[4 mm ] & + \lambda+2q^2(3m_s^2+s_1+s_2))m_c^3 + 2s_2m_s[4s_1 ^ 2 - 8s_2s_1 + 4s_2 ^ 2 - 2q^4\\[4 mm ] & + 6(s_1-s_2)m_s^2 - 3\lambda-2q^2(3m_s^2+s_1+s_2)]m_c^2+[6(s_1\\[4 mm ] & -s_2)s_2m_s^4 + 2(4s_2 ^ 3 - 8s_1s_2
^ 2 + 4s_1 ^ 2s_2 - 2\lambda s_2+s_1\lambda)m_s^2+(s_1\\[4 mm ] & -s_2)s_2(2s_1 ^ 2 - 4s_2s_1 + 2s_2 ^ 2-\lambda)-2s_2q^4(2m_s^2 + 2s_1\\[4 mm ] & + s_2)+q^2(-6s_2m_s^4 - 2(2s_2 ^ 2 + 2s_1s_2+\lambda)m_s^2+s_2(2s_1 ^ 2 - 6s_2s_1\\[4 mm ] & + 4s_2 ^ 2-\lambda))]m_c+m_s[2s_2 ^ 4 - 6s_1s_2 ^ 3 + 6s_1 ^ 2s_2 ^ 2 - 3\lambda s_2 ^ 2 + 6(s_2\\[4 mm ] &
-s_1)m_s^4s_2-s_1 ^ 3s_2 + 3s_1\lambda s_2 + 2q^4(2m_s^2 + 2s_1+s_2)s_2+\lambda^2\\[4 mm ] & -2(4s_2 ^ 3 - 8s_1s_2 ^ 2 + 4s_1 ^ 2s_2 - 4\lambda s_2+s_1\lambda ) m_s^2+q^2(6s_2m_s^4 + 2(2s_2 ^ 2\\[4 mm ] & + 2s_1s_2+\lambda ) m_s^2+s_2(-2s_1 ^ 2 + 6s_2s_1 - 4s_2 ^ 2 + 3\lambda ) ) ] \}\ ; , \end{array } \eqno(a5)\ ] ] @xmath196 & + m_2 ^ 2)+q^2)m_s^4m_2 ^ 2-(m_1 ^ 2+m_2 ^ 2)m_s^6]m_1 ^ 4+m_2 ^ 2(m_1 ^ 2+m_2 ^ 2)m_c^2m_s^2(3m_2 ^ 2\\[4 mm ] & -m_s^2)m_1 ^ 2-m_2 ^ 2m_cm_s(3m_1 ^ 2m_2 ^ 4+(m_1 ^ 2+m_2 ^ 2-q^2)m_s^2m_2 ^ 2\\[4 mm ] & + ( m_1 ^ 2+m_2 ^ 2)m_s^4)m_1 ^ 2-m_2 ^ 4(m_1 ^ 2+m_2 ^ 2)m_c^3m_s^3 \ }
\times\langle \bar{s}s\rangle\ ; , \end{array } \eqno(a6)\ ] ] @xmath197 & + 5m_1 ^ 2m_2 ^ 2)m_s^2]m_1 ^ 4 + m_2 ^ 2(m_1 ^ 2+m_2 ^ 2)m_c^2(3m_2 ^ 2-m_s^2)m_1 ^ 2-m_2 ^ 2m_cm_s\\[4 mm ] & [ 2(m_1 ^ 2 + 2m_2 ^ 2)m_2 ^ 2+(m_1 ^ 2+m_2 ^ 2)m_s^2 ] m_1 ^ 2-m_2 ^ 4(m_1 ^ 2+m_2 ^ 2)m_c^3m_s+q^2[m_2 ^ 2(m_s^2\\[4 mm ] & -3m_2 ^ 2)m_1 ^ 4 + m_2 ^ 4m_cm_sm_1 ^ 2]\}\times g\langle \bar{s}\sigma tg s \rangle\ ; , \end{array } \eqno(a8)\ ] ] @xmath198 & -2m_1 ^
2)m_c^5m_s^3m_2 ^ 4 + m_1 ^ 2(m_1 ^ 2+m_2 ^ 2)m_c^4m_s^4m_2 ^ 2+m_1 ^ 2q^4m_s^3(m_1 ^ 2m_s\\[4 mm ] & -2m_2 ^ 2m_c)m_2 ^ 2 + m_1 ^ 2m_c^3m_s^3[-26m_2 ^ 4 + 4m_1 ^ 2m_2 ^ 2+(m_1 - 2m_2 ^ 2)m_s^2]\\[4 mm ] & m_2 ^ 2 + m_1 ^ 4m_c^2[-54(-1+e^{m_s^2/m_2 ^ 2})m_1 ^ 2m_2 ^ 6 + 54m_1 ^ 2m_s^2m_2 ^ 4
-4(m_1 ^ 2\\[4 mm ] & -2m_2 ^ 2)m_s^4m_2 ^ 2+(m_1 ^ 2+m_2 ^ 2)m_s^6 ] + q^2[54(-1+e^{m_s^2/m_2 ^ 2})m_1 ^ 6m_2 ^ 6\\[4 mm ] & -54m_2 ^ 4m_s^2m_1 ^ 6 -(m_1 ^ 2+m_2 ^ 2)m_s^6m_1 ^ 4-m_2 ^ 2(m_1 ^ 2 - 2m_2 ^ 2)m_cm_s^5m_1 ^ 2\\[4 mm ] & + m_2 ^
2(4(m_1 ^ 4 - 2m_1 ^ 2m_2 ^ 2)-(2m_1 ^ 2+m_2 ^ 2)m_c^2)m_s^4m_1 ^ 2 + m_2 ^ 4m_c\\[4 mm ] & ( -4m_1 ^ 4 + 26m_2 ^ 2m_1 ^ 2+(4m_1 ^ 2 - 7m_2 ^ 2)m_c^2)m_s^3]\ } \times g^2\langle \bar{s}s\rangle^2\ ; . \end{array } \eqno(a9)\ ] ] @xmath199 with @xmath200 & + 3s_2-q^2)m_sm_c^4 + 2s_2[-4s_1 ^ 2 - 4s_2s_1 + 8s_2 ^ 2 + 2q^4 - 6(s_1 +
3s_2)\\[4 mm ] & m_s^2+\lambda + 2q^2(3m_s^2+s_1 - 5s_2)]m_c^3 + 2s_2m_s[4s_1 ^ 2 + 4s_2s_1 - 8s_2 ^ 2\\[4 mm ] & -2q^4 + 6(s_1 + 3s_2)m_s^2+\lambda -2q^2(3m_s^2+s_1 - 5s_2)]m_c^2 + [ 6s_2(s_1\\[4 mm ] & + 3s_2)m_s^4 + 2(s_1 + 2s_2)(-4s_2 ^ 2 + 4s_1s_2+\lambda ) m_s^2 + ( s_1-s_2)s_2(2s_1 ^ 2\\[4 mm ] & -2s_2 ^ 2-\lambda ) + 2s_2 q^4(-2m_s^2 - 2s_1+s_2 ) + q^2(-6s_2m_s^4 - 2(-10s_2 ^ 2\\[4 mm ] & + 2s_1s_2+\lambda ) m_s^2-s_2(-2s_1 ^ 2 - 10s_2s_1 + 4s_2 ^
2+\lambda))]m_c -m_s[2s_2 ^ 4\\[4 mm ] & -2s_1s_2 ^ 3 - 2s_1 ^ 2s_2 ^ 2-\lambda s_2 ^ 2 + 6(s_1 + 3s_2)m_s^4s_2 + 2s_1 ^ 3s_2+s_1\lambda s_2\\[4 mm ] & + 2q^4(-2m_s^2 - 2s_1+s_2)s_2-\lambda^2 + 2(-8s_2 ^ 3 + 4s_1s_2 ^ 2 + 4s_1 ^ 2s_2\\[4 mm ] & + 4\lambda s_2+s_1\lambda ) m_s^2 + q^2(-6s_2m_s^4 - 2(-10s_2 ^ 2 + 2s_1s_2+\lambda ) m_s^2\\[4 mm ] & + s_2(2s_1 ^ 2 + 10s_2s_1 - 4s_2 ^ 2+\lambda ) ) ] \}\ ; , \end{array } \eqno(a10)\ ] ] @xmath201 & + m_2 ^ 2)+q^2)m_s^4m_2 ^ 2-(m_1 ^ 2+m_2 ^ 2)m_s^6]m_1 ^ 4+m_2 ^ 2(m_1 ^ 2+m_2 ^ 2)m_c^2\\[4 mm ] & m_s^2(3m_2 ^ 2-m_s^2)m_1 ^ 2-m_2 ^ 2m_cm_s(3m_1 ^ 2m_2 ^ 4+(m_1 ^ 2 - 3m_2 ^ 2-q^2)m_s^2\\[4 mm ] & m_2 ^ 2+(m_1 ^ 2+m_2 ^ 2)m_s^4)m_1 ^ 2-m_2 ^ 4(m_1 ^ 2+m_2 ^ 2)m_c^3m_s^3 \ }
\times\langle \bar{s}s\rangle\ ; , \end{array } \eqno(a11)\ ] ] @xmath202 & -m_2 ^ 4)m_s^2]m_1 ^ 4 + m_2 ^ 2(m_1 ^ 2+m_2 ^ 2)m_c^2(3m_2 ^ 2-m_s^2)m_1 ^ 2\\[4 mm ] & -m_2 ^ 4(m_1 ^ 2+m_2 ^ 2)m_c^3m_s - m_c[2m_1 ^ 4m_sm_2 ^ 4+m_1 ^ 2(m_1 ^ 2+m_2 ^ 2)m_s^3m_2 ^ 2 ] \\[4 mm ] & + q^2[m_2 ^ 2(m_s^2 - 3m_2 ^ 2)m_1 ^ 4 + m_2 ^ 4m_cm_sm_1 ^ 2]\}\times g\langle \bar{s}\sigma tg s \rangle\ ; , \end{array } \eqno(a13)\ ] ] @xmath203 & -2m_1 ^ 2)m_c^5m_s^3m_2 ^ 4 + m_1 ^ 2(m_1 ^ 2+m_2 ^ 2)m_c^4m_s^4m_2 ^ 2+m_1 ^ 2q^4m_s^3(m_1 ^ 2m_s\\[4 mm ] & -2m_2 ^ 2m_c)m_2 ^ 2 + m_1
^ 2m_c^3m_s^3[-18m_2 ^ 4 + 4m_1 ^ 2m_2 ^ 2+(m_1 - 2m_2 ^ 2)m_s^2]\\[4 mm ] & m_2 ^ 2 + m_1 ^ 4m_c^2[-54(-1+e^{m_s^2/m_2 ^ 2})m_1 ^ 2m_2 ^ 6 + 54m_1 ^ 2m_s^2m_2 ^ 4 -4(m_1 ^ 2\\[4 mm ] & + 4m_2 ^ 2)m_s^4m_2 ^ 2+(m_1 ^ 2+m_2 ^ 2)m_s^6 ] + q^2[54(-1+e^{m_s^2/m_2 ^ 2})m_1 ^ 6m_2 ^ 6\\[4 mm ] & -54m_2 ^ 4m_s^2m_1 ^ 6 -(m_1 ^ 2+m_2 ^ 2)m_s^6m_1 ^ 4-m_2 ^ 2(m_1 ^ 2 - 2m_2 ^ 2)m_cm_s^5m_1 ^ 2\\[4 mm ] & + m_2 ^ 2(4m_1 ^ 2(m_1 ^ 2 + 4m_2 ^ 2)-(2m_1 ^ 2+m_2 ^ 2)m_c^2)m_s^4m_1 ^ 2 + m_2 ^ 4m_c\\[4 mm ] & ( -4m_1 ^ 4 + 18m_2 ^ 2m_1 ^
2+(4m_1 ^ 2 - 7m_2 ^ 2)m_c^2)m_s^3]\ } \times g^2\langle \bar{s}s\rangle^2\ ; . \end{array } \eqno(a14)\ ] ] @xmath204 with @xmath205 & -s_2+q^2)m_c^3 + 2s_2m_s(2m - s^2 + s_1-s_2+q^2)m_c^2+[2s_2m_s^4\\[4 mm ] & + 2(-s_2 ^ 2+s_1s_2+\lambda ) m_s^2 + s_2\lambda + 2s_2q^2 ( m_s^2+s_1)]m_c - m_s[2s_2m_s^4\\[4 mm ] & + 2(-s_2 ^ 2+s_1s_2+\lambda ) m_s^2 -s_1\lambda + q^2(2s_2m_s^2 + 2s_1s_2+\lambda ) ] \ } \ ; , \end{array } \eqno(a15)\ ] ] @xmath206 & [ 3m_1 ^ 2m_2 -(m_1 ^ 2 + 2m_2 ^ 2)m_s^2]m_c^4 + [ 3m_1 ^ 4m_sm_2 ^ 6 - 9m_1 ^ 2(m_1 ^ 2+m_2 ^ 2)\\[4 mm ] & m_s^3m_2 ^ 4+(m_2 ^ 6 + 4m_1 ^ 2m_2 ^ 4 + 3m_1 ^ 4m_2 ^ 2)m_s^5 ] m_c^3+m_1 ^ 2[-6m_1 ^ 4m_2 ^ 6+(m_1 ^ 4\\[4 mm ] & + 4m_2 ^ 2m_1 ^ 2 + 3m_2 ^ 4)m_s^6-(6m_2 ^ 6 + 11m_1 ^ 2m_2 ^ 4 + 5m_1 ^ 4m_2 ^ 2)m_s^4\\[4 mm ] & + 3(3m_1 ^ 2m_2 ^ 6 + 2m_1 ^ 4m_2 ^ 4)m_s^2]m_c^2+m_1 ^ 2m_s[-15m_1 ^ 4m_2 ^ 6 + ( 2m_1 ^ 4\\[4 mm ] & + 3m_2 ^ 2m_1 ^ 2+m_2 ^ 4)m_s^6-(2m_2 ^ 6 + 13m_1 ^
2m_2 ^ 4 + 11m_1 ^ 4m_2 ^ 2)m_s^4\\[4 mm ] & + ( 8m_1 ^ 2m_2 ^ 6 + 11m_1 ^ 4m_2 ^ 4)m_s^2]m_c+m_1 ^ 4m_s^2[3m_1 ^ 2m_2 ^ 6 - 4(m_1 ^ 2+m_2 ^ 2)\\[4 mm ] & m_s^4m_2 ^ 2+(m_1 ^ 2+m_2 ^ 2)m_s^6 + ( 2m_2 ^ 6 + 5m_1 ^ 2m_2 ^ 4)m_s^2]+m_1 ^ 2m_2 ^ 2q^4m_s^2\\[4 mm ] & [ ( m_s^2 - 3m_2 ^ 2)m_1 ^ 2+m_2 ^ 2m_cm_s]+ q^2[(6m_1 ^ 2m_2 ^ 6-(m_1 ^ 2 + 2m_2 ^ 2)m_s^6\\[4 mm ] & + ( 7m_2 ^ 4 + 5m_1 ^ 2m_2 ^ 2)m_s^4 - 3(m_2 ^ 6 + 2m_1 ^ 2m_2 ^ 4)m_s^2)m_1 ^ 4 + m_2 ^ 2m_c^2m_s^2\\[4 mm ] &
( 3m_2 ^ 2(2m_1 ^ 2+m_2 ^ 2)-(2m_1 ^ 2 + 3m_2 ^ 2)m_s^2)m_1 ^ 2+m_2 ^ 2m_cm_s ( -3m_1 ^ 2m_2 ^ 4\\[4 mm ] & -(3m_1 ^ 2 + 2m_2 ^ 2)m_s^4+(2m_2 ^ 4 + 9m_1 ^ 2m_2 ^ 2)m_s^2)m_1 ^ 2-m_2 ^ 4(2m_1 ^ 2+m_2 ^ 2)\\[4 mm ] & m_c^3m_s^3]\ } \times\langle \bar{s}s\rangle\ ; , \end{array }
\eqno(a16)\ ] ] @xmath207 & -(m_1 ^ 2 + 2m_2 ^ 2)m_s^2]m_c^4 + m_2 ^ 2m_s[(3m_1 ^ 4 + 4m_2 ^ 2m_1 ^ 2+m_2 ^ 4)m_s^2\\[4 mm ] & -2m_1 ^ 2m_2 ^ 2(5m_1 ^ 2 + 3m_2 ^ 2)]m_c^3 + [ 2m_1 ^ 4(m_1 ^ 2 + 3m_2 ^ 2)m_2 ^ 4+(m_1 ^ 6\\[4 mm ] & + 4m_2 ^ 2m_1 ^ 4 + 3m_2 ^ 4m_1 ^ 2)m_s^4- 2(3m_2 ^ 2m_1 ^ 6 + 5m_2 ^ 4m_1 ^ 4)m_s^2]m_c^2+m_1 ^ 2m_s\\[4 mm ] & [ 4m_1 ^ 2(m_1 ^ 2 + 3m_2 ^ 2)m_2 ^ 4 + 2(m_1 ^ 4 + 3m_2 ^ 2m_1 ^ 2+m_2 ^ 4)m_s^4+(m_2 ^ 6\\[4 mm ] & -8m_1 ^ 2m_2 ^ 4 - 13m_1 ^ 4m_2 ^ 2)m_s^2]m_c + m_1 ^ 4[4m_1 ^ 2m_2 ^ 6+(m_1 ^ 2+m_2 ^ 2)\\[4 mm ] & m_s^2(2m_2 ^
4+m_s^4)-(m_2 ^ 4 + 5m_1 ^ 2m_2 ^ 2)m_s^4]+q^4[m_2 ^ 2(m_s^2\\[4 mm ] & -3m_2 ^ 2)m_1 ^ 4+m_2 ^ 4m_cm_sm_1 ^ 2]-q^2[(2(m_1 ^ 2 + 3m_2 ^ 2)m_2 ^ 4 + ( m_1 ^ 2 + 2m_2 ^ 2)m_s^4\\[4 mm ] & -2(2m_2 ^ 4 + 3m_1 ^ 2m_2 ^ 2)m_s^2)m_1 ^ 4+m_2 ^ 2m_cm_s(m_2 ^ 4 -10m_1 ^ 2m_2 ^ 2\\[4 mm ] & + ( 3m_1 ^ 2 + 2m_2 ^ 2)m_s^2)m_1 ^ 2+m_2 ^ 2m_c^2((2m_1 ^ 2 + 3m_2 ^ 2)m_s^2\\[4 mm ] & -3(m_2 ^ 4 + 2m_1 ^ 2m_2 ^ 2))m_1 ^ 2+m_2 ^ 4(2m_1 ^ 2+m_2 ^ 2)m_c^3m_s ] \}\times
g\langle \bar{s}\sigma tg s \rangle\ ; . \end{array } \eqno(a18)\ ] ] 99 m.a .
shifman , a.i .
vainshtein and v.i .
zakharov , nucl .
phys . b147 ( 1979 ) 385 , 448 .
aliev , v.l .
eletskii , and ya .
i. kogan , sov .
40 ( 1984 ) 527 .
ovchinnikov and v.a .
slobodenyyuk , z. phys .
c44 ( 1989 ) 433 ; v.n .
baier and grozin , z. phys .
c47 ( 1990 ) 669 .
ovchinnikov , sov . j. nucl
. phys . 50
( 1989 ) 519 .
p. ball , v.m .
braun and h.g .
dosch , phys .
d44 ( 1991 ) 3567 .
p. ball , phys .
d48 ( 1993 ) 3190 .
p. colangelo and f. de fazio , phys .
b 520 ( 2001 ) 78 .
k. kodama et al .
( fnal e653 collab . ) , phys .
b309 ( 1993 ) 483 .
frabetti et al .
( fnal e687 collab . ) , phys .
b328 ( 1994 ) 187 .
p. avery et al .
( cleo collab . )
b337 ( 1994 ) 405 .
aitala et al .
( fnal e791 collab . ) , phys .
b450 ( 1999 ) 294 . i. bediaga and m. nielsen , eprint hep - ph/0304193 .
reinders , h. rubinstein and s. yazahi , phys . rep .
127 , no.1 ( 1985)1 - 97 .
r.e.cutkosky , j. math .
phys . 1 ( 1960 ) 429 .
ioffe and a.v .
smilga , nucl .
b216 ( 1983 ) 373 .
fock , phys .
z. su 12 ( 1937 ) 404 ; j. schwinger , particles , sources , and fields , addison - wesley ( 1973 ) ; m.a .
shifman , nucl .
b173 ( 1980 ) 13 ; m.s .
dubovikov and a.v .
smilga , nucl .
b185 ( 1981 ) 109 .
smilga , yad .
35 ( 1982 ) 473 [ sov
. j. nucl .
35 ( 1982 ) 271 .
shifman , nucl .
phys . b173 ( 1980 ) 13 .
landau , nucl .
( 1959 ) 181 .
s. narison , _ qcd spectral sum rules _ ( world scientific , singapore , 1989 ) . for a recent review article , see , p. colangelo and a. khodjamirian , in : at the frontier of particle physics / handbook of qcd , edited by m. shifman ( world scientific , singapore , 2001 ) , hep - ph/0010175 .
particle data group , k. hagiwara et al .
d66 ( 2002 ) 010001 . | we calculate @xmath0 transition form factors @xmath1 , @xmath2 , @xmath3 and @xmath4 , and study semileptonic decay of @xmath5 based on qcd sum rule method .
we compare our results of the ratios of @xmath6 , @xmath7 , @xmath8 , and the total decay branching ratio of @xmath5 with experimental data , and find that they are consistent .
( 0,0 ) ( 340,230)bihep - th-2003 - 30 ( 340,210 ) * * |
10 url # 1#1urlprefix[2][]#2 gaillard s a , kluge t , flippo k a _ et al .
_ 2011 _ phys .
plasmas _ * 18 * 056710 dhumieres e , lefebvre e , gremillet l _
_ 2005 _ phys . plasmas _
bf 12 062704 ; maksimchuk a , gu s , flippo k _ et al . _ 2000 _ phys .
lett . _ * 84 * 4108 ; wilks s c , langdon a b , cowan t e _ et al .
_ 2001 _ phys . plasmas _ * 8 * 542 | the interaction of ultra - intense lasers with solid foils can be used to accelerate ions to high energies well exceeding 60 mev @xcite .
the non - linear relativistic motion of electrons in the intense laser radiation leads to their acceleration and later to the acceleration of ions .
ions can be accelerated from the front surface , the foil interior region , and the foil rear surface ( tnsa , most widely used ) , or the foil may be accelerated as a whole if sufficiently thin ( rpa ) . + here ,
we focus on the most widely used mechanism for laser ion - acceleration of tnsa . starting from perfectly flat foils we show by simulations how electron filamentation at or inside the solid leads to spatial modulations in the ions .
the exact dynamics depend very sensitively on the chosen initial parameters which has a tremendous effect on electron dynamics . in the case of step - like density gradients
we find evidence that suggests a two - surface - plasmon decay of plasma oscillations triggering a raileigh - taylor - like instability . in laser ion acceleration ,
an ultra intense laser interacts with a solid and quickly ionizes its surface .
the laser light creates an immense pressure on the surface , produces large and fast electron currents into the solid and excites plasma waves . in such a scenario many instabilities develop which can disrupt the surface and break up the electron currents .
+ it has been shown previously that instabilities in _ ultrathin _ solid foils can imprint onto the spatial structure of ions in the regime of relativistically induced transparency target normal sheath acceleration ( rit tnsa @xcite ) or radiation pressure acceleration ( rpa @xcite ) @xcite mostly attributed to the rayleigh - taylor instability ( rti ) @xcite .
instabilities behind the foils can also develop during the ps timescales of the plasma expansion of tnsa @xcite .
here we show that instabilities developing in solid foils can also be of significance for the laser absorption and ion acceleration for thicker foils where no rit occurs .
we present results from particle - in - cell ( pic ) simulations using the code ipicls2d @xcite and refer the reader to our ref .
@xcite for experimental evidence .
we use dimensionless units with @xmath0 and measure distances in @xmath1-direction ( the laser direction ) relative to the initial foil front surface position and times relative to the time when the laser intensity maximum reaches @xmath2 . in most of our analysis
we will focus on two simulations , comparing the cases of a flat foil with exponential preformed plasma on the surface ( simulation a ) and without ( simulation b ) , the full simulation parameters are given in the caption of fig .
[ fig:1 ] .
+ one main finding is that instabilities which are prone to evolve in all of our simulations imprint on the ion density distribution behind the foil slab even for the micron thick foils used .
the instabilities we discuss in this paper are developing in the electrons first since the heavier ions remain almost at rest during the relevant time scales of electron motion .
of course this is not strictly correct since otherwise no rti instability could develop .
fig .
[ fig:1](a ) shows exemplary part of the electron density distribution of simulation a @xmath3 after the laser maximum has reached the foil front surface , and fig .
[ fig:1](b ) shows the corresponding ion density distribution . in this simulation
the electrons pushed into the foil at @xmath4 by the laser lorentz force are prone to a transverse weibel - like instability @xcite .
the @xmath4 electron slabs become filamented transversely and create energetic channels almost normal to the foil surface which are surrounded by strong quasi - static magnetic fields .
this happens during the first few plasma wavelengths @xmath5 as can be seen in fig .
[ fig:3]a .
the ion acceleration at the foil rear surface then follows this structure(cp .
fig , [ fig:1]a , b ) .
it is this that is different to the mechanism described in @xcite : the rear surface ion s acceleration itself is structured transversely _ from the beginning _ since the driving energetic electrons reach the rear surface filamented already in contrast to the relatively slow weibel instability witnessed in @xcite _ during _ the ion acceleration .
of course both can happen concurrently .
we now study the case of step - like density gradient at the front surface ( simulation b ) . in comparison to simulation a , where the bulk electrons remained distributed almost homogeneously transversely at the surface and only energetic electrons became filamented inside the foil , here the foil surface already shows transverse structure both in electron density ( fig .
[ fig:3]b , c ) and energy density ( fig .
[ fig:3]a ) , with the latter not increasing during passage through the foil contrarily to simulation a. the reason for this lies in the density ripples which give rise to filamented injection of electrons , similar to the mechanism described in @xcite .
we therefore conclude that in simulation b another filamentation mechanism must be present .
density ripples usually are created by rti and the density distribution in fig .
[ fig:3]c shows a pattern consistent with the linear stage of rti . yet
, typically the average distance @xmath6 between density maxima ( wave number @xmath7 ) should be given by the smallest transverse scale that could seed the instability , since the growth rate @xmath8 ( @xmath9 is the acceleration of the surface ) is largest for the largest @xmath10 .
hence we are looking for a mechanism that can seed a transverse density fluctuation which is then prone to a rti growth @xcite .
such a mechanism was described e.g. in @xcite where a generation of 2d electron surface oscillations ( 2desos ) by a parametric decay of @xmath4 oscillations of the foil surface in the laser field was found . there from initially purely longitudinal driving oscillations of the surface ( for the @xmath4 oscillations the driver frequency is @xmath11 ) , oscillations in a standing wave along the surface are generated via excitation of upward and downward moving and interfering surface waves with @xmath12 and @xmath13 given by [ eqn .
( 6 ) in @xcite ] . for our simulation conditions
this would correspond to @xmath14 , much smaller than the value extracted from the pic simulation @xmath15 .
this decay of @xmath4 oscillations with the characteristic @xmath16 longitudinal oscillation is observed in our simulation only at later times growing only slowly , which we attribute to larger @xmath10 .
however , considering not the @xmath4 oscillation as a driving source but the plasma oscillations excited by the disturbances we find good numerical agreement .
fig .
[ fig:4 ] shows a comparison of @xmath17 for a number of simulations we performed with different parameters ( varying @xmath18 and @xmath19 ) with the analytical value @xmath10 .
the agreement is generally rather good .
also , we find oscillations at @xmath20 in the spectrum of surface oscillations , fig .
[ fig:4](b ) .
here we defined a density level and recorded the position in x - direction ( laser direction ) where the rising density reaches that value .
we then computed the temporal fourier transform of this position averaged over the positions along the y - direction at the density ripple maxima , neglecting the regions in between .
the result would be a strong signal around the laser harmonics ( overlying all other signals ) which however are expected to be transversely flat here .
therefore we took the oscillation at the `` inner surface '' . by this
we mean the oscillation of the falling density slope behind the region of laser pressure steepened electron density , which can screen the lower laser harmonics .
as expected we find a signal at the plasma frequency @xmath21 computed using the maximum density and temperature at the peak of the laser pressure steepened density profile averaged over the distance of a plasma period .
additionally we also found another pronounced peak at @xmath20 .
since there is still a significant contribution from laser harmonics , we performed another simulation with @xmath22 and @xmath23 ( fig .
[ fig:4](c ) ) . due to the higher density
the skin depth @xmath24 is much shorter , screening laser harmonics , and additionally the plasma frequency and its half value are shifted to higher frequencies and hence are at higher and hence weaker harmonics . here
we again obtain a very clear signal at @xmath25 and @xmath20 . + it is interesting to mention that the development of surface ripples seen in simulation b can be artificially seeded . introducing an initial surface roughness containing a mixture of several spatial frequencies around @xmath26 we first find that the smallest spatial frequencies have the fastest growth , as expected for rti , and that this growth is most dominant when the spatial frequencies are initialized around @xmath10 ( fig .
[ fig:5 ] ) . for yet higher spatial frequencies
the effect of seeding abruptly ceases . + .
color scale is the same as in fig .
[ fig:1 ] .
largest surface ripple growth occurs when the spatial frequency @xmath10 is seeded , with a characteristic mushroom shape ( non - linear rti ) growing at the front ( central panel ) . here ,
@xmath27 , @xmath28.,width=302 ] summarizing , we found by simulations that the laser ion acceleration process in the tnsa regime can be influenced by electron instabilities at the surface or inside the foil .
the exact mechanism and scale of spatial modulations in the beam of hot electrons and ions sensitively depends on the initial parameters . for
the pursuit of higher ion energies at higher laser intensities , e.g. available at ( future ) petawatt laser systems , a deeper understanding and further characterization of laser and target parameter dependent plasma instabilities will be essential .
the work has been partially supported by ec fp7 laserlab - europe / charpac ( contract 284464 ) and by the german federal ministry of education and research ( bmbf ) under contract number 03z1o511 . |
the vmc estimator can be factorized as , @xmath94 where @xmath36 is the number fluctuation contribution , @xmath33 is the sign contribution , and @xmath37 is the magnitude contribution .
the kronecker delta , @xmath95 , is one when copy @xmath96 and @xmath97 have the same number of particles in region @xmath24 and zero otherwise .
+ to more efficiently compute the sign we sample a different probability distribution , the swap magnitude , @xmath98 where @xmath30 represents the electronic spin and spatial coordinates in region @xmath23 for copy @xmath10 , @xmath31 and @xmath32 .
this computes the sign of the swap operator from the regions in @xmath50 , @xmath99 that contribute most to @xmath28 .
the magnitude and particle number contributions decay more slowly than the sign contribution and are sampled over the original distribution , @xmath100 .
+ because the expectation value of the swap operator can be negative when the value is small and the error is large , we are unable to plot the full range of values for @xmath101 and @xmath102 in figure [ delta_sn_plot_supp ] . it can be seen that for the region sizes we are able to compute , the qualitative trends for @xmath64 shown for @xmath28 in the letter also hold for @xmath101 and @xmath102 , @xmath103 .
and @xmath104 plotted against @xmath39 .
the smallest region contains @xmath105 particles and the largest @xmath106 .
[ delta_sn_plot_supp],title="fig : " ] and @xmath104 plotted against @xmath39 .
the smallest region contains @xmath105 particles and the largest @xmath106 .
[ delta_sn_plot_supp],title="fig : " ] | we perform quantum monte carlo calculations to determine how the renyi entropies , @xmath0 , of the interacting fermi liquid depend on renyi order , @xmath1 , and scale as a function of system size , @xmath2 . using the swap operator and an accurate slater - jastrow wave function ,
we compute renyi entropies for spinless fermions interacting via the coulomb and modified pschl - teller potentials across a range of correlation strengths .
our results show that interactions increase the renyi entropies and increase the prefactor of their scaling laws .
the relationships between renyi entropies of different order @xmath1 are also modified .
additionally , we investigate the effect of the swap operator on the fermi liquid wave function to determine the source of the @xmath3 scaling form . entanglement is a central concept in quantum information theory , and has become increasingly important in studies of quantum many body models of condensed matter @xcite .
the renyi entropies computed from a spatially reduced density matrix provide a measure of entanglement , through non - local correlations in the ground state wave function @xcite .
the scaling laws for these entropies provide a `` fingerprint '' for quantum phases @xcite . for 1d systems these scaling laws
can often be computed analytically using results from conformal field theory @xcite . in higher dimensions
the renyi entropies are more difficult to compute and are often calculated using numerical techniques @xcite . no general results for interacting systems in higher dimension are available
, models must be addressed individually and results can differ greatly from their non - interacting counterparts @xcite .
one of the most fundamental models in condensed matter is the fermi liquid @xcite .
the renyi entropies of the non - interacting fermi liquid in two dimensions have been conjectured to scale as @xmath4 @xcite .
this scaling law is equivalent to the widom conjecture and has been verified numerically for lattice and continuum hamiltonians @xcite .
high dimensional bosonization and conformal field theory studies have predicted that the scaling law of the renyi entropies for the interacting and non - interacting fermi liquid are identical @xcite .
these results imply that the relationship between @xmath0 of different order , @xmath5 , should also hold when interactions are included @xcite .
in this work , we use variational quantum monte carlo ( vmc ) to test these predictions . specifically , we compute the renyi entropy scaling forms for two different inter - particle potentials and compare them to the non - interacting results .
vmc provides an accurate framework for the calculation of interacting and non - interacting quantum states in a many body framework , and is expected to produce reliable results for well optimized trial wave functions @xcite .
high quality trial wave function forms and robust optimization schemes have been developed to accurately compute observables @xcite .
these have been used to calculate benchmark energies for the electron gas and fermi liquid parameters for the coulomb potential , helium-3 , and several related fermi liquid systems @xcite .
_ hamiltonians and trial wave functions- _ we choose two inter - particle potentials , the coulomb and the modified pschl - teller potential ( mpt ) , to test if our results are sensitive to the range of the interaction potential .
spin polarized electrons interact via the long range coulomb potential @xcite .
the mpt potential is a well tested short range potential for the fermi gas @xcite .
if the range of the potential is important to the renyi entropies then we could expect some qualitative difference between the scaling laws computed for these potentials .
the hamiltonian for the spinless non - interacting electron gas in hartree units is @xmath6 .
the exact fermionic ground state for this hamiltonian is @xmath7 . for the interacting electron gas @xmath8 , where @xmath9 is the distance between electron @xmath10 and @xmath11 , and the constant , @xmath12 , is due to a uniform positive background charge . for the mpt potential @xmath13 .
the parameter @xmath14 can be tuned to increase the correlation of the system and the effective range is set to the wigner - seitz radius , @xmath15 @xcite .
we use a slater - jastrow trial wave function , @xmath16 , for these interacting hamiltonians .
the jastrow is of the form @xmath17 and adds additional correlation between particle pairs .
it is a reasonably accurate form that can be quickly computed @xcite . for the electron gas wave function , @xmath18 ,
we eliminate the divergence in the electron - electron energy as @xmath19 by including a cusp in the jastrow @xcite . for the mpt wave function , @xmath20
, there is no singularity in the potential and therefore no cusp in the jastrow . _
renyi entropy and the swap operator- _ the renyi entropies are defined as @xmath21 with the spatially reduced density matrix , @xmath22 , of a system partitioned into two regions , @xmath23 and @xmath24 . using vmc we are able to compute unbiased estimates of @xmath0 for trial wave functions using the swap operator @xcite , @xmath25 the swap operator computes @xmath26 by cyclically swapping coordinates between @xmath1 statistically independent copies of the system , each sampling @xmath27 .
for instance , we can compute @xmath28 using , @xmath29 where @xmath30 represents the electronic spin and spatial coordinates in region @xmath23 for copy @xmath10 , @xmath31 and @xmath32 . _ computing renyi entropies- _ in practice
, the expectation value of @xmath26 decays rapidly as region @xmath23 increases in size @xcite . to compute renyi entropies for large regions
we factorize the swap operator into two parts : sign , @xmath33 , and magnitude , @xmath34 @xcite .
these two pieces are computed separately then summed to obtain the full value .
this factorization is not an approximation and does not change the values we compute for @xmath0 .
the magnitude component can be further factorized into @xmath35 , with @xmath36 due to swapped copies having different numbers of particles in region @xmath24 , and @xmath37 from the absolute value of the swap when they do have the same number .
the estimators for the full factorization , @xmath38 can be found in the supplementary material . as a function of @xmath39 , @xmath40 , for ( _ left _ ) the non - interacting electron gas and ( _ right _ ) the interacting gas at @xmath41 .
the bold black line is @xmath28 .
@xmath36 , @xmath33 , and @xmath37 are the factorization of @xmath28 as in eqn .
[ factored_swap ] .
the @xmath39 boundary law violation comes from the sign of the swap , it is the only term that continues to increase as @xmath2 gets large . for the interacting system at @xmath41 ,
the jastrow increases the contribution from @xmath37 .
[ spo0_orb ] ] through the circular region @xmath23 and plot @xmath42 , defined in the letter body , for @xmath43 .
the white dot is the scanned particle at it s original position , @xmath44 , in the lower left of region @xmath23 .
the swap operator exchanges the red ( blue ) particles from copy @xmath45 ( @xmath46 ) outside of region @xmath23 .
the red ( blue ) regions are positive ( negative ) , with saturation representing magnitude .
( _ bottom _ ) the average amplitude of @xmath47 , where @xmath48 is the distance from the center of region @xmath23 .
@xmath47 is larger than it s average value in the interior and decays rapidly near the boundary .
the sign behaves similarly .
[ swap_nodes ] ] _ source of the area law violation- _ in fig .
[ spo0_orb ] we plot the contributions to @xmath28 from the factorization in eqn .
[ factored_swap ] for the non - interacting fermi liquid and the interacting electron gas at @xmath41 .
the contribution to @xmath28 from the sign of the swap operator grows as @xmath2 increases , while the other contributions plateau or decrease .
this indicates that as @xmath2 gets large the sign of the swap operator is the only component that can be responsible for the leading scaling behavior .
we note that previous studies found the same qualitative behavior @xcite .
we investigate the spatial structure of the amplitude of the swap operator in fig .
[ swap_nodes ] .
the upper plot shows @xmath49 , with @xmath44 a coordinate in @xmath50 : @xmath51 .
this plot is produced by swapping all particles in region @xmath24 then scanning a single particle initially located at @xmath44 , through region @xmath23 and plotting the amplitude of the swap operator .
the lower plot shows the normalized average amplitude of the swap operator as a function of distance from the center of region @xmath23 .
this average amplitude , just as the average sign , is approximately constant in the center and decays as particles are moved out to the edge .
this suggests that the change in sign of the swap operator due to particles near the boundary drives the @xmath52 violation of the area law .
heuristically , we can justify the jastrow modifying the @xmath3 and @xmath2 scaling terms .
the jastrow can change the leading scaling law of the renyi entropies by changing the particle distribution .
it can also create additional @xmath2 dependence by destroying the short range correlations in the jastrow for the particles in region @xmath23 near region @xmath24 .
we expect the difference in the renyi entropies between the interacting and non - interacting systems to be well fit by these corrections , @xmath53 .
_ computational details- _ all qmc calculations are performed in qmcpack@xcite .
we parameterize the jastrow by a flexible b - spline function with fixed cusp conditions depending on the potential and optimize them using a variant of the linear method @xcite .
a more accurate slater - jastrow - backflow wave function was also tested @xcite . using this wave function renyi
entropy results for @xmath28 at @xmath41 were the same as the slater - jastrow values .
the only changes were in the orbital contributions to the renyi entropy .
backflow increased the determinant and decreased the jastrow contributions to the swap magnitude .
for the electron gas , the swap operator for @xmath28 was used without factorization for circular regions @xmath23 up to @xmath54 for a simulation cell of @xmath55 electrons and with the sign factorization technique for @xmath54 , @xmath56 , and @xmath57 in @xmath58 electron systems .
we computed @xmath28 for regions of @xmath59 though @xmath60 for the @xmath55 particle simulation cell using the unfactorized estimator for the mpt potential . when factorization is unnecessary we are able to increase our computational efficiency by computing the swap operator for several regions sizes at the same time .
approximately @xmath61 cpu hours were spent on the smaller systems and the same on the larger . for the higher order renyi entropies we were restricted to @xmath62 or fewer electrons for the @xmath58 electron system .
we rescale all lengths , @xmath63 , so that @xmath64 for the non - interacting fermi liquid has no @xmath65 dependence . under this rescaling
, all interacting @xmath0 will collapse to the non - interacting @xmath0 if interactions are irrelevant .
we use a @xmath39 x - axis to highlight the log linear nature of @xmath66 and @xmath67 . _
results- _ we find that interactions increase the renyi entropies and increase the prefactor in their scaling law .
more strongly correlated hamiltonians produce ground state wave functions with stronger particle correlations and larger renyi entropies .
we also find that coulomb interactions change the simple relationship between the renyi entropies of different order @xmath1 .
the non - interacting data is well fit by the leading scaling form predicted by the widom conjecture , @xmath68 with @xmath69 and @xmath70 @xcite . as shown in fig .
[ s2_plot ] , coulomb interactions modify the renyi entropies when the coulomb energy becomes larger in magnitude that the kinetic energy . at @xmath15
the coulomb energy is approximately equal to the kinetic energy and interactions do not significantly alter @xmath28 . as coulomb interactions become stronger , the renyi entropy @xmath28 increases . for @xmath71
the coulomb energy is an order of magnitude larger than the kinetic energy , and the renyi entropy is increased relative to the non - interacting fermi gas . the higher order renyi entropies , reported in the supplementary material and shown in fig . [ sn_plot ] , show the same qualitative trend .
we find @xmath72 is well fit to the two parameter form @xmath73 .
the fitting parameters for the coulomb potential at all @xmath65 are presented in table [ cshifts ] and for the mpt at all @xmath14 in table [ ptshifts ] .
the coulomb fit is plotted in the inset of fig .
[ s2_plot ] .
as the strength of the inter - particle potential is increased so too is the prefactor of the leading scaling term , @xmath74 .
we compute the quasi particle renormalization factor , @xmath75 , to determine how strong inter - particle correlations are and compare scaling laws between different model potentials @xcite .
the scaling laws for the mpt potentials show qualitative agreement with the interacting electron gas with the same @xmath75 . as @xmath75 increases so does the slope of @xmath76 .
these slopes are almost linear with @xmath75 and are plotted in fig .
[ z_m ] .
we note that finite size corrections to @xmath75 may not preserve this relationship @xcite . on the left of fig .
[ sn_plot ] , @xmath77 is plotted for the coulomb potential for a region of size @xmath78 . the simple relationship between non - interacting @xmath0 does not hold when coulomb interactions are included .
interactions increase the renyi entropies for all @xmath1 . as shown on the right of fig .
[ sn_plot ] , upon rescaling , @xmath79 , the non interacting renyi entropy is made constant .
the difference between the interacting and non - interacting @xmath0 decreases as @xmath1 gets larger .
the growth of the error bars as @xmath2 and @xmath1 increase prevent a good fit for their scaling laws and the relationship between @xmath0 of different order @xmath1 . for the spin polarized interacting electron gas at @xmath80 and the non - interacting fermi liquid .
@xmath2 is scaled so that without interactions all lines lie on top of @xmath43 results .
the high density @xmath15 data falls on top of the non - interacting liquid while the lower density , more strongly correlated , gasses have larger renyi entropies .
the @xmath43 reference is the exact non - interacting @xmath28 computed using the same number of particles as the interacting system for each @xmath2 .
( _ inset _ ) @xmath81 plotted against @xmath39 .
the renyi entropy for the interacting case appears to scale the same as the non - interacting case with a larger prefactor .
the lines are from a fit to @xmath82 shown in table [ cshifts ] .
[ s2_plot ] ] , for @xmath83 for region @xmath23 size @xmath84 , @xmath85 .
@xmath0 increases as correlations increase and decreases as @xmath1 increases .
( _ right _ ) @xmath86 for @xmath83 .
besides some small @xmath2 oscillations , this scaling makes the non - interacting renyi entropies linear in @xmath1 .
interactions modify this relationship and the rescaled renyi entropies decrease as @xmath1 gets larger .
[ sn_plot ] ] ( @xmath87 ) , as a function of @xmath88 , one minus the quasi particle renormalization factor , for the coulomb and modified pschl - teller ( mpt ) potential at all correlation strengths considered .
lines are guides to the eye from linear fits constrained to go through the origin .
[ z_m ] ] .@xmath65 is the wigner - seitz radius of the electron gas .
@xmath89 is the non - interacting system .
@xmath90 .
@xmath75 is the quasi particle renormalization factor .
corr@xmath91 is the ratio of correlation energy the wave function recovers and provides a metric for wave function quality.[cshifts ] [ cols="^,^,^,^,^,^",options="header " , ] _ error analysis- _ for these calculations there are two primary sources of error , finite size effects , and trial wave function bias .
finite size effects come from two places , one is a function of the number of electrons in the system , the other is from subleading corrections to the renyi entropy scaling law . by integrating out regions that are a small fraction of the system size , we minimize interactions of region @xmath23 and it s periodic neighbors . for small subregions
we see oscillations , small relative to @xmath28 , that decay as the subregion length increases for both the interacting and non - interacting cases @xcite .
fits are performed on @xmath76 to minimize the effect of oscillations on the fitted parameters .
the degree to which the trial wave function represents the fermi liquid depends on it s overlap with the exact ground state wave function .
we estimate the quality of our trial wave function by looking at the amount of correlation energy the vmc recovers relative to the fixed node diffusion quantum monte carlo result , @xmath92 @xcite .
we recover more than @xmath93 of this correlation energy for all densities and potentials , shown in table [ cshifts ] and [ ptshifts ] .
also because the slater - jastrow and slater - jastrow - backflow renyi entropies are similar , we expect them to be accurate .
_ conclusions and future work- _ in this work we have performed the first _ ab initio _ calculation of the renyi entropy of the interacting fermi liquid .
our results show that , for this system , interactions modify the scaling behavior of the renyi entropies . as particle correlations , quantified by @xmath75 ,
are strengthened , the prefactor for the leading scaling behavior of the renyi entropies increases .
these results hold for all wave function forms , correlation strengths , and inter particle potentials considered in this work .
we also find the relationship between @xmath0 of different order @xmath1 changed .
the source of the disagreement with recent theoretical predictions is not clear and deserves further investigation @xcite .
work is currently underway improving the efficiency of computing higher order renyi entropies and investigating related systems such as the paramagnetic electron gas and the wigner crystal . _
acknowledgements- _ we would like to thank david ceperley , eduardo fradkin , sarang gopalakrishnan , and benjamin hsu for conversations .
this work was supported by the national science foundation and efrc center for defect physics sponsored by the u.s .
doe , office of basic energy sciences .
this work used the extreme science and engineering discovery environment ( xsede ) , which is supported by national science foundation grant number oci-1053575 . , and resources provided by the innovative and novel computational impact on theory and experiment ( incite ) at the oak ridge leadership computing facility at the oak ridge national laboratory , which is supported by the office of science of the u.s .
department of energy under contract no .
de - ac05 - 00or22725 .
55ifxundefined [ 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 _
_
( , , ) link:\doibase 10.1103/physrevlett.100.070502 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.010504 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.90.227902 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.110405 [ * * , ( ) ] link:\doibase 10.1103/physreva.66.032110 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.010504 [ * * , ( ) ] @noop http://stacks.iop.org/1742-5468/2004/i=06/a=p06002 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.050404 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.110404 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.82.277 [ * * , ( ) ] link:\doibase 10.1103/physreva.78.032329 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.69.2863 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.77.259 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.220601 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.245119 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.165134 [ * * , ( ) ] @noop ( ) , link:\doibase 10.1103/physrevb.85.125308 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.067202 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.235127 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.020402 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.205105 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.245128 [ * * , ( ) ] http://www.google.co.in/books?id=rf59qgaacaaj[__ ] , theory of quantum liquids ( , ) link:\doibase 10.1103/physrevlett.98.220603 [ * * , ( ) ] link:\doibase 10.1063/1.2800167 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.100503 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.010404 [ * * , ( ) ] link:\doibase 10.1103/physreva.74.022329 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.073103 [ * * , ( ) ] http://stacks.iop.org/0295-5075/97/i=2/a=20009 [ *
* , ( ) ] link:\doibase 10.1103/physrevlett.105.050502 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.035116 [ * * , ( ) ] link:\doibase 10.1103/physrevx.2.011012 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.73.33 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.45.566 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.88.256601 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.98.110201 [ * * , ( ) ] link:\doibase 10.1103/physrevb.39.5005 [ * * , ( ) ] link:\doibase 10.1103/physrevb.53.7376 [ * * , ( ) ] link:\doibase 10.1103/physrevb.16.3081 [ * * , ( ) ] @noop _ _ ( , , ) link:\doibase 10.1103/physrevlett.91.050401 [ * * , ( ) ] link:\doibase 10.1103/physreva.81.033619 [ * * , ( ) ] link:\doibase 10.1103/physreva.84.023615 [ * * , ( ) ] link:\doibase 10.1103/physrevb.50.1684 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.041308 [ * * , ( ) ] link:\doibase 10.1103/physrevb.18.3126 [ * * , ( ) ] link:\doibase 10.1002/cpa.3160100201 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.157201 [ * * , ( ) ] @noop `` , '' @noop _ _ ( , ) link:\doibase 10.1103/physrevb.48.12037 [ * * , ( ) ] |
one of the most exciting prospects of particle physics is the discovery of the source of electroweak symmetry breaking ( ewsb ) . in the standard model ( sm ) ,
a single @xmath1 complex scalar doublet is responsible for both gauge boson masses ( via the higgs mechanism ) and fermion masses ( via yukawa interactions ) .
the only scalar physical remnant of the process is the higgs boson ( @xmath2 ) whose mass is a free parameter of the model . in contrast , the minimal supersymmetric standard model ( mssm ) contains two complex @xmath1 scalar doublets , as required to give masses to both up- and down - type quarks and to cancel gauge anomalies . once ewsb occurs in the mssm , the scalar sector contains _ five _ physical higgs bosons : a light scalar @xmath3 , a heavy scalar @xmath4 , a pseudoscalar @xmath5 and a pair of charged bosons @xmath6 . finding experimental evidence for the existence of one or more higgs bosons is , thus , a major goal of current and future collider experiments .
in addition , measuring the couplings of the higgs boson(s ) to gauge bosons , leptons and quarks would be imperative to uncover the true structure of the higgs sector .
direct searches for higgs bosons at lep2 have placed lower mass limits on both the sm higgs ( @xmath7 114.4 gev at @xmath8 c.l . ) @xcite and the lightest scalar and pseudoscalar higgs bosons of the mssm ( @xmath9 93 gev at @xmath8 c.l . )
in addition , the mass of the lightest scalar higgs boson of the mssm is also bounded from above by theory to less than @xmath10 130 gev . finally
, precision electroweak measurements , which are sensitive to the effects of higgs boson loop contributions , imply that the mass of the sm higgs boson should be less than @xmath10 186 gev at @xmath8 c.l .
@xcite ( @xmath10 219 gev when including the lep-2 direct search limit ) .
hence , in both the sm and the mssm , a higgs boson should lie in a mass range which can be explored at the large hadron collider ( lhc ) @xcite .
the dominant production mechanism for a sm higgs boson in hadronic interactions ( @xmath11 or @xmath12 ) is gluon fusion ( @xmath13 ) , which proceeds predominantly through a loop of top quarks . among the production modes with smaller rates , the associated production with electroweak gauge bosons ( @xmath14 , where @xmath15 ) or with top quark pairs ( @xmath16 ) , as well as weak boson fusion ( @xmath17 ) , play crucial roles .
the next - to - leading order ( nlo ) qcd corrections have been calculated for all of the above processes @xcite , while for gluon fusion and associated @xmath18 production corrections are also known at next - to - next - to - leading order ( nnlo ) in perturbative qcd @xcite . in the mssm
, the hierarchy of production modes can be significantly different from that of the sm .
in fact , for large values of @xmath0 ( the ratio of the two mssm higgs vacuum expectation values ) , the bottom quark yukawa couplings to both scalar and pseudoscalar higgs bosons become strongly enhanced and the production of higgs bosons with bottom quarks dominates over all other production modes . and @xmath19 production at tree level . ]
recently , the production of higgs bosons in association with bottom quarks has received much interest from both the theoretical and experimental communities @xcite . at tree level , the cross section is almost entirely dominated by the sub - process @xmath20 , with only a small contribution from @xmath19 , at both the tevatron and lhc ( see fig .
[ fg : ggbbh_feyn ] ) .
the theoretical prediction of @xmath21 production at hadron colliders , however , involves several subtle issues , as will be described in detail in this review , and depends on the number of bottom quarks identified , or _
tagged _ , in the final state . in case of no or only one tagged bottom quark there are two approaches available for calculating the cross sections to @xmath21 production , dubbed the four flavor ( 4fns ) and five flavor ( 5fns ) number schemes .
the main difference between these two approaches is that the 4fns is a fixed - order calculation of qcd corrections to the @xmath22-induced @xmath21 production processes , while in the 5fns the leading processes are induced by @xmath23 ( @xmath24 ) and @xmath25 initial states and large collinear logarithms present at tree level as well as at each order of qcd corrections are resummed by using a pertubatively defined bottom quark parton distribution function ( pdf ) .
the two schemes are described in more detail in section [ sec : fourfns ] and [ sec : fivefns ] , respectively . in the 5fns . ] in the 5fns . ] for purposes of differentiating the final states , we introduce the following terminology . _ exclusive _ @xmath21 production refers to the case when both bottom quark jets are tagged ( or are at high transverse momentum , @xmath26 ) .
note that , in this case , the computation of the cross section solely relies on the 4fns .
_ semi - inclusive _ production ( denoted by @xmath27 ) pertains to the case when _ at least _
one bottom quark is at high @xmath26 . here , the dominant leading - order ( lo ) process for the 4fns ( 5fns ) is @xmath28 ( @xmath29 or @xmath30 , see fig .
[ fg : bghb_feyn ] ) .
finally , the _ inclusive _ production process ( denoted by @xmath31 ) involves the case where no bottom quark jets are tagged .
the dominant lo process for this mode is @xmath32 ( @xmath33 , see fig .
[ fg : bbh_feyn ] ) in the 4fns ( 5fns ) .
it should be mentioned that , although the cross section for the inclusive production mode dominates over the less inclusive modes by one to two orders of magnitude , tagging bottom quark jets greatly reduces backgrounds , making the less inclusive modes experimentally more appealing .
requiring one or two high-@xmath26 bottom quark jets in the final state also ensures that the detected higgs boson has been radiated off a bottom or anti - bottom quark and the corresponding cross section is unambiguously proportional to the bottom quark yukawa coupling .
qcd corrected cross sections are available for all three final states and the inclusive and semi - inclusive production processes have been computed in both the 4fns and 5fns . for the inclusive case , the nlo qcd corrected 4fns @xcite and the nnlo qcd corrected 5fns cross sections @xcite have been compared and are found to be in good agreement within theoretical uncertainties .
the nlo predictions of the semi - inclusive cross sections for the 4fns @xcite and 5fns @xcite have also been extensively compared @xcite and the agreement between the two is spectacular .
the compatibility of these two seemingly different calculational schemes in the prediction of inclusive and semi - inclusive higgs boson production rates is indeed a powerful check of the theory .
the 4fns and 5fns represent different perturbative expansions of the same physical observables , and therefore should agree at sufficiently high order .
indeed , we see that considering the first ( second ) order of corrections already brings agreement between the two schemes within the respective theoretical uncertainties . finally , two independent calculations of the nlo qcd corrections for the exclusive mode have been compared and agreement has been found @xcite . the above discussion for the production of a _ scalar _ higgs boson with bottom quarks applies equally well to the production of a _
_ higgs boson .
in fact , for massless bottom quarks , the predictions for @xmath34 would be identical to those of @xmath35 given a rescaling of the yukawa couplings . for massive bottom quarks , however , the situation becomes complicated due to the @xmath36 appearing in the @xmath34 yukawa coupling .
the @xmath36 dirac matrix is intrinsically a four - dimensional object and care must be taken in its treatment when dimensionally regularizing the calculation . in any case , bottom quark mass effects are expected to be small , i.e. @xmath37@xmath38 , and predictions for @xmath35 are good indicators for @xmath34 production ( modulus yukawa couplings ) .
the remainder of this paper is organized as follows . in section [ sec : theory ] , we discuss the framework of the calculations in the four- and five - flavor - number schemes in detail . given the importance it plays in the prediction of the cross section through the yukawa coupling
, we also discuss different renormalization schemes of the bottom quark mass .
section [ sec : mssmcoups ] details the susy - corrected yukawa couplings obtained from the program feynhiggs @xcite that were used in this study to produce results for mssm higgs bosons . in section [ sec : numbers ] , we review the results for inclusive , semi - inclusive and exclusive higgs boson production with bottom quarks . in section [ sec :
pdfs ] , we investigate the uncertainties arising from the pdfs for the semi - inclusive @xmath21 production process .
these uncertainties are calculated using the algorithm developed by the cteq collaboration @xcite .
finally , we summarize and conclude in section [ sec : summary ] .
final state bottom quarks are identified imposing cuts on their transverse momentum ( @xmath39 ) and pseudorapidity ( @xmath40 ) . on the other hand , when a final state bottom quark is not identified , as in the inclusive and semi - inclusive @xmath21 production processes , the corresponding integration over its phase space may give rise to large logarithms of the form : @xmath41 where @xmath42 and @xmath43 represent the lower and upper bounds on the integration over the transverse momentum , @xmath26 , of the final state bottom quark .
this happens when the final state bottom quark is directly originating from the @xmath44 splitting of an initial state gluon , and corresponds to the collinear singularity that would be present in the case of a gluon splitting into massless quarks .
the scale @xmath43 is typically of @xmath45 and therefore , due to the smallness of the bottom quark mass , these logarithms can be quite large . additionally , the same logarithms appear at every order in the perturbative expansion of the cross section in the strong coupling , @xmath46 , due to recursive gluon emission from internal bottom quark lines
. these logarithms could severely hinder the convergence of the perturbative expansion of the total and differential cross sections . in the 5fns the convergence
is improved by introducing a perturbatively defined bottom quark pdf @xcite , defined at lowest order in @xmath46 as @xmath47 where @xmath48 is the ( experimentally - measured ) gluon pdf and @xmath49 is the altarelli - parisi splitting function for @xmath50 given by @xmath51\,\,\,.\ ] ] subsequently , the @xmath52 logarithms are resummed through the dglap evolution equation @xcite , such that contributions proportional to @xmath53 can be absorbed , to all orders in @xmath54 , into a leading logarithms bottom quark pdf , while subleading logarithms are recursively resummed when higher order corrections are considered in the dglap equation . since @xmath55 is not a small expansion parameter , the use of a bottom quark pdf should improve the stability of the total and differential cross sections for inclusive and semi - inclusive higgs boson production with bottom quarks .
the 5fns is based on the approximation that any _ spectator _ bottom quarks ( i.e. , bottom quarks which are not tagged in the final state ) are at small transverse momentum , since this is the region where the @xmath52 logarithms dominate over other @xmath26-dependent contributions . at lowest order ,
the spectator quarks are produced with zero @xmath26 , and a transverse momentum spectrum for the outgoing bottom quarks is generated at higher orders @xcite . with the use of a bottom quark pdf
, the 5fns effectively reorders the perturbative expansion to be one in @xmath46 _ and _ @xmath56 . to see how this works ,
let us consider the perturbative expansion of the inclusive process @xmath57 ( fig .
[ fg : bbh_feyn ] ) , which is at tree level of order @xmath58 . at nlo ,
the virtual and real corrections to the tree level process make contributions of @xmath59 .
however , at nlo , we must also consider the contribution from @xmath29 where the final state bottom quark is at high @xmath26 .
this process makes a contribution of order @xmath60 and is , thus , a correction of @xmath61 to the tree level cross section .
similarly , at nnlo , among the myriad of radiative corrections of @xmath62 , we must also include the contribution from the process @xmath20 , where both @xmath63 and @xmath64 are at high @xmath26 .
the contribution from these diagrams are of order @xmath65 , and are , thus , @xmath66 ( or nnlo ) corrections to the tree level process @xmath33 @xcite .
the above discussion for @xmath33 also applies to the perturbative expansion of @xmath29 . in this case , the tree level process is of order @xmath67 and the contribution from @xmath20 is a nlo correction of @xmath68 @xcite .
finally , we observe that , in the existing 5fns calculations , all bottom quark masses other than the one appearing in the yukawa coupling are set to zero . in the 4fns
, the initial state quarks are constrained to be the four lightest quarks , i.e. there are no bottom quarks in the initial state .
no kinematic approximations are made and the cross section for @xmath69 is computed at fixed order in qcd without resumming higher - order collinear logarithms .
moreover , the bottom quark mass is always considered to be non - zero .
similarly to the 5fns , for the less inclusive final states discussed above ( @xmath27 or @xmath21 ) , identification cuts are imposed on the final state bottom quark(s ) which constrain its transverse momentum and pseudorapidity . applying these cuts to a final state bottom quark
eliminates large logarithms that may appear from gluon splitting and it ensures that the bottom quark jets can be tagged experimentally . while the inclusive and semi - inclusive production of a higgs boson with bottom quarks can be calculated in both the 4fns and 5fns , the exclusive production can only be calculated in the 4fns , since both final state bottom quarks are tagged .
the nlo qcd corrections to the hadronic process @xmath70 in the 4fns consist of calculating the @xmath71 virtual and real qcd corrections to the tree level ( partonic ) processes @xmath72 @xcite .
we note that the nlo calculation of @xmath69 is identical to the calculation of the nlo qcd corrections to @xmath73 @xcite , with the global interchange of the top quark and bottom quark masses ( @xmath74 )
. one potential source of theoretical uncertainty in the calculation of @xmath69 involves the renormalization of the bottom quark mass .
the bottom quark mass counterterm has to be used twice in the renormalization of the calculation : once to renormalize the bottom quark mass appearing in internal propagators and once to renormalize the sm bottom quark yukawa coupling @xmath75 , where @xmath76 gev .
indeed , if one only considers qcd corrections , the counterterm for the bottom quark yukawa coupling coincides with the counterterm for the bottom quark mass , since the vacuum expectation value @xmath77 is not renormalized at 1-loop in qcd .
the renormalization scheme dependence has been studied in ref .
@xcite , in the 4fns , for the case of the exclusive @xmath78 production .
two schemes for the renormalization of the bottom quark mass have been chosen : the @xmath79 scheme and an on - shell ( @xmath80 ) scheme .
the two are perturbatively consistent at nlo with the only difference being at higher orders and , thus , part of the theoretical uncertainty of the nlo calculation .
the main reason for investigating this renormalization scheme dependence is the large sensitivity of the @xmath79 bottom quark mass and the prominent role it plays in the @xmath21 production cross section through the overall bottom quark yukawa coupling .
a summary of the results of this investigation are presented in section [ subsec:2btag ] .
the benefit of using the @xmath79 scheme for the yukawa coupling is that it potentially gives control over higher - order corrections beyond the 1-loop corrections .
this is often reflected in the better perturbative behavior of physical observables calculated using the @xmath79 bottom quark yukawa coupling . therefore , unless otherwise stated all the results presented in this review are given using the @xmath79 scheme for the bottom quark mass and yukawa coupling .
this is accomplished by replacing the @xmath79 bottom quark mass in the yukawa coupling with the corresponding 1-loop and 2-loop renormalization group improved @xmath79 masses : @xmath81^{c_0/b_0}\,\,\,,\ ] ] @xmath82^{c_0/b_0 } \left [ 1+\frac{c_0}{b_0}(c_1-b_1)\frac{\alpha_s(\mu)-\alpha_s(m_b)}{\pi } \right ] \left(1-\frac{4}{3}\frac{\alpha_s(m_b)}{\pi}\right)\,\,\,,\ ] ] where we take the bottom quark pole mass to be @xmath83 gev and @xmath84 are the one and two loop coefficients of the qcd @xmath85 function and mass anomalous dimension @xmath86 , while @xmath87 is the number of colors and @xmath88 is the number of light flavors .
so far , most of the results for @xmath21 production have been obtained in the sm , in order to simplify the comparison of different calculations and approaches ( for a review see , e.g. , ref .
@xcite ) . as stated earlier , though , these processes are relevant discovery modes at hadron colliders only for susy models with large @xmath0 . therefore , we will present most of the following results for the case of the mssm with large @xmath0 .
the prediction for the mssm case can easily be derived from the sm results by rescaling the yukawa couplings . in the 4fns , due to the existence of diagrams where the higgs is radiated off a closed top quark loop
, some care must be taken when performing this rescaling . in this review
, all 4fns results for the neutral mssm higgs bosons , are obtained from the sm results by carefully rescaling the contributions _ with _ ( @xmath89 ) and _ without _ top quark loops ( @xmath90 ) as : @xmath91 examples of the closed top quark loop contributions described by @xmath92 are shown in fig .
[ fg : bbhtriangle ] .
note that in the 5fns , as implemented in the program mcfm @xcite , the corresponding contributions to @xmath92 are considered to be zero . while this could affect the comparison in the sm , in susy models with large bottom quark yukawa couplings ( and consequently small top quark yukawa couplings ) the contributions of the closed top quark diagrams is very small , and @xmath92 does not play a significant role . .
the circled cross denotes all possible insertion of the final state higgs boson leg , each insertion corresponding to a different diagram . ] in the mssm the bottom and top quark yukawa couplings to the scalar , neutral higgs bosons are given by : @xmath93 where @xmath94 and @xmath95 are the sm bottom and top quark yukawa couplings , @xmath3 and @xmath4 are the lighter and heavier neutral scalars of the mssm , and @xmath96 is the angle which diagonalizes the neutral scalar higgs mass matrix @xcite .
the dominant susy radiative corrections to @xmath78 production can be taken into account by including the susy corrections to the @xmath97 vertex only , i.e. by replacing the tree level yukawa couplings by the radiative corrected ones .
we follow the treatment of the program feynhiggs @xcite and take into account the leading , @xmath0 enhanced , radiative corrections that are generated by gluino - sbottom and chargino - stop loops , as follows @xcite : @xmath98\,\,\ , , \nonumber\\ g_{b\bar{b}h^0}^{{\scriptscriptstyle}m\!s\!s\!m } & = & \,\,\,\,g_{b\bar{b}h}^{{\scriptscriptstyle}s\!m}\frac{1}{1+\delta_b } \left[\frac{\cos\alpha}{\cos\beta}+\delta_b \frac{\sin\alpha}{\sin\beta}\right]\,\,\,,\end{aligned}\ ] ] with @xmath99 \,\,\,,\end{aligned}\ ] ] where @xmath100 gev , @xmath101 denote the sbottom , stop , and gluino masses , respectively , @xmath102 is the stop and @xmath43 the higgs mixing parameter .
the strong coupling constant and the running top quark mass , @xmath103 , are evaluated at the on - shell top mass , @xmath104 gev . the vertex function @xmath105 is given by @xmath106}{(a^2-b^2)(b^2-c^2)(a^2-c^2)}\,\,\,.\end{aligned}\ ] ] in tables [ tab : ratiolight ] and [ tab : ratioheavy ] we provide the ratios of the sm and mssm bottom and top quark yukawa couplings at @xmath107 , as provided by feynhiggs , where we assumed cp conserving couplings . in the following
, we will use these ratios to derive the mssm cross sections from the sm results as described in eq .
( [ eq : rescale ] ) . using feynhiggs ,
the mssm higgs boson masses and the mixing angle @xmath96 have been computed up to two - loop order . as expected , for this choice of mssm input parameters
the bottom yukawa coupling is strongly enhanced and the top yukawa coupling is suppressed , resulting in a mssm @xmath21 cross section that is about three orders of magnitude larger than the sm cross section .
@ccccc@ + @xmath108 [ gev ] & 100 & 110 & 120 & 130 + @xmath96 [ rad ] & -1.5063 & -1.4716 & -1.3798 & -0.7150 + @xmath109&33.913 & 33.823 & 33.387 & 22.390 + @xmath110 & 0.0645 & 0.0991 & 0.1899 & 0.7553 + @cccccccc@ + @xmath111 [ gev ] & 120 & 200 & 300 & 400 & 500 & 600 & 800 + @xmath96 [ rad ] & -0.3315 & -0.0454 & -0.0318 & -0.0282 & -0.0266 & -0.0258 & -0.0251 + @xmath112&25.787 & 27.338 & 27.356 & 27.360 & 27.362 & 27.363 & 27.364 + @xmath113&-0.3255 & -0.0454 & -0.0318 & -0.0282 & -0.0266 & -0.0258 & -0.0251 +
if not stated otherwise , the numerical results are obtained using cteq6 m parton distribution functions for the calculation of the nlo cross section , and cteq6l parton distribution functions for the calculation of the lowest order cross sections @xcite .
the nlo ( lo ) cross section is evaluated using the 2-loop ( 1-loop ) evolution of @xmath114 with @xmath115 . for the exclusive and semi - inclusive channels ( @xmath116 and @xmath117 production )
, it is required that the final state bottom quarks have @xmath118gev and pseudorapidity @xmath119 for the tevatron and @xmath120 for the lhc . in the nlo real gluon
emission contributions , the final state gluon and bottom quarks are considered as separate particles only if their separation in the pseudorapidity - azimuthal angle plane , @xmath121 , is larger than @xmath122 . for smaller values of @xmath123 ,
the four momentum vectors of the two particles are combined into an effective bottom / anti - bottom quark momentum four - vector . in the following numerical discussion ,
the sm nlo qcd corrected cross sections for semi - inclusive and exclusive @xmath78 production in the 4fns are taken from refs .
@xcite and @xcite , respectively .
the 5fns sm results for semi - inclusive @xmath124 production have been produced with the program mcfm @xcite and for the inclusive case have been taken from ref .
@xcite .
if the outgoing bottom quarks can not be observed then the dominant mssm higgs production process at large @xmath0 is @xmath126 ( 4fns ) or @xmath33 ( 5fns ) . at the lhc
, this process can be identified by the decays to @xmath127 and @xmath128 for the heavy higgs bosons , @xmath4 and @xmath5 , of the mssm where the higgs couplings are enhanced at large @xmath0 .
recently , also at the tevatron this process , with @xmath129 , has been used to search for the neutral mssm higgs boson @xcite . in the 5fns
this process has been computed to nnlo @xcite , and has only a small renormalization and factorization scale dependence . in the 4fns ,
the production processes @xmath130 , where the outgoing bottom quarks are not observed , are known at nlo @xcite and have been recalculated for the purpose of this review .
a comparison of the total production rates in the 4fns and 5fns within the mssm for @xmath131 is shown in fig .
[ fg : mhdep0b ] .
the bands represent the theoretical uncertainty due to the residual scale dependence at nlo ( nnlo ) . in the 4fns
they have been obtained by varying the renormalization ( @xmath132 ) and factorization ( @xmath133 ) scales independently from @xmath134 to @xmath135 , where @xmath136 . in the 5fns ,
the renormalization scale is fixed to @xmath137 and the factorization scale is varied around the central scale @xmath138 from @xmath139 to @xmath140 .
there is good agreement between the results of the two schemes within their respective scale uncertainties , although the 5fns at nnlo is slightly higher than the 4fns for all values of the higgs boson masses .
[ cols= " > , < " , ] in fig .
[ fg : mhdep ] we compare the production of the sm higgs boson with that of the neutral scalar higgs bosons of the mssm and again observe a significant enhancement of the rate in the mssm for large @xmath0 .
the mssm results have been obtained from the sm results as described in eq .
[ eq : rescale ] by using the rescaling factors of tables [ tab : ratiolight ] , [ tab : ratioheavy ] for @xmath131 .
finally , in figs . [ fg : bbh_ptrel ] , [ fg : bbh_etarel ] we illustrate the impact of nlo qcd corrections on the transverse momentum and pseudorapidity distribution of the sm higgs boson and the bottom quark by showing the relative correction , @xmath141 ( in percent ) . for the renormalization / factorization scale we choose @xmath142 at the tevatron and @xmath143 at the lhc .
these two scale choices are well within the plateau regions where the nlo cross sections vary the least with the value of @xmath43 .
as can be seen , the nlo qcd corrections can considerably affect the shape of kinematic distributions , and their effect can not be obtained from simply rescaling the lo distributions with a k - factor of @xmath144 ( tevatron , @xmath145 ) and @xmath146 ( lhc , @xmath147 ) .
the errors only include the statistical uncertainty from the monte carlo integration .
the kinematic distributions have been calculated within the sm and using the @xmath80 scheme , but we see a similar behavior when using the @xmath79 bottom quark yukawa coupling or consider the mssm case . for the @xmath26 distribution of the bottom or anti - bottom quark with the largest @xmath26 ( @xmath148 ) ( left ) and of the sm higgs boson ( @xmath149 ) ( right ) to @xmath150 production in the sm at the tevatron ( with @xmath151 tev and @xmath152 ) and the lhc ( with @xmath153 tev and @xmath154).,title="fig : " ] for the @xmath26 distribution of the bottom or anti - bottom quark with the largest @xmath26 ( @xmath148 ) ( left ) and of the sm higgs boson ( @xmath149 ) ( right ) to @xmath150 production in the sm at the tevatron ( with @xmath151 tev and @xmath152 ) and the lhc ( with @xmath153 tev and @xmath154).,title="fig : " ] for the @xmath155 distribution of the bottom quark @xmath156 ( left ) and of the sm higgs boson ( @xmath157 ) ( right ) to @xmath150 production in the sm at the tevatron ( with @xmath151 tev and @xmath152 ) and the lhc ( with @xmath153 tev and @xmath154).,title="fig : " ] for the @xmath155 distribution of the bottom quark @xmath156 ( left ) and of the sm higgs boson ( @xmath157 ) ( right ) to @xmath150 production in the sm at the tevatron ( with @xmath151 tev and @xmath152 ) and the lhc ( with @xmath153 tev and @xmath154).,title="fig : " ]
besides the residual renormalization / factorization scale dependence after the nlo corrections have been included , another major source of theoretical uncertainty for cross section predictions at hadron colliders comes from the parton distribution functions .
unfortunately , pdfs are plagued by uncertainties either from the non - perturbative starting distributions used to fit the data or from the perturbative dglap evolution to the higher energies relevant at hadron colliders @xcite .
recently , several collaborations have introduced new schemes which allow an estimate of theoretical uncertainties on physical observables due to the uncertainty in the pdfs . here
, we focus on the scheme introduced by the cteq collaboration based on the hessian matrix method @xcite and study the uncertainties of semi - inclusive total @xmath78 production rates that are induced by the uncertainties in the pdfs . the details of this method are beyond the scope of this review , however , we give a brief explanation below .
first , the nominal set of pdfs ( e.g. cteq6 ) is constructed by fitting a _ non - perturbative core equation _ to data from low - energy experiments designed to measure pdfs .
the core equation , in the method used by cteq , is parameterized by 20 independent parameters which are dialed to fit the data . once the nominal set is fixed , the 20 parameters are then varied in a well - defined manner to produce an additional 40 sets of pdfs .
these sets serve as a _ map _ of the neighborhood around the nominal fit to the data . indeed
, one can then use the 40 sets to estimate the uncertainty from the pdfs on a physical observable in the following way @xcite : first , the central value cross section @xmath158 is calculated using the global minimum pdf ( i.e. cteq6 m ) .
the calculation of the cross section is then performed with the additional 40 sets of pdfs to produce 40 different predictions , @xmath159 . for each of these
, the deviation from the central value is calculated to be @xmath160 when @xmath161 .
finally , to obtain the uncertainties due to the pdfs the deviations are summed quadratically as @xmath162 and the cross section including the theoretical uncertainties arising from the pdfs is quoted as @xmath163 . recently , similar analyses have been performed for the dominant sm higgs production modes @xcite and for the fully inclusive higgs production modes in both the sm ( @xmath164 ) and the mssm ( @xmath165 ) @xcite . in figs .
[ fg : tevpdf ] and [ fg : lhcpdf ] we plot the total sm nlo cross section for @xmath29 obtained with mcfm @xcite at the tevatron and lhc respectively . here , we compare the spread uncertainties from residual scale dependence and the pdfs both for the total cross section ( top ) and the total cross section normalized to the central value calculated with cteq6 m ( bottom ) . from fig .
[ fg : lhcpdf ] one can see that , at the lhc , the theoretical uncertainty is dominated by the residual scale dependence . due to the large center of mass ( c.o.m . )
energy of the lhc , the gluons and bottom quarks in the initial state have small momentum fraction ( @xmath166 ) values and , hence , small pdf uncertainties typically in the 5 - 10% range .
in contrast , due to the smaller c.o.m .
energy , the pdf uncertainties at the tevatron ( fig .
[ fg : tevpdf ] ) are comparable and even larger than the uncertainties due to residual scale dependence over the full higgs mass range .
the smaller c.o.m .
energy results in higher-@xmath166 gluons and bottom quarks in the initial state which corresponds to large pdf uncertainties in the 10 - 30% range . finally , in fig .
[ fg : ggvbg ] , we plot the normalized total sm nlo cross sections of @xmath29 and @xmath167 and compare their respective uncertainties due to the pdfs . we see that , at both the tevatron and the lhc , the pdf uncertainties are almost identical for both the @xmath168 and @xmath23 initial states .
we discussed the status of theoretical predictions for higgs boson production in association with bottom quarks for the cases where zero , one or both outgoing bottom quarks are identified .
these processes are important higgs boson discovery modes in models with enhanced bottom quark yukawa couplings , such as the mssm at large values of @xmath0 .
we presented results for the mssm @xmath21 ( @xmath169 ) total and differential production rates at @xmath131 and discussed the theoretical uncertainties of the state - of - the - art predictions due to the residual renormalization / factorization scale dependence at nlo ( nnlo ) in qcd and the uncertainty in the pdfs .
we have shown that the total cross section for the inclusive case , @xmath170 , and both the total and differential cross sections for the semi - inclusive case , @xmath171 , within a 4fns and 5fns are fully compatible within the existing theoretical errors . all important higgs production processes at hadron colliders are now known at nlo , and some even at nnlo , and the theoretical uncertainties of the qcd predictions are well under control .
we thank s. dittmaier , m. krmer , and m. spira for comparing results ; r. harlander and w. kilgore for providing us with the results of their calculation , and j. campbell , f. maltoni , and s. willenbrock for discussions .
the work of s.d . and c.b.j .
is supported in part by the u.s .
department of energy under grant de - ac02 - 98ch10886 ( de - fg02 - 97er41022 ) .
the work of d.w . is supported in part by the national science foundation under grant nsf - phy-0244875 .
l.r . and d.w .
would also like to thank the aspen center for physics , where part of this work was done , for their hospitality and financial support .
the lep electroweak working group , hep - ex/0412015 , update summer 2005 at webpage : lepewwg.web.cern.ch/lepewwg/stanmod/summer2005@xmath172results .
m. carena and h.e .
haber , _ prog .
phys . _ * 50 * , 63 ( 2003 )
. t. han and s. willenbrock , _ phys .
* b273 * , 167 ( 1991 ) .
t. han , g. valencia , and s. willenbrock , _ phys .
lett . _ * 69 * , 3274 ( 1992 ) .
s. dawson , _ nucl .
phys . _ * b359 * , 283 ( 1991 ) .
a. djouadi , m. spira , and p.m. zerwas , _ phys .
lett . _ * b264 * , 440 ( 1991 ) .
d. graudenz , m. spira , and p.m. zerwas , _ phys .
lett . _ * 70 * , 1372 ( 1993 ) .
w. beenakker , s. dittmaier , m. krmer , b. plmper , m. spira , and p.m. zerwas , _ phys .
_ * 87 * , 201805 ( 2001 ) .
l. reina and s. dawson , _ phys .
* 87 * , 201804 ( 2001 ) .
l. reina , s. dawson , and d. wackeroth , _ phys .
rev . _ * d65 * , 053017 ( 2002 ) .
w. beenakker , s. dittmaier , m. krmer , b. plmper , m. spira , and p.m. zerwas , _ nucl .
phys . _ * b653 * , 151 ( 2003 ) .
s. dawson , l.h .
orr , l. reina , and d. wackeroth , _ phys
. rev . _ * d67 * , 071503 ( 2003 ) .
s. dawson , c. jackson , l.h .
orr , l. reina , and d. wackeroth , _ phys .
rev . _ * d68 * , 034022 ( 2003 ) .
harlander and w.b .
kilgore , _ phys .
lett . _ * 88 * , 201801 ( 2002 ) . c. anastasiou and k. melnikov , _ nucl .
phys . _ * b646 * , 220 ( 2002 ) .
harlander and w.b .
kilgore , _ phys .
rev . _ * d68 * , 013001 ( 2003 ) .
s. catani , d. de florian , and m. grazzini , _ jhep _ * 05 * , 025 ( 2001 ) .
v. ravindran , j. smith , and w.l .
van neerven , _ nucl .
phys . _ * b634 * , 247 ( 2002 ) .
o. brein , a. djouadi , and r. harlander , _ phys .
lett . _ * b579 * , 149 ( 2004 ) .
j. campbell _
et al . _ , in 3rd les houches workshop _ physics at tev colliders _ , 2003 , hep - ph/0405302 . the d0 collaboration , hep - ex/0504018 . the cdf collaboration , `` search for neutral mssm higgs bosons decaying to tau pairs '' , cdf 7676 ( 2005 ) .
s. dittmaier , m. krmer , and m. spira , _ phys .
rev . _ * d70 * , 074010 ( 2004 ) .
s. dawson , c. jackson , l. reina , and d. wackeroth , _ phys .
lett . _ * 94 * , 031802 ( 2005 ) .
j. campbell , r.k .
ellis , f. maltoni , and s. willenbrock , _ phys .
rev . _ * d67 * , 095002 ( 2003 ) .
s. dawson , c. jackson , l. reina , and d. wackeroth , _ phys .
rev . _ * d69 * , 074027 ( 2004 ) .
s. heinemeyer , _ eur .
phys . j. _ * c22 * , 521 ( 2001 ) .
m. frank , s. heinemeyer , w. hollik , and g. weiglein , hep - ph/0212037 . s. heinemeyer , hep - ph/0407244 .
j. pumplin , d.r .
stump , j. huston , h.l .
lai , p. nadolsky , and w .- k .
tung , _ jhep _ * 0207 * , 012 ( 2002 ) .
r. barnett , h. haber , d. soper , _ nucl .
phys . _ * b306 * , 697 ( 1988 ) .
v. gribov and l. lipatov , _ sov .
_ * 15 * , 438 ( 1972 ) .
g. altarelli and g. parisi , _ nucl .
* b126 * , 298 ( 1977 ) .
y. dokshitzer , _ sov .
j. nucl . _ * 46 * , 641 ( 1977 ) .
d. dicus , t. stelzer , z. sullivan , s. willenbrock , _ phys .
rev . _ * d59 * , 094016 ( 1999 ) .
j. campbell and r.k .
ellis , webpage : mcfm.fnal.gov .
m. carena , d. garcia , u. nierste , and c. e. m. wagner , _ nucl .
_ * b577 * , 88 ( 2000 ) .
a. djouadi , j. kalinowski , and m. spira , _ comput .
* 108 * , 56 ( 1998 ) .
j.s . lee _
_ , _ comput .
commun . _ * 156 * , 283 ( 2004 ) .
j.s . lee _
et al . _ ,
webpage : theory.ph.man.ac.uk/@xmath10jslee/cpsuperh.html .
_ , the cteq collaboration , _ eur .
j. _ * c12 * , 375 ( 2000 ) .
s. catani and b.r .
webber , _ jhep _ * 10 * , 005 ( 1997 ) . v.
del duca , f. maltoni , z. nagy , and z. trocsanyi , _ jhep _ * 04 * , 059 ( 2003 ) .
e. boos and t. plehn , _ phys .
_ * d69 * , 094005 ( 2004 ) .
f. maltoni , z. sullivan , and s. willenbrock , _ phys . rev .
_ * d67 * , 093005 ( 2003 ) .
s. ferrag and b. lagorge , atlas - com-2003 - 13 .
martin , r.g .
roberts , w.j .
stirling , and r.s .
thorne , _ eur .
phys . j. _ * c28 * , 455 ( 2003 ) .
a. djouadi and s. ferrag , _ phys .
* b586 * , 345 ( 2003 ) .
a. belyaev , j. pumplin , w.k .
tung and c.p .
yuan , hep - ph/0508222 . | we review the present status of the qcd corrected cross sections and kinematic distributions for the production of a higgs boson in association with bottom quarks at the fermilab tevatron and cern large hadron collider .
results are presented for the minimal supersymmetric standard model where , for large @xmath0 , these production modes can be greatly enhanced compared to the standard model case . the next - to - leading order qcd results are much less sensitive to the renormalization and factorization scales than the lowest order results , but have a significant dependence on the choice of the renormalization scheme for the bottom quark yukawa coupling .
we also investigate the uncertainties coming from the parton distribution functions and find that these uncertainties can be comparable to the uncertainties from the remaining scale dependence of the next - to - leading order results .
we present results separately for the different final states depending on the number of bottom quarks identified . |
the quantum theory of gravity is expected to enjoy very special properties some of which are tentatively stated by the _ holographic principle _ @xcite . for more details , we refer to @xcite . _ ( a ) holographic principle . in quantum gravity , classical lorentzian space - time with matter should emerge in such a way that the number of independent quantum states associated with the light sheets of any surface , is bounded by the exponential of the surface area .
_ associating the independent degrees of freedom with @xmath1-surfaces rather than @xmath2-volumes is in outright contrast to the familiar properties of local quantum field theory in a fixed metric background .
the following statement @xcite can therefore often be found in conjunction with ( a ) .
_ ( b ) breakdown of local quantum field theory ( qft ) .
qft whose degrees of freedom are associated with the points of space , vastly overestimates the number of degrees of freedom of the true quantum theory of gravity .
_ principle ( a ) essentially implies ( b ) . in this letter
, we explain how ( b ) can already be resolved independently of ( a ) if the diffeomorphism symmetry of general relativity is implemented into the path integral quantum theory . in order to resolve ( b ) alone ,
no new physical postulate is needed .
we first recall that any path integral quantization yields a partition function that is an invariant of smooth manifolds .
results on the classification of smooth and piecewise - linear ( pl ) manifolds show that it can be computed from a triangulation of space - time which characterizes the topology and the differentiable structure .
naively , one would assume that the degrees of freedom were associated with the simplices of the triangulation , that their number increased with the number of simplices and that the metric size of the simplices provided a short - distance cut - off .
theorems on the classification of pl - manifolds , however , imply that the triangulation in the bulk can be arbitrarily refined without affecting any physical prediction .
the actual degrees of freedom are therefore already determined by a suitable _ minimal _ triangulation .
only refinements at the boundary are physically relevant as long as the experimenters who make their observations through this boundary , can increase the resolution of their measurements . whereas the precise form of the path integral for quantum gravity in @xmath3 remains elusive
, the spin foam models in @xmath4 precisely implement the framework developed here .
all properties described so far , already follow from the symmetries and are independent of the particular choice of variables or of the action .
for simplicity , we consider general relativity without matter .
space - time is a smooth oriented @xmath0-manifold with boundary @xmath5 .
the classical theory @xcite is the study of the existence and uniqueness of ( smooth ) metric tensors @xmath6 on @xmath7 that satisfy the einstein equations subject to suitable boundary conditions . in the first order hilbert
palatini formulation , one specifies an @xmath8-connection @xmath9 together with a cotetrad field @xmath10 rather than a metric tensor .
fixing @xmath11 at the boundary , one can derive first order field equations in the bulk which are equivalent to the einstein equations provided that the cotetrad is non - degenerate .
the theory is invariant under space - time diffeomorphisms @xmath12 . for the path integral , consider a smooth @xmath0-manifold @xmath7 whose boundary @xmath13 ( figure [ fig_manifolds](a ) ) is the disjoint union of two smooth @xmath2-manifolds @xmath14 and @xmath15 to which we associate hilbert spaces @xmath16 of @xmath2-geometries , @xmath17 .
these contain suitable wave functionals of connections @xmath18 .
we denote the connection eigenstates by @xmath19 .
the path integral , @xmath20 is the sum over all connections @xmath9 matching @xmath11 , and over all @xmath10 .
it yields the matrix elements of a linear map @xmath21 between states of @xmath2-geometry .
for fixed boundary data , is diffeomorphism invariant in the bulk .
if @xmath22 are diffeomorphic , we can identify @xmath23 and @xmath24 and consider @xmath25 ( figure [ fig_manifolds](b ) ) .
provided that the trace over @xmath26 can be defined , the _ partition function _ , @xmath27 where the integral is now unrestricted , is a dimensionless number which depends only on the diffeomorphism class of the smooth manifold @xmath25 @xcite . due to the potentially infinite - dimensional hilbert space @xmath26 , the partition function may be formally divergent . in @xmath28 space - time dimensions ,
the following examples are known : the turaev viro model @xcite ( signature @xmath29 and quantized positive cosmological constant ) yields finite - dimensional hilbert spaces and a well - defined @xmath30 whereas the ponzano regge model @xcite ( signature @xmath29 and zero cosmological constant ) and the model with lorentzian signature @xmath31 @xcite have infinite - dimensional hilbert spaces and divergent @xmath30 .
we assume that we can first focus on the partition function and then learn how to fix @xmath9 on sub - manifolds in order to go back to which is the actual physically interesting object .
any @xmath32-dimensional smooth space - time manifold @xmath7 is characterized by its _
whitehead triangulation_. this is a purely combinatorial triangulation in terms of @xmath33-simplices , @xmath34 ( vertices , edges , triangles , tetrahedra , @xmath0-simplices ) , which characterizes the topology and the differentiable structure of @xmath7 .
combinatorial here means that we specify which triangles are contained in the boundary of which tetrahedra , . under well - known conditions @xcite , we can glue the simplices according to the combinatorial data and obtain a @xmath0-manifold whose differentiable structure is unique up to diffeomorphism @xcite .
conversely , any two diffeomorphic manifolds have equivalent ( pl - isomorphic ) triangulations @xcite .
if @xmath7 is compact with boundary @xmath5 , the triangulation can be chosen to consist of only a finite number of simplices .
the generic quantum geometry measurement involves such a compact @xmath7 and is given by the generalization of the r.h.s . of to generic boundary @xcite .
let @xmath35\times b^3 $ ] where @xmath36 is the @xmath2-ball .
the boundary @xmath5 is the union of @xmath37 ( initial preparation of a space - like @xmath2-geometry ) , @xmath38\times s^2 $ ] ( time - like geometry which represents a classical clock in the laboratory surrounding the experiment while the system is kept isolated ) , and @xmath39 ( final measurement of the space - like @xmath2-geometry ) .
measurements of this type involve both classical observers and a classical clock in the laboratory .
they can in principle be defined operationally .
notice that some authors call other diffeomorphism invariant expressions ` observables ' , for example , expectation values of scalars with respect to the measure of .
we do not discuss these global expressions since it is not known whether their measurement can be defined operationally .
let us sketch when triangulations are equivalent : two finite triangulations are equivalent if and only if they are related by a finite sequence of local modifications , called _ pachner moves _
@xcite , which are elementary shellings for manifolds with boundary and bistellar moves in the bulk .
the latter change the triangulation only inside some given polyhedron and replace one possible subdivision of its interior by another one .
figure [ fig_pachner ] shows the moves in @xmath40 .
the moves in @xmath41 are illustrated in detail in @xcite .
sequences of pachner moves are the combinatorial equivalent of applying space - time diffeomorphisms .
the partition function and the path integral in the bulk therefore have to be invariant under pachner moves .
the same holds , for example , for expectation values of scalars under the measure of .
having understood the moves , we can interpret the role of the triangulation physically . given the experience with lattice field theory , the naive idea would be to think that the triangulation provided a cut - off by introducing a physical length for the edges of the triangulation .
in view of the move of figure [ fig_pachner](a ) which subdivides one simplex into four and which leaves all physical predictions invariant , this can not be true !
the triangulation is rather purely combinatorial .
it does not carry any metric information .
the choice of one of several equivalent triangulations is completely analogous to the choice of one of several atlases of coordinate systems for the same smooth manifold .
it has no physical significance at all .
the metric data encoded in the path integral and in particular the expected planck scale cut - off rather have a dynamical origin .
the known models in @xmath28 dimensions mentioned above , dynamically assign metric information to the simplices of the triangulation .
the turaev viro and the ponzano regge model , for example , endow the @xmath42-simplices ( edges ) with lengths @xmath43 where @xmath44 denotes the planck length and @xmath45 is a half - integer . in @xmath46 dimensions , it is expected that the planck area @xmath47 plays the role of the fundamental geometric quantum .
as an application , let us consider a tubular space - time region @xmath48 , figure [ fig_tube](a ) , with boundary @xmath49 . figure [ fig_tube](b ) shows a slice of ( a ) for some foliation parameter @xmath50 . the experiment performed on the space - time region @xmath7 and measured through @xmath5 , requires a certain resolution which determines how many simplices are needed in @xmath5 . beyond these boundary conditions , we are free to choose any equivalent triangulation in the bulk , in particular a very coarse one , without affecting any physical prediction . figure [ fig_tube](b ) is deliberately drawn in a very suggestive way .
of course , the apparent size of the simplices in the figure is completely arbitrary .
what matters is only the combinatorics .
since the path integral dynamically assigns multiples of the planck length @xmath51 to the edges in @xmath4 or multiples of the planck area @xmath52 to the triangles in @xmath3 , there is the following indirect bound on the number of simplices in @xmath5 that is compatible with a given classical geometry .
whenever the quantum theory of gravity on @xmath7 has a classical limit in which the area of @xmath53 is @xmath9 , it makes sense to increase the number @xmath54 of triangles in @xmath53 only up to @xmath55 .
triangulations with more triangles in @xmath5 are always assigned larger total areas and their states therefore be suppressed in the very same classical limit .
we have seen that the degrees of freedom are associated with the simplices only up to equivalence of triangulations . in particular
, their number does not increase with the volume of the spatial region @xmath36 ( consider @xmath56 as ` time ' ) .
the minimum number of simplices is rather determined by the boundary conditions we impose on @xmath53 , and the number of simplices in the boundary is naturally bounded by @xmath57 whenever there emerges a classical geometry such that @xmath53 has a total area of @xmath9 .
standard qft , in contrast , essentially assigns harmonic oscillators , each with an infinite - dimensional hilbert space , to the uncountably many points of space .
we have reduced this to hilbert spaces associated with only a finite number of simplices .
the major conceptual difference is that standard qft constructs hilbert spaces for the fields _ relative _ to a fixed riemannian @xmath2-manifold , whereas the hilbert spaces of represent equivalence classes of @xmath2-geometries up to spatial diffeomorphisms .
we note that none of our arguments changes substantially if we add matter or gauge fields or if we formulate the classical theory in terms of other variables .
recall that the symmetry under space - time diffeomorphisms is not special to general relativity , but shared by many other classical field theories , indicating a potential impact on particle physics in general .
further details on path integral quantization of general relativity and invariants of smooth manifolds will be presented in @xcite .
999 g. t hooft : dimensional reduction in quantum gravity . in _ salamfestschrift : a collection of talks _ , eds .
a. ali , j. ellis and s. randjbar - daemi .
world scientific , singapore , 1993 , pp .
284296 , ` gr - qc/9310026 ` .
r. bousso : the holographic principle .
* 74 * ( 2002 ) 825874 , ` hep - th/0203101 ` .
r. m. wald : general relativity .
the university of chicago press , chicago , 1984 .
m. f. atiyah : topological quantum field theories .
_ ihes publ .
_ * 68 * ( 1988 ) 175186 .
j. w. barrett : quantum gravity as a topological quantum field theory
. _ j. math .
phys . _ * 36 * , no .
11 ( 1995 ) 61616179 , ` gr - qc/9506070 ` .
l. crane : clock and category : is quantum gravity algebraic ? _ j. math
. phys . _ * 36 * , no .
11 ( 1995 ) 61806193 , ` gr - qc/9504038 ` . v. g. turaev and o. y. viro : state sum invariants of @xmath2-manifolds and quantum @xmath58-symbols .
_ topology _ * 31 * , no . 4 ( 1992 ) 865902 .
g. ponzano and t. regge : semiclassical limit of racah coefficients . in _ spectroscopic and group theoretical methods in physics _ ,
f. bloch .
north holland publications , amsterdam , 1968 , pp .
l. freidel : a ponzano regge model of lorentzian 3-dimensional gravity .
* b 88 * ( 2000 ) 237240 , ` gr - qc/0102098 ` .
s. davids : a state sum model for @xmath59 lorentzian quantum gravity .
thesis , university of nottingham ( 2001 ) .
preprint ` gr - qc/0110114 ` .
rourke and b. j. sanderson : introduction to piecewise - linear topology .
ergebnisse der mathematik und ihrer grenzgebiete 69 .
springer , berlin , 1972 .
m. w. hirsch and b. mazur : smoothings of piecewise linear manifolds .
annals of mathematics studies 80 .
princeton university press , princeton , 1974 .
n. h. kuiper : on ` la nullit de @xmath60 ' by j. cerf . mathematical reviews
6641 a d vol . 33
( 1963 ) 1126 . j. h. c. whitehead : on @xmath61-complexes .
_ * 41 * , no . 4 ( 1940 ) 809824 .
r. oeckl : schrdinger s cat and the clock : lessons for quantum gravity .
_ class .
grav . _ * 20 * ( 2003 ) 53715380 , ` gr - qc/0306007 ` .
u. pachner : pl homeomorphic manifolds are equivalent by elementary shellings . _ eur .
j. comb . _
* 12 * ( 1991 ) 129145 .
j. s. carter , l. h. kauffman and m. saito : structures and diagrammatics of four dimensional topological lattice field theories .
_ * 146 * ( 1999 ) 39100 , ` math.gt/9806023 ` .
h. pfeiffer : quantum general relativity and the classification of smooth manifolds ( 2004 ) .
preprint ` gr - qc/0404088 ` . | if the diffeomorphism symmetry of general relativity is fully implemented into a path integral quantum theory , the path integral leads to a partition function which is an invariant of smooth manifolds .
we comment on the physical implications of results on the classification of smooth and piecewise - linear @xmath0-manifolds which show that the partition function can already be computed from a triangulation of space - time .
such a triangulation characterizes the topology and the differentiable structure , but is completely unrelated to any physical cut - off .
it can be arbitrarily refined without affecting the physical predictions and without increasing the number of degrees of freedom proportionally to the volume .
only refinements at the boundary have a physical significance as long as the experimenters who observe through this boundary , can increase the resolution of their measurements .
all these are consequences of the symmetries .
the planck scale cut - off expected in quantum gravity is rather a dynamical effect . |
the concept of `` holography '' is first introduced by d. gabor in 1948 @xcite as a new technique of the optical physics to playback the three - dimensional information onto the two - dimensional plate as a hologram . in 1997 , this concept of holography comes to get a new appearance in the framework of the superstring theory as the duality between two theories belonging to the different spatial dimensions .
it is first proposed by maldacena @xcite as the ads / cft correspondence between @xmath7 supergravity and @xmath8=4 susy yang - mills theory through @xmath9 brane in type iib superstring theory . in more general point of view , an essential element of the holography is `` @xmath10 brane '' as the @xmath11-dimensional membrane in the ten - dimensional space - time .
in fact , @xmath10 brane appears as the soliton , i.e. , the condensed object of the fundamental strings .
the @xmath10 brane has two important aspects as follows : @xmath11-dimensional gauge theory appears _ on _ a surface of the @xmath10 brane , and @xmath12-dimensional supergravity appears _ around _
the @xmath10 brane .
( the italic `` _ 1 _ '' denotes the radial dimension with non - trivial curvature around the @xmath10 brane , indicating the existence of the gravity ) .
actually , the concept of holography indicates the duality between the @xmath11-dimensional gauge theory without the gravity and @xmath12-dimensional supergravity mediated by the @xmath10 brane , and the gauge interaction as a `` hologram '' on the surface of the @xmath10 brane is to give the supergravity as a `` vision '' in the extra - dimension .
one of the most essential properties of the holography is the `` strong - weak duality '' between the gauge theory and the supergravity : the coupling strengths are transversely related with each other .
therefore the holography provides a remarkable possibility that non - perturbative aspects of one side can be analyzed by the other dual side just with the tree - level calculations .
then , if we find the special configurations of @xmath13 branes reflecting qcd on their surfaces , non - perturbative aspects of qcd can be successfully examined from the tree - level dual supergravity side .
this is the strategy of the holographic qcd .
there exist several trials to find the special configurations of @xmath13 branes reflecting qcd .
eventually , in 2005 , sakai and sugimoto succeeded in constructing qcd with massless quarks and gluons from the fluctuation modes of the open strings on the d4/d8/@xmath14 multi - d brane configurations in type iia superstring theory @xcite , called sakai - sugimoto model , which is one of the most realistic model of holographic qcd . by using this model , many phenomenological properties of _ mesons _ belonging to the non - perturbative aspects of qcd like meson mass spectra , hidden local symmetry @xcite , vector meson dominance @xcite , ksrf relation @xcite , gsw model @xcite , etc .
, are successfully derived from the tree - level dual supergravity calculations . in this sense , holographic qcd is often regarded as the `` unified meson theory '' . on the other hand ,
baryon description is not straightforward in this approach since the classical supergravity is found to be dual with the strong - coupling `` large-@xmath3 '' qcd , where baryons do not directly appear as dynamical degrees of freedom @xcite . in our previous work
, we gave the first study of the baryon as a non - trivial topological soliton in the four - dimensional meson effective action derived from holographic qcd .
we call this topological soliton as a `` brane - induced skyrmion '' @xcite . especially we included pions and @xmath1 mesons appearing below the kaluza - klein mass scale @xmath2 , which is often called the `` truncated - resonance model '' for the baryon analysis .
actually , @xmath15 plays the role as the ultraviolet cutoff scale of the holographic approach .
in fact , there appear infinite number of `` non - qcd modes '' with mass scale @xmath16 in holographic qcd like gluinos and kaluza - klein modes . in this sense
, the duality with qcd could be maintained below @xmath15 as the ultraviolet cutoff .
the appearance of certain cutoff scale @xmath15 should be essential for the holographic approach to be dual of realistic qcd with confinement and chiral symmetry breaking as the non - susy natures . in the holographic model with @xmath0 multi - d brane system ,
d4 branes are @xmath17-compactifiled with @xmath15 scale , to give the complete susy breaking and its resulting non - conformal natures of qcd like finite string tension and chiral condensate .
these considerations suggest that @xmath15 should be so much respected , giving our truncation of meson resonances at @xmath15 for baryon analysis ( see , sec .
[ mea ] for details ) . recently
a baryon is also described as an `` instanton '' on the five - dimensional gauge theory of d8 branes with d4 supergravity background @xcite .
the instanton is introduced _ before _ the mode expansion of the five - dimensional gauge field into mesons , so that the baryon as the instanton is to be composed by the infinite number of color - singlet modes with mesonic quantum number even above @xmath15 scale .
however , we consider that such color - singlet modes above @xmath15 might not directly correspond to physical mesons in qcd , because the duality with qcd is mentained below @xmath15 . furthermore , there also exist infinite number of other non - qcd modes above @xmath15 , which could also affect the baryon properties if they were included . therefore , in contrast to instanton models , we severely respect the cutoff scale @xmath15 and truncate the meson resonances at @xmath15 , to keep the duality with qcd .
more comprehensive discussions with instantons are summarized in sec .
[ so ] . in present work ,
we newly consider the extension of the holographic model to dense qcd . due to the non - abelian nature of qcd ,
various realizations are expected in the vacuum itself with finite temperature and density , called `` qcd phase diagram '' . up to now ,
interesting phase structures are proposed by using some low - energy effective theories of qcd , e.g. , confined phase with mesons and baryons , deconfined phase with quark - gluon plasma ( qgp ) , chiral symmetry broken phase with mass generation @xcite , color superconductivity as diquark condensation @xcite , etc .
in fact , some wisdom about qgp gives the insight for the early universe just after the big bang @xcite .
furthermore , the possible qcd phase transitions in the core region of neutron stars could affect a lot of their macroscopic features like moment of inertia , angle velocity , and breaking index @xcite .
there also exist several experimental projects to search the qgp in the ultra - relativistic heavy ion collisions in rhic ( relativistic heavy ion collider ) at bnl , and lhc ( large hadron collider ) at cern .
there will also appear relatively low - energy collision experiments to make the low - temperature high - density object in fair ( facility for antiproton and ion research ) at gsi , giving some knowledge about the core region of the compact stars . with these background
, it should be urgently important to make clear the structure of qcd phase diagram more explicitly from qcd itself with the rich help of experimental data , which will eventually bring about the fundamental understanding of our whole nature .
there exists the lattice qcd numerical study as the first principle calculation of the strong interaction .
however , because of the `` sign problem '' , its applicability is severely restricted near the zero - density at finite - temperature regime of the wide qcd phase diagram ( for some review , see , ref .
@xcite ) . therefore ,
if one succeeds in the extension of holographic approach to the dense regime , it should give a new analytical tool for non - perturbative aspects of the finite density qcd , where the holography provides the duality between the strong - coupling gauge theory and the weak - coupling supergravity .
this is the main aim of our study . in this work ,
we consider the baryonic matter in holographic qcd as the extension of the holographic approach to dense qcd .
especially , we treat the baryonic matter with large-@xmath3 because the holographic qcd is derived as a large-@xmath3 effective theory . as the general property of large-@xmath3 qcd @xcite ,
the static baryon mass is proportional to @xmath18 , so that its kinetic energy becomes @xmath19 .
there also exist the quantum effects like the zero point quantum fluctuation energy @xmath20 and also the baryon mass splitting @xmath21 within the baryonic matter , while these correspond to the higher order contributions of the @xmath22 expansions as @xmath23 and @xmath24 @xcite .
such large-@xmath3 countings indicate that , for sufficiently large @xmath3 , the kinetic energy and the quantum effects within the baryonic matter can be suppressed relative to the static mass , and the baryonic matter comes into the static skyrme matter .
such static skyrme matter was first analyzed by klebanov @xcite , by placing the skyrme soliton solutions periodically along the three - dimensional cubic lattice , which would correspond to the nuclear crystal in the deeper interior of neutron stars .
since the cubic lattice treatment is rather cumbersome , we employ a mathematical trick to analyze such static skyrme matter , proposed by manton and ruback @xcite . in order to represent some high - density state of the multi - skyrmion system on the three - dimensional flat space @xmath25 , single skyrmion
is alternately placed on a surface of the three - dimensional closed manifold @xmath4 with a finite radius .
in fact , the multi - skyrmion system on @xmath25 and the single skyrmion on @xmath4 can be related with each other through the compactification of the boundary of a unit cell on @xmath25 shared by one skyrmion .
the interactions between baryons on @xmath25 are simulated by the curvature of the manifold @xmath4 , and decreasing the radius of @xmath4 represents the increase of the baryon - number density in the medium in this modeling .
actually , by taking such mathematical simplification , one could avoid some complicated analysis like monte carlo simulations on the three - dimensional cubic lattice @xcite , and he can get some physical intuitions as for the baryonic matter qualitatively and even quantitatively @xcite . therefore , by placing the single brane - induced skyrmion on the closed manifold @xmath4 , we try to analyze the typical features of baryonic matters in holographic qcd .
especially , in this analysis , the roles of @xmath1 mesons in the dense baryonic matter will be examined in detail from the holographic point of view .
there exist a lot of works about the extension of the holographic approaches to dense susy qcd ( see , ref .
@xcite and references therein ) . with the bottom - up construction of the ads / qcd models @xcite ,
it was applied to dense qcd by introducing the bi - nucleon condensate @xcite .
after the discovery of the sakai - sugimoto model @xcite as one of the most reliable holographic top - down approaches to non - susy qcd with massless quark flavors , there exist several proposals about the extension of this model to dense qcd .
for example , in refs .
@xcite , the baryon chemical potential @xmath26 is introduced by the asymptotic value of a @xmath27 gauge field on the d8-@xmath28 branes as @xmath29 , similarly to the introduction of the chemical potentials to the chiral perturbation theory by promoting the global chiral symmetry to local gauge one @xcite . in these holographic analyses , the density dependence of the `` local '' properties of mesons and baryons like their masses and coupling constants , and also the phase structure of qcd
have been successfully discussed .
now , in our paper , we treat the baryon as a `` non - local '' solitonic object as the skyrmion in the sakai - sugimoto model , and we analyze the dense qcd by the skyrme matter . therefore , adding to the informations about the local natures of hadrons and qcd phase structure , we can extensively examine the internal structure of the baryon like the size and the field configurations .
here we show the organization of this paper and its brief summary . in sec .
[ mea ] , we overview the holographic derivation of the four - dimensional meson effective action with the pion and @xmath1 meson fields , reemphasizing the role of cutoff scale @xmath15 in the holographic framework to be dual of qcd . in sec .
[ bisrs ] , we analyze the properties of baryons and baryonic matter in holographic qcd . in sec .
[ bis_r3 ] , we introduce the concept of the brane - induced skyrmion on the three - dimensional flat space @xmath25 . then , in sec
. [ bis_s3 ] , we describes the baryonic matter as the system of single brane - induced skyrmion on the three - dimensional closed manifold @xmath4 .
we derive the expression of the hedgehog mass and euler - lagrange equations for the pion and @xmath1 meson fields as the brane - induced skyrmion on @xmath4 .
[ nr ] is devoted to the numerical results and their physical interpretations about the baryon nature in dense qcd . in sec .
[ ecpro ] , the baryon - number density dependence of the energy density and field configuration profiles of single baryon are discussed . in sec .
[ pdnc ] , we propose a new striking picture of the `` pion dominance '' near the critical density , i.e. , all the ( axial ) vector meson fields disappear and only pion field survives . in sec .
[ mp ] , the baryon - number density dependence of the mass and root - mean - square mass radius of single baryon are analyzed .
we find some non - linear increase in the size of the baryon near the critical density as a `` swelling '' phenomena . in sec .
[ smpi ] , we explain the swelling mechanism in the general context of qcd , and consider its effects on the stability of @xmath30-@xmath31 mixed matter . in sec .
[ crdpt ] , we examine the features of the delocalization phase transitions and the chiral symmetry restoration by choosing proper order parameters , through which the relations between deconfinement and chiral symmetry restoration are reconsidered . in sec .
[ cdt ] , we calculate the critical densities of the phase transitions in the physical units with the experimental inputs for the pion decay constant @xmath32 and @xmath1 meson mass @xmath33 .
we find the critical density @xmath34 in the holographic approach . through all of the sections , by comparing the brane - induced skyrmion and standard skyrmion without @xmath1 meson fields , the roles of the vector mesons in the dense baryonic matter are examined from the holographic point of view .
[ so ] is devoted to summary and outlook . in this final section ,
we compare our truncated - resonance approach with another `` instanton '' description of baryons @xcite in the holographic qcd , paying attention on the role of cutoff scale @xmath35 .
in this section , we overview the derivation of the four - dimensional meson effective action from holographic qcd with @xmath0 multi-@xmath13 brane system in type iia superstring theory , called the sakai - sugimoto model @xcite .
the meson effective action is derived without small amplitude expansion to discuss a baryon as a large amplitude chiral soliton .
( for more comprehensive derivations , see our previous paper @xcite ) . here
we especially emphasize the roles of cutoff scale @xmath15 in holographic qcd .
first , we review the construction of @xmath0 multi - d brane system and also the supergravity description of d4 branes as the holographic dual of qcd . as the first step ,
@xmath3 sheets of d4 branes are prepared to construct the gluon sector of qcd .
the d4 branes are @xmath36-compactified along one extra - dimension with a radius as the inverse of the kaluza - klein mass scale @xmath15 .
there appear ten independent fluctuation modes from the open strings on the surface of d4 branes , i.e. , gauge fields @xmath37 , scalar fields @xmath38 and @xmath39 , and also their superpartners as fermions . by imposing the anti - periodic boundary conditions for all the fermions along the @xmath36-compactified direction
, they acquire large masses @xmath16
. then supersymmetry ( susy ) is completely broken and , due to the radiative corrections , all the scalar fields @xmath38 and @xmath40 also get large masses @xmath16 . because of the @xmath36-compactification , there also appear the infinite number of the kaluza - klein modes with masses @xmath16 .
therefore , below the @xmath15 scale , only massless gauge fileds @xmath41 appears . in this sense
, the system of @xmath3 sheets of d4 branes with the @xmath36-compactification can be viewed as the @xmath42 yang - mills theory below @xmath15 scale , corresponding to the pure gauge sector of qcd . as the next step , @xmath43 sheets of d8 and @xmath28 branes
are added to introduce the massless quark flavors of qcd . @xmath28
has opposite chirality relative to d8 , providing @xmath44 chiral symmetry in this model . from the fluctuation modes of open strings between d4 and d8 ( @xmath28 ) ,
there appear massless chiral fermions as quarks in qcd . as a whole
, massless qcd appears as the `` hologram '' on the surface of @xmath0 branes .
then we shift into the gravitational description of d branes from the extra - dimensions .
the d brane is originally introduced as the fixed edges of open strings with dirichlet boundaries .
this also indicates that , through the `` open - closed duality '' for the fundamental strings , the d brane can also be regarded as the source of closed strings , giving the graviton in the extra - dimensions outside of the d brane . in this sense ,
the d brane can be identified as a highly gravitational system , i.e. , the `` black brane '' , allowing the gravitational description from the extra - dimensions .
in fact , the mass of the d branes is proportional to its sheets number , so that , by assuming @xmath45 , only d4 branes can be represented by the gravitational background and d8 ( @xmath28 ) branes are introduced as the probes called `` probe approximation '' , which corresponds to the quenched approximation in lattice qcd study @xcite . especially the classical supergravity description of d4 branes is tractable , followed by the local approximation of the strings and also the suppressions of the string loop effect .
these conditions in the gravitational side around the d branes give the constraints for the gauge theory side as the qcd on the surface of d branes as @xmath46 which is achieved by @xmath47 , @xmath48 , and thooft coupling : @xmath49 fixed and large . in this sense
, the strong - coupling large-@xmath3 qcd is found to be dual with the classical supergravity of d4 branes with probe d8 ( @xmath28 ) branes , which is one of the realization of the strong - weak duality between the gauge theory and gravitational theory through the holography .
therefore , by analyzing the effective action of d8 branes with d4 supergravity background around the d branes , one can analyze the nonperturbative aspects of qcd on the surface of d branes .
then we start formal discussions from the @xmath50 non - abelian dirac - born - infeld ( dbi ) action of probe @xmath51 brane with @xmath52 supergravity background as a probe approximation . after dimensional reductions
, the nine - dimensional dbi action of the probe @xmath51 brane with the @xmath52 supergravity background can be reduced into a five - dimensional yang - mills theory , belonging to the flat four - dimensional euclidean space - time @xmath53 , i.e. , @xmath54 and the other fifth dimension @xmath55 with curved measures as follows @xcite : @xmath56 where @xmath57 is the gauge field and @xmath58 } $ ] ( @xmath59 ) is the field strength tensor in five - dimensional space - time @xmath60 of the probe @xmath51 branes . in the action ( [ 5dimdbi_2 ] ) ,
@xmath61 unit is taken , and the overall factor @xmath62 is defined as @xmath63 ( note here that we use the value of @xmath62 as a half of that in ref .
@xcite , taking away the misleading factor @xmath64 in eq.(16 ) of ref .
all the formula and numerical results can be scaled by the factor @xmath62 , so that the discussions in ref .
@xcite are not altered . )
the functional @xmath65 in the action ( [ 5dimdbi_2 ] ) expresses the nontrivial curvature in the extra fifth dimension @xmath55 induced by the supergravity background of the d4 brane .
the gravitational energy of @xmath51 brane , i.e. , @xmath66 is subtracted in the action ( [ 5dimdbi_2 ] ) as the vacuum relative to the gauge sectors . in the @xmath0 multi-@xmath13 brane configurations ,
color quantum number is carried only by the @xmath3-folded @xmath52 branes .
therefore , after the supergravity description of @xmath52 branes , there is no colored particles from the fluctuation modes of open strings on the residual probe @xmath51 branes .
this is regarded as some holographic manifestation of `` color confinement '' in low - energy scale of qcd .
in fact , gauge field @xmath67 in the action ( [ 5dimdbi_2 ] ) is color - singlet and obeys the adjoint representation of @xmath68 group , eventually producing the meson degrees of freedom after some proper mode expansions in the holographic qcd . in this paper
, we treat the non - trivial leading order of @xmath22 and @xmath69 expansions in the holographic qcd as the five - dimensional yang - mills action ( [ 5dimdbi_2 ] ) with @xmath70 . in general
, there also exists the chern - simons ( cs ) term to avoid anomalies in the superstring theory . by introducing @xmath71 meson degrees of freedom as the @xmath72 sector of @xmath1 meson fields in the holographic approach , @xmath71 meson indirectly couples with pions via ( axial ) vector mesons after the proper mode expansions of cs term , which is regarded as a general representation of the gell - mann - sharp - wagner ( gsw ) model @xcite .
the roles of @xmath71 mesons for low - energy meson dynamics and also chiral solitons have been traditionally examined in some of qcd phenomenologies @xcite .
actually , however , the cs term is found to be @xmath73 , i.e. , the higher order contributions of thooft coupling expansion relative to the @xmath70 of the yang - mills action ( [ 5dimdbi_2 ] ) with @xmath74 , which is manifestly shown in holographic qcd .
furthermore , most part of cs term includes one time - derivative of pion fields and does not affect the static properties of hedgehog solitons .
therefore we neglect the cs term along the discussions below for the argument of the non - perturbative ( strong coupling ) properties of qcd . in the holographic approach
, the pion field is introduced as the wilson line of the fifth gauge field @xmath75 , i.e. , a path - ordered product of the fifth gauge field along the @xmath55 direction @xcite as @xmath76 one can also introduce the variables @xmath77 as @xmath78 where @xmath79 is a single - valued arbitrary function of @xmath80 .
then the pion field ( [ pion1 ] ) can be written as @xmath81 some resemble formula of which can also be found in the traditional approach of hidden local symmetry @xcite .
now we take `` @xmath82 gauge '' and also `` @xmath83 gauge '' for the @xmath68 gauge symmetry in the action ( [ 5dimdbi_2 ] ) of the probe @xmath51 brane .
the @xmath82 gauge is similar to the unitary gauge in the non - abelian higgs theory ; fifth gauge field @xmath75 performs as a scalar field in four - dimensional space - time @xmath84 , and it is eaten by the four dimensional gauge field @xmath85 to give the mass generation of gauge field , especially the ( axial ) vector mesons as a part of @xmath85 . in this sense
the masses of the ( axial ) vector mesons come from the higgs mechanism in five - dimensional space - time with the @xmath68 gauge symmetry breaking on the probe @xmath51 brane .
the @xmath86 gauge is also essential to get the low - energy effective theory of qcd with proper parity and g - parity classification in a manifest way @xcite . with these gauge fixings ,
the five - dimensional gauge field @xmath87 can be mode - expanded into the four - dimensional parity and g - parity eigen states with proper complete orthogonal basis @xmath88 and @xmath89 ( @xmath90 ) as follows @xcite : @xmath91 where @xmath92 and @xmath93 are left and right currents of pion fields , respectively .
the basis @xmath88 are introduced to support whole of the gauge field @xmath87 at the boundary @xmath94 as @xmath95 and @xmath96 as @xmath97 in order to diagonalize the five - dimensional yang - mills action ( [ 5dimdbi_2])with the induced measures @xmath98 and @xmath99 in the fifth dimension @xmath55 , the basis @xmath100 @xmath101 are taken to be the normalizable eigen - function satisfying @xmath102 with normalization condition as @xmath103 in the holographic model , the fields @xmath104 in the mode expansion ( [ limit2_lr ] ) are regarded as ( axial ) vector mesons , belonging to the adjoint representation of the @xmath68 gauge group as @xmath105 . by substituting the expansion ( [ limit2_lr ] ) into the action ( [ 5dimdbi_2 ] ) , the mass of @xmath106 field is found with the eigenvalue of oscillating fifth basis in eq.([eigen1 ] ) as @xmath107 , indicating that the origin of meson mass is the oscillation of meson wave function in the extra fifth dimension .
furthermore , @xmath108 is the five - dimensional vector and @xmath89 are the parity eigen state in the @xmath55 direction as @xmath109
. therefore , from the mode expansion ( [ limit2_lr ] ) , @xmath106 fields have four - dimensional parity transformation as @xmath110 .
these consideration indicates that vector and axial vector mesons appear alternately in the excitation spectra about index @xmath111 as @xmath112 . in our study
, we construct the four - dimensional meson effective action only with pion field @xmath113 and @xmath1 meson field @xmath114 below the kaluza - klein mass scale @xmath2 .
recall that there appear infinite number of non - qcd modes with large mass @xmath16 in holographic qcd , e.g. , scalar fields , gluino , and also the kaluza - klein modes , discussed in the first part of this section .
therefore , the @xmath0 multi - d brane system can be viewed as qcd as far as low energy phenomenology below @xmath115 is concerned . in this sense , @xmath15 plays the roles as the ultraviolet cutoff scale of the theory , so that we include the meson degrees of freedom below @xmath15 for the baryon analysis in later sections , called `` truncated - resonance model '' .
actually , the appearance of @xmath15 scale with finite value seems to be essential in the recent holographic analysis . in the framework of ads
/ cft correspondence without kaluza - klein compactification , the @xmath116 susy and its resulting conformal symmetry protect the emergence of the dimensional quantities like string tension and chiral condensate at the ground state , because of the cancellation of the radiative corrections between bosons and fermions . in this sense ,
susy breaking is at lest needed to be dual of qcd with confinement and chiral symmetry breaking as its vacuum nature . in the sakai - sugimoto model , @xmath3 sheets of d4 branes
are kaluza - klein compactified with radius @xmath117 , and the susy is completely broken by the field boundary condition along the compactified direction . therefore confinement and chiral symmetry breaking _ could _ occur as the non - susy gauge theory .
in fact , the compactified d4 brane is `` non - bps '' , having a `` horizon '' in the supergravity description . in the holographic framework , confinement and chiral symmetry breaking _
do _ occur on this horizon , giving the lost of `` colored '' information , and also the geometrical connection of d8 and @xmath28 branes . as a whole , the appearance of @xmath15 would be essential for the holographic model to be dual of realistic qcd .
now , one may regret about the finiteness of @xmath15 as almost 1gev , which is comparable with the qcd mass scale @xmath118 . in the holographic approach ,
the qcd mass scale is introduced by the experimental inputs for the pion decay constant and @xmath1 meson mass as @xmath119 and @xmath120 .
these experimental values come from our hadronic world with @xmath121 and large but finite thooft coupling @xmath122 , which may infringe the condition ( [ cond1 ] ) to give the effects of string length and string loops in the gravitational side .
therefore , too large @xmath15 can not be taken to neglect the internal structure of strings on the surface of d4 branes compactified with radius @xmath117 , which might eventually gives the scale @xmath123 .
these considerations indicate that , by including the effects of string length and loops in the gravitational side , @xmath15 would be taken sufficiently large relative to @xmath118 with fixed @xmath124 and @xmath125 .
anyway , @xmath123 essentially appears as the ultraviolet cutoff scale in the system of probe d8 brane with d4 `` classical '' supergravity background to be dual of qcd with proper dimensional quantities . by neglecting the higher mass excitation modes of ( axial ) vector mesons rather than @xmath1 meson sector with @xmath126 in the expansions ( [ limit2_lr ] )
, the five - dimensional gauge field @xmath87 can be written as @xmath127 by taking this mode expansion ( [ exp_rhomode ] ) with the @xmath82 gauge , five - dimensional field strength @xmath128 and @xmath129 can be written as @xmath130 } \nonumber\\ & = & \!\!{\left ( } \partial_\mu l_\nu-\partial_\nu l_\mu { \right)}\psi_+ + { \left ( } \partial_\mu r_\nu-\partial_\nu r_\mu { \right)}\psi_- + { \left ( } \partial_\mu \rho_\nu-\partial_\nu \rho_\mu { \right)}\psi_1 + i{\left\ { } \newcommand{\rtk } { \right\ } } [ l_\mu , l_\nu ] \psi_+^2 + [ r_\mu , r_\nu ] \psi_-^2 + [ \rho_\mu,\rho_\nu ] \psi_1 ^ 2 \rtk\nonumber\\ & & + i\ { { \left ( } [ l_\mu , r_\nu ] + [ r_\mu , l_\nu ] { \right)}\psi_+\psi_- + { \left ( } [ l_\mu,\rho_\nu ] + [ \rho_\mu , l_\nu ] { \right)}\psi_+\psi_1 + { \left ( } [ r_\mu,\rho_\nu ] + [ \rho_\mu , r_\nu ] { \right)}\psi_-\psi_1 \}\nonumber\\ & = & \!\!-i[\alpha_\mu,\alpha_\nu]\psi_+\psi_- + { \left ( } \partial_\mu\rho_\nu -\partial_\nu \rho_\mu { \right)}\psi_1 + i[\rho_\mu,\rho_\nu]\psi_1 ^ 2 + i\{{\left ( } [ \alpha_\mu,\rho_\nu]+[\rho_\mu,\alpha_\nu]{\right)}\hat{\psi}_0\psi_1\nonumber\\ & & \hspace{-1.5mm}+{\left ( } [ \beta_\mu,\rho_\nu]+[\rho_\mu,\beta_\nu]{\right)}\psi_1 \},\label{fieldst1}\\ f_{z\mu}\!\ ! & = & \!\!\partial_z a_\mu = \alpha_\mu\partial_z\hat{\psi}_0+\rho_\mu\partial_z\psi_1,\label{fieldst2 } \end{aligned}\ ] ] with axial vector current @xmath131 and vector current @xmath132 of pion field as @xmath133 in the derivation of ( [ fieldst1 ] ) and ( [ fieldst2 ] ) , we have used the maurer - cartan equations , @xmath134=0 $ ] and @xmath135=0 $ ] . by substituting eqs.([fieldst1 ] ) and ( [ fieldst2 ] ) into the five - dimensional yang - mills action ( [ 5dimdbi_2 ] ) with @xmath70 , we eventually get the four - dimensional euclidean meson effective action with pions and @xmath1 mesons from holographic qcd as follows ( derivations in more detail can be found in our previous paper @xcite ) : @xmath136 @xmath137 } ^2 + \frac{1}{2 } \int d^4 x \mbox{tr}{\left ( } \partial_{\mu}\rho_{\nu}-\partial_{\nu}\rho_{\mu}{\right)}^2 \nonumber\\ & & + ig_{3\rho } \int d^4 x \mbox{tr}{\left\ { } \newcommand{\rtk } { \right\ } } { \left ( } \partial_{\mu}\rho_{\nu}-\partial_{\nu}\rho_{\mu}{\right)}{\left [ } \rho_{\mu } , \rho_{\nu } { \right ] } \rtk - \frac{1}{2}g_{4\rho } \int d^4 x \mbox{tr}{\left [ } \rho_{\mu } , \rho_{\nu } { \right ] } ^2 -i g_1 \int d^4 x \mbox{tr}{\left\ { } \newcommand{\rtk } { \right\ } } { \left [ } \alpha_{\mu } , \alpha_{\nu } { \right ] } { \left ( } \partial_{\mu}\rho_{\nu}-\partial_{\nu}\rho_{\mu}{\right)}\rtk \nonumber\\ & & + g_2 \int d^4 x \mbox{tr}{\left\ { } \newcommand{\rtk } { \right\ } } { \left [ } \alpha_{\mu } , \alpha_{\nu } { \right ] } { \left [ } \rho_{\mu } , \rho_{\nu } { \right ] } \rtk + g_3 \int d^4 x \mbox{tr}{\left\ { } \newcommand{\rtk } { \right\ } } { \left [ } \alpha_{\mu } , \alpha_{\nu } { \right ] } { \left ( } { \left [ } \beta_{\mu } , \rho_{\nu } { \right ] } + { \left [ } \rho_{\mu } , \beta_{\nu } { \right ] } { \right)}\rtk \nonumber\\ & & + i g_4 \int d^4 x \mbox{tr}{\left\ { } \newcommand{\rtk } { \right\ } } { \left ( } \partial_{\mu}\rho_{\nu}-\partial_{\nu}\rho_{\mu}{\right)}{\left ( } { \left [ } \beta_{\mu } , \rho_{\nu } { \right ] } + { \left [ } \rho_{\mu } , \beta_{\nu } { \right ] } { \right)}\rtk - g_5 \int d^4 x \mbox{tr}{\left\ { } \newcommand{\rtk } { \right\ } } { \left [ } \rho_{\mu } , \rho_{\nu } { \right ] } { \left ( } { \left [ } \beta_{\mu } , \rho_{\nu } { \right ] } + { \left [ } \rho_{\mu } , \beta_{\nu } { \right ] } { \right)}\rtk \nonumber\\ & & -\frac{1}{2 } g_6 \int d^4 x \mbox{tr}{\left ( } { \left [ } \alpha_{\mu } , \rho_{\nu } { \right ] } + { \left [ } \rho_{\mu } , \alpha_{\nu } { \right ] } { \right)}^2 -\frac{1}{2 } g_7 \int d^4 x \mbox{tr}{\left ( } { \left [ } \beta_{\mu } , \rho_{\nu } { \right ] } + { \left [ } \rho_{\mu } , \beta_{\nu } { \right ] } { \right)}^2 , \label{f11_1}\end{aligned}\ ] ] where @xmath138 is the 1-form of pion fields as @xmath139 there exist twelve kinds of coupling constants : @xmath124 , @xmath125 , @xmath140 , @xmath141 , @xmath142 and @xmath143 in the action ( [ f11_1 ] ) .
however all the coupling constants are uniquely determined from the properties of meson wave functions in the extra fifth dimension @xmath55 , i.e. , the complete orthogonal basis @xmath144 and @xmath145 with oscillating eigen value @xmath146 as follows : @xmath147 the holographic model has two parameters @xmath148 and the kaluza - klein mass @xmath15 as the ultraviolet cutoff scale of this theory .
@xmath62 appears in front of the effective action ( [ 5dimdbi_2 ] ) because the effective action of @xmath51 brane with @xmath52 supergravity background expanded up to @xmath70 corresponds to the leading order of @xmath22 and @xmath69 expansions .
therefore , by fixing two parameters @xmath62 and @xmath6 to adjust experimental inputs for @xmath124 and @xmath125 , then all the coupling constants ( [ f_pi_1])@xmath149([g_7 ] ) are uniquely determined through the background of extra fifth dimension @xmath55 .
such uniqueness of the action is one of the remarkable consequences in holographic qcd .
in this section , we discuss baryons and baryonic matter in holographic qcd . in sec .
[ bis_r3 ] , we describe the baryon as a chiral soliton in the four - dimensional meson effective action @xmath150 in ( [ f11_1 ] ) including pion and @xmath1 meson fields derived from holographic qcd .
we call this topological soliton as the `` brane - induced skyrmion '' . the hedgehog mass of the brane - induced skyrmion on the flat coordinate space @xmath25 is derived , which is originally given in our previous paper @xcite . in sec .
[ bis_s3 ] , we newly discuss the baryonic matter in holographic qcd by analyzing the system of single brane - induced skyrmion on the three - dimensional closed manifold @xmath4 . through the projection procedure from the flat space @xmath25 onto the curved space @xmath4 , the hedgehog mass and the euler - lagrange equations of the brane - induced skyrmion on @xmath4 are derived .
all the numerical results and their physical interpretations are presented in secs .
[ nr ] , [ crdpt ] , [ cdt ] . in this work
, we describe the baryon as the four - dimensional chiral soliton , i.e. , the skyrmion in holographic qcd . to see the validity of this approach for the baryon
, we now compare the meson effective action induced by holographic qcd with that in a chiral perturbation theory ( chpt ) as a low - energy effective theory of qcd @xcite .
the chpt is phenomenologically constructed respecting the symmetry of qcd , the chiral symmetry and the lorentz invariance in the four - dimensional space - time . with these symmetry constraints , there are three possible terms as the four - derivative terms of pion fields : @xmath151 ^ 2,\hspace{4mm}{\rm tr}\{l_\mu , l_\nu\}^2,\hspace{4 mm } { \rm tr}(\partial_\mu l_\nu)^2 , \label{deri1}\ ] ] where @xmath152 is the 1-form pion fields in eq .
( [ u_def ] ) . the first term @xmath153 ^ 2 $ ] ,
called the `` skyrme term '' @xcite , is to give the stability of the skyrme soliton solution with finite size in the coordinate space . on the other hand ,
the other two terms are known to give the instability of skyrme solitons @xcite .
the symmetry constraints in the chpt can not determine which terms should appear because all the terms in ( [ deri1 ] ) are chiral symmetric and lorentz invariant . therefore skyrme deliberately takes only the first term @xmath153 ^ 2 $ ] in the meson effective action as an effective `` model '' for the baryon as the chiral soliton , which was called the `` skyrme model '' @xcite .
now , by starting from the holographic qcd with the five - dimensional yang - mills action ( [ 5dimdbi_2 ] ) of the probe @xmath51 brane , one can find only the skyrme term without the other two in ( [ deri1 ] ) , which is manifestly seen in the action ( [ f11_1 ] ) .
actually the five - dimensional yang - mills action ( [ 5dimdbi_2 ] ) with @xmath70 includes two - time derivatives at most , so that the appearances of the other two terms : @xmath154 and @xmath155 with four - time derivatives are forbidden at the leading order of @xmath22 and @xmath69 expansion in holographic qcd .
these comparisons with the chiral perturbation theory clearly indicate that holographic qcd is not just the low - energy effective theory of qcd only with the constraint of symmetries in four - dimensional space - time : actually , it obeys the @xmath68 symmetry extending to the extra fifth dimension @xmath55 .
furthermore , one can see that the chiral soliton picture for the baryon is now supported by the holographic approach , retaining the direct connection with qcd . with these considerations ,
we employ the concept of chiral soliton picture for the baryon analysis in holographic qcd @xcite .
now we begin with the hedgehog ansatz for pion field @xmath156 and @xmath1 meson field @xmath157 as a baryon configuration @xcite : @xmath158 where @xmath159 is pauli matrix , and @xmath160 is a dimensionless profile function of the pion field with boundary conditions @xmath161 and @xmath162 , giving topological charge equal to unity . ansatz ( [ hh ] ) means @xmath163 for the pion field .
@xmath164 is also a dimensionless profile function of the @xmath1 meson field .
this ansatz for @xmath1 meson field is also called `` wu - yang-thooft - polyakov ansatz '' @xcite , and the same configuration ansatz can be seen for the gauge field of the thooft - polyakov monopole @xcite . by substituting the configuration ansatz ( [ hh ] ) and ( [ wytp ] ) into the four - dimensional meson effective action @xmath150 in ( [ f11_1 ] ) with the euclidean metric
, we can get the static hedgehog mass of a brane - induced skyrmion on the flat space @xmath25 as follows ( detailed derivations can be found in our previous paper @xcite ) : @xmath165&\equiv & { \left [ } s_{{\rm d}8}^{\rm dbi}-s_{{\rm d}8}^{\rm dbi}|_{a_m\rightarrow 0 } { \right ] } _ { { \rm hedgehog } } \equiv \int_0^{\infty}4\pi dr r^2\cdot\varepsilon [ f(r ) , g(r ) ] , \label{bis_energy}\\ r^2\cdot\varepsilon [ f(r ) , g(r ) ] & = & \frac{f_\pi^2}{4}{\left [ } 2{\left ( } r^2 f'^{2}+2\sin^2f { \right ) } { \right ] } + m_{\rho}^2 { \left [ } 4 r^2 \tilde{g}^2 { \right ] } + \frac{1}{32e^2}{\left [ } 16 \sin^2f { \left ( } 2 f'^{2}+\frac{\sin^2f}{r^2}{\right ) } { \right ] } \nonumber \\ & + & \frac{1}{2 } { \left [ } 8{\left\ { } \newcommand{\rtk } { \right\ } } 3\tilde{g}^2 + 2 r \tilde{g}(\tilde{g } ' ) + r^2\tilde{g}'^{2}\rtk { \right ] } - g_{3\rho } { \left [ } 16 r \tilde{g}^3 { \right ] } + \frac{1}{2}g_{4\rho } { \left [ } 16 r^2 \tilde{g}^4 { \right ] } \nonumber \\ & + & g_1 { \left [ } 16 { \left\ { } \newcommand{\rtk } { \right\ } } f'\sin f \cdot{\left ( } \tilde{g}+ r \tilde{g}'{\right)}+\sin^2f\cdot\tilde{g}/r\rtk { \right ] } - g_2 { \left [ } 16 \sin^2f\cdot\tilde{g}^2 { \right ] } \nonumber \\ & - & g_3 { \left [ } 16 \sin^2f\cdot{\left ( } 1-\cos f{\right)}\tilde{g}/r { \right ] } - g_4 { \left [ } 16 { \left ( } 1-\cos f{\right)}\tilde{g}^2 { \right ] } + g_5 { \left [ } 16 r { \left ( } 1-\cos f { \right)}\tilde{g}^3 { \right ] } \nonumber \\ & + & g_6 { \left [ } 16 r^2 f'^{2}\tilde{g}^2 { \right ] } + g_7 { \left [ } 8 { \left ( } 1-\cos f{\right)}^2\tilde{g}^2 { \right ] } , \label{energy_dense_an}\end{aligned}\ ] ] where @xmath166 and @xmath167 .
now we take the `` adkins - nappi - witten ( anw ) unit '' for energy and length as @xmath168 and @xmath169 @xcite , and we rewrite all variables in this anw unit as @xmath170 and @xmath171 . by taking this scaled unit , the hedgehog energy density ( [ energy_dense_an ] ) of single brane - induced skyrmion on @xmath172 can be rewritten as follows ( overlines of @xmath173 and @xmath174 below are abbreviated for simplicity ) : @xmath175 & = & { \left ( } r^2 f'^{2}+2\sin^2f { \right)}+ 2 { \left ( } \frac{m_{\rho}}{f_{\pi}}{\right)}^2 { \left [ } 4 r^2 \tilde{g}^2 { \right ] } + \sin^2f { \left ( } 2 f'^{2}+\frac{\sin^2f}{r^2}{\right)}\nonumber \end{aligned}\ ] ] @xmath176 } - { \left ( } 2e^2{\right)}g_{3\rho } { \left [ } 16 r \tilde{g}^3 { \right ] } + { \left ( } 2e^2{\right)}\frac{1}{2}g_{4\rho } { \left [ } 16 r^2 \tilde{g}^4 { \right ] } \nonumber \\ & + & { \left ( } 2e^2{\right)}g_1 { \left [ } 16 { \left\ { } \newcommand{\rtk } { \right\ } } f'\sin f \cdot{\left ( } \tilde{g}+ r \tilde{g}'{\right)}+\sin^2f\cdot\tilde{g}/r\rtk { \right ] } - { \left ( } 2e^2{\right)}g_2 { \left [ } 16 \sin^2f\cdot\tilde{g}^2 { \right ] } \nonumber \\ & -&{\left ( } 2e^2{\right)}g_3 { \left [ } 16 \sin^2f\cdot{\left ( } 1-\cos f{\right)}\tilde{g}/r { \right ] } - { \left ( } 2e^2{\right)}g_4 { \left [ } 16 { \left ( } 1-\cos f{\right)}\tilde{g}^2 { \right ] } \nonumber \\ & + & { \left ( } 2e^2{\right)}g_5 { \left [ } 16 r { \left ( } 1-\cos f { \right)}\tilde{g}^3 { \right ] } + { \left ( } 2e^2{\right)}g_6 { \left [ } 16 r^2 f'^{2}\tilde{g}^2 { \right ] } + { \left ( } 2e^2{\right)}g_7 { \left [ } 8 { \left ( } 1-\cos f{\right)}^2\tilde{g}^2 { \right ] } .\label{energy_dense_re_r3}\end{aligned}\ ] ] here we comment about a scaling property of the brane - induced skyrmion .
the holographic qcd has just two parameters : @xmath148 and @xmath15 , so that the pion decay constant @xmath124 , the @xmath1 meson mass @xmath125 , and the skyrme parameter @xmath140 in eqs.([f_pi_1])@xmath149([sky_para ] ) can be explicitly written by @xmath62 and @xmath15 in holographic qcd as @xmath177 } ^{-1/2 } \simeq \frac{1}{4\sqrt{0.157}}\frac{1}{\sqrt{\kappa}},\nonumber\\ \label{e_kappa}\end{aligned}\ ] ] where the energy unit @xmath15 is recovered . by using these relations ( [ fpi_mkk])@xmath149([e_kappa ] ) , anw unit for energy and length , i.e. , @xmath178 and @xmath179 can be written by @xmath62 and @xmath15 as @xmath180 the five - dimensional yang - mills action ( [ 5dimdbi_2 ] ) with @xmath70 is proportional to @xmath148 , as the leading order of @xmath22 and @xmath69 expansion .
furthermore @xmath15 is the sole energy scale of the holographic approach .
therefore , in the energy unit @xmath181 , the total energy appears as a scale invariant variable .
in fact , by introducing the rescaled @xmath1 meson field @xmath182 as @xmath183 and considering the @xmath62-dependence of the basis @xmath184 as @xmath185 in the normalization condition ( [ normal_psi_n ] ) , one can analytically show that every energy density in each term of eq.([energy_dense_re_r3 ] ) and meson field configurations @xmath160 and @xmath182 are scale invariant variables , being independent of the holographic two parameters , @xmath62 and @xmath15 . with the considerations above , we give most discussions below in the anw unit as the universal features of baryonic matter in holographic qcd , being independent of the definite values of @xmath124 and @xmath125 .
the recovering of physical unit with the experimental inputs for @xmath124 and @xmath125 is discussed in sec .
[ cdt ] , with respect to the critical densities of the phase transitions in the baryonic matter within holographic approach .
now we study the baryonic matter in holographic qcd by analyzing the system of single brane - induce skyrmion on a three - dimensional closed manifold @xmath4 . in this study
, we consider the baryonic matter with large @xmath3 because holographic qcd is the large @xmath3 effective theory , derived from the classical supergravity justified in the large @xmath3 and large thooft coupling @xcite .
according to the general analysis of large-@xmath3 qcd , a baryon mass is found to become @xmath18 @xcite , so that its kinetic energy becomes @xmath19 .
as for the quantum effects of the baryonic matter , zero point quantum fluctuation energy @xmath20 and baryon mass splitting @xmath21 in the isospin projection correspond to the higher - order effects of @xmath22 expansion : @xmath186 and @xmath24 @xcite .
therefore , with large-@xmath3 condition , we can consider that the kinetic energy and quantum effects are suppressed relative to the static mass , and the baryonic matter comes into the `` static skyrme matter '' .
such static skyrme matter was first analyzed by klebanov @xcite , placing skyrme soliton configurations periodically along the three - dimensional cubic lattice , which could be related with `` nuclear crystal '' with pion condensation in the deep interior of neutron stars .
therefore , by analyzing the static skyrme matter , one can see some typical features of baryonic matter with large-@xmath3 conditions .
+ in this paper , we take certain mathematical trick to analyze such static skyrme matter suggested by manton and ruback @xcite . to analyze the features of multi - skyrmion system on the flat coordinate space @xmath25
, they alternately treat the system of single skyrmion on a three - dimensional closed manifold @xmath4 with finite radius @xmath5 as shown in fig .
[ fig_projection1 ] . actually , the multi - skyrmion system on @xmath25 and the system of single skyrmion on @xmath4 can be related with each other through the compactification of the boundary for a unit cell on @xmath25 shared by the single skyrmion as in fig .
[ fig_projection1 ] .
the interaction between the baryons in the medium on @xmath25 is simulated by the curvature of the closed manifold @xmath4 .
the baryon - number density can be represented as @xmath187 on @xmath4 , so that , as the size of @xmath4 decreases , the increase of the baryon - number density in the medium is represented in this modeling . actually , as the radius @xmath5 of @xmath4 decreases ,
the energy density of single baryon is found to delocalize due to the medium effects in the baryonic matter @xcite , and , below the critical radius @xmath188 of @xmath4 , the energy density of the baryon coincides with the uniform distribution as the `` identity map '' , which is called the `` delocalization phase transition '' shown in fig .
[ fig_projection1 ] . such delocalization phase transition in the skyrme model can be related with the deconfinement of the baryon and also the chiral symmetry restoration in qcd , which will be inclusively discussed by taking the order parameters in sec .
[ crdpt ] . with these considerations , by analyzing the system of single brane - induced skyrmion on @xmath4 , we can see some typical features of baryonic matter in holographic qcd . especially , by comparing the standard skyrmion without @xmath1 mesons and the brane - induced skyrmion on @xmath4 ,
the roles of ( axial ) vector mesons in high density phase of baryonic matter can be discussed from the holographic point of view .
now we introduce the projection procedure from the flat space @xmath25 onto the curved space @xmath4 @xcite , to get a hedgehog mass of a brane - induced skyrmion on @xmath4 .
first , we consider the three - dimensional orthogonal space @xmath25 in polar coordinates as @xmath189 the integral operator @xmath190 and the derivation @xmath191 can be written in polar coordinates as @xmath192 so that the differential operator @xmath193 can be written as @xmath194 second , we consider the four - dimensional orthogonal space @xmath195 in polar coordinates as @xmath196 the integral operator @xmath190 and the derivation @xmath191 can be written in polar coordinates as @xmath197 so that the differential operator @xmath193 can be written as @xmath198 now , by limiting the four - dimensional orthogonal space @xmath195 onto the surface of a three - dimensional closed manifold @xmath4 with fixed radius @xmath5 , the coordinate @xmath199 in eq .
( [ coordr4 ] ) becomes dependent on the other coordinates @xmath200 .
furthermore , @xmath201 and @xmath202 can be regarded as zero in eqs .
( [ 4dimio ] ) and ( [ 4dimdo ] ) because the radial coordinate @xmath5 is fixed on @xmath4 .
therefore , by comparing eqs .
( [ coordr3 ] ) , ( [ 3dimio ] ) , ( [ 3dimdo ] ) on @xmath25 , and eqs .
( [ coordr4 ] ) , ( [ 4dimio ] ) , ( [ 4dimdo ] ) on @xmath4 with fixed radius @xmath5 , we find the projection procedure from @xmath25 to @xmath4 as follows : @xmath203 recall that the hedgehog mass on @xmath25 with energy density in eq .
( [ energy_dense_re_r3 ] ) can be written with its explicit arguments for the energy density as @xmath204 } .
\nonumber\\ \label{bis_energy_r3}\end{aligned}\ ] ] by applying the projection procedure ( [ pro1 ] ) , ( [ pro2 ] ) and ( [ pro3 ] ) to the hedgehog mass on @xmath25 in eq .
( [ bis_energy_r3 ] ) , we can get the hedgehog mass on @xmath4 as @xmath205 } .
\label{bis_energy_s3}\\ & = & \int_0^{\pi r}4\pi dr r^2\sin^2\frac{r}{r}\cdot\varepsilon { \left [ } f(r ) , g(r ) , \frac{\partial}{\partial r}f(r ) , \frac{\partial}{\partial r}g(r ) , r\sin\frac{r}{r } { \right ] } .
\label{bis_energy_s3_re}\end{aligned}\ ] ] in eq .
( [ bis_energy_s3_re ] ) , we introduce a new variable @xmath206 as the arc length on @xmath4 as @xmath207 and @xmath206-dependent dimensionless functions @xmath208 and @xmath209 are renamed again as @xmath160 and @xmath164 .
therefore , by comparing eq .
( [ bis_energy_r3 ] ) on @xmath25 and eq .
( [ bis_energy_s3_re ] ) on @xmath4 , we can get the simple projection procedure for the hedgehog energy density from @xmath25 to @xmath4 as @xmath210 } \longrightarrow dr r^2\sin^2\frac{r}{r}\cdot\varepsilon { \left [ } f(r ) , g(r ) , \frac{\partial}{\partial r}f(r ) , \frac{\partial}{\partial r}g(r ) , r\sin\frac{r}{r } { \right ] } , \label{edensity_pro1}\end{aligned}\ ] ] where the topological boundary for the chiral field @xmath160 in ( [ hh ] ) is also projected on @xmath4 as @xmath211 now , by applying the projection procedure ( [ edensity_pro1 ] ) for the hedgehog energy density on @xmath25 in eq .
( [ energy_dense_re_r3 ] ) , we can eventually get the hedgehog energy density on @xmath4 with anw units as follows ( note here that the dimensional profile function @xmath212 on @xmath25 is to be naturally introduced on @xmath4 through the projection procedure as @xmath213 ) : @xmath165&= & \int_0^{\pi r}4\pi dr r^2\sin^2\frac{r}{r}\cdot\varepsilon [ f(r ) , g(r ) ] , \label{bis_energy_s3_again}\\ r^2\sin^2\frac{r}{r}\cdot\varepsilon [ f(r ) , g(r ) ] & = & { \left ( } r^2\sin^2\frac{r}{r}\cdot f'^{2}+2\sin^2f { \right)}+ 2 { \left ( } \frac{m_{\rho}}{f_{\pi}}{\right)}^2 { \left [ } 4 r^2\sin^2\frac{r}{r}\cdot \tilde{g}^2 { \right ] } \nonumber \\ & + & \sin^2f { \left ( } 2 f'^{2}+\frac{\sin^2f}{r^2\sin^2\frac{r}{r}}{\right)}\nonumber \\ & + & { \left ( } 2e^2{\right)}\frac{1}{2 } { \left [ } 8{\left\ { } \newcommand{\rtk } { \right\ } } { \left ( } 2+\cos^2\frac{r}{r}{\right)}\tilde{g}^2 + 2r\sin\frac{r}{r}\cos\frac{r}{r}\cdot\tilde{g}(\tilde{g } ' ) + r^2\sin^2\frac{r}{r}\cdot\tilde{g}'^{2}\rtk { \right ] } \nonumber \\ & - & { \left ( } 2e^2{\right)}g_{3\rho } { \left [ } 16 r\sin\frac{r}{r}\cdot \tilde{g}^3 { \right ] } + { \left ( } 2e^2{\right)}\frac{1}{2}g_{4\rho } { \left [ } 16 r^2\sin^2\frac{r}{r}\cdot \tilde{g}^4 { \right ] } \nonumber \\ & + & { \left ( } 2e^2{\right)}g_1 { \left [ } 16 { \left\ { } \newcommand{\rtk } { \right\ } } f'\sin f \cdot{\left ( } \cos\frac{r}{r}\cdot\tilde{g}+ r\sin\frac{r}{r}\cdot \tilde{g}'{\right)}+\sin^2f\cdot\tilde{g}/{\left ( } r\sin\frac{r}{r}{\right)}\rtk { \right ] } \nonumber \\ & -&{\left ( } 2e^2{\right)}g_2 { \left [ } 16 \sin^2f\cdot\tilde{g}^2 { \right ] } - { \left ( } 2e^2{\right)}g_3 { \left [ } 16 \sin^2f\cdot{\left ( } 1-\cos f{\right)}\tilde{g}/{\left ( } r\sin\frac{r}{r}{\right ) } { \right ] } \nonumber \\ & -&{\left ( } 2e^2{\right)}g_4 { \left [ } 16 { \left ( } 1-\cos f{\right)}\tilde{g}^2 { \right ] } + { \left ( } 2e^2{\right)}g_5 { \left [ } 16 r\sin\frac{r}{r}\cdot { \left ( } 1-\cos f { \right)}\tilde{g}^3 { \right ] } \nonumber \\ & + & { \left ( } 2e^2{\right)}g_6 { \left [ } 16 r^2\sin^2\frac{r}{r}\cdot f'^{2}\tilde{g}^2 { \right ] } + { \left ( } 2e^2{\right)}g_7 { \left [ } 8 { \left ( } 1-\cos f{\right)}^2\tilde{g}^2 { \right ] } , \label{energy_dense_re}\end{aligned}\ ] ] where @xmath166 and @xmath167 . here
measures @xmath214 newly appear in the energy density ( [ energy_dense_re ] ) relative to eq.([energy_dense_re_r3 ] ) on the flat space @xmath25 , indicating the existence of the curvature of the closed manifold @xmath4 .
we also construct the euler - lagrange equations for the pion field @xmath160 and @xmath1 meson field @xmath215 from the energy density in eq .
( [ energy_dense_re ] ) as follows : @xmath216 } \nonumber\\ & - & { \left ( } 2e^2{\right)}g_2 { \left [ } 16 { \left ( } 2\sin f\cdot\cos f\cdot \tilde{g}^2{\right ) } { \right ] } - { \left ( } 2e^2{\right)}g_3 { \left [ } 16 { \left ( } \sin f + 2\sin f \cdot \cos f-3\sin f\cdot \cos^2f{\right)}\tilde{g}/{\left ( } r\sin\frac{r}{r}{\right ) } { \right ] } \nonumber\\ & - & { \left ( } 2e^2{\right)}g_4 { \left [ } 16 { \left ( } \sin f
\cdot\tilde{g}^2{\right ) } { \right ] } + { \left ( } 2e^2{\right)}g_5 { \left [ } 16 { \left ( } r\sin\frac{r}{r } \sin f \cdot\tilde{g}^3{\right ) } { \right ] } \nonumber\\ & + & { \left ( } 2e^2{\right)}g_6 { \left [ } 16 { \left ( } -4r\sin\frac{r}{r}\cos\frac{r}{r}\cdot f'\tilde{g}^2 -2r^2\sin^2\frac{r}{r}\cdot f''\tilde{g}^2 -4r^2\sin^2\frac{r}{r}\cdot f'\tilde{g}\tilde{g}'{\right ) } { \right ] } \nonumber\\ & + & { \left ( } 2e^2{\right)}g_7 { \left [ } 8 { \left\ { } \newcommand{\rtk } { \right\ } } 2 { \left ( } 1-\cos f { \right)}\sin f \cdot\tilde{g}^2\rtk { \right ] } = 0 , \label{el_f } \end{aligned}\ ] ] @xmath217 } \nonumber\\ & + & { \left ( } 2e^2{\right)}\frac{1}{2 } { \left [ } 8 { \left ( } 4\tilde{g } + 2\sin^2\frac{r}{r}\cdot\tilde{g } -4r\sin\frac{r}{r}\cos\frac{r}{r}\cdot\tilde{g}'-2r^2\sin^2\frac{r}{r}\cdot\tilde{g}''{\right ) } { \right ] } \nonumber\\ & - & { \left ( } 2e^2{\right)}g_{3\rho } { \left [ } 16 { \left ( } 3r\sin\frac{r}{r}\cdot\tilde{g}^2 { \right ) } { \right ] } + { \left ( } 2e^2{\right)}\frac{1}{2 } g_{4\rho } { \left [ } 16 { \left ( } 4 r^2\sin^2\frac{r}{r}\cdot \tilde{g}^3 { \right ) } { \right ] } \nonumber\\ & + & { \left ( } 2e^2{\right)}g_1 { \left [ } 16 { \left\ { } \newcommand{\rtk } { \right\ } } \sin^2 f/{\left ( } r\sin\frac{r}{r}{\right)}-r\sin\frac{r}{r}\cos f\cdot f'^{2}- r\sin\frac{r}{r}\sin f\cdot f''\rtk { \right ] } \nonumber\\ & - & { \left ( } 2e^2{\right)}g_2 { \left [ } 16 { \left ( } 2\sin^2f \cdot\tilde{g } { \right ) } { \right ] } - { \left ( } 2e^2{\right)}g_3 { \left [ } 16 \sin^2 f ( 1-\cos f)/{\left ( } r\sin\frac{r}{r}{\right ) } { \right ] } \nonumber\\ & - & { \left ( } 2e^2{\right)}g_4 { \left [ } 16 { \left\ { } \newcommand{\rtk } { \right\ } } 2(1-\cos f)\tilde{g}\rtk { \right ] } + { \left ( } 2e^2{\right)}g_5 { \left [ } 16 { \left\ { } \newcommand{\rtk } { \right\ } } 3r\sin\frac{r}{r}{\left ( } 1-\cos f{\right)}\tilde{g}^2 \rtk { \right ] } \nonumber\\ & + & { \left ( } 2e^2{\right)}g_6 { \left [ } 16 { \left\ { } \newcommand{\rtk } { \right\ } } 2r^2\sin^2\frac{r}{r}\cdot f'^{2}\tilde{g}\rtk { \right ] } + { \left ( } 2e^2{\right)}g_7 { \left [ } 8 { \left\ { } \newcommand{\rtk } { \right\ } } 2 { \left ( } 1-\cos f { \right)}^2 \tilde{g}\rtk { \right ] } = 0 .\label{el_g}\end{aligned}\ ] ] now one can show that the euler - lagrange equations ( [ el_f ] ) and ( [ el_g ] ) always have the analytical solution for arbitrary value of radius @xmath5 of @xmath4 as @xmath218 which is the `` identity map '' solution in the standard skyrme model without @xmath1 meson field @xcite . in the skyrme model ,
the energy density of the baryon is generated by the spatial gradient of the pion fields .
therefore , such linear configuration as in eq.([bis_ide ] ) gives the uniform energy distributions with its classical mass as @xmath219 which can be given by substituting the linear configuration ( [ bis_ide ] ) into the energy density ( [ energy_dense_re ] ) .
hence the change of absolute minimum solution from the localized skyrmion into the identity map ( [ bis_ide ] ) can be regarded as a signal of transitions from localized phase to the uniform phase of the baryonic matter , referred as the delocalization phase transition .
note that the identity map solution ( [ bis_ide ] ) has no @xmath1 meson configuration : @xmath220 .
therefore , if the identity map is realized as the absolute minimum solution , it indicates that @xmath1 meson field absolutely disappear in high density phase , which is discussed with numerical results in sec .
in this section , we show the numerical results and discussions about the baryon nature in dense qcd by solving the euler - lagrange equations ( [ el_f ] ) and ( [ el_g ] ) derived from holographic qcd .
the energy density and the field configuration profiles of the baryon are analyzed in sec .
[ ecpro ] .
a new striking picture of `` pion dominance '' near the critical density is proposed in sec .
the mass and root - mean - square mass radius of the baryon are examined in sec .
we explain the `` swelling '' mechanism of the baryon in the general context of qcd in sec . [ smpi ] . through all of the sections below , by comparing the skyrme model without @xmath1 meson field and the brane - induced skyrme ( bis ) model on @xmath4 , the roles of ( axial ) vector mesons in the baryonic matter are discussed from a holographic point of view .
cc & in this section , we discuss the baryon - number density dependence of the energy density and the field configuration profiles of single baryon for the skyrme model and the bis model , by changing the size of the manifold @xmath4 . actually , the baryon - number density is represented as @xmath221 on @xmath4 , so that , as the radius @xmath5 of @xmath4 decreases , the increase of total baryon - number density in the medium is represented in this modeling .
first , we show in fig.[fig_density1 ] the energy density of the baryon for the skyrme model with radius @xmath5 of @xmath4 .
one can see that the energy density tends to delocalize as @xmath5 decreases , which can be regarded as some medium effects in the dense baryonic matter .
below the critical radius , @xmath222 , the energy density distribution of the baryon exactly coincides with the uniform one as identity map denoted by the dashed lines in fig .
[ fig_density1 ] .
the transition from localized energy density into uniform one is called the `` delocalization phase transition '' .
next we show in fig.[fig_density2 ] the same plot in the case of the bis model .
the energy density of the baryon also tends to delocalize as @xmath5 decreases .
however the delocalization along with the decrease of @xmath5 is delayed relative to the skyrme model without @xmath1 meson field , and the energy distribution is still localized even around @xmath223 , which is the critical radius for the skyrme model .
in fact , the heavy @xmath1 meson field appearing in the core region of the baryon is to provide the attraction with the pion field , which leads to the shrinkage of the total size of the baryon @xcite .
therefore , the smaller radius of @xmath4 , i.e. , the larger baryon - number density is needed for the bis model to give the delocalization phase transition , which is discussed in sec .
[ cdt ] with recovering the physical units .
then , below the critical radius , @xmath224 , the energy density distribution of the baryon in the bis model becomes the uniform one as identity map ( [ bis_ide ] ) .
these delocalization phase transitions in the skyrmion picture can be related with the deconfinement of the baryon and also the chiral symmetry restoration , which will be discussed with the order parameters in sec .
[ crdpt ] .
+ we can also show in fig.[fig_t_rho ] the comparison between the energy density of the baryon and @xmath1 meson contributions in the interaction terms of the bis model ( [ energy_dense_re ] ) for @xmath225 , @xmath226 , and 1.19 . for @xmath225 and @xmath226 , one can see the manifest contributions from the @xmath1 meson field in the core region of the baryon . on the other hand , at the critical radius @xmath227 , all the contributions from the @xmath1 meson field absolutely disappear in the uniform phase .
we also show the field configuration profiles of the pion and the @xmath1 meson in fig .
[ fig_conf1 ] and fig .
[ fig_conf2 ] , respectively , in the bis model for @xmath225 , @xmath226 , and @xmath228 . as @xmath5 decreases , the pion field @xmath160 in fig .
[ fig_conf1 ] approaches to the linear configuration as the identity map in eq.([bis_ide ] ) , giving the uniform energy density distribution . as for the @xmath1 meson field in fig [ fig_conf2 ]
, the amplitude of @xmath1 meson field @xmath229 tends to decrease as @xmath5 decreases , and it absolutely disappears below the critical radius , @xmath224 as the identity map in eq .
( [ bis_ide ] ) .
these results indicate that the amplitude of the @xmath1 meson field decreases as the baryon - number density increases in the medium , and it absolutely disappears and only the pion field survives near the critical density . in the previous section , we find that the @xmath1 meson field would disappear near the critical density .
now we propose a conjecture that such disappearance of the @xmath1 meson field near the critical density can be generalized to all the other ( axial ) vector meson fields : @xmath230 , denoted by the field @xmath231 in eq .
( [ limit2_lr ] ) by the following reasons within holographic approach : the kinetic term of ( axial ) vector meson field @xmath106 on the closed manifold @xmath4 is proportional to @xmath232 as @xmath233 which can be derived from a simple dimensional analysis . in general ,
the kinetic energy indicates the `` kink '' energy of field configurations . therefore the kinetic energy in eq.([kine_pro ] ) suppresses the spatial dependence of the field configuration @xmath106 for small @xmath5 , i.e. , for high density state , giving flat configuration in the high density phase .
the mass term @xmath234 suppresses the absolute value of @xmath106 field more severely in accordance with its larger mass @xmath235 .
the couplings between pions and heavier ( axial ) vector mesons @xmath106 with larger index @xmath111 are found to become smaller , which is suggested by the `` bottom - up '' dimensional deconstruction model @xcite and also the `` top - down '' holographic approach @xcite .
in fact , within the holographic models , the origin of the meson mass in four dimensional space - time can be regarded as the oscillation of meson wave function in the extra fifth dimension @xmath55 , denoted by the mass relation @xmath236 in sec .
such larger oscillations of heavier ( axial ) vector mesons wave functions in the fifth dimension have the smaller overlap with that of pions , giving the smaller coupling constants with pions in four dimensional space - time @xcite .
in fact , some recent experiments with the hadron reactions provide some interesting data , showing that the heavier ( axial ) vector mesons tend to have smaller width for the decay into pions despite of larger phase space @xcite , which may be consistent with the prediction of holographic qcd as their smaller coupling constants with pions as mentioned above .
hence the effects of heavier ( axial ) vector mesons should be smaller for a baryon as a large - amplitude pion field , i.e. , the chiral soliton .
these considerations 1 ) , 2 ) , 3 ) about the meson effective action in holographic qcd should support our conjecture about the general disappearance of ( axial ) vector meson fields in the high density phase . in other words , _ only the pion fields survive near the critical density in the large-@xmath3 baryonic matter_. we call this phenomena as `` pion dominance '' near the critical density . in sec.[cdt ] , we will show that , even if the @xmath1 meson field disappears near the critical point , it affects the critical density of the phase transition through its mass and also the interactions with the pion field in the action @xmath150 in eq.([dbi ] ) . in this sense , even with the pion dominance near the critical point proposed above , the ( axial ) vector mesons may still affect the critical phenomena like the critical density through their contributions in the effective action if they are included . as for the survival of the pion fields near the critical density as the pion dominance
, we give the following reason : the unit baryon - number on the manifold @xmath4 comes from the boundary conditions of the pion hedgehog configuration @xmath160 at the north ( @xmath237 ) and the south ( @xmath238 ) poles on @xmath4 as @xmath161 and @xmath239 in ( [ bondary_s3 ] ) . in this sense
, pions play the essential roles for the baryon - number current with their field boundaries .
therefore the pion field can not disappear because of the baryon - number constraint ( [ bondary_s3 ] ) for each unit cell of the manifold @xmath4 . on the other hand
, there is no constraints for the other ( axial ) vector meson fields , and they can disappear in the high density phase as some representation of `` deconfinement '' . such dominance of the pion field near the transition point might be somehow related with the appearance of chiral plasma modes @xmath240 and @xmath241 above @xmath242 of the deconfinement phase transition observed in the lattice qcd stuty @xcite .
in fact , the screening masses were measured for a variety of color - singlet channels with the quantum number of @xmath240 , @xmath241 , @xmath1 , @xmath243 , and @xmath244 mesons .
the chiral multiplet of @xmath240 and @xmath241 was found to appear as the bound state even above @xmath242 , whereas the other ( axial ) vector mesons would be in the continuum with the threshold as the twice of the lowest matsubara frequency , i.e. , @xmath245 .
such survival of the chiral plasma modes above @xmath242 would inspire the concept of the strong - coupling quark - gluon plasma ( sqgp ) due to the long - range , non - perturbative effects of the strong interaction even at the highest temperature , instead of the naive free gas picture of quarks and gluons only due to the asymptotic freedom .
such strong correlation would also develop even at dense regime of qcd , which may give some linking of our analysis with `` pion dominance '' in dense qcd . by seeing some qcd phenomenologies
, one can also find that quark degrees of freedom are often represented by `` pions '' . for example , in the case of the chiral quark model @xcite , quark - antiquark correlations are represented by pionic collective modes through the bosonization scheme for qcd , giving the `` cheshire cat picture '' where the quark dynamics are represented by pions . in fact , it have provided a foundation of skyrme soliton picture for the baryon as a large - amplitude pion field .
these traditional phenomenologies may also support the dominance of the pions near the critical density in the baryonic matter , with the appearance of quark - gluon dynamics .
+ in this section , we discuss the baryon - number density dependence of the mass and the root - mean - square ( rms ) mass radius of single baryon for the skyrme model and the bis model , by changing the size of the manifold @xmath4 . in fig.[fig_masspro1 ] , we show the mass of the baryon for the skyrme model and the bis model with radius @xmath5 of @xmath4 . in the case of the skyrme model ,
the mass decreases from the value on the flat space @xmath25 : @xmath246 $ ] @xcite as @xmath5 decreases , and it coincides with the mass of identity map below the critical radius , @xmath222 @xcite .
this coincidence of numerical solution with identity map indicates the transition from the localized phase to the uniform phase as the delocalization phase transition . in the case of the bis model ,
the mass decreases from the value on @xmath25 : @xmath247 $ ] @xcite as @xmath5 decreases , and it coincides with the mass of identity map below the critical radius , @xmath224 , which is smaller than @xmath248 due to the shrinkage of baryon by the @xmath1 meson effects . from a physical point of view , the decrease of the baryon mass as @xmath5 decreases in fig .
[ fig_masspro1 ] might come from the partial restoration of chiral symmetry in the dense baryonic matter , which will be discussed with the order parameter in sec .
[ crdpt ] .
this result is related with the well - known picture of the baryon mass generation from the quark condensate in the vacuum , suggested in some qcd phenomenologies , e.g. , the qcd sum rules with ioffe formula @xcite , and also other chiral effective models @xcite .
in fact , the decrease of the baryon mass with the chiral symmetry restoration is observed in the finite temperature lattice qcd study @xcite .
note also that the decrease of the baryon mass with the chiral symmetry restoration in dense qcd is proposed in the framework of ads / qcd model as the phenomenological bottom - up constructions , called `` hard - wall '' model @xcite .
+ next we analyze of the rms mass radius of the baryon for the skyrme model and the bis model . by using the normalized energy density @xmath249 ( @xmath250 is total energy density and @xmath251 is the mass of single baryon )
, the rms mass radius can be naturally introduced on @xmath4 with radius @xmath5 as @xmath252 } ^{1/2}\nonumber\\ & \equiv & { \left [ } \int_0^{\pi r } 4\pi dr r^2\sin^2\frac{r}{r}\cdot\bar{\varepsilon}(r ) r^2 { \right ] } ^{1/2},\label{rms_1}\end{aligned}\ ] ] where @xmath214 denotes the measure of the curved manifold @xmath4 . in the case of the identity map ,
the rms mass radius can be analytically calculated by using the normalized uniform energy density for the identity map : @xmath253 as @xmath254 } ^{1/2 } \nonumber\\ & = & \pi r\sqrt{\frac{1}{3}-\frac{1}{2\pi^2}}. \label{rms_ide}\end{aligned}\ ] ] in fig.[fig_rms ] , we show the rms mass radius for the skyrme model and the bis model with radius @xmath5 of @xmath4 .
one can find that the rms mass radius in each model increases non - linearly around its critical radius as @xmath5 decreases , indicating the `` swelling '' phenomenon of the baryon in high density phase .
such swelling of the baryon can also be seen in the analysis of the skyrme crystal on the three - dimensional cubic lattice @xcite and also the finite density bag model @xcite .
after the delocalization phase transition , the energy density of single baryon is uniformly saturated over the surface of @xmath4 .
then the rms mass radius decreases linearly with the decrease of @xmath5 as seen in eq.([rms_ide ] ) .
( in the uniform phase , baryons are no longer localized , and the rms mass radius may have less physical meaning . )
+ in the previous section , we numerically find the swelling of the baryon by taking the skyrmion picture for the baryon analysis .
actually , as discussed in sec . [ bis_r3 ] , the skyrmion picture is naturally derived from qcd through the new concept of holography . therefore one can expect the swelling phenomenon in qcd itself . in this section , we try to give a qualitative explanation about the swelling mechanism of a baryon in terms of qcd , by especially referring to contexts of the lattice qcd @xcite and also the finite density bag picture @xcite . first , in lattice qcd study , the non - perturbative qcd vacuum is analyzed within a finite box of periodic boundaries ( and also dirichlet boundaries ) in the euclidean space .
the non - perturbative effects with long wave length tends to be `` blocked out '' from such finite box of definite boundaries , so that the qcd vacuum in the finite box is found to approach to the perturbative one .
also , the finite - temperature phase transition in lattice qcd occurs because of the definite boundaries along the imaginary time axis in euclidean space ; as the temperature increases , the qcd vacuum in the euclidean space is more closely packed along the imaginary time axis to block out the non - perturbative effects with long wave length , which eventually gives the transition into the perturbative qcd vacuum , like the chiral symmetry restoration and also the deconfinement .
next we combine the above picture into the finite density bag model , to explain the mechanism of the swelling in qcd . in the bag model , the baryon
is represented by a `` bag '' with the perturbative qcd vacuum surrounded by the dirichlet boundaries , which is supported by the non - perturbative qcd vacuum with the bag pressure as shown in fig .
[ fig_fbag](a ) . as a number of bags increases , representing the high density baryonic matter ,
the non - perturbative qcd vacuum around the bags is more closely packed within a definite region , imitated by a cube with dashed lines in fig .
[ fig_fbag](b ) .
then , in analogy with the case of lattice qcd study discussed above , the qcd vacuum around the bags should approach to the perturbative one by blocking out the non - perturbative effects with long wave length as shown by the several waves in fig .
[ fig_fbag ] .
therefore the bag pressure as a non - perturbative effect decreases as the number of bags increases , so that it eventually gives the swelling of bags .
such swelling in turn decreases the region of non - perturbative qcd vacuum around the bags to give the decrease of the bag pressure again . in this sense
, such swelling could occur _ non - linearly _ as shown in fig .
[ fig_rms ] , giving some drastic change of features like `` deconfinement '' in the baryonic matter .
one can expect some resemblance between the finite density bag model and the skyrme matter in our study with respect to baryonic matter , so that the swelling phenomena as shown in fig .
[ fig_rms ] could be understood on the same footing provided above .
now we also suggest the physical effects of the swelling in the high density baryonic matter . here
we take into account possible baryon excitation with the chiral soliton picture . since a moment of inertia of the baryon
is related with the size of the baryon , some non - linear increase in the moment of inertia around critical density is expected along with the swelling of a baryon . in the semiclassical quantization procedure of a skyrme soliton , the baryon mass spectra is given @xcite as @xmath255 where @xmath256 is static hedgehog mass and @xmath257 is the moment of inertia of a baryon .
the quantum number @xmath258 in eq .
( [ sqp ] ) denotes the spin @xmath259 and the isospin @xmath260 of the baryon as @xmath261 for @xmath30 and @xmath262 for @xmath31 .
the baryon mass splitting in the second term in eq.([sqp ] ) is proportional to the inverse of the moment of inertia , indicating that a larger object can not be easily rotated .
therefore we can now propose that the baryon mass splitting decreases as the density increases in the medium , because the moment of inertia @xmath257 increases due to the swelling .
in other words , the _ individuality _ of each baryon like @xmath30 , @xmath31 , etc .
, would be _ lost _ at least with respect to their mass spectra as the density increases .
such lost of individuality may be regarded as some precursory phenomena of deconfinement , where the dominant degrees of freedom in the hadronic matter shifts from the confined composites as baryons ( and mesons ) into quarks and gluons .
such swelling of the baryon leads to some interesting phenomenological realizations .
as one possibility , here we propose the stable @xmath30-@xmath31 mixed matter in dense qcd . for simplicity , we consider the `` symmetric nuclear matter '' , where only the strong interaction is taken into account without the electro - magnetic interaction . in the low density case ,
there exists large @xmath30-@xmath31 splitting as @xmath263 , and the fermi surface of the nucleons @xmath264 may locate below the threshold of @xmath31 isobars , i.e. , its mass @xmath265 .
this situation gives only the nucleon degrees of freedom as nuclear matter .
now , if the swelling occurs as the density increases , the @xmath30-@xmath31 splitting should decrease as discussed in eq .
( [ sqp ] ) .
then , the fermi surface of the nucleons @xmath264 would excess @xmath265 to give the @xmath31 isobar degrees of freedom in the medium through the equilibrium process @xmath266 . in this sense ,
@xmath30-@xmath31 mixed matter would be realized in dense qcd because of the swelling of a baryon .
such @xmath30-@xmath31 mixed matter may appear in the deep interior of neutron stars between the nuclear crust and the core of quark matter as the precursor of deconfinement , giving some softening of the eos of neutron stars relative to the analysis without the mixed matter . while these are qualitative discussions
, we feel that these explanations catch some essential aspects of baryonic matter with baryon excitation in view of its `` swelling '' nature .
in this section we discuss the delocalization phase transition in view of the deconfinement , and also the chiral symmetry restoration by introducing the proper order parameters in the skyrme model and the bis model .
the relations between these phase transitions are also considered by referring to other qcd phenomenologies in the end of sec .
[ csr ] .
first we consider the delocalization phase transition with the order parameter . by using the normalized energy density @xmath267 ( @xmath268 is total energy density and @xmath251 is the mass of single skyrmion )
, one can introduce the spatial fluctuation @xmath269 of the energy density around the uniform energy density distribution @xcite as @xmath270 where @xmath271 is the normalized uniform energy density for the identity map .
if the skyrmion is well localized like the delta function , the energy density @xmath272 is almost decoupled with uniform energy density distribution @xmath273 through the spatial integral over @xmath4 . therefore its spatial fluctuation @xmath269 becomes , @xmath274 on the other hand , if the skyrmion gives uniform energy density distribution as the identity map , the energy density @xmath272 coincides with @xmath273 , giving @xmath275 in this sense , @xmath269 in eq.([lop1 ] ) measures the amount of localization in the energy density distribution of the skyrmion , regarded as the order parameter of delocalization phase transition in the baryonic matter .
+ in fig .
[ fig_df ] , we show the value of @xmath269 for the skyrme model and the bis model with radius @xmath5 of @xmath4 . for sufficiently large @xmath5 as the low density state , the skyrme soliton is well localized for both the skyrme model and the bis model , giving @xmath276 . with the decrease of @xmath5 , the baryon tends to delocalize with the decrease of @xmath269 , regarded as some medium effect in the baryonic matter .
the delocalization phase transition into the uniform phase with @xmath277 occurs at the critical radius @xmath278 for the skyrme model and at @xmath227 for the bis model . actually , the delocalization phase transition in the bis model is delayed along with the decrease of @xmath5 because of the heavy @xmath1 mesons in the core region of the baryon , discussed in sec .
[ ecpro ] . here
one may have some temptation to relate the delocalization phase transition with the `` deconfinement '' of the baryon in qcd , which should be carefully discussed .
actually , there have been several conflicts about the meaning of the delocalization phase transition in the standard skyrme model with finite density @xcite . within the traditional works in 1970 , the direct linking of the skyrme model with qcd was still uncertain .
furthermore , the standard skyrme model does not manifestly include quarks and gluons , so that large care should be taken to relate the delocalization of the skyrmion with the deconfinement of the baryon as the appearance of quark - gluon dynamics . in our case
, however , the new concept of holography allows us to derive the skyrmion picture from qcd itself , which is discussed in sec .
[ bis_r3 ] .
moreover , one should also note that quark degrees of freedom can be represented by `` pions '' , suggested in some qcd phenomenologies with `` cheshire cat picture '' as in the chiral quark model @xcite derived from the bosonization scheme for qcd . with these backgrounds , here we propose that the delocalization phase transition in the bis model can be more admissibly related with the deconfinement of the baryons in qcd , relative to the standard skyrme models in 1970. next we consider the chiral symmetry restoration with the order parameter .
normally , in case of the linear sigma model , the chiral symmetry restoration can be signaled by the vanishing of the expectation value of the meson fields at the matter ground state as @xmath279 where the sigma meson field @xmath280 and three pseudoscalar pion fields @xmath281 are called as the `` linear '' realization of chiral symmetry .
we relate this terminology to the non - linear sigma model , since the meson effective action and also the skyrmion picture are based on the `` non - linear '' realization of the chiral symmetry . in the non - linear sigma model ,
the meson fields @xmath280 and @xmath281 are introduced from the chiral field @xmath282 as @xmath283 since @xmath284 is unitary matrix with the condition @xmath285 , the squared sum of the meson fields @xmath280 and @xmath281 in eq .
( [ sp1 ] ) should be constant everywhere as @xmath286 therefore , as far as the action is written in the chiral field @xmath284 with fixed @xmath287 , the meson fields @xmath280 and @xmath281 are forced on a surface of a three - dimensional closed manifold @xmath288 with finite radius @xmath124 in the internal space .
in fact , the existence of such closed manifold @xmath288 is essential for the concept of a skyrmion , which is a non - trivial winding of the compactified physical manifold @xmath4 around the other closed manifold @xmath288 with conserved topological charge , belonging to the homotopical classification @xmath289 @xcite .
however , by comparing eq .
( [ nls1 ] ) with eq .
( [ ncr1 ] ) , one would always encounter a problem of how to describe the chiral symmetry restoration with its non - linear realization . now in this paper , we take a spatially - averaged condensate of the meson fields @xmath280 and @xmath281 as the order parameter of the chiral symmetry restoration with the non - linear realization @xcite as @xmath290 here , the bracket in eq.([exp1 ] ) denotes the three - dimensional spatial average in the medium with volume @xmath291 as @xmath292 such spatially - averaged condensate of the meson fields is somehow similar to the spatially - averaged magnetization @xmath293 as the global order parameter of the ferromagnetic material like bulk iron . within the bulk iron at sufficiently high temperature ,
the spins of the ions orient randomly to get the entropy gain in the free energy . as the temperature decreases below the critical temperature without the external magnetic field ,
there appear magnetic domains with non - zero local magnetization @xmath294 .
however , because of the total angular momentum conservation , the spatially - averaged magnetization @xmath293 should vanish with the appearance of a complex structure of magnetic domain walls within the bulk iron , which gives _ no _ breaking of the global spatial symmetry macroscopically in the bulk material as @xmath295 . in this sense , the spatially - averaged condensate of the meson fields in eq.([exp1 ] ) can be regarded as the `` global '' order parameter of the chiral symmetry in the bulk hadronic matter . by taking the hedgehog configuration ansatz @xmath296 as in eq.([hh ] ) , the meson fields @xmath280 and @xmath281 can be written as @xmath297 note here that the pion field with the hedgehog ansatz in eq.([hh_pi ] ) is proportional to the unit directional vector @xmath298 , so that its spatial average becomes trivially zero .
therefore only the sigma meson field should be considered as the global order parameter of the chiral symmetry .
if a skyrmion is well localized around the north pole at @xmath237 on the manifold @xmath4 , the classical meson filed configuration @xmath160 becomes zero almost everywhere except for the localized point @xmath299 .
therefore the sigma meson configuration in eq.([hh_sigma ] ) becomes @xmath300 almost everywhere .
in other words , the vacuum is so much oriented even globally in the medium , thus its spatial average becomes non - zero as @xmath301 indicating the appearance of the chiral symmetry broken phase . on the other hand , if the skyrmion gives uniform energy density distribution as the identity map @xmath302 in eq.([bis_ide ] ) , the sigma meson configuration becomes @xmath303 , changing monotonously from @xmath304 to @xmath305 over the coordinate space @xmath4 through the arc distance @xmath306 $ ] .
therefore its spatial average vanishes as @xmath307 indicating the appearance of the chiral symmetry restored phase .
these considerations about eqs .
( [ op1 ] ) and ( [ dop1 ] ) also indicate that the chiral symmetry restoration is indirectly related with the energy density distribution in the hadronic matter through the classical meson field configurations .
+ in fig .
[ fig_cc ] , we show the spatially - averaged condensate @xmath308 for the skyrme model and the bis model with radius @xmath5 of @xmath4 .
the spatial average with the hedgehog configuration ( [ hh_sigma ] ) is explicitly taken for each closed manifold @xmath4 as @xmath309 for sufficiently large @xmath5 as the low density state , the skyrme soliton is well localized for both the skyrme model and the bis model , as previously shown in fig [ fig_df ] . therefore , as discussed in eq .
( [ op1 ] ) , the spatially - averaged condensate of the meson field becomes @xmath310 as the chiral symmetry breaking in both models as shown in fig .
[ fig_cc ] . in this sense , @xmath1 meson field ( and also the other vector and axial vector meson fields , if they are included ) would have less importance for the chiral symmetry breaking in the bulk hadronic matter as far as sufficiently localized baryon appears in the medium . with the decrease of @xmath5
, @xmath308 monotonously decreases , which corresponds to the partial chiral symmetry restoration as the medium effect in the baryonic matter .
the chiral symmetry is fully restored with @xmath311 at the critical radius @xmath278 for the skyrme model and at @xmath227 for the bis model .
actually the small radius @xmath5 , i.e. , the larger baryon - number density is needed to give the chiral symmetry restoration due to the @xmath1 meson fields , in similar to the case of the delocalization phase transition as previously discussed . finally , we comment about the relation between the deconfinement and the chiral symmetry restoration in qcd , which has been discussed so long in these decades of the hadron physics .
the relation between two phase transitions is nontrivial since they are characterized by the different symmetries : the global chiral symmetry @xmath312 , and z(3 ) symmetry as the center of @xmath313 gauge group .
note that the former can be defined in the massless quark limit , while the latter in the heavy quark limit , leading to the difficulty to discuss their relation . despite of the difficulty ,
several implications have been obtained : in finite temperature , the lattice qcd studies suggest that the two phase transitions occur at the same critical temperature @xmath314 @xcite .
such simultaneous occurrence is supported by the analyses of the nambu - jona - lasinio model with a polyakov loop ( pnjl model ) as a low - energy effective theory of qcd @xcite .
actually , these two phase transitions are separated with respect to their dominant symmetries as mentioned above , so that such mysterious coherence between two phase transitions is often compared to some kinds of `` entanglement '' @xcite .
furthermore , the simultaneous occurrence of the two phase transitions is also suggested by the recent analysis of the holographic qcd with d4/d8/@xmath14 multi - d brane configurations : two independent chirality spaces on the d8 and @xmath14 branes respectively are connected with each other ( chiral symmetry breaking ) by the `` worm hole '' of the d4 supergravity background into which the colored objects like quarks and gluons are absorbed ( color confinement ) @xcite . in this sense , these two phase transitions are more directly related with each other as just single event on the `` worm hole '' in the extra - dimensions , the effects of which might appear as some kinds of mysterious entanglement in view of the four - dimensional space - time .
now , the situation has not been clear in finite density . in this work ,
we indirectly relate the two phase transitions through the meson field configurations .
admittedly our approach includes only meson fields and baryons appear as mesonic solitons .
however , if the cheshire cat picture holds and quark - gluon dynamics can be indirectly expressed in the meson dynamics as an effective model , our results could be interpreted as the simultaneous occurrence of two phase transition in the finite density qcd .
investigations in this direction are interesting and should be developed further .
in this section , we show the critical densities of the phase transitions for the skyrme model and the bis model with recovering the physical units . in this study , single baryon is placed on a closed manifold @xmath4 with the surface volume @xmath315 , so that the total baryon - number density @xmath316 can be given as @xmath317 } .
\label{rho_cri1}\end{aligned}\ ] ] now , in the holographic model , by fixing two parameters , e.g. , experimental inputs for @xmath124 and @xmath125 , all the physical quantities like masses and the coupling constants are uniquely determined , which is a remarkable consequence of the holographic approach discussed in sec .
first , we take @xmath124 and @xmath125 as experimental values , @xmath318 then the skyrme parameter @xmath140 in eq.([e_kappa ] ) can be uniquely determined as @xmath319 with these experimental inputs , we show in fig .
[ fig_rho_c ] the total baryon - number density @xmath316 in eq .
( [ rho_cri1 ] ) as a function of radius @xmath5 of @xmath4 , divided by the normal nuclear density @xmath320 . from fig.[fig_rho_c ] and also eq.([rho_cri1 ] ) , the critical densities for the skyrme model ( @xmath321 ) and the bis model ( @xmath322 ) are found as follows : @xmath323 the heavy @xmath1 meson fields in the core region of the baryon tend to decrease the total size of the baryon @xcite , so that the @xmath1 meson field has a significant role to increase the critical density into the uniform phase as in eq .
( [ rho_c_bis ] ) relative to eq .
( [ rho_c_ss ] ) .
+ in sec . [ ecpro ] , we discuss the disappearance of the @xmath1 meson field in high density phase of baryonic matter , while we now find the significant roles of @xmath1 meson field as for the critical density in eq .
( [ rho_c_bis ] ) relative to eq .
( [ rho_c_ss ] ) .
this situation is somehow resemble to the `` two - higgs model '' with scalar fields @xmath324 and @xmath325 in the finite temperature . on the critical temperature of a phase transition with the dynamics of a scalar filed @xmath325 ,
the condensate of the @xmath325 field might become trivial as @xmath326 .
however , in general , the mass of @xmath325 field and also the interactions between @xmath325 and @xmath324 fields affect the critical phenomena like critical temperature . in this sense , even if @xmath1 meson field disappear near critical point , they could affect the critical phenomena as in eq .
( [ rho_c_bis ] ) through its mass and also the interactions with pions in the action @xmath150 in eq .
( [ dbi ] ) . here
we briefly comment about the value of the critical density by taking other parameter set used by adkins _
et al . _
@xcite . in 1983 ,
_ analyzed the baryon mass spectra with the semiclassical quantization of the skyrme model , whereas the @xmath30-@xmath31 splitting corresponds to the higher order contribution @xmath19 relative to the static skyrme action @xmath327 in the large-@xmath3 expansion . to reproduce the proper amount of the @xmath30-@xmath31 splitting ,
they take the smaller pion decay constant and skyrme parameter @xcite as @xmath328 now if the parameter set in eq.([anw_input1 ] ) is used for the baryon - number density in eq .
( [ rho_cri1 ] ) , we find too small critical density @xmath329 for the skyrme model . if one can relate the phase transition in the skyrme model with the deconfinement of the baryon in qcd , it might propose , e.g. , that quark degrees of freedom are already manifest in the normal nuclei .
this unreasonable results may come from the fact that semiclassical quantization procedure corresponds to extracting higher order contributions @xmath19 from the leading order skyrme action @xmath327 as for the large-@xmath3 expansion .
same comments can also be applied to the holographic model if one try to perform the semiclassical quantization as the baryon analysis .
in fact , holographic qcd is derived as the large-@xmath3 effective theory even by starting from superstring framework .
higher order effects of the large-@xmath3 expansion like the baryon mass splitting correspond to the string loop effects beyond the classical supergravity , which should be fairly untractable . with these considerations
above , we do not intentionally proceed to the semiclassical quantization of skyrme soliton in the present work , and we employ the experimental inputs ( [ exp_input1 ] ) and skyrme parameter ( [ e_sky_exp_input ] ) in the analysis of the critical densities with recovering the physical units .
actually , due to the scaling property of the bis model discussed in sec .
[ mea ] , all the results in anw unit in the previous sections should not be altered , being independent of the definite values for @xmath124 and @xmath125 .
we have studied baryonic matter in holographic qcd with @xmath0 multi - d brane configurations , by analyzing the system of single brane - induced skyrmion on the three - dimensional closed manifold @xmath4 . by changing the size of @xmath4 , the density dependence of the baryon properties
are examined from a holographic point of view .
first we begin with the dirac - born - infeld ( dbi ) action of the probe d8 brane with d4 supergravity background . with the dimensional reductions , we get the five - dimensional yang - mills action with a curved fifth dimension , as the leading order of the large-@xmath3 and large thooft coupling expansions in dual of non - perturbative ( strong - coupling ) qcd . through the mode expansions of the five - dimensional gauge fields , we get the four - dimensional meson effective action from holographic qcd .
especially , we emphasize the appearance of the ultraviolet cutoff scale @xmath2 in the holographic approach to be dual of qcd .
therefore we construct the four - dimensional meson effective action with pion and @xmath1 meson fields below the cutoff scale @xmath15 .
then we discuss the baryon in holographic qcd as the `` brane - induced skyrmion '' in the four - dimensional meson effective action .
the analyses of the baryon in the meson effective action with restricted degrees of freedom below the cutoff scale @xmath15 is called the `` truncated resonance model '' for the baryon . taking the hedgehog configuration ansatz for pion and @xmath1 meson fields
, we can investigate many properties of single baryon as the brane - induced skyrmion , which are inclusively summarized in ref .
then we consider the baryonic matter in holographic qcd , as the extension of holographic approach to dense qcd .
especially we treat the baryonic matter with large-@xmath3 , because holographic qcd is derived as the large-@xmath3 effective theory , in dual of the classical supergravity . for sufficiently large @xmath3
, the kinetic energy and the quantum effects can be suppressed relative to the static mass from simple large-@xmath3 countings @xcite , where the baryonic matter comes into the static skyrme matter . in this sense , we analyze the static skyrme matter to see the typical features of the baryonic matter with large-@xmath3 conditions . in order to analyze the static skyrme matter on the flat coordinate space @xmath25 , we alternately treat the system of single brane - induced skyrmion on the three - dimensional closed manifold @xmath4 .
the interactions between the baryons are simulated by the curvature of the closed manifold @xmath4 , and , as the size of @xmath4 decreases , the baryon - number density increases in this modeling .
actually , through the projection procedure from the flat space @xmath25 onto the curved space @xmath4 , we get the hedgehog mass and also the euler - lagrange equations for pion and @xmath1 meson fields of the brane - induced skyrmion on @xmath4 . by numerically solving the euler - lagrange equations for pion and @xmath1 meson fields on @xmath4 , we find a stable soliton solution as the brane - induced skyrmion on @xmath4 . by using this solution
, we analyze many properties of the baryon within the baryonic matter . especially by comparing the standard skyrme model without @xmath1 mesons and the brane - induced skyrme ( bis ) model ,
the roles of ( axial ) vector mesons in dense qcd are discussed from a holographic point of view .
first we show the baryon - number density dependence of the energy density and the field configuration profiles of single baryon , by changing the size of @xmath4 . as the size of @xmath4 decreases ,
the localized energy density distribution of single baryon becomes uniform one as the identity map , which is called the `` delocalization phase transition '' .
the critical radii of the phase transitions are given as @xmath330 $ ] for the skyrme model and @xmath331 $ ] for the bis model .
because of the shrinkage of the baryon size due to the @xmath1 meson effects @xcite , the smaller critical radius of @xmath4 , i.e. , the larger baryon - number density is needed for the bis model to give the delocalization phase transition .
we also find that @xmath1 meson field absolutely disappears and only pion field survives near the critical density .
then , with the mathematical arguments , we propose a remarkable conjecture that all the ( axial ) vector meson fields would disappear and only pion field survives near the critical density , referred as the `` pion dominance '' in dense baryonic matter .
we also investigated the baryon - number density dependence of the mass and the root - mean - square ( rms ) mass radius of single baryon .
we find the the `` swelling '' phenomenon of the baryon , as the non - linear increase of the rms mass radius near the critical density .
actually , non - perturbative qcd vacuum around the baryons are closely packed as the baryon - number density increases .
therefore , the non - perturbative effect with long wave length like the bag pressure would be blocked out from such definite region to give the swelling of the baryon .
such swelling provides the the decrease of the baryon mass splitting , indicating the lost of the individualities for the baryons as the precursor of the deconfinement .
we also propose the stable @xmath30-@xmath31 mixed matter in dense qcd because of the swelling , which could , e.g. , soften the eos of the neutron stars .
the features of the delocalization phase transition and the chiral symmetry restoration are also analyzed with the order parameters in the holographic qcd .
we conjecture with careful arguments that the delocalization phase transition can be related with the deconfinement of the baryons in qcd .
then we find the coherence of the deconfinement and the chiral symmetry restoration through the meson field configurations .
such coherence of the two phase transitions are also suggested in the lattice qcd study in the finite temperature @xcite , the pnjl model as a low - energy effective theory of qcd @xcite , and also the sakai - sugimoto model as one of the reliable holographic approaches @xcite .
we also calculate the critical densities of the phase transitions with the experimental inputs for the pion decay constant @xmath124 and @xmath1 meson mass @xmath125 as @xmath332 and @xmath333 .
we find the critical densities as @xmath334 for the skyrme model and @xmath335 for the bis model .
we can see that the larger baryon - number density is needed for the bis model to give the phase transitions , because of the shrinkage of total size of the baryon with heavy @xmath1 mesons in its core region .
finally we compare our truncated resonance approach for the baryons with the other works of baryons as the _ instantons _ in holographic qcd .
in fact , there seem to exist some conflicts for the baryon analysis in holographic qcd , especially between the truncated resonance model and the instanton models . in refs .
@xcite , the baryons are studied as the instantons on the five - dimensional gauge theory of the probe d8 brane with d4 supergravity background as the holographic dual of qcd .
the properties of the baryonic matter are also analyzed by the system of the single instanton on the three - dimensional closed manifold @xmath4 as the wigner - seiz approximation @xcite . in these analyses ,
the instanton is found to shrink into zero size only for the dbi sector of the effective action of the d8 brane , which corresponds to the leading order of large-@xmath3 and large thooft coupling expansions . in the holographic approach ,
infinite tower of color - singlet modes with mesonic quantum numbers as @xmath1 , @xmath244 , @xmath336 , @xmath337 , @xmath338 , @xmath339 , appears in the mode expansions of the five - dimensional gauge field @xmath340 on the d8 brane as in eq.([limit2_lr ] ) . in this sense ,
the baryon as the instanton on the five - dimensional gauge theory before the mode expansion is in principle to be composed by the infinite tower of such color - singlet modes with mesonic quantum numbers .
therefore , the instability of the instanton only with the dbi sector is often regarded that the infinite number of ( axial ) vector mesons @xmath1 , @xmath244 , @xmath336 , @xmath337 , @xmath338 , @xmath339 , would affect the low - energy soliton feature to give the zero size . then ,
in order to describe the baryon as an instanton with finite size in the holographic approach , the inclusion of the chern - simons ( cs ) sector of the effective action of the d8 brane is claimed , whereas the cs sector corresponds to the higher order contribution of the thooft coupling expansion relative to the dbi sector .
however , as emphatically noted in sec .
[ mea ] , the appearance of certain cutoff scale like @xmath2 in holographic approach should be essential to be dual of qcd .
in fact , the mesonic mass spectra predicted from holographic qcd starts to deviate from the experimental data beyond the @xmath15 scale , indicating that the holographic duality with qcd is mainly maintained almost below @xmath15 .
this tendency of the mass spectra also denotes that such color - singlet modes beyond cutoff scale @xmath15 do not directly correspond to physical mesons in qcd .
therefore , one _ can not _ manifestly conclude that the instability of the instanton only with the dbi sector really arises from the physical effect in qcd .
furthermore , there also exist the infinite number of non - qcd modes , e.g. , the kaluza - klein modes with large mass @xmath16 through the kaluza - klein compactification of the d4 brane @xcite . therefore
, if the baryon as the instanton is to be really composed by the infinite tower of the color - singlet non - qcd modes even beyond @xmath15 , there is no reason to cast away the kaluza - klein modes in the baryon analysis .
in fact , such kaluza - klein modes still have the possibility to affect baryon properties , giving , e.g. , the stability of the instanton even without the cs sector of the effective action of the d8 brane , through it is not to be manifestly proved yet . as a whole , there still seems to remain puzzling conflicts between the truncated resonance model and instanton models for the baryon analyses in holographic qcd . by looking back the long history of hadron physics , baryons seem to have inherent difficulties to be described relative to mesons . for example , the quark model succeeded in the systematic classification of the hadrons in terms of their valence quarks @xcite .
however , while the dynamics of mesons as the two - body composites can be relatively easily analyzed by solving the bethe - salpeter equation @xcite , we have to treat the complicated faddeev equation @xcite to describe the dynamics of the baryons as the three - body composites .
same difficulties are also found in the lattice qcd studies , for instance , with respect to the inter - quark potential of mesons @xcite and baryons @xcite .
in fact , the quark - antiquark potential as the internal nature of the meson could be successfully measured in lattice qcd in 1980 @xcite , while it takes almost twenty years after that to find the well measurement of three - quark potential with the appearance of gluonic y - type flux tube as the internal nature of the baryon @xcite .
the difficulty of the baryons gets more apparent in the large-@xmath3 qcd @xcite , which has provided a powerful perturbative treatment with the @xmath22 expansion for the non - perturbative aspects of the strong interaction .
in fact , large-@xmath3 qcd achieved large successes in the explanations for the hadron phenomenology , e.g. , the okubo - zweig - iizuka ( ozi ) rule @xcite , the @xmath341 rule @xcite , and the narrowness of meson resonances @xcite .
it also provided a quantitative formula for the large @xmath342 meson mass as the veneziano - witten formula @xcite . on the other hand
, the baryon does not even appear as the dynamical degrees of freedom in the large-@xmath3 qcd , because the baryon mass is proportional to @xmath18 .
in other words , @xmath3 quarks are needed to construct the su(@xmath3 ) color - singlet composite as a baryon only from quarks , while a meson can still be constructed as the two - body color - singlet composite from a quark and an antiquark . in the end
, the baryon appears as a `` non - local object '' , i.e. , the soliton of the meson fields in the large-@xmath3 qcd . as a whole ,
the baryons tend to suffer from the many - body difficulties relative to the mesons , by seeing the long hadron history .
now , we suppose here that the recent baryon analyses in holographic qcd may suffer from the same kind of difficulties inherent in the baryon itself .
in fact , meson properties are successfully described in the framework of holographic qcd as the sakai - sugimoto model @xcite whereas the baryons seem to be not sufficiently described yet .
one should note that the baryon mass splitting corresponds to the higher - order contributions of the large-@xmath3 expansion . in terms of the superstring theory ,
such baryon mass splitting corresponds to the string loop effect beyond the classical supergravity , which is fairly untractable .
these naive considerations imply that the baryon still drags the essential difficulties even by starting from the superstring theory .
we have to sincerely reconsider the meaning of the difficulty existing in the baryon itself and identify an origin of the problems , which would inspire us to cure the recent conflicts of the baryon analyses in the holographic approach for the future .
authors thank josh erlich and dong - pil min for their communications about our truncated resonance model as the baryon analysis in holographic qcd in comparison with the instanton models .
is also indebted to atsushi hosaka and hiroshi toki for their discussions in research center for nuclear physics ( rcnp ) .
is supported by a grant - in - aid for scientific research [ ( c ) no .
19540287 ] in japan .
is supported by special postdoctoral research program of riken .
this work is supported by the global coe program , `` the next generation of physics , spun from universality and emergence '' .
k. nawa , h. suganuma and t. kojo , phys .
d * 75 * , 086003 ( 2007 ) [ arxiv : hep - th/0612187 ] . k. nawa , h. suganuma and t. kojo , prog .
. suppl . * 168 * , 231 ( 2007 ) [ arxiv : hep - th/0701007 ] ; mod .
a23 * , 2364 ( 2008 ) [ arxiv:0806.3040 [ hep - th ] ] ; prog .
. suppl . * 174 * , 347 ( 2008 ) [ arxiv:0806.3041 [ hep - th ] ] ; h. suganuma , k. nawa and t. kojo , nucl .
* b * ( proc . suppl . )
( _ to appear _ ) [ arxiv:0809.0805 [ hep - th ] ]
. h. hata , t. sakai , s. sugimoto and s. yamato , prog .
phys . * 117 * , 1157 ( 2007 ) . c. t. hill , phys .
. lett . * 88 * , 041601 ( 2002 ) [ arxiv : hep - th/0109068 ] .
h. boschi - filho , n. r. f. braga , eur .
j. c*32 * , 529 ( 2004 ) [ hep - th/0209080 ] ; jhep * 05 * , 009 ( 2003 ) [ arxiv : hep - th/0212207 ] .
d. t. son and m. a. stephanov , phys . rev .
d * 69 * , 065020 ( 2004 ) [ arxiv : hep - ph/0304182 ] .
y. brihaye , c. t. hill , c. k. zachos , phys .
rev . d*70 * , 111502 ( 2004 ) [ arxiv : hep - th/0409222 ] .
j. erlich , e. katz , d. t. son and m. a. stephanov , phys .
* 95 * , 261602 ( 2005 ) [ arxiv : hep - ph/0501128 ] .
g. f. de teramond and s. j. brodsky , phys .
lett . * 94 * , 201601 ( 2005 ) [ arxiv : hep - th/0501022 ] .
s. j. brodsky and g. f. de teramond , phys .
lett . * 96 * , 201601 ( 2006 ) [ arxiv : hep - ph/0602252 ] .
a. wulzer and a. pomarol , jhep * 03 * , 051 ( 2008 ) [ arxiv:0712.3276 [ hep - th ] ] ; nucl
b809 * , 347 ( 2009 ) [ arxiv:0807.0316 [ hep - ph ] ] .
y. kim , c .- h .
lee and h .- u .
yee , phys .
d * 77 * , 085030 ( 2008 ) .
y. igarashi , m. johmura , a. kobayashi , h. otsu , t. sato and s. sawada , nucl . phys .
b * 259 * , 721 ( 1985 ) ; t. fujiwara , y. igarashi , a. kobayashi , h. otsu , t. sato and s. sawada , prog .
. phys . * 74 * , 128 ( 1985 ) . c. ratti , m.a .
thaler and w. weise , phys .
d * 73 * , 014019 ( 2006 ) [ arxiv : hep - ph/0506234 ] ; s. rner , c. ratti and w. weise , phys .
d * 75 * , 034007 ( 2007 ) [ arxiv : hep - ph/0609281 ] ; c. ratti , s. rner and w. weise , phys . lett .
* b649 * , 57 ( 2007 ) [ arxiv : hep - ph/0701091 ] ; k. fukushima , phys . lett . * b591 * , 277 ( 2004 ) [ arxiv : hep - ph/0310121 ] .
m. creutz , phys .
lett . * 43 * , 553 ( 1979 ) , _ erratum - ibid .
_ * 43 * , 890 ( 1979 ) ; phys . rev .
d * 21 * , 2308 ( 1980 ) .
t. t. takahashi _
et al . _ , phys .
* 86 * , 18 ( 2001 ) ; phys . rev .
d * 65 * , 114509 ( 2002 ) .
buras , j.m .
gerard , nucl .
b264 * , 371 ( 1986 ) ; w.a .
bardeen , a.j .
buras and j .-
gerard , phys .
b*180 * , 133 ( 1986 ) . | we study baryonic matter in holographic qcd with @xmath0 multi - d brane system in type iia superstring theory .
the baryon is described as the `` brane - induced skyrmion '' , which is a topologically non - trivial chiral soliton in the four - dimensional meson effective action induced by holographic qcd .
we employ the `` truncated - resonance model '' approach for the baryon analysis , including pion and @xmath1 meson fields below the ultraviolet cutoff scale @xmath2 , to keep the holographic duality with qcd .
we describe the baryonic matter in large @xmath3 as single brane - induced skyrmion on the three - dimensional closed manifold @xmath4 with finite radius @xmath5 .
the interactions between baryons are simulated by the curvature of the closed manifold @xmath4 , and the decrease of the size of @xmath4 represents the increase of the total baryon - number density in the medium in this modeling .
we investigate the energy density , the field configuration , the mass and the root - mean - square radius of single baryon on @xmath4 as the function of its radius @xmath5 .
we find a new picture of `` pion dominance '' near the critical density in the baryonic matter , where all the ( axial ) vector meson fields disappear and only the pion field survive .
we also find the `` swelling '' phenomena of the baryons as the precursor of the deconfinement , and propose the mechanism of the swelling in general context of qcd .
the properties of the deconfinement and the chiral symmetry restoration in the baryonic matter are examined by taking the proper order parameters .
we also compare our truncated - resonance model with another `` instanton '' description of the baryon in holographic qcd , considering the role of cutoff scale @xmath6 . |
the main motivation of this work is to find the density of the first time @xmath0 , that a one - dimensional , standard brownian motion @xmath1 , reaches the moving boundary @xmath2 : @xmath3 where @xmath4 , @xmath5 , and @xmath6 for all @xmath7 . alternatively , assuming the previous assumptions : given that @xmath8 is the density of the first time that @xmath1 reaches the fixed level @xmath9 , the process @xmath10 is a 3-dimensional bessel bridge , which has the following dynamics : @xmath11 ( where @xmath12 is a wiener process . )
then , the distribution of @xmath0 equals @xmath13\\ \label{cota}&&\quad\times \exp\left\{-\frac{1}{2}\int_0^s(f'(u))^2du - f'(0)a\right\}h(s , a)ds\end{aligned}\ ] ] from girsanov s theorem , the fact that the boundary @xmath2 is twice continuously differentiable , the following radon - nikodym derivative : @xmath14 is indeed a _ martingale _ [ novikov s condition ] and induces the following relationship : @xmath15\\ & = & \tilde{\mathbb{e}}\left[\frac{d\mathbb{p}}{d\tilde{\mathbb{p}}}(t)\mathbb{i}_{(t\leq t)}\right]\\ & = & \int_0^t\tilde{\mathbb{e}}\left[\frac{d\mathbb{p}}{d\tilde{\mathbb{p}}}(t)\big{|}t = s\right]h(s , a)ds,\end{aligned}\ ] ] where the second equality follows from the optional sampling theorem , and the third from conditioning with respect to @xmath0 under @xmath16 , and @xmath17 is defined in ( [ level ] ) . in revuz and yor ( 1999 ) ( chapter 11 )
, it is shown that calculating : @xmath18\ ] ] is equivalent to finding the expected value of a functional of a 3-dimensional bessel bridge @xmath10 , which at time 0 starts at @xmath9 and at time @xmath19 equals 0 .
indeed , @xmath20\\ & & \nonumber\quad= \tilde{\mathbb{e}}\bigg{[}\exp\bigg{\{}-f'(t)a+\int_0^tf''(u)\tilde{b}_udu\\ & & \nonumber\quad\qquad -\frac{1}{2}\int_0^t(f'(u))^2du\bigg{\}}\big{|}t = s\bigg{]}\\ & & \nonumber\quad= \tilde{\mathbb{e}}\left[\exp\left\{-f'(s)a+\int_0^sf''(u)(a-\tilde{x}_u)du-\frac{1}{2}\int_0^s(f'(u))^2du\right\}\right]\\ & & \nonumber\quad = \tilde{\mathbb{e}}\bigg{[}\exp\bigg{\{}-f'(s)a+(f'(s)-f'(0))a-\int_0^sf''(u)\tilde{x}_udu\\ & & \nonumber\quad\qquad-\frac{1}{2}\int_0^s(f'(u))^2du\bigg{\ } } \bigg{]}\\ & & \label{problem}\quad = \tilde{\mathbb{e}}\left[\exp\left\{-\int_0^sf''(u)\tilde{x}_udu\right\}\right]\\ & & \nonumber\qquad\quad\times\exp\left\{-\frac{1}{2}\int_0^s(f'(u))^2du - f'(0)a\right\}\end{aligned}\ ] ] thus the process @xmath21 equals 0 at @xmath22 , and at @xmath19 reaches level @xmath9 for the first time , as required [ see figures [ f1][f3 ] ] . from equation ( [ problem ] )
it is clear that our next goal is to compute the following expected value : @xmath23.\ ] ] suppose that @xmath24\times \mathbb{r}^+\to\mathbb{r}^+$ ] is continuous , is of class @xmath25 and satisfies the cauchy problem @xmath26 as well as @xmath27 then @xmath28 admits the stochastic representation @xmath29\end{aligned}\ ] ] we proceed as in the proof of theorem 5.7.6 , pp .
366367 , karatzas & shreve ( 1991 ) .
applying ito s rule to the process @xmath30
@xmath31 $ ] , and obtain , with @xmath32 , @xmath33\\ & & + \tilde{\mathbb{e}}^{t , a}\left[v(\tau_n , x_{\tau_n})\exp\left\{-\int_t^{\tau_n}f''(u)\tilde{x}_udu\right\}\mathbb{i}_{(\tau_n\leq s)}\right]\end{aligned}\ ] ] the second term converges to zero as @xmath34 , since @xmath35\\&&\qquad\qquad\leq \tilde{\mathbb{e}}^{t , x}\left[v(\tau_n,\tilde{x}_{\tau_n})\mathbb{i}_{(\tau_n\leq s)}\right]\\ & & \qquad\qquad\leq\tilde{\mathbb{p}}^{t , x}(\tau_n\leq s)\\ & & \qquad\qquad=\tilde{\mathbb{p}}^{t , x}\left(\max\limits_{t\leq\theta\leq s}\tilde{x}_\theta\geq n\right)\\ & & \qquad\qquad\leq \frac{\tilde{\mathbb{e}}^{t , x}\left[\max\limits_{t\leq\theta\leq s}\tilde{x}_\theta^{2m}\right]}{n^{2m}}\end{aligned}\ ] ] [ see pitman and yor ( 1998 ) for the moments of the running maximum of @xmath10 ] .
finally , the first term converges to @xmath37\end{aligned}\ ] ] either by the dominated or by the monotone convergence theorem . ,
width=240,height=201 ] , width=240,height=201 ] , width=240,height=201 ]
solutions to ( [ cau ] ) are of the following form @xmath38 where @xmath39 and @xmath40 as in ( [ level ] ) . setting @xmath41 and @xmath42 , we have that @xmath43u(t , a)\\ u_a(t , a)&=&-\left[\frac{1}{a}-\frac{a}{s - t}\right]u(t , a)\\ u_{aa}(t , a)&=&\left[\frac{2}{a^2}+\frac{a^2}{(s - t)^2}-\frac{1}{s - t}\right]u(t , a)\end{aligned}\ ] ] and @xmath44 hence from ( [ cau ] ) and ( [ schro ] ) it follows that @xmath45w=\left[u_a+\left(\frac{1}{a}-\frac{a}{s - t}\right)u\right]w_a,\end{aligned}\ ] ] as claimed .
let @xmath48 applying the fourier transform to ( [ schro ] ) we have @xmath49 next , set @xmath50 and @xmath51 , that is : @xmath52 which after substitution in ( [ fou ] ) leads to @xmath53 consequently @xmath54 which alternatively implies that @xmath55 as claimed .
xx karatzas , i. and s. shreve .
brownian motion and stochastic calculus , springer - verlag , new york .
pitman , j. , and m. yor ( 1998 ) .
the law of the maximum of a bessel bridge .
technical report no .
revuz , d. , and m. yor .
continuous martingales and brownian motion , springer - verlag , new york . | in this work we relate the density of the first - passage time of a wiener process to a moving boundary with the three dimensional bessel bridge process and a solution of the heat equation with a moving boundary .
we provide bounds . |
for a graph @xmath2 , a function @xmath3 is called a _ distribution on the vertices of @xmath4 _ , or a _ distribution on @xmath4_.
we usually imagine that @xmath5 pebbles are placed on @xmath6 for each vertex @xmath7 .
let @xmath8 denote the _ size _ of @xmath9 , i. e. @xmath10 . for two distributions @xmath9 and @xmath11 on @xmath4
, we say that @xmath9 _ contains _
@xmath11 if @xmath12 for all @xmath13 .
we allow pebbling moves on the graph , and define the _ pebbling number _ , the _ optimal pebbling number _ , the _ @xmath0-pebbling number _ , and the _ optimal @xmath0-pebbling number _ of a graph as follows : * definitions * : a _ pebbling move _ in @xmath4 takes two pebbles from a vertex @xmath7 , which contains at least two pebbles , and places a pebble on a neighbor of @xmath6 . for two distributions @xmath14 and @xmath15 , we say that @xmath15 is _ reachable _ from @xmath14 if there is some sequence of pebbling moves beginning with @xmath14 and resulting in a distribution which contains @xmath15 .
we say the distribution @xmath9 is _ solvable _ , ( respectively , _
@xmath0-solvable _ ) , if every distribution with one pebble ( respectively , @xmath0 pebbles ) on a single vertex is reachable from @xmath9 .
the traditional _ pebbling number _ , and _ @xmath0-pebbling number _ of a graph @xmath4 , denoted @xmath16 and @xmath17 respectively , were defined by chung @xcite .
the _ optimal pebbling number _ and _ optimal @xmath0-pebbling number _ of @xmath4 , denoted @xmath18 and @xmath19 respectively , were defined by pacther , snevily , and voxman @xcite .
we give those definitions now .
* definitions ( chung @xcite and pachter _ et al . _
@xcite ) * : the _ @xmath0-pebbling number _ of @xmath4 is the smallest number @xmath17 such that every distribution @xmath9 with @xmath20 is @xmath0-solvable .
the _ optimal @xmath0-pebbling number _ of @xmath4 , denoted @xmath19 , is the smallest number such that some distribution with @xmath17 pebbles is @xmath0-solvable . in both cases
we omit the @xmath0 when @xmath21 .
thus , the _ pebbling number _ of @xmath4 is @xmath22 and the _ optimal pebbling number _ of @xmath4 is @xmath23 .
the pebbling number was generalized in @xcite to allow for an arbitrary set of target distributions .
we define this generalization and extend it to define the optimal pebbling number of a set of distributions on @xmath4 .
* definitions ( @xcite ) * : let @xmath24 be a set of distributions on a graph @xmath4 .
we say a distribution @xmath9 is _
@xmath24-solvable _ if every distribution in @xmath24 is reachable from @xmath9 .
the _ pebbling number of @xmath24 in @xmath4 _ , denoted @xmath25 , is the smallest number such that every distribution @xmath9 with @xmath26 is @xmath24-solvable .
the _ optimal pebbling number of @xmath24 in @xmath4 _ , denoted @xmath27 , is the smallest number such that some distribution @xmath9 with @xmath28 is @xmath24-solvable . if @xmath29 consists of all distributions with @xmath0 pebbles on a single vertex , we have @xmath30 , @xmath31 , @xmath32 , and @xmath33 .
graham s conjecture asserts a bound on the pebbling number of the cartesian product of two graphs .
* definition * : if @xmath34 and @xmath35 are two graphs , their cartesian product is the graph @xmath36 whose vertex set is the product @xmath37 and whose edges are given by @xmath38 we also write @xmath39 for the graph @xmath40 with @xmath41 copies of @xmath4 in the product . throughout this paper
we follow that convention that @xmath34 and @xmath35 .
chung @xcite attributed conjecture [ graham s conjecture ] to graham .
[ graham s conjecture ] for any graphs @xmath4 and @xmath42 , we have @xmath43 .
conjecture [ graham s conjecture ] was generalized in @xcite to accommodate the more general definitions of pebbling numbers with arbitrary sets of target distributions .
the following definition of products of distributions first appeared in @xcite and the definition of products of sets of distributions appeared in @xcite .
* definition ( @xcite ) * : if @xmath9 and @xmath11 are distributions on @xmath4 and @xmath42 respectively , then we define @xmath44 as the distribution on @xmath45 such that @xmath46 for every vertex @xmath47 .
similarly , if @xmath24 and @xmath48 are sets of distributions on @xmath4 and @xmath42 respectively , then @xmath49 is the set of distributions on @xmath45 given by @xmath50 also , for any integer @xmath51 we define the distribution @xmath52 by @xmath53 for all @xmath54 . for all graphs @xmath4 and @xmath42 , and all sets of distributions @xmath24 and @xmath48 on @xmath4 and @xmath42 respectively , we have @xmath55 . [ general graham ] in this section we prove the analog of conjecture [ general graham ] for optimal pebbling .
let @xmath9 be an @xmath24-solvable distribution on @xmath4 and let @xmath11 be an @xmath48-solvable distribution on @xmath42 . then @xmath44 is an @xmath56-solvable distribution on @xmath57 .
in particular , we have @xmath58 .
[ optimal pebbling distributions ] to show this , we first establish a few lemmas . if @xmath14 and @xmath15 are distributions on the graph @xmath4 such that @xmath15 is reachable from @xmath14 , then for any integer @xmath51 , the distribution @xmath59 is reachable from @xmath60 .
[ multiplying by s ] * proof * : the distribution @xmath60 may be regarded as @xmath51 distinct copies of @xmath14 .
we can reach @xmath15 from each copy of @xmath14 , so @xmath61 is reachable from @xmath60 .
@xmath62 let @xmath4 and @xmath42 be graphs . if @xmath14 and @xmath15 are distributions on @xmath4 such that @xmath15 is reachable from @xmath14 , then for any distribution @xmath11 on @xmath42 , @xmath63 is reachable from @xmath64 .
[ multiplying by d ] * proof * : for each @xmath65 , the number of pebbles on @xmath66 in the distribution @xmath64 is given by @xmath67 .
fix @xmath68 .
we write @xmath69 for the subgraph of @xmath45 induced by the vertices whose second coordinate is @xmath70 .
then @xmath71 , and if we restrict @xmath64 to @xmath72 , we obtain the distribution @xmath73 . since @xmath70 is fixed , @xmath74 is a constant , so by lemma [ multiplying by s ] , the distribution @xmath75 is reachable in @xmath76 . repeating this for each @xmath77 , we end up with a distribution in which each @xmath66 has at least @xmath78 pebbles , so @xmath63 is reachable from @xmath64 .
@xmath62 we are now ready to prove theorem [ optimal pebbling distributions ] .
* proof of theorem [ optimal pebbling distributions ] * : let @xmath9 and @xmath11 be @xmath24- and @xmath48-solvable distributions on @xmath4 and @xmath42 respectively . to show that @xmath44 is @xmath56-solvable on @xmath36 , let @xmath79 be a distribution in @xmath49 .
then we can write @xmath80 for some @xmath81 and @xmath82 .
also , @xmath83 is reachable from @xmath9 and @xmath84 is reachable from @xmath11 .
thus , by lemma [ multiplying by d ] , @xmath85 is reachable from @xmath86 , which is reachable from @xmath44 . if we choose @xmath9 and @xmath11 so that @xmath87 and @xmath88 , we have @xmath89 thus , @xmath44 is an @xmath56-solvable distribution on @xmath45 with @xmath90 ; therefore @xmath58 , as desired .
@xmath62 corollaries [ optimal st - pebbling graphs ] and [ optimal pebbling graphs ] follow immediately from theorem [ optimal pebbling distributions ] .
fu and shiue @xcite announced corollary [ optimal pebbling graphs ] , the optimal pebbling analog to graham s conjecture .
shiue proved it in @xcite .
for all graphs @xmath4 and @xmath42 and all positive integers @xmath51 and @xmath0 , we have @xmath91 .
[ optimal st - pebbling graphs ] for all graphs @xmath4 and @xmath42 , we have @xmath92 .
[ optimal pebbling graphs ]
our work in section [ graham ] puts an upper bound on @xmath93 . in this section ,
we improve those bounds when @xmath4 and @xmath42 are complete graphs .
our main result is theorem [ bounds for pi^*(gxkn ) ] . for any graph @xmath4 and any positive integer @xmath0
, we have @xmath94 .
[ bounds for pi^*(gxkn ) ] lemma [ at most one odd vertex ] helps us find the optimal @xmath0-pebbling number of a complete graph ( theorem [ pi t(kn ) ] ) .
* definition * : given any distribution of pebbles on the vertices of the graph @xmath4 , we say the vertex @xmath6 is _ odd _ or _ even _ , depending on whether it has an odd or an even number of pebbles on it .
let @xmath0 and @xmath95 be positive integers , and suppose we have a @xmath0-solvable distribution with @xmath96 pebbles on the vertices of @xmath97 . then : 1 .
if the vertex @xmath98 is odd , then every other vertex has at least as many pebbles as @xmath98 .
there are at most two odd vertices .
if there are two odd vertices in @xmath97 , then moving a pebble from one of these vertices to the other creates another @xmath0-solvable distribution .
in particular , some @xmath0-solvable distribution of @xmath96 pebbles on @xmath97 has at most one odd vertex .
[ at most one odd vertex ] * proof * : removing a pebble from an odd vertex @xmath98 does not affect the number of pebbles that may be moved to any other vertex ; thus , every other vertex may still receive @xmath0 pebbles . since there would now be fewer than @xmath96 pebbles ,
@xmath98 could no longer receive @xmath0 pebbles .
if another vertex @xmath99 started with fewer pebbles than @xmath98 , we could use the pebbles now on @xmath98 and @xmath99 to put at least as many pebbles on @xmath98 as on @xmath99 , and any pebbles that could be moved to @xmath99 from other vertices could also be moved to @xmath98 .
thus , we could put at least as many pebbles on @xmath98 as on @xmath99 , contradicting our assertion that @xmath0 pebbles can be moved to @xmath99 , but not to @xmath98 .
therefore , every other vertex has at least as many pebbles as the odd vertex @xmath98 .
if there are two or more odd vertices in @xmath97 , we remove a pebble from each of these vertices and add two pebbles to any vertex , say @xmath100 .
now every vertex can receive at least as many pebbles as it could from the original distribution : if the target originally was odd , the first move would be from @xmath100 to the target .
we therefore have a @xmath0-solvable distribution in which every vertex is even . furthermore , if we originally had three or more odd vertices , this distribution would have fewer pebbles , contradicting the hypothesis that the original distribution had @xmath96 pebbles .
@xmath62 for any positive integers @xmath95 and @xmath0 , let @xmath101 and let @xmath102 .
thus , @xmath103 . then @xmath96 is given by @xmath104 in particular , @xmath105 if and only if @xmath106 .
[ pi t(kn ) ] * proof * : first note that if we put @xmath107 pebbles on one vertex and we put @xmath108 pebbles on every other vertex , then we can move an additional @xmath109 pebbles onto any vertex that starts with @xmath108 pebbles , and we can move @xmath110 additional pebbles onto the vertex that starts with @xmath107 pebbles . in either case , we can move at least @xmath111 pebbles to any target , including the pebbles that start there . thus , @xmath112 .
we now consider whether a @xmath0-solvable distribution in @xmath97 could have fewer than @xmath113 pebbles .
let @xmath98 be the vertex with the fewest pebbles , and suppose it has @xmath114 pebbles .
adding @xmath115 pebbles to @xmath98 costs at least @xmath116 pebbles .
therefore , including the pebbles that started on @xmath98 , the original distribution has at least @xmath117 pebbles . if this is less than @xmath118 , then @xmath119 . if @xmath120 , every vertex has at least @xmath121 pebbles , and so the distribution uses @xmath122 pebbles .
therefore , we assume @xmath123 . now by lemma [ at most one odd vertex ] , we may assume every other vertex has at least @xmath121 pebbles , so we have already accounted for @xmath124 pebbles .
the only way this can be smaller than @xmath113 is if @xmath125 .
now we simply observe that if @xmath126 and @xmath127 pebbles are on @xmath98 and @xmath121 pebbles are on every other vertex , then a total of @xmath128 pebbles can be moved to @xmath98 , and similarly , @xmath129 pebbles can be move to any other vertex .
finally , @xmath105 if and only if @xmath130 and @xmath131 , i. e. if and only if @xmath106 .
@xmath62 the optimal @xmath0-pebbling number is not generally monotone , in the following sense . if it is large for a particular graph , it can be reduced significantly by the addition of a single vertex adjacent to all others . however , for complete graphs the parameter is nondecreasing . for every graph @xmath4 and every positive integer @xmath95
, we have @xmath132 .
[ k(n+1 ) x g ] * proof * : given any distribution @xmath133 , let @xmath134 be the distribution on @xmath135 defined by @xmath136 then any pebbling move from @xmath9 to @xmath11 in @xmath137 can be shadowed by moves from @xmath138 to a distribution that contains @xmath139 in @xmath135 : moves from @xmath140 to @xmath141 or _
vice versa _ may be ignored , other moves from @xmath9 to @xmath11 either from , to , or within @xmath142 can be made from @xmath138 to @xmath139 using @xmath143 instead , and moves from @xmath9 to @xmath11 that do not use @xmath144 can be made unchanged from @xmath138 to @xmath139 .
therefore , if @xmath9 is a @xmath0-solvable distribution on @xmath137 then @xmath138 is a @xmath0-solvable distribution on @xmath135 . since @xmath145 , we have @xmath146 .
@xmath62 corollary [ pi(g x k_m ) < = pi(g x k_n ) ] follows from proposition [ k(n+1 ) x g ] by induction on @xmath95 , starting with @xmath147 as a basis . for every graph @xmath4 and all positive integers
@xmath148 and @xmath95 with @xmath149 , we have @xmath150.@xmath62[pi(g x k_m ) < = pi(g x k_n ) ] * definitions * : given a distribution @xmath151 on @xmath45 and a subset @xmath152 , we define the distribution @xmath153 on @xmath4 by @xmath154 for every @xmath155 . in other words , we count every pebble on a vertex whose coordinate in @xmath42 is in @xmath156 twice and every other pebble once . if the vertices of @xmath42 are @xmath157 , we define @xmath158 by @xmath159 lemmas [ matching sequence from pi(d ) ] and [ useless pebble ] are key to proving theorem [ g x kn ] , which is the upper bound in theorem [ bounds for pi^*(gxkn ) ] . let @xmath156 be any nonempty subset @xmath160 , and suppose there is a sequence of pebbling moves in @xmath45 from @xmath161 to @xmath162 . then there is a sequence of pebbling moves in @xmath4 from @xmath163 to a distribution that contains @xmath164 . in particular , if @xmath165 , then @xmath161 can not be @xmath0-solvable in @xmath57 . [ matching sequence from pi(d ) ]
* proof * : let @xmath166 be the sequence of distributions in @xmath45 after each pebbling move .
we show by induction that we can shadow each pebbling move in @xmath45 with moves in @xmath4 . toward that end , suppose that there is a sequence of pebbling moves in @xmath4 from @xmath163 to a distribution that contains @xmath167 .
the basis @xmath168 is trivial .
suppose going from @xmath83 to @xmath169 requires a move from @xmath170 to @xmath171 .
then the pebbles involved in the move add either four or two pebbles to @xmath172 in @xmath173 , depending on whether @xmath174 , and they add either two pebbles or one pebble to @xmath172 in @xmath175 , depending on whether @xmath176 or not . in either case , @xmath173 contains @xmath177 , and we can simply ignore the extra pebbles . otherwise , going from @xmath83 to @xmath169 requires a move from @xmath178 to @xmath179 .
if @xmath180 the pebbles involved in this move add four pebbles to @xmath181 and two pebbles to @xmath182 in @xmath167 and @xmath183 , and if @xmath184 , they add two pebbles to @xmath181 and one pebble to @xmath182 in @xmath167 and @xmath183 , respectively .
the latter case simply requires a pebbling move from @xmath181 to @xmath182 in @xmath4 to get from @xmath167 to @xmath183 ; the former case requires two such moves . in any of these cases , we can go from @xmath163 to a distribution that contains @xmath167 to one that contains @xmath183 . continuing this process
, we reach a distribution that contains @xmath164 .
now if @xmath165 , there is some vertex @xmath54 such that @xmath185 pebbles can not be moved onto @xmath186 by any sequence of pebbling moves starting from @xmath163 .
therefore , we can not reach any distribution @xmath162 in @xmath45 for which @xmath187 .
in particular , for any @xmath188 , we can not move @xmath0 pebbles onto the vertex @xmath189 .
@xmath62 lemma [ useless pebble ] tells us that if some copy of @xmath4 in @xmath190 starts with a single pebble , then that pebble does not help us reach vertices in any other copy of @xmath4 .
let @xmath191 be a distribution of pebbles on @xmath135 , and suppose there is at most one pebble on some @xmath192 .
let @xmath11 be the distribution on @xmath135 obtained by removing that pebble , or let @xmath193 if there is no such pebble .
let @xmath194 , and let @xmath195 be any configuration of pebbles on @xmath196 that we can reach from @xmath9 .
then we can reach a configuration that contains @xmath195 starting from @xmath11 .
[ useless pebble ] * proof * : if there are no pebbles on @xmath197 and @xmath198 , there is nothing to prove , so we assume there is a pebble on @xmath199 in @xmath9 . paint this pebble gold , and assume it survives every pebbling move in the sequence from @xmath9 to @xmath195 in which it participates .
if the gold pebble never leaves @xmath197 , we can make the same moves in @xmath11 as in @xmath9 and ignore the moves involving the gold pebble .
otherwise , let @xmath99 be the vertex in @xmath97 involved in the first move of the gold pebble from @xmath200 to @xmath201 .
we examine the moves by the gold pebble before it leaves @xmath197 .
note that every such move consumes a nongold pebble that was moved onto @xmath197 from a different copy of @xmath4 .
our approach is to move those pebbles to @xmath202 instead .
thus , from @xmath11 , we ignore all moves from @xmath9 involving the gold pebble before it first leaves @xmath197 .
we replace all other moves to , from , or within @xmath197 with moves to , from , or within @xmath202 , ignoring moves between @xmath197 and @xmath202 .
now the pebble that would have been removed from @xmath200 when the gold pebble moved to @xmath201 reaches @xmath201 in place of the gold pebble .
this pebble can replace of the gold pebble on all subsequent moves .
the result of these changes is that all pebbles that ended up on @xmath196 starting from @xmath9 end up on the same vertices starting from @xmath11 , except that the gold pebble is replaced by a different pebble .
@xmath62 * notation * : suppose we have a distribution of pebbles on @xmath190 .
for each @xmath203 with @xmath204 , we let @xmath114 be the number of pebbles on @xmath197 , and we assume without loss of generality that @xmath205 .
theorem [ g x kn ] gives the upper bound from theorem [ bounds for pi^*(gxkn ) ] .
for any graph @xmath4 and any positive integer @xmath95 , we have @xmath206 .
furthermore , equality holds when @xmath207 .
[ g x kn ] * proof * : we first note that if @xmath9 is a @xmath185-solvable distribution on @xmath4 , then placing @xmath208 pebbles on the vertex @xmath209 for every @xmath54 creates a distribution from which @xmath0 pebbles can be moved to the vertex @xmath210 since we can first move @xmath185 pebbles to @xmath211
. therefore , @xmath212 . now suppose @xmath213 , and let @xmath9 be a distribution on @xmath135 with @xmath214 pebbles or fewer . then either @xmath215 or @xmath216 ; otherwise , we would have @xmath217 and @xmath218 .
but then @xmath219 , contrary to our assumption that @xmath9 has at most @xmath214 pebbles .
if @xmath215 , then @xmath220 has at most @xmath214 pebbles , so @xmath185 pebbles can not be moved onto some @xmath54 starting from @xmath221 .
therefore , by lemma [ matching sequence from pi(d ) ] , we can not move @xmath0 pebbles onto @xmath222 starting from @xmath9 . on the other hand , if @xmath216 , let @xmath11 be the distribution on @xmath190 with the lone pebble on @xmath223 removed .
then @xmath224 , since the pebble on @xmath225 that is counted twice is offset by the pebble that is removed from @xmath223 .
as before , theorem [ matching sequence from pi(d ) ] , shows that @xmath0 pebbles can not be moved to some @xmath222 in @xmath226 starting from @xmath11 .
but now applying lemma [ useless pebble ] with @xmath227 shows that @xmath0 pebbles can not be moved to @xmath222 from @xmath9 in this case either .
therefore , @xmath228 .
@xmath62 applying theorem [ g x kn ] inductively gives corollary
[ g x k_n1 x k_n2 x ... x k_nd ] .
for any graph @xmath4 , any positive integer @xmath0 , and any sequence of integers @xmath229 , we have @xmath230 furthermore , equality holds if @xmath231 for each @xmath232 .
[ g x k_n1 x k_n2 x ... x k_nd ] * proof * : we fix @xmath41 and @xmath0 , and prove by induction on @xmath233 that @xmath234 and that equality holds when each @xmath232 satisfies @xmath235 .
the basis @xmath236 is trivial , so we assume that ( [ g x k_n1 x ... x k_nd equation ] ) holds for some @xmath233 with @xmath237 . applying theorem [ g x kn ] and then applying ( [ g x k_n1 x ... x k_nd equation ] ) gives @xmath238 as desired .
furthermore , equality continues to hold if @xmath239 satisfies @xmath240 .
@xmath62 for all positive integers @xmath0 , and any product of @xmath41 complete graphs , we have @xmath241 if and only if each @xmath242 .
[ kn products ] * proof * : applying corollary [ g x k_n1 x k_n2 x ... x k_nd ] with @xmath4 equal to the trivial graph gives @xmath243 .
furthermore , equality holds when each @xmath232 satisfies @xmath244 , or equivalently , @xmath242 . on the other hand , if @xmath245 for some @xmath203 , we assume without loss of generality that @xmath246 . now applying corollary [ g x k_n1 x k_n2 x ...
x k_nd ] with @xmath247 gives @xmath248 , and by proposition [ pi t(kn ) ] , @xmath249 when @xmath246 .
@xmath62 we can now prove theorem [ bounds for pi^*(gxkn ) ] .
* proof of theorem [ bounds for pi^*(gxkn ) ] * : the upper bound is given by theorem [ g x kn ] . to establish the lower bound ,
suppose we have a @xmath0-solvable distribution @xmath9 of @xmath250 pebbles on @xmath135 .
since @xmath205 , we have @xmath251 . now by lemma [ matching sequence from pi(d ) ] , @xmath252 . since @xmath253 must be an integer , we have @xmath254 . @xmath62 for the smallest of products , we are able to get exact results for all @xmath0 .
these exhibit a nice pattern that we will say more about subsequently .
first we present an obvious proposition . for any graph @xmath4 and any positive integers @xmath51 and @xmath0 , we have @xmath255 .
similarly , for regular pebbling , we have @xmath256 . [ pi^*_s+t ] * proof * : we can place @xmath257 red pebbles and @xmath19 blue pebbles on @xmath4 in such a way that @xmath51 red pebbles and @xmath0 blue pebbles can be moved to any target vertex . for regular pebbling
, we note that from any placement of @xmath258 pebbles , if we arbitrarily paint @xmath259 pebbles red and @xmath17 pebbles blue , then @xmath51 red pebbles and @xmath0 blue pebbles can be moved to any target vertex .
@xmath62 to find the optimal @xmath0-pebbling number of @xmath260 , let @xmath261 and @xmath262 . then @xmath263 in each case except @xmath264 , the lower bound from theorem [ bounds for pi^*(gxkn ) ] is tight .
[ pi t(k2xk2 ) ] * proof * : if @xmath21 , we note that two pebbles are not enough to reach every vertex : if we put them on different vertices , the unoccupied vertices can not be reached , and if we put them on the same vertex , the antipodal vertex is unreachable . on the other hand ,
three pebbles are sufficient , since we can put two pebbles on @xmath265 and one on @xmath266 . for @xmath267 , we consider table [ k2 x k2 table ] : @xmath268 the second row of this table gives the lower bound for @xmath269 from theorem [ bounds for pi^*(gxkn ) ] , and the last row gives a solvable distribution with the given number of pebbles .
therefore , the bound is tight . finally , for @xmath270 , we assume by induction on @xmath0 that the lower bound is tight for @xmath271 , and we show that @xmath272 . comparing the computation of the lower bound for @xmath273 to that of @xmath274
, we have @xmath275 , so @xmath276 , and @xmath277 .
thus , @xmath278 , and the lower bound from theorem [ bounds for pi^*(gxkn ) ] gives @xmath279 . on the other hand , proposition [ pi^*_s+t ] tells us that @xmath280 .
therefore , @xmath281 , as required .
@xmath62 we can compute @xmath282 similarly . the optimal @xmath0-pebbling number of @xmath283 is @xmath284 in particular , if @xmath285 and @xmath286 , then @xmath287 [ pi t(k2xk3 ) ] * proof * : for @xmath288 , we use table [ k2 x k3 table ] .
@xmath289 for larger @xmath0 , we note from proposition [ pi^*_s+t ] that @xmath290 , which agrees with the asserted lower bound .
@xmath62 corollary [ kn products ] shows that for small values of @xmath0 , the upper bound in theorem [ bounds for pi^*(gxkn ) ] is tight for products of complete graphs .
it was obtained by applying theorem [ g x kn ] inductively , with @xmath4 being the trivial graph .
if we apply the lower bound in theorem [ bounds for pi^*(gxkn ) ] inductively with @xmath4 being the trivial graph , we get a lower bound on the optimal @xmath0-pebbling number of a product of complete graphs .
theorem [ asymptotic bound for complete products ] shows that this lower bound is asymptotically tight as @xmath0 gets large .
we begin with lemma [ 2^d pebbles on each vertex ] [ 2^d pebbles on each vertex ] let @xmath291 be a sequence of nonnegative integers , and let @xmath292 .
then for any integer @xmath233 , putting @xmath293 pebbles on each vertex of @xmath294 creates a @xmath295-solvable configuration .
thus , @xmath296 . * proof * :
if @xmath297 the products are all empty , so @xmath298 and @xmath4 is the trivial graph . clearly , putting @xmath233 pebbles on the lone vertex gives an optimal @xmath233-solvable configuration , as required . for larger @xmath41
, we first show the specified configuration is @xmath295-solvable . toward that end ,
let @xmath299 be the target vertex in @xmath300 .
if we have @xmath293 pebbles on each vertex , then for each @xmath301 , we have @xmath302 pebbles on each vertex of @xmath303 .
therefore , by induction on @xmath41 , we assume that we can put @xmath304 pebbles on @xmath305 .
but now we have @xmath306 pebbles on @xmath307 and we can move an additional @xmath308 pebbles from @xmath309 to @xmath307 for every vertex @xmath310 .
thus , we can move a total of @xmath311 pebbles onto @xmath307 , as required , and so @xmath312 .
conversely , we know from theorem [ bounds for pi^*(gxkn ) ] that @xmath313 now @xmath314 , and we may assume by induction on @xmath41 that @xmath315 multiplying this number by @xmath316 gives an integer , so taking the ceiling is irrelevant .
therefore , @xmath317 which agrees with our upper bound .
therefore , @xmath318.@xmath62 [ asymptotic bound for complete products ] let @xmath291 be a sequence of nonnegative integers .
then @xmath319 * proof * : let @xmath320 , let @xmath321 , and let @xmath322 be given by @xmath323 for any @xmath324 , we can let @xmath325 and @xmath326 . then applying proposition [ pi^*_s+t ] , we have @xmath327 from lemma [ 2^d pebbles on each vertex ] , we know that @xmath328 , and from the definition of @xmath322 , we have @xmath329 thus , @xmath330 , and so @xmath331 , as desired.@xmath62
_ fractional distributions _ and _ fractional pebbling moves _ were defined in @xcite .
these are continuous analogs of pebbling concepts .
moews @xcite previously called them _ continuous distributions _ , and _ continuous pebbling moves _ , and he defined the _ continuous optimal pebbling number _ of a graph .
we give these definitions now .
* definitions @xcite * : a _ fractional distribution _ on @xmath4 is a function @xmath332 .
again , a distribution represents a placement of pebbles on the vertices of @xmath4 , though we now allow a nonintegral number of pebbles . a _ fractional pebbling move _ consists of removing @xmath333 pebbles from one vertex and adding @xmath233 pebbles to an adjacent vertex . as in an integer - valued distribution , the _ size _ of @xmath9
is given by @xmath334 , and @xmath9 is _ fractionally solvable _ , ( or simply _ solvable _ if there is no ambiguity ) , in case for every vertex @xmath6 , it is possible to reach @xmath6 with one pebble through some sequence of fractional pebbling moves , starting from @xmath9 .
moews @xcite defined the _ continuous optimal pebbling number _ of a graph , which we denote @xmath335 .
the _ optimal fractional pebbling number _ of the graph @xmath4 , which we denote @xmath336 , was defined in @xcite .
we give these definitions now .
* definitions @xcite * : the _ continuous optimal pebbling number _ of a graph @xmath4 , is the smallest number @xmath335 such that some fractional distribution @xmath9 with @xmath337 is solvable using fractional pebbling moves
. the _ optimal fractional pebbling number _
@xmath336 is given by @xmath338 theorem [ optimal fractional theorem ] was shown in @xcite .
[ optimal fractional theorem ] every graph @xmath4 satisfies @xmath339 .
furthermore , @xmath336 is rational for any graph @xmath4 and every graph has an optimal fractional distribution in which the number of pebbles on each vertex is rational.@xmath62 moews @xcite proved theorem [ continuous optimal graham ] , and used it to give a nonconstructive proof of theorem [ optimal products ] , which relates the optimal pebbling number of @xmath39 to the continuous optimal pebbling number of @xmath4 .
[ continuous optimal graham ] for all graphs @xmath4 and @xmath42 , we have @xmath340 . [ optimal products ] for all graphs @xmath4 , we have @xmath341 for some constants @xmath342 and @xmath233 . theorem [ optimal fractional ] generalizes some of our results from section [ kn products section ] .
we begin with lemma [ optimal fractional lemma ] .
[ optimal fractional lemma ] for every graph @xmath4 we have @xmath343 for all @xmath0 . * proof * : suppose by contradiction that there is some @xmath0 such that @xmath344 .
let @xmath9 be a @xmath0-solvable distribution on @xmath4 with @xmath345 .
then the fractional distribution @xmath346 given by @xmath347 for all @xmath6 is fractionally solvable .
so , @xmath348 , contradicting theorem [ optimal fractional theorem ] .
@xmath62 + the definition of @xmath336 implies that @xmath349 .
here we tighten the lower order term . for every graph @xmath4 we have @xmath350 .
[ optimal fractional ] * proof * : let @xmath351 and let @xmath352 and @xmath353 be integers satisfying @xmath354 . from theorem [ optimal fractional theorem ] , there is some fractionally solvable fractional distribution @xmath346 on @xmath4 with @xmath355 such that @xmath356 for some integers @xmath357 and @xmath358 . under @xmath346 ,
if @xmath359 , then @xmath360 pebbles could be sent from @xmath98 to @xmath99 by making fractional pebbling moves toward @xmath99 . since @xmath346 is fractionally solvable , every vertex @xmath6 satisfies @xmath361 let @xmath362 and let @xmath363 , where @xmath364 . given an integer @xmath0 , the division algorithm produces integers @xmath365 and @xmath366 such that @xmath367 and @xmath368 . consider the distribution @xmath9 on @xmath4 given by @xmath369 for all @xmath203 . under @xmath9 , we have @xmath370 for all @xmath203 . since @xmath371 is a multiple of @xmath358
, @xmath372 is a multiple of @xmath373 .
so , under the distribution @xmath11 , it is possible to send @xmath374 pebbles from @xmath98 to @xmath99 .
so , starting from @xmath11 , the number of pebbles that can be sent to a root @xmath6 is given by @xmath375 thus , @xmath11 is @xmath376-solvable on @xmath4 , meaning @xmath9 is @xmath0-solvable on @xmath4 .
since @xmath95 and @xmath233 are constants , we have @xmath377 in connection with lemma [ optimal fractional lemma ] , this gives us the desired result .
@xmath62 + we note that @xmath378 , @xmath379 , and @xmath380 .
thus , these specific cases of theorem [ optimal fractional ] are witnessed by theorem [ pi t(kn ) ] and propositions [ pi t(k2xk2 ) ] and [ pi t(k2xk3 ) ] , respectively .
if we apply theorem [ optimal products ] to @xmath1 , we find @xmath381 for some constant @xmath233 , since @xmath382 .
however , moews s proof of theorem [ optimal products ] was nonconstructive
. it does not give distributions for small values of @xmath41 , and it gives no information for small values of @xmath41 .
we give distributions that show that @xmath383 .
we let the vertices of @xmath1 be @xmath384 .
we begin by finding @xmath0-solvable distributions @xmath385 on @xmath386 for @xmath21 , @xmath387 , and @xmath388 .
* notation * : we denote by @xmath389 the distribution with four pebbles on @xmath265 and two pebbles each on @xmath390 and @xmath391 , by @xmath392 the distribution with four pebbles each on @xmath265 , @xmath390 , and @xmath391 , and by @xmath393 the distribution with 4 pebbles on each @xmath394 , @xmath395 .
we write @xmath396 for the @xmath397 , i. e. the distribution with one pebbles on each @xmath398 .
@xmath396 is shown in figure [ b ] ( filled - in vertices are occupied , dark edges give the neighborhoods of the occupied vertices , and the edges wrap around in the obvious ways ) .
[ t - solvable on c5xc5 , t=1,2,4 ] for each @xmath399 , the distribution @xmath385 is @xmath0-solvable on @xmath386 .
* proof * : first note that @xmath400 .
we call these vertices the _ corners _ of the graph , imagining @xmath265 to be the center .
if we have two pebbles each on @xmath390 and @xmath391 , we have the @xmath401-solvable distribution in proposition [ pi t(k2xk2 ) ] , so two pebbles can be moved any of the corners , and one pebble can be moved to any vertex adjacent to these corners .
the rest of the vertices are within two steps from @xmath265 , so they can be reached from the four pebbles from there .
this takes care of the @xmath21 case . for @xmath387
, we instead have four pebbles each on @xmath390 and @xmath391 , so we can consider these to be two groups which each have two pebbles on both @xmath390 and @xmath391 .
therefore , we can put four pebbles on any corner .
then the vertices whose distance from @xmath265 is zero or one can receive two pebbles from that vertex .
the vertices whose distance from @xmath265 is three or four can receive two pebbles from the nearest corner , and those whose distance from @xmath265 is two can receive one pebble from @xmath265 and one from the nearest corner .
when @xmath388 , the symmetry of @xmath393 allows us to consider only one target , say @xmath402 .
this vertex can receive two pebbles from @xmath265 , and one each from @xmath403 and @xmath404 .
@xmath62 ( 0,0 ) on @xmath386 ] [ b dot d ] let @xmath4 be any graph and let @xmath9 be a @xmath0-solvable distribution on @xmath4 in which the number of pebbles on every vertex is a multiple of four .
then the distribution @xmath406 is a @xmath0-solvable distribution in @xmath407 in which the number of pebbles on every vertex is a multiple of four .
note that the number of pebbles in @xmath408 is @xmath409 .
in particular , by induction on @xmath148 , we have @xmath410 .
* proof * : let the target vertex in @xmath411 be @xmath412 . since @xmath208 is a multiple of four , we write @xmath413 for the distribution with @xmath414 pebbles on @xmath54 , and we write @xmath415 for the set @xmath416 .
now @xmath417 . by proposition [ t - solvable on c5xc5 , t=1,2,4 ] , @xmath418 is @xmath419-solvable
, so by theorem [ optimal pebbling distributions ] , @xmath408 is @xmath420-solvable in @xmath411 .
that is , from @xmath406 we can reach the distribution in which @xmath421 pebbles are on @xmath422 for every @xmath423 .
but now the distribution on the vertices in @xmath424 is @xmath9 .
since @xmath9 is @xmath0-solvable , we can put @xmath0 pebbles on @xmath425 .
clearly , @xmath426 is a multiple of four , since @xmath427 is a multiple of four.@xmath62 [ odd product of c5 ] for all integers @xmath428 , we have @xmath429 . * proof *
: we apply theorem [ b dot d ] to the distribution with four pebbles on a single vertex of @xmath430 .
@xmath62 a natural question at this point is what bounds we can get for @xmath431 .
we create a solvable distribution @xmath432 on @xmath433 , and we use @xmath432 to start an induction with theorem [ b dot d ] for even products similar to the argument for corollary [ odd product of c5 ] . * notation * : let @xmath432 be the distribution of @xmath434 pebbles on @xmath433 given by @xmath435 note that if we denote the empty distribution by @xmath436 , then we may more simply write @xmath437 where @xmath438 .
[ f is 4-solvable ] every occupied vertex in @xmath432 has four pebbles , and @xmath432 is solvable in @xmath433 . *
proof * : every occupied vertex of both @xmath396 and @xmath392 has four pebbles in @xmath386 , so this holds for @xmath432 in @xmath433 as well .
to show @xmath432 is solvable , let the target vertex in @xmath433 be @xmath439 . by construction ,
the distribution of pebbles on @xmath440 is @xmath396 .
therefore , by proposition [ t - solvable on c5xc5 , t=1,2,4 ] , four pebbles can be moved to @xmath441 using only the pebbles on @xmath442 .
similarly , and simultaneously , by proposition [ t - solvable on c5xc5 , t=1,2,4 ] , two pebbles can be moved to both @xmath443 and @xmath444 .
at this point , the distribution of pebbles on @xmath445 is @xmath389 , so one pebble may be moved to @xmath446 , again by proposition [ t - solvable on c5xc5 , t=1,2,4].@xmath62 the use of theorem [ b dot d ] on the distribution @xmath389 would give a better coefficient of @xmath447 in the theorem [ all products of c5 ] ; however , @xmath389 does not qualify since some vertices get only 2 pebbles .
[ all products of c5 ] we have @xmath448 . for all @xmath449 we have @xmath450 . * proof * : if @xmath451 , proposition [ t - solvable on c5xc5 , t=1,2,4 ] shows that @xmath452 . if @xmath453 , proposition [ f is 4-solvable ] shows that @xmath454 . for @xmath455 with @xmath456 , theorem [ b dot d ] implies that @xmath457 . since @xmath458 ,
the second part follows for even @xmath41 . if @xmath459 , corollary [ odd product of c5 ] gives us @xmath460 .
@xmath62 we can generalize the construction of @xmath432 and the proof of proposition [ f is 4-solvable ] to obtain theorem [ using a solvable distribution ] . [ using a solvable distribution ]
let @xmath24 be a set of distributions on @xmath4 , suppose @xmath9 is an @xmath24-solvable distribution , and suppose @xmath461 is a family of distributions on @xmath42 such that each @xmath462 is @xmath366-solvable .
let @xmath463 be the distribution on @xmath45 defined by @xmath464 then @xmath79 is @xmath465-solvable in @xmath45 .
that is , for any distribution @xmath466 and any vertex @xmath467 , a copy of @xmath468 can be moved to the vertices of @xmath469 . * proof * : let @xmath470 be the chosen vertex in @xmath42 . then for any @xmath471 , restricting @xmath79 to the vertices @xmath472 gives the distribution @xmath473 in @xmath472 .
since @xmath473 is @xmath5-solvable in @xmath42 , we can move @xmath5 pebbles to @xmath474 for each @xmath7 . after these moves
, the distribution of pebbles on @xmath475 is @xmath9 .
since @xmath9 is @xmath24-solvable , we can put a copy of any @xmath476 on @xmath477 , as desired.@xmath62 we note that the proofs of corollaries [ optimal st - pebbling graphs ] and [ optimal pebbling graphs ] essentially involved letting each @xmath478 , where @xmath11 is the distribution of pebbles on @xmath42 defined in the proof of theorem [ optimal pebbling distributions ] .
corollary [ optimally using solvable distribution ] is a stronger result .
[ optimally using solvable distribution ] let @xmath24 be a set of distributions on @xmath4 , and suppose @xmath9 is an @xmath24-solvable distribution .
then for any graph @xmath42 , we have @xmath479 * proof * : we simply apply theorem [ using a solvable distribution ] and use a family of distributions @xmath480 in which each @xmath481 is optimal , i. e. @xmath482.@xmath62
in this section , we give optimal pebbling distributions on the @xmath41-dimensional hypercube @xmath483 .
we consider the vertices of @xmath484 to be all bitstrings of length @xmath41 , or equivalently , all vectors in the @xmath41-dimensional vector space @xmath485 over the two - element field @xmath486 .
there is an edge between two vertices when the hamming distance between the corresponding bitstrings is @xmath487 . given two bitstrings @xmath488 and @xmath489 , we write @xmath490 for the bitstring in @xmath491 obtained by concatenating the bits in @xmath492 and @xmath493 .
we also write @xmath494 and @xmath495 for the bitstrings @xmath496 and @xmath497 , respectively , and we call the number of @xmath487 s in a bitstring its _ weight_. theorem [ moews bound ] gives the best known bound for hypercubes , but it does not give explicit distributions , which is our aim .
the @xmath501 root of moews s result tends to about @xmath502 , and proposition [ antipodal solvable ] gives an example , the @xmath501 root of whose size is roughly @xmath503 .
our new construction in theorem [ weighted geometric mean ] improves that number below @xmath504 .
proposition [ antipodal solvable ] gives a solvable distribution on @xmath484 for all @xmath41 .
these were first given in pachter , snevily , and voxman @xcite .
[ antipodal solvable ] if @xmath505 , the distribution on @xmath484 obtained by putting @xmath506 pebbles on @xmath507 and @xmath508 pebbles on @xmath509 is solvable .
if @xmath510 , the distribution on @xmath484 given by putting @xmath506 pebbles on both @xmath507 and @xmath509 is solvable .
thus , the optimal pebbling number of a hypercube satisfies @xmath511 in particular , @xmath512 .
* proof * : in both cases whether @xmath41 is even or odd , every vertex whose weight is at most @xmath233 can receive at least one pebble from the pebbles on @xmath494 in the given distribution , and every vertex with larger weight can receive a pebble from those on @xmath495.@xmath62 we give a construction for extending the distributions in proposition [ antipodal solvable ] to distributions on larger cubes with better asymptotic bounds than those in the proposition .
this construction is based on an argument similar to the proof of theorem [ using a solvable distribution ] using distributions on @xmath498 obtained from theorem [ pi t(kn ) ] .
first recall the distributions on @xmath498 from theorem [ pi t(kn ) ] ; we will use these in theorem [ splitting construction ] . in the spirit of theorem
[ using a solvable distribution ] , we want to extend a solvable distribution @xmath9 on a graph @xmath4 to a distribution @xmath11 on @xmath522 .
we hope that @xmath523 .
unfortunately , the extra @xmath524 in proposition [ dr r - solvable ] can cause problems . for example , if @xmath9 has a single pebble on a large number of vertices , those pebbles each give rise to two pebbles in @xmath11 .
we can get an extra @xmath524 for each occupied vertex in @xmath9 .
we define the _ support _ of @xmath9 to keep track of this information . [ splitting construction ]
let @xmath9 be a @xmath0-solvable distribution on @xmath484 .
for each @xmath527 , define @xmath528 and @xmath529 by @xmath530 then @xmath11 is @xmath0-solvable on @xmath531 .
furthermore , the number of pebbles in @xmath11 is at most latexmath:[$\frac{4}{3 } |d| + \frac{2}{3 } * proof * : let the target in @xmath531 be @xmath533 , where @xmath534 and @xmath535 . for each @xmath536 ,
the distribution of pebbles on @xmath537 is @xmath538 .
since this distribution is @xmath539-solvable in @xmath498 , we can put @xmath539 pebbles on @xmath540 . if we do this for each @xmath536 , the distribution of pebbles on @xmath541 is @xmath9 .
since @xmath9 is @xmath0-solvable on @xmath484 , we can put @xmath0 pebbles on @xmath533 .
the total number of pebbles in @xmath11 is @xmath542 finally , @xmath543.@xmath62 [ many splits ] let @xmath9 be a solvable distribution on a graph @xmath4 with @xmath544 @xmath148 times to @xmath9 . then @xmath545 is a solvable distribution on @xmath546 such that @xmath547 , and @xmath548 . * proof * : there is nothing to show if @xmath549 , so we suppose by induction that for some @xmath550 , @xmath83 is a solvable distribution on @xmath551 with @xmath552 , and latexmath:[$|\sigma ( d_i)| \leq 2^i @xmath83 , we find that @xmath169 is a solvable distribution on @xmath554 with @xmath555 .
furthermore , we have @xmath556 + \frac{2}{3 } ( 2^i s).\ ] ] multiplying through by the @xmath557 and noting that @xmath558 , we have @xmath559 completing the induction.@xmath62 [ splitting antipodal ] let @xmath161 be the distribution with @xmath506 pebbles on both @xmath560 and @xmath561 in @xmath562 , and let @xmath545 be the resulting distribution in @xmath563 obtained by applying theorem [ splitting construction ] @xmath148 times .
then @xmath564 for large @xmath148 , the term @xmath567 in corollary [ splitting antipodal ] is small compared to @xmath568 . by controlling the relationship between @xmath233 and @xmath148 , we can ensure that the first two terms are roughly equal . using logarithms to solve the equation @xmath569 , or @xmath570 together with the observation that @xmath571 .
we obtain the constants in theorem [ weighted geometric mean ] . [
weighted geometric mean ] given an integer @xmath41 , let @xmath572 , and let @xmath573 .
then the distribution @xmath545 on @xmath574 from corollary [ splitting antipodal ] with these values of @xmath233 and @xmath148 satisfies @xmath575 .
* proof * : we define @xmath576 and @xmath577 by @xmath578 and @xmath579 .
we note that @xmath580 since @xmath581 , we have @xmath582 , and since @xmath583 , this implies @xmath584 . furthermore , @xmath585 ; therefore , @xmath586 , or equivalently , @xmath587 . in particular , @xmath588 .
thus , @xmath589 . from corollary
[ splitting antipodal ] , this implies that @xmath590.@xmath62 | we prove a generalization of graham s conjecture for optimal pebbling with arbitrary sets of target distributions .
we provide bounds on optimal pebbling numbers of products of complete graphs and explicitly find optimal @xmath0-pebbling numbers for specific such products .
we obtain bounds on optimal pebbling numbers of powers of the cycle @xmath1 . finally , we present explicit distributions which provide asymptotic bounds on optimal pebbling numbers of hypercubes
. * keywords . * pebbling distribution , pebbling number , fractional pebbling * 2000 msc . * 05c99 |
most metal - poor stars have relatively high abundances of the so - called `` @xmath2-elements '' ( o , ne , mg , si , s , ar , ca , and ti ) which is to say that the number abundance ratios o / fe , ne / fe , etc .
, are higher than one finds in the sun ( see , e.g. , the reviews by @xcite , @xcite , and @xcite ) . for instance , @xcite concluded from his review of the available observational data that globular clusters ( gcs ) with [ fe / h ] @xmath11 are overabundant in the @xmath2-elements by a factor of about two ( i.e. , [ @xmath2/fe ] @xmath12 in the standard logarithmic notation ) .
since then , some gcs have been discovered to have solar proportions of these elements ( see @xcite ) .
moreover , recent field star studies ( e.g. , @xcite ) have revealed a significant number of halo stars with [ @xmath2/fe ] @xmath13 , though the majority of such stars show the same level of enhancement as is typically found in gcs .
given that the abundances of the individual @xmath2-elements appear to scale with the overall @xmath2-element abundance,/fe ] value generally represents the abundance ratios determined for the _ individual _
@xmath2-elements to within 0.1 dex .
oxygen may turn out to be an exception to this rule , but current estimates of the abundance of this element have large error bars associated with them because they depend sensitively on which spectral features are analyzed : much higher o / fe ratios are found from the ultraviolet oh bands ( @xcite , @xcite ) than from the @xmath14 nm [ o i ] lines ( @xcite ) . in this investigation
we assume that all of the @xmath2-elements , including oxygen , vary together .
the possibility that [ o / fe ] @xmath15 in extremely metal - deficient stars , and the consequences of such high oxygen abundances for synthetic c - m diagrams and for turnoff luminosity versus age relations , is considered by @xcite . ] the chemical composition of a globular cluster or population ii star is often described in terms of the two quantities [ fe / h ] and [ @xmath2/fe ] .
it goes without saying that the models which are used to interpret stellar data should be computed for the run of chemical abundances that is actually observed and given that there are star - to - star differences in the abundances of the @xmath2-elements at a fixed iron content , it is important to treat such variations explicitly .
accordingly , a new grid of stellar evolutionary tracks has been computed by @xcite for 17 [ fe / h ] values between @xmath16 and @xmath1 assuming , in each case , [ @xmath2/fe ] @xmath17 , 0.3 , and 0.6 .
these calculations have also taken into account the most important non - ideal effects in the equation of state as well as recent improvements to opacity data and nuclear reaction rates ( though gravitational settling and radiative acceleration processes are not treated ) . moreover , as fully discussed in paper i , the properties of local sudwarfs , the loci for gc red giants on the ( @xmath18)plane as inferred from infrared photometry , and the luminosities of rr lyrae variables as deduced from baade - wesselink , statistical parallax , and trignometric parallax studies are reproduced quite well by the models .
reference should be made to that study for a full description of how variations in [ @xmath2/fe ] affect the tracks and zero - age horizontal - branches , both at a fixed [ fe / h ] and as a function of [ fe / h ] . in paper ii , @xcite examined the implications of the new model grids for the ages of gcs and of field halo stars .
the main results of that study are ( 1 ) the most metal - deficient clusters _ and _ field subgiants have ages @xmath19 gyr , ( 2 ) ages decrease with increasing [ fe / h ] ( by 23 gyr between [ fe / h ] @xmath20 and @xmath21 ) ( 3 ) the dispersion in age at a fixed metallicity appears to be small ( especially at [ fe / h ] @xmath22 ) , and ( 4 ) there is no more than a weak variation of age with galactocentric distance , if that .
the photometric data for many of the clusters are still not as good as they need to be , relative age estimates based on the horizontal method , as described by @xcite , sometimes seem to be inconsistent with those derived from the @xmath23 technique ( suggesting that something besides , or in addition to , age is varying ) , and the uncertainties associated with most reddening , metallicity , and distance determinations remain large .
consequently , the conclusions that were reached must still be considered tentative even though they seem well - supported by the observations _ as they presently exist_. ( many age - related issues , such as how best to intercompare synthetic and observed c - m diagrams for the determination of both absolute and relative ages , are discussed in paper ii . ) the primary purpose of the present investigation is to make the @xmath2-element enhanced isochrones and what we have called `` isochrone population functions '' ( ipfs ) readily available to the astronomical community .
an ipf is derived from the slope of the predicted mass_distance _ relation in @xmath24/mag , where the distance along an isochrone , on any color magnitude plane for which good color@xmath25 relations are available , is defined with respect to some arbitrary , well - defined point . in conjunction with an assumed mass spectrum ,
ipfs thus provide the means to generate synthetic stellar populations in a very straightforward way or to calculate luminosity functions ( lfs ) or color functions ( cfs ) .
the value and importance of lfs have long been appreciated ( see , e.g. , @xcite ) , but it is only recently that @xcite have pointed out that the distibution in color of stars along the subgiant branch has the potential to provide an age constraint that is not only completely distance - independent , but is also nearly metallicity - independent . to obtain meaningful results from observed cfs , superb photometry and very large subgiant samples are needed , but it should be possible to meet these requirements in the case of the nearest and most massive of the galaxy s gcs .
certainly , all aspects of observed c - m diagrams should be fully investigated in order to achieve the best possible understanding of stellar systems . in the following section , an improved algorithm for interpolating in evolutionary sequences for isochrones and ipfs
is described : this is based in large part on the scheme introduced by @xcite .
section 3 outlines how to execute the software that has been developed to obtain isochrones , ipfs , lfs , and cfs , and provides a fairly detailed description of the output of these codes . in
4 , some additional discussion beyond that contained in papers i and ii is presented on how well the present computations appear to satisfy observational constraints .
brief concluding remarks are given in 5 .
the mathematical formalism for the derivation of accurate ipfs parallels that for the derivation of luminosity functions presented in bv92 , except that the distance along an isochrone is used as the observable coordinate instead of luminosity . in the @xmath26@xmath27 plane ,
the distance along an isochrone is defined by @xmath28^{1/2 } , \eqno(1)\ ] ] where @xmath29 and @xmath30 are chosen to stretch the sgb region and to obtain a convenient range of distances .
( we have chosen @xmath31 and @xmath32 , and we arbitrarily define @xmath33 at the lowest mass point on the isochrone . )
the slope of the mass distance relation is then obtained from@xmath34^{1/2}.\eqno(2)\ ] ] consequently , equation ( 6 ) from bv92 is recast as @xmath35 where @xmath36 is the number of stars with mass @xmath37 , and @xmath38 is the differential isochrone population function .
( note that the second of these two equations becomes the familiar definition of a differential luminosity function if @xmath39 is replaced by @xmath40 see bv92 . )
the evaluation of the partial derivatives in equation ( 2 ) is performed as described in bv92 . in the observer s plane ,
the distance is calculated with respect to a well - defined , but arbitrary , point on the isochrone . for this paper ,
a point on the subgiant branch ( sgb ) 0.05 mag redward of the main - sequence turnoff ( regardless of color index ) , has been adopted as the distance zero - point with the idea that it should be relatively easy to identify in observed cmds .
a convenient definition of distance is @xmath41^{1/2 } , \eqno(4)\ ] ] where @xmath42 is the absolute magnitude in the @xmath43 passband and @xmath44 is the color index for passbands @xmath45 and @xmath46 .
however , the transformation of equation ( 3 ) from the theoretical plane to the observer s plane involves the derivatives of both the color transformations and the bolometric corrections .
rather than attempt to evaluate these numerically , it is easier to construct a @xmath47@xmath48 calibration which can be used to translate bin limits from the observer s plane to the theoretical plane , and then to perform integrations across bins on the theoretical plane .
isochrones , ipfs , lfs , and cfs are interpolated from grids of evolutionary tracks for stars of the same initial chemical composition but different masses . consequently
, the most fundamental interpolation is the one in which the mass is derived from the age
mass relation for a given evolutionary phase .
the obvious advantage of an interpolation scheme in which mass is a monotonic function of age is that the mass is uniquely determined once the age and evolutionary phase are specified .
the corresponding luminosities , effective temperatures , and their temporal derivatives then can be interpolated with respect to mass .
the basic approach to setting up the interpolation scheme remains as it was described in bv92 , with up to seven primary equivalent evolutionary phase " ( eep ) points ( see , in particular , their fig .
2 ) located on each track between the zero - age main sequence and the tip of the rgb .
the definitions of two of these points have been revised slightly .
the first primary eep is now taken to be the point at which the central hydrogen content falls to @xmath49 , where @xmath50 is the adopted h abundance in the initial fully convective model on the hayashi line .
this corresponds quite well with the zero - age main - sequence location of the model on the h - r diagram .
the more noteworthy change resulted from the discovery that the age
mass relation defined by the second primary eeps can be non - monotonic just where the blue hook begins to manifest itself ( at @xmath51 ) , if the amount of convective overshoot is assumed to increase with increasing mass .
( depending on whether the star has a radiative or convective core during the main - sequence phase , the second primary eep corresponds to either the turnoff point or the minimum @xmath25 that is reached just prior to the blue hook , respectively . )
it is however , relatively easy to avoid this difficulty .
as shown in figure [ f1 ] , the competition between the @xmath52-@xmath52 chain and the cno cycle produces a small dip in the derivative @xmath53 for the lower mass tracks that develops into the blue hook at higher mass . when this feature appears _ after _ the turnoff point , it is used to define the second primary eep .
track in fig . 1 ) .
perturbations to the model arising from the procedures used to optimize the mesh point distribution or to calculate the atmospheric boundary conditions are the most likely causes of this behavior . however , the amplitude of such variations is clearly very small and of little consequence for computed ipfs . ] over the range 0.51.8 @xmath54 in stellar mass and for metallicities @xmath55}\le + 0.12 $ ] that we have investigated , the age
mass relations for the primary eeps remain monotonic
. will be the subject of a forthcoming paper by d.a .
vandenberg , p.a .
bergbusch , and p.d .
dowler ( in preparation ) . ]
thus , it is possible to maintain the monotonic behaviour in the secondary eep relations that are obtained by dividing each evolutionary phase interval , bracketed by a pair of primary eeps , into an equal number of regularly placed sub - intervals . for the evolutionary tracks of low - mass stars which achieve ages of 30 gyr before reaching the turnoff point ,
the placement of the last secondary eep is determined by matching it with the secondary eep having the same central hydrogen content on the track of the next higher mass . in bv92 , linear interpolation was found to be acceptable for all of the eep relations , except for the evaluation of derivatives along the direction of interpolation . in the absence of a better solution at the time
, these derivatives were evaluated from a cubic polynomial fitted through the eep points in other words , the derivatives were obtained from smoothed versions of the eep relations .
however , for this paper , the interpolations and evaluations of derivatives were performed with the akima spline ( akima , 1970 ) , for complete internal consistency . among the various spline routines tested ,
the akima spline was found to be the best for tracing the subtle morphology of the eep relations , as it is particularly good at avoiding the typical spline overshooting that occurs when the behaviour of the spline points is non - polynomic .
in addition , as one may infer from fig . 6 of bv92 , linear interpolation tends to produce the onset of the convective hook in the isochrones at ages slightly greater than those predicted by the evolutionary sequences .
akima spline interpolation reduces this discrepancy over the entire range of metallicities that we have investigated .
the differences between isochrones derived via akima spline interpolation and those derived using linear interpolation can be readily appreciated by comparing the eep relations that give the turnoff effective temperatures and luminosities as a function of mass .
the @xmath3 eep relations , plotted in upper panel of figure [ f2 ] , reveal small , but detectable , differences between linear interpolation and akima spline interpolation at the turnoff , even for low mass stars .
moreover , the largest differences always occur when there is an abrupt change in the behaviour of the eep relation .
for example , in the [ fe / h]@xmath56 , [ @xmath2/fe]@xmath17 grid , isochrones with ages 8 and 12 gyr , interpolated by either method , are virtually identical at the turnoff because they have turnoff masses near 0.9 and 0.8 solar masses , respectively . however , the turnoff masses for ages 10 , 14 , and 16 gyr , lie near @xmath57 , @xmath58 , and @xmath59 , respectively ; i.e. , approximately midway between the computed tracks .
the @xmath3 eep relations shows a distinct change in slope at @xmath60 , around which the spline deviates noticeably from the linear relations . as can be seen from the lower panel of fig . 2 , the difference in @xmath27 for a mass of @xmath58 is @xmath61 .
this means that the temperature predicted by the akima spline interpolation , for the turnoff of a 14 gyr isochrone , is @xmath62 cooler than that derived by linear interpolation .
similar observations can be made about the luminosity eep relations , plotted in the upper panel of figure [ f3 ] .
the differences in @xmath63 between the two interpolation methods , plotted in the lower panel , reveal that turnoff luminosities are consistent to within 0.02 mag in @xmath64 over the entire range of masses plotted , and to within 0.008 mag over the range of turnoff masses relevant to globular cluster studies .
it should be emphasized that the points plotted in the eep relations of figs . 2 and 3 are taken _ directly _ from the model tracks . in most instances ,
they were identified automatically by software designed to recognize the primary eeps .
thus , they directly reflect the behaviour of the models at the masses specified .
one problem that is exposed by the spline approach occurs as a result of the fact that there are usually more tracks with zero - age main sequence eeps than turnoff eeps , and more with turnoff eeps than with eeps on the subgiant branch ( because lower mass tracks are not extended past an age of 30 gyr ) .
consequently , the nature of the spline fit to the eeps changes abruptly when the number of points used to construct the fit changes .
this sometimes manifests itself as a small discontinuity on an isochrone generated via the akima spline , that would not be evident on one generated by linear interpolation .
( we emphasize that such discontinuities are virtually undetectable ; we have ensured this by extending low mass tracks to the base of the rgb in the grids where discontinuities appeared on isochrones near the turnoff . ) while the akima spline provides an accurate means of interpolating the points along an isochrone , its chief advantage over linear interpolation is that derivatives can be evaluated directly at each point . a comparison between differential ipfs on the theorist s plane , derived from akima spline interpolation via the approach defined by equations ( 2 ) and ( 3 ) , and ipfs calculated directly from the tabulated masses and distances obtained via linear interpolation , is shown in figure [ f4 ] .
the discontinuities evident on the main sequence portion of the linearly interpolated ipfs occur wherever an isochrone crosses one of the evolutionary tracks .
they are caused by the abrupt changes in the slopes of the interpolation relations at these points .
such discontinuities would also be evident in linearly interpolated lfs and cfs ; they did not appear in the lfs presented in bv92 because the slopes of the eep relations were derived from cubic polynomial fits to the eep points .
first we describe the general morphology of the ipfs with respect to the expected evolutionary behavior , using the examples plotted in figure [ f5 ] .
the transition from the main sequence to the rgb occurs near @xmath65 for both the metal - rich and the metal - poor cases , because the distance , as we have defined it , is calculated with reference to a point on the subgiant branch 0.05 mag redward of the turnoff .
the relatively broad local maximum immediately following the transition , at @xmath66 , corresponds to the inflection point in the temperature evolution at the base of the rgb .
the evolutionary pause on the rgb is located at a distance near @xmath67 to @xmath68 in both cases .
the extension of the upper rgb to distances beyond @xmath69 in the @xmath70=+0.12 $ ] grid occurs because giant branch evolution in color is stronger than in more metal - poor stars , and color evolution is weighted more strongly than luminosity evolution in the calculation of distance along an isochrone . in the theorist s plane ,
the derivatives of luminosity and temperature with respect to time along evolutionary sequences tend to have some numerical noise on the lower rgb , more obviously in the metal - rich grids .
the very fine wiggles seen on the lower rgb portion of the interpolated metal - rich ipf plotted in the upper panel of fig .
5 are a manifestation of this .
all of the other structures not predicted by canonical theory arise from the adopted color@xmath3 relations and the bolometric corrections .
this is particularly evident at the faint end of the models in the upper panel of fig . 5 , where the plateau - like structure near @xmath71 in the ipf has no counterpart at the faint end of the theorist s ipfs plotted in fig . 4 .
similar irregularities can be seen on the upper rgb , such as the one that is weakly evident near @xmath72 in the upper panel .
it turns out that this anomaly is also found near @xmath73 in the corresponding cf ( not shown ) , and that it occurs over the same range in ( ) as the plateau - like structure on the main sequence .
moreover , artifacts of the bolometric corrections , evident near the bright and faint ends of the corresponding lf ( not shown ) , are also found to occur over the same range in ( ) .
one conclusion that can be drawn from fig .
5 is that neither the color@xmath3 relations nor the bolometric corrections are solely responsible for a number of the features on the computed ipfs .
most of them are probably not real ; indeed , it is our impression that the transformations to the observed plane are not as consistently determined for the metal - rich grid as for the metal - poor grid .
for instance , only very minor artifacts appear on the ipfs for the latter case .
the small glitch near @xmath74 on the ipf can be traced to a corresponding glitch near @xmath75 on the cfs , etc .
comparisons with observations are needed for definitive conclusions : at this point , we can only say that , for the metal - rich grids in particular , the discontinuous behavior on both the main - sequence and on the rgb portions of the cfs between @xmath76 , and the corresponding wiggles at the bright and faint ends of the lfs originate in the same @xmath77 region of the transformation tables in which the color indices and bolometric corrections are interpolated . @xmath78 is prepared for publication . ]
for this paper , the grids of evolutionary tracks described in paper i have already been processed with software that automatically locates the primary eeps . in this section ,
we describe how to use the software and we give examples of the file formats currently available . each grid of models
is contained in an eep file ( denoted by the file extension .eep ) with the format illustrated in table [ tab1 ] .
the first seven lines have the same format in all of the subsequent files produced from the eep file .
the first line specifies that this particular file contains seven tracks ; lines 2 through 6 list the adopted chemical abundances and line 7 gives the assumed mixing length parameter .
for each track , there is a header line listing the mass , the number of models in the track ( npts ) , the last model generated via the lagrangian code ( match ) , the shifts in age and in @xmath79 applied to the subsequent giant branch models computed via the @xcite method , the age of the first model on the track , and the model numbers of the primary eeps . in the example illustrated , the primary eeps for the @xmath80 track are located at models 1 , 36 , 134 , 198 , 278 , and 434 this track does not have a convective hook and thus the three spaces following the turnoff point ( 36 ) are blank . within each track , the entries are the model number , the luminosity , the effective temperature , the age , the central hydrogen content , the derivative of luminosity with respect to time , and the derivative of the effective temperature with respect to time .
[ tab1 ] isochrones on the @xmath81@xmath79 plane are generated with the program mkiso , which interactively prompts the user for the input eep file and a series of isochrone ages .
table [ tab2 ] illustrates the format of the isochrone files ( denoted by the extension .iso ) generated .
the first line indicates that the file contains 6 isochrones . the header line for each isochrone lists its age and
the number of points generated .
the individual lines for each isochrone list the luminosity , the effective temperature , the mass , the distance along the isochrone , and two values of the derivative of the mass with respect to distance along the isochrone .
the first of these values is computed via equation ( 2 ) ; the second is computed directly via five point differentiation along the isochrone .
[ tab2 ] isochrones , ipfs , lfs , and cfs are generated with the program mkipf , which prompts the user for the input isochrone file , the passband ( @xmath82 , @xmath83 , @xmath84 , or @xmath85 ) for the magnitudes , the color index ( , , , , or ) , and three values for the exponent , @xmath86 , in the power law mass spectrum given by @xmath87 .
the user then specifies the type of output desired and is asked to specify the bin size .
for example , if the option to produce an ipf file ( with the extension .ipf ) is chosen , the bin size is specified in terms of the distance along the isochrone , in whatever passband and color index were selected ; if the lf option ( with the extension .lfn ) is chosen , the bin size is specified in terms of the passband selected ; if the cf option ( with the extension .cfn ) is chosen , the bin size is specified in terms of the color index selected . the format of an ipf file is shown in table [ tab3 ] . in this example , the magnitude passband and the color index selected were @xmath85 and , respectively .
the three values supplied for the mass - spectrum exponent were @xmath88 , @xmath89 , and @xmath90 . for each age specified in the input ioschrone file ,
the isochrones and ipfs are tabulated at equal distance intervals .
the entries on a line are the magnitude , the color index , the mass , the bolometric magnitude , @xmath27 , @xmath91 , the distance along the isochrone ( the bin size is specified with respect to this distance ) , and three pairs of differential and cumulative ( logarithmic ) ipfs , one pair for each value of @xmath86 specified . the last line for each age
specified gives the parameters for a model at the rgb tip .
[ tab3 ] the format of an lf file is shown in table [ tab4 ] . in this example , both the differential and cumulative distributions are tabulated at the centers of 0.2 mag bins , except for the bin at the rgb tip .
the format of a cf file is almost the same .
however , since stars in the post - turnoff stages of evolution have the same colors as those along the main sequence , the distributions have been separated at the color of the turnoff .
for the example cf file shown in table [ tab5 ] , there are 53 ( nms ) bins associated with the main sequence in the 8 gyr model ; the post - turnoff portion begins at bin 54 .
[ tab4 ] [ tab5 ]
in order to make reliable stellar populations models , the predicted luminosity , @xmath3 , and color scales of the underlying evolutionary tracks and isochrones must be as accurate as possible . in this section
, we describe some further tests of the theoretical calculations , beyond those presented in papers i and ii , which show that they seem to satisfy most observational constraints rather well .
as discussed in paper i , the value of the mixing - length parameter , @xmath92 , that has been assumed in these calculations , was set by the requirement that a standard solar model reproduce the observed temperature of the sun .
( to be consistent with the models presented in this series of papers , the reference solar model did not take diffusive processes into account . ) in support of the resultant @xmath25 scale , it was shown that the location on the ( @xmath93)plane of the well - observed subdwarf hd@xmath6103095 ( perhaps better known by the name groombridge 1830 ) is matched quite well by models for the appropriate chemical abundances if its temperature is @xmath94 k. although there is considerable support for this estimate ( e.g. , @xcite ; @xcite ) , many would argue that something closer to 5000 k is more realistic ( see , e.g. , @xcite ; @xcite ) . indeed
, the effective temperatures of most stars , including the best - studied population ii subdwarfs , appear to be uncertain at the level of about 100 k , which clearly limits the ability of such data to provide stringent constraints on predicted @xmath95 scales .
however , it is worthwhile to explore this issue further for the following reason .
paper i concluded that the present models provide a good fit to the local subdwarfs on the [ @xmath96diagram , _ if _ metallicities close to those adopted by gratton et al . (
1997 ) are assumed for them . but whether or not comparable agreement is found on the [ @xmath93]plane , if the _ temperatures _ derived by gratton et al
. are assumed , was not investigated .
in fact , it turns out that there is quite good consistency between these two plots . to illustrate this ,
we have constructed `` mono - metallicity '' subdwarf sequences consisting of those subdwarfs that have the smallest uncertainties in their parallaxes , along with a few subgiants to permit an evaluation of the ages of stars that have evolved past the main - sequence turnoff .
that is , we have used our isochrones ( only in a differential sense ) to correct the inferred @xmath25 and intrinsic color of each field halo star , at its observed @xmath98 , to the temperature and color it would have if it had [ fe / h ] = @xmath0 .
( we chose the lowest metallicity in our grid as the reference [ fe / h ] value because the few subgiants that we have considered are very metal - deficient and it is desirable that the @xmath25/color adjustments which are applied to them be as small as possible to avoid introducing significant errors into the determination of their ages . ) the result of this exercise on the [ @xmath99plane is shown in figure [ f6 ] .
the @xmath98 , @xmath100 , and [ fe / h ] values that have been adopted are from the updated list of parameters given by @xcite,188510 appears to be incorrect : the _ hipparcos _ catalogue gives @xmath101 mas for this star ( instead of 22.80 mas ) , which implies that its @xmath98 is @xmath102 , if @xmath103 is taken to be its apparent magnitude and a lutz - kelker correction of @xmath104 mag is applied . ] while the sources of the temperature data are gratton et al .
( 1996 ) and , in the case of hd@xmath6134439 and hd@xmath6134440 , @xcite.132475 and hd@xmath6145417 were kindly provided to us by raffaele gratton : they are , in turn , 5800 k and 4953 k. ] as the @xmath2-elements appear to be overabundant by @xmath105 dex in the vast majority of field halo stars ( see , e.g. , the clementini et al . study ) , the offsets in @xmath25 [ and in @xmath106 color , to be discussed shortly ] that were applied to the selected stars were derived from our isochrones for [ @xmath2/fe ] @xmath107 , an age of 14 gyr , and the range in [ fe / h ] between @xmath0 and @xmath108 . [ with the exception of hd@xmath6132475 , the locations of stars in fig .
6 are nearly , or completely , independent of the age of the isochrones which are used to determine the offsets .
moreover , the position of hd@xmath6132475 remains indicative of high age ( @xmath109 gyr ) , irrespective of which isochrones are employed in this analysis . ]
the main conclusion to be reached from fig .
6 is that the theoretical and `` observed '' @xmath25 scales at @xmath110 agree extremely well .
this is very encouraging from our perspective because , as discussed in their paper , the gratton et al .
( 1996 ) determinations of effective temperature are consistent with those derived from a number of methods including the fitting of balmer line profiles ( e.g. , @xcite ) , which is widely considered to be one of the more accurate techniques .
it is interesting to note that hd@xmath619445 is believed to have quite high oxygen and @xmath2-element abundances ( [ o / fe ] @xmath111 and [ @xmath2/fe ] @xmath112 , according to gratton et al .
1997 ) , which may explain why this star sits to the right of the isochrones . on the other hand , it is puzzling that the locations of hd@xmath6134439 and hd@xmath6134440 are not discrepant , given that they are examples of the small minority of population ii field stars that apparently have [ @xmath2/fe ] @xmath113 ( @xcite ) .
however , we would not expect their displacements to hotter temperatures to be more than @xmath1140.010 , judging from our models , and such small shifts are probably consistent with the uncertainties associated with the @xmath25 and [ fe / h ] determinations .
( we have no explanation for the somewhat anomalous location of hd@xmath6145417 relative to the locus defined by the other subdwarfs . ) as illustrated in figure [ f7 ] , the same subdwarfs define a mono - metallicity sequence on the @xmath115$]plane that is also in good agreement with the same isochrones , when the predicted effective temperatures are converted to @xmath116 colors using very close to @xcite transformations ( see paper i for some discussion of the adopted color@xmath25 relations ) .
thus , there is no obvious need to adjust either the temperatures or the colors of the models in order to satisfy the constraints posed by the subdwarf sample that has been considered .
granted , a couple of the stars ( hd@xmath625329 , hd@xmath6188510 ) are slightly displaced from the theoretical loci in fig . 7 , but not in fig . 6 , which suggests that the adopted temperatures and colors are not perfectly consistent with one another . but this could simply be an indication that the reddenings of these two stars are slightly higher than the values adopted by carretta et al .
an underestimation of the reddenings of hd@xmath6134439 and hd@xmath6134440 could also explain why they sit on the same locus as hd@xmath6103095 despite the latter having [ @xmath2/fe ] @xmath117 ( gratton et al .
1997 ) , and the former two stars having [ @xmath2/fe ] @xmath13 , as already noted . the other striking feature of figs .
6 and 7 is that high ages are implied by the locations of the post - turnoff stars .
as emphasized by @xcite , hd@xmath6140283 provides an especially strong argument that there are very old stars and , furthermore , that the most metal - deficient globular clusters must be of comparable age if the local subgiants are representative of cluster subgiants of similar metal abundance .
indeed , paper ii has shown that excellent agreement between synthetic and observed c - m diagrams is obtained for such systems as m@xmath692 if its age is @xmath118 gyr . to be sure
, there is sufficient flexibility in the mixing length theory of convection to obtain good fits to the morphology of an observed c - m diagram for any assumed distance ( within reason ) .
but if m@xmath692 is as young as 12 gyr , say , then the zero - point of our adopted color@xmath25 relations would require a large correction , in conflict with our findings from figs . 6 and 7 , and m@xmath692 would be @xmath119 gyr younger that field halo stars of similar [ fe / h ] , which seems unlikely . to elaborate on this point , we show in figure [ f8 ] two fits of isochrones to the m@xmath692 c - m diagram where the only difference in the models is the choice of @xmath92 , the usual mixing - length parameter .
the left - hand panel reproduces the same isochrones that were plotted in fig . 7 and it is obvious that a superb match to the cluster fiducial is obtained if the adopted distance is near @xmath120 and if @xmath121 mag ( @xcite ) .
this case assumes a value of @xmath92 that is needed to satisfy the solar constraint ( see paper i ) , and the distance is approximately what one would infer from hd@xmath6140283 if m@xmath692 and the field subgiant are close to being coeval .
however , even if the cluster distance modulus is actually as high as @xmath122 , it is still possible to obtain an equally fine fit to the _ shape _ of the observed c - m diagram simply by adopting @xmath123 ( see the right - hand panel ) .
is assumed .
hence , in this instance , one may have to postulate some variation in the value of the mixing - length parameter with evolutionary state in order to match the entire c - m diagram .
of course , there is no reason why this parameter should be completely independent of @xmath25 and/or @xmath91 , but it is interesting that no such difficulty is found if @xmath124 . ]
the main problem with this scenario is that a very large correction must be applied to the model colors in order for the relevant isochrone ( for an age of @xmath125 gyr ) to match the observed turnoff color .
such a large color shift would seem to be precluded by the comparisons given in figs . 6 and 7 .
while paper i and the present discussion appear to offer strong support for the @xmath25 and [ fe / h ] scales derived by gratton et al .
( 1996 ) and carretta et al .
( 2000 ) for the local subdwarfs , paper ii has suggested that the metallicities determined by @xcite for globular clusters , especially those with @xmath126 [ fe / h ] @xmath127 , are too high .
something much closer to the @xcite scale was favored . because this is such an unexpected ( and controversial ) result , some additional comments on this matter are worthwhile .
in particular , if our inference from figs .
6 and 7 is correct , that our isochrones require little or no adjustment of the predicted colors to match the local calibrators , then we should obtain reliable estimates of globular cluster distances by performing main - sequence fits of the cluster fiducials to our isochrones .
quite understandingly , the results of such an exercise depend sensitively on what metal abundances are assumed .
this is illustrated in figure [ f9 ] , using m@xmath63 as an instructive example .
recall that carretta & gratton ( 1997 ) have determined [ fe / h ] @xmath128 for this system whereas the zinn & west ( 1984 ) estimate is @xmath129 .
hence , the most appropriate isochrones in our grid to use in this analysis are those having [ fe / h ] @xmath130 and @xmath131 .
if the @xcite fiducial for m@xmath63 is fitted to the models of higher metallicity ( see the left - hand panel ) , then one obtains @xmath132 and an age of @xmath133 gyr .
( a main - sequence fit to local subdwarfs will necessarily yield the same relatively young age for m@xmath63 if the cluster iron abundance is as high as the carretta & gratton estimate . )
this is to be compared with @xmath134 and an age of @xmath135 gyr if a fit is performed to the lower - metallicity isochrones ( see the right - hand panel ) .
[ our transformations to @xmath136 are precisely as given by bell & gustafsson ( 1989 ) and , as shown in papers i and ii , very similar interpretations of both cluster and subdwarf c - m diagrams are generally found regardless of whether the color used is @xmath116 or @xmath136 . ]
the isochrones for [ fe / h ] @xmath137 clearly fail to match the morphology of the m@xmath63 fiducial whereas the predicted and observed loci agree extremely well if the assumed [ fe / h ] is @xmath131 . by itself , this is not a particularly strong argument against the comparison shown in the left - hand panel of fig .
9 , given that isochrone shapes are so easily manipulated ( as we have already demonstrated ) .
a more compelling reason for rejecting this possibility is the fairly strong case that has been made in support of an age of @xmath118 gyr for m@xmath692 and the finding that isochrones for this age , and a metallicity within the uncertainties of most determinations , do not require any _ ad hoc _ adjustments to either the model temperatures or colors to achieve nearly perfect coincidence with the cluster ( and subdwarf ) c - m diagrams .
it seems highly improbable that m@xmath63 is 56 gyr younger than m@xmath692 and that the isochrone fit would would be so problematic for the one cluster , but not for the other .
there are no such difficulties if the [ fe / h ] value obtained by zinn & west ( 1984 ) for m@xmath63 is more realistic than that derived by carretta & gratton ( 1997 ) .
this conclusion does depend on the accuracy of the assumed reddenings and @xmath2-element abundances as well as , among other things , the photometric zero - points , so it may well be revised somewhat as further progress is made .
however , it would not be too surprising if globular cluster abundances , as derived from spectroscopic data for the brightest giants , are not quite on the same scale as that of local field halo dwarfs .
( note that the age estimates discussed above should probably be reduced by about 10% to take into account the effects of diffusion , though current diffusive models are not without their difficulties : for some discussion of this point , see grundahl et al .
values for large samples of turnoff and lower rgb stars in two globular clusters , ngc@xmath66397 and ngc@xmath66752 . the failure to find any variation of [ fe / h ] across the cluster subgiant branches ,
together with the unexpectedly high li abundances found in both cluster ( @xcite ) and field turnoff stars ( e.g. , @xcite , @xcite ) , provide extremely important constraints on diffusion in population ii stars that have yet to be properly evaluated .
it seems very likely that the age effect , which arises from the settling of helium in the deep interior , will persist , but the effective temperatures predicted by currently available diffusive models may well need to be revised significantly . a collaboration between the university of montreal ( g. michaud and colleagues ) and d.a.v .
is underway to study this problem . ] ) paper i has already shown that , both in zero - point and slope , the computed giant branches on the ( @xmath18)plane agree very well with the loci inferred for globular clusters from infrared photometry .
our intention here is merely to point out that reasonably good agreement between theory and observations is also found on the various c - m diagrams that can be constructed from @xmath138 magnitudes . as an example , we show in figure [ f10 ] how well our isochrones are able to reproduce the @xcite fiducials for m@xmath615 , ngc@xmath66752 , ngc@xmath61851 , and 47 tuc on the [ @xmath7diagram . in preparing this figure , we have adopted the reddenings derived by schlegel et al .
( 1998 ) , relations giving @xmath139 and @xmath140 as a function of @xmath141 from @xcite , and the distance moduli that were derived in paper ii ; thus , the fiducials tabulated by da costa & armandroff have been have corrected to take the small differences in @xmath141 and @xmath142 between their adopted parameters and ours into account .
although we have plotted the giant - branch segments of isochrones for certain ages , to be consistent with the assumed distances , the comparison is fairly independent of this choice .
it should be noted , as well , that the implied metallicities of the four clusters are within 0.1 dex of the [ fe / h ] values listed in the compilation of such data by @xcite . at [ fe / h ]
@xmath143 , the adopted transformations to @xmath136 are exactly as given by bell & gustafsson ( 1989 ) , whereas some adjustment of their color@xmath25 relations ( only at low gravities ) was required at higher metal abundances to achieve consistency between the theoretical and observed planes .
these corrections were always to the red and they tended to be systematically larger at higher [ fe / h ] values , which may indicate the lack of sufficient blanketing in the model atmospheres constructed by bell & gustafsson for cool , relatively metal - rich giants . in any case , it is desirable to force the _ colors _ of the upper giant branches to agree with those observed in order to enhance the usefulness of our models in stellar population studies .
[ even though there are some discrepancies between the [ fe / h ] @xmath144 isochrone and the 47 tuc fiducial at the brightest magnitudes , this does not necessarily indicate a problem with the adopted color transformations
. it may be , for instance , that 47 tuc has a slightly higher iron abundance , and given the huge sensitivity of the @xmath136 color index ( as well as the bolometric corrections ) to @xmath25 at low temperatures , a better fit to the upper end of the cluster fiducial would be obtained if either the predicted temperatures were reduced slightly or the adopted [ fe / h ] value were increased by @xmath145 dex . ]
in contrast to the colors , which were constrained to some extent to satisfy empirical data ( for giants only ) , the bolometric corrections to @xmath83 magnitudes do represent purely theoretical predictions ( at all temperatures , gravities , and [ fe / h ] values ) .
in particular , we have adopted the bolometric corrections given by @xcite , which are based on kurucz model atmospheres and which span a very wide range in parameter space .
they agree extremely well in a systematic sense with those computed recently by r.a .
bell ( see , e.g. , the discussion by @xcite ) . for instance , we have found that bell s transformation of selected isochrones to fit the m@xmath692 and ngc@xmath62419 c - m diagrams in the study by @xcite is nearly indistinguishable with ours .
hence , a comparison of predicted and observed giant - branch tip magnitudes does provide a valid test of the theoretical models .
( bolometric corrections in @xmath85 were obtained from @xmath146 . ) considering that cluster distance moduli are currently uncertain by at least 0.2 mag , the luminosity at which giant - branch evolution is predicted to terminate , and its variation with metallicity , is in satisfactory agreement with observations ( see fig .
10 ) . if anything , slightly shorter distances , implying somewhat higher ages , may be indicated for m@xmath615 and ngc@xmath61851 .
it should be appreciated that the _ bolometric _ magnitudes for the giant - branch tip are within 0.05 mag of those predicted by a.v .
sweigart for stars of the same mass and chemical composition ( see paper i ) and within @xmath147 mag of those reported by e.g. , @xcite , when very similar physics is adopted .
( sweigart s models are fainter than ours while those by cassisi et al .
are brighter . ) as the same neutrino cooling rates have been used by the aforementioned investigators , we suspect that small differences in the respective equation - of - state formulations are responsible for the minor variations in the predicted tip magnitudes .
this should be investigated further .
photometry for the red - giant and horizontal - branch populations in 10 globular clusters .
we hope to examine how well our models are able to reproduce their observations at some future date , once we have fully investigated the transformations to @xmath148 and @xmath149 . ]
another important diagnostic in observed c - m diagrams is the luminosity on the red - giant branch where the h - burning shell in evolving stars passes through the chemical abundance discontinuity left behind when the convective envelope reached its maximum penetration ( near the base of the rgb ) .
when this occurs , the evolution up the giant branch either slows down or reverses direction for a short time as the structure of the star adjusts to a somewhat larger fuel supply ( higher h abundance ) .
this manifests itself as a local enhancement in the number of stars in an otherwise smoothly decreasing differential luminosity function for the giant - branch component of a populous system such as a globular cluster where , with few exceptions ( e.g. , @xmath150 cen ) , the member stars are essentially coeval and chemically homogeneous . as discussed by @xcite ,
the luminosity of this so - called `` rgb bump '' is a strong function of metallicity and age , and depends only weakly on other factors such as the helium abundance and the mix of heavy elements .
however , one must be wary of placing too much reliance on this feature as a distance ( and age ) constraint , given the possibility that there may be some amount of overshooting below the convective envelope ( e.g. , @xcite ) , or that opacities in the critical @xmath151 k regime are still inadequate , or that oxygen abundances are much higher than we have assumed .
( as shown in paper i , oxygen is the dominant contributor to the opacity at @xmath152 , and its abundance will therefore have important consequences for the depth of the envelope convection zones in giants ; also see vandenberg & bell 2001 ) .
table [ tab6 ] lists the bump magnitudes predicted by our models for ages from 10 to 16 gyr and the entire range of [ fe / h ] and [ @xmath2/fe ] values considered in this investigation .
the considerable sensitivity of @xmath153 to these three parameters is clearly evident , and a careful inspection of this table , in conjunction with table 2 of paper i ( which gives the @xmath154 value for each mixture ) , reveals that the chemical composition dependence is primarily with @xmath154 , the total mass - fraction abundance of the elements heavier than helium .
that is , the detailed distribution of the metals affects the results at only the level of a few to several hundredths of a magnitude .
compare , for instance , the tabulated magnitudes for [ fe / h ] @xmath155 and [ @xmath2/fe ] @xmath156 ( @xmath157 ) with those for [ fe / h ] @xmath158 and [ @xmath2/fe ] @xmath17 ( @xmath159 ) : the differences in @xmath153 are @xmath160 mag . only at relatively high values of @xmath154
do these differences rise to @xmath147 mag ; e.g. , compare the tabulated magnitudes for [ fe / h ] @xmath130 and [ @xmath2/fe ] @xmath156 ( @xmath161 ) with those for [ fe / h ] @xmath144 and [ @xmath2/fe ] @xmath17 ( @xmath162 ) .
[ tab6 ] at the same @xmath154 , the bump magnitudes predicted our models agree quite well ( generally to within @xmath163 mag ) with those reported by cassisi & salaris ( 1997 , see their table 1 ) and with those computed by @xcite , as noted by @xcite .
( some differences can be expected due to the fact that the different studies assume slightly different helium contents and heavy - element mixtures . ) thus , the interpretation of this feature in observed c - m diagrams should be nearly independent of whose models are used .
much more critical is the adopted metallicity . for instance , rood et al .
have found that our models for [ fe / h ] @xmath130 , [ @xmath2/fe ] @xmath107 , and an age of @xmath164 gyr accurately predict the location of the rgb bump in m@xmath63 ( as do the models by salaris et al .
for the same age and nearly the same @xmath154 , but with the metals in solar proportions ) . in this cluster , @xmath165 ( @xcite ) , and an age of 12 gyr
is obtained if the cluster distance modulus is close to @xmath166 , which implies @xmath167 , in excellent agreement with the relevant entry in table 6 . however , as we have argued above , such a high metal abundance for m@xmath63 is problematic .
indeed , rood et al . have found that it is impossible to choose a distance and metallicity such that the corresponding models are able to reproduce the entire luminosity function ( lf ) . in particular
, they found that a good fit to the subgiant lf resulted in a poor fit to the luminosity of the rgb bump , and vice versa .
an appreciably higher age ( @xmath19 gyr ) , and/or a lower metallicity , is indicated by the slope of the lf between the turnoff and lower giant branch .
it is especially unfortunate that the chemical composition of m@xmath63 is so uncertain at the present time : the [ fe / h ] values determined by zinn & west ( 1984 ) , carretta & gratton ( 1997 ) , and @xcite differ from one another by @xmath168 dex and span a range @xmath169 dex .
until these differences are understood , it will be difficult to say whether or not models accurately predict the observed bump magnitude in this cluster .
however , it does seem unlikely that such a reconciliation of theory and observation is possible in the case of m@xmath692 , _ if _ it is older than @xmath170 gyr . for this system ,
current metallicity determinations are much more consistent with one another e.g. , high - resolution spectroscopy of bright giants by @xcite and carretta & gratton ( 1997 ) has yielded [ fe / h ] @xmath171 and @xmath172 , respectively , while zinn & west give [ fe / h ] @xmath173 .
moreover , most spectroscopic studies seem to find [ @xmath2/fe ] @xmath117 for most globular clusters , including m@xmath692 ( see the spectroscopic studies mentioned above , as well as the reviews by kraft 1994 and carney 1996).92 stars have [ o / fe ] @xmath174 ( israelian et al . 1998 , boesgaard et al .
1999 ) , then the cluster age could be as low as @xmath175 gyr _ and _ there would be no discrepancy between the predicted and observed bump magnitudes ( see vandenberg & bell 2001 ) . ] as shown in table [ tab7 ] , our models are able to match the luminosity of the rgb bump in m@xmath692 only if the cluster metallicity is at the high end of the observed range and its age is @xmath176 gyr . according to ferraro et al .
( 1999 ) , @xmath177 in m@xmath692 , which corresponds to the absolute visual magnitudes given in the fifth column if the distance moduli in the third column are assumed . at these distances , the ages obtained from our models and the corresponding bump luminosities ( from table 6 )
are as tabulated in the fourth and sixth columns , respectively .
the last column , which contains the differences between the observed and predicted bump @xmath153 values , shows that the latter are too bright by 0.05 to 0.31 mag , depending on the distance and iron abundance that is assumed . in view of the fairly strong case that has been made in support of an age near 16 gyr for m@xmath692 ( see 4.1 above , and grundahl et al .
2000 ) , we are inclined to conclude that the our models fail to predict the luminosity of the rgb bump in this cluster by at least 0.2 mag .
[ tab7 ] in fact , a very similar conclusion is reached from an analysis of the c - m diagram for 47 tuc , which has [ fe / h ] @xmath178 .
it has @xmath179 mag ( ferraro et al .
1999 ) , and if we adopt @xmath180 , which is within 0.1 mag of most estimates ( and slightly higher than our preferred value : see paper ii ) , then @xmath181 . at this distance , the age and
bump magnitude predicted by our isochrones for [ fe / h ] @xmath182 and [ @xmath2/fe ] @xmath107 ( @xcite ) are @xmath183 gyr and 0.91 mag , respectively . the corresponding numbers for isochrones having [ fe / h ] @xmath184 and the same @xmath2/fe ratio ( carretta & gratton 1997 ) are @xmath185 gyr and 1.00 mag .
thus , the discrepancy between the predicted and observed values of @xmath153 is , in turn , 0.24 and 0.15 mag .
although the agreement could be improved by adopting a larger distance , any age younger than @xmath186 gyr at these metallicities poses problems for other features in the observed c - m diagram .
for instance , the subgiant branch in 47 tuc has a gentle upward slope between the turnoff and lower rgb on the [ @xmath187diagram , while isochrones for ages @xmath188 gyr predict that the variation in @xmath83 across the subgiant branch is non - monotonic ( see fig .
34 in paper ii ) .
if our models truly are incapable of explaining the location of the rgb bump in both m@xmath692 and 47 tuc , then it seems unlikely that they would fare any better at intermediate metal abundances .
that is , there is some basis for expecting that the bump luminosity in m@xmath63 _ should _ be @xmath189 mag fainter than that predicted by the most appropriate isochrone in our grid .
it turns out to be quite difficult to force such a discrepancy unless the cluster is more metal poor than [ fe / h ] @xmath190 .
this becomes readily apparent when one considers figure [ f11 ] , which illustrates how well isochrones for [ fe / h ] @xmath130 and [ @xmath2/fe ] @xmath107 are able to reproduce the m@xmath63 c - m diagram , on the assumption of smaller distances than that derived from a main - sequence fit ( see the left - hand panel of fig .
9 ) . to identify which isochrone has the same turnoff luminosity as the cluster , small offsets to the synthetic colors were applied ( as indicated).3 has a lower [ fe / h ] , because little or no adjustment to the model colors seems to be necessary to match the properties of the local subdwarfs ( as discussed above and in paper i ) . however , as noted in paper ii ( see fig . 12 therein ) , the three best photometric data sets currently available for m@xmath63 have zero - points that differ by up to @xmath191 mag in @xmath136 : the stetson et al .
( 1999 ) fiducial , which we have used , is the bluest , while those by @xcite and rood et al .
( 1999 ) are redder .
had we used either of the latter , the offsets applied to the isochrones in fig .
11 would have been smaller in an absolute sense by 0.010.015 mag ( and a smaller distance moduli by @xmath192 mag would have been derived in fig .
the accuracy of photometric zero - points is obviously an important concern in the main - sequence fitting technique . ] since @xmath193 , the assumption of @xmath194 and 14.85 imply that @xmath195 and 0.60 mag , respectively .
the observed turnoff luminosities in these two cases are clearly very close to those predicted by 12 and 14 gyr isochrones , for which the predicted bump magnitudes ( see table 6 ) are , in turn , 0.45 and 0.52 mag .
because of the substantial dependence of @xmath153 on age , the difference between the theoretical and observed bump luminosities varies relatively slowly with distance modulus .
thus , in order to obtain @xmath196 mag , m@xmath63 would have to be older than 16 gyr , if it has [ fe / h ] @xmath197 and [ @xmath2/fe ] @xmath117 .
but such a high age would imply that the isochrone colors need to be corrected blueward by @xmath198 mag in @xmath136 , which is completely unacceptable .
there is no such difficulty if a metallicity close to the zinn & west ( 1984 ) estimate is assumed ( see the right - hand panel of fig .
note , as well , how much better the isochrones for lower @xmath154 match the observed c - m diagram .
interestingly , the 14 gyr isochrone for [ fe / h ] @xmath130 and [ @xmath2/fe ] @xmath107 _ is _ able to reproduce both the location of the main sequence and the rgb , once a large zero - point adjustment is applied to the synthetic colors , but not the slope of the subgiant branch ( see the right - hand panel of fig .
11 ) . here is an example where it would not be possible to match the entire c - m diagram , if @xmath199 , simply by assuming a different value of @xmath92 in the construction of the isochrones : there would have to be at least two things wrong with the models .
all things considered , the [ fe / h ] value determined by carretta & gratton ( 1997 ) for m@xmath63 seems to be too high .
[ even the somewhat higher value obtained by the lick - texas group from high - resolution spectroscopy of bright m@xmath63 giants ( i.e. , [ fe / h ] @xmath200 , kraft et al . 1998 ) would seem to be too high by 0.10.2 dex according to the present analysis . ]
we conclude this section with figure [ f12 ] , which plots the @xmath153 versus [ fe / h ] relations given in table 6 for the [ @xmath2/fe ] @xmath107 case .
superimposed on this diagram are the predicted and observed bump magnitudes for m@xmath692 , m@xmath63 , m@xmath65 , and 47 tuc on the assumption of [ fe / h ] values that are within 0.1 dex of those given by zinn & west ( 1984 ) and which present relatively few difficulties for our models .
we have also adopted fairly high ages for these clusters , as indicated by the location of the _ open circles _ relative to the _
solid curves_. this is for the following reason .
paper ii found that , when zero - age horizontal branch loci are used to set the globular cluster distance scale , there was a significant dependence of mean age with metallicity . to be specific , the mean age varied smoothly from @xmath201 gyr at [ fe / h]@xmath202 to @xmath203 gyr at [ fe / h]@xmath204 .
however , it was also noted that this relation should be displaced to higher ages if the distances are set instead according to empirical determinations of the luminosities of rr lyrae stars from , e.g. , baade - wesselink and statistical parallax studies .
as argued therein and in the present investigation , the short distance scale is also implied by field population ii subdwarfs and subgiants .
thus , in order to satisfy other constraints , we conclude that our models fail to match the observed luminosities of the rgb bump in clusters by @xmath205 mag .
this is certainly speculation on our part , but not without justification .
the location of radiative convective boundaries in stars is subject to a lot of uncertain physics and it would not be too surprising that current models are somewhat lacking in this regard . whether or not the scenario that we have described is realistic may be difficult to confirm or refute until the uncertainties associated with current distance , @xmath25 , and chemical abundance determinations are significantly reduced .
the algorithms presented in bv92 to produce isochrones , lfs , and cfs from grids of evolutionary sequences have been improved with the implementation of the akima spline to replace linear interpolation .
experiments with the interpolating algorithms show that they work well , over a wide range in metallicity ( @xmath206[fe / h]@xmath207 ) and ages ( from @xmath15 gyr , which is the youngest age that has been examined , to 20 gyr ) , to produce the morphology of isochrones accurately .
moreover , what we have called the isochrone population function " , which literally gives the number of stars at any location along the corresponding isochrone , has been built into the interpolation algorithms .
software ( fortran 77 ) has been developed to implement the algorithms for grids of evolutionary sequences with @xmath208\le -0.30 $ ] .
each grid of sequences is available with three @xmath2-element enhancements : [ @xmath2/fe]@xmath17 , @xmath209 , and @xmath210 .
[ note that the diffusion of helium and heavy elements has not been treated in the models .
however , as mentioned at the end of section 4.2 ( also see grundahl et al .
2000 ) , current observations of population ii stars appear to be inconsistent with the predictions of diffusive models ( like those recently reported by salaris , groenewegen , & weiss 2000 ) , for reasons that have yet to be understood . ]
the evolutionary grids , together with the software are available to the astronomical community , so that users can generate isochrones , ipfs , lfs , and cfs for the particular set of abundances and ages of interest to them .
the ipf is a particularly powerful tool for stellar astrophysics and for stellar population studies because it contains all of the information found in isochrones , lfs , and cfs in a compact form .
for example , magnitudes and color indices for synthetic stellar populations can easily be generated from the cumulative form of the ipf for any pair of @xmath82 , @xmath83 , @xmath84 , and @xmath85 , for which empirically constrained color@xmath3 relations and bolometric corrections are supplied as part of the package .
( transformations to the strmgren system , presently being developed by j. clem at the university of victoria , will be provided in the near future .
those involving the @xmath211 , and possibly @xmath212 , bands will be added at a later date . ) a side benefit of the differential ipfs , lfs , and cfs is that they provide a way of fine tuning the transformation tables that are used to transpose the models from the @xmath81@xmath77 plane to the observer s plane .
the additional tests of the models that we have carried out to augment those described in papers i and ii suggest that the computed temperatures and adopted color@xmath3 relations are quite realistic .
the apparent agreement between the predicted and observed properties of local subdwarfs having well - determined parallaxes ( @xmath213 , as measured by _ hipparcos _ ) is especially encouraging .
indeed , there should be no difference between the distances which are derived to e.g. , globular star clusters when the observed fiducials are fitted either to the subdwarf calibrators or to our theoretical isochrones .
however , in practice , the large uncertainty that presently exists in the gc [ fe / h ] scale ( see , e.g. , @xcite ) is a severe limitation of the main - sequence fitting technique . be that as it may , it is our impression that something close to the zinn & west ( 1984 ) scale is required to achieve the best overall consistency between synthetic and observed cmds ( and between distance estimates based on the subdwarf and rr lyrae standard candles " see paper ii ) .
both in zero - point and in slope , and on either the theoretical or observed planes , our computed giant branches agree well with those derived observationally .
the models do not reproduce the luminosity of the rgb bump as well as they should , but this may simply be an indication that there is some overshooting of the convective envelopes of red giants into their radiative interiors as suggested some time ago by , e.g. , the padova group ( alongi et al .
1991 ) . in conclusion , we believe that the present grid of models is particularly well constrained and it should provide a valuable tool for the interpretation of stellar data , whether through the fitting of isochrones , lfs , and cfs or through the construction of integrated stellar population models whose properties can be compared with those of distant systems .
we thank raffaele gratton for providing his estimates of the effective temperatures for a few field halo stars .
the support of operating grants ( to p.a.b . and to d.a.v . ) from the natural sciences and engineering research council of canada is also gratefully acknowledged .
akima , h. 1970 , journal of the association for computing machinery , 17 , 589 alongi , m. , bertelli , g. , bressan , a. , & chiosi , a. 1991 , , 244 , 95 alonso , a. , arribas , s. , & martnez - roger , c. 1996 , , 117 , 227 balachandran , s.c . , & carney , b.w .
1996 , , 111 , 946 bell , r.a . , & gustafsson , b. 1989 , , 236 , 653 bergbusch , p.a . , &
vandenberg , d.a .
1992 , , 81 , 163 bv92 bergbusch , p.a .
, & vandenberg , d.a . 1997 , , 114 , 2604 bessell , m.s . , castelli , f. , & plez , b. 1998 , , 333 , 231 boesgaard , a.m. , king , j.r . ,
deliyannis , c.p . , & vogt , s.s . 1999 , , 117 , 492 bolte , m. 1994 , , 431 , 223 brown , j.a . ,
wallerstein , g. , & oke , j.b .
1990 , , 100 , 1561 brown , j.a . ,
wallerstein , g. , & zucker , d. 1997 , , 114 , 180 carretta , e. , & gratton , r.g .
1997 , , 121 , 95 carretta , e. , gratton , r.g . , clementini , g. , & fusi pecci , f. 2000 , , 533 , 215 carney , b.w .
1996 , , 108 , 900 cassisi , s. , castellani , v. , deglinnocenti , s. , & weiss , a. 1998 , , 129 , 267 cassisi , s. , & salaris , m. 1997 , , 285 , 593 clementini , g. , gratton , r.g .
, carretta , e. , & sneden , c. 1999 , , 302 , 22 da costa , g.s . , & armandroff , t.e . 1990 , , 100 , 162 eggleton , p.p .
1971 , , 151 , 351 ferraro , f.r . ,
messineo , m. , fusi pecci , f. , de palo , m.a . ,
straniero , o. , chieffi , a. , & limongi , m. 1999 , , 118 , 1738 ferraro , f.r . ,
montegriffo , p. , origlia , l. , & fusi pecci , f. 2000 , , 119 , 1282 fuhrmann , k. , axer , m. , & gehren , t. 1994 , , 285 , 585 fulbright , j.p . , & kraft , r.p .
118,527 gratton , r.g .
, bonifacio , p. , bragaglia , a. , et al .
2001 , astro - ph/0012457 gratton , r.g .
, carretta , e. , & castelli , f. 1996 , , 314 , 191 gratton , r.g .
, fusi pecci , f. , carretta , e. , clementini , g. , corsi , c.e . , & lattanzi , m. 1997 , , 491 , 749 grundahl , f. , vandenberg , d.a . ,
bell , r.a . ,
andersen , m.i . , & stetson , p.b . , 2000 , , submitted harris , w.e .
1996 , , 112 , 1487 harris , w.e . ,
bell , r.a . ,
vandenberg , d.a .
, et al . 1997 , , 114 , 1030 israelian , g. , garcia lopez , r.j . , & rebolo , r. 1998 , , 507 , 855 jimenez , r. & padoan , p. 1998
, , 498 , 704 johnson , j.a . , & bolte , m. 1998 , , 115 , 693 king , j.r .
1993 , , 106 , 1206 king , j.r .
1997 , , 113 , 2302 kraft , r.p .
1994 , , 106 , 553 kraft , r.p . ,
sneden , c. , smith , g.h . ,
shettrone , m.d . , & fulbright , j. 1998 , , 115 , 1500 mcwilliam , a. 1997 , , 35 , 503 molaro , p. , & pasquini , l. 1994 , , 281 , l77 nissen , p.e . , & schuster , w.j .
1997 , , 326 , 751 padoan , p. & jimenez , r. 1997 , , 475 , 580 renzini , a. , & fusi pecci , f. 1988 , , 26 , 199 rood , r.t . ,
carretta , e. , paltrinieri , b. , et al . 1999 , , 523 , 752 rutledge , g.a . ,
hesser , j.e . , & stetson , p.b .
1997 , , 109 , 907 salaris , m. , groenewegen , m.a.t . , &
weiss , a. 2000 , , 355 , 299 schlegel , d.j . ,
finkbeiner , d.p .
, & davis , m. 1998 , , 500 , 525 sneden , c. , kraft , r.p . , prosser , c.f .
, & langer , g.e .
1991 , , 102 , 2001 spite , f. , & spite , m. 1982 , , 115 , 357 swenson , f.j .
1995 , , 438 , l87 stetson , p.b . ,
bolte , m. , harris , w.e . , et al . 1999 , , 117 , 247 straniero , o. , chieffi , a. , & limongi , m. 1997 , , 490 , 425 vandenberg , d.a .
1992 , , 391 , 685 vandenberg , d.a . 2000 , , 129 , 315 ( paper ii ) vandenberg , d.a . , & bell , r.a .
2001 , new astron .
, in press vandenberg , d.a . ,
bolte , m. , & stetson , p.b . 1990 , , 100 , 445 vandenberg , d.a . , & irwin , a.w . 1997 , in advances in stellar evolution , eds .
rood & a. renzini ( cambridge : cambridge univ .
press ) , p. 22
vandenberg , d.a . , swenson , f.j . ,
rogers , f.j . , iglesias , c.a .
, & alexander , d.r .
2000 , , 532 , 430 ( paper i ) wheeler , j.c . , sneden , c. , & truran , j.w .
1989 , , 27 , 279 wood , p.r .
, & bessell , m.s . 1994 , private communication by anonymous @xmath214 zinn , r. , & west , m.j . 1984 , , 55 ,
45 l + tracks 7 + -0.525 + + 0.3 + z 1.010d-02 z = 6.000d-03 + alpha - element enhancement + x 0.742900 + y 0.247000 + alpha(mlt ) 1.89 + + mass npts match d(age ) d(log teff ) zage primary eeps + 1.100 434 194 + 4.4901d-04 + 2.7616d-04 0.0391 1 36 134198278434 + 1 0.132689 3.792568 3.9100d-02 0.7407 2.5689242d-02 1.4486261d-03 + 2 0.133692 3.792632 2.5526d-03 0.7404 2.4341290d-02 1.3111739d-03 + 3 0.135469
3.792717 8.9341d-03 0.7393 2.0199367d-02 8.8881749d-04 + 4 0.137098 3.792766 1.3401d-02 0.7376 1.6208193d-02 4.8355037d-04 + 5 0.138709 3.792789 2.0102d-02 0.7352 1.5496454d-02 4.0982141d-04 + . .
. . . .
. + . .
. . . .
. + . .
. . . .
. + 434 3.408267 3.530332 1.9975d-04 0.4746 1.9413841d+03 -2.1235817d+02 + + mass npts match d(age ) d(log teff ) zage primary eeps + 1.000 432 178 + 5.1563d-04 + 1.4076d-04 0.0500 1 49 134188268432 + 1 -0.067340 3.770430 5.0000d-02 0.7407 2.3175805d-02 -5.0453201d-04 + 2 -0.066481 3.770415 3.3192d-03 0.7404 2.2202865d-02 -4.4192319d-04 + 3 -0.064795 3.770383 1.1617d-02 0.7394 1.9219138d-02 -2.4991998d-04 + 4 -0.063101 3.770375 1.7426d-02 0.7379 1.5988973d-02 -6.7321489d-05 + 5 -0.061384 3.770385 2.6138d-02 0.7356 1.5124787d-02 1.7622827d-04 + . .
. . . . .
+ . .
. . . .
. + . .
. . . .
. + . . .
. .
. .
+ + l + isochrones 6 + -0.525 + + 0.3 + z 1.010d-02 z = 6.000d-03 + alpha - element enhancement + x 0.742900 + y 0.247000 + alpha(mlt ) 1.89 + + age npts + 8.0 256 + 1 -1.333304 3.593967 0.5068505908 0.0000000 4.4097889d-01 4.2796556d-01 + 2 -1.285759 3.594365 0.5313098653 0.0595653 3.7309103d-01 3.6969990d-01 + 3 -1.236702 3.596696 0.5532147760 0.1251682 2.9388266d-01 3.0552844d-01
+ 4 -1.187757 3.600882 0.5724189213 0.1992959 2.2815211d-01 2.4122070d-01 + 5 -1.140842 3.606441 0.5888593464 0.2801004 1.8235053d-01 1.9184230d-01 + . .
. .
+ . .
. . .
+ . .
. .
. . .
+ 256 3.408755 3.529098 1.0791668980 7.7891261 2.0186710d-04 1.5983838d-04 + + age npts + 10.0 253 + 1 -1.339361 3.594616 0.5002053380 0.0000000 4.5027950d-01 4.5082604d-01 + 2 -1.299791 3.594533 0.5213256448 0.0494690 3.9824028d-01 3.9623085d-01 + 3 -1.258991 3.595736 0.5405685847 0.1018699 3.3393240d-01 3.3839995d-01
+ 4 -1.217881 3.598282 0.5578690324 0.1592193 2.7157814d-01 2.8065268d-01 + 5 -1.177527 3.601987 0.5732006971 0.2218015 2.2171010d-01 2.3005911d-01 + . .
. .
. .
+ . .
. .
. .
+ . .
. . . .
. + . .
. . . . .
+ + l + ipfs 6 i v - i x = -0.5 + 0.5 + 1.5 + -0.525 + + 0.3 + z 1.010d-02 z = 6.000d-03 + alpha - element enhancement + x 0.742900 + y 0.247000 + alpha(mlt ) 1.89 + + age npts + 8.0 99 + i v - i mass mbol log te log g d x = -0.5 x = + 0.5 x = + 1.5 + 1 7.262 1.651 0.5418861 7.907 3.5952 4.768 6.400 4.52271 5.00000 4.66645 5.00000 4.79193 5.00000 + 2 7.101 1.621 0.5669044 7.756 3.5995 4.745 6.200 4.32906 4.97005 4.45372 4.95771 4.55961 4.94257 + 3 6.980 1.581 0.5833506 7.643 3.6043 4.731 6.000 4.16448 4.94973 4.27571 4.92960 4.36835 4.90503 + 4 6.891 1.536 0.5956772 7.550 3.6094 4.723 5.800 4.06875 4.93525 4.17060 4.90989 4.25386 4.87902 + 5 6.811 1.490 0.6063663 7.462 3.6147 4.717 5.600 4.01700 4.92327 4.11100 4.89376 4.18639 4.85793 + . . . . . . . . . . . . . .
+ . .
. .
. . .
. . . . . .
. + . .
. . . . . . . . . . .
. + 99 -4.046 3.070 1.0791651 -3.757 3.5297 0.140 -13.187 0.51774 -0.24401 0.36126 -0.40049 0.18615 -0.57560 + -4.064 3.091 1.0791669 -3.772 3.5291 0.132 -13.273 + + age npts + 10.0 101 + i v - i mass mbol log te log g d x = -0.5 x = + 0.5 x = + 1.5 + 1 7.302 1.653 0.5314385 7.948 3.5950 4.775 6.200 4.61279 5.00000 4.74746 5.00000 4.86529 5.00000 + 2 7.133 1.626 0.5592527 7.786 3.5986 4.747 6.000 4.41954 4.96284 4.53063 4.94850 4.62527 4.93112 + 3 6.998 1.590 0.5778846 7.661 3.6034 4.731 5.800 4.22658 4.93724 4.32342 4.91398 4.40341 4.88590 + 4 6.904 1.545 0.5909741 7.564 3.6083 4.721 5.600 4.12586 4.91999 4.21263 4.89109 4.28255 4.85631 + 5 6.822 1.500 0.6021440 7.474 3.6135 4.714 5.400 4.06100 4.90581 4.13958 4.87249 4.20130 4.83250 + . .
. . . . . . . . . . .
. + . .
. . . . . . . . . . .
. + . .
. . . . . . . . . . .
. + . .
. . . . . . . . . . . .
+ + l + lfs 6 i v - i x = -0.5 + 0.5 + 1.5 + -0.525 + + 0.3 + z 1.010d-02 z = 6.000d-03 + alpha - element enhancement + x 0.742900 + y 0.247000 + alpha(mlt ) 1.89 + + age npts + 8.0 57 + i x = -0.5 x = + 0.5 x = + 1.5 + 1 7.200 4.54403 5.00000 4.68367 5.00000 4.80591 5.00000 + 2 7.000 4.46666 4.96849 4.58374 4.95591 4.68324 4.94056 + 3 6.800 4.43036 4.94023 4.52787 4.91738 4.60780 4.88966 + 4 6.600 4.32834 4.91252 4.40940 4.88043 4.47284 4.84172 + 5 6.400 4.27108 4.88928 4.33876 4.85003 4.38882 4.80289 + . . . . . . . .
+ . .
. . . . .
. + . .
. . . . .
. + 56 -3.800 1.17235 0.73334 1.02008 0.58107 0.85016 0.41115 + 57 -3.982 1.17301 0.38695 1.02073 0.23467 0.85081 0.06475 + + age npts + 10.0 57 + i x = -0.5 x = + 0.5 x = + 1.5 + 1 7.200 4.59146 5.00000 4.71916 5.00000 4.83211 5.00000 + 2 7.000 4.51210 4.96470 4.61695 4.95194 4.70681 4.93658 + 3 6.800 4.47533 4.93293 4.56047 4.90980 4.63058 4.88213 + 4 6.600 4.36671 4.90154 4.43538 4.86909 4.48901 4.83051 + 5 6.400 4.30264 4.87542 4.35813 4.83585 4.39858 4.78902 + . . . . . . . .
+ . .
. . . . .
. + . .
. . . . .
. + . .
. . .
. .
+ + l + cfs 6 i v - i x = -0.5 + 0.5 + 1.5 + -0.525 + + 0.3 + z 1.010d-02 z = 6.000d-03 + alpha - element enhancement + x 0.742900 + y 0.247000 + alpha(mlt ) 1.89 + + age npts nms + 8.0 177 53 + v - i x = -0.5 x = + 0.5 x = + 1.5 + 1 1.661 6.15467 5.00000 6.31226 5.00000 6.44939 5.00000 + 2 1.650 5.48941 4.98464 5.62727 4.97775 5.74506 4.96919 + 3 1.630 5.16197 4.95595 5.28373 4.93716 5.38481 4.91399 + 4 1.610 4.96835 4.94176 5.08033 4.91742 5.17185 4.88751 + 5 1.590 4.85784 4.93242 4.96383 4.90460 5.04942 4.87047 + . . .
. . .
. .
+ . .
. . . . .
. + . .
. . . . .
. + 52 0.650 5.54940 4.26728 5.42737 4.12937 5.28500 3.97155 + 53 0.640 6.23807 4.05760 6.10449 3.90953 5.95048 3.74122 + 54 0.640 6.23139 4.01804 6.09502 3.86857 5.93823 3.69882 + 55 0.650 5.20946 3.97522 5.06719 3.82436 4.90450 3.65320 + . .
. . . . . .
+ . .
. . . . .
. + . .
. . . . .
. + 176 3.070 1.11244 -0.39442 0.94847 -0.55839 0.76407 -0.74279 + 177 3.085 1.12639 -0.84117 0.96242 -1.00514 0.77801 -1.18955 + + age npts nms + 10.0 179 50 + v - i x = -0.5 x = + 0.5 x = + 1.5 + 1 1.649 5.78583 5.00000 5.92022 5.00000 6.03681 5.00000 + 2 1.630 5.24754 4.95007 5.35566 4.93044 5.44625 4.90657 + 3 1.610 5.06810 4.93249 5.16642 4.90668 5.24674 4.87538 + 4 1.590 4.88341 4.92045 4.97514 4.89059 5.04888 4.85445 + 5 1.570 4.81476 4.91240 4.90146 4.87991 4.97015 4.84065 + . . .
. . . .
. + . .
. . . . .
. + . .
. . . . . .
+ . .
. . .
. .
. + + cccccccccccc @xmath0 & 0.00 & @xmath215 & @xmath216 & @xmath217 & @xmath218 & @xmath219 & 0.00 & @xmath220 & @xmath221 & @xmath222 & @xmath223 & 0.30 & @xmath224 & @xmath225 & @xmath226 & @xmath223 & & 0.30 & @xmath227 & @xmath228 & @xmath229 & @xmath230 & 0.60 & @xmath222 & @xmath223 & @xmath231 & @xmath232 & & 0.60 & @xmath233 & @xmath234 & @xmath235 & @xmath236 @xmath237 & 0.00 & @xmath238 & @xmath222 & @xmath239 & @xmath240 & @xmath241 & 0.00 & @xmath222 & @xmath242 & @xmath243 & @xmath244 & 0.30 & @xmath245 & @xmath229 & @xmath232 & @xmath104 & & 0.30 & @xmath243 & @xmath246 & @xmath236 & @xmath247 & 0.60 & @xmath248 & @xmath249 & @xmath250 & @xmath251 & & 0.60 & @xmath252 & @xmath253 & @xmath254 & @xmath255 @xmath131 & 0.00 & @xmath256 & @xmath232 & @xmath235 & @xmath257 & @xmath258 & 0.00 & @xmath259 & @xmath235 & @xmath252 & @xmath260 & 0.30 & @xmath104 & @xmath261 & @xmath260 & @xmath262 & & 0.30 & @xmath263 & @xmath251 & @xmath264 & @xmath255 & 0.60 & @xmath253 & @xmath265 & @xmath266 & @xmath267 & & 0.60 & @xmath268 & @xmath269 & @xmath270 & @xmath271 @xmath272 & 0.00 & @xmath246 & @xmath261 & @xmath273 & @xmath262 & @xmath274 & 0.00 & @xmath275 & @xmath276 & @xmath265 & @xmath277 & 0.30 & @xmath260 & @xmath278 & @xmath268 & @xmath266 & & 0.30 & @xmath279 & @xmath280 & @xmath267 & @xmath281 & 0.60 & @xmath266 & @xmath270 & @xmath282 & @xmath283 & & 0.60 & @xmath284 & @xmath285 & @xmath286 & @xmath287 @xmath288 & 0.00 & @xmath251 & @xmath265 & @xmath266 & @xmath289 & @xmath108 & 0.00 & @xmath277 & @xmath289 & @xmath282 & @xmath290 & 0.30 & @xmath269 & @xmath284 & @xmath291 & @xmath290 & & 0.30 & @xmath285 & @xmath292 & @xmath287 & @xmath293 & 0.60 & @xmath283 & @xmath294 & @xmath295 & @xmath296 & & 0.60 & @xmath297 & @xmath298 & @xmath299 & @xmath300 @xmath301 & 0.00 & @xmath284 & @xmath285 & @xmath286 & @xmath287 & @xmath302 & 0.00 & @xmath303 & @xmath297 & @xmath304 & @xmath305 & 0.30 & @xmath306 & @xmath307 & @xmath304 & @xmath308 & & 0.30 & @xmath309 & @xmath310 & @xmath311 & @xmath312 & 0.60 & @xmath305 & @xmath313 & @xmath312 & @xmath314 & & 0.60 & @xmath315 & @xmath316 & @xmath317 & @xmath318 @xmath319 & 0.00 & @xmath320 & @xmath321 & @xmath310 & @xmath311 & @xmath322 & 0.00 & @xmath305 & @xmath313 & @xmath323 & @xmath324 & 0.30 & @xmath325 & @xmath323 & @xmath326 & @xmath327 & & 0.30 & @xmath328 & @xmath316 & @xmath329 & @xmath330 & 0.60 & @xmath331 & @xmath318 & @xmath332 & @xmath333 & & 0.60 & @xmath318 & @xmath334 & @xmath335 & @xmath336 @xmath337 & 0.00 & @xmath338 & @xmath312 & @xmath326 & @xmath327 & @xmath339 & 0.00 & @xmath340 & @xmath341 & @xmath342 & @xmath343 & 0.30 & @xmath344 & @xmath345 & @xmath346 & @xmath347 & & 0.30 & @xmath318 & @xmath347 & @xmath348 & @xmath349 & 0.60 & @xmath332 & @xmath348 & @xmath350 & @xmath351 & & 0.60 & @xmath352 & @xmath353 & @xmath354 & @xmath355 @xmath226 & 0.00 & @xmath356 & @xmath318 & @xmath357 & @xmath358 & & & & & & & 0.30 & @xmath347 & @xmath348 & @xmath349 & @xmath359 & & & & & & & 0.60 & @xmath353 & @xmath360 & @xmath361 & @xmath362 & & & & & & ccccccc @xmath0 & 0.30 & 14.90 & 12 & @xmath228 & @xmath225 & @xmath260 & & 14.70 & 14 & @xmath246 & @xmath226 & @xmath265 & & 14.57 & 16 & @xmath275 & @xmath223 & @xmath277 @xmath219 & 0.30 & 14.85 & 12 & @xmath363 & @xmath228 & @xmath261 & & 14.67 & 14 & @xmath104 & @xmath229 & @xmath253 & & 14.54 & 16 & @xmath364 & @xmath230 & @xmath264 | an isochrone population function ( ipf ) gives the relative distribution of stars along an isochrone .
ipfs contain the information needed to calculate both luminosity functions and color functions , and they provide a straightforward way of generating synthetic stellar populations .
an improved algorithm for interpolating isochrones and isochrone population functions , based on the scheme introduced by bergbusch & vandenberg ( 1992 , , 81 , 163 ) , is described .
software has been developed to permit such interpolations for any age encompassed by an input grid of stellar evolutionary tracks .
our first application of this software is to the models presented in this series of papers for 17 [ fe / h ] values between @xmath0 and and @xmath1 , with three choices of [ @xmath2/fe ] at each iron abundance ( specifically , 0.0 , 0.3 , and 0.6 ) .
[ these models do not treat gravitational settling or radiative acceleration processes , but otherwise they are based on up - to - date physics .
additional grids will be added to this data base as they are completed . ] the computer programs ( written in fortran 77 ) and the grids of evolutionary tracks which are presently available for processing by these codes into isochrones and ipfs are freely available to interested users .
in addition , we add to the evidence presented in previous papers in this series in support of the @xmath3 and color scales of our models .
in particular , the temperatures derived by gratton et al .
( 1996 , a&a , 314 , 191 ) for local population ii subdwarfs with accurate ( _ hipparcos _ ) parallaxes are shown to be in excellent agreement with those predicted for them , when the gratton et al .
[ fe / h ] scale is also assumed .
interestingly , the locus defined by local subdwarfs and subgiants on the @xmath4 plane and the morphologies of globular cluster c - m diagrams are well matched by the present models , despite the neglect of diffusion which suggests that some other process(es ) must be at play to limit the expected effects of gravitational settling on predicted temperatures .
the three field halo subgiants in our sample all appear to have ages @xmath5 gyr , which is favored for the galaxy s most metal - poor globular clusters ( gcs ) as well .
( the settling of helium and heavy elements in the central regions of stars is expected to cause about a 10% reduction in these age estimates : this effect should persist even if some process , such as turbulence at the base of the convective envelope , counteracts diffusion in the surface layers . ) furthermore , our isochrones accurately reproduce the da costa & armandroff ( 1990 , aj , 100 , 162 ) red - giant branch fiducials for m@xmath615 , ngc@xmath66752 , ngc@xmath61851 , and 47 tuc on the [ @xmath7-diagram .
however , our models fail to predict the observed luminosities of the red - giant bump by @xmath8 mag : this could be an indication that there is some amount of inward overshooting of convective envelopes in red giants . for consistency reasons , the zinn & west ( 1984 , apjs , 55 , 45 ) metallicities for intermediate metal - poor gcs ( @xmath9 [ fe / h ] @xmath10 ) seem to be preferred over recent spectroscopic results ( based on the brightest cluster giants ) , suggesting that there is an inconsistency between current subdwarf and gc [ fe / h ] scales . |
on a finite state space @xmath0 we consider q - reversible ( quasi - reversible ) markov matrices , i.e. markov matrices @xmath1 with elements denoted @xmath2 , @xmath3 , such that @xmath4 if , and only if , @xmath5 , @xmath6 . and directed graph @xmath7[fig:2arcs ] ] the support of @xmath1 is the graph @xmath8 , where @xmath9 , @xmath6 , is an edge if , and only if , @xmath2 and @xmath10 are both positive .
we associate to each edge @xmath9 the two directed arcs @xmath11 and @xmath12 to get a directed graph without loops ( i.e. arcs from @xmath13 to @xmath13 ) that we denote by @xmath14 , see figure [ fig:2arcs ] .
the neighborhood of @xmath13 is @xmath15 , the degree of @xmath13 is @xmath16 , the set of arcs leaving @xmath13 is @xmath17 , the set of arcs entering @xmath13 is @xmath18 .
viceversa , given a connected graph @xmath8 , a q - reversible markov matrix @xmath1 with structure @xmath19 is such that @xmath20 if @xmath21 and @xmath4 if , and only if , @xmath5 , @xmath22 .
such a markov matrix is characterized as a mapping @xmath23 such that @xmath24 , @xmath25 .
as we assume @xmath20 , @xmath26 , we have @xmath27 .
the support of @xmath1 will be a sub - graph of @xmath19 .
we are going to need this detailed classification in the following . for the time being , we note that the set of @xmath1 s is parameterized by a product of symplexes @xmath28 , where @xmath29 denotes the ( solid ) simplex over the index set @xmath30 . a markov matrix @xmath1 on @xmath0 satisfies the _ detailed balance _
conditions if there exists @xmath31 , @xmath25 , such that @xmath32 it follows that @xmath1 is q - reversible and that @xmath33 is an invariant probability with full support .
equivalently , the markov chain @xmath34 , @xmath35 , with invariant probability @xmath36 and transition matrix @xmath37 $ ] , @xmath3 , has _
reversible _ bivariate joint distribution @xmath38 such a markov chain , and its transition matrix are called reversible .
reversible markov chains ( mcs ) are relevant in statistical physics , e.g. in the theory of entropy production , and in applied probability , e.g. the simulation method monte carlo markov chain ( mcmc ) .
the main aim of this paper is to find useful parameterizations of the reversible markov matrices of a given structural graph . in section [ sec : bgd ]
we review some basics from , @xcite , ( * ? ? ?
* ch 5 ) , , @xcite , @xcite , @xcite . in section [ sec : algebra ]
we discuss the algebraic theory prompted by the detailed balance condition .
the results pertain to the area of algebraic statistics , see e.g. , , .
previous results in the same algebraic spirit were presented in @xcite and @xcite .
we believe the results here are new ; some proofs depend on classical notions of graph theory that are reviwed in some detail because of our particular context .
the discussion and the conclusions are briefly presented in section [ sec : discussion ] .
the reversibility of the bivariate joint distribution in equation gives a second parameterization of reversible markov chains .
in fact , the stochastic process @xmath39 with state space @xmath0 is 2-reversible and 2-stationary if , and only if , equation holds .
when the process is a markov chain , the distribution depends on the bivariate distributions only . in particular
, the process is 1-stationary : by summing over @xmath40 , we have @xmath41 let @xmath42 be the set of all subset of @xmath0 of cardinality 2 .
the elements of @xmath42 are the edges of the full graph on @xmath0 .
the following parameterization of the 2-dimensional distributions has been used in : @xmath43 the number of parameters is @xmath44 ; moreover it holds @xmath45 hence @xmath46 belongs to @xmath47 , where @xmath48 denotes the ( flat ) simplex on the index set @xmath30 .
given an undirected graph @xmath8 such that @xmath49 if @xmath21 , then the vector of parameters @xmath50 belongs to the convex set @xmath51 .
we note that the vertices @xmath0 are identified with loops of the transitions because @xmath52 .
the marginal probability @xmath36 can be written using the @xmath53 parameters : @xmath54 or , in matrix form , @xmath55 where @xmath56 is the incidence matrix of the graph @xmath19 .
[ ex : running ] consider the graph @xmath8 with @xmath57 and @xmath58 , see left side of figure [ fig:2arcs ] . here
@xmath59 } ] { & \overline{12 } & \overline{23 } & \overline{34 } & \overline{14 } & \overline{24 } \cr 1 & 1 & 0 & 0 & 1 & 0\cr 2 & 1 & 1 & 0 & 0 & 1\cr 3 & 0 & 1 & 1 & 0 & 0\cr 4 & 0 & 0 & 1 & 1 & 1\cr } .\ ] ]
all @xmath36 s are obtained this way if we admit positive probability on loops : 1 .
the map @xmath60 is a surjective markov map .
the image of @xmath61 , @xmath62 , is the convex hull of the half points of each edge of the simplex @xmath63 whose vertices are connected in @xmath19 . 1 .
each probability @xmath36 is the image of @xmath64 .
2 . if @xmath65 , then @xmath66 , where @xmath67 and @xmath68 is the @xmath9-column of @xmath56 . hence , @xmath69 is the the middle point of the @xmath13- and the @xmath70-vertex of the simplex @xmath63
. item 1 of the proposition leaves open the question of the existence of an element @xmath53 such that @xmath71 for each @xmath72 .
this is discussed in the next subsection .
item 2 shows that , while all @xmath36 s can be obtained if loops are allowed ( @xmath73 ) , only a convex subset of @xmath63 is obtained if we do not allow for loops ( @xmath65 ) .
given @xmath36 , the fiber @xmath74 is contained in an affine space parallel to the subspace @xmath75 each fiber @xmath74 , @xmath72 , contains special solutions .
the solution @xmath76 is not of interest because we want @xmath77 .
if the graph has full connections , @xmath78 , there is the independence solution @xmath79 , @xmath80 .
e if @xmath81 , @xmath25 , a strictly positive solution is obtained as follows .
let @xmath16 be the degree of the vertex @xmath13 and define a transition probability by @xmath82 if @xmath22 , @xmath83 , and @xmath84 otherwise .
@xmath85 is the transition matrix of a random walk on the graph @xmath19 , stopped with probability 1/2 .
define a probability on @xmath86 with @xmath87 .
if @xmath88 , @xmath89 , we are done : we have found a 2-reversible probability with marginal @xmath36 and such that @xmath90 if , and only if @xmath91 . otherwise , if @xmath92 for some @xmath89 , we turn to the following hastings - metropolis construction .
[ prop : m - h ] let @xmath93 be a probability on @xmath86 such that @xmath90 if , and only if , @xmath91 or @xmath94 . write @xmath95 . given @xmath960,1[\times]0,1 [ \to ] 0,1[$ ] a symmetric function
such that @xmath97 , then @xmath98 is a 2-reversible probability on @xmath86 such that @xmath99 and @xmath100 if , and only if , @xmath91 . for @xmath91
we have @xmath101 , otherwise zero .
as @xmath102 , @xmath103 , it follows @xmath104 we have @xmath105 by construction and , in particular , @xmath106 is a probability on @xmath86 . 1
. the proposition applies to 1 .
this is the standard hastings choice .
2 . @xmath108 .
this was suggested by barker .
3 . @xmath109 .
in fact , as @xmath110 , we have @xmath111 .
2 . given a joint probability @xmath1 , the corresponding parameters @xmath112 are strictly positive for @xmath113 and @xmath114 , otherwise zero .
we have shown the existence of a mapping from @xmath36 in the interior of @xmath63 to a vector of parameters @xmath53 in the interior of @xmath51 .
an q - reversible markov matrix @xmath1 supported on a graph @xmath19 is parameterised by its non - zero extradiagonal values @xmath2 , i.e. by the elements of @xmath115 , where @xmath116 denotes the open solid simplex .
as @xmath19 is connected , the invariant probability of the markov matrix @xmath1 is unique , therefore the joint 2-distribution is uniquely defined .
if moreover the markov matrix is reversible , the joint 2-distribution is symmetric and the @xmath53 parameters are computed .
viceversa , given the @xmath53 s , the transition matrix is given by @xmath117 the mapping @xmath118 is a rational mapping and the number of degrees of freedom is @xmath119 . denoting @xmath120 by @xmath121 , from ,
the detailed balance conditions follow @xmath122 and @xmath123 . the detailed balance condition @xmath124 is equivalent to @xmath1 being adjoint as a linear operator on @xmath125 @xmath126 where @xmath127 . as @xmath125 is isomorphic to the canonical euclidian space @xmath128 via the linear mapping @xmath129
the markov matrix @xmath1 is mapped to the symmetric matrix @xmath130 this implies that a reversible markov matrix is diagonalizable ; in particular the left eigenvector @xmath36 and the right eigenvector 1 of @xmath1 are both mapped to the eigenvector @xmath131 of @xmath132 .
each element of the symmetric matrix @xmath132 is positive if , and only if , the corresponding element of @xmath1 is positive .
each reversible markov matrix @xmath1 with invariant probability @xmath36 is parameterized by a unique symmetric matrix @xmath132 .
viceversa , let @xmath133 be defined for @xmath22 .
extend @xmath134 to all couples @xmath103 , @xmath21 by @xmath135 .
if @xmath136 we can define @xmath137 to get a symmetric non - negative matrix @xmath138 $ ] , @xmath89 .
the matrix @xmath139 has non - negative entries ; is a markov matrix because @xmath140 satisfies the detailed balance equations @xmath141 we can rephrase the computations above as follows .
[ prop : squarerootpi ] the set of markov matrices which have structure @xmath19 and are reversible with invariant probability @xmath36 is parameterized by @xmath142 the polytope of all positive weight functions @xmath143 and ( unnormalized ) positive probability @xmath36 satisfying the inequalities .
let @xmath8 be a connected graph . for each closed path @xmath144 ,
that is a path on the graph such that the last vertex coincides with the first one , @xmath145 , we denote by @xmath146 the reversed path @xmath147 .
the kolmogorov s characterization of reversibility based on closed paths is well known .
however , we give here a variation of the proof by @xcite as it is an introduction to the algebraic arguments in the next section .
the proof has been modified to allow null transitions .
if @xmath148 is any path connecting @xmath149 to @xmath13 we write @xmath150 to denote the product of transitions along @xmath151 , i.e. @xmath152 . the markov irreducible matrix @xmath1 is reversible if , and only if , for all closed path @xmath144 @xmath153 assume that the process is reversible . by multiplying together all detailed balance equations @xmath154 and clearing the @xmath155 s we obtain .
viceversa , assume that all closed path have property .
fix once for all a vertex @xmath149 and consider a generic path @xmath151 from @xmath149 to @xmath13 .
first we prove that there exists a positive constant @xmath121 , depending only on @xmath13 , such that @xmath156 .
in fact , for any other path @xmath157 with the same endpoints @xmath149 and @xmath13 , @xmath158 is a closed path .
denoting by @xmath159 the corresponding constant , kolmogorov s condition implies @xmath160 .
moreover , for any vertex @xmath70 connected with @xmath13 , consider the path @xmath161 and the corresponding constant @xmath162 .
we have : @xmath163 i.e. the detailed balance condition on @xmath70 and @xmath13 . in the next section
we will discuss the algebraic interpretation of kolmogorov condition .
the present section is devoted to the algebraic structure implied by the kolmogorov s theorem for reversible markov chains .
we refer mainly to the textbooks @xcite and @xcite for graph theory , and to the textbooks and for computational commutative algebra . the theory of toric ideals is treated in detail in @xcite and .
general references for algebraic methods in stochastics are e.g. , .
the relevance of graver bases , see @xcite , has been pointed out to us by shmuel onn has in view of the applications discussed in and @xcite .
we denote by @xmath8 an undirected graph .
we split each edge into two opposite arcs to get a connected directed graph ( without loops ) denoted by @xmath14 .
the arc going from vertex @xmath13 to vertex @xmath70 is denoted by @xmath11 or @xmath164 .
the graph @xmath7 is such that @xmath165 and @xmath166 if , and only if , @xmath167 . because of our application to markov chains
, we want two arcs on each edge , as it was explained in figure [ fig:2arcs ] .
the _ reversed _ arc is the image of the 1-to-1 function @xmath168 defined by @xmath169 .
a _ path _ is a sequence of vertices @xmath170 such that @xmath171 , @xmath172 .
the reversed path is denoted by @xmath173 .
equivalently , a path is a sequence of inter - connected arcs @xmath174 , @xmath175 , and @xmath176
. a _ closed path _
@xmath177 is any path going from a vertex @xmath149 to itself ; @xmath178 is the reversed closed path . in a closed path
any vertex can be the initial and final vertex .
if we do not distinguish any initial vertex , the equivalence class of paths is called a _ circuit_. a closed path is _ elementary _ if it has no proper sub - closed - path , i.e. if does not meet twice the same vertex except the initial one @xmath149 .
the circuit of an elementary closed path is a _
cycle_. we denote by @xmath179 the set of cycles of @xmath7 .
consider the commutative indeterminates @xmath180 $ ] , @xmath181 , and the polynomial ring @xmath182 $ ] , i.e. the set of all polynomials in the indeterminates @xmath1 and coefficients in the number field @xmath183 . for each path @xmath184 , @xmath185 , @xmath172
, we define the monomial term @xmath186 for each @xmath187 , let @xmath188 be the number of traversals of the arc @xmath189 by the path @xmath144 .
hence , @xmath190 note that @xmath191 is a representation of the non - commutative concatenation of arcs on the commutative product of indeterminates .
two closed paths associated to the same circuit are mapped to the same monomial term because they have the same traversal counts .
the monomial term of a cycle is square - free because no arc is traversed twice .
figure [ fig:6 ] presents the 6 cycles in the running example of figure [ fig:2arcs ] .
, @xmath192 , @xmath193 , @xmath194 , @xmath195,@xmath196.[fig:6 ] ] this list of cycles is larger than a basis of cycles in the undirected graph @xmath19 , for instance @xmath197 .
we will see below that all directed cycles are needed for the algebraic argument .
the _ kolmogorov s ideal _ or _ k - ideal _ of the graph @xmath19 is the ideal of the ring @xmath198 $ ] generated by the binomials @xmath199 , where @xmath144 is _ any circuit_. the k - variety is the @xmath183-affine variety of the k - ideal .
our main application concerns the real case @xmath200 , but the combinatorial structure of the k - ideal does not depend on the choice of a specific field . a interesting choice for computations could be the galois field @xmath201 .
let the markov matrix @xmath1 with structure @xmath19 be reversible . 1 .
the real vector @xmath2 , @xmath202 , is a point of the intersection of the variety of the k - ideal with @xmath203 , where @xmath204 2 .
let @xmath205 be the stationary markov chain with transition @xmath1 .
then the real vector of joint probabilities @xmath206 , @xmath181 , is a point in the intersection of the k - variety and the simplex @xmath207 1 .
it is the first part of the kolmogorov s theorem .
2 . let @xmath208 be a closed path .
if @xmath36 is the stationary probability , by multiplying the kolmogorov s equations by the product of the initial probabilities at each transition , we obtain @xmath209 hence @xmath210 however , in this case the kolmogorov s equations are trivially satisfied as @xmath211 .
the k - ideal has a finite basis because of the hilbert s basis theorem .
precisely , a finite basis is obtained by restricting to cycles , which are finite in number .
we underline that here we consider all the cycles , not just a generating set of cycles
. the related result by ( * ? ? ? * th .
1 ) is discussed in the next subsection .
[ prop : cyclebasis ] the k - ideal is generated by the set of binomials @xmath212 , where @xmath144 is cycle .
let @xmath213 be a closed path which is not elementary and consider the least @xmath214 such that @xmath215 for some @xmath216 .
then the sub - path @xmath217 between the @xmath218-th vertex and the @xmath183-th vertex is an elementary closed path and the residual path @xmath219 is closed and shorter than the original one .
the arcs of @xmath144 are in 1-to-1 correspondence with the arcs of @xmath217 and @xmath220 , hence @xmath221 , @xmath187 .
the procedure can be iterated and stops in a finite number of steps . hence , given any closed path @xmath144 , there exists a finite sequence of cycles @xmath222 , such that the list of arcs in @xmath144 is partitioned into the lists of arcs of the @xmath223 s . from @xmath224 , @xmath225
, it follows @xmath226 the k - ideal is generated by a finite set of binomials and this set has the same number of elements as the set of undirected cycles of @xmath19 .
if the graph is a tree , the cycles reduce to those of the type @xmath227 and the binomials are identities , hence the k - ideal is equal the the full ring .
this is an involved way to prove all stationary markov process on a tree is reversible ( * ? ? ?
* lemma 1.5 ) .
the cycle basis of proposition [ prop : cyclebasis ] belongs to the special class of bases , namely grbner bases .
we refer to the textbooks and for a detailed discussion .
we review the basic definitions of this theory , which is based on the existence of a monomial order @xmath228 , i.e. a total order on monomial terms which is compatible with the product .
given such an order , the leading term @xmath229 of the polynomial @xmath230 is defined .
a generating set is a _ grbner basis _ if the set of leading terms of the ideal is generated by the leading terms of monomials of the generating set .
a grbner basis is _ reduced _ if the coefficient of the leading term of each element of the basis is 1 and no monomial in any element of the basis is in the ideal generated by the leading terms of the other element of the basis .
the grbner basis property depends on the monomial order . however , a generating set is said to be a _
universal _ grbner basis if it is a grbner basis for all monomial orders .
the finite algorithm for testing the grbner basis property depends on the definition of _ syzygy_. the syzygy of two polynomial @xmath230 and @xmath231 is the polynomial @xmath232 a generating set of an ideal is a grbner basis if , and only if , it contains the syzygy @xmath233 whenever it contains the polynomials @xmath230 and @xmath231 , see or .
[ prop : cycles ] the cycle basis of the k - ideal is a reduced universal grbner basis . choose any monomial order @xmath228 and let @xmath217 and @xmath220 be two cycles with @xmath234 , @xmath235 .
assume first they do not have any arc in common .
in such a case @xmath236 and the syzygy is @xmath237 which belongs to the k - ideal .
let now @xmath238 be the common part , that is @xmath239 .
the syzygy of @xmath240 and @xmath241 is @xmath242 which again belongs to the k - ideal because @xmath243 is a cycle .
in fact @xmath244 and @xmath245 have in common the extreme vertices , corresponding to the extreme vertices of @xmath238 .
notice that @xmath238 is the common part of @xmath217 and @xmath220 only if it is traversed in the same direction by the both cycle .
the previous proof does not depend on the choice of the leading term of the binomials , therefore the grbner basis is universal . the grbner basis is reduced because no monomial of a cycle can divide a monomial of a different cycle .
figure [ fig:1 ] is an illustration of the proof . [ cols="^ " , ] on this example we can see why the two @xmath246 and @xmath247 do not generate the k - ideal .
in fact , from @xmath248 it follows @xmath249 which , in turn , gives the binomial of @xmath250 if @xmath251 and , therefore , the factor @xmath252 can be cleared .
this is confirmed by the use of a symbolic algebraic software such as , see @xcite .
this computation shows that the @xmath250 equation does not belong to the ideal generated by the @xmath253 equations unless we add the condition @xmath251 . notice that @xmath246 and @xmath247 are the cycles obtained from the spanning tree @xmath254 .
take any order on vertexes , e.g. @xmath255 , and derive a lexicographic order on arcs : @xmath256 .
take the same order on indeterminates @xmath257 , @xmath187 , and the lexicographic order on monomials .
we check that the leading terms of the there binomials in the g - basis are @xmath258 , see figure [ fig:6 ] .
the exponents of the leading terms of the g - basis are @xmath259 each monomial @xmath260 , @xmath261 , is reduced by the k - ideal to a monomial whose exponent does not contain any of the counts in the table .
e.g. @xmath262 is an element of the monomial basis of the polynomial ring @xmath263 the k - ideal .
we adapt to our context some standard tools of algebraic graph theory , namely the cycle an cocycle spaces , see e.g. ( * ? ? ?
* ch 2 ) and ( * ? ? ?
* ii.3 ) .
let @xmath264 be the set of cycles . for each cycle
@xmath265 we define the _ cycle vector _ of @xmath144 to be @xmath266 , where @xmath267 note that @xmath268 .
if @xmath269 and @xmath270 are the positive and the negative part of @xmath271 , respectively , then @xmath272 and @xmath273 .
it follows that @xmath274 and @xmath275 more generally , the definition can be is extended to any circuit @xmath144 by defining @xmath276 the equality @xmath277 holds if , and only if , @xmath278 implies @xmath279 , @xmath187 .
let @xmath280 be the _ cycle space _
, i.e. the vector space generated in @xmath281 by the cycle vectors . for each proper subset @xmath282 of the set of vertices , @xmath283 we define the _ cocycle vector _ of @xmath282 to be @xmath284 , with @xmath285 see an example in figure [ fig : cocycle ]
. ] note that @xmath286 .
let @xmath287 be the _ cocycle space _
, i.e. the vector space generated in @xmath281 by the cocycle vectors .
let @xmath288 be the matrix whose rows are the cocycle vectors @xmath289 , @xmath283 .
the matrix @xmath290_{\emptyset \ne b\subsetneq v , a\in { \mathcal a}}$ ] is the _
cocycle matrix_. the cycle space and the cocycle space are orthogonal in @xmath291 .
in fact , for each cycle vector @xmath292 and cocycle vector @xmath289 , we have @xmath293 so that @xmath294 = 0.\end{aligned}\ ] ] it is shown e.g. in the previous references that the cycle space is the orthogonal complement of the cocycle space for undirected graphs . in our setting it is the orthogonal complement relative to the subspace of vectors @xmath295 such that @xmath296 .
as we are interested in elements of the cycle space with integer entries , i.e. those elements @xmath297 of the cycle space that can be exponents of a monomial @xmath298 , we are going to use the following matrix encoding of our problem .
[ def : modelmatrix ] consider the matrix @xmath299_{e\in{\mathcal e } , a\in{\mathcal a}}$ ] whose element @xmath300 in position @xmath301 is 1 if the arc @xmath189 is une of the directions of the edge @xmath302 , zero otherwise .
let @xmath288 be the cocycle matrix .
the _ model martix _ is the block matrix @xmath303 it follows @xmath304 .
lattice basis _ of the lattice @xmath305 is a linear basis of @xmath306 with integer entries .
the matrix @xmath85 has dimension @xmath307 . in fact
@xmath308 can be re - arranged as @xmath309 $ ] , with @xmath310 the identity matrix , and @xmath288 has @xmath311 linearly independent rows , the dimension of the cocycle space .
@xmath312 table [ tab : a ] shows the matrix @xmath85 and a lattice basis of @xmath313 computed with .
the top matrix is the @xmath308 matrix ; the bottom matrix is the @xmath288 matrix , where three linearly independent rows are highlighted .
the two bottom row vectors are the lattice basis .
we want to show that the k - ideal is the toric ideal of the model matrix @xmath85 , see definition [ def : modelmatrix ] .
basic definitions and theory are in ( * ? ?
* ch 4 ) , see also .
consider the polynomial ring @xmath314 $ ] and the laurent polynomial ring @xmath315 $ ] , together with their homomorphism @xmath316 defined by @xmath317 as @xmath318 for all edge @xmath302 and arc @xmath189 , the first factor in is a symmetric function @xmath319 .
we could write @xmath320 the kernel @xmath321 of @xmath316 is called the _ toric ideal _ of @xmath85 , @xmath322}{f(t^{a(a ) } \colon a \in { \mathcal a})=0}.\ ] ] the toric ideal @xmath321 is a prime ideal and the binomials @xmath323 are a generating set of @xmath321 as a @xmath324-vector space .
a finite generating set of the ideal is formed by selecting a finite subset of such binomials .
the basis we find is a graver basis .
we recall the definition of graver basis as it is given in and we apply it to the _ cycle lattice _
we introduce a partial order and its set of minimal elements as follows .
let @xmath325 and @xmath326 be two element of the cycle lattice @xmath313 . 1 .
@xmath325 is _ conformal _ to @xmath326 , @xmath327 , if the component - wise product is non - negative ( i.e. @xmath325 and @xmath326 are in the same quadrant ) and @xmath328 component - wise , i.e. @xmath329 and @xmath330 for all @xmath331 .
a _ graver basis _ of @xmath313 is the set of the minimal elements with respect to the conformity partial order @xmath332 .
[ pr : graver ] 1 .
[ item : graver1 ] for each cycle vector @xmath333 , @xmath334 , @xmath335 , there exist cycles @xmath336 and positive integers @xmath337 , such that @xmath338 , @xmath339 , @xmath340 , and @xmath341 2 .
the set @xmath342 is a _
graver basis _ of @xmath343 . 1 . for all
@xmath344 we have @xmath345 , so that we can assume all the @xmath346 s to be non - negative .
notice also that we can arrange things in such a way that at most one of the two direction of each cycle has a non - zero coefficient .
we define @xmath347 and consider two subgraph of @xmath7 with a the set of arcs restricted to @xmath348 , @xmath349 , respectively .
we note that @xmath350 and @xmath351 ; in particular , both @xmath352 , @xmath353 are not empty .
we drop from now on the dependence on @xmath271 for ease of notation . + we show first there is a cycle whose arcs are in @xmath348 . if not ,
if a cycle of full graph @xmath7 has one arc in @xmath348 , it would exists vertex @xmath13 such that @xmath354 while @xmath355 .
let @xmath356 be the cocycle vector of @xmath357 ; we derive a contradiction to the assumption @xmath358 .
in fact , @xmath359 because each of the terms @xmath360 , @xmath361 , is either 0 or equal to @xmath362 if @xmath363 .
+ let @xmath217 be a cycle in @xmath348 and define an integer @xmath364 such that @xmath365 and it is zero for at least one @xmath189 .
the vector @xmath366 belongs to the cycle space @xmath313 , and moreover @xmath367 .
+ by repeating the same step a finite number of times we obtain a new representation of @xmath271 in the form @xmath368 where the support of each @xmath369 is contained in @xmath348 .
it follows @xmath370 2 . in the previous decomposition
each @xmath371 , @xmath340 , is conformal to @xmath271 .
in fact , from @xmath338 and @xmath372 , it follows @xmath373 and @xmath374
. therefore @xmath375 .
we give an illustration of the previous proof .
consider the cycle vectors @xmath376 and the element of the cycle space @xmath377 , see figure [ fig : zeta - omega ] .
, width=340 ] we have @xmath378 as it is illustrated in figure [ fig : zeta - omega2 ] . of figure
[ fig : zeta - omega][fig : zeta - omega2],width=340 ] [ pr : toric ] 1 .
the k - ideal is the toric ideal of the matrix @xmath85 .
2 . the binomials of the cycles form a graver basis of the k - ideal . 1 . for each cycle
@xmath144 the cycle vector @xmath292 belongs to @xmath313 . from equation , @xmath379 ,
therefore the k - ideal is contained in the toric ideal @xmath321 because of proposition [ prop : cyclebasis ] .
+ to prove the equality we must show that each binomial in @xmath321 belongs to the k - ideal . from proposition [ pr : graver].[item : graver1 ] , it follows that @xmath380 belongs to the k - ideal .
the graver basis of a toric ideal is the set of binomials whose exponents are the positive and negative parts of a graver basis of @xmath313 . from propositions [ pr : graver ] and the previous item the proof follows .
the knowledge that the k - ideal is toric is relevant , because the homomorphism definition in equation provides a parametric representation of the the variety .
in particular , the strictly positive @xmath257 , @xmath331 , are given by : @xmath381 we observe that the first set of parameters , @xmath382 , is a function of the edge , while the second set of parameters , @xmath383 , represents the deviation from symmetry .
in fact , as @xmath384 we have @xmath385 , so that @xmath386 if , and only if , @xmath387 as the rows of @xmath308 are linearly independent , the @xmath382 s parameters carry @xmath388 degrees of freedom to represent a generic symmetric matrix .
the second set of parameters is not identifiable because the rows of the @xmath288 matrix are not linearly independent .
the parameterization can be used to derive an explicit form of the invariant probability .
all properties of the parameterization are collected in the following proposition .
[ th : parameters ] consider the strictly non - zero points on the k - variety . 1 .
the symmetric parameters @xmath389 , @xmath390 , are uniquely determined in equation .
the parameters @xmath391 , @xmath392 are confounded by @xmath393 2 .
[ item : param ] an identifiable parameterization is obtained by taking a subset of parameters corresponding to linearly independent rows , denoted by @xmath391 , @xmath394 : @xmath395 3 .
[ item : param1 ] the detailed balance equations , @xmath396 , are verified by @xmath397 1 .
we have @xmath398 for @xmath399 , @xmath400 , @xmath401 . if @xmath402 , then @xmath403 because the rows of @xmath308 are orthogonal to the rows of @xmath288 .
hence , @xmath404 because @xmath308 has full rank .
finally , @xmath405 .
the sub - matrix of @xmath85 formed by @xmath308 and by the rows of @xmath288 in @xmath406 has full rank .
3 . using equations
, we have : @xmath407 which is equivalent to @xmath408 by multiplying both terms in the equality by @xmath409 , we obtain @xmath410 so that @xmath411 depends only on @xmath13 and satisfy the detailed balance condition .
we are now in the position of stating an algebraic version of kolmogorov s theorem .
the _ detailed balance ideal _ is the ideal of the ring @xmath412\ ] ] generated by the polynomials @xmath413 the first polynomial in the list of generators , @xmath414 , assures that the @xmath155 s are not zero . 1 . a point @xmath415_{v\to w \in { \mathcal a}}$ ] with non - zero components belongs to the variety of the k - ideal if , and only if , there exists @xmath416 such that @xmath417 belongs to the variety of the detailed balance ideal .
the detailed balance ideal is a toric ideal .
the k - ideal is the @xmath155-elimination ideal of the detailed balance ideal . 1 .
one direction is the kolmogorov s theorem .
the other direction is a rephrasing of item [ item : param1 ] of theorem [ th : parameters ] .
this ideal is the kernel of the homomorphism defined by , i.e. @xmath418 together with @xmath419 .
the elimination ideal is generated by dropping the parametric equations of the indeterminates to be eliminated .
the parameterization in theorem [ th : parameters].[item : param ] is to be compared to that in proposition [ prop : squarerootpi ] .
both split the parameter space into a representation of the invariant probability and a generic symmetric function .
the latter is a special case of the former .
in fact , it is obtained by the use of the basis of the cocycle space where each @xmath282 is the set containing one vertex .
the parameterization of reversible markov matrices is obtained by adding to the representation in equation the relevant inequalities .
let us check first the degrees of freedom .
as a reversible markov matrix supported by a connected graph @xmath19 is parameterized by the joint 2-distributions of the stationary markov chain , the number of degrees of freedom is @xmath119 , i.e. the cyclotomic number of the graph @xmath19 . in the parameterization of proposition [ prop : squarerootpi ] the probability @xmath36 carries @xmath311 degrees of freedom , therefore @xmath134 must carry @xmath420 degrees of freedom .
par - k ] let @xmath1 be a matrix with supporting graph @xmath8 .
let @xmath406 be a family of subsets of @xmath0 such that the cocycle vectors @xmath421 , @xmath422 , span the coclycle space . 1 .
@xmath1 is a reversible markov matrix if , and only if , there exists a non - negative symmetric function @xmath423 which is supported on the edges @xmath424 and there exist positive parameters @xmath425 , @xmath426 , such that @xmath427 and the invariant probability is proportional to @xmath428 .
2 . for @xmath25 ,
@xmath429 3 .
the parameters @xmath389 , @xmath390 and @xmath121 , @xmath25 , are identifiable , while the parameters @xmath383 , @xmath426 , are identifiable if @xmath421 , @xmath422 , is a basis of the cocycle space . 1
. follows from and and @xmath430 2 .
it follows from equation as @xmath1 is a transition probability .
3 . assume there exists two set of parameters @xmath431 , @xmath432 , @xmath235 giving the same @xmath1 , and define @xmath433 , @xmath434 , @xmath435 .
it follows @xmath436 hence @xmath437 . in turn
we get @xmath438 the identifiability of the @xmath383 s follows from @xmath439 an over - parameterization of two transition probabilities of the @xmath440-variety is : @xmath441 by choosing the cocycle basis @xmath442 , we have : @xmath443 the transition matrix parameterized by @xmath389 , @xmath444 and @xmath383 , @xmath445 is @xmath446 } ] { & 1 & 2 & 3 & 4\cr 1 & \star & s(1,2)\ t_{\{1\}}^{-1 } & 0 & s(1,4)\ t_{\{1\}}^{-1 } \
t_{\{1,2\}}^{-1}\cr 2&s(1,2)\ t_{\{1\ } } & \star & s(2,3)\ t_{\{1,2\}}^{-1}\ t_{\{3\ } } & s(2,4)\ t_{\{1,2\}}^{-1}\cr 3 & 0 & s(2,3)\ t_{\{3\}}^{-1}\ t_{\{1,2\ } } & \star & s(3,4 ) \ t_{\{3\}}^{-1 } \cr 4 & s(1,4 ) \
t_{\{1\}}\ t_{\{1,2\ } } & s(2,4 ) \ t_{\{1,2\ } } & s(3,4 ) \ t_{\{3\ } } & \star \cr } .\ ] ] where the diagonal terms are uniquely defined if the following set of inequalities is true @xmath447
the algebraic analysis of statistical models of the type @xmath448 where @xmath85 is an integer matrix has been first introduced in , where an implicit binomial form of the model is derived from the monomial form . our analysis differs from that in two respect .
first , we move backwards from the binomial form represented by the kolmogorov s conditions to the monomial form .
second , we parameterize the transition probabilities , so that the normalization requires more than one constant .
our parameterization of a reversible markov matrix is based on a generic weight function @xmath389 on the edges @xmath390 of the structure graph @xmath19 and a monomial form of the invariant probability .
when the basis of the cocycle space is given by the vertexes of the graph , the parameterization is identical to that the classical form @xmath449 .
the monomial form of the unnormalized invariant probability @xmath450 suggests the use of a family of sets @xmath451 smaller than a cocycle basis @xmath406 in order to get a parsimonious statistical model .
for example , if the graph @xmath19 is a square @xmath452 grid , a coarse - grained model could use @xmath453 sub - grids , @xmath454 .
one distinct advantage of the implicit binomial form is its ability to fully describe the closure of its strictly positive part , i.e. the extended exponential family .
the computation of an hilbert basis of the non - negative integer kernel of the cocycle matrix @xmath288 leads to a parameterization of the extended exponential family as in , see also .
however , in this case the border of the model appears to consist simply on the deletion of edges in the support graph .
it follows from general properties of toric ideals that a graver basis is a universal grbner basis and that a universal grbner basis is a markov basis , @xcite .
the markov basis property is related with the connectedness of random walks on the fibers of @xmath85 , see and subsequent literature on mcmc simulation . in this case
it would be a simulation of a random reversible markov matrix . during this research
we have discussed various items with s. sullivan , b. sturmfels , g. casnati and we want to thank them for providing us with supporting advice .
preliminary version of this paper have been presented at the workshop wogas2 , warwick university 2010 and the conference crest - sbm 2 , osaka 2010 .
we wish to thank g. letac for bringing reference @xcite to our attention and f. rigat for the reference @xcite .
also they thank s. onn for pointing out the relevance of his own work on graver bases . after the publication of the first draft on the arxiv , y. a. mitrophanov provided us with the reference to his own work on the subject , we where not aware of and that with gratefully acknowledge . | for a markov chain both the detailed balance condition and the cycle kolmogorov condition are algebraic binomials . this remark suggests to study reversible markov chains with the tool of algebraic statistics , such as toric statistical models .
one of the results of this study in an algebraic parameterization of reversible markov transitions and their invariant probability . |
randomness and computation are related to two basic elements of bounded rationality modeling : ( a ) players can not generate truly random actions ; ( b ) they can not implement arbitrarily complex strategies .
when rationality is computationally constrained , true randomness can be replaced by pseudo - randomness : one need not generate truly random actions , just actions whose pattern is too complicated for any rationally bounded agent to be able to recognize .
algorithmic randomness is the field of computer science that studies the relations between computability and randomness .
it provides the rigorous definitions and analytic tools to address questions involving the tension between the two above elements of bounded rationality . on the other hand
, the present paper applies game theory to solve a problem from algorithmic randomness by recasting the problem as an extensive form game .
gambler 0 , the cousin of the casino owner , repeatedly bets on red / black .
she is only allowed to wager even ( positive ) integers , i.e. , she can bet 4 dollars on black , then 10 on red , then 2 on black , etc .
( `` betting 4 on black '' means that you gain 4 if black occurred , and lose 4 if red did ) .
then the _ regular _ gamblers make their bets , and they are only allowed to wager odd integers .
the red / black outcome is not decided by a lottery ; rather , the casino owner chooses red or black .
he attempts to aid gambler 0 by his choices .
all gamblers start with some finite , not necessary equal , wealth . going into debt
is not allowed , i.e. , you can not bet more than you have . say that a gambler `` succeeds '' if , along the infinite stream of bets , her wealth tends to infinity .
the `` home team '' , consisting of the casino owner and gambler 0 , wins if and only if gambler 0 succeeds while the other gamblers do not . in the case
depicted above , gambler 0 uses even integers as wagers , and the others use odd wagers .
we will see that the home team can guarantee a win in this case and that they could not had it been the other way round , namely , odd integers for gambler 0 and even integers for the others . in general , let @xmath0 be the wagers allowed to gambler 0 , and @xmath1 those allowed to the other gamblers ( @xmath4 being any pair of subsets of the positive real numbers ) .
we ask for which @xmath0 and @xmath1 the home team can _ guarantee a win _ , in the following sense : gambler 0 can choose a strategy that guarantees that for any strategies that the other gamblers choose the casino can choose a sequence so that the home team wins . as a simple example ,
suppose that apart from gambler 0 there is only one more , regular , gambler ( call him gambler 1 ) , and first , let @xmath5 and @xmath6 .
then the home team can not guarantee a win : for any pure strategy of gambler 0 , let gambler 1 employ the strategy that is a copy of it , only wagering 1 instead of 2 .
then the gains ( or losses ) of 1 are exactly half the gains of 0 .
therefore , if gambler 0 succeeds , so does gambler 1 .
second , let @xmath1 be as before , and let @xmath7 . then the home team can guarantee a win as follows . the strategy of gambler 0 is simple : she always bets on red ; at the start she wagers two dollars , and keeps doing this as long as she wins .
then , after the first loss , she switches to wagering one dollar ad infinitum .
the casino strategy chooses , in a first phase , red every time , until the wealth of 0 exceeds 1 s wealth by more than three dollars ( note that this is bound to eventually happen , since in the first phase 0 gains two every time , while 1 gains at most one ) .
then it chooses black once ( after which the wealth of 0 still exceeds the wealth of 1 , as 0 lost two dollars and 1 gained no more than one dollar ) . then comes the second phase , where the casino always chooses the opposite color to what gambler 1 bet on , and chooses red if 1 did not bet anything .
this strategy of the casino ensures that , during the second phase , 1 can not make a non - zero bet more than a finite number of times before he goes broke ( and if 1 does go broke , 0 still does not ) .
hence , whatever strategy he chooses , 1 will not succeed , while 0 will .
this casino game is related to the notions of randomness and computability . in the theory of _ algorithmic randomness _ , a sequence of zeros and ones is called _ computably random _ if there exists no computable strategy that succeeds against it ( thus , the sequence is in some sense unpredictable ) . for a set @xmath0 of positive real numbers ,
a sequence is _
@xmath0-valued random _ if there exists no computable gambling strategy ( aka `` martingale '' ) , with wagers in @xmath0 , that succeeds against it .
hence we may compare two sets of reals @xmath0 and @xmath1 , and ask whether @xmath1-valued randomness implies @xmath0-valued randomness . in other words , whether for any casino sequence @xmath8 , and a computable @xmath0-martingale ( gambling strategy with wagers in @xmath0 ) that succeeds on @xmath8 , there exists a @xmath1-martingale that succeeds on @xmath8 . and in terms of our casino game
, suppose that _ every _ computable @xmath1-martingale is employed by some regular gambler .
then the question is whether there exist a computable @xmath0-martingale and a casino sequence , such that the home team wins .
there are countably many computable strategies . in many cases
this countability , rather than the specific type of admissible strategies , is the essential point in the analysis .
this naturally leads to the following setting ( definition [ def countable anticipation ] ) : gambler 0 s wagers are in @xmath0 , and there are countably many regular players , whose wagers are in @xmath1 .
we will say that @xmath0 _ evades _
@xmath1 if the home team can guarantee a win .
this is the central notion in our paper .
thus , we are not directly concerned with computability , or other complexity considerations and @xmath1 are recursive sets of rational numbers our analysis applies to the question whether every @xmath1-valued random is @xmath0-valued random . ] , but rather take a more abstract view : we only require that the number of strategies against which the home team needs to concurrently combat is countable ( see also remark [ rem : effective ] below ) .
one may consider a few variants of this settings ( see below ) .
if @xmath0 does not evade @xmath1 , we say that @xmath1 _ anticipates _ @xmath0 . if @xmath1 contains @xmath0 , then clearly @xmath1 anticipates @xmath0 , because the strategy of gambler 1 can be an exact copy of 0 s strategy ( and there is even no need for countably many regular gamblers , one is enough ) . therefore
, if 0 succeeds , so does 1 .
similarly , if @xmath1 contains a multiple of @xmath0 , i.e. , @xmath9 for some @xmath10 , then @xmath1 anticipates @xmath0 : as in the example above , the @xmath1-strategy can be the same as the @xmath0-strategy up to the multiple @xmath3 ( i.e. , the wagers are multiplied by @xmath3 ) , and therefore the gains are the same , multiplied by @xmath3 .
for example , the even integers anticipate the odd integers ( or the whole set of integers , for that matter ) .
thus , @xmath1 containing a multiple of @xmath0 is a sufficient condition for @xmath1 to anticipate @xmath0 .
is it also a necessary condition ?
@xcite showed that if @xmath0 and @xmath1 are finite , then it is necessary .
they asked whether this characterization extends to infinite @xmath0 and @xmath1 ( note that their framework is that of algorithmic randomness , i.e. , it only allows for computable martingales ) .
in other words : * does @xmath1 not containing a multiple of @xmath0 imply that @xmath0 evades @xmath1 ?
theorem [ thm : suffic ] implies a negative answer to this question . ) .
] thus , for example , the segment @xmath11 $ ] anticipates the set @xmath12 , although it does not contain any multiple of it .
still , although @xmath13 does not hold in general , we will see classes of pairs @xmath4 where @xmath13 does hold .
theorem [ theorem bounded ] says that if @xmath0 is bounded , and @xmath14 is bounded away from 0 , then @xmath13 holds .
thus , for example , the set @xmath15 evades the positive integers .
theorem [ theorem well ordered ] says that if @xmath1 is well - ordered , then @xmath13 holds .
thus , for example , the set of even integers evades the odds .
now let us consider the following version of our casino game , whose description is more like that of a `` classic '' repeated game .
at each stage of the repeated game , first gambler 0 makes a bet ; then , after observing it , the ( team of ) regular gamblers each make their bets ; and then the casino chooses a red / black outcome .
all of our results apply to this version as well .
the difference between our definition of evasion and the description of this game is that in the former setting , the regular gamblers knew the _ strategy _ of gambler 0 ( i.e. , including her future actions ) , gambler 0 did not observe their past actions , and the casino knew the strategies of all gamblers
. one may consider many such variations to the settings , and our results are robust to all of them whether gambler 0 observes anything or not , whether the regular gamblers know her future bets or not , and whether the casino knows the future bets or not .
previous work on martingales with restricted wagers ( @xcite , @xcite , @xcite ) employed various solutions in various specific situations .
the present paper proposes a systematic treatment that applies to most of the previously studied situations , as well as many others .
our proofs are elementary and constructive .
also , the proposed construction is recursive : it does not rely on what the strategies do in the future ; therefore we believe our method can be useful in many frameworks where computational or other constraints are present . the rest of the paper is organized as follows .
section [ sec : martingales ] presents the definitions , results , and a few examples , as well as a discussion of related previous work .
the next two sections contain proofs .
section [ sec : discussion ] points to open problems and further directions .
a _ martingale _ is a gambling strategy that bets on bits of a binary sequence .
formally , it is a function @xmath16 that satisfies @xmath17 for every string @xmath18 .
the _ increment _ of @xmath19 at @xmath18 is defined as @xmath20 for @xmath21 , we say that @xmath19 is an _ @xmath0-martingale _ if @xmath22 , for every @xmath18 .
the empty string is denoted @xmath23 and @xmath24 is called the _ initial value _ of @xmath19 .
note that a martingale is determined by its initial value and its increments .
the initial sub - string of length @xmath25 of a binary sequence , @xmath26 , is denoted @xmath27 .
a martingale @xmath19 _ succeeds _ on @xmath28 , if @xmath29 and @xmath30 , for every @xmath25 .
the latter condition asserts that @xmath19 does nt bet on money it does nt have .
the set of sequences on which @xmath19 succeeds is denoted @xmath31 .
a martingale @xmath32 _ dominates _
@xmath19 if @xmath33 , and a set of martingales @xmath34 dominates @xmath19 if @xmath35 . the following are non - standard definitions .
[ def single anticipation ] a set @xmath36 _ singly anticipates _ a set @xmath37 , if every @xmath0-martingale is dominated by some @xmath1-martingale .
if @xmath0 singly anticipates @xmath1 and @xmath1 singly anticipates @xmath0 , we say that @xmath0 and @xmath1 are _ strongly equivalent_. [ def countable anticipation ] a set @xmath36 _ countably anticipates _
( _ anticipates _ , for short ) a set @xmath37 , if every @xmath0-martingale is dominated by a countable set of @xmath1-martingales .
if @xmath0 anticipates @xmath1 and @xmath1 anticipates @xmath0 , we say that @xmath0 and @xmath1 are _
( weakly ) equivalent_. if @xmath1 does not anticipate @xmath0 we say that @xmath0 _ evades _ @xmath1 .
note that both `` singly anticipates '' and `` anticipates '' are reflexive and transitive relations ( namely , they are preorders ) , and that single anticipation implies anticipation . also , if @xmath38 then @xmath1 singly anticipates @xmath0 .
[ rem : effective ] the motivation for the definition of countable anticipation comes from the study of algorithmic randomness ( see , e.g. , @xcite ) , where it is natural to consider the countable set of all @xmath1-martingales that are computable relative to @xmath19 ( see @xcite ) . formally , a set @xmath36 _ effectively anticipates _ a set @xmath37 , if for every @xmath0-martingale @xmath19 and every sequence @xmath39 , there is a @xmath1-martingale , computable relative to @xmath19 , that succeeds on @xmath28 .
more generally , one could define in the same fashion a `` @xmath40-anticipation '' relation with respect to any complexity class @xmath40
. our main focus will be on countable anticipation , and specifically on sets @xmath0 and @xmath1 such that @xmath1 does not countably anticipate @xmath0 ; and therefore @xmath1 does not @xmath40-anticipate @xmath0 for any complexity class @xmath40 .
when we present cases in which anticipation does hold , the dominating martingales will usually be fairly simple relative to the dominated martingales .
we do not presume to rigorously address the computational complexity of those reductions , though .
the topological closure of a set @xmath37 is denoted @xmath41 .
the following lemma says that we can restrict our attention to closed subsets of @xmath12 .
[ lem close ] every subset of @xmath12 is strongly equivalent to its closure .
let @xmath21 and let @xmath19 be an @xmath41-martingale . define an @xmath0-martingale @xmath42 by @xmath43 where @xmath44 is the length of @xmath45
. clearly , @xmath46 , for every @xmath47 ; therefore @xmath48 .
another simple observation is that for every @xmath49 and @xmath50 , @xmath0 is strongly equivalent to @xmath51 .
this observation leads to the next definition .
let @xmath52 .
we say that @xmath0 and @xmath1 are _ proportional _ , if there exists @xmath50 such that @xmath53 .
if we only require that @xmath54 , then @xmath0 is proportional to a subset of the closure of @xmath1 . in that case
we say that @xmath0 _ scales _ into @xmath1 . from the above and the fact that @xmath38 implies that @xmath1 singly anticipates @xmath0
, we have the following lemma .
[ lem similarity ] if @xmath0 scales into @xmath1 , then @xmath1 anticipates @xmath0 , for every @xmath52 .
the next two theorems provide conditions under which the converse of lemma [ lem similarity ] also holds .
[ theorem bounded ] for every @xmath52 , if @xmath55 and @xmath56 , then @xmath1 anticipates @xmath0 only if @xmath0 scales into @xmath1 .
[ theorem well ordered ] for every @xmath52 , if @xmath1 is well ordered , namely , @xmath57}$ ] , then @xmath1 anticipates @xmath0 only if @xmath0 scales into @xmath1 .
@xcite studied effective anticipation between finite subsets of @xmath58 .
they showed that ( effective ) anticipation is equivalent to containing a proportional set on the domain of finite sets .
they further asked whether their result extends to infinite sets .
in particular , they asked if @xmath59 anticipates the set @xmath60 .
@xcite showed that the set @xmath61 evades ( i.e. , is not anticipated by ) @xmath59 , and the set @xmath62 evades @xmath63 .
all of the above results follow immediately from theorem [ theorem bounded ] .
furthermore , any set that contains two @xmath64-linearly independent numbers ( e.g. , @xmath15 ) evades @xmath59 . in light of lemma [ lem close ] , one may rephrase the above question of @xcite and ask whether , for infinite sets , anticipation is equivalent to scaling .
the answer is still negative ( theorem [ thm : suffic ] ) , although theorems [ theorem bounded ] and [ theorem well ordered ] above gave some conditions under which the equivalence does hold .
theorem [ theorem well ordered ] says that if , for example , @xmath65 and @xmath1 is a subset of @xmath59 whose density is zero , then @xmath0 evades @xmath1 .
this is because @xmath1 is well ordered , and @xmath0 does not scale into @xmath1 ( by the zero density ) .
another example is when @xmath1 is the set of all odd integers where , again , @xmath0 evades @xmath1 .
note that the set of even integers is proportional to @xmath59 and hence is equivalent to @xmath59 .
therefore the even integers evade the odd integers , but not vice versa .
furthermore , we can take any subset of @xmath59 that does not contain an `` ideal '' ( i.e. , all the integer multiples of some number , namely , a set proportional to @xmath59 ) .
if @xmath1 is of the form @xmath66 for some function @xmath67 , then @xmath59 evades @xmath1 even when , for example , the function @xmath68 grows very rapidly .
in particular , the density of @xmath1 could equal one .
the previous theorems gave some necessary conditions for anticipation .
the next theorem gives a sufficient condition .
[ thm : suffic ] let @xmath69 .
for @xmath70 , let @xmath71 ) \subseteq \bar{b}\cup { \{{0}\ } } \,\}.\ ] ] for any @xmath72 let @xmath73)$ ] .
if for some @xmath19 , @xmath74 , then @xmath1 singly anticipates @xmath0 .
this is equivalent to the following seemingly stronger theorem .
[ lemma f ] let @xmath69 .
suppose there is a non - increasing function @xmath75 , such that 1 . @xmath76\right)$ ] , for every @xmath77 ; and 2 .
@xmath78 .
then , @xmath1 singly anticipates @xmath0 .
let @xmath19 be an @xmath0-martingale .
by lemma [ lem close ] we may assume that @xmath1 is closed .
define a @xmath1-martingale , @xmath42 , by @xmath79 let @xmath80 such that @xmath81 , and let @xmath82 . since @xmath83 is non - increasing , @xmath84 it follows by induction that @xmath85 which concludes the proof , since @xmath86 and @xmath87 .
if , for example , @xmath0 scales into @xmath1 , namely , @xmath88 , the function @xmath83 in theorem [ thm : suffic]@xmath89 can be taken to be simply @xmath90 .
also note that when @xmath0 and @xmath1 are finite , such a function @xmath83 as in the theorem exists if and only if @xmath0 scales into @xmath1 .
the theorem tells us , for example , that although @xmath91 does not scale into the interval @xmath11 $ ] , these two sets are ( strongly ) equivalent : to see that @xmath11 $ ] singly anticipates @xmath12 , apply theorem [ thm : suffic]@xmath89 with @xmath92 .
another example is the set @xmath93 being ( strongly ) equivalent to @xmath94 , although @xmath0 does not scale into @xmath1 . to see this , apply theorem [ thm : suffic]@xmath89 with @xmath95 .
we previously saw , by theorem [ theorem well ordered ] , that @xmath96 evades @xmath66 .
now look at this example in the context of theorem [ thm : suffic ] , to see that there is no contradiction .
suppose @xmath97 is increasing , and @xmath98 .
then for any @xmath99 , the set @xmath100 in the theorem is unbounded , i.e. , @xmath101 .
nevertheless , fix any choice of @xmath19 , and note that for every @xmath8 large enough , @xmath100 does not contain any nonzero number smaller than @xmath19 , i.e. , @xmath102 = { \{{0}\}}$ ] .
thus , for any choice of @xmath19 , @xmath103 for every @xmath8 large enough .
in particular , @xmath104 .
throughout this section @xmath105 are two sets satisfying @xmath106 we must show @xmath0 ( countably ) evades @xmath1 .
since @xmath0 does not scale into @xmath1 , one thing that @xmath1-martingales can not do in general is to mimic @xmath0-martingales , not even up to a constant ratio .
we use this idea in order to construct a sequence of heads and tails that will separate between the two types of martingales .
let @xmath32 and @xmath19 be martingales with @xmath32 non - negative .
we say that @xmath107 is the _ @xmath108-ratio - minimizing outcome _ at @xmath109 ( assuming @xmath110 ) if either 1 .
@xmath111 , or 2 .
@xmath112 and @xmath113 , or 3 .
@xmath114 and @xmath115 . in words :
our first priority is to make the ratio @xmath108 decrease ; if this is impossible ( i.e. , the increments of @xmath32 and @xmath19 are proportional to their value at @xmath45 , and so @xmath108 does nt change ) , then we want @xmath19 to increase so as to insure that @xmath116 ; if that is impossible as well ( i.e. , both increments are 0 ) , we set @xmath8 to be @xmath117 , for completeness of definition only .
our definition extends to finite / infinite extensions of @xmath45 by saying that @xmath28 is the length @xmath118 ( possibly @xmath119 ) _ @xmath108-ratio - minimizing extension _ of @xmath45 , if @xmath120 is the @xmath108-ratio - minimizing outcome at @xmath121 , for every @xmath122 . for any such @xmath32 , @xmath19 , and @xmath45 the infinite @xmath108-ratio - minimizing extension of @xmath45 , @xmath28 , makes the ratio @xmath123 monotonically converging to a limit @xmath124 , as @xmath125 .
the next lemma will help us argue that the ratio between the increments @xmath126 also converges to @xmath127 in a certain sense .
[ lem : lhopital ] let @xmath128 and @xmath129 be sequences of real numbers .
assume that @xmath130 , for every @xmath25 .
if @xmath131 monotonically converges to a limit @xmath132 , and @xmath133 , then @xmath134 since @xmath135 is bounded , @xmath136 is bounded , too ; therefore @xmath137 it remains to prove that @xmath138_+=0,\ ] ] where @xmath139 and similarly @xmath140 .
we may assume w.l.o.g . that @xmath141 ( otherwise consider the sequence @xmath142 ) .
namely , @xmath143 , which implies that @xmath144 ; hence @xmath145_+ \leq \left[\left(\frac{b_k}{a_k}-l\right)a'_k\right]_+\leq \left(\frac{b_k}{a_k}-l\right)|a'_k|.\ ] ] now follows since @xmath146 converges to @xmath127 and @xmath147 is bounded .
[ cor ratio ] let @xmath19 and @xmath32 be a pair of martingales and @xmath148 .
assume that @xmath32 is non - negative , @xmath149 , and @xmath150 is bounded .
is bounded can be relaxed by assuming only that @xmath151 is bounded .
] let @xmath28 be the infinite @xmath108-ratio - minimizing extension of @xmath45 and @xmath152 .
for every @xmath153 the set @xmath154 has zero density .
note that @xmath155 by lemma [ lem : lhopital ] , we have @xmath156 which implies that the density of the set @xmath157 is zero , for every @xmath153 . the next step is to construct the history - independent @xmath0-martingale prescribed by theorem [ theorem bounded ] .
we formalize the properties of this martingale in the following lemma .
[ lemma 0 ratio ] let @xmath105 .
suppose that @xmath158 and @xmath0 does not scale into @xmath1 ; then there exists a history - independent @xmath0-martingale with positive initial value , @xmath19 , such that for every non - negative @xmath1-martingale , @xmath32 , and every @xmath47 such that @xmath149 , the infinite @xmath108-ratio - minimizing extension of @xmath45 , @xmath28 , satisfies @xmath159 note that the @xmath0-martingale , @xmath19 , does not depend on @xmath32 or @xmath45 , so the same @xmath19 can be used against any @xmath32 at any @xmath45 that leaves @xmath160 positive .
let @xmath161 be a countable dense subset of @xmath162 .
for every positive integer @xmath163 , let @xmath164 be the largest integer such that @xmath165 divides @xmath163 .
define @xmath166 .
the sequence @xmath167 has the property that the set @xmath168 has positive density , for every @xmath153 and @xmath169 .
let @xmath19 be a history - independent martingale whose increment at time @xmath163 is @xmath170 , for every @xmath171 ( with an arbitrary positive initial value ) .
let @xmath32 be an arbitrary non - negative @xmath1-martingale .
suppose that @xmath110 and let @xmath28 be the infinite @xmath108-ratio - minimizing extension of @xmath45 and let @xmath172 .
corollary [ cor ratio ] and guarantee that @xmath173 . by the assumption that @xmath0 does not scale into @xmath1
, we conclude that @xmath174 .
in the rest of this section we assume that @xmath175 .
this is w.l.o.g . since proportional sets are ( strongly ) equivalent .
we begin with an informal description of the casino sequence .
lemma [ lemma 0 ratio ] provides a history - independent @xmath0-martingale , @xmath19 , that can be used against any @xmath1-martingale .
given a sequence of @xmath1-martingales , @xmath176 , we start off by ratio - minimizing against @xmath177 .
when @xmath19 becomes greater than @xmath177 , we proceed to the next stage .
we want to make sure that @xmath177 no longer makes any gains .
this is done by playing adversarial to @xmath177 whenever he wagers a positive amount . at times when @xmath177 wagers nothing ( i.e. , @xmath178 ) , we are free to choose either @xmath117 or @xmath179 without risking our primary goal . at those times
we turn to ratio - minimizing against @xmath180 , while always considering the goal of keeping @xmath177 from making gains a higher priority . since @xmath181
, it is guaranteed that at some point we will no longer need to concern @xmath177 , and hence , at some even further point , @xmath19 will become greater than @xmath182 .
the process continues recursively , where at each stage our highest priority is to prevent @xmath177 from making gains , then @xmath180 , @xmath183 , and so on until @xmath184 ; and if none of @xmath185 wagers any positive amount , we ratio - minimize against @xmath186 .
when a positive wager of some @xmath187 , @xmath188 , is answered with an adversarial outcome , a new index @xmath189 must be calculated , so that @xmath19 is sufficient to keep @xmath190 from making gains .
that is , @xmath191 .
an inductive argument shows that for every fixed @xmath192 , there is a point in time beyond which none of @xmath185 will ever wager a positive amount ; therefore , at some even further point , @xmath19 becomes greater than @xmath193 ; hence the inductive step .
the above explains how the value of each @xmath187 converges to some @xmath194 , and the limit inferior of the value of @xmath19 is at least @xmath195 . in order to make sure that @xmath19 goes to infinity we include , among the @xmath187s , infinitely many martingales of constant value @xmath196 .
we turn now to a formal description .
as mentioned above , we assume without loss of generality that @xmath197 .
we additionally assume that @xmath198 , and so we can convert arbitrary @xmath1-martingales to non - negative ones by making them stop betting at the moment they go bankrupt .
let @xmath19 be a history - independent @xmath0-martingale provided by lemma [ lemma 0 ratio ] .
let @xmath176 be a sequence of non - negative @xmath1-martingales .
assume without loss of generality that infinitely many of the @xmath187s are the constant @xmath196 martingale .
we define a sequence @xmath199 recursively .
assume @xmath121 is already defined .
first we introduce some notation .
denote the value of @xmath19 at time @xmath163 by @xmath200 , and similarly @xmath201 , for every @xmath202 .
let @xmath203 note that the maximum is well defined , since the @xmath204 include infinitely many 1s .
we are now ready to define @xmath120 .
we distinguish between two cases : case i : there exists @xmath205 such that @xmath206 ; case ii : @xmath207 . in case i ,
let @xmath208 and define @xmath209 in case ii , @xmath120 is the @xmath210-ratio - minimizing outcome at @xmath121 .
explicitly , @xmath211 consider the tuple @xmath212 . in case i ,
@xmath213 is strictly less that @xmath214 according to the lexicographic order . in case
ii , @xmath214 is a prefix of @xmath213 , and so under a convention in which a prefix of a tuple is greater than that tuple , we have that @xmath215 is a non - increasing sequence .
is appended with an infinite sequence of @xmath216 elements .
] let @xmath217 .
it follows that from some point in time , @xmath218 consists of at least @xmath192 elements ; therefore the first @xmath192 elements of @xmath218 must stabilize at some further point in time .
namely , for @xmath163 large enough we have @xmath219 , for every @xmath220 .
since the increments of @xmath204 are bounded below by 1 , @xmath221 stabilize , too .
also , since @xmath222 , we have @xmath223 , for every @xmath224 .
since there are infinitely many @xmath225 s for which @xmath204 is constantly @xmath196 , the proof of theorem [ theorem bounded ] is concluded by showing that @xmath226 .
assume by negation that @xmath227 .
there is a time @xmath228 such that @xmath229 and @xmath230 , for every @xmath231 .
there can not be a @xmath231 , for which @xmath232 and @xmath233 .
that would mean a case i transition from time @xmath163 to @xmath234 and we would have @xmath235 , for some @xmath236 .
it follows that @xmath237 , for every @xmath231 . from time @xmath228 ratio - minimization against @xmath238
takes place .
let @xmath239 . if @xmath240 , then @xmath241 , for some @xmath231 ; at this point @xmath238 stabilizes ( otherwise @xmath186 would go bankrupt ) ; therefore @xmath242 ; therefore @xmath232 , which is not possible .
if @xmath243 , then by lemma [ lemma 0 ratio ] , there must be some time @xmath231 in which @xmath244 , which contradicts the definition of @xmath245 .
to show that if @xmath1 is well ordered and @xmath0 does not scale into @xmath1 , then @xmath0 evades @xmath1 , we construct an @xmath0-martingale @xmath19 , s.t . for any @xmath1-martingales
@xmath176 we construct a sequence @xmath28 on which @xmath19 succeeds , while every @xmath187 does not .
we begin with a rough outline of the proof ideas . @xmath19
always bets on `` heads . '' before tackling every @xmath187 , we first gain some money and `` put it aside . ''
then we ratio - minimize against @xmath187 .
it will eventually make @xmath19 sufficiently richer than @xmath187 , so that we can declare @xmath187 to be `` fragile '' now .
this means that from now on the casino can make @xmath187 lose whenever it is `` active '' ( i.e. , makes a non - zero bet ) , since @xmath19 can afford losses until @xmath187 is bankrupt .
when @xmath187 is not active , we can start tackling @xmath246 , while constantly making sure that we have enough money kept aside for containing the fragile opponents .
an important point is to show that once some @xmath187 becomes fragile , it remains fragile unless a lower - index martingale becomes active .
let @xmath247 be a sequence that is dense within @xmath248 , and such that each number in the sequence appears infinitely many times .
for example , given a dense sequence @xmath249 in @xmath248 , the sequence @xmath250 can be used . since
multiplying @xmath0 or @xmath1 by a positive constant does not make a difference , we may assume w.l.o.g . that @xmath251 , and that @xmath252 ( @xmath1 is well ordered ; hence in particular @xmath253 is bounded away from @xmath254 ) .
we construct @xmath19 and @xmath28 as follows .
denote @xmath255 , and similarly @xmath256 .
take integers @xmath257 we take @xmath258 .
for the first @xmath259 stages , @xmath260 .
then , after stage @xmath261 , @xmath262 for the next @xmath263 periods , and after stage @xmath264 , @xmath265 for @xmath266 stages , and so on .
but this goes on only as long as no `` tails '' appears .
whenever a `` tails '' appears , namely , at a stage @xmath163 where @xmath267 , we revert to playing from the beginning of the sequence , i.e. , @xmath260 for the next @xmath268 stages ( or until another `` tails '' appears ) , then @xmath262 , etc . for @xmath269
we define the function @xmath270 , which is similar to @xmath271 , but modifies sudden increases of @xmath272 into more gradual ones . and the wagers ] at the beginning of a block of stages where @xmath273 is wagered ( i.e. , where @xmath274 ) , @xmath275 equals @xmath273 .
if @xmath276 , then @xmath275 remains @xmath273 throughout this block .
otherwise , it linearly increases until reaching @xmath277 exactly at the beginning of the next block .
i.e. , suppose @xmath278 , and let @xmath279 be the beginning of the block ( of length @xmath280 ) of @xmath273 wagers .
if @xmath276 then @xmath281 . otherwise , @xmath282 @xmath283 the last one follows from @xmath278 , @xmath284 , and @xmath285 , by the definition of @xmath83 .
let @xmath176 be @xmath1-martingales ( and assume these martingales never bet on money that they do not have ) .
denote @xmath286 . to define the sequence @xmath28 , denote @xmath287 , and let @xmath288 be the largest integer such that @xmath289@xmath290 are the `` fragile '' martingales at time @xmath163 .
define @xmath291 and consider two cases .
( i ) if there exists some index @xmath292 s.t .
@xmath293 , let @xmath225 be the smallest such index , and @xmath120 is chosen adversely to @xmath294 .
( ii ) otherwise , @xmath120 is chosen by @xmath295-ratio - minimizing , i.e. , if @xmath296 and @xmath297 then @xmath298 , and otherwise @xmath299 .
we now show that these @xmath19 and @xmath28 indeed work .
[ lem : fragile ] for any @xmath163 , if @xmath300 and @xmath301 for every @xmath302 , then @xmath303 .
we prove the following equivalent claim : + ( i ) if @xmath304 is the smallest index such that @xmath305 , then @xmath306 .
( ii ) if @xmath307 for any @xmath308 , then @xmath309 . in case ( ii ) , denote @xmath310
. then in both cases @xmath311 is known .
let @xmath312 designate the lhs of this inequality .
we need to show that @xmath313 .
note that for any @xmath314 , @xmath315 , and that if @xmath316 then @xmath317 , since @xmath318 and @xmath319 .
\(i ) in this case the casino makes @xmath225 lose , hence @xmath320 ( recall that @xmath321 ) ; therefore @xmath322 .
if @xmath316 we are done . if @xmath323 then @xmath324 , and @xmath325 ; therefore , @xmath326 .
\(ii ) in this case @xmath327 ; hence @xmath328 . if @xmath316 we are done . if @xmath323 then @xmath329 and @xmath330 . but
@xmath323 also implies ( by the definition of @xmath28 ) that @xmath331 and @xmath332 .
since @xmath333 is always @xmath334 , we get that @xmath335 .
now , @xmath336 , because @xmath337 is @xmath338 .
thus , @xmath339 .
remark : the above argument also proves that @xmath19 is never bankrupt , i.e. , @xmath340 , and moreover @xmath341 , as follows . in the beginning @xmath342 .
as long as @xmath343 we are in case ( ii ) . in this case , if @xmath344 then @xmath312 does not decrease ; hence @xmath345 does not decrease . and if @xmath346 we just saw that actually @xmath347 , which implies that @xmath348 .
once @xmath349 , then @xmath350 , hence @xmath351 implies that @xmath348 ; and lemma [ lem : fragile ] implies that @xmath352 remains @xmath353 .
[ lem : inductive ] for any @xmath225 there exists a stage @xmath354 s.t . for any @xmath355 , @xmath356 , and @xmath357 .
we proceed by induction over @xmath358 ; namely , the induction hypothesis is that the lemma holds for @xmath359 .
note that the induction base case @xmath360 holds vacuously .
, as @xmath341 and @xmath361 imply that @xmath362 . ]
if @xmath363 for some stage @xmath364 , then @xmath357 for every @xmath365 , by lemma [ lem : fragile ] . from this stage on , the casino chooses adversely to @xmath225 whenever @xmath225 is active ( because the lower - index players are not active ) .
therefore , @xmath225 will be active at no more than @xmath366 stages after @xmath367 , since afterwards @xmath225 has nothing to wager , and we are done .
so assume by way of contradiction that @xmath368 for every @xmath369 .
then @xmath120 is @xmath370-ratio - minimizing . as long as @xmath371 we get `` heads ''
; therefore , from some stage on , @xmath372 ( as every @xmath273 appears infinitely many times , the sum of the wagers will not converge ) .
denote @xmath373 .
@xmath374 is non - increasing and therefore converges to a limit @xmath127 .
denote @xmath375 .
suppose there exists some @xmath192 s.t .
the wagers @xmath271 never reach beyond @xmath376 . hence , there are infinitely many stages @xmath163 where @xmath225 over - bets , i.e. , @xmath377 . for @xmath378 ,
let @xmath379 .
since @xmath1 is well ordered , there exists some @xmath380 s.t .
@xmath381 . since @xmath382 , @xmath383 for @xmath163 large enough . when @xmath225 over - bets and @xmath384 , then @xmath385 . when @xmath225 does not over - bet , then @xmath386 , and @xmath387 . therefore @xmath388 does not increase , and infinitely many times it decreases by at least @xmath389 ; hence eventually @xmath390 , which is a contradiction .
therefore , there does not exist an index @xmath192 as above .
this implies that for any @xmath391 , there is a stage @xmath163 after which @xmath392 is wagered @xmath393 consecutive times , and @xmath187 does not over - bet ( otherwise @xmath394 can not be reached ) .
now suppose that @xmath395 .
let @xmath396 be the set of all the values that the sequence @xmath247 takes .
@xmath397 is dense in @xmath0 , and @xmath398 ( since @xmath0 does not scale into @xmath1 ) ; therefore , also @xmath399 .
hence , there exists an @xmath400 s.t .
the distance between @xmath401 and @xmath402 is @xmath403 .
let @xmath404 .
for @xmath163 larger than some @xmath405 , @xmath406 . since @xmath407 appears infinitely many times in the sequence @xmath247 , there exist @xmath408 and an index @xmath391 , s.t .
@xmath409 is wagered @xmath410 consecutive times , and at each of these times @xmath411 , as otherwise @xmath412 , but that is over - betting since @xmath413 .
hence , @xmath414 .
but @xmath415 ; therefore @xmath416 ; hence @xmath417 . by the definition of @xmath83 , @xmath418 ; therefore after those @xmath410 times , @xmath419
. this can not be ; therefore @xmath174 . as @xmath420 , surely @xmath421 for large enough @xmath163 , namely , @xmath422 .
since @xmath423 , we get @xmath424 . at some stage
@xmath163 , @xmath19 starts wagering @xmath196 . for this @xmath163 , @xmath425 ; hence @xmath426 , contradicting our assumption that @xmath225 is not fragile .
lemma [ lem : inductive ] states that any @xmath187 is only active a finite number of times , and therefore it is bounded ; it also states that for any @xmath225 and large enough @xmath163 , @xmath427 , hence @xmath428 , and therefore @xmath429 .
the definition of anticipation corresponds to the game where gambler 0 first announces her @xmath0-martingale , then the regular gamblers announce their @xmath1-martingales , and then the casino chooses a sequence .
thus , the regular gamblers know the future actions of gambler 0 , the casino knows the future actions of all gamblers , and gambler 0 does not even observe the past bets of other gamblers . our proofs , however , do not rely on this state of affairs .
one may consider variants of this game in which the casino does not know the future actions , or the regular gamblers do not know the future actions of gambler 0 , or gambler 0 does observe the bets of others ; or any combination thereof .
our results hold for all these variants .
one of these variants looks more like a `` classic '' repeated game : at each stage of the game , first gambler 0 makes a bet , then the regular gamblers make bets , and then the casino chooses red or black ( and everything is observed by all ) .
another generalization is that the set @xmath1 need not be the same for all regular gamblers . that is , the results still apply for sets @xmath430 of wagers for gamblers @xmath431 , if the conditions hold for each @xmath432 separately ( of course , the conditions may not hold for their union ) .
@xcite consider the case of computable martingales , with @xmath0 being the set of all integers , and @xmath432 finite sets of integers .
they use a probabilistic argument to show evasion .
our proof provides , in particular , a construction of the casino sequence for this case .
the constructed sequence is computable relative to the history of bets .
it seems possible that our proof of theorem [ theorem well ordered ] could be modified so as to avoid the assumption that @xmath1 is well ordered .
we conjecture that theorems [ theorem bounded ] and [ theorem well ordered ] can be unified as the following statement .
let @xmath433 .
if @xmath434 , then @xmath1 anticipates @xmath0 only if @xmath0 scales into @xmath1 .
the present paper strove to understand the effect of restricting the wager sets on the prediction power of martingales .
as is often the case , understanding one thing brings up many new questions .
we list just a few .
* under the assumptions of theorem [ theorem bounded ] , @xmath0 can evade @xmath1 through a history - independent martingale .
is this also the case under the assumptions of theorem [ theorem well ordered ] ? * are single anticipation and countable anticipation different ? that is , are there sets @xmath105 , such that @xmath1 countably , but not singly , anticipates @xmath0 ?
* what can be said about the anticipation relation between sets that _ do _ include 0 as an accumulation point , for example , @xmath435 , @xmath436 , and @xmath12 ?
* @xcite introduced martingales defined by probabilistic strategies .
how do these martingales behave in our framework ?
we wish to thank abraham neyman and bernhard von stengle for their useful comments . | a decision maker observes the evolving state of the world while constantly trying to predict the next state given the history of past states .
the ability to benefit from such predictions depends not only on the ability to recognize patters in history , but also on the range of actions available to the decision maker .
we assume there are two possible states of the world .
the decision maker is a gambler who has to bet a certain amount of money on the bits of an announced binary sequence of states .
if he makes a correct prediction he wins his wager , otherwise he loses it .
we compare the power of betting strategies ( aka _ martingales _ ) whose wagers take values in different sets of reals . a martingale whose wagers take values in a set @xmath0
is called an @xmath0-martingale .
a set of reals @xmath1 _ anticipates _ a set @xmath0 , if for every @xmath0-martingale there is a countable set of @xmath1-martingales , such that on every binary sequence on which the @xmath0-martingale gains an infinite amount at least one of the @xmath1-martingales gains an infinite amount , too .
we show that for two important classes of pairs of sets @xmath0 and @xmath1 , @xmath1 anticipates @xmath0 if and only if the closure of @xmath1 contains @xmath2 , for some positive @xmath3 .
one class is when @xmath0 is bounded and @xmath1 is bounded away from zero ; the other class is when @xmath1 is well ordered ( has no left - accumulation points ) .
our results generalize several recent results in algorithmic randomness and answer a question posed by @xcite .
* keywords : * repeated games , gambling , algorithmic randomness , pseudo - randomness , predictability . + * jel classification : * c72 , c73 . |
frustrated magnetic systems have attracted a lot of research interest because such systems often exhibit interesting low temperature properties.@xcite the zig - zag spin chain with antiferromagnetic interactions between nearest- and next - nearest - neighbors is about the most simple frustrated system . in a zig - zag spin chain system with spin @xmath11 , the ground state phase diagram ( at temperature @xmath12 )
as a function of xxz anisotropy and ratio between nearest - neighbor ( nn ) and next - nearest - neighbor ( nnn ) interactions exhibits six different phases.@xcite in addition to two nel ordered phases and two phases with a haldane gap , there exists a large phase region called a gapless chiral phase where the chirality exhibits long range order without accompanying spin order , and a small phase region where there is a gapped chiral phase .
cav@xmath1o@xmath2 is a possible candidate for a zig - zag spin @xmath11 chain system.@xcite it has an orthorhombic crystal structure ( space group _ pnam _ ) at room temperature as shown in fig .
[ fig : structure ] .
vanadium moments at two crystallographically inequivalent sites respectively form two inequivalent zig - zag spin chains along the @xmath13-axis . in one of the two chains , the distances between nn and nnn vanadium atoms are 3.01 and 3.08 , respectively , while in the other chain
, these two distances are 3.01 and 3.06 , respectively .
the smallest interchain vanadium distances are 3.58 and 3.62 .@xcite thus one might expect a much smaller interchain coupling as compared to nn and nnn interactions within the chain .
o@xmath2 at room temperature .
large spheres two inequivalent v sites , small dark spheres ca , small light spheres o. the crystallographic @xmath14 and @xmath8 axes are along the vertical and horizontal directions in the plane of the page , respectively .
the @xmath13 axis is perpendicular to the page .
the cuboid denotes the size of a unit cell.,width=288 ] previous magnetization and @xmath0v nmr studies of cav@xmath1o@xmath2 suggested that the ground state of the system might be a gapless chiral phase.@xcite however , this finding contradicts earlier powder neutron diffraction studies which showed that the system is in an antiferromagnetic state at 4.2 k.@xcite the neutron measurements indicated a magnetic unit cell in which the @xmath8 and @xmath13 lattice constants are doubled and the spin directions in each chemical unit cell are reversed relative to their orientations in neighboring chemical unit cells along the @xmath8- and @xmath13-axes .
each chemical unit cell contains 8 vanadium spins and the magnetic spin structure within each chemical unit cell could not be uniquely determined .
the neutron diffraction pattern was found to be consistent with three different collinear models with spins parallel to the @xmath8-axis . by assuming the same spin moment at all vanadium sites ,
the magnetic moment of each vanadium spin was determined to be @xmath15 , only about half the value expected for a vanadium spin with @xmath16-factor @xmath17 and @xmath11.@xcite the presence of a low temperature antiferromagnetic phase is also supported by recent magnetization measurements on annealed cav@xmath1o@xmath2 single crystals , which showed a clear signature of an antiferromagnetic phase transition at temperature @xmath5 k.@xcite in order to obtain further evidence of the magnetic phase transition and to study the magnetic properties in the ordered state , we performed @xmath3o and @xmath0v nmr studies on @xmath3o - enriched powder and single crystal samples , respectively .
the @xmath3o nmr spectrum and the nuclear spin - lattice relaxation rate measurements give strong evidence for a magnetic transition at @xmath4 k in the powder sample .
in contrast to early @xmath0v nmr measurements,@xcite we could not detect a @xmath0v nmr signal in the powder sample around the normal larmor frequency in an applied field of 1.67 t in the temperature range of @xmath18 k. instead , we observed a _ zero - field _
@xmath0v nmr signal at @xmath19 k ( @xmath20 mhz at 4.2 k ) .
the zero - field @xmath0v nmr signal is observed because of a strong local field at @xmath0v nuclear sites ( @xmath21 t at 4.2 k ) in the ordered state .
the local field arises mainly from interaction between nuclei and vanadium core electrons , which are polarized by the ordered 3@xmath22 electronic spins.@xcite this local field points antiparallel to the direction of the local electronic spin moment . by studying how the resonance frequency changes as a function of the direction and magnitude of the applied magnetic field
, one can obtain information on the vanadium spin structure , as will be demonstrated below .
we measured the temperature dependence of the ordered moment to study the anisotropy gap of the antiferromagnetic spin wave excitations .
we also attempted to measure the temperature dependence of the @xmath0v nuclear spin - lattice relaxation rate @xmath23 .
however , due to the very broad line and the presence of nuclear quadrupole splitting ( the nuclear spin of @xmath0v is @xmath24 ) , the relaxation curves are highly nonexponential and depend strongly on the saturation condition .
thus , a reliable measurement of the @xmath0v @xmath25 was not achieved .
the remainder of the paper is organized as follows .
experimental details are explained in sec .
[ sec : exp ] .
@xmath3o and @xmath0v nmr results are presented in secs . [
sec : o17 ] and [ sec : v51 ] , respectively . in sec .
[ sec : summary ] , we give a summary of the main results of the paper .
polycrystalline single phase cav@xmath1o@xmath2 ( sample an-2 - 116 ) was synthesized via the solid - state route by reacting v@xmath1o@xmath26 ( 99.995% , mv labs ) with cao obtained by calcining caco@xmath26 ( 99.995% , aithaca chemicals ) at 1100 @xmath27c .
the chemicals were ground inside a he glove - box , then pressed and sintered at 1200 @xmath27c for 96 hours in flowing 4.5% h@xmath1-he with intermediate grindings .
phase purity was confirmed by powder x - ray diffraction ( xrd ) on a rigaku geigeflex diffractometer using cu k@xmath28 radiation in the @xmath29 range of 10@xmath30@xmath31.@xcite 72.1 atomic% @xmath3o - enriched oxygen ( msd isotopes ) was used for @xmath3o - enrichment .
about 1 g of the precursor cav@xmath1o@xmath2 was placed in a pt foil - lined alumina boat in an evacuated furnace tube , which was then preheated to 750 @xmath32c under dynamic vacuum .
the pumping line was then closed and the tube backfilled with the @xmath3o - enriched o@xmath1 . the mass gain on oxidation indicated a nominal composition of cav@xmath1o@xmath33 .
this was placed in flowing 4.5% h@xmath1-he and reduced as before to cav@xmath1o@xmath2 ( sample an-2 - 180e ) .
the final @xmath3o content of the enriched cav@xmath1o@xmath2 was about 25% .
powder xrd was used to confirm that the sample was single phase .
@xmath3o nmr measurements were performed utilizing a phase - coherent pulse spectrometer in applied fields of 3.0 and 4.7 t. the typical @xmath34 pulse length is @xmath35 .
the echo signal was produced by a sequence of a @xmath34 and a @xmath36 pulse , which produce the maximum echo signal intensity .
the separation between these two echo generating pulses was 40 @xmath37 .
the @xmath3o nmr spectra were measured by either fourier transform of half the echo signal or by plotting the area of the echo as a function of the rf frequency ( frequency sweep ) .
the nuclear spin - lattice relaxation rates were measured by monitoring the recovery of the echo intensity following a comb sequence of @xmath38 saturation pulses .
static magnetization versus temperature was measured in a quantum design squid magnetometer in a field of @xmath39 t and in the temperature range 5100 k to confirm the low temperature magnetic behavior and the ordering temperature @xmath40 k. the magnetic susceptibility of the powdered @xmath3o - enriched sample is shown in fig .
[ fig : chi ] .
the transition temperature is revealed by a small kink in the @xmath41 data at @xmath42 . versus temperature @xmath43 of the @xmath3o - enriched cav@xmath1o@xmath2 powder sample measured in a field of @xmath44 t.
the vertical arrow indicates the position of the antiferromagnetic transition temperature @xmath40 k.,width=288 ] two cav@xmath1o@xmath2 crystals were used in @xmath0v nmr measurements .
crystal # 1 was grown in an optical floating zone furnace while crystal # 2 was grown using a tri - arc crystal pulling method.@xcite the sizes of the crystals # 1 and # 2 are about @xmath45 mm@xmath46 and @xmath47 mm@xmath46 , respectively .
both crystals were annealed at 1200 @xmath30c under 4.5% h@xmath1-he flow and the antiferromagnetic phase transition temperatures of the annealed crystals were found from magnetic susceptibility and heat capacity measurements to be 69 k.@xcite the difference in @xmath42 between the powder and single crystal samples may be related to different unit cell volumes of these samples .
the unit cell volumes in the powder and single crystal cav@xmath1o@xmath2 samples are 296.0(5 ) and 298.0(5 ) @xmath46 , respectively .
the reason for this difference is , however , unknown at present .
a similar correlation between @xmath42 and the lattice volume was observed in eucu@xmath1ge@xmath1.@xcite the magnetic susceptibility of crystal # 2 is shown in fig .
[ fig : xstal ] with the field along @xmath14 and @xmath8 directions .
the antiferromagnetic transition temperature @xmath42 is clearly seen as a bifurcation in the susceptibilities along the two directions .
we note that , when the field is along the @xmath8 direction , a splitting between zero - field - cooled and field - cooled susceptibility is observed below @xmath48 k. versus temperature @xmath43 of cav@xmath1o@xmath2 crystal # 2 measured with the applied field @xmath44 t along the @xmath14 and @xmath8 directions .
the vertical arrow indicates the antiferromagnetic transition temperature @xmath5 k. the measurements were carried out under either field - cooled or zero - field - cooled conditions , as indicated.,width=288 ] a search for a zero - field @xmath0v nmr echo signal was performed at 4.2 k and was found to be located close to a frequency of 237 mhz at that temperature .
the echo was produced by a sequence of two pulses with the same pulse length , which was typically 4 @xmath49s and about half the length of a @xmath38 pulse .
the separation between the pulses was fixed to @xmath50s .
@xmath0v nmr spectra were measured by plotting the echo intensity as a function of the magnetic field . a variable magnetic field from 0 to 2.0 t was produced by an electromagnet .
the value of the magnetic field was measured by a hall magnetometer attached to one of the two magnet pole caps .
the difference between the measured field and the field at the position of the sample ( measured by the resonance frequency of protons in water ) was less than 0.005 t over the whole field range .
individually , the crystals were placed inside copper solenoid coils with the crystal @xmath13-axis parallel to the coil axis .
measurements of crystal # 1 involved rotation of the field in the @xmath14-@xmath8 and @xmath8-@xmath13 planes .
measurements of crystal # 2 involved rotation of the field in the @xmath14-@xmath8 plane .
the rotation of the field was achieved by rotating the cryostat together with the crystal about the crystallographic axes perpendicular to the field plane .
the misalignment between the rotation axis and the intended crystallographic axis is estimated to be less than @xmath51 .
figure [ fig : o17spectrum ] displays the @xmath3o nmr spectra for the @xmath3o - enriched powder sample of cav@xmath1o@xmath2 in @xmath53 t at three different temperatures . the spectrum at @xmath54 k was obtained via fourier transform of half the echo signal while the spectra at @xmath55 and 77 k were obtained by frequency sweep .
nuclei have spin @xmath56 and thus possess a nonzero electric quadrupole moment . since the local environments of all oxygen sites do not possess cubic symmetry ( point group @xmath57 ) , one expects a quadrupole splitting of the @xmath3o resonance frequencies .
we attribute the lack of a powder pattern of the first order quadrupole splitting in the observed spectra to a smaller quadrupole splitting compared to the magnetic broadening of the spectra .
the positions of the spectrum peak position show negligible field dependence .
the chemical shift @xmath58 of the peak position is given by @xmath59 , where @xmath60 is the isotropic hyperfine coupling constant and @xmath61 is the magnetic susceptibility .
using the data in figs .
[ fig : chi ] and [ fig : o17spectrum ] and additional @xmath41 data ( not shown ) , we obtain the upper limit @xmath62 . the line gets broader below @xmath42 with decreasing temperature . as shown in fig .
[ fig : o17spectrum ] , the absorption line at @xmath63 k is much broader than the lines at @xmath55 and 296 k. as will be further shown below , this broadening is a signature of an antiferromagnetic phase transition at @xmath64 k , where the nuclear spin - lattice relaxation rate @xmath23 exhibits a peak .
as the temperature approaches the phase transition temperature , the electronic vanadium spins slow down dramatically and thus induce an inhomogeneous static ( on the nmr time scale ) dipolar or transfered hyperfine field on neighboring @xmath3o sites and broaden the @xmath3o nmr line . due to the fast decrease of signal intensity below @xmath42 with decreasing temperature ,
detailed measurements of the temperature dependence of the line broadening in this temperature range were not pursued .
o nmr signal for an @xmath3o - enriched powder sample of cav@xmath1o@xmath2 at three different temperatures in an applied magnetic field @xmath65 t. a strong inhomogeneous broadening is observed close to the magnetic transition temperature @xmath4 k. the solid line at 77 k is a guide to the eye.,width=288 ] the recovery of the @xmath3o longitudinal nuclear magnetization @xmath66 following the saturation pulses is a single exponential function at @xmath67 k. this shows that all the @xmath3o nmr transitions are saturated and thus there are no satellite peaks located away from the main peaks displayed in fig .
[ fig : o17spectrum ] , consistent with a small quadrupole splitting as discussed above .
below 100 k , deviation from single exponential behavior is observed , indicating a distribution of @xmath23 values . in order to extract a characteristic relaxation rate at all temperatures , we fitted the recovery curves to a stretched exponential function @xmath68 .
\label{eqn : exponential}\ ] ] the relation of the parameters @xmath69 and @xmath70 to the @xmath23 distribution underlying a stretched exponential function has been discussed in a recent paper.@xcite figure [ fig : o17t1 ] shows the temperature dependence of @xmath71 and @xmath70 in @xmath65 and 4.7 t. at @xmath67 k , the relaxation rate is almost temperature independent .
below 100 k , @xmath69 exhibits a strong enhancement and reaches a peak at @xmath4 k. combining the above nmr results with magnetization studies of powder and single crystals,@xcite and with the magnetic susceptibility data in fig .
[ fig : chi ] , we identify the @xmath69 peak temperature as the temperature of an antiferromagnetic phase transition @xmath42 .
the peak in the nuclear spin - lattice relaxation rate at @xmath42 results from an enhancement and slowing down of the electronic spin fluctuations at wave vectors close to the antiferromagnetic ordering vector as the temperature approaches @xmath42 from either side.@xcite dependence of the @xmath3o nuclear spin - lattice relaxation rate @xmath69 [ see eq .
( [ eqn : exponential ] ) ] of @xmath3o - enriched powder cav@xmath1o@xmath2 in applied magnetic fields @xmath65 and 4.7 t. a strong peak is observed close to the antiferromagnetic transition temperature @xmath4 k , as labeled by the arrow .
inset : the stretching exponent @xmath70 versus @xmath43.,width=288 ]
in an external magnetic field @xmath72 , the resonance frequency @xmath73 of the @xmath0v nuclear spins is given by @xmath74 where @xmath75 is the hyperfine coupling constant between the nuclear spin and the vanadium electronic spins @xmath76 , @xmath77 is the gyromagnetic ratio of @xmath0v nuclear spins , and @xmath78 denotes the average electronic spin value in thermal equilibrium . in our experiments ,
the local field is much larger than the applied field : @xmath79 .
depending on whether the applied field component along @xmath78 is parallel or antiparallel to @xmath80 , the resonance frequency shifts to higher or lower values , respectively .
if cav@xmath1o@xmath2 is a collinear antiferromagnet at low temperatures , where opposite spin directions exist , upon application of an external field along the ordering axis , the zero - field spectrum should split into two peaks .
these peaks should be symmetrically displaced above and below the zero - field peak frequency .
-@xmath8 plane at rf frequencies of 231 mhz ( left panels ) and 243 mhz ( right panels ) .
the angles between the field and the average of the two projections of the two spin directions onto the @xmath14-@xmath8 plane ( @xmath81 in fig .
[ fig : spinstrctr ] ) are labeled in each panel .
the arrows indicate the positions of the peaks .
@xmath81 is approximately parallel to the crystallographic @xmath8-axis .
the spectra were measured at 4.2 k on crystal # 1.,width=288 ] figure [ fig : specab ] shows the field - swept spectra with the field at different angles in the @xmath14-@xmath8 plane , measured with rf frequencies both higher and lower than the zero - field peak frequency @xmath82 mhz .
all measurements in this subsection were performed on cav@xmath1o@xmath2 crystal # 1 .
in contrast to a single peak above and a single peak below the zero - field peak frequency @xmath83 when an applied field is present , as expected for a collinear antiferromagnet , instead we see two peaks above @xmath83 and two peaks below @xmath83 in applied fields as shown by the vertical arrows in fig .
[ fig : specab ] . whether each set of two peaks is resolved depends on the angle of the applied field in the @xmath14-@xmath8 plane as shown .
we infer below that the spectra in fig .
[ fig : specab ] ( and [ fig : specbc ] ) are consistent with a magnetic structure at 4.2 k that consists of two antiferromagnetic substructures , each of which is a collinear antiferromagnetic arrangement where the angle between the ordering axes of the two substructures is @xmath7 . the angle labeled in each panel of fig .
[ fig : specab ] is the angle between the applied field and the axis @xmath81 , which is the average of the projections onto the @xmath14-@xmath8 plane of the two spin ordering directions ( see fig . [
fig : spinstrctr ] below ) .
@xmath81 is approximately parallel to the @xmath8-axis and is determined by fitting the peak positions versus angle , as will be explained below . in our discussions of the @xmath0v nmr results , we assume that the applied magnetic field only shifts the nmr frequency without affecting the electronic spins .
in fact , the ordered electronic moments can be tilted by the applied field due to the presence of a torque .
however , we can show that the tilting angle is indeed negligibly small . from the magnetization measurements,@xcite at 4.2 k ,
the susceptibility @xmath61 of single crystal cav@xmath1o@xmath2 with applied field in the @xmath14 , @xmath8 or @xmath13 directions is @xmath84 cm@xmath46/mol , which corresponds to an induced moment of @xmath86 for each vanadium spin in a 2 t field .
the tilting angle required to produce such a moment is only @xmath87 , assuming an ordered moment of @xmath88.@xcite both this angle and the induced moment are negligible to our studies .
the spectra in fig .
[ fig : specab ] exhibit a two - peak structure when the field points away from the @xmath89 axis .
when measured by field sweep at a fixed frequency @xmath73 , each peak @xmath90 should appear at the field value @xmath91 where @xmath92 is the angle between the field and the respective electronic spin ordering direction of a magnetic substructure noted above .
the observed two - peak structures for @xmath93 and @xmath94 strongly indicate the presence of these two different antiferromagnetic spin ordering directions in the system .
the ability to resolve the two peaks at the larger angles in fig .
[ fig : specab ] ( and [ fig : specbc ] ) is related to the larger partial derivative of @xmath95 of eq .
( [ eqn : hpk ] ) at the larger @xmath92 values for @xmath96 rad .
the two peaks in the same spectrum have different heights , as can be clearly seen in fig .
[ fig : specab ] ( b ) , ( d ) , ( g ) , and ( i ) .
the reason behind this difference is currently not understood .
it may be due to the change of the nuclear spin - spin relaxation times at different field values , since we fix the separation between the two rf pulses for echo generation to be @xmath50s .
in fact , we observed a field - dependent oscillation of the spin echo intensity as a function of the separation between the two echo - generating pulses . figure [ fig : t2 ] displays two nuclear spin - spin relaxation curves measured under identical conditions except for different external magnetic fields of 1.346 and 1.86 t , respectively .
the oscillation pattern shows a clear field dependence .
however , without knowing the detailed functional form of the decay curves , we can not extrapolate the signal intensity back to zero pulse separation to correct for the nuclear spin - spin relaxation effect .
-@xmath8 plane and form an angle of @xmath97 from the @xmath81 direction .
inset : the field swept spectrum measured at a fixed pulse separation of 18 @xmath49s in the same field direction .
the vertical arrows show the field positions of 1.346 and 1.86 t. the measurements were performed at 4.2 k on crystal # 2.,width=288 ] we further note that the different heights of the two peaks can not be attributed to the different percentage of spins in the two spin substructures . in such a scenario ,
the spins contributing to the left peaks in figs .
[ fig : specab](b ) and ( g ) should contribute to the right peaks in figs .
[ fig : specab](d ) and ( i ) ( see fig .
[ fig : pkvsq ] ) , and the left peaks should be higher in one orientation while lower in the other in figs .
[ fig : specab](b ) and ( d ) , and in figs . [ fig : specab](g ) and ( i ) , respectively .
however , the spectra in figs .
[ fig : specab](b ) and ( d ) , and in figs . [ fig : specab](g ) and ( i ) are almost the same . the symmetry in the spectra with the field on opposite sides of @xmath81 , such as in figs .
[ fig : specab](b ) and ( d ) and in figs .
[ fig : specab](g ) and ( i ) , indicates that the number of spins in the two substructures are the same .
since the above two - peak structure is observed with the field in the @xmath14-@xmath8 plane , these measurements can only detect the difference of the spin projections of the two substructures onto the @xmath14-@xmath8 plane . in order to determine whether or not the projections onto the @xmath8-@xmath13 plane are also different , we also measured the spectra with the field in the @xmath8-@xmath13 plane .
some representative spectra with the field in different directions are displayed in fig .
[ fig : specbc ] , where the angles listed are described in the caption .
as one can see , a two peak structure is still observed when the field is at a large angle from the @xmath98 axis .
however , the separations between the two peaks are smaller than in fig . [
fig : specab ] , indicating a smaller angle between the two easy axis projections onto the @xmath8-@xmath13 plane than onto the @xmath14-@xmath8 plane . -@xmath13 plane at rf frequencies of 231 mhz ( left panels ) and 243 mhz ( right panels ) .
the angles between the field and the average of the two projections of the two spin directions onto the @xmath8-@xmath13 plane ( @xmath98 in fig .
[ fig : spinstrctr ] ) are labeled in each panel .
the arrows indicate the positions of the peaks .
the spectra were measured at 4.2 k on crystal # 1.,width=288 ] in order to study whether there exists canting and/or an imbalance in the number of spins in opposite directions for each of the two ordered magnetic substructures , we compared the spectra with those measured with the field rotated by @xmath99 .
figure [ fig : fieldinverse ] displays two spectra measured at @xmath100 mhz with the field parallel to the @xmath14-@xmath8 plane and @xmath101 and @xmath102 away from the @xmath81 direction , respectively .
these two spectra are identical within experimental error , indicating the absence of spin canting and the same number of spins in opposite directions within each magnetic substructure .
mhz with and without the field direction reversed .
the fields are parallel to the @xmath14-@xmath8 plane and form angles of @xmath101 and @xmath102 from the @xmath81 direction , respectively .
the spectra were measured at 4.2 k on crystal # 1.,width=288 ] thus we propose a model of the spin structure as shown in fig .
[ fig : spinstrctr ] .
various notations used in the model are explained in the caption of fig .
[ fig : spinstrctr ] .
there are equal numbers of spins in the two antiferromagnetic substructures , each of which consists of collinear antiparallel spins also with equal number .
the plane defined by the two ordering directions is parallel neither to the @xmath14-@xmath8 nor the @xmath8-@xmath13 plane .
the average ordered moment direction @xmath103 is approximately parallel to the @xmath8-axis .
this is consistent with single crystal anisotropic magnetization measurements versus temperature which showed that below @xmath42 , the average easy axis of the magnetic structure is approximately the @xmath8-axis.@xcite note that in our nmr study , we can not determine the location in the lattice of the two different magnetic substructures .
there are two different antiferromagnetic ordering substructures with equal numbers of spins , each of which has a collinear antiferromagnetic spin arrangement .
@xmath104 , @xmath105 , and @xmath106 are the angles between these two directions , and their projections on @xmath14-@xmath8 and @xmath8-@xmath13 planes , respectively .
@xmath103 , @xmath81 , and @xmath98 are the average of the two directions and their projections on @xmath14-@xmath8 and @xmath8-@xmath13 planes , respectively . @xmath107 ( @xmath108 ) and @xmath109 ( @xmath110 ) are the angles formed between a fixed arbitrary experimental reference direction in the @xmath14-@xmath8 ( @xmath8-@xmath13 ) plane and the applied field @xmath72 and @xmath81 ( @xmath98 ) , respectively . , width=288 ] to extract the angle between the spin ordering directions of the two substructures , we measured the dependence of the peak positions versus the field directions at rf frequencies of 231 and 243 mhz .
the results are shown in fig .
[ fig : pkvsq ] . for small angles between @xmath72 and @xmath81 or @xmath98 ,
only one peak is observed . at larger angles ,
the positions of two peaks can be resolved .
two different symbols are used to represent the two different spin substructures , while for spectra with single peaks , a third symbol is used . since the angle between the spin direction and the @xmath8-axis is much less than one radian , @xmath111 in eq .
( [ eqn : hpk ] ) for the field in the @xmath14-@xmath8 plane can be approximated by @xmath112 where @xmath113 are the angles between the field and the two spin directions @xmath114 and @xmath115 , respectively .
similarly , for the field in the @xmath8-@xmath13 plane , one has @xmath116 we fitted eq .
( [ eqn : hpk ] ) with @xmath117 given by eqs .
( [ eqn : hpkab ] ) and ( [ eqn : hpkbc ] ) to the data in fig .
[ fig : pkvsq ] .
the free parameters in the fit were @xmath83 , @xmath109 , @xmath110 , @xmath118 , and @xmath119 .
the best fit results are @xmath120 mhz , @xmath121 , and @xmath122 .
the fits are shown in fig .
[ fig : pkvsq ] .
since the angles between the spins and the @xmath8-axis are much less than one radian , we have @xmath123 from which one obtains the angle between the easy axes of the two magnetic substructures to be @xmath124 .
-@xmath13 ( top two panels ) and @xmath14-@xmath8 ( bottom two panels ) planes of crystal # 1 , where the rf frequencies are equal to 231 and 243 mhz , respectively . for definitions of the angles @xmath107 ,
@xmath109 , @xmath108 , and @xmath110 , see fig .
[ fig : spinstrctr ] . circles and filled squares correspond to the two different spin ordering directions of the two magnetic substructures , respectively .
the symbol @xmath125 is used when the two peaks from the two spin directions overlap and only a single peak can be observed .
the error in @xmath126 is comparable to the size of the symbols unless shown explicitly .
the solid and dotted lines represent the fits by the theoretical prediction in eq .
( [ eqn : hpk]).,width=240 ] in addition to the study of the angular dependence of the peak positions , we also measured their frequency dependences to further confirm the proposed spin structure .
figures [ fig : spechparb](a ) and ( b ) show the field swept spectra with the field @xmath72 pointing along the @xmath98 direction , at rf frequencies lower and higher than @xmath82 mhz , respectively .
note that when @xmath127 , the two magnetic substructures have the same peak positions of the spectra ( see the zero - angle data in fig . [
fig : pkvsq ] ) . the peaks in figs .
[ fig : spechparb](a ) and ( b ) both shift to higher fields when the frequency shifts further away from @xmath83 , respectively .
the peak positions @xmath126 versus rf frequency are plotted in fig .
[ fig : spechparb](c ) .
the two sets of data points can be well fitted by the two linear equations @xmath128 where @xmath83 is the peak frequency of the spectrum at zero applied field .
a fit of eq .
( [ eqn : linear ] ) to the data gives @xmath129 mhz / t , and @xmath130 mhz .
assuming @xmath103 to be parallel to the @xmath8-axis , the value of @xmath131 should be @xmath132 mhz / t , where @xmath133 is the above angle between the ordering directions of the two magnetic substructures .
this value of @xmath134 is very close to the above fitting result .
v nmr spectra at four different frequencies at 4.2 k. the frequencies are given under each spectrum in units of mhz .
the field is applied parallel to the @xmath98 direction .
( c ) : the frequency versus the peak field of the spectra .
the solid lines are linear fits by eq .
( [ eqn : linear ] ) .
the measurements were done on crystal # 1 at 4.2 k.,width=288 ] in this subsection , we will study the temperature dependence of the vanadium ordered moment , which provides evidence for an energy gap in the antiferromagnetic spin wave excitation spectrum , arising from anisotropy effects .
then we will discuss the value of the saturation vanadium spin moment at low temperatures .
measurements in this subsection were performed on crystal # 2 .
the experiment was set up to allow field rotation in the @xmath14-@xmath8 plane . by rotating the field in the @xmath14-@xmath8 plane , the @xmath81 direction ( see fig . [
fig : spinstrctr ] ) is identified as the direction along which the peak position in the spectrum is a minimum at a fixed rf frequency away from @xmath83 ( see fig .
[ fig : pkvsq ] ) . after identifying the @xmath81 direction , all subsequent measurements of the spectra
were performed versus @xmath135 at fixed rf frequencies with the field along the @xmath81 direction . with the field along @xmath81
, only a single peak is observed in the spectrum at each frequency ( see fig . [
fig : spechparb ] ) . in fig .
[ fig : fvshpk2 ] , we display the frequency dependence of the peak position at 4.2 k. similar to the case of crystal # 1 ( fig . [
fig : spechparb ] ) , the data points can be well fitted by two straight lines .
a fit of eq .
( [ eqn : linear ] ) to the data gives @xmath136 mhz and @xmath137 mhz / t .
this value of @xmath131 is in agreement with the fitting value in crystal # 1 .
however , the value of @xmath83 is slightly larger than in crystal # 1 .
this slight difference may be due to sample - dependent differences . versus peak @xmath126 in field swept spectrum in crystal # 2 at 4.2 k. the field
is applied along the @xmath138 direction .
the solid lines are fits with eq .
( [ eqn : linear]).,width=192 ] figure [ fig : spectrat ] displays representative spectra measured at four different temperatures . for comparison between the different spectra
, the @xmath139-axis has been converted to the quantity @xmath140 , with @xmath141 mhz / t . as the temperature increases , the signal intensity decreases rapidly and the spectra can only be measured below 45 k. in order to more accurately determine the peak position of the spectra , we fitted the original field swept spectra ( with the @xmath139-axis being @xmath135 ) by a gaussian function @xmath142 , \label{eqn : gauss}\ ] ] with @xmath75 , @xmath143 , @xmath126 , and @xmath144 as fitting parameters .
the zero field peak frequency @xmath83 is then determined from @xmath145 where @xmath141
mhz / t and the @xmath146 and @xmath147 signs correspond to the cases of @xmath94 and @xmath93 , respectively . in order to determine whether @xmath94 or @xmath93 , spectra were measured with at least two different frequencies at each temperature . with the correct choices of the @xmath146 or @xmath147 signs , the obtained @xmath83 values for different spectra as in fig .
[ fig : spectrat ] are the same within experimental error at each temperature .
the final @xmath83 value is an average over all calculated @xmath83 values for various spectra at the same given temperature . ) to extract the peak positions.,width=288 ] figure [ fig : v51fvst ] shows the temperature dependence of @xmath83 . since the temperature dependence of the hyperfine coupling constant
can be ignored,@xcite @xmath148 is directly proportional to the ordered local moment .
the ordered moment is almost temperature independent at @xmath149 k. in the spin wave theory of a three - dimensional antiferromagnet _ without anisotropy _ , the initial decrease of the ordered moment with temperature @xmath43 should follow a @xmath150 dependence.@xcite fitting the data by a power law gives an exponent @xmath151 , an unphysically large value ( not shown ) .
the temperature independence below @xmath149 k thus indicates the presence of an anisotropy - induced energy gap for spin wave excitations.@xcite v nmr spectra peak position in zero applied field in cav@xmath1o@xmath2 crystal # 2 .
the dotted and solid curves are fits by eqs .
( [ eqn : mtfit1 ] ) with one - dimensional spin wave dispersion and ( [ eqn : mtfit3 ] ) with three - dimensional dispersion , respectively.,width=288 ] before estimating the gap energy , we will first estimate the value of the saturation moment at @xmath12 k. the local moments reach their saturation value at @xmath149 k. from the value of @xmath152 mhz at @xmath149 k , one obtains a local field value of @xmath153 t. in order to infer the value of local moment from the local field value , the hyperfine coupling constant @xmath75 and the @xmath16-factor have to be determined . with known values of @xmath75 and @xmath16 , the ordered moment @xmath154 is @xmath155 .
the local field is dominated by the contact interaction through the polarized core electrons , which is approximately proportional to the number of unpaired electronic spins in the @xmath156 orbitals with a proportionality constant of 12.5 t per unpaired electron to within 20%.@xcite however , beside the contact interaction , orbital effects@xcite and transfered hyperfine coupling with the neighboring v@xmath157 ions@xcite may also contribute significantly to the local field at the @xmath0v nuclear site . in the absence of a knowledge of the orbital effects and transfered interactions
, we will estimate a possible range of the @xmath158 value in cav@xmath1o@xmath2 using the known values of @xmath158 in other v@xmath157 compounds .
the values of @xmath158 in lavo@xmath26 , yvo@xmath26 , and v@xmath1o@xmath26 are @xmath159 , @xmath160 , and @xmath161 t/@xmath162 , respectively.@xcite for v@xmath157 in al@xmath1o@xmath26 , epr measurements gave @xmath163 t/@xmath162.@xcite the range of @xmath158 in the above four compounds is between @xmath164 and @xmath160 t/@xmath162 . assuming @xmath158 in cav@xmath1o@xmath2 lies in the same range , the low temperature ordered vanadium moment in cav@xmath1o@xmath2 is thus in the range of 1.021.59 @xmath162 .
this @xmath165 range is consistent with the value @xmath15 extracted from the previous neutron powder diffraction study.@xcite next we estimate the energy gap for the spin wave excitations . from the neutron diffraction
studies,@xcite we know that the spins reverse their ordering direction upon moving along the @xmath13-axis .
therefore the spin structure in the ordered state within a zig - zag chain should look as in fig .
[ fig : zigzag ] , where we ignore the possible small misalignment of @xmath166 between spins in the two magnetic substructures discussed above . because each spin in one leg of the chain couples by the same exchange constant @xmath167 to two spins in the other leg that are ordered in opposite directions , we expect the effective coupling between the two legs within a zig - zag chain
is greatly reduced in the ordered state . as a result , we will consider the nearest - neighbor interactions within the leg @xmath168 as the only dominant magnetic interaction and treat the effect of interleg interaction within a zig - zag chain as a weak interchain interaction .
for simplicity , we will use a single exchange constant @xmath169 to characterize the effect of the interchain interactions . to include the effect of anisotropy
, we assume a single ion anisotropy in the system with a single direction of easy axis .
then , the hamiltonian can be written as @xmath170 where @xmath58 is the anisotropy constant , the index @xmath171 runs through the spins in one leg of the chain , and the summation @xmath172 runs through all interleg and interchain nearest - neighbor pairs .
o@xmath2 . due to the alternation of the spin directions along the @xmath13-aixs ,
the interaction between the spins in the two legs of the zig - zag chain are essentially decoupled .
the possible misalignment of @xmath166 between spins in the two legs of the zig - zag chain is ignored . @xmath167 and @xmath168 are the nearest - neighbor inter and intra - leg exchange interactions , respectively . ]
the fourier transform of the exchange interactions is @xmath173 where @xmath174 connects two spins in opposite sublattices , @xmath175 is the nearest - neighbor exchange constant , and the index @xmath176 runs through all the nearest - neighbor spins of spin @xmath171 in the opposite sublattice ( each sublattice consists of spins in the same direction ) .
the spin wave dispersion relation is given by@xcite @xmath177 ^ 2 - [ 2s\mathcal{j}(q)]^2\right\}^{1/2 } , \label{eqn : disp}\ ] ] where we ignored interactions between spins in the same sublattice .
the spin wave gap value is given by the value of @xmath178 at @xmath179 . in the limit of small anisotropy @xmath180 , the gap energy is given by @xmath181^{1/2}. \label{eqn : eg}\ ] ] in the spin wave theory , the decrease of sublattice magnetization is due to the thermal activation of spin wave excitations . in
the above bipartite antiferromagnetic system,@xcite @xmath182 where the integral is limited to the first brillouin zone of one sublattice , @xmath183 is the sample volume per sublattice site , and @xmath184 is the number of thermally excited antiferromagnetic magnons . the dispersion relation in eq . ( [ eqn : disp ] )
depends on the spin lattice structure and the exchange interactions @xmath185 . for a quasi one - dimensional chain with interchain coupling @xmath169 , at temperatures @xmath186
, one can ignore the dispersion perpendicular to the chain direction .
then for small values of @xmath187 , which is the @xmath188 vector component along the chain , one can perform a taylor series expansion of @xmath189 as @xmath190 , \label{eqn : smallk}\ ] ] where @xmath191 is the nearest - neighbor distance within the leg . at @xmath192 ,
only spin waves at small @xmath187 values have significant contributions to the integral in eq .
( [ eqn : integral ] ) , so one can change the limits of integral for @xmath187 in eq .
( [ eqn : integral ] ) to @xmath193 .
the small @xmath194 approximation is valid only at temperatures where @xmath195,@xcite which is satisfied within our experimental temperature range . substituting eq .
( [ eqn : smallk ] ) into eq .
( [ eqn : disp ] ) , and changing the limits of integral for @xmath187 in eq .
( [ eqn : integral ] ) to @xmath193 , one obtains in the limit of small anisotropy @xmath196 and @xmath197 @xmath198 where @xmath199 equation ( [ eqn : st1d ] ) is valid at temperatures @xmath200 . in cav@xmath1o@xmath2 , @xmath201 might fall within the experimental temperature range in fig .
[ fig : v51fvst ] ( @xmath202 k ) .
so it is useful to consider the other limit of @xmath203 , where a three dimensional dispersion is more appropriate . applying a small @xmath194 approximation ,
one obtains @xmath204\right\ } , \label{eqn : smallk3}\ ] ] where for simplicity , we assumed an isotropic dispersion in the @xmath14-@xmath8 plane , @xmath205 is a geometrical factor of order one which depends on the spin structure,@xcite and @xmath206 is of the order of @xmath207 . by combining eqs .
( [ eqn : disp ] ) , ( [ eqn : integral ] ) , and ( [ eqn : smallk3 ] ) , and changing the three limits of integrations to @xmath208 in eq .
( [ eqn : integral ] ) , then instead of eq .
( [ eqn : st1d ] ) , we have in the limit of small anisotropy @xmath196 and @xmath197@xcite @xmath209 where @xmath210 with @xmath211 $ ] . using eqs .
( [ eqn : st1d ] ) and ( [ eqn : st3d ] ) and the relation @xmath212 one obtains the variation of the zero - field @xmath0v nmr resonance frequency @xmath83 as ( for 1d ) @xmath213 \label{eqn : mtfit1}\ ] ] or ( for 3d ) @xmath214 , \label{eqn : mtfit3}\ ] ] depending on whether a one - dimensional ( 1d ) or three - dimensional ( 3d ) dispersion is used for @xmath178 .
we fitted eqs .
( [ eqn : mtfit1 ] ) and ( [ eqn : mtfit3 ] ) to the @xmath148 versus @xmath43 data in fig .
[ fig : v51fvst ] at @xmath215 k with @xmath216 , @xmath143 , and @xmath217 as free parameters .
the best fit results are @xmath218 mhz , @xmath219 k , @xmath220 for the 1d dispersion with eq .
( [ eqn : mtfit1 ] ) , and @xmath221 mhz , @xmath222 k , and @xmath223 for the 3d dispersion with eq .
( [ eqn : mtfit3 ] ) .
the best fit curves are shown as the dotted and solid curves in fig .
[ fig : v51fvst ] , respectively . since eqs .
( [ eqn : mtfit1 ] ) and ( [ eqn : mtfit3 ] ) are derived under the two limiting conditions of @xmath224 and @xmath225 , respectively , one can expect that the actual @xmath217 value might lie somewhere between 64 and 98 k. given a value of @xmath217 , we can make a rough estimate of the anisotropy constant @xmath58 . from the magnetization study ,
one estimates that the intrachain nearest - neighbor exchange constant to be of the order of @xmath226 k.@xcite taking @xmath11 , @xmath227 k , and @xmath228 , we thus have from eq .
( [ eqn : eg ] ) that @xmath229 k. the above fitting value of @xmath220 from 1d dispersion is similar to the calculated value of 0.80 . the fitting value of @xmath223 from the 3d dispersion constrains the value of @xmath230 in the 3d model .
taking @xmath231 , @xmath232 , and @xmath233 , then from eq .
( [ eqn : b3 ] ) one has @xmath234 and @xmath235 k. however , this value of @xmath230 seems inconsistent with the initial assumption of @xmath225 required for the three - dimensional model to be valid .
therefore , eq . ( [ eqn : mtfit1 ] ) of the 1d model might provide a better approximation to the @xmath148 data .
we have presented @xmath3o and @xmath0v nmr results on the zig - zag spin chain compound cav@xmath1o@xmath2 .
the strong inhomogeneous broadening and a peak in the nuclear spin - lattice relaxation rate versus temperature of @xmath3o nmr confirm the presence of an antiferromagnetic phase transition at 78 k in a powder sample .
the crystals we studied have @xmath5 k. @xmath0v nmr in the ordered state of crystals reveals the presence of two antiferromagnetic substructures at 4.2 k , each of which is collinear and which form an angle of @xmath236 between them with the average direction approximately parallel to the @xmath8-axis .
the origin and location in the lattice of the different spin substructures remain unknown . however , we speculate that the two magnetic substructures are associated with the two inequivalent v@xmath157 @xmath11 zig - zag spin chains in the orthorhombic crystal structure , respectively .
magnetic neutron diffraction studies are required to further characterize the magnetic structure below @xmath42 .
the temperature dependence of the zero - field resonance frequency at low temperatures suggests the presence of magnetic anisotropy and an energy gap in the spin wave excitation spectrum .
the energy gap is estimated from spin wave theory to be between 64 and 98 k. a. niazi , s. budko , a. honecker , j. q. yan , d. l. schlagel , t. a. lograsso , and d. c. johnston , ( unpublished ) ; a. niazi , d. c. johnston , s. budko , d. l. schlagel , t. a. lograsso , and a. honecker , bull . am .
* 51 * , 192 ( 2006 ) , abstract b23.6 . | @xmath0v nmr studies on cav@xmath1o@xmath2 single crystals and @xmath3o nmr studies on @xmath3o - enriched powder samples are reported .
the temperature dependences of the @xmath3o nmr line width and nuclear spin - lattice relaxation rate give strong evidence for a long - range antiferromagnetic transition at @xmath4 k in the powder .
magnetic susceptibility measurements show that @xmath5 k in the crystals .
a zero - field @xmath0v nmr signal was observed at low temperatures ( @xmath6 mhz at 4.2 k ) in the crystals .
the field swept spectra with the field in different directions suggest the presence of two antiferromagnetic substructures .
each substructure is collinear , with the easy axes of the two substructures separated by an angle of @xmath7 , and with their average direction pointing approximately along the @xmath8-axis of the crystal structure .
the two spin substructures contain equal number of spins .
the temperature dependence of the ordered moment , measured up to 45 k , shows the presence of an energy gap @xmath9 in the antiferromagnetic spin wave excitation spectrum .
antiferromagnetic spin wave theory suggests that @xmath10 lies between 64 and 98 k. |
in this section we explicitly give the circuits simulating the magnetization of a @xmath31-qubit spin chain using @xmath16 qubits .
we measure the magnetization of the spin chain at 12 equidistantly distributed values of @xmath32 . in particular , we choose @xmath33 , as in @xcite .
we also choose the parameters of the adiabatic evolution , @xmath34 and @xmath35 .
see main text for an explanation of these parameters . as explained in the main text ,
we compute the unitary @xmath36 performing the whole adiabatic evolution and decompose this unitary into the available gates set , as a step - wise implementation of the adiabatic evolution is not possible at the moment due to the current limit in circuit depth .
we entangle qubit 2 with an auxiliary qubit , qubit 3 , which is discarded afterwards in order to prepare @xmath37 on qubit 2 . in each circuit
we measure qubit 1 in order to obtain the magnetization @xmath38 .
the explicit circuits for each value @xmath32 are given in figure [ fig : cheatingcircuits ] .
in the following , we argue that with the current version of the ibm quantum computer it seems unfeasible to run the compressed simulation of the ising spin chain using three or more qubits and hence , simulating a eight or more - qubit spin chain .
nevertheless , we show that the computation will become possible once several improvements that ibm announced are implemented . at the moment , performing the computation using three or more qubits seems not possible due to the restriction in circuit depth , the limited gate set , and the fact that gates can not be implemented probabilistically .
we exemplarily show , that even preparing the initial state @xmath39 is a difficult task . to obtain the initial state , two of the qubits have to be prepared in a completely mixed state , while one qubit is prepared in @xmath40 .
see figure [ fig : preparation3 ] for a possible , but very uneconomical way to do so using a circuit of circuit depth six and consuming two auxiliary qubits that are discarded in the process .
note that there seems to be no less wasteful way to prepare @xmath39 as applying gates probabilistically is not possible at the moment . nevertheless , once the improvements that ibm announced are available , implementing the circuit for more qubits will become possible . here
, we exemplarily show how to implement the circuit for three qubits , the method can be generalized to more qubits , though . to this end , we assume that the following improvements are available .
we assume that advanced classical processing is available .
in particular , we assume that it is possible to apply gates probabilistically .
furthermore , we assume that arbitrary single qubit gates are available and that subroutines are available , i.e. , user - defined gates can be declared and used . in this case
the circuit can be implemented as follows .
the initial state @xmath41 is prepared by performing either a pauli @xmath4 or @xmath37 with probability @xmath42 on both of the qubits for which we want to prepare @xmath43 individually ( which we will denote as qubits 1 and 2 in the following ) , and furthermore performing a single qubit unitary that rotates @xmath3 to @xmath40 for the remaining qubit , which we will denote as qubit 3 in the following .
after the initial state is prepared , the system is evolved adiabatically . in each step of this adiabatic evolution the unitary @xmath44 has to be applied .
the unitaries @xmath45 , @xmath46 , and @xmath47 are given in the main text .
the unitary @xmath48 is a single qubit unitary and hence can be implemented easily .
we have @xmath49 , where @xmath50 denotes a gate @xmath51 controlled by qubits @xmath52 .
a possible implementation of this controlled phase gate is depicted in figure [ fig : ccphase ] @xcite . recall that the two swaps can be implemented using three @xmath13 gates , while phase gates that are controlled by one qubit may be decomposed into two controlled not gates and three single qubit unitaries @xcite .
implementing @xmath46 is more tricky .
first , one performs a basis transformation by applying @xmath53 , where @xmath54 . in the new basis the unitary @xmath46
is given by the unitary @xmath55 , where @xmath56 followed by a single qubit unitary @xmath57 @xcite .
finally , the basis change has to be undone , i.e. , @xmath19 is applied .
the controlled rotation can be implemented in a similar way as shown above for @xmath45 . in order to implement @xmath19
, we use that this unitary can be decomposed into a toffoli gate @xmath58 followed by a @xmath59 , and a pauli @xmath60 @xcite .
this circuit can be further decomposed using the decomposition of the toffoli gate suitable for the ibm quantum computer @xcite and some simplifications , yielding the circuit depicted in figure [ fig : agate ] .
this circuit can be packed into a user - defined three - qubit gate depending on the free parameter @xmath61 and the adiabatic evolution is performed by applying these gates with increasing @xmath62 successively . finally , measuring @xmath6 on qubit 3 yields the magnetization of the eight - qubit spin chain . as ibm announced , that advanced classical processing , arbitrary single qubit unitaries , and user - defined gates will become available in future , implementing the circuit for three or more qubits will become feasible
, as long as the number of steps ( recall that we used l=2400 steps before ) is not an issue .
otherwise , similar methods as those used in the two qubit circuit will have to be applied . | the notion of compressed quantum computation is employed to simulate the ising interaction of a 1d chain consisting out of @xmath0 qubits using the universal ibm cloud quantum computer running on @xmath1 qubits .
the external field parameter that controls the quantum phase transition of this model translates into particular settings of the quantum gates that generate the circuit .
we measure the magnetization , which displays the quantum phase transition , on a two qubit system , which simulates a four qubit ising chain , and show its agreement with the theoretical prediction within a certain error .
we also discuss the relevant point of how to assess errors when using a cloud quantum computer . as a solution
, we propose to use validating circuits , that is to run independent controlled quantum circuits of similar complexity to the circuit of interest .
the faithful simulation of quantum systems remains one of the most interesting problems that can be addressed with a full - fledged quantum computer .
phenomena such as superconductivity in two dimensions , highly frustrated condensed matter systems or the effect of topology in quantum systems are out of the reach of classical simulation .
the emergence of new designs for quantum computation further motivates more detailed studies of mapping quantum problems to realistic quantum computation .
a particular instance of relevant physics that can be addressed with a quantum computation is the study of quantum phase transitions @xcite .
indeed , some systems undergo a quantum phase transition which is characterized by large quantum correlations at zero temperature . at their critical point , conformal symmetry is restored and correlations decay algebraically and become long - ranged .
furthermore , the entanglement entropy of the ground state of the system diverges in the thermodynamic limit at the phase transition . in general
, such a large amount of entanglement can not be described correctly in two dimensions by classical means . several experimental set
ups are currently employed to investigate quantum phase transitions @xcite .
this poses the problem of designing refined experiments , which have so far tended to exploit the avenue of quantum simulation , rather than quantum computation .
one of the reasons for that is that universal quantum computation , which can be used to simulate any system , is currently restricted to approximately ten qubits @xcite .
however , as we will also exploit here , certain simulations can be compressed and run on an exponentially smaller universal quantum computer .
a particularly interesting new approach to the use of quantum computers is the advent of cloud quantum computation .
the free access to run quantum circuits on a remote cloud computer opens the door to design new algorithms , to improve them by trial and error , and to refute or consolidate non - obvious ideas .
it may be argued that cloud quantum computation plays a similar role to the introduction of personal computers or open mainframes in the early stages of informatics . at present ,
the availability of cloud quantum computation is limited to the ibm quantum experience project @xcite .
it is a universal five - qubit quantum computer based on superconducting transmon qubits .
the ibm quantum computer has already been tested in various ways , e.g. , how well it performs in violating mermin inequalities @xcite . moreover ,
error correction codes , fourier addition , preparing graph states , and fault tolerant circuit design have been considered @xcite . in the present paper
we test the performance of the ibm quantum computer with a compressed simulation of the transverse field 1d ising interaction .
the quantum ising model is an integrable model and an exact circuit can construct its ground state @xcite .
moreover , the notion of compressed quantum computation @xcite can be employed to simulate the ising chain of @xmath0 qubits by using only @xmath1 qubits @xcite .
it has also been applied to the xy - model and compressed quantum metrology @xcite . moreover ,
the compressed simulation of the ising spin chain ( consisting of @xmath2 qubits ) has been realized in an experiment using nmr quantum computing @xcite . on the available cloud quantum computer ,
it is now possible to simulate a four qubit ising chain utilizing only two qubits . in order to realize this computation
, we decompose the circuits for the compressed simulation into the available gate set .
we run these circuits on the quantum computer and measure the order parameter that displays the quantum phase transition . given that the size of the system is finite
, we do observe smoothed changes of the order parameter that agree with the theoretical predictions within errors .
an important aspect to be addressed here is the assessment of errors .
there are two sources of errors that have to be considered separately .
first , it is necessary to run an experiment often enough so that statistical errors are reduced .
this is an easy task since it only implies repetition of experiments .
second , systematic errors must be estimated .
the situation here is particularly subtle , as a cloud computer is run by teams unrelated to its users .
the problem of how to estimate a systematic error without knowing the detail of the computer is non - trivial .
nevertheless , an approach to the correct assessment of systematic errors can be done , using independent controlled circuits of similar complexity to the one of interest .
this idea of estimating systematic errors produced by a black box might be of relevance for all future cloud quantum computation .
the outline of the remainder of the paper is the following .
we first discuss some of the features and constraints of the ibm quantum computer .
then , we analyze errors occurring when single gates are applied as well as errors in more complex circuits . there , we introduce the concept of a validating circuit set in order to estimate the error of a quantum computation in case the user does not have direct access to the computer . after reviewing then the notion of compressed quantum simulation and the explicit circuits for the compressed simulation of the ising interaction , we derive a two - qubit circuit suitable for the ibm quantum computer , which simulates a four - qubit spin chain .
we present and discuss the results of the simulation and show that they agree with the theoretical prediction within the error estimated before . in the appendix
, we outline how circuits simulating the eight and more - qubit spin chain can be constructed and argue that performing the computation will become possible once the announced improvements to the ibm quantum computer are implemented .
let us begin by discussing some of the features and constraints of the ibm quantum computer , which consists currently of five qubits @xcite .
the qubits are initiated in the computational basis state @xmath3 . as mentioned before ,
a limited but universal set of gates is available , namely the well known clifford+t set .
this set consists of the pauli operators ( @xmath4 , @xmath5 , and @xmath6 ) , the hadamard gate ( @xmath7 ) , phase gates ( @xmath8 , @xmath9 ) , @xmath10 gates ( @xmath11 , @xmath12 ) , as well as entangling controlled not gates ( @xmath13 ) .
measurements in @xmath6-basis as well as bloch vector measurements ( see below ) are performable .
currently , only limited classical control is available , e.g. , implementing gates probabilistically is not supported .
moreover , the depth of the circuit , i.e. , the number of gates that have to be applied successively and can not be parallelized , is limited to 39 . due to the architecture of the quantum computer , one qubit , which we denote as qubit 2 in the following , plays a special role .
it is the only qubit that can be target of a @xmath13 gate .
note that the gate set is nevertheless universal .
however , applying e.g. a @xmath13 gate between qubits @xmath14 and @xmath15 ( both different than @xmath16 here ) , which we denote by @xmath17 in the following , is very uneconomical .
one first needs to swap qubit @xmath15 and qubit @xmath16 ( which requires three @xmath13s ) , apply @xmath18 and swap again qubits @xmath15 and @xmath16 .
as computations are naturally subject to both systematic and statistical errors , ibm provides access to a classical simulator that implements an error model of the quantum computing hardware and therefore allows simulation of a circuit before actually performing the computation .
let us now investigate the errors , which occur in the computation .
we will first analyze the errors that occur after applying a single gate and then consider those , which occur in an actual quantum computation , involving many gates .
note that the maximum allowed number of runs of one computation is limited to 8192 , which allows to estimate the statistical error . in order to get an estimation of the error that occurs after applying a single gate from the gate set , we perform the following procedure .
we apply the single gate @xmath19 to the initial state @xmath20 .
ideally this would yield the state @xmath21 .
however , due to systematic errors , in the preparation as well as in the application of the gate , a state @xmath22 is obtained .
we perform tomography , which is of course also subject to both systematic and statistical errors , on the state @xmath22 .
that is , we perform three experiments measuring @xmath23 , @xmath24 , and @xmath25 with 8192 runs each . in order to measure @xmath4 and
@xmath5 the gates @xmath7 and @xmath26 are applied respectively prior the @xmath6 measurement .
an estimate , @xmath27 , of the state @xmath22 is then determined using the direct inversion method , i.e. , @xmath28 , where @xmath29 .
the fidelity , @xmath30 , of the estimate @xmath27 with respect to the ideal state @xmath21 is presented in table [ tab : fidelities ] for different choices of @xmath19 .
.fidelities of the estimate of the real state , @xmath27 , with respect to the ideal state @xmath21 . [ cols="^,^,^,^,^,^,^,^",options="header " , ] |
the gamma function , defined as @xmath8 , @xmath9 , has attracted much interest since its introduction by euler , appearing frequently in both mathematics and natural sciences problems .
the transcendental nature of this function at rational values of @xmath1 in the open interval @xmath10 , to which we shall restrict our attention hereafter , is enigmatic , just a few special values having their transcendence established .
such special values are : @xmath11 , whose transcendence follows from the lindemann s proof that @xmath12 is transcendental ( 1882 ) @xcite , @xmath13 , as shown by chudnovsky ( 1976 ) @xcite , @xmath14 , as proved by le lionnais ( 1983 ) @xcite , and @xmath15 , as can be deduced from a theorem of schneider ( 1941 ) on the transcendence of the beta function at rational entries @xcite .
the most recent result in this line was obtained by grinspan ( 2002 ) , who showed that at least two of the numbers @xmath16 , @xmath17 and @xmath12 are algebraically independent @xcite . for other rational values of @xmath1 in the interval @xmath10 ,
not even irrationality was established for @xmath18 .
the function @xmath19 , known as the log - gamma function , on the other hand , received less attention with respect to the transcendence at rational points . in a recent work , however , gun , murty and rath ( gmr ) have presented a `` theorem '' asserting that @xcite : [ conj : gmr ] the number @xmath20 is transcendental for any rational value of @xmath1 , @xmath21 , with at most * one * possible exception .
this has some interesting consequences . for a better discussion of these consequences ,
let us define a function @xmath22 as follows : @xmath23 note that @xmath24 , which implies that @xmath25 is symmetric with respect to @xmath26 . by taking into account the well - known _ reflection property _ of the gamma function @xmath27 valid for all @xmath28 , and being @xmath29 } = \log{\gamma(x ) } + \log{\gamma(1-x)}$ ]
, one easily deduces that @xmath30 } = \log{\pi } -\log{\sin{(\pi \ , x ) } } \ , .
\label{eq : seno}\ ] ] from this logarithmic expression , one promptly deduces that @xmath25 is differentiable ( hence continuous ) in the interval @xmath10 , its derivative being @xmath31 .
the symmetry of @xmath25 around @xmath26 can be taken into account for proofing that , being conjec .
[ conj : gmr ] true , the only exception ( if there is one ) has to take place for @xmath26 ( see the appendix ) . from eq .
( [ eq : seno ] ) , we promptly deduce that @xmath32 is transcendental for all rational @xmath1 in @xmath10 , the only possible exception being @xmath33 all these consequences would be impressive , but the proof presented in ref .
@xcite for conjec .
[ conj : gmr ] is _
incorrect_. this is because those authors implicitly assume that @xmath34 for every pair of distinct rational numbers @xmath35 in @xmath10 , which is not true , as may be seen in fig .
[ fig : fx ] , where the symmetry of @xmath25 around @xmath26 can be appreciated . to be explicit , let me exhibit a simple counterexample : for the pair @xmath36 and @xmath37 , eq .
( [ eq : seno ] ) yields @xmath38 and then @xmath39 . with @xmath40 ( i.e. , symmetric with respect to @xmath26 ) . ]
this _ null _ result clearly makes it invalid their conclusion that @xmath41 is a _ non - null _ baker period . here in this short paper ,
i take conjec .
[ conj : gmr ] on the transcendence of @xmath42 into account for setting up a theorem establishing that there are at most _ two _ possible exceptions for the transcendence of @xmath25 , @xmath1 being a rational in @xmath10 .
this theorem is proved here based upon a careful analysis of the monotonicity of @xmath25 , taking also into account its obvious symmetry with respect to @xmath26 .
interestingly , this yields a criteria for the transcendence of @xmath4 , an important number in the study of the algebraic nature of special values of a general class of @xmath43functions @xcite .
this reformulation of the gmr `` theorem '' allows us to exhibit an infinity of pairs of logarithms of certain algebraic multiples of @xmath12 whose elements are not both algebraic .
for simplicity , let us define @xmath45 as @xmath46 , i.e. the set of all rational numbers in the real open interval @xmath10 , which is a countable infinite set .
my theorem on the transcendence of @xmath44 depends upon the fundamental theorem of baker ( 1966 ) on the transcendence of linear forms in logarithms .
we record this as : [ lem : baker ] let @xmath47 be nonzero algebraic numbers and @xmath48 be algebraic numbers .
then the number @xmath49 is either zero or transcendental .
the latter case arises if @xmath50 are linearly independent over @xmath51 and @xmath48 are not all zero .
@xmath52 see theorems 2.1 and 2.2 of ref .
@xcite .
now , let us define a _ baker period _ according to refs .
@xcite .
[ def : bakerperiod ] a baker period is any linear combination in the form @xmath53 , with @xmath47 nonzero algebraic numbers and @xmath48 algebraic numbers . from baker s theorem , it follows that [ cor : naonulo ] any _ non - null _ baker period is necessarily a transcendental number .
now , let us demonstrate the following theorem , which comprises the main result of this paper .
[ teo : meu ] the number @xmath20 is transcendental for all @xmath54 , with at most * two * possible exceptions .
@xmath52 let @xmath25 be the function defined in eq .
( [ eq : fx ] ) . from eq .
( [ eq : seno ] ) , @xmath55 for all real @xmath56 .
let us divide the open interval @xmath10 into two adjacent subintervals by doing @xmath57 \ , \bigcup \ , [ \frac12,1)$ ] .
note that @xmath58 and thus @xmath25 is either a monotonically increasing or decreasing function in each subinterval . now , suppose that @xmath59 and @xmath60 are both algebraic numbers , for some pair of distinct real numbers @xmath61 and @xmath62 in @xmath63 $ ] .
then , the difference @xmath64 will , itself , be an algebraic number .
however , as the sine of any rational multiple of @xmath12 is an algebraic number @xcite , then lemma [ lem : baker ] guarantees that , being @xmath65 , then @xmath66 is either null or transcendental . since @xmath67 is a continuous , monotonically increasing function in @xmath68 , then @xmath69 for all @xmath70 in @xmath63 $ ] .
therefore , @xmath71 and then @xmath66 is a _ non - null _ baker period . from corol .
[ cor : naonulo ] , we know that non - null baker periods are transcendental numbers , which contradicts our initial assumption .
then , there is at most one exception for the transcendence of @xmath25 , @xmath72 $ ] .
clearly , as @xmath67 is a continuous and monotonically decreasing function for @xmath73 , an analogue assertion applies to this complementary subinterval , which yields another possible exception for the transcendence of @xmath25 , @xmath74 .
it is most likely that not even one exception takes place for the transcendence of @xmath44 with @xmath54 .
if this is true , it can be deduced that @xmath75 is transcendental .
if there are exceptions , however , then their number either one or two , according to theorem [ teo : meu ] will determine the transcendence of @xmath75 .
the next theorem summarizes these connections between the existence of exceptions to the transcendence of @xmath25 , @xmath54 , and the transcendence of @xmath75 .
[ teo : excecoes ] with respect to the possible exceptions to the transcendence of @xmath76 , @xmath54 , exactly one of the following statements is true : * there are no exceptions , hence @xmath4 is a transcendental number ; * there is only one exception and it has to be for @xmath26 , hence @xmath4 is an algebraic number ; * there are exactly two exceptions for some @xmath77 , hence @xmath4 is a transcendental number .
@xmath52 if @xmath78 is a transcendental number for every @xmath54 , item(i ) , it suffices to put @xmath26 in eq .
( [ eq : seno ] ) for finding that @xmath79 is transcendental .
if there is _ exactly one _ exception , item ( ii ) , then it has to take place for @xmath80 , otherwise ( i.e. , for @xmath77 ) the symmetry property @xmath81 would yield algebraic values for _ two _ distinct arguments , namely @xmath1 and @xmath82 .
therefore , @xmath83 is the only exception , thus it is an algebraic number .
if there are two exceptions , item ( iii ) , both for @xmath77 , then they have to be symmetric with respect to @xmath84 , otherwise , by the property @xmath81 , we would find more than two exceptions , which is prohibited by theorem [ teo : meu ] .
indeed , if one of the two exceptions is for @xmath26 , then the other , for @xmath77 , would yield a third exception , corresponding to @xmath85 , which is again prohibited by theorem [ teo : meu ] .
then the two exceptions are for values of the argument distinct from @xmath86 and then @xmath79 is a transcendental number . from this theorem , it is straightforward to conclude that the number @xmath4 is algebraic if and only if @xmath76 is a transcendental number for every @xmath54 , except @xmath80 .
the symmetry of the possible exceptions for the transcendence of @xmath44 around @xmath26 yields the following conclusion .
[ cor : logpares ] every pair @xmath5 } , \ , \log{[\pi/\sin(\pi y)]}\}$ ] , with both @xmath1 and @xmath6 rational numbers in the interval @xmath10 , @xmath87 , contains at least one transcendental number . by fixing @xmath80 in this corollary
, one has [ cor : pares ] every pair @xmath88}\}$ ] , @xmath6 being a rational in @xmath10 , @xmath89 , contains at least one transcendental number .
an interesting consequence of corol . [ cor : pares ] , together the famous hermite - lindemann ( hl ) theorem , is that the algebraicity of @xmath90 for some @xmath91 implies the transcendence of @xmath92 , another number whose irrationality is not established yet .
let me proof this assertion based upon a logarithmic version of the hl theorem .
[ lem : hl ] for any non - zero complex number @xmath93 , one at least of the two numbers @xmath93 and @xmath94 is transcendental .
@xmath52 see ref .
@xcite and references therein .
[ lem : hl1 ] for any positive real number @xmath95 , @xmath96 , one at least of the real numbers @xmath95 and @xmath97 is transcendental .
@xmath52 it is enough to put @xmath98 , @xmath95 being a non - negative real number , in lemma [ lem : hl ] and to exclude the singularity of @xmath97 at @xmath99 .
if the number @xmath90 is algebraic for some @xmath91 , then the number @xmath100 is transcendental .
@xmath52 let us denote by @xmath101 the set of all algebraic numbers and by @xmath102 the set of all non - null algebraic numbers .
first , note that @xmath103 for every @xmath91 and that , from eq .
( [ eq : seno ] ) , @xmath104}$ ] . now , note that if @xmath105 } \in \overline{\mathbb{q}}$ ] for some @xmath106 , then @xmath107}$ ] is also an algebraic number .
therefore , @xmath108 } = \log{[k(\tilde{y } ) \ , \pi \ , e ] } \in \overline{\mathbb{q}}$ ] and , by lemma [ lem : hl1 ] , the number @xmath109 has to be either transcendental or @xmath3 .
however , it can not be equal to @xmath3 because this would imply that @xmath110 , which is not possible because @xmath111 implies that @xmath112 .
therefore , the product @xmath109 is a transcendental number . since @xmath113
, then @xmath7 has to be transcendental .
in this work , the transcendental nature of @xmath44 for rational values of @xmath1 in the interval @xmath10 has been investigated .
i have first shown that the proof presented in ref .
@xcite for the assertion that @xmath20 is transcendental for any rational value of @xmath1 , @xmath21 , with at most _ one _ possible exception is incorrect .
i then reformulate their conjecture , presenting and proofing a theorem that establishes the transcendence of @xmath20 , @xmath1 being a rational in @xmath10 , with at most _ two _ possible exceptions .
the careful analysis of the number of possible exceptions has yielded a criteria for the number @xmath75 to be algebraic .
i have also shown that each pair @xmath5 } , \ , \log{[\pi/\sin(\pi y)]}\}$ ] , @xmath114 , @xmath87 , contains at least one transcendental number .
this occurs , in particular , with the pair @xmath88}\}$ ] , @xmath89 .
at last , i have shown that if @xmath115}\}$ ] is algebraic for some @xmath89 , then the product @xmath116 has to be transcendental .
the author acknowledges a postdoctoral fellowship from cnpq ( brazilian agency ) .
let us show that the assumption that conjec .
[ conj : gmr ] is true i.e. , that @xmath117 is transcendental with at most _ one _ possible exception , @xmath1 being a rational in @xmath10 implies that if one exception exists then it has to be just @xmath79 .
the fact that @xmath81 for all @xmath56 implies that , if the only exception would take place for some rational @xmath1 distinct from @xmath86 , then automatically there would be another rational @xmath82 , distinct from @xmath1 , at which the function would also assume an algebraic value ( in fact , the same value obtained for @xmath1 )
. however , conjec .
[ conj : gmr ] restricts the number of exceptions to at most _
one_. then , we have to conclude that if an exception exists , it has to be for @xmath26 , where @xmath25 evaluates to @xmath75 .
@xmath118 99 f. lindemann , sur le rapport de la circonference au diametre , et sur les logarithmes neperiens des nombres commensurables ou des irrationnelles algebriques , c. r. acad .
paris * 95 * ( 1882 ) 7274 .
g. v. chudnovsky , algebraic independence of constants connected with the exponential and elliptic functions , dokl .
nauk ukrain .
a * 8 * ( 1976 ) 698701 . f. le lionnais , les nombres remarquables , hermann , 1979 , p. 46 .
t. schneider , zur theorie der abelschen funktionen und integrale , j. reine angew
* 183 * ( 1941 ) 110128 .
p. grinspan , measures of simultaneous approximation for quasi - periods of abelian varieties , j. number theory * 94 * ( 2002 ) 136176 .
s. gun , m. ram murty , p. rath , transcendence of the log gamma function and some discrete periods , j. number theory * 129 * ( 2009 ) 21542165 .
s. gun , r. murty , p. rath , transcendental nature of special values of @xmath43-functions , canad .
, in press . a. baker , transcendental number theory , cambridge university press , 1975 , chap .
m. ram murty , n. saradha , transcendental values of the digamma function , j. number theory * 125 * ( 2007 ) 298318 .
m. kontsevich , d. zagier , periods , in : _
mathematics unlimited-2001 and beyond _ , springer - verlag , 2001 , pp . 771808 .
i. niven , irrational numbers , maa , 2005 , pp . 2930 .
g. p. dresden , a new approach to rational values of trigonometric functions .
available at http://arxiv.org/abs/0904.0826 a. baker and g. wstholz , logarithmic forms and diophantine geometry , cambridge university press , 2007 , sec .
} $ ] in the interval @xmath10 . since @xmath24 ,
the graph is symmetric with respect to @xmath26 .
note that , as @xmath119 for all @xmath56 , then @xmath120 , and then @xmath121 and the minimum of @xmath25 , @xmath1 being in the interval @xmath10 , is attained just for @xmath26 , where @xmath25 evaluates to @xmath75 .
the dashed lines highlight the coordinates of this point . ] | in a very recent work [ jnt * 129 * , 2154 ( 2009 ) ] , gun and co - workers have claimed that the number @xmath0 , @xmath1 being a rational number between @xmath2 and @xmath3 , is transcendental with at most _ one _ possible exception , but the proof presented there in that work is _
incorrect_. here in this paper , i point out the mistake they committed and i present a theorem that establishes the transcendence of those numbers with at most _ two _ possible exceptions . as a consequence ,
i make use of the reflection property of this function to establish a criteria for the transcendence of @xmath4 , a number whose irrationality is not proved yet .
i also show that each pair @xmath5 } , \ , \log{[\pi/\sin(\pi y)]}\}$ ] , @xmath1 and @xmath6 being rational numbers between @xmath2 and @xmath3 , contains at least one transcendental number .
this has an interesting consequence for the transcendence of the product @xmath7 , another number whose irrationality is not proved .
submitted to : _ acta aritmetica _ |
we used density - functional theory in conjunction with generalized gradient approximation for the exchange correlation functional@xcite and projector augmented waves@xcite for the c(@xmath43 ) electrons , as implemented in gpaw code using real - space grids@xcite .
converged energies were obtained with grid spacing of @xmath44 and @xmath45 @xmath46-points in the periodic direction . in the perpendicular directions the system is not periodic and the space between the atoms and the wall of the simulation cell was @xmath47 .
the energies were converged to @xmath48 ev / atom and structures were optimized until forces were less than @xmath49 ev / . our calculations agree well with previous relevant experimental as well as theoretical energetic and geometric properties@xcite .
the activation barrier was calculated with nudged elastic band method@xcite by fixing the atoms beyond the first two zigzag - rows with unit cell of length @xmath30 .
the constant current stm of fig .
[ fig : electr ] shows the height profile of electron density isosurface of the occupied states within @xmath50 ev energy window below the fermi - level .
the isosurface value corresponds to average density 2 above the graphene plane ( @xmath51 electrons / @xmath52 ) .
we acknowledge support from the academy of finland ( projects 121701 and 117997 ) and from the finnish cultural foundation .
we thank matti manninen for discussions , michael walter for technical assistance with gpaw calculations and karoliina honkala for reading of the manuscript .
the computational resources were provided by the finnish it center for science ( csc ) in espoo . | planar reconstruction patterns at the zigzag and armchair edges of graphene were investigated with density functional theory .
it was unexpectedly found that the zigzag edge is metastable and a planar reconstruction spontaneously takes place at room temperature
. the reconstruction changes electronic structure and self - passivates the edge with respect to adsorption of atomic hydrogen from molecular atmosphere .
carbon is one of the most prominent elements in nature , vital for biology and life .
although macroscopic carbon has been important since ancient times@xcite , only modern materials design , utilizing nanotubes@xcite , fullerenes@xcite and single graphene sheets@xcite , fully attempts to use its flexible chemistry . in applications for nanoscale materials and devices , it is often the atomic and electronic structure of boundaries and surfaces that is responsible for mechanical , electronic and chemical properties . since the properties of nanomaterial depend on the precise atomic geometry , its knowledge is crucial for focused preparation of experiments and for worthy theoretical modeling . only this enables the further development of nanoelectronic components , nanoelectromechanical devices and hydrogen storage materials@xcite , or the usage of carbon in compound designs@xcite .
the importance of precise geometry is emphasized in low - dimensional systems .
the strong correlations are known to bring up novel phenomena@xcite , and such should be expected also for the quasi - one - dimensional edges of graphene .
the edge chemistry plays even crucial role in the catalyzed growth of carbon nanotubes@xcite .
specifically , as two - dimensional carbon has the honeycomb lattice , edge behaviour ultimately boils down to the properties of graphene edges .
hence it is relevant to explore different graphene edge geometries and their chemical properties beyond the standard zigzag and armchair ones .
this relevance is evident from the abundant literature .
the electronic properties of graphene as well as carbon nanotube armchair and zigzag edges have been studied extensively@xcite , often in connection to nanotube growth@xcite or the so - called electronic `` edge states''@xcite .
there has been experimental and theoretical work done even on the reconstruction of graphene edges , but they have differed from the basic reconstruction patterns studied in this work . they have involved either edge roughness@xcite or more dramatic folding of the edge into a loop@xcite . , @xmath0 , @xmath1 and @xmath2 .
all geometries are strictly in plane.,width=321 ] the edges discussed in this work are shown in figure [ fig : edges ] , and were investigated by modeling an infinitely long carbon nanoribbon of a given width ( see methods ) .
tight - binding method@xcite was used to explore number of other edge candidates , but density - functional analysis only for the relevant ones is reported here .
the most important edge is zz(57 ) , a reconstruction of zigzag edge where two hexagons transform into a pentagon and a heptagon , like an edge cut through a haeckelite structure or a line of stone - wales defects@xcite .
the edge ac(677 ) is a reconstruction of the armchair edge where two separate `` armrest '' hexagons merge into adjacent heptagons by stone - wales mechanism .
the pentagonal reconstruction ac(56 ) of armchair has a slightly different nature , since it requires the diffusion of carbon atoms from distant `` armrests '' to `` seat '' positions .
let us start analysis by looking at edge energy @xmath3 , which is calculated from the total energy of the graphene ribbon @xmath4 where @xmath5 is the number of carbon atoms , @xmath6 the total length of edges ( twice the length of simulation cell ) , and @xmath7 ev is the cohesion energy of graphene .
the edge energies converge rapidly as shown in figure [ fig : width ] and justify the reference to ( semi - infinite ) graphene .
the energy of armchair edge is @xmath8 ev / lower than for zigzag edge , in accord with a similar value for nanotubes@xcite .
however , the principal result is that by edge reconstruction zigzag may lower its energy by @xmath9 ev / .
this implies that the _ reconstructed zigzag is the best edge for graphene_. to the best of our knowledge , this is the first report on the metastability of the zigzag edge , which is surprising in view of the abundant literature
. thermal stability of this novel reconstruction was also confirmed with tight - binding simulations@xcite .
the ac(677 ) edge has only slightly higher energy than the armchair edge , whereas ac(56 ) reconstruction has the highest edge energy .
we remark that the reconstructions appear to be stable with respect to out - of - plane motion , a situation somewhat different from small - diameter nanotubes@xcite .
.,width=302 ] these energetics can be understood by looking at the geometries of figure [ fig : edges ] .
previous studies have shown that armchair has low energy due to triple bonds in the `` armrests''@xcite , as realized by comparing their short bonds ( @xmath10 ) to the bond in acetylene ( @xmath11 ) .
zigzag does not have such triple bonds and ends up with strong and expensive dangling bonds .
the reconstructed zz(57 ) has triple bonds ( @xmath10 ) but also wider bond angles ( @xmath12 ) which reduces the hybridization energy cost .
triple bonds with wide angles can be observed also in ac(677 ) edge , but strain in other parts makes the reconstruction unfavoured .
the ac(56 ) edge suffers from dangling bonds like zigzag , and additional high strain energy makes this reconstruction the most expensive one .
regardless , ac(56 ) edge has relevance during the growth of armchair edges@xcite . with coverage of one hydrogen per edge atom and lower
faint symbols are shifted by subtracting the h@xmath13 binding energy ; positive @xmath14 means the hydrogen atom is more strongly bound to the edge than to h@xmath13 molecule.,width=302 ] these observations are supported by hydrogen atom adsorption energies , shown in figure [ fig : ads ] .
the weak adsorption for armchair ( @xmath15 ev ) compared to zigzag ( @xmath16 ev ) stems from the triple bonds in the armchair edge . similarly the weak adsorption for zz(57 ) ( @xmath17 ev ) and ac(677 ) ( @xmath18 ev ) witnesses the weakening of dangling bonds .
the adsorption for ac(56 ) is large because of the strongest dangling bonds . for insight
, the adsorption energies in figure [ fig : ads ] are replotted by subtracting hydrogen molecule binding energy @xmath19 ev . the resulting negative number for zz(57 ) means that the adsorption of hydrogen atom from h@xmath13 molecule is not favored due to cost of h@xmath13 dissociation energy , unless the adsorption process should be complicated@xcite .
this amounts to the conclusion that _ the edge reconstruction chemically passivates zigzag edge_. however , hydrogen adsorption for zigzag@xmath20hydrogen edge has yet smaller adsorption energy of @xmath21 ev .
the edge energetics are summarized in table [ tab : ads ] .
note that the ordering of edges changes when expressed as energy per edge atom ( @xmath22 ) due to different edge atom densities @xmath23 and @xmath24 .
more interesting is to look at edges with hydrogen termination ( klein edge)@xcite . for this case
the edge energy is @xmath25 , where the reference is to bulk graphene and h@xmath13 molecules .
the best edges in this case are normal armchair and zigzag edges because of dangling bonds . on the contrary ,
the weak dangling bonds and small adsorption energy causes high energies for zz(57 ) and ac(677 ) klein edges .
.summary of the edge and hydrogen adsorption energies .
@xmath26 is the energy per edge atom , @xmath27 the hydrogen adsorption energy with full edge coverage , and @xmath28 is the edge energy with hydrogen termination .
note that for ac(56 ) the edge atom density is @xmath29 .
[ cols="<,^,^,^,^,^",options="header " , ] wide zigzag and zz(57 ) nanoribbons , respectively , with unit cell width of @xmath30 .
note that for zigzag this is twice the minimum unit cell and the reciprocal space is thus only half of the normal representation . the dashed line is the fermi - level .
the bands coloured were identified directly by visual inspection of the wave functions .
( c ) and ( d ) show the height profiles of simulated scanning tunneling microscope images in constant current mode of the respective edges ( height variations @xmath31
) , formed by integrating the electron density from occupied bands within 0.1 ev of the fermi energy .
the degeneracies at the gamma - point are 2 and 4 for the dangling bonds and the flat band , respectively.,width=321 ] let us now concentrate on the thermodynamic and electronic properties of zigzag edges .
we would like to clarify an aspect which is ambiguous in the literature : there are two types of electronic zigzag edge states . the so - called `` flat band '' in figure [ fig : electr]a comes from the bulk @xmath32-electrons and is indeed localized at the edge .
but the band due to dangling bonds is the one seen in the scanning tunneling microscope ( stm ) image and is located spatially even beyond the edge ( figure [ fig : electr]c ) . in zz(57 ) formation of triple bonds is evidenced by the nearly isolated dimers in figure [ fig : electr]d , and the reconstruction removes the dangling bond bands away from the fermi - level by lifting the degeneracy almost by @xmath33 ev .
hence for zz(57 ) the stm image shows only the `` flat band '' states . because the dangling bond bands shift to elusive energies , also chemical reactivity reduces . in thermodynamic sense the spontaneous reconstruction of zigzag into zz(57 ) should be possible , since the activation barrier from the zigzag side is only @xmath34 ev , from the reconstructed side @xmath35 ev .
the g - mode vibration of graphene at @xmath36 @xmath37 gives an attempt frequency of @xmath38 s@xmath39 , and elementary approach yields the rapid rate @xmath40 s@xmath39 at room temperature .
the reconstructions predicted in this work are expected to survive on graphite terraces due to the weak interaction ( @xmath41 mev / atom ) between the basal planes@xcite . using appropriate sample preparation it should be thus possible to observe the reconstruction .
stm images often show irregular and blurred edges , yielding no atomic resolution , but at least for passivated edges armchair predominance is claimed , in agreement with table [ tab : ads]@xcite . so far samples have been prepared intentionally with hydrogen passivation during heat treatment@xcite , a situation where reconstruction would not be favored .
alternate routes for observing the reconstruction would be the radial distribution function from neutron diffraction experiments without deuterium atmosphere@xcite , or detection of triple bond -spawned high - energy modes around @xmath42 @xmath37 with raman spectroscopy .
a further topic is the study of the conductance of various graphene edges , particularly the zz(57 ) edge found here .
the presence of the edge state around the fermi level makes zz(57 ) ideal for conductance measurements , in contrast to armchair ribbons where the edge state is absent .
this may render zz(57 ) as an interesting stable model for quasi - one - dimensional carbyne with alternating single and triple bonds .
furthermore , the novel thermodynamically and chemically stable reconstruction could play a role in formation of angular joints in nanoribbons@xcite , closure of the ends of nanotubes after cutting@xcite and any other system where graphene sheets are joined to produce systems with nano - scale morphology . |
polarons as quasiparticles , in which the electronic excitations are strongly coupled to the lattice modes , have been studied theoretically for well over 50 years .
nearly all these studies have been confined to models with only one or two electrons , and do not include the electron spin .
such models have limited applicability and there is a need for studies of more realistic many - electron situations@xcite .
there are technical problems , however , in handling such models as non - perturbative methods are required to describe the effects of strong electron - lattice interactions and few such methods are applicable to the many - electron case .
progress has been made in recent years in dealing with models with strong local interactions via the dynamical mean field theory ( dmft)@xcite .
calculations based on the holstein and holstein - hubbard models , where the electrons are coupled to local einstein lattice modes , have been made using the dmft to examine the effect of strong electron - phonon interactions on the mott transition@xcite , for example .
the holstein model@xcite , however , was originally introduced to describe small polaron effects .
within dmft , only the single electron case can be solved exactly@xcite .
there have been few studies at finite electron densities and including the electron spin@xcite , except in one dimension@xcite .
the problem is that if spin is included in the holstein model , there is an effective attractive local interaction between the electrons , which leads to local bipolaron formation and , away from half filling , superconductivity .
once the electron spin is included the more realistic model is the holstein - hubbard model which includes an local electron repulsion @xmath2 . in most situations the repulsion @xmath2 dominates and undermines any tendency to the formation of local bipolarons . in the holstein - hubbard model at half filling with a large @xmath2 , the electron - phonon interaction plays no important role as charge fluctuations are suppressed ; its main effect is to renormalise and reduce the effective value of @xmath2@xcite . at half filling and
small @xmath2 , where the electron - phonon coupling dominates , again there is little evidence@xcite of purely polaronic effects , because of the formation of bipolarons@xcite .
the question arises , therefore , whether there is any parameter regime where purely polaronic effects can be observed .
if there is , then does the behaviour correspond to the predictions based on the spinless models with only one or two electrons , or do many - electron correlations play a significant role ? in this letter we present calculations to show clearly that small strongly renormalized polaronic excitations do exist in the holstein - hubbard model and form a distinct narrow band at the fermi level .
the hamiltonian for this model is @xmath3 where @xmath4 describes the dispersion of the band electrons and @xmath2 is the on - site inter - electron repulsion as in the hubbard model@xcite .
the electron density @xmath5 at site @xmath6 couples linearly to the local displacement operator @xmath7 with an electron - phonon coupling @xmath8 .
the phonons are assumed to be dispersionless ( local einstein phonons ) with energy @xmath9 and @xmath10 is the mass of the vibrating ions .
our calculations are based on dmft using a semi - elliptical density of states @xmath11 of the non - interacting system given by @xmath12 for @xmath13 and zero elsewhere , corresponding to a bethe lattice , where @xmath14 is the bandwidth and @xmath15 is the chemical potential .
the effective impurity problem is solved using the numerical renormalization group ( nrg)@xcite , as generalized for the calculation of dynamical response functions@xcite , and incorporating the density matrix generalization@xcite due to hofstetter . to suppress any tendency for bipolaron formation
we take a large value of the on - site repulsion , @xmath16 in units where @xmath17 , which for half filling corresponds to a mott insulator @xcite .
we choose a value of @xmath18 corresponding to quarter filling @xmath19 .
the local density of states for @xmath20 is shown in fig . [
fig : g_spec_u60_x50 ] , with the fermi level such that the lower hubbard band is @xmath21 filled . in this regime
the strong correlation effects are largely suppressed due to the charge fluctuations ; there is no sharp quasiparticle peak at the fermi level , which occurs for the small doping regime near half filling , and there is a moderately reduced quasiparticle renormalization factor @xmath22 .
the small spectral weight between the upper and lower bands is due to the tails in the broadening in the higher energy peaks of the discrete nrg results . to study the impact of electrons coupling to phonons we concentrate on the lower hubbard band . in fig .
[ fig : g_spec_u60_gx_x50 ] we show the effect of increasing the electron - phonon interaction @xmath8 , taking @xmath23 in all cases . as @xmath8 increases to the value @xmath24 a sharp feature develops at the fermi level which we attribute to a polaronic renormalization .
due to the large @xmath2 , there is negligible double occupancy @xmath25 for this range of @xmath8 indicating no local bipolaron formation .
to look at the band aspects of these excitations , we introduce the @xmath27-resolved one - electron green s function @xmath28 with a local self - energy @xmath29 , @xmath30 figure [ fig : g_spec_k_u60_g60_x50 ] shows a plot of the spectral density of the green s function @xmath31 for @xmath32 and @xmath33 . when integrated over @xmath27 , it gives the local density of states for the corresponding @xmath8 shown in fig .
[ fig : g_spec_u60_gx_x50 ] .
we plot the positions of the maxima in the quasiparticle peaks as a function of @xmath4 in fig .
[ fig : qp_band_u60_gx_x50 ] .
the flat dispersion in the region of the fermi level is clear evidence of a narrow polaronic band ( full line ) .
we see a kink at @xmath34 which we discuss later in relation with the phonon spectra .
none of this is seen in the pure hubbard case , shown as the dashed line in fig .
[ fig : qp_band_u60_gx_x50 ] . of the electron green s function for the quarter - filled holstein - hubbard model ( @xmath26 , full line ) and the quarter - filled hubbard model ( @xmath35 , dashed line ) .
the arrow indicates @xmath36.,scaledwidth=35.0% ] the locus of points @xmath37 on the fermi surface of the interacting system is given by @xmath38 according to luttinger s
theorem@xcite the volume of the fermi surface should be the same as that of the non - interacting system . for a local self energy @xmath39
, this implies that the fermi surface given by eq .
( [ eq : fs ] ) should be the same as the non - interacting one , @xmath40 . if these are to coincide , then @xmath41 . to check the theorem we have calculated the density @xmath42 directly from the nrg expectation value in the ground state , and then from the non - interacting fermi surface using the value of @xmath41 from equation ( [ eq : fs ] ) .
the two sets of results are compared for a range of hole dopings in fig .
[ fig : n_vs_mu_gx.eps ] ( left ) for the pure hubbard model , and in fig .
[ fig : n_vs_mu_gx.eps ] ( right ) for the holstein - hubbard model with @xmath43 .
the agreement between the two sets of results provide a convincing evidence of the applicability of luttinger s theorem in strongly correlated systems both with and without the electron - phonon interaction .
if we assume fermi liquid theory@xcite and expand the self - energy in eq .
( [ eq : gf ] ) about the fermi level @xmath44 , and retain the first two terms , we can define a quasiparticle green s function @xmath45 where @xmath46 , and @xmath0 is the usual wavefunction renormalization factor @xmath47 , and @xmath48 is a renormalized chemical potential .
the corresponding density of states @xmath49 for the non - interacting quasiparticles is given by @xmath50 where @xmath51 plays the role of a renormalized band width .
we interpret this as a renormalized band of free polaronic quasiparticles .
integrating this free quasiparticle band up to the fermi level gives the quasiparticle occupation @xmath52 .
the resulting @xmath53 , demonstrating the one - to - one correspondence of the electron and quasiparticle states of fermi liquid theory@xcite ( see fig . [
fig : n_vs_mu_gx.eps ] ) . calculated directly ( full lines ) and from the volume of the fermi surface ( circles ) for the hubbard model ( left - hand plot , @xmath16 ) and the holstein - hubbard model ( right - hand plot , @xmath54).,scaledwidth=45.0% ] in fig .
[ fig : z_u60_gx_x50 ] we give the results for the change in the quasiparticle renormalization factor @xmath0 with increase of @xmath8 for quarter filling and @xmath16 .
the values of @xmath0 we calculated by differentiating the self - energy , and also from an analysis of the renormalized parameters@xcite on the approach to the low energy fixed point .
the two types of estimates are in reasonable agreement ; the deviation for smaller values of @xmath8 is largely due to fluctuations in the numerical differentiation of @xmath39 in this range .
the behaviour of @xmath0 with increase of @xmath8 is quite different from that near half filling where @xmath0 increases with increase of @xmath8 . near half filling the strong correlation effects stem from the enhancement of spin fluctuations and the suppression of charge fluctuations due to the large value of @xmath2 . in this regime
the increase of @xmath0 with @xmath8 can be understood as due to the reduction in the effective value of @xmath2 , @xmath55 ; hence an increase in @xmath8 reduces @xmath56 and hence reduces the renormalization@xcite . the results for @xmath0 at quarter filling shown in fig .
[ fig : z_u60_gx_x50 ] , by contrast , show a large decrease of @xmath0 with increase of @xmath8 .
this increased renormalization is a pure polaronic effect ; the coupling to the lattice causing an increase in the effective mass of the conduction electrons and a reduced mobility .
the renormalized parameter approach@xcite also allows us to calculate the local interaction between the quasiparticles @xmath1 .
the results for @xmath1 over this range are shown in fig .
[ fig : utilde_u60_gx_x50 ] .
there is a progressive reduction with increase of @xmath8 as might be expected for the induced attactive interaction due to phonon exchange .
for the largest value of @xmath8 the value of @xmath1 becomes negative , even though @xmath60 .
this can be interpreted as due to the fact the very low energy quasiparticles hardly ever occupy the same site due to correlation effects ( double occupancy @xmath61 ) and their effective attraction is due to retardation effects . for values of @xmath62
the dynamical mean field theory no longer converges satisfactorily , the iterations fluctuate between occupied and unoccupied states .
this indicates the on - set of a broken symmetry state , probably some form of charge ordering or phase separation .
this instability is also revealed in the one - phonon propagator @xmath63 whose spectral density is shown in fig . [
fig : b_spec_u60_gx_x50 ] . with increasing @xmath8
there is a progressive softening of the phonon mode from its bare value @xmath23 . for @xmath64 considerable spectral
weight develops for @xmath65 .
this weight relates to the expected number of excited phonons in the ground state .
it clearly indicates an incipient lattice instability for @xmath62 .
the kink to be seen at @xmath66 in the quasiparticle dispersion in fig .
[ fig : qp_band_u60_gx_x50 ] is clearly related to the frequency of the renormalized phonon excitation at the corresponding @xmath67 ( fig .
[ fig : b_spec_u60_gx_x50 ] ) .
an analysis for different values of @xmath8 shows a clear correlation of the kink energy to the _ renormalized _ phonon frequency .
this is similar to the experimentally observed kink in the hole - doped copper oxides@xcite . to summarize : we have shown that an accurate treatment of the holstein - hubbard model at _ finite electron density _ and @xmath68 reveals a clear polaronic quasiparticle band induced at the fermi level .
luttinger s theorem is shown to be satisfied .
a kink found in the quasiparticle dispersion relation is related to the renormalized phonon frequency . | we show that a strongly renormalized band of polaronic quasiparticle excitations is induced at the fermi level of an interacting many - electron system on increasing the coupling of the electrons to local phonons .
we give results for the local density of states at zero temperature both for the electrons and phonons .
the polaronic quasiparticles satisfy luttinger s theorem for all regimes considered , and their dispersion shows a kink similar to that observed experimentally in copper oxides .
we calculate the quasiparticle weight factor @xmath0 and deduce the local effective inter - quasiparticle interaction @xmath1 .
our calculations are based on the dynamical mean field theory and the numerical renormalization group for the hole - doped holstein - hubbard model and large on - site repulsion . |
quantum electrodynamics ( qed ) was the first quantum field theory to be formulated and has successfully passed every experimental test at low and intermediate fields .
a well - known example of qed effects at low fields ( @xmath0 v / cm ) is the lamb shift in hydrogen @xcite . at low fields , the qed effects ( self - energy and vacuum polarisation )
can still be treated as a perturbation , only taking into account lower order terms @xcite . however , up to now qed calculations have never been tested at high fields ( @xmath1 v / cm ) because such fields can not be produced in a laboratory , nor by the strongest lasers available . at high fields , perturbative qed is no longer
valid and higher order terms become important as well @xcite .
experiments carried out at high fields therefore test different aspects of qed calculations and are complementary to high - precision tests of the lower order terms .
heavy atoms that have been stripped of almost all their electrons , the so - called highly - charged ions ( hci ) , are ideal ` laboratories ' for tests of qed at high fields .
these ions have , for example , electric field strengths of the order of @xmath2 v / cm close to the nucleus @xcite and can be produced at high velocities at the gesellschaft fr schwerionenforschung ( gsi ) in darmstadt , germany . at the hitrap facility , which is currently being built at gsi
, ions coming from the experimental storage ring ( esr ) with mev energies will be slowed down by linear and radiofrequency stages to kev kinetic energies , trapped and cooled down to sub - ev energies , and finally made available for experiments . within the hitrap project
, instrumentation is being developed for high - precision measurements of atomic and nuclear properties , mass and g - factor measurements and ion - atom and ion - surface interaction studies @xcite .
hydrogen- and lithium - like ions are the best candidates for our studies , since they have @xmath3-electrons which are very close to the nucleus .
the ( higher order ) qed effects are most pronounced at the high fields close to the nucleus , therefore the best measurable quantity is the ground state hyperfine splitting ( hfs ) . due to the simple electronic structure of h- and li - like species , accurate ( higher order )
calculations of ground state hfs can be done , which will then be compared with accurate experimental results . as a first approximation , good within about 4% ,
the energy of the @xmath4s@xmath5 ground state hfs of hydrogen - like ions is given by @xcite : @xmath6 where @xmath7 is the fine structure constant , @xmath8 is the nuclear @xmath9-factor ( with @xmath10 the nuclear magnetic moment and @xmath11 the nuclear magneton ) , @xmath12 the nuclear spin , @xmath13 and @xmath14 are the electron and proton mass , respectively , and @xmath15 is the speed of light .
equation ( [ eq1 ] ) represents the normal ground state hfs multiplied by a correction @xmath16 for the relativistic energy of the @xmath3-electron , and by a factor @xmath17 , which takes the ` breit - schawlow ' ( bs ) effect into account .
the bs effect is due to the spatial distribution of the nuclear charge .
it corrects for the fact that we can not assume a homogeneous charge distribution over a spherical nucleus .
the values for @xmath18 were taken from @xcite , those for @xmath19 and @xmath12 from @xcite . in principle
eq.([eq1 ] ) should also contain a correction for the finite nuclear mass , but since this correction is very small it can be neglected @xcite .
the energy of the @xmath20s@xmath5 ground state hfs of lithium - like ions only differs from eq.([eq1 ] ) by a factor @xmath21 and by the @xmath16-value @xcite .
however , eq.([eq1 ] ) requires two further important corrections , the one of most interest to us being that which corrects for the qed effects .
the other correction takes the ` bohr - weisskopf ' ( bw ) effect into account @xcite .
the bw effect is due to the spatial distribution of the nuclear magnetisation and is only known with an accuracy of 20 - 30 % , which is mainly due to the single - particle model used for its calculation @xcite .
unfortunately , the qed effects are of the same order of magnitude as the uncertainty in the bw effect @xcite .
thus , from a hfs measurement of a single species ( _ i.e. _ h- or li - like ) the qed effects can not be determined accurately . equation ( [ eq1 ] ) can also be written as @xmath22 , where the constant @xmath23 includes everything except the qed effects . since the equations for the @xmath24 and @xmath25 states are so similar , it is possible to write the difference between the two hfs as @xmath26 @xcite .
the factor @xmath27 only contains non - qed terms and can be calculated to a high precision @xcite . from the difference between the hfs measurements of h- and li - like ions of the same isotope , the qed effects can thus be determined within a few percent .
however , this requires measurements of transition wavelengths with an experimental resolution of the order of @xmath28 .
the transition lifetime @xmath29 is defined as @xmath30 ( see _ e.g. _ @xcite ) .
the transition probability @xmath31 , for an @xmath32 transition from the excited to the lower hyperfine state , is given by @xcite @xmath33 where @xmath34 is planck s constant divided by @xmath35 and @xmath36 is related to the electron s angular momentum @xcite . from eq.([eq2 ] ) it can be seen that @xmath31 scales with the transition frequency as @xmath37 , whereas @xmath38 is proportional to @xmath39 , see eq.([eq1 ] ) .
therefore , the transition lifetime scales with @xmath40 as @xmath41 and is roughly of the order of milliseconds for @xmath42 .
.calculated hfs transition wavelengths ( @xmath43 ) and lifetimes ( @xmath29 ) of the most interesting ion species for systematic studies .
also shown are the nuclear spin ( @xmath12 ) and magnetic moment ( @xmath10 ) , taken from @xcite .
the half - lives of these species are longer than 10 minutes .
( the values listed are truncated and the qed and bw effects are not included . ) [ cols="<,<,^,^,^,^,^",options="header " , ] [ tab1 ] in table [ tab1 ] the calculated transition wavelengths ( @xmath43 ) and lifetimes ( @xmath29 ) , together with their corresponding @xmath12 and @xmath10 values , of the most interesting species for our laser spectroscopy studies are listed .
( the qed and bw effects are not taken into account . )
the half - lives of these species exceed 10 minutes , which corresponds to the minimum time required for a measurement .
although the wavelengths span a broad range , roughly from 200 to 1600 nm , these transitions are still accessible with standard laser systems .
the three species ( pb @xcite , cl @xcite and ar @xcite ) at the bottom of the table are considered as candidates for pilot experiments .
they are singly charged ions , which are easily produced , have @xmath32 transitions at convenient wavelengths , and can be used to test the laser spectroscopy part of the experiment .
a measurement of the hfs in @xmath44pb@xmath45 is of special interest , because it will be possible to extract the value of @xmath10 .
currently two different values exist , which unfortunately leads to a 2% difference in the hfs calculations @xcite . in principle ,
similar experiments could be carried out with metastable hafnium ( @xmath46hf , level energy 1141 kev , half - life 5.5 h @xcite ) . for h - like hafnium ,
the transition values are @xmath47 nm and @xmath48 ms . for the li - like ion , @xmath49 nm and @xmath50
ms are obtained .
the difficulty with this isotope is that its nucleus is in an excited state , which is difficult to produce .
figure [ fig1 ] shows the calculated transition wavelengths of all h - like lead , and all h- and li - like bismuth isotopes with half - lives exceeding 10 minutes .
( the qed and bw effects are not included . ) the isotopes are labelled by their corresponding atomic mass units ( in @xmath51 ) and the stable isotopes ( @xmath44pb and @xmath52bi ) are indicated by the small arrows . for pb ,
only the h - like isotopes are accessible with standard lasers , because the transition wavelengths of li - like isotopes are much longer than 1600 nm . for bi , many isotopes of both ion species are accessible , although their transition wavelengths differ considerably . from fig.[fig1
] it is clear that both elements offer many candidates for laser spectroscopy measurements of ground state hfs and that bismuth , in particular , allows for a systematic study of the ( higher order ) qed effects at high fields .
furthermore , a systematic study of different isotopes of the same species , for example a study of the h - like pb isotopes , will make it possible to study trends in nuclear properties across a range of isotopes . .
( the qed and bw effects are not included.),width=283 ] there already exist two previous measurements of the @xmath53 ground state hfs in @xmath52bi .
a direct measurement @xcite was carried out using the esr at gsi ( darmstadt ) , but unfortunately no resonance was found at the predicted value of @xmath54 nm @xcite . an indirect measurement @xcite was performed in an electron beam ion trap ( ebit ) and yielded a value of @xmath55 nm , but the error in the measurement was rather large ( @xmath56 nm ) . in the esr
the ions have relativistic velocities ( @xmath57 mev / u ) , which are used to shift the transition wavelength to a lower value ( @xmath58 nm ) , and the transitions are doppler - broadened ( @xmath59 ghz ) . in the ebit
the ions have temperatures of several hundreds of ev ( @xmath60 k ) , which lead to considerable doppler broadening ( @xmath61 ghz ) .
the resolution obtained in previous measurements at the esr is of the order of @xmath62 , whereas that of the ebit measurement is of the order of @xmath63 .
a detailed description of the proposed experiments , as well as a treatment of the techniques used , can be found elsewhere @xcite .
briefly , an externally produced bunch of roughly @xmath64 hci at an energy of a few kev is loaded into a cylindrical open - endcap penning trap @xcite on axis , _ i.e. _ along the magnetic field lines .
electron capture ( neutralisation ) by collisions is strongly reduced by operating the trap at cryogenic temperatures under uhv conditions .
the hci are captured in flight , confined , cooled by ` resistive cooling ' @xcite and radially compressed by a ` rotating wall ' @xcite technique . after these steps a cold and dense ion cloud is obtained .
the spectroscopy laser enters the trap axially through an open - endcap and will fully irradiate the ion cloud .
the fluorescence from the excited hci is detected perpendicular to the cooled axial motion ( trap axis ) through segmented ring electrodes , which are covered by a highly - transparent copper mesh .
( the ring is segmented for the rotating wall technique . )
the above mentioned transition lifetimes imply that , for a detection efficiency of @xmath65 , acceptable fluorescence rates , up to a few thousand counts per second , from @xmath32 transitions can be expected from a ( @xmath66 mm diameter ) cloud of @xmath64 ions @xcite .
confining the hci in a trap , and cooling and compressing the cloud , will thus enable fluorescence detection and ensure long interrogation times by the laser .
however , due to the high density of hci in the cloud , space charge effects will play a role and will lead to shifts of the motional frequencies of the trapped ions .
we have studied this effect in detail and understand the corresponding frequency shifts well @xcite .
since these shifts are fairly small , the ( frequency dependent ) cooling and compression techniques can still be applied .
the hci also need to be strongly cooled to reduce doppler broadening of the transitions .
this will be achieved by resistive cooling of the ( axial ) ion motion in the trap .
for example , for the @xmath67 transition in @xmath44pb@xmath68 at @xmath69 hz , the doppler broadened linewidth at a temperature of 4 k is @xmath70 hz . the anticipated resolution is therefore of the order of @xmath71 .
this is three orders of magnitude better than any previous measurement , see _
@xcite , and good enough to measure the qed effects within a few percent .
this work is supported by the european commission within the framework of the hitrap project ( hpri - ct-2001 - 50036 ) .
w.n . acknowledges funding by the helmholtz association ( vh - ng-148 ) .
lamb jr . and r.c .
rhetherford , phys . rev . * 72 * , ( 1974 ) 241 .
t. beier , phys . rep . * 339 * , ( 2000 ) 79 .
w. quint j. dilling , s. djekic , h. hffner , n. hermanspahn , h .- j .
kluge , g. marx , r. moore , d. rodriguez , j. schnfelder , g. sikler , t. valenzuela , j. verd , c. weber and g. werth , hyp . int .
* 132 * , ( 2001 ) 457 .
t. beier , l. dahl , h .- j .
kluge , c. kozhuharov , w. quint and the hitrap collaboration , nucl .
instr . meth .
b * 235 * , ( 2005 ) 473 .
kluge , t. beier , k. blaum , m. block , l. dahl , s. eliseev , f. herfurth , s. heinz , o. kester , c. kozhuharov , t. khl , g. maero , w. nrtershuser , t. sthlker , w. quint , g. vorobjev , g. werth , and the hitrap collaboration , _ proceedings of the memorial symposium for gerhard soff _ , topics in heavy ion physics ( eds walter greiner and joachim reinhardt ) , pages 89 - 101 ( 2005 ) , ep systema ( budapest ) .
shabaev , j. phys .
b * 27 * , ( 1994 ) 5825 .
firestone and v.s .
shairley , _ table of isotopes _ ( appendix e ) , wiley ( 1998 ) .
a. bohr and v.f .
weisskopf , phys . rev . *
77 * , 94 ( 1950 ) .
shabaev , m. tomaselli , t. khl , a.n .
artemyev and v.a .
yerokhin , phys .
a * 56 * , ( 1997 ) 252 .
shabaev , a.n .
artemyev , v.a .
yerokhin , o.m .
zherebtsov and g. soff , phys .
lett . * 86 * , ( 2001 ) 3959 .
w. demtrder , _ laser spectroscopy _ , springer , new york ( 1996 ) .
a. roth , ch .
gerz , d. wilsdorf and g. werth , z. phys .
d * 11 * , ( 1989 ) 283 .
bowen , astrophys .
j. * 132 * , ( 1960 ) 1 .
prior , phys .
a * 30 * , ( 1984 ) 3051 .
p. seelig , s. borneis , a. dax , t. engel , s. faber , m. gerlach , c. holbrow , g. huber , t. khl , d. marx , k. meier , p. merz , w. quint , f. schmitt , m. tomaselli , l. vlker , h. winter , m. wrtz , k. beckert , b. franzke , f. nolden , h. reich , m. steck and t. winkler , phys .
* 81 * , ( 1998 ) 4824 .
s. borneis , a. dax , t. engel , c. holbrow , g. huber , t. khl , d. marx , p. merz , w. quint , f. schmitt , p. seelig , m. tomaselli , h. winter , k. beckert , b. franzke , f. nolden , h. reich and m. steck , hyp .
int . * 127 * , ( 2000 ) 305 .
p. beiersdorfer , a.l .
osterheld , j.h .
scofield , j.r .
crespo lpez - urrutia and k. widmann , phys .
* 80 * , ( 1998 ) 3022 .
winters , a.m. abdulla , j.r .
castrejn pita , a. de lange , d.m .
segal and r.c .
thompson , nucl .
instr . meth .
b * 235 * , ( 2005 ) 201 .
m. vogel , d.f.a .
winters , d.m . segal and r.c .
thompson , rev .
instrum . * 76 * , ( 2005 ) 103102 .
g. gabrielse , l. haarsma and s.l .
rolston , int .
j. mass spectr
. ion proc . *
88 * , ( 1989 ) 319 .
wineland and h.g .
dehmelt , j. appl . phys . * 46 * , ( 1975 ) 919 .
itano , j.j .
bollinger , j.n .
tan , b. jelenkovi , x .-
huang and d.j .
wineland , science * 279 * , ( 1998 ) 686 .
winters , m. vogel , d.m .
segal and r.c .
thompson , j. phys .
b : at . mol .
opt . phys .
* 39 * , ( 2006 ) 3131 .
i. klaft , s. borneis , t. engel , b. fricke , r. grieser , g. huber , t. khl , d. marx , r. neumann , s. schrder , p. seelig and l. vlker , phys .
* 73 * , ( 1994 ) 2425 . | an overview is presented of laser spectroscopy experiments with cold , trapped , highly - charged ions , which will be performed at the hitrap facility at gsi in darmstadt ( germany ) .
these high - resolution measurements of ground state hyperfine splittings will be three orders of magnitude more precise than previous measurements .
moreover , from a comparison of measurements of the hyperfine splittings in hydrogen- and lithium - like ions of the same isotope , qed effects at high electromagnetic fields can be determined within a few percent .
several candidate ions suited for these laser spectroscopy studies are presented . ,
qed , highly - charged ions , hyperfine structure , laser spectroscopy , trapping and cooling 12.20.fv , 21.10.ky , 32.10.fn , 32.30.-r |
reaction diffusion equations are used to model transport phenomena in a variety of contexts such as population dynamics , transmission lines , flame propagation among others .
the prototype of such equations , for which a thorough understanding @xcite of its properties has been achieved is the parabolic reaction diffusion equation @xmath4 where subscripts denote derivatives . for positive reaction terms ,
sufficiently localized initial conditions evolve into a monotonic decaying travelling front joining the stable @xmath5 to the unstable @xmath6 equilibrium points . for bistable reaction terms , which satisfy @xmath7 in @xmath8 and @xmath9 in @xmath10 , with @xmath11 ,
it is possible to find initial conditions for which the system will evolve into a monotonic decaying front joining the two stable equilibrium points @xmath5 to @xmath6 . in the first case
there is a continuum of speeds for which there exist monotonic fronts , the system evolves into the front of minimal speed . in the second case
there is a unique speed . the use of eq .
( [ parab ] ) to model a physical process involves assumptions on the stochastic process that describes the motion of the individuals .
more specifically , brownian motion is assumed @xcite .
if instead a more realistic process is considered @xcite , then , in one spatial dimension , the differential equation that describes the motion is a hyperbolic reaction diffusion equation of the form @xmath12 it has been shown that , as in the usual reaction diffusion equation , for positive @xmath1 , the hyperbolic equation admits monotonic decaying fronts of speed @xmath13 and that the speed of the front is determined from a related parabolic equation @xcite .
equations of this form have been studied for particular cases of @xmath1 obtained assuming specific functions for memory effects in the diffusion term @xcite .
recent work has dealt with the stability of the traveling front @xcite . for constant @xmath1 and positive , concave , @xmath14
the traveling fronts are stable to local perturbations .
the asymptotic behavior of the front of minimal speed has also been found @xcite .
numerical solutions for positive but non concave @xmath14 indicate that the transition from linear to nonlinear marginal stability occurs at the same parameter values than for the parabolic equation eq .
( [ parab])@xcite .
detailed analysis for a piecewise linear reaction term has given insight into the nature of possible travelling waves @xcite . in systems with exponential transport memory @xmath1
is of the form @xmath15 for this case a variational principle for the front of minimal speed was constructed for positive reaction terms @xmath2 ; it was proved that for concave reaction terms , as in the parabolic equation , linear marginal stability holds @xcite . in this work
we construct a variational principle for any positive function @xmath1 and general reaction terms for which the parabolic equation admits a front . from this principle upper and lower bounds
are constructed which permit the a priori determination of the transition from linear to nonlinear marginal stability @xcite .
we recover all known results as particular examples , and construct others that show different types of behavior depending on the explicit functions @xmath2 and @xmath1 .
our main result is the following , consider eq .
( 2 ) , with @xmath16 , and @xmath2 a reaction term for which fronts of the parabolic equation exist , as described above .
the minimal ( or unique , for the bistable case ) speed of the fronts joining the stable to the unstable points @xmath17 to @xmath6 is given by @xmath18 where @xmath19 and @xmath20 the supremum is taken over all positive decaying functions @xmath21 for which the integrals exist . from this result
it follows that , for positive reaction terms which satisfy @xmath22 , the following bound holds , @xmath23 which enables one to characterize functions for which marginal stability holds .
the lower bound , the marginal stability value , is that imposed by linear considerations alone . improved lower bounds valid for all reaction terms
are obtained by direct use of the variational principle .
consider the hyperbolic reaction diffusion equation @xmath24 with @xmath25 where @xmath2 and @xmath1 @xmath26 $ ] .
we shall assume that @xmath14 belongs to the class for which monotonic fronts joining the equilibrium points @xmath5 to @xmath6 exist .
the precise conditions have been spelled above in the introduction , and in more precisely elsewhere @xcite .
we wish to find the minimal or unique speed for which there is a monotonic decaying travelling wave solution @xmath27 of eq.([hypereq ] ) .
the speed satisfies @xmath28 , @xmath29 , @xmath30 , @xmath31 , where @xmath32
. it is known that when @xmath33 there is a continuum of fronts for a range of speeds @xmath34 , where @xmath35 is the minimal speed of a related parabolic equation @xcite . there is a unique speed in the bistable case .
we may , as usual , consider the trajectory in phase space by defining @xmath36 .
the monotonic decaying front obeys [ phase ] @xmath37 before going any further we recall the constraints posed by linearization around the fixed points .
linearizing around @xmath6 we find that @xmath38 approaches zero as @xmath39 , where @xmath40 is the positive root of @xmath41 that is , @xmath42 when @xmath22 , since it is known that monotonic fronts exist with @xmath43 , @xmath40 is real if the term in square brackets is positive , that is , @xmath44 where we have called @xmath45 the minimal speed derived from linear theory or , as it is known , linear marginal stability value . when @xmath46 , linear theory imposes no constraint on the speed .
analysis near the equilibrium point @xmath5 imposes no constarint on the speed as we assume that @xmath14 is positive or of the bistable type .
the simplest , but not unique , method to obtain a variational characterization of the speed is to introduce a stretching of coordinates which reduces the equation to the standard parabolic reaction diffusion equation @xcite . since @xmath16
, we may introduce a new independent coordinate defined by the following transformation , @xmath47 with @xmath48 defined in eq.([k ] ) .
the transformation is invertible , and @xmath49 varies between 0 and 1 . in the new coordinates , eq.([phase ] ) reads @xmath50 @xmath51 this equation is the equation for fronts of the parabolic equation , of speed @xmath52 and a reaction term @xmath53 we now check that @xmath54 satisfies all the requirements of the existence of fronts , if @xmath14 does .
effectively , @xmath55 ; for @xmath56 and @xmath57 , sign(f ) = sign(f ) , and @xmath58 so that if @xmath14 is of the bistable type , for which fronts exist , the same holds for @xmath59 if @xmath60 which we know to hold .
moreover @xmath61 we may then apply directly the variational principle for fronts of the parabolic reaction diffusion equation @xcite , that is , @xmath62 where the maximum is taken over all positive decaying functions @xmath63 in ( 0,1 ) for which the integrals exist .
the maximum is attained for @xmath64 provided that @xmath65 , or , if @xmath66 , when @xmath67 .
this is , @xmath68.\ ] ] written in terms of the original quantities , @xmath69 , this condition reads @xmath70 as in the usual parabolic case the maximum is attained except at the linear marginal stability value . here
too one can show that the linear marginal stability value is obtained by taking the supremum instead of the maximum . in terms of the original functions
@xmath14 and @xmath71 , eq.([varhat ] ) reads , @xmath72 notice that the simpler form eq.([simpler ] ) is obtained by changing to the original independent variable @xmath73 .
for the parabolic reaction diffusion equation eq([parab ] ) , for @xmath74 , we know that the minimal speed , @xmath75 of the propagating front is bounded below and above . when these two bounds coincide , then one can determine the speed unambiguosly .
a first estimate comes from the bound @xcite @xmath76 when these two bounds coincide linear marginal stability holds .
it may still hold if these two bounds do not coincide .
this can be decided making use of the integral variational principle which improves the lower bound @xcite , and of a minimax variational principle which improves the upper bound @xcite .
combined use of the two permits the exact determination of the speed . for the bistable case ,
we can obtain the speed making use of the integral variational principle . in this section the analogue of the above results is obtained for the hyperbolic fronts .
we present here a simple derivation of the upper bound , valid both for positive and bistable reaction terms .
the following evident inequality holds , @xmath77 @xmath78 \frac { \int_0 ^ 1 y g(y ) dy}{\int_0 ^ 1 - g^2(y)/g'(y ) dy}.\ ] ] but , as shown in the appendix , @xmath79 from where we obtain @xmath80.\ ] ] going back to the original independent variable , this is @xmath81 .
\label{upbound}\ ] ] observe that this upper bound coincides with the linear value when the supremum occurs at @xmath6 .
the lower bounds have already been obtained , one is the linear bound eq.([linearbound ] ) , and improved lower bounds can be obtained by choosing particular trial functions @xmath63 .
equality , in the expression below , holds for a certain @xmath63 . @xmath82
as a first example consider the case @xmath83 , a constant . as
explained in the introduction the stability of the traveling front for this equation has been proven recently @xcite .
numerical investigations have shown that the transition from linear to nonlinear marginal stability occur at the same parameter values as for the parabolic equation with the same reaction term @xcite .
this last fact follows directly from the bounds above . for a constant @xmath1
, we obtain @xmath84 and @xmath85 .
then , the bounds read @xmath86 the upper and lower bounds are identical to those for the speed of fronts of the parabolic equation parabolic equation ; the speed is given by @xmath87 the speed of the front is determined in terms of the speed of the parabolic front with the same reaction term , therefore the transition from linear to nonlinear marginal stability occurs at the same parameter values as for the parabolic case . the speed itself , however , is lower than the speed of the front of the parabolic equation , @xmath88 as a second example , which shows that substantially different behavior may occur , take the reaction function @xmath89 .
the front of minimal speed for the parabolic equation is given by @xmath90 if we now consider the hyperbolic equation with @xmath91 , the upper and lower bounds coincide and the speed @xmath92 effectively we obtain @xmath93 the minimal speed is the linear marginal stability value , no transition from linear to nonlinear marginal stability occurs .
a significant slowdown of the speed takes place . in systems with time delayed diffusion
the function @xmath1 adopts the form @xmath94 form which arises when considering diffusion with exponential time delay . here
@xmath95 is a parameter . for this function @xmath96 and @xmath97 .
the speed of the front was studied for positive reaction terms , and it was shown that for concave reaction terms marginal stability holds @xcite .
this can be proved easily from the present approach . as eq.([upbound ] ) shows , linear marginal stability holds for all positive reaction terms with @xmath22 provided that the supremum of @xmath98 occurs on @xmath6 , since @xmath99 , is equal to the lower bound . to prove
then that linear marginal stability holds it suffices to show that @xmath100 for @xmath101 $ ] . taking the derivative we obtain , after some algebra , @xmath102\ ] ] where @xmath103 .
it follows immediately that for concave reaction terms @xmath100 . for concave terms @xmath104
, so the first term in the square brackets is negative .
and @xmath105 is negative for concave functions as well , @xmath106 , and @xmath107 .
so we have recovered the result that for concave functions with @xmath22 , linear marginal stability holds . as it occurs for the parabolic case .
it may hold for noncave functions as well , fact which can be determined by use of the variational principle .
consider again the reation term @xmath108 as in subsection * b*. in the parabolic equation transition from linear to nonlinear marginal stability occurs at @xmath109 .
choose here @xmath110 which guarantees @xmath111 for @xmath112 .
we show in fig .
1 that the transition from linear to nonlinear marginal stability occurs at a lower value of @xmath113 than that for the parabolic case .
we chose as a trial function @xmath114 .
the dot - dash line shows the lower bound imposed by the linear marginal stability value , the solid line the lower bound from the variational principle .
the transition occurs at at @xmath113 as low as 1.6 at least .
the precise transition value can be found by using a better trial function .
the dashed line shows the upper bound , which guarantees that linear marginal stability holds for @xmath115 , values for which @xmath14 is concave . the variational principle and the bounds obtained above hold for positive and bistable reaction systems , no assumption on the sign of @xmath14 is made in the derivations .
we may therefore apply them to a bistable reaction term .
we take as an example a case for the speed of the parabolic front can be determined exactly , @xmath116 the unique value of the speed for which a front exists is @xmath117 if @xmath83 a constant , the speed of the front is given by eq([constfi2 ] ) , and the exact speed is known with @xmath75 given above . for this bistable reaction term , with @xmath118 , which is positive for @xmath119 , with a simple trial function @xmath120 , all integrals are elementary , we obtain , using eq.([var ] ) , @xmath121 numerical determination of the speed can be made with any desired accuracy by use of improved trial functions .
we wish to point that , in recent work , a variational principle valid only for positive reaction terms has been applied to a bistable example @xcite .
bistable reaction terms , by definition , are not positive throughout the whole interval , so the hypothesis of their derivation is violated .
we attribute the nearly perfect agreement between the speed of the front obtained from their ( nonapplicable ) variational principle to that obtained from direct numerical integration to a coincidence for the specific trial function chosen .
it is not difficult to find an acceptable trial function , sharply peaked at the origin , for which the integrand of their variational expression , and so the speed , becomes complex .
we have studied the speed of travelling fronts of hyperbolic reaction diffusion equations ( [ hypereq ] ) for reaction terms for which the parabolic reaction diffusion equation ( [ parab ] ) admits fronts . a variational principle for
the speed of the fronts has been obtained for a wide class of systems .
explicit upper and lower bounds , obtained from the variational expression allow one to characterize sytems for which marginal stability holds .
we have seen that depending on the nonlinearities of the equation the range of validity of linear marginal stability , can be increased or decreased , compared to the comparable range for a parabolic equation with the same reaction term .
the stability of the fronts has been proved for a limited class of such equations @xcite , numerical solutions have been explored for systems with memory effect .
these results will be more relevant if the stability of fronts of more general systems is proved . at present numerical results
indicate that the front of minimal or unique speed can be selected .
the general initial value problem for hyperbolic reaction diffusion equations is an open problem .
let @xmath122 be a positive decaying function , which satisfies @xmath123 , and call @xmath124 , so that @xmath125 .
then @xmath126 ^ 2 & = & \left [ \int _ 0 ^ 1 \left(\frac{g}{\sqrt{h}}\right ) ( \sqrt{h } y ) du \right]^2 \\
& \le & \int_0 ^ 1 \frac{g^2(y)}{h(y ) } d y \int_0 ^ 1 y^2 h(y ) d y,\end{aligned}\ ] ] where we used schwarz s inequality .
the second integral on the right side is , integrating by parts , @xmath127 where we used the fact that @xmath63 diverges at @xmath128 slower than @xmath129 ( this is equivalent to @xmath130 ) and that @xmath123 which can be seen form eq.([ghat ] ) .
replacing this in the inequality above we have the desired result , @xmath131 | we study travelling fronts of equations of the form @xmath0 . a criterion for the transition from linear to nonlinear marginal stability
is established for positive functions @xmath1 and for any reaction term @xmath2 for which the usual parabolic reaction diffusion equation @xmath3 admits a front . as an application , we treat reaction diffusion systems with transport memory . |
gravitational lensing provided the first experimental verification of general relativity ( gr ) through observations of starlight bending around the sun during an eclipse in 1919 @xcite and continues to be a major source of insight into gravitation @xcite .
excitingly , increasingly precise observations of the compact radio source sgr a * at the galactic center and its surrounding stars have given us very high confidence that only a very gross deviation from gr could allow for the absence of a black hole there @xcite . according to @xcite , the black hole is estimated to have a mass of about @xmath6 and a distance of about 8.33 kpc from earth .
black holes are unique laboratories for gravitational lensing because their compactness allows light to closely approach the photon sphere and its path will bend significantly there due to gravity .
most studies of gravitational lensing @xcite are in the weak deflection limit , when the point of closest approach of any lensed photons is far from the lensing mass .
this allows for the simple expression of the bending angle as @xmath7 , where @xmath8 is the point of closest approach of the null geodesic ( this paper uses geometric units @xmath9 ) .
this expression is valid in the limit @xmath10 .
when a photon closely approaches a black hole s photon sphere , the weak deflection limit does not hold and using this approximation for the bending angle leads to inaccurate results . for a spherically symmetric ,
static metric with line element @xmath11 the bending angle is an elliptic integral based on the functions of the metric @xcite and is @xmath12^{-1/2 } \\ & \times & \frac{dr}{r}- \pi \label{bending}\end{aligned}\ ] ] if the point of closest approach is very close to the photon sphere , the strong deflection limit approximation of this integral @xcite can be used . in all cases ,
a full numerical treatment of the bending angle can be used @xcite , and it should be used for studies , such as this one , where neither the strong nor weak deflection limits are satisfied .
this is further explained in sec .
[ sec : lensing ] .
large bending angles yield interesting results : when photons approach close enough to the photon sphere , they can loop around the lens before reaching the observer and produce an infinite sequence of images on both sides of the optic axis @xcite .
the properties of these images are sensitive to the form of the black hole metric @xcite with @xcite studying the effect of the rn metric
. there has been further study of an astrophysical scenario which utilizes lensing with a large bending angle .
s2 was the first star that was examined for a retrolensing " @xcite effect ( lensing with a bending angle of @xmath13 ) . by using orbital parameters of the s stars in the galactic center provided by @xcite ,
@xcite study the properties of secondary images of those s stars .
this study showed 9 stars with secondary images with a maximum brightness in the k - band brighter than @xmath14 , a cutoff that maximizes observational prospects . in this paper , we use a black hole metric with an additional term that is @xmath0 .
the metric is @xmath15 with @xmath16 a free parameters often expressed by @xmath17 .
when @xmath18 is positive , this represents the reissner - nordstrom solution for a charged black hole .
static black holes with a large amount of electric charge are not expected to exist in nature , and the existence of rotating , charged black holes is controversial @xcite .
in addition , the amount of charge is limited to @xmath19 or @xmath20 because the saturation of this bound would lead to a naked singularity and the violation of cosmic censorship @xcite .
however , since the calculation of the properties of secondary images of s stars has only been done with a schwarzschild metric , it is useful to examine such a fundamental case .
we have also found it useful to explore negative values of @xmath18 .
there are only weak constraints on a lower bound for @xmath18 that come from studies of neutron star binary systems @xcite .
using a @xmath1 term is interesting because it is motivated by alternative gravity frameworks , particularly braneworld theories that construct gravity as a higher - dimensional theory .
as these theories usually predict a correction that strengthens gravity , this would correspond to a negative value for @xmath18 .
for any non - trivial value of @xmath18 , observational constrains from solar system observations @xcite disallow the @xmath1 term from applying anywhere but near the black hole . strengthening gravity using a negative value for @xmath18 yields a brighter secondary image ; therefore , observations of these secondary images can place constraints on the size of the @xmath1 term near the black hole .
an extra @xmath1 term in the metric comes directly from a proposed black hole metric in the randall - sundrum ii theory .
the randall - sundrum i model @xcite is a braneworld scenario inspired by heterotic m - theory @xcite , a 5 dimensional effective framework that arises from the dimensional reduction of 11-dimensional harova - witten theory on a calabi - yau manifold with the imposition of @xmath21 symmetry @xcite .
six dimensions are compactified , making gravity effectively 5 dimensional .
the randall - sundrum ii scenario is this model with the second brane taken to infinity @xcite .
it is not clear whether a static black hole solution exists for a rs ii braneworld .
although @xcite has shown the existence of a static black hole in a 2 + 1 brane setup , it is unclear what the solution is and whether there even is a static black hole solution in the 3 + 1 brane scenario @xcite .
attempts to study the problem have yielded several possible black hole metrics @xcite .
the effect of these metrics on lensing has been examined in several studies @xcite . for a supermassive black hole such as sgr a * , the garriga - tanaka @xcite and black string " @xcite metrics will not show any results @xcite and this paper aims to show that a modification of the metric near the horizon can affect lensing observables .
one particular metric that has been studied in connection with lensing by the black hole at sgr a * is the tidal " reissner - nordstrom metric @xcite which is of the form of eq .
( [ eq : trnmetric ] ) .
we therefore use it as an example of a metric with a @xmath1 potential term .
there are several possible objections to using the trn metric for a supermassive black hole : studies suggest that supermassive black holes in the braneworld should have induced metrics that are no more than negligibly different than the schwarzschild metric @xcite .
in addition , the trn metric has no known completion in the bulk @xcite and there are likely naked singularities in the bulk for the trn metric @xcite .
however , we will study lensing properties of the trn metric to gain understanding of the effects of adding a @xmath0 term in the potential , whether it comes from the braneworld scenario or any other gravitational framework . in sec .
ii , we discuss s stars near the galactic center . now that their orbital parameters are well known from decades of observation @xcite , we show how to reconstruct the orbits of these stars and characterize the variables in the ohanian lens equation @xcite in terms of the star s orbital parameters .
although current uncertainties in lensing parameters can interfere with the test we are proposing , future instruments and observations will further constrain these orbital parameters . in sec .
iii , we use the orbital parameters to construct a light curve for the secondary images of s2 , s6 , and s14 . for these stars ,
we show the light curve of the secondary image when assuming a schwarzschild spacetime , a trn spacetime with @xmath22 , and , for the case of s2 , an extremal rn spacetime with @xmath23 .
we show that for high enough values of the tidal charge parameter @xmath18 , an appreciable difference appears for these light curves in the trn spacetime .
we also show the relevant properties , such as image magnitude and image position at peak brightness for several values of @xmath18 . in sec .
iv , we discuss the observational prospects for these images as well as the observing the difference in brightness for these images . we conclude that observation of these images is possible , as is observing the difference in image properties due to a modified gravity theory .
there is a large stellar population in the central parsec of the milky way .
most stars are old red main sequence stars , but there are many young wolf - rayet stars which present challenges to theories of star formation @xcite .
some of these young stars orbit close to the black hole at sgr a * , labelled s stars , have been observed carefully .
their orbits can be reconstructed from orbital parameters published in @xcite .
for the purposes of this paper , at each point in time , we treat the star as a source being lensed by the black hole . to study the affect of the metric near the black hole , we are interested in the light that passes across the optic axis ( line connecting the lens and the observer ) , forming a secondary image @xcite . using orbital parameters and the intrinsic brightness of the stars in the k - band , @xcite calculated the position and magnitude of the secondary images , assuming a schwarzschild metric .
observationally , the most promising images come from the stars s6 and s27 , which have the brightest predicted secondary images with @xmath24 and @xmath25 respectively .
the peak brightness occurs when each star is at periapse . although it will not get as bright
, the periapse of s2 will occur relatively soon in 2018 and its secondary image will be at peak brightness then , with predicted @xmath26 .
although s27 is brighter , we will instead study s14 because its peak brightness comes at an earlier date than the peak of s27 .
in addition , the secondary image of s14 is a better candidate for differentiating between a schwarzschild spacetime and an alternative one , as will be explained in sec .
[ sec : results ] . in calculating the positions of the secondary image , we use the improved ohanian lens equation @xcite , apply the small angle approximation , and throw out negligible terms , leaving @xmath27 where @xmath28 is the angle between the optic axis and the line connecting the source to the lens , @xmath29 is the image position ( to the observer ) , @xmath30 is the constant distance from the observer to the lens ( in this case , the distance between us and sgr a * ) and @xmath31 is the distance , which varies over time , between the lens and the source star .
the anomaly angle from the periapse ( @xmath32 ) is determined by the initial conditions and differential equation : @xmath33 where @xmath34 is the mass enclosed within the orbit and @xmath35 is the eccentricity of the orbit .
@xmath34 can be calculated in relation to the orbital period @xmath36 and semi - major axis @xmath37 using kepler s third law @xmath38 the relationship between @xmath28 and @xmath31 and @xmath32 is given by : @xmath39\end{aligned}\ ] ] in eq .
( [ eq : ohanian ] ) , @xmath40 is the bending angle as a function of @xmath29 .
as noted in @xcite , the ohanian lens equation is accurate to better than @xmath41 . applying the small - angle approximations and throwing out the @xmath29 term to obtain eq .
( [ eq : ohanian ] ) affects outcomes by less than that . for a given source position , we numerically solve eq .
( [ eq : ohanian ] ) which yields the position of the secondary image .
the image magnification is @xcite : @xmath42 analytical formulas for image position and magnification are given by @xcite .
however , the analytical formulas are not accurate because the bending angle in this astrophysical scenario often lies neither in the strong or weak deflection limit .
therefore , numerical evaluation of the image positions and magnifications are necessary . to calculate this
, we must numerically evaluate eq .
( [ bending ] ) . writing down the equations of motion for photons in this spacetime
@xmath43 the coordinates @xmath32 and @xmath44 are cyclic , leading to the conserved quantities @xmath45 which characterize the motion of null geodesics throughout this spacetime @xcite .
importantly , we can create a relationship between the coordinate of closest approach ( @xmath8 ) and the angular position of images @xcite : @xmath46 we can then take the derivative of @xmath47 with respect to @xmath29 @xmath48 which is necessary to evaluate eq .
( [ eq : mag ] ) .
this method is described in greater detail in @xcite . [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] the effect of black hole spin on the secondary images was studied by @xcite and commented on by @xcite . in this paper , we have not studied the effects of the kerr metric for several reasons .
firstly , evaluation of lensing in a kerr spacetime is far more challenging than the study of a trn spacetime , so we have started with a less ambitious project . also , while for higher - order images , spin can greatly change the magnification , this is not true for secondary images @xcite , and the effect of the black hole s spin is less significant .
still , the magnification of secondary images depends on the magnitude and direction of the black hole s spin and secondary images in the kerr spacetime are worthy of further study .
some work on rotating black holes and their gravitational lensing effects has already been done @xcite , including some work on rotating braneworld black holes @xcite .
of the three stars studied in this paper , s14 and s6 have brighter secondary images at periapse compared to s2 s image because of the more edge - on nature of their orbits relative to our line of sight with sgr a*. while the orbit of s2 is not aligned close to edge - on , its periapse is the closest amongst known stars , causing its secondary image to be very bright .
in addition , s2 will next be at periapse in 2018 , allowing for a more immediate study of its image s properties .
as mentioned above , the brightness of secondary images is dependent on @xmath18 because the images are either pushed closer or further from the optic axis when there is a @xmath1 term in the metric .
this is directly related to the difference in the size of the event horizon and photon sphere due to the value of @xmath18 .
another , and perhaps easier , way of determining the metric around the black hole would be to measure the shape and size of the black hole s horizon or photon sphere .
a preliminary attempt at this has been made @xcite , but they were unable to identify the observed structure with the black hole itself . at present , there are no constraints which would prevent the size of the event horizon from being significantly larger or smaller than predicted in a schwarzschild spacetime ( but not by an order of magnitude @xcite ) . in addition
, since the photon sphere is so small , even a relatively large percentage change in its size corresponds to only a few @xmath49as .
resolving the difference between two proposed photon sphere sizes may be beyond the capabilities of projected future instruments . in this case , observing secondary images may provide more information . there has been some discussion about observing these secondary images with the upcoming generation of telescopes @xcite . a very promising project is the mcao imaging camera for deep observations ( micado ) telescope @xcite at e - elt .
it is expected to be able resolve images as faint as @xmath50 and will have a resolution of up to 6 mas in imaging mode ( 10 mas in the k - band ) and an astrometric accuracy of up 10 @xmath49as .
it also will have a photometric accuracy of 0.03 magnitudes . from the data shown above , trying to resolve the difference between an image s position in the schwarzschild metric and an image s position in the trn
metric is next to impossible given the small separation between these positions .
therefore , it is important to explore not only image ( and photon sphere ) positions , but to use image magnfications as a complementary avenue for exploring the metric . if the secondary images in this paper were isolated point sources , micado would not have any problems detecting them and even differentiating between two predicted values for image brightness ( assuming the difference is larger than the photometric accuracy ) .
however , these images will be very close to and essentially unresolvable from sgr a * and its crowded environment .
this is not a fatal flaw , because sgr a * is very faint in the near infared k - band ( @xmath51 m ) , and it may be feasible to subtract out the quiescient state of sgr a * in the k - band .
some studies @xcite claim that sgr a * has a highly variable quiescient " state of @xmath52 .
in addition , there are occasional flares that can be brighter than @xmath53 and last on a scale of hours .
while the flares are thought to originate very close to the black hole , it is not clear whether the quiescient " radiation in the k - band comes from sgr a * itself or whether it comes from the fact that the lower resolution ( 65 mas ) in the survey includes one or more sources in the area near sgr a*. the currently known star with the closest approach is s2 , which appears to be 11 mas from sgr a * at the point of closest approach- at all points , micado should be able to resolve s2 and many other closely moving stars from sgr a*. there may be other , unidentified stars that are currently conflated with radiation from sgr a * but will be separately with micado s increased resolution .
better resolution , combined with further understanding of the k luminosity function near the galactic center means that observations of secondary images is a real possibility .
micado s photometric accuracy should be able to distinguish many of the image brightness differences discussed in this paper .
this assumes that when viewed with a fine enough resolution , sgr a * does not emit too brightly in the k - band and source crowding is not insurmountable .
even if flares are observed , they persist for a time scale much shorter than secondary images ( which remain bright for months ) and should be easily distinguished from the quiescient state .
additionally , it is expected that an additional @xmath54-@xmath55 s stars will be found in the central milli - parsec of the galaxy @xcite , yielding additional and perhaps even better candidates to observe secondary images with the right properties to give us insight into the metric near the black hole .
it may not be possible to accurately treat stars that close to sgr a * with the thin - lens approximation , but in that case , a more exact numerical treatment can be utilized @xcite .
observing secondary images of s stars will be challenging , but we may very well find that it is possible to observe faint secondary images and use their properties to give us insight into the true nature of gravity .
this exciting prospect should be an additional motivation for the next generation of observational instruments aimed at the galactic center .
i would like to thank r.k .
sheth and j. khoury for their unwavering support of my research , r. davies for very helpful discussions related to the observational aspects of this paper , and g. bernstein , s. doelman , t. pumard , a. lidz , b. jain , a. ghez , b. greenbaum , and f. melia for helpful conversations . | we show that a possible astrophysical experiment , detection of lensed images of stars orbiting close to sgr a * , can provide insight into the form of the metric around a black hole .
we model sgr a * as a black hole and add in a @xmath0 term to the schwarzschild metric near the black hole .
we then attempt to determine the effect of this extra term on the properties of the secondary images of the s stars in the galactic center .
when the @xmath1 term is positive , this represents a reissner - nordstrom ( rn ) metric , and we show that the there is little observational difference between a schwarzschild black hole and a rn black hole , leading to the conclusion that secondary images may not be a useful probe of electrical charge in black holes . a negative value for the @xmath1 term can enter through modified gravity scenarios .
although physically unlikely to apply in the case of a large black hole , the randall - sundrum ii braneworld scenario admits a metric of this form , known as tidal reissner - nordstrom ( trn ) metric .
we use values of tidal charge ( @xmath2 in @xmath3 ) ranging from @xmath4 to @xmath5 .
a negative value of @xmath2 enhances the brightness of images at all times and creates an increase in brightness of up to 0.4 magnitudes for the secondary image of the star s2 at periapse .
we show that for other stars with brighter secondary images and positions more aligned with the optic axis , using the tidal reissner - nordstrom metric with negative @xmath2 enhances the images as well , but the effect is less pronounced .
this effect is related to the increase in the size of the photon sphere , and therefore , should be noticeable in other metrics with a similar effect on the photon sphere . with the next generation of instruments and
increased knowledge of radiation from sgr a * , using properties of secondary images to place constraints on the size of the @xmath1 term .
this knowledge will be useful in constraining any modified gravity theory that adds a similar term into the strong field near a black hole . |
for a permutation @xmath0 , let @xmath1 , where the sum is over all permutations @xmath2 below @xmath3 in the strong bruhat order . geometrically , the polynomial @xmath4 is the poincar polynomial of the schubert variety @xmath5 in the flag manifold @xmath6 .
define the _ inversion hyperplane arrangement _
@xmath7 as the collection of the hyperplanes @xmath8 in @xmath9 , for all inversions @xmath10 , @xmath11 .
let @xmath12 be the generating function that counts regions @xmath13 of the arrangement @xmath7 according to the distance @xmath14 from the fixed initial region @xmath15 .
the main result of the paper is the claim that @xmath16 if and only if the schubert variety @xmath17 is smooth . according to well - known lakshmibai - sandhya s criterion @xcite ,
the schubert variety @xmath17 is smooth if and only if the permutation @xmath3 avoids two patterns @xmath18 and @xmath19 .
( let us say that the permutation @xmath3 is smooth in this case . )
also carrell - peterson @xcite proved that @xmath17 is smooth if and only if the poincar polynomial @xmath4 is palindromic , that is @xmath20 .
if @xmath3 is not smooth then the polynomial @xmath4 is not palindromic , but the polynomial @xmath21 is always palindromic .
so @xmath22 in this case .
on the other hand , we show that , for smooth @xmath3 , the polynomials @xmath21 and @xmath4 satisfy the same recurrence relation . for the poincar polynomials
@xmath4 , this recurrence relation was given by gasharov @xcite .
this implies that @xmath16 in this case . for smooth @xmath3 , we present an explicit factorization of the polynomials @xmath16 as a product of @xmath23-numbers @xmath24_q\cdots [ e_n+1]_q$ ] , where @xmath25 can be computed using the left - to - right maxima ( aka records ) of the permutation @xmath3 . in this case , the _ inversion graph _ @xmath26
, whose edges correspond to inversions in @xmath3 , is a chordal graph .
the numbers @xmath25 are the roots of the chromatic polynomial @xmath27 of the inversion graph .
the polynomial @xmath27 is also the characteristic polynomial of the inversion hyperplane arrangement @xmath7 .
we call the numbers @xmath25 the _ exponents .
_ we thank vic reiner and jonas sjstrand for helpful conversations .
the ( strong ) _ bruhat order
_ `` @xmath28 '' on the _ symmetric group _
@xmath29 is the partial order generated by the relations @xmath30 if @xmath31 . here
@xmath32 is the transposition of @xmath33 and @xmath34 ; and @xmath35 denotes the _ length _ of a permutation @xmath0 , i.e. , the number of inversions in @xmath3 .
intervals in the bruhat order play a role in schubert calculus and in kazhdan - lusztig theory . in this paper we concentrate on bruhat intervals of the form @xmath36:=\{u\in s_n \mid u\leq w\}$ ] ( where @xmath37 is the identity permutation ) , that is , on lower order ideals of the bruhat order .
they are related to _
schubert varieties _ @xmath5 in the flag manifold @xmath6 . here
@xmath38 denotes the borel subgroup of @xmath39 .
the _ poincar polynomial _ of the schubert variety @xmath17 is the rank generating function for the interval @xmath36 $ ] , e.g. , see @xcite : @xmath40 the well - known smoothness criterion for schubert varieties , due to lakshmibai and sandhya , is based on pattern avoidance .
a permutation @xmath0 contains a _ pattern _
@xmath41 if there is a subword with @xmath42 letters in @xmath3 with the same relative order of the letters as in the permutation @xmath43 .
a permutation @xmath3 _ avoids the pattern @xmath43 _ if @xmath3 does not contain this pattern .
( lakshmibai - sandhya @xcite ) for a permutation @xmath44 , the schubert variety @xmath17 is smooth if and only if @xmath3 avoids the two patterns @xmath18 and @xmath45 .
we will say that @xmath0 is a _
smooth permutation _ if it avoids these two patterns @xmath18 and @xmath45 .
another smoothness criterion , due to carrell and peterson , is given in terms of the poincar polynomial @xmath4 .
let us say that a polynomial @xmath46 is _ palindromic _ if @xmath47 , i.e. , @xmath48 for @xmath49 .
[ th : peterson_criterion ] ( carrell - peterson @xcite , see also ( * ? ? ?
* sect . 6.2 ) ) for a permutation @xmath44 , the schubert variety @xmath17 is smooth if and only if the poincar polynomial @xmath4 is palindromic .
for a graph @xmath50 on the vertex set @xmath51 , the _ graphical arrangement _
@xmath52 is the hyperplane arrangement in @xmath9 with hyperplanes @xmath53 for all edges @xmath54 in @xmath50 .
characteristic polynomial _
@xmath55 of the graphical arrangement @xmath52 is also the _ chromatic polynomial _ of the graph @xmath50 .
the value of @xmath55 at a positive integer @xmath56 equals the number of ways to color the vertices of the graph @xmath50 in @xmath56 colors so that all neighboring pairs of vertices have different colors .
the value @xmath57 is the number of regions of @xmath52 .
the regions of @xmath52 are in bijection with acyclic orientations of the graph @xmath50 .
recall that an _ acyclic orientation _ is a way to direct edges of @xmath50 so that no directed cycles are formed .
the region of @xmath52 associated with an acyclic orientation @xmath58 is described by the inequalities @xmath59 for all directed edges @xmath60 in @xmath58 .
we will study a special class of graphical arrangements . for a permutation @xmath0 , the _ inversion arrangement _
@xmath7 is the arrangement with hyperplanes @xmath53 for each inversion @xmath61 , @xmath11 .
define the _ inversion graph _
@xmath26 as the graph on the vertex set @xmath51 with the set of edges @xmath62 .
the arrangement @xmath7 is the graphical arrangement @xmath52 for the _ inversion graph _ @xmath63 .
let @xmath64 be the number of regions in the inversion arrangement @xmath7 .
let @xmath65=p_w(1)$ ] be the number of elements in the bruhat interval @xmath66 $ ] .
interestingly , the numbers @xmath64 and @xmath67 are related to each other .
[ conj : r = b ] ( hultman - linusson - shareshian - sjstrand @xcite ) \(1 ) for any permutation @xmath0 , we have @xmath68 .
\(2 ) the equality @xmath69 holds if and only if @xmath3 avoids the following four patterns @xmath45 , @xmath70 , @xmath71 , @xmath72 .
this result was conjectured in @xcite and verified on a computer for all permutations of sizes @xmath73 .
this conjecture was announced as an open problem in a workshop in oberwolfach in january 2007 .
a. hultman , s. linusson , j. shareshian , and j. sjstrand reported that they proved the conjecture .
their proof will appear on arxiv .
it was proved in @xcite that @xmath74 for all grassmannian permutations @xmath3 , which agrees with the above result . in this case
, @xmath67 counts the number of totally nonnegative cells in the corresponding schubert variety in the grassmannian , see @xcite .
the four patterns from theorem [ conj : r = b ] came up earlier in the literature in at least two places .
firstly , gasharov and reiner @xcite showed that the schubert variety @xmath17 can be described by simple inclusion conditions exactly when @xmath3 avoids these four patterns .
secondly , sjstrand @xcite showed that the bruhat interval @xmath66 $ ] can be described as the set of permutations associated with rook placements that fit inside a skew ferrers board if and only if @xmath3 avoids the same four patterns .
note that each of the four patterns from theorem [ conj : r = b ] contains one of the two patterns from lakshmibai - sandhya s smoothness criterion .
thus the theorem implies the equality @xmath74 for all smooth permutations @xmath3 .
let us define the @xmath23-analog of the number of regions of the graphical arrangement @xmath52 , where @xmath50 is a graph on the vertex set @xmath51 . for two regions @xmath13 and
@xmath75 of the arrangement @xmath52 , let @xmath76 be the number of hyperplanes in @xmath52 that separate @xmath13 and @xmath75 . in other words , @xmath76 is the minimal number of hyperplanes we need to cross to go from @xmath13 to @xmath75 .
let @xmath15 be the region of @xmath52 that contains the point @xmath77 .
define @xmath78 where the sum is over all regions @xmath13 of the arrangement @xmath52 .
equivalently , the polynomial @xmath79 can be described in terms of acyclic orientations of the graph @xmath50 . for an acyclic orientation @xmath58 ,
let @xmath80 be the number of edges of @xmath50 oriented as @xmath60 in @xmath58 where @xmath81 ( descent edges ) .
then @xmath82 where the sum is over all acyclic orientations @xmath58 of @xmath50 .
indeed , for the acyclic orientation @xmath58 associated with a region @xmath13 we have @xmath83 . for @xmath0 ,
let @xmath84 be the polynomial that counts the regions of the inversion arrangement @xmath85 .
we are now ready to formulate the first main result of this paper .
recall that @xmath1 is the poincar polynomial of the schubert variety .
[ th : p = r ] for a permutation @xmath0 , we have @xmath86 if and only if @xmath3 is a smooth permutation , i.e. , if and only if @xmath3 avoids the patterns @xmath18 and @xmath45 .
this result was initially conjectured during a conversation of the second author ( a.p . ) with vic reiner . the `` only if '' part of theorem [ th : p = r ] is straightforward . indeed , if @xmath3 is not smooth then by carrell - peterson s smoothness criterion ( theorem [ th : peterson_criterion ] ) the poincar polynomial @xmath4 is not palindromic . on the other hand
, the polynomial @xmath21 is always palindromic , which follows from the involution on the regions induced by the map @xmath87 .
thus @xmath22 in this case
. we will prove the `` if '' part of theorem [ th : p = r ] in section [ sec : recurrence ] .
our second result is an explicit non - recursive formula for the polynomials @xmath86 , when @xmath3 is smooth .
let us say that an index @xmath88 is a _ record position _ of a permutation @xmath0 if @xmath89 .
the values @xmath90 are called the _ records _ or _ left - to - right maxima _ of @xmath3 . for @xmath91 ,
let @xmath13 and @xmath75 be the record positions of @xmath3 such that @xmath92 and there are no other record positions between @xmath13 and @xmath75 .
( set @xmath93 if there are no record positions greater than @xmath33 . )
let @xmath94 [ th : formula ] let @xmath3 be a smooth permutation in @xmath29 , and let @xmath25 be the numbers constructed from @xmath3 as above .
then @xmath95_q \ , [ e_2 + 1]_q \cdots [ e_n+1]_q \,.\ ] ] here @xmath96_q : = ( 1-q^a)/(1-q ) = 1 + q+ q^2 + \cdots + q^{a-1}$ ]
. we will prove theorem [ th : formula ] in section [ sec : speo ] .
[ examp : exps ] let @xmath97 .
the record positions of @xmath3 are @xmath98 .
we have @xmath99 theorem [ th : formula ] says that @xmath100_q\ , [ 2]_q\,[3]_q\,[4]_q\,[1]_q\,[2]_q\,[3]_q$ ] .
it was known before that the poincar polynomial @xmath4 for smooth @xmath3 factors as a product of @xmath23-numbers @xmath96_q$ ] .
gasharov @xcite ( see proposition [ prop : gasharov_factorization ] below ) gave a recursive construction for such factorization .
on the other hand , carrell gave a closed non - recursive expression for @xmath4 as a ratio of two polynomials , see @xcite and ( * ? ? ?
however , it is not immediately clear from that expression that its denominator divides the numerator .
one benefit of the formula in theorem [ th : formula ] is that it is non - recursive and it involves no division .
another combinatorial formula for @xmath4 that has these features was given by billey , see @xcite and ( * ? ? ?
a graph is called _ chordal _ if each of its cycles with four or more vertices has a _ chord , _ which is an edge joining two vertices that are not adjacent in the cycle .
a _ perfect elimination ordering _ in a graph @xmath50 is an ordering of the vertices of @xmath50 such that , for each vertex @xmath101 of @xmath50 , all the neighbors of @xmath101 that precede @xmath101 in the ordering form a clique ( i.e. , a complete subgraph ) .
( fulkerson - gross @xcite ) a graph is chordal if and only if it has a perfect elimination ordering .
it is easy to calculate the chromatic polynomial @xmath55 of a chordal graph @xmath50 .
let us pick a perfect elimination ordering @xmath102 of the vertices of @xmath50 .
for @xmath91 , let @xmath103 be the number of the neighbors of the vertex @xmath104 among the preceding vertices @xmath105 .
the numbers @xmath25 are called the _ exponents _ of @xmath50 .
the following formula is well - known .
[ prop : chromatic ] the chromatic polynomial of the chordal graph @xmath50 equals @xmath106 .
thus the graphical arrangement @xmath52 has @xmath107 regions . for completeness sake
, we include the proof , which also well - known .
it is enough to prove the formula for a positive integer @xmath56 .
let us count the number of coloring of vertices of @xmath50 in @xmath56 colors .
the vertex @xmath108 can be colored in @xmath109 colors .
then the vertex @xmath110 can be colored in @xmath111 colors , and so on .
the vertex @xmath104 can be colored in @xmath112 colors , because the @xmath113 preceding neighbors of @xmath104 already used @xmath113 _ different _ colors .
[ rem : does_not_depend_on_peo ] a chordal graph can have many different perfect elimination orderings that lead to different sequences of exponents .
however , the multiset ( unordered sequence ) @xmath114 of the exponents does not depend on a choice of a perfect elimination order .
indeed , by proposition [ prop : chromatic ] , the exponents @xmath103 are the roots of the chromatic polynomial @xmath55 .
[ lem : r = mr ] ( cf .
bjrner - edelman - ziegler @xcite ) suppose that a graph @xmath50 on the vertex set @xmath51 has a vertex @xmath101 adjacent to @xmath115 vertices that satisfy the two conditions : 1 .
the set of all neighbors of @xmath101 is a clique in @xmath50 .
2 . 1 .
all neighbors of @xmath101 are less than @xmath101 , or 2 .
all neighbors of @xmath101 are greater than @xmath101
. then @xmath116_q\ , r_{g\setminus v}(q ) , $ ] where @xmath117 is the graph @xmath50 with the vertex @xmath101 removed .
this claim follows from general results of @xcite on supersolvable hyperplanes arrangements . for completeness
, we give a simple proof .
the polynomials @xmath79 and @xmath118 are @xmath119-generating functions for acyclic orientations of the graphs @xmath50 and @xmath117 .
let us fix an acyclic orientation @xmath58 of the graph @xmath117 , and count all ways to extend @xmath58 to an acyclic orientation of @xmath50 .
the vertex @xmath101 is connected to a subset @xmath120 of @xmath115 vertices of the graph @xmath117 , which forms the clique @xmath121 .
clearly , there are @xmath122 ways to extend an acyclic orientation of the complete graph @xmath123 to an acyclic orientation of @xmath124 .
moreover , for each @xmath125 , there is a unique extension of @xmath58 to an acyclic orientation @xmath126 of @xmath50 such that there are exactly @xmath34 edges oriented towards the vertex @xmath101 in @xmath126 ( and @xmath127 edges oriented away from @xmath101 ) .
all vertices in @xmath120 are less than @xmath101 or all of them are greater than @xmath101 . in both cases we have @xmath128_q\ ,
q^{\des(\o)}$ ] , where the sum is over extensions @xmath126 of @xmath58 .
thus @xmath129_q\ , r_{g\setminus v}(q)$ ] .
[ def : nicepeo ] for a chordal graph @xmath50 on the vertex set @xmath51 , we say that a perfect elimination ordering @xmath102 of the vertices of @xmath50 is _ nice _ if it satisfies the following additional property . for @xmath91 , all neighbors of the vertex @xmath104 among the vertices @xmath130 are greater than @xmath104 ( in the usual order on @xmath131 ) , or all neighbors of @xmath104 among @xmath130 are less than @xmath104 . for a nice perfect elimination ordering @xmath102 of @xmath50 ,
the last vertex @xmath132 satisfies the conditions of lemma [ lem : r = mr ] .
moreover , @xmath133 is a nice perfect elimination ordering of the graph @xmath134 . in this case
, we can inductively use lemma [ lem : r = mr ] to completely factor the polynomial @xmath79 as @xmath135_q \,[m'+1]_q\cdots$ ] .
the numbers @xmath136 are exactly the exponents @xmath137 ( written backwards ) coming from this perfect elimination ordering .
[ cor : nice_r ] suppose that @xmath50 has a nice perfect elimination ordering of vertices .
let @xmath25 be the exponents of @xmath50 .
then we have @xmath138_q\ , [ e_2 + 1]_q \cdots [ e_n+1]_q\,.\ ] ]
it is convenient to represent a permutation @xmath0 as the _ rook diagram _
@xmath139 , which the placement of @xmath140 non - attacking rooks into the boxes @xmath141 of the @xmath142 board .
see an example on figure [ fig : rook_diagram ] .
we assume that boxes of the board are labelled by pairs @xmath54 in the same way as matrix elements .
the rooks are marked by @xmath143 s .
@xmath144 the inversion graph @xmath26 contains an edge @xmath54 , with @xmath145 , whenever the rook in the @xmath33-th column of @xmath139 is located to the south - west of the rook in the @xmath34-th column . in this case
, we say that this pair of rooks _ forms an inversion .
_ here are the rook diagrams of the two forbidden patterns @xmath18 and @xmath45 for smooth permutations : @xmath146 a permutation @xmath3 is smooth if and only if its diagram @xmath139 does not contain four rooks located in the same relative order as in one of these diagrams @xmath147 or @xmath148 .
let @xmath149 be the rook located in the last column of @xmath139 , and let @xmath150 be the rook located in the last row of @xmath139 .
the row containing @xmath149 and the column containing @xmath150 subdivide the diagram @xmath139 into the four sectors @xmath151 , as shown on figure [ fig : rooks_abcd ] . in the case when @xmath152 , we assume that @xmath153 and the sectors @xmath154 are empty .
let @xmath3 be a smooth permutations
. then its rook diagram @xmath139 has the following two properties .
( 1 ) each pair of rooks located in the sector @xmath155 forms an inversion .
( 2 ) at least one of the sectors @xmath38 or @xmath156 contains no rooks . for example , for the rook diagram @xmath157 shown on figure [ fig : rook_diagram ] ,
the sector @xmath38 contains one rook , the sector @xmath156 contains no rooks , and the sector @xmath155 contains two rooks that form an inversion .
\(1 ) if the sector @xmath155 contains a pair of rooks that do not form an inversion , then these two rooks together with the rooks @xmath149 and @xmath150 form a forbidden pattern as in the diagram @xmath148 .
( 2 ) if the sector @xmath38 contains at least one rook and the sector @xmath156 contains at least one rook , then these two rooks together with the rooks @xmath149 and @xmath150 form a forbidden pattern as in the diagram @xmath147
. let @xmath158 and @xmath159 be the vertices of the inversion graph @xmath26 corresponding to the rooks @xmath149 and @xmath150 . also let @xmath160 be the vertices of @xmath26 corresponding to the rooks inside the sector @xmath155 .
if the sector @xmath38 of the rook diagram @xmath139 is empty , then the vertex @xmath159 is connected only with the vertices @xmath161 , that form a clique in the graph @xmath26 , and all these vertices are greater than @xmath159 . on the other hand ,
if the sector @xmath156 of the rook diagram @xmath139 is empty , then the vertex @xmath162 is connected only with the vertices @xmath163 , that form a clique , and all these vertices are less than @xmath162 . in both cases ,
the inversion graph @xmath26 satisfies the conditions of lemma [ lem : r = mr ] , where @xmath164 if @xmath38 is empty , and @xmath165 if @xmath156 is empty .
( if both @xmath38 and @xmath156 are empty then we can pick @xmath165 or @xmath164 . ) for @xmath0 and @xmath166 , let @xmath167 be the _ flattening _ of the sequence @xmath168 , that is , the permutation @xmath169 has the same relative order of elements as in this sequence .
equivalently , the rook diagram @xmath170 is obtained from the rook diagram @xmath139 by removing its @xmath42-th column and the @xmath171-th row .
lemma [ lem : r = mr ] , together with the above discussion , implies the following recurrence relations for the polynomials @xmath172 . [ prop : factor ] let @xmath0 be a smooth permutation , and assume that @xmath173 and @xmath174 .
then ( at least ) one of the following two statements is true : 1 . @xmath175 , or 2 .
@xmath176 . in both cases , the polynomial @xmath21 factors as @xmath177_q \,r_{w'}(q),\ ] ] where @xmath178 and @xmath179 in case ( 1 ) , or @xmath180 and @xmath181 in case ( 2 ) .
in this proposition , case ( 1 ) means that the sector @xmath38 of the rook diagram @xmath139 is empty , and case ( 2 ) mean that the sector @xmath156 is empty . clearly , if @xmath3 is smooth , then the flattening @xmath182 is smooth as well .
the inversion graph @xmath183 is isomorphic to the graph @xmath184 .
this means that , for smooth @xmath0 , one can inductively use proposition [ prop : factor ] to completely factor the polynomial @xmath21 as in corollary [ cor : nice_r ] .
[ cor : rwq_factor ] for a smooth permutation @xmath0 , the inversion graph @xmath26 is chordal and , moreover , it has a nice perfect elimination ordering .
we have @xmath185_q\ , [ e_2 + 1]_q \cdots [ e_n+1]_q$ ] , where @xmath25 are the exponents of the inversion graph @xmath26 .
interestingly , gasharov @xcite found exactly the same recurrence relations for the poincar polynomials @xmath4 .
[ prop : gasharov_factorization ] ( gasharov @xcite , cf .
lascoux @xcite ) the poincar polynomials @xmath4 , for smooth permutations @xmath3 , satisfy exactly the same recurrence relation as in proposition [ prop : factor ] . note that lascoux @xcite gave a factorization of the kazhdan - lusztig basis elements , that implies proposition [ prop : gasharov_factorization ] .
propositions [ prop : factor ] and [ prop : gasharov_factorization ] , together with the trivial claim @xmath186 , imply that @xmath16 for all smooth permutations @xmath3 .
this finishes the proof of theorem [ th : p = r ] .
section [ sec : recurrence ] gives a recursive construction for a nice perfect elimination ordering of the graph @xmath26 , for smooth @xmath3 . in this section
we give a simple non - recursive construction for another perfect elimination ordering of @xmath26 .
this simple ordering may not be nice ( see definition [ def : nicepeo ] ) .
however , one still can use it for calculating the exponents of the graph @xmath26 and factorizing the polynomials @xmath16 as in corollary [ cor : rwq_factor ] .
indeed , the multiset of the exponents does not depend on a choice of a perfect elimination ordering ( see remark [ rem : does_not_depend_on_peo ] ) .
recall that a _ record position _ of a permutation @xmath0 is an index @xmath88 such that @xmath89 .
let @xmath187 $ ] denote the interval @xmath188 with the usual @xmath131-order of entries .
[ lem : speo ] for a smooth permutation @xmath0 with record positions @xmath189 , the ordering @xmath190,\ [ r_{s-1},\,r_s-1],\dots,[r_2,\,r_3 - 1],\ [ r_1,\,r_{2}-1]\ ] ] of the set @xmath51 is a perfect elimination ordering of the inversion graph @xmath26 .
example [ examp : exps ] ) the permutation @xmath97 has records @xmath191 and record positions @xmath98 .
lemma [ lem : speo ] says that the ordering @xmath192 is a perfect elimination ordering of the inversion graph @xmath26 .
figure [ fig : inv_graph ] displays this inversion graph @xmath26 . for each vertex @xmath193 of @xmath26
, we wrote @xmath33 inside a circle and @xmath194 below it .
the exponents of this graph ( i.e. , the numbers of edges going to the left from the vertices ) are @xmath195 .
suppose that this ordering of vertices of @xmath26 is not a perfect elimination ordering .
this means that there is a vertex @xmath33 connected in @xmath26 with vertices @xmath34 and @xmath42 , preceding @xmath33 in the order , such that the vertices @xmath34 and @xmath42 are not connected by an edge in @xmath26 .
let us consider three cases .
i. the vertices @xmath196 belong to the same interval @xmath197 $ ] , for some @xmath198 .
( here we assume that @xmath199 . )
we have @xmath200 and @xmath201 , @xmath202 , but @xmath203 , because @xmath204 and @xmath205 are edges of @xmath26 but @xmath206 is not an edge .
the value @xmath207 is the maximal value of @xmath3 on the interval @xmath208 .
since @xmath203 is not the maximal value of @xmath3 on @xmath208 , we have @xmath209 and so @xmath210 .
thus @xmath211 and the values @xmath212 form a forbidden @xmath45 pattern in @xmath3 .
so @xmath3 is not smooth .
contradiction .
the vertices @xmath213 are in the same interval @xmath208 and the vertex @xmath42 belongs to a different interval @xmath214 .
then @xmath215 , because the vertex @xmath42 precedes @xmath33 in the order .
in this case we have @xmath216 , @xmath202 , @xmath217 .
this implies that @xmath218 that is @xmath205 is an edge in the inversion graph @xmath26 .
contradiction .
the vertex @xmath33 belongs to the interval @xmath208 and the vertices @xmath219 do not belong to @xmath208 .
assume that @xmath220 and that @xmath34 belongs to @xmath214
. then @xmath215 . in this case ,
@xmath221 , @xmath11 , @xmath217 , and @xmath222 .
the record value @xmath223 is greater than @xmath194 .
this implies that @xmath224 . in particular , @xmath225 and , thus , @xmath226 .
we have @xmath227 and the values @xmath228 form a forbidden @xmath18 pattern . contradiction .
let us calculate the exponents of the inversion graph @xmath26 for a smooth permutation @xmath0 using the perfect elimination ordering from lemma [ lem : speo ] .
suppose that @xmath229 .
then the exponent @xmath103 of the vertex @xmath33 equals the number of neighbors of the vertex @xmath33 in the graph @xmath26 among the preceding vertices , that is among the vertices in the sets @xmath230 and @xmath231 . in other words ,
the exponent @xmath103 equals @xmath232 this is exactly the expression for @xmath103 from theorem [ th : formula ] .
the result follows from corollary [ cor : rwq_factor ] .
our proof of theorem [ th : p = r ] is based on a recurrence relation .
it would be interesting to give more direct combinatorial proof of theorem [ th : p = r ] based on a bijection between elements of the bruhat interval @xmath66 $ ] and regions of the arrangement @xmath7
. it would be interesting to better understand the relationship between bruhat intervals @xmath66 $ ] and the hyperplane arrangement @xmath7 .
one can construct a directed graph @xmath233 on the regions of @xmath7 .
two regions @xmath13 and @xmath75 are connected by a directed edge @xmath234 if these two regions are adjacent ( i.e. , separated by a single hyperplane ) and @xmath13 is more close to @xmath15 than @xmath75 . for example , for the longest permutation @xmath235 , the graph @xmath236 is the hasse diagram of the _ weak _ bruhat order .
it is true that , for any smooth permutation @xmath0 , the graph @xmath233 is isomorphic to a subgraph of the hasse diagram of the bruhat interval @xmath66 $ ] ?
it would be interesting to explain theorem [ th : p = r ] from a geometrical point of view .
is it possible to link the arrangement @xmath7 and the polynomial @xmath21 with the cohomology ring of of the schubert variety @xmath17 ?
is it possible to define a related ring structure on the regions of @xmath7 ?
the statement of theorem [ th : p = r ] can extended to any finite weyl group @xmath237 , as follows . for a weyl group element @xmath238 ,
let @xmath239 , where the sum is over all @xmath240 such that @xmath241 in the bruhat order on @xmath237 .
define the arrangement @xmath7 as the collection of hyperplanes @xmath242 for all roots @xmath243 in the corresponding root system such that @xmath244 and @xmath245 .
let @xmath15 be the region of @xmath7 that contains the fundamental chamber of the corresponding coxeter arrangement .
define @xmath246 , where the sum is over all regions of the arrangement @xmath7 and @xmath14 is the number of hyperplanes separating @xmath15 and @xmath13 .
let @xmath5 be the schubert variety in the corresponding generalized flag manifold @xmath247 .
details about ( rational ) smoothness of schubert varieties @xmath17 can be found in @xcite . | the aim of this article is to link schubert varieties in the flag manifold with hyperplane arrangements . for a permutation
, we construct a certain graphical hyperplane arrangement .
we show that the generating function for regions of this arrangement coincides with the poincar polynomial of the corresponding schubert variety if and only if the schubert variety is smooth .
we give an explicit combinatorial formula for the poincar polynomial .
our main technical tools are chordal graphs and perfect elimination orderings . |
there has been a growing database of direct supermassive black hole ( smbh ) mass ( @xmath0 ) estimates from the centers of nearby galactic bulges ( e.g. , * ? ? ?
while the limits of our current abilities to significantly expand this database may have been reached @xcite , the last decade has seen a wealth of @xmath0 estimates that have increased the smbh catalog from 13 or 26 @xcite to @xmath670 @xcite .
an intense interest in populating the smbh database was sparked by observed correlations between @xmath0 and fundamental properties of their host bulges .
these @xmath0 scaling relations have generated numerous theoretical investigations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and have possibly added valuable limits to evolutionary models ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the degree to which a smbh s sphere of influence , @xmath3 , is resolved has been used as a quality measure for @xmath0 estimates @xcite . the only method available to a priori determine
if a @xmath0 estimate can be made is to assume @xmath7 @xcite .
however , to calculate @xmath3 in galaxies with known @xmath1 , @xmath0 is estimated using the @xmath4 relation given by @xmath8 , where @xmath9 and @xmath10 .
the observed scatter , @xmath11 , is 0.4 dex @xcite . following this , figure [ fig:1 ] demonstrates where @xmath3 will be resolved , given a spatial resolution of @xmath12 .
a value of @xmath12 is used here as that is the typical fwhm of the _ hst _ psf . to date _
hst _ has been responsible for most @xmath0 estimates .
* ff05 ) discussed the limited abilities of _ hst _ to resolve @xmath3 , and ( * ? ? ? * g09 ) found the @xmath4 relation to be biased when applying the @xmath3 argument .
the influence of @xmath3 cuts on the @xmath4 relation is continued in this _
letter _ with two significant advances .
first , a sample of all galaxies with a known @xmath1 ( @xmath13100 mpc ) is used .
it is important to only use these galaxies as there are no @xmath1=400 km s@xmath14galaxies at 1 mpc , for example .
second , random @xmath0 estimates are applied to each galaxy , i.e. , no galaxy is assumed to intrinsically fall on the @xmath4 relation .
this ensures no pre - selection of galaxies that lie on the @xmath4 relation . throughout
, a distinction is made between the _ observed _
@xmath4 relation ( published values ) and the _ observable _
@xmath4 relation ( the relation that can be fitted using the data simulated here ) .
in short , these simulations take a galaxy with a known @xmath1 and distance , assign a @xmath0 , then calculate @xmath3 . the @xmath3 argument
is then applied ; if @xmath3 is unresolved the galaxy is removed from the sample ( the low mass cut ) .
if the assigned @xmath0 generates a galaxy that lies above the _ observed _ @xmath4 relation , it is also removed from the sample ( the high mass cut ) .
the _ observable _
@xmath4 relation is then fit to the remaining galaxies using a levenberg - marquardt least - squares add - on to idl . an sql search of the hyperleda catalog was performed .
all galaxies with a kinematical distance modulus less than 35.0 ( 100 mpc ) were found . of
the 49740 galaxies returned , 2518 have published values of @xmath1 .
the incompleteness of this sample , due to difficulties in measuring @xmath1 in faint galaxies , is noted .
values of @xmath3 ( in arcsecs ) were then calculated for each galaxy by assigning a @xmath0 . in the first cases ,
the low mass cut was made by applying @xmath12 to represent the _ hst _ psf .
however , several values of @xmath15 are applied later .
the high mass cuts were then made by considering the _ observed _ distribution of galaxies in the @xmath4 plane ; there is a clear absence of over - massive smbhs , i.e. , there are no @xmath1=50 km s@xmath14 dwarf spheroids in the local group with @xmath0@xmath16
. such a nearby over - massive smbh population would have been detected , and therefore it is assumed that these objects do not exist .
galaxies selected for the high mass cuts were determined by assigning a scatter ( @xmath17 dex ) and upper limits to the @xmath4 relation , @xmath18 .
galaxies with @xmath19 were removed from the sample , where @xmath20m_\odot$ ] .
each galaxy was first assigned 90 separate values of @xmath0 using a step size of @xmath21 from @xmath22 to @xmath23 . in this case , the high mass cut was made by applying the @xmath24 relation of g09 and the @xmath25 relation of ff05 .
figure [ fig:2 ] shows the distribution of _ observable _ galaxies in the @xmath4 plane assuming every galaxy within 100 mpc can host any @xmath0 and that @xmath12 .
the fit to this sample ( red line ) is @xmath26 .
however , galaxies host single ( or binary ) smbhs , therefore it is shrewd to assign each galaxy a single ( uniformly sampled ) random value for @xmath0 in the range @xmath27 .
the distribution of _ observable _ galaxies in the @xmath4 plane using a random sample of @xmath0 is shown in figure [ fig:3 ] using the red open circles . in this case , the low cut was made using @xmath12 , and the high mass cut was made using a @xmath18 relation of @xmath28 ( g09 ) .
the @xmath4 relation fit to these _ observable _ galaxies ( red line ) is @xmath29 dex .
an _ observable _
@xmath4 relation of @xmath30 dex is found using the high mass @xmath18 relation of ff05 .
figure [ fig:3 ] also shows all _ observed _ galaxies based on the combined catalogs of @xcite , @xcite and g09 .
no distinction is made between `` good '' and `` bad '' @xmath0 estimates , or differences in quoted @xmath1 ; all estimates are plotted ( 143 total ) .
lccc + 005 & @xmath31 & @xmath32 & 0.41 dex + 010 & @xmath33 & @xmath34 & 0.27 dex + 015 & @xmath35 & @xmath36 & 0.22 dex + 020 & @xmath37 & @xmath38 & 0.18 dex + + + 005 & @xmath39 & @xmath40 & 1.82 dex + 010 & @xmath41 & @xmath42 & 1.22 dex + 015 & @xmath43 & @xmath44 & 0.97 dex + 020 & @xmath45 & @xmath46 & 0.85 dex + the similarity between the @xmath4 distribution of _ observable _ random mass smbhs and _ observed _ smbhs masses is striking .
however , as this result derives from a single random sampling of @xmath0 , it represents a single possible _ observable _
@xmath4 relation .
if galaxies intrinsically have random @xmath0 values , then the range of _ observable _ smbhs could have a distribution given by figure [ fig:2 ] .
therefore , 50 separate random @xmath0 samplings were then made for each galaxy , and in each case a fit to the _ observable _
@xmath4 relation was performed .
in addition , as the previously used value of @xmath12 only applies to _ hst _ observations , the analysis was repeated for a range of @xmath3 lower cut offs .
finally , as the high mass cuts may not actually be defined by the g09 and ff05 fits , values of @xmath47 from 7.8 to 8.7 ( in 0.1 steps ) and @xmath48 from 3.6 to 5.4 ( in 0.2 steps ) were used to create an addition 100 individual @xmath18 cutoffs .
the mean values of @xmath49 and @xmath11 , derived by applying these different @xmath15 criteria , and from using these variable upper limits , are presented in table [ tab:1 ] .
table [ tab:1 ] has several notable features .
first , as @xmath15 rises @xmath11 falls .
this is expected because the range of _ observable _ smbhs decreases with larger values of @xmath15 .
second , the slope ( @xmath50 ) remains consistent .
finally , in the case of the variable upper limits , the zero - point ( @xmath51 ) remains constant and is consistent with the mid - value of the @xmath18 limits imposed on the sample . to test whether these values of @xmath51 are a consequence of the high mass cut conditions , the analysis was repeated using @xmath47 limits of 7.0 and 8.8 ( in 0.2 steps ) . applying a @xmath12 low mass cut , an _
observable _ relation with @xmath52 was found .
again , the values of @xmath51 are consistent with the mid - value of the imposed limits .
however , an estimate of the _ observed _
@xmath18 relation can be made . the _ observed _ smbh sample in figure [ fig:3
] was used to define an upper limit by fitting all _ observed _ smbhs that fall above the ff05 and g09 relations , and above @xmath17 .
the fit to these over - massive smbhs produce an upper limit of @xmath53 ( fig .
[ fig:4 ] ) .
this @xmath18 fit was then applied to 50 random values of @xmath0 in each galaxy , and the _ observable _
@xmath4 relations were then fitted in each case .
the average of these fits are also presented in table [ tab:1 ] . in this case , the fitted values of @xmath51 and @xmath50 are consistent with the variable upper limits , but the values of @xmath11 are significantly larger .
this is expected , as these fits were applied to data that had the maximum range in the @xmath4 plane due to the high _ observed _ values of @xmath47 and @xmath48 .
in summary , as a result of applying the @xmath3 argument , a @xmath4 relation consistent with _ observed _ values ( @xmath54 dex ) can be fitted to a sample of galaxies that contain random mass smbhs , and as a consequence do not follow a @xmath4 relation .
the @xmath3 argument removes low mass smbhs where @xmath3 would not be resolved , and high mass smbhs where @xmath3 should have been resolved if such a population were present . therefore , for scaling relations to be of value in constraining galaxy evolution models , @xmath0 estimates must not be solely proposed on the basis of critically resolving values of @xmath3 derived from @xmath1 .
it is unlikely that an intrinsic @xmath4 relation has been _ observed _ as a result of critically sampling @xmath3 in the @xmath4 plane , and it will be some time until we are able to sample a significant area rightward of the _ observed _ @xmath4 relation ( e.g. , ff05 ; * ? ? ? * ) .
therefore , it is clear that caution must be used in the application of the _ observed _
@xmath4 relation until this issue can be resolved .
the possibility of there being no tight intrinsic @xmath4 relation , and that the distribution of _ observed _ galaxies in the @xmath4 plane determine an upper limit only , is now discussed . as the same @xmath0 estimates are used to define the other @xmath0 scaling relations
, this also implies that all _ observed _ scaling relations are upper limits .
there are a number of questions arising from this possibility , beginning with the most fundamental .
why do _ observed _ smbhs fall on the @xmath4 relation ?
there are several smbhs with such high quality @xmath0 estimates ( sgr a * , ngc 4258 , m87 ) , that there clearly is a relation between @xmath0 and @xmath1 in some galaxies .
however , the @xmath4 plane has been increasingly populated with less certain @xmath0 estimates that have potentially been included on the assumption that @xmath3 is spatially resolved , i.e. , that they follow a previously estimated @xmath4 relation defined by higher quality @xmath0 estimates . in these cases , as the simulations presented here show , an _ observed _
relation will arise simply as a result of the @xmath3 selection effect , even if @xmath0 is randomly distributed within galaxies .
does a population of under - massive smbhs exist , and have they been detected ? if the @xmath4 relation is an upper limit , then there should be galaxies that host low mass smbhs , as suggested by simulations @xcite .
indeed , both the simulations presented here , and the observations plotted in figure [ fig:3 ] , show that under - massive smbhs can be , and have been , detected .
some notable cases are that of ngc 4435 @xcite , a sample of barred galaxies @xcite and narrow - line seyfert 1 galaxies @xcite .
in fact , the evidence to suggest the presence of under - massive smbhs is not matched by any evidence for a significant population of over - massive smbhs leftward of the _ observed _
@xmath4 relation .
it is important to note some intricacies with the two dominant techniques for measuring @xmath0 ( stellar and gas dynamics ) . in the case of stellar dynamics , there are systematics to the models that may allow a large range of @xmath0 in a given bulge @xcite . in the case of gas dynamics , it is unclear what the inclination of the nuclear gas disk may be ( e.g. , * ? ? ?
both methods allow the potential for many of the current @xmath0 estimates to in fact be upper limits , i.e. , the true @xmath4 plane may have a large distribution of under - massive smbhs .
therefore , under - massive smbhs may have been observed , their mass over - estimated , and their impact over - looked .
all smbh models allow stringent upper limits to be placed on @xmath0 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . however , these @xmath0 upper limits will also be dependent on the available spatial resolution .
any kinematical data will have two velocity points spatially separated on a scale of @xmath15 .
an upper limit to @xmath0 is estimated by including an increasing dark mass until the derived model becomes inconsistent with the data , i.e. , a higher upper limit to @xmath0will be estimated using a lower @xmath15 .
there are still important constraints that can be added to the @xmath4 plane from estimates of @xmath0 upper limits , however , as upper limits that fall below the _ observed _
@xmath4 relation provide the same evidence for under - massive smbhs as would a tightly constrained low mass @xmath0 . at present , most @xmath0 upper limits are based on data derived from gas dynamics .
consequently , these limits generally fall above the _ observed _
@xmath4 relation due to unknown amounts of line broadening from non - gravitational processes , and uncertainties in the inclination if the nuclear gas disk .
if the @xmath4 relation represents an upper limit in the @xmath4 plane , then what is this limit ? in [ mands ]
an upper limit of @xmath53 was found based on the distribution of _ observed _ smbhs leftward of the _ observed _
@xmath4 relation .
this is likely a good approximation to the @xmath18 limit as there are no reports of a steeper relation .
however , this estimate does not include the potential over - massive smbhs from the upper limits calculated by @xcite . including these limits to the upper limit sample
gives values of @xmath55 to @xmath18 .
as expected , due to the addition of smbh limits at lower @xmath0 , this @xmath18 limit is more shallow with a higher zero - point . including these limits at the lower @xmath0 end of the @xmath4 plane addresses , in part , a limitation of the sample used here .
as already noted , the @xmath1 catalog used here is likely incomplete due to the difficulty of measuring @xmath1 in faint galaxies at greater distances .
in addition , the @xmath1 catalog likely contains inhomogenuous measurements that may not translate from bulge to bulge .
what are the consequences to galaxy evolution models if there is only a @xmath18 relation ?
first , galaxies will no longer be required to obey the @xmath4 relation , and could host a smbh with any @xmath0 below @xmath18 .
models that include feedback from the smbh to the galaxy will then need to be carefully reconsidered . while the smbh will undoubtedly have some influence on a portion of the host galaxy
, it would not need to affect large scale properties ; evolution of the smbh would be a result of host galaxy evolution .
an upper limit in the @xmath4 plane would also represent the pinnacle of smbh evolution as a function of @xmath1 , in which galaxies evolve up to the @xmath18 limit .
a signature of such a scenario could be a cosmic variation in @xmath11 ( the scatter would increase with redshift ) and an _ observed _ @xmath4 relation that does not exceed @xmath18 .
if the distribution of @xmath0 is random within bulges , then when compared with the local _ observed _ @xmath4 relation , @xmath0 estimates from higher redshift could fall to the left or the right .
@xcite find a population of z=0.36 seyfert 1 galaxies that lie above the local _ observed _
@xmath4 relation by @xmath56@xmath0@xmath57 , but this population still lies below the @xmath18 limit estimated here . finally , if smbhs can reside anywhere below the @xmath18 limit , then the local black hole mass function may have been over - estimated .
this would relax the observation that merging is not important and that smbh growth is dominated by accretion ( e.g. , * ? ? ?
this potentially allows anti - hierarchical smbh growth to no longer present problems for hierarchical galaxy formation models . the use of the hyperleda database ( http://leda.univ-lyon1.fr ) is acknowledged .
db thanks alessandro marconi , david merritt and andy robinson for useful discussions , and acknowledges the referee for suggesting improvements . support for this work was provided by proposal number hst - ar-10935.01 awarded by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 .
adams , f. c. , graff , d. s. , mbonye , m. , & richstone , d. o. 2003 , , 591 , 125 baes , m. , buyle , p. , hau , g. k. t. , & dejonghe , h. 2003 , , 341 , l44 batcheldor , d. , & koekemoer , a. m. 2009 , , 121 , 1245 beifiori , a. , sarzi , m. , corsini , e. m. , dalla bont , e. , pizzella , a. , coccato , l. , & bertola , f. 2009 , , 692 , 856 cattaneo , a. , blaizot , j. , devriendt , j. , & guiderdoni , b. 2005 , , 364 , 407 ciotti , l. 2008 , arxiv:0808.1349 ciotti , l. , & van albada , t. s. 2001 , , 552 , l13 coccato , l. , sarzi , m. , pizzella , a. , corsini , e. m. , dalla bont , e. , & bertola , f. 2006 , , 366 , 1050 ferrarese , l. 2002 , , 578 , 90 ferrarese , l. , & ford , h. 2005 , space science reviews , 116 , 523 ferrarese , l. , & merritt , d. 2000 , , 539 , l9 gebhardt , k. , et al .
2000 , , 539 , l13 graham , a. w. 2008a , , 680 , 143 graham , a. w. 2008b , publications of the astronomical society of australia , 25 , 167 graham , a. w. , erwin , p. , caon , n. , & trujillo , i. 2001 , , 563 , l11 graham , a. w. , & li , i. 2009 , , 698 , 812 gltekin , k. , et al .
2009 , , 698 , 198 hring , n. , & rix , h .- w .
2004 , , 604 , l89 heckman , t. m. , kauffmann , g. , brinchmann , j. , charlot , s. , tremonti , c. , & white , s. d. m. 2004 , , 613 , 109 hu , j. 2008 , , 386 , 2242 kormendy , j. , & richstone , d. 1995 , , 33 , 581 magorrian , j. , et al .
1998 , , 115 , 2285 marconi , a. , et al .
2003 , , 586 , 868 marconi , a. , & hunt , l. k. 2003 , , 589 , l21 marconi , a. , risaliti , g. , gilli , r. , hunt , l. k. , maiolino , r. , & salvati , m. 2004 , , 351 , 169 mathur , s. , & grupe , d. 2005 , , 633 , 688 novak , g. s. , faber , s. m. , & dekel , a. 2006 , , 637 , 96 paturel , g. , petit , c. , prugniel , p. , theureau , g. , rousseau , j. , brouty , m. , dubois , p. , & cambrsy , l. 2003 , , 412 , 45 peebles , p. j. e. 1972 , , 178 , 371 pizzella , a. , corsini , e. m. , dalla bont , e. , sarzi , m. , coccato , l. , & bertola , f. 2005 , , 631 , 785 robertson , b. , hernquist , l. , cox , t. j. , di matteo , t. , hopkins , p. f. , martini , p. , & springel , v. 2006 , , 641 , 90 sarzi , m. , et al .
2002 , , 567 , 237 treu , t. , woo , j .- h . , malkan , m. a. , & blandford , r. d. 2007 , , 667 , 117 valluri , m. , merritt , d. , & emsellem , e. 2004 , , 602 , 66 vittorini , v. , shankar , f. , & cavaliere , a. 2005 , , 363 , 1376 volonteri , m. 2007 , , 663 , l5 wyithe , j. s. b. , & loeb , a. 2005 , , 634 , 910 | the observed relation between supermassive black hole ( smbh ) mass ( @xmath0 ) and bulge stellar velocity dispersion ( @xmath1 ) is described by @xmath2 .
as this relation has important implications for models of galaxy and smbh formation and evolution , there continues to be great interest in adding to the @xmath0 catalog .
the `` sphere of influence '' ( @xmath3 ) argument uses spatial resolution to exclude some @xmath0 estimates and pre - select additional galaxies for further smbh studies .
this _
letter _ quantifies the effects of applying the @xmath3 argument to a population of galaxies and smbhs that do not follow the @xmath4 relation .
all galaxies with known values of @xmath1 , closer than 100 mpc , are given a random @xmath0 and selected when @xmath3 is spatially resolved .
these random smbhs produce a @xmath4 relation of @xmath5 , consistent with observed values .
consequently , future proposed @xmath0 estimates should not be justified solely on the basis of resolving @xmath3 . this _
letter _ shows the observed @xmath4 relation may simply be a result of available spatial resolution .
however , it also implies the observed @xmath4 relation defines an upper limit .
this potentially provides valuable new insight into the processes of galaxy and smbh formation and evolution . |
in this paper , we consider the martin compactification of a complete non - compact surface @xmath4 with negative curvature .
in particular , we investigate positive eigenfunctions with eigenvalue one of the laplace operator @xmath2 of @xmath4 and prove a certain uniqueness result .
the problem arises in our study of deformations of three - dimensional collapsed gradient steady ricci solitons @xcite . a classical result due to g. herglotz @xcite says that any positive harmonic function @xmath5 on the unit disk @xmath6 has an integral representation @xmath7 where @xmath8 is the poisson kernel of @xmath6 and @xmath9 is a finite positive borel measure .
a fundamental generalization of the above formula to bounded domains in @xmath10 was given by r. martin in @xcite .
he showed that an analogue of the above representation formula holds in complete generality , where the integral is taken over an ideal boundary , called the martin boundary , defined in terms of the limiting behavior of a green s function .
later , martin s results were extended to the second order elliptic differential operators on complete riemannian manifolds , see , e.g. , @xcite and the references therein . on a riemannian manifold of non - positive curvature ,
there is a well - known compactification by attaching its geometric boundary that is determined by the asymptotic behavior of geodesics at infinity , see @xcite .
a celebrated result by m. anderson and r. schoen @xcite states that on a negatively curved manifold @xmath11 , its martin boundary with respect to the laplace operator is homeomorphic to its geometric boundary , provided the sectional curvature is pinched between two negative constants .
thus the martin compactification of such a manifold associated to the laplace operator coincides with its geometric compactification .
anderson - schoen s result has been generalized by a. ancona in @xcite to weakly coercive elliptic operators @xmath12 using different techniques . in @xcite , w. ballmann constructed examples of non - positively curved manifolds containing flats of dimension @xmath13 such that the martin compactification agrees with the geometric compactification .
the cartan - hadamard surface @xmath4 we are concerned with is diffeomorphic to the upper half - plane @xmath14 with the length element given by @xmath15 for @xmath16 .
we are especially interested in characterizing non - negative functions @xmath17 on @xmath1 such that @xmath18 where @xmath19 note that the laplace operator of @xmath4 is given by @xmath20 hence such a function @xmath21 is in fact an eigenfunction of @xmath2 with eigenvalue one , @xmath22 this problem of characterizing non - negative solutions of ( [ eqn : pdew ] ) arises in our study of the infinitesimal rigidity of deformations of the collapsed 3d cigar soliton @xmath23 , the product of hamilton s cigar soliton @xmath24 and the real line @xmath25 with the product metrics , as follows : consider any one - parameter family of complete three - dimensional gradient steady ricci solitons @xmath26 ( @xmath27 ) , with @xmath28 being the 3d cigar soliton @xmath29 , satisfying the following two conditions : * the metric @xmath30 admits a non - trivial killing vector field for all @xmath31 , * the scalar curvature @xmath32 of @xmath30 attains its maximum at some point on @xmath33 for each @xmath31 .
then , the first variation of the sectional curvatures of @xmath26 produces a certain non - negative function @xmath17 on @xmath1 satisfying ( [ eqn : pdew ] ) .
the infinitesimal rigidity of the deformation essentially reduces to proving the uniqueness of such nonnegative eigenfunctions up to a positive constant multiple ; see @xcite for the details . this led us to consider the martin compactification of @xmath4 associated with the operator @xmath0 .
it turns out that the gauss curvature of @xmath4 is negative , bounded from below , but approaches zero along some paths to infinity .
so we can not apply the results of anderson - schoen @xcite and ancona @xcite in this case .
nevertheless , we shall see that the martin boundary with respect to the operator @xmath34 is the same as the geometric boundary of @xmath4 .
our first results are the following martin and geometric compactifications of @xmath1 .
[ thm : martinboundary ] the martin compactification of @xmath1 associated with the operator @xmath0 is homeomorphic to the closed half disk @xmath35 . on the martin boundary @xmath36
we have 1 . the real line @xmath37 is identified with @xmath38 ; 2 .
the @xmath39-axis with @xmath40 is identified with the point @xmath41 ; 3 .
the asymptotic ray @xmath42 is identified with the point @xmath43 on the semi - circle @xmath44 a. each @xmath43 in case ( 3 ) can be approached by geodesic that is asymptotic to the ray @xmath45 as @xmath46 .
b. under the topology in the martin compactification , we have @xmath47 these two points can be approached by geodesics which are asymptotic to @xmath48 as @xmath49 . c. we show that @xmath36 is the minimal martin boundary and determine the kernel function @xmath50 , see proposition [ prop : martinkernel ] and theorem [ thm : martincompsigma ] .
[ thm : geomcompactificationintro ] the geometric compactification @xmath51 of @xmath1 with the metric in ( [ eqn : ds2surface ] ) is homeomorphic to the martin compactification @xmath52 .
the detailed description of @xmath51 is given in theorem [ thm : geometriccompactification ] .
in @xcite , using an elementary method , l. a. caffarelli and w. littman showed that for any positive solution @xmath5 to the equation @xmath53 on the euclidean space @xmath10 there exists a unique non - negative borel measure @xmath54 on the unit sphere @xmath55 such that @xmath56 it follows that the ( minimal ) martin boundary of @xmath10 with respect to @xmath57 is @xmath55 .
this is similar to the semi - circle part of @xmath36 when @xmath58 ; note that , on @xmath1 , along the ray @xmath59 the gauss curvature approaches zero as @xmath49 .
the geometric boundary of @xmath10 with the flat metric is also @xmath55 , i.e. , any point on the martin boundary can be reached by a geodesic .
[ rem : laplace ] since @xmath60 is conformal to the hyperbolic plane , the martin boundary of @xmath1 associated with the laplace operator @xmath2 is given by the union of the real line @xmath61 and @xmath62 , the point at infinity , thus is different from the geometric boundary @xmath63 .
note that the laplace operator @xmath2 has zero as the bottom of the @xmath64-spectrum , i.e. , @xmath65 ( see remark [ rem : bottoml2spectrum ] ) , and is not weakly coercive .
the martin compactification of a complete riemannian manifold @xmath66 with respect to an operator @xmath12 is also related to the dirichlet problem at infinity , i.e. , given a continuous function @xmath67 on the geometric boundary @xmath68 of @xmath66 , whether there is a unique @xmath12-harmonic function @xmath69 on @xmath66 such that @xmath70 on @xmath68 .
when @xmath12 is the laplace operator @xmath2 , the dirichlet problem at infinity is always solvable if @xmath66 has negatively pinched curvature @xmath71 ( which was first proved independently by m. anderson @xcite and d. sullivan @xcite ) , or if @xmath66 is one of the examples in @xcite .
note that in both cases , as we mentioned before , there holds the stronger conclusion that martin and geometric compactifications are homeomorphic .
we remark that the dirichlet problem at infinity for @xmath2 on a symmetric space @xmath66 of noncompact type was investigated by h. frstenberg @xcite ; in particular , he showed that the problem can be solved if and only if @xmath66 has rank one .
see also the earlier work of l .- k .
hua @xcite on bounded symmetric domains . for more general cartan - hadamard manifolds of rank one ( in the sense of @xcite ) , the solvability of the dirichlet problem at infinity was proved by ballmann @xcite .
moreover , the poisson integral representation formula was established by ballmann - ledrappier @xcite .
meanwhile , h. i. choi @xcite , ding - zhou @xcite , and e. p. hsu @xcite have shown that the dirichlet problem at infinity for the laplace operator @xmath2 is solvable on certain negatively curved manifolds whose curvature approaches zero with certain rate .
very recently , r. neel @xcite has shown that the asymptotic dirichlet problem for the laplace operator on a cartan - hadamard surface is solvable under the curvature condition @xmath72 ( in polar coordinates with respect to a pole ) outside of a compact set , for some @xmath73 .
however , in our case , theorem [ thm : geomcompactificationintro ] and remark [ rem : laplace ] imply the following [ cor : dirichlet ] the dirichlet problem at infinity for the laplace operator @xmath74 on @xmath1 is not always solvable .
\(a ) we also characterize those functions @xmath75 for which the dirichlet problem at infinity has a unique solution , see remark [ rem : dirichletuniqueness ] .
\(b ) corollary [ cor : dirichlet ] gives a new example of a cartan - hadamard manifold for which the dirichlet problem at infinity is not solvable in general .
note that the curvature of @xmath1 is bounded from below by a negative constant , but approaches zero exponentially fast as @xmath40 for each fixed @xmath76 ( see remark [ rem : gaussdecay ] ) .
previously , when the curvature of a cartan - hadamard manifold is assumed to be bounded from above by a negative constant but not from below , ancona @xcite constructed a counterexample to the solvability of dirichlet problem at infinity .
see also the work of borbly @xcite . on the other hand ,
there are several papers on the solvability of the dirichlet problem at infinity when the curvature lower bound has a quadratic growth condition @xcite , or certain exponential growth conditions @xcite .
an eigenfunction @xmath77 of the laplace operator @xmath2 with eigenvalue one is also referred as an @xmath34-harmonic function since @xmath78 .
each boundary point @xmath79 associates a kernel function @xmath50 that is positive on @xmath80 and @xmath34-harmonic .
the martin integral representation theorem implies that for any positive @xmath34-harmonic function @xmath69 , there is a ( unique ) finite non - negative borel measure @xmath81 on @xmath36 such that @xmath82 from this integral representation we derive the following uniqueness result of positive eigenfunctions .
[ thm : uniquenesswpos ] suppose that @xmath21 is a non - negative @xmath34-harmonic function on @xmath80 which vanishes on the boundary @xmath61 and satisfies the following inequality : @xmath83 for some point @xmath84 .
then either @xmath85 , or it is a positive constant multiple of @xmath86 a. theorem [ thm : uniquenesswpos ] is used in @xcite to prove an infinitesimal rigidity result of the gradient steady ricci soliton @xmath87 , where @xmath24 is hamilton s cigar soliton @xcite .
b. in @xcite we used the statement of theorem [ thm : uniquenesswpos ] with ( [ eqn : wgrowth ] ) replaced by the inequality @xmath88 which is a stronger assumption .
now we outline the main steps in the proofs .
since the metric of @xmath1 is explicit we obtain an integral formula of green s function @xmath89 of @xmath34 by the classical work of e. titchmarsh in @xcite .
the martin kernel function is derived by asymptotic expansions of @xmath90 along various paths . in turn
, it determines the martin compactification of @xmath80 .
see theorem [ thm : martincompsigma ] for theorem [ thm : martinboundary ] and theorem [ thm : geometriccompactification ] for theorem [ thm : geomcompactificationintro ] , while corollary [ cor : dirichlet ] is proved at the end of section 6 , and theorem [ thm : uniquenesswpos ] is proved in section 7 .
we refer to the table of contents for an overview of the paper s organization . * acknowledgement .
* we are grateful to werner ballmann , ovidiu munteanu , christian remling , jiaping wang , xiaodong wang and meijun zhu for helpful discussions .
part of the work was carried out while the first author was visiting the university of macau , where he was partially supported by science and technology development fund ( macao s.a.r . )
grant fdct/016/2013/a1 , as well as rdg010 of university of macau .
in this section we collect some basics of positive solutions to linear elliptic equations on complete riemannian manifolds and martin compactification of a complete riemannian manifold @xmath11 with respect to an elliptic operator @xmath12 .
let @xmath11 be an @xmath91-dimensional complete non - compact riemannian manifold and consider the operator @xmath92 .
the results in this subsection hold for any second order linear elliptic operator @xmath12 with uniformly hlder continuous coefficients and @xmath93 , see for example @xcite
. denote @xmath94 the diagonal set .
recall the following [ defn : greenfunction ] let @xmath12 be a second order linear elliptic operator on @xmath11 .
a _ green s function _ of @xmath12 is a function @xmath95 such that the following two equations hold : @xmath96 and @xmath97 for any smooth function @xmath98 with compact support on @xmath66 .
let @xmath99 be an exhaustion of @xmath66 by relatively compact subsets with @xmath100 boundary , and let @xmath101 be the unique green s function of @xmath12 on @xmath102 with the dirichlet boundary condition , then we have @xmath103 whenever the green s function @xmath90 exists , see for example ( * ? ? ?
* proposition 5.6 ) .
such a green s function is called minimal , see @xcite when @xmath104 .
we assume that @xmath12 admits a green s function @xmath90 . in
the following we describe the martin compactification from the limiting behavior of the green s function @xmath90 , see also @xcite .
fix a reference point @xmath105 and consider the normalized green s function @xmath106 for any fixed @xmath107 , the function @xmath108 is harmonic on @xmath109 and equals to @xmath110 at @xmath111 . by the harnack principle ,
any sequence @xmath112 with @xmath113 has a subsequence @xmath114 such that @xmath115 converges .
the limit is a positive @xmath12-harmonic function on @xmath66 with value @xmath110 at @xmath111 .
now consider the space of all sequences @xmath116 in @xmath66 with @xmath117 such that @xmath118 converges .
two such sequences @xmath116 and @xmath119 in @xmath66 are equivalent if their corresponding limit functions coincide . the _
martin boundary _ @xmath120 is defined as the space of equivalence classes .
the _ martin topology _ on @xmath121 induces the given topology on @xmath66 and is such that a sequence @xmath122 converges to @xmath123 if and only if @xmath118 converges to @xmath50 .
the space @xmath124 is compact with respect to the martin topology and is called _ martin compactification _ of @xmath66 .
the martin topology on @xmath124 is equivalent to the one induced by the following metric : @xmath125 where @xmath126 $ ] is any positive continuous function and integrable on @xmath66 .
recall a positive @xmath12-harmonic function @xmath69 on @xmath127 is called _ minimal _ if for any @xmath12-harmonic function @xmath128 , @xmath129 on @xmath66 implies @xmath130 for some constant @xmath131 .
a boundary point @xmath123 is called _ minimal _ if @xmath50 is a minimal @xmath12-harmonic function .
denote @xmath132 the set of all minimal boundary points .
next we collect some basic results on martin kernels and martin integral representations , see , e.g. , ( * ? ? ?
* theorem 1.10 ) and ( * ? ? ?
* section 6 ) .
[ thm : martinintegralrep ] let @xmath11 be a complete riemannian manifold and @xmath12 an elliptic operator on @xmath66 with the green s function @xmath90 .
then , a. any positive minimal @xmath12-harmonic function is a positive constant multiple of @xmath50 for some @xmath133 .
b. @xmath134 is a @xmath135 set , i.e. a countable intersection of open subsets of @xmath120 . c. @xmath136 is continuous on @xmath137 .
d. for any positive @xmath12-harmonic function @xmath5 , there exists a unique finite borel measure @xmath81 on @xmath120 such that @xmath138 and @xmath139 let @xmath140 be a bounded domain .
then , the first eigenvalue of @xmath2 on @xmath141 with the dirichlet boundary condition is given by @xmath142 denote @xmath143 the bottom of the @xmath64-spectrum of @xmath2 , then we have @xmath144 where @xmath145 is any exhaustion of @xmath66 by relatively compact subsets with @xmath100 boundary .
we recall the following well - known result of existence of positive @xmath12-harmonic function for @xmath146 , see for example , ( * ? ? ?
* theorem 1 ) .
the equation @xmath147 for @xmath148 has a positive solution @xmath5 on @xmath66 if and only if @xmath149 .
[ rem : bottoml2spectrum ] note that on the surface @xmath1 , the metric defined by ( [ eqn : ds2surface ] ) is asymptotic to the flat one as @xmath39 approaches infinity , so we have @xmath150 . for an elliptic operator @xmath12 , while a green s function always exists locally , the existence of global green s function requires extra conditions .
the result below follows from ( * ? ? ?
* corollary 5.13 ) .
a riemannian manifold @xmath11 admits a green s function if either one of the following conditions holds : 1 .
the function @xmath110 is not @xmath12-harmonic , or 2 .
there are two non - proportional positive @xmath12-harmonic functions on @xmath66 . in some special case where @xmath127 is diffeomorphic to @xmath10 and @xmath151
is a separable equation , titchmarsh showed that the green s function of @xmath12 has an explicit integral form , see @xcite or ( * ? ? ?
* chapter 15 ) .
[ thm : titchmarshgreen ] let @xmath152 be a continuous function on @xmath153 . denote the green s functions @xmath154 and @xmath155 of the differential operators @xmath156 and @xmath157 respectively with @xmath158 assume that the spectra of @xmath156 , @xmath157 are bounded below at @xmath159 and @xmath160 respectively . then for @xmath161 , the green function @xmath162 associated with the operator @xmath163 has the following integral form @xmath164 where the integral is taken along a straight line @xmath165 with @xmath166 .
a. the green s function @xmath162 is constructed using the exhaustion of @xmath167 by rectangles of finite size .
it agrees with the construction of the minimal positive green s function in definition [ defn : greenfunction ] for complete riemannian manifolds .
b. similar formulae hold when there are more than two independent variables .
in this section we determine all geodesics on @xmath1 and then the geometric compactification of @xmath1 , see proposition [ prop : sigmageodesics ] and theorem [ thm : geometriccompactification ] .
recall that the surface @xmath168 has the length element , see ( [ eqn : ds2surface ] ) , of the form @xmath169 with @xmath170 clearly , @xmath171 is a positive definite warped product metric on @xmath14 so it is complete in @xmath76-direction for any fixed @xmath39 .
it is also complete as @xmath40 since @xmath172 converges to @xmath173 .
when @xmath174 we have @xmath175 hence it follows that the metric is also complete as @xmath174 . therefore , @xmath4 is a complete surface .
moreover , its gauss curvature is given by @xmath176 thus @xmath177 note that the minimal value @xmath178 of @xmath179 is achieved at @xmath180 .
to summarize , we have the following [ prop : surfacecomplete ] @xmath181 , with the metric @xmath182 given by ( [ eqn : ds2surface ] ) , is a complete surface with negative gauss curvature bounded below by @xmath183 .
[ rem : gaussdecay ] let @xmath184 denote the distance function to a fixed horizontal line @xmath185 , say @xmath186 .
then , by ( 3.1 ) and ( 3.2 ) , we have the following asymptotic properties : @xmath187 as @xmath40 .
in particular , @xmath188 approaches zero exponentially fast when @xmath40 .
figure [ fig : geodesicsigma ] shows the typical geodesics on @xmath1 .
we sketch the geodesics passing through the @xmath39-axis at the point @xmath189 with @xmath190 .
the others can be obtained by translation in @xmath76-direction . *
the vertical dashed blue line is of type ( i ) and it has constant value of @xmath76 .
* the red curves are of type ( ii ) and they have horizontal tangent vector . * the green curves are of type ( iii ) and they are asymptotic to @xmath48 for large @xmath191 . * the purple curves are of type ( iv ) and they are asymptotic to the rays @xmath45 for large @xmath191 with @xmath192 . ] for the analytic formula of the geodesics of each type , see proposition [ prop : sigmageodesics ] below .
[ prop : sigmageodesics ] all geodesics of @xmath1 can be obtained by translation in @xmath76-direction , the reflection about the @xmath39-axis or their combinations from the following ones through some point @xmath189 with @xmath190 : a. the @xmath39-axis .
b. the geodesic has the horizontal tangent vector at @xmath189 and it is given by the following formula @xmath193 with @xmath194 $ ] . c. the geodesic has the slope of the tangent vector @xmath195 at @xmath189 and it is given by the following formula @xmath196 with @xmath197 .
d. the geodesic has the slope of the tangent vector @xmath198 at @xmath189 and it is given by the following formula @xmath199 with @xmath197 and @xmath200 the geodesics of type ( i ) , ( iii ) and ( iv ) have no horizontal tangent vector at any point .
if a geodesic has a horizontal tangent vector somewhere , then it can be obtained by translation in @xmath76-direction from a geodesic of type ( ii ) .
first of all , the nonzero christoffel symbols at any point @xmath201 are given by @xmath202 with @xmath203 so the geodesic equations are given by @xmath204 it is obvious that all vertical ( half ) lines are geodesics .
next , for any @xmath205 and @xmath190 we consider the geodesic @xmath206 passing through @xmath207 with tangent vector @xmath208 . since the metric is invariant under the translation in @xmath76-coordinate and the reflection about the @xmath39-axis , we may assume that @xmath209 and @xmath210 . from equation ( [ eqn : geodesicx ] )
we get @xmath211 for some constant @xmath212 .
since @xmath213 and @xmath214 we have @xmath215 let @xmath216 , then the second equation ( [ eqn : geodesicxi ] ) can be written as @xmath217 it follows that @xmath218 for some constant @xmath219 . when @xmath220 , since @xmath221 with @xmath214 the denominator of @xmath222 vanishes and thus we have @xmath223 equations ( [ eqn : geodesicdxt ] ) and ( [ eqn : dsxi ] ) imply that @xmath224 integrating the equation above yields the formula ( [ eqn : geodesicxyhorizontal ] ) . when @xmath225 , the constant @xmath219 can be solved from equation @xmath226 as @xmath227 and we have @xmath228 we separate the discussion into three different cases according to the sign of the coefficient @xmath229 . if @xmath230 , i.e. , @xmath231 , then equation ( [ eqn : dxdym ] ) has the solution @xmath232 with @xmath233 $ ] and the function @xmath234 is given by @xmath235 with @xmath236 note that the geodesic @xmath237 has horizontal tangent vector at @xmath238 and it can be obtained by translation in @xmath76-direction @xmath239 from the geodesic in case ( ii ) with @xmath240 . if @xmath241 , i.e. , @xmath242 , then equation ( [ eqn : dxdym ] ) has the solution @xmath243 given in case ( iii ) . if @xmath244 , i.e. , @xmath245 , then equation ( [ eqn : dxdym ] ) has the solution as in case ( iv ) .
this finishes the proof .
[ thm : geometriccompactification ] the geometric compactification of @xmath1 is homeomorphic to the half disk @xmath246 . on the geometric boundary @xmath247
we have 1 .
the geodesics that approach to points of the real line @xmath248 are identified with points of the interval @xmath249 , 2 .
the vertical half lines with @xmath40 are identified with the point @xmath250 , 3 .
the geodesics asymptotic to @xmath251 for some @xmath190 are identified with the point @xmath252 , 4 . the geodesics asymptotic to @xmath253 with @xmath254 given in equation ( [ eqn : geodesicf ] ) for some @xmath190 are identified with the point @xmath255(@xmath256 ) with @xmath257 and @xmath198 . without loss of generality
we consider the geodesic @xmath258 starting from the point @xmath259 , i.e. , @xmath260 .
we choose the parameter @xmath261 such that @xmath262 . if @xmath263 , then @xmath258 with @xmath264 is the @xmath39-axis with @xmath265 .
if @xmath266 , then it approaches to @xmath267 along the @xmath39-axis with @xmath268 .
next we assume that @xmath269 is not parallel to the @xmath39-axis .
denote by @xmath98 the angel from the positive @xmath76-axis to @xmath269 and @xmath270 .
we consider the case when @xmath271 .
the argument for @xmath272 is similar . from the proof of proposition [ prop : sigmageodesics ]
, we see that when @xmath273 , @xmath258 approaches to the positive @xmath76-axis as @xmath274 and the limits @xmath275 cover the whole positive @xmath76-axis @xmath276 . when @xmath277 , the geodesic is given by @xmath278 and it is asymptotic to the curve @xmath279 for large @xmath280 .
when @xmath281 , the geodesic is given by @xmath282 as @xmath254 in equation ( [ eqn : geodesicf ] ) .
it is asymptotic to the ray @xmath283 for large @xmath280 .
so we have the homeomorphism from the directions at @xmath284 to @xmath285 such that @xmath286 is identified with @xmath287 , @xmath288 with @xmath289 , @xmath290 with @xmath291 , @xmath292 with @xmath41 , and @xmath293 with @xmath294(@xmath295 ) by equation ( [ eqn : tanthetam ] ) with the plus sign .
in this section we prove an integral formula of the green s function @xmath296 of @xmath297 , see theorem [ thm : greenintegral ] .
recall that the laplace operator of @xmath4 is given by @xmath298 so we have @xmath299 here and in section 5 we shall use various properties of certain special functions , e.g. , the gamma function , the gauss hypergeometric function , etc .
we refer the reader to @xcite and @xcite for more details .
first of all , we define a few relevant functions for the rest of the paper .
these functions arise in the study of the spectral properties of the differential operator @xmath300 defined on @xmath276 , see appendix [ sec : stspectrum ] . for any complex number @xmath301
, we write @xmath302 and define the function @xmath303 if we use the phase angle @xmath304 instead and write @xmath305 then we have @xmath306 the function @xmath307 has branch cut along @xmath308 . using @xmath309
we define the following functions for @xmath310 and @xmath311 : @xmath312 @xmath313 here @xmath314 is the gauss hypergeometric function .
we also denote @xmath315 the wronskian of @xmath316 and @xmath317 is computed in the proof of theorem a.2 and is given by @xmath318 we define the following differential operators : @xmath319 the green functions @xmath320 of @xmath321 ( @xmath322 ) are given by @xmath323 for @xmath324 . here
@xmath325 if @xmath326 with @xmath327 and @xmath328 .
+ moreover the green function @xmath329 has the branch cut along @xmath330 $ ] .
the green function @xmath331 of @xmath156 is well - known .
we now show the formula of @xmath332 . from the proof of theorem [ thm : expansions ] , we know that @xmath333 , @xmath316 is square integrable near @xmath334 , and @xmath317 is square integrable near @xmath335 .
thus , we have @xmath336 and it gives the desired formula after the substitution by @xmath337 and @xmath338 . in the following we determine the singularities of @xmath155 for fixed @xmath339 .
first we prove an identity of hypergeometric functions in @xmath337 and @xmath338 .
[ lem : connectionformula ] we have the formula @xmath340 for any @xmath341 and @xmath197 . here
the function @xmath342 is defined by @xmath343 it follows from the connection formula @xcite of legendre functions that @xmath344 where @xmath345 and @xmath346 . from @xcite , when @xmath347 the legendre functions can be represented by the hypergeometric functions as @xmath348 in the equation of @xmath349 , it is assumed that @xmath350 . applying the connection formula with @xmath351 , @xmath352 and @xmath353 , we obtain @xmath354 the transformation formula @xmath355 implies that @xmath356 which gives us the desired identity .
when @xmath357 , the first term on the left hand side vanishes as @xmath358 goes to infinity and the identity still holds where the hypergeometric functions reduces to polynomials . for any fixed @xmath359 ,
the green function @xmath360 has the branch cut along @xmath308 and is analytic on the principal branch . since @xmath332 is the green function of the differential operator @xmath157 , it is analytic for non - real @xmath361 .
we discuss the singularities of @xmath332 on the real axis . from the formulas of @xmath362 and @xmath363 given in ( 4.3)-(4.5 ) ,
it has branch cut along @xmath364 . by lemma [ lem : connectionformula ] ,
@xmath332 can be written as @xmath365 since @xmath366 when @xmath367 is an integer and the gamma and hypergeometric functions have only simple poles , the possible poles of @xmath332 are given by @xmath368 for @xmath369 , i.e. , @xmath370 .
note that @xmath371 with @xmath372 , this case does not exist and so @xmath332 is analytic in @xmath373 .
note that the green function of @xmath297 is the same as the one of @xmath374 . from theorem [ thm :
titchmarshgreen ] , for any real constant @xmath375 , the green function of the operator @xmath376 is the given by @xmath377 [ thm : greenintegral ] the green function @xmath296 of the operator @xmath0 and @xmath378 has the following integral form : @xmath379 or @xmath380 for @xmath339 and @xmath381 with @xmath382 , and the integral in ( [ eqn : greenr ] ) takes the principal value . here , for @xmath383 , @xmath384 and @xmath385 for @xmath197 .
we need a lemma that will be used in the proof of theorem [ thm : greenintegral ] .
[ lem : asymptoticgammaf ] we have the following asymptotic expansions : @xmath386 when @xmath387 . here
we show the first formula which involves gamma functions .
since @xmath389 and @xmath390 we have @xmath391 since @xmath392 we have @xmath393 it follows that when @xmath387 we have @xmath394 in the last step we have used the following expansion for large @xmath395 with @xmath396 : @xmath397 next we consider the hypergeometric function . from the transformation formulas we have @xmath398 write @xmath399 with @xmath400
. then we have @xmath401 and @xmath402 for all @xmath403 .
so from ( * ? ? ?
* section 15.12 ) we have the following asymptotic expansion @xmath404 for any fixed @xmath405 .
letting @xmath406 in the formula above gives us the desired expansion .
first we show that formulas ( [ eqn : greenre ] ) and ( [ eqn : greenr ] ) are equivalent .
note that @xmath407 and @xmath408 .
it follows that the real part of @xmath409 is an even function in @xmath410 and its imaginary part is odd in @xmath410 .
so the two integral formulas are equal . in the following we show formula ( [ eqn : greenre ] ) .
we first assume that @xmath411 .
consider the simple closed contour @xmath412 in figure [ fig : contourg ] .
used for the green function @xmath296 ] it consists of the circular arcs @xmath212 and @xmath219 with radius @xmath414 centered at @xmath173 , the vertical line from @xmath415 to @xmath416 cut by the arcs @xmath212 and @xmath219 , the small circle @xmath417 with radius @xmath418 centered at @xmath173 , and the horizontal line segments @xmath419 , @xmath420 . since the green functions @xmath331 and @xmath332 have no singularities in the region bounded by @xmath412 , the cauchy theorem implies that @xmath421 we have @xmath422 on the circle @xmath417 we have @xmath423 with @xmath304 and @xmath328 . as @xmath424 we have @xmath425 and @xmath426 .
it follows that @xmath427 and then @xmath428 so we have @xmath429 for some constant @xmath430 as @xmath424 . for the green function @xmath332 , we have @xmath431 that converges to @xmath432 as @xmath424 .
it follows that when @xmath424 we have the following limits @xmath433 and then @xmath434 where the constant @xmath435 does not depend on @xmath436 .
so we have @xmath437 and the integral along @xmath417 converges to zero as @xmath424 .
this finishes the proof of the claim on @xmath417 .
we have @xmath438 on the arcs @xmath212 and @xmath219 we have @xmath439 and @xmath440 $ ] .
it follows that @xmath441 in the following we consider the asymptotic expansions of various functions as @xmath442 . for the fundamental solutions @xmath443 s , from lemma [ lem : asymptoticgammaf ] we have @xmath444
so we have @xmath445 for some constants @xmath446 . for the exponential functions we have @xmath447 and
@xmath448 we also have @xmath449 the real part is given by @xmath450 on the arc @xmath212 since @xmath451 and @xmath452 we have @xmath453 .
on the arc @xmath219 we have @xmath454 . on both arcs we have @xmath455 for some constant @xmath456 and
so the real part of @xmath457 diverges to infinite and we have the following limit @xmath458 it follows that for large @xmath459 we have @xmath460 for some constant @xmath461 . combining these estimates , we have @xmath462 write @xmath463 with @xmath328 and we have @xmath464 and @xmath465 .
it follows that @xmath466 and then we have @xmath467 when @xmath468 , since @xmath469 $ ] we have @xmath470 .
note that @xmath471 by assumption and @xmath472 , so we have @xmath473 for some constant @xmath474 when @xmath414 is large . on the other hand we have @xmath475 and so we have @xmath476\cup [ 3\pi/2 , 2\pi ] } e^{-k \sqrt r } d\theta\end{aligned}\ ] ] which converges to zero as @xmath442 . when @xmath477 , we have @xmath478 .
let @xmath479 be a constant and we show that @xmath480 on @xmath212 .
since @xmath481 on @xmath212 for some constant @xmath482 , we have @xmath483 which converges to zero as @xmath442 .
a similar argument shows that the integral converges to zero on the arc @xmath219 .
apply this convergence in equation ( [ eqn : g1g2estimate1 ] ) by taking @xmath484 and @xmath485 and then we have @xmath486 this finishes the proof of the claim on @xmath487 .
the identity ( [ eqn : cauchyidentity ] ) implies that @xmath488 let @xmath489 with @xmath490 .
then on @xmath419 and @xmath420 we have @xmath491 on @xmath419 we have @xmath492 and then the green function @xmath332 is given by @xmath493 on @xmath420 we have @xmath494 and the green function @xmath332 is given by @xmath495 so we have @xmath496 which gives formula ( [ eqn : greenre ] ) when @xmath411 .
since the real part of @xmath497 is even in @xmath410 , the kernel of the integral is symmetric in @xmath39 and @xmath498 .
therefore the same formula holds when @xmath499 .
for each fixed @xmath500 , the green s function @xmath501 has the limit zero as @xmath502 diverges to infinity . in this section
we determine the asymptotic expansions along various paths when @xmath502 diverges to infinity , see theorems [ thm : asymgeta0 ] , [ thm : asymgetainfty ] and [ thm : asymgreenrays ] . for the techniques of asymptotic expansion of integrals , we refer to the book @xcite .
first we derive a new formula of the integrand in the green s function .
let @xmath503 and @xmath504 for @xmath505 and @xmath506 .
then the green function in theorem [ thm : greenintegral ] is given by @xmath507 [ lem : ksyeta ] for @xmath505 and @xmath508 we have @xmath509 and @xmath510 from lemma [ lem : connectionformula ] we have @xmath511 by the transformation formula @xmath512 we have @xmath513 which gives us the formula of @xmath514 . for the real part of @xmath514 ,
we have @xmath515 and @xmath516 so they give us the desired formula of @xmath517 .
recall pochhammer s symbol @xmath519 and the psi function ( or digamma function ) @xmath520 [ thm : asymgeta0 ] for any fixed @xmath521 and @xmath522 , the green function @xmath523 has the following expansion near @xmath518 : @xmath524 note that when @xmath525 we have the asymptotic expansion @xmath526 so the improper integral in theorem [ thm : asymgeta0 ] does not converge when @xmath477 . for any @xmath527 with @xmath406
, we recall the following formula for @xmath528 : @xmath529 it follows from the above formula and ( [ eqn : fsy ] ) that @xmath530 here we have also used the identity @xmath531 so we have @xmath532\left[1- \left({\frac{1}{2}}- is\right)(\eta - \eta^2)\right]\right .
\\ & & \left . - \frac{1 + 4s^2}{8}\eta^2\left(\log(1-e^{-2\eta } )
-\psi(1 ) - \psi(3 ) + \psi(3/2 ) + \psi(3/2 + is)\right)\right\ } + o(\eta^3 ) \\ & = & \frac{\gamma(1+is)}{\gamma(3/2)\gamma(3/2 + is)}\left\{1-{\frac{1}{2}}\eta + { \frac{1}{2}}\left(1- i s + s^2\right ) \eta^2 \right . \\ & & \left . - \frac{1 + 4s^2}{8}\eta^2\left(\log(1-e^{-2\eta } ) -\psi(1 ) - \psi(3 ) + \psi(3/2 ) + \psi(3/2 + is)\right)\right\ } + o(\eta^3)\end{aligned}\ ] ] and then @xmath533 where the function in @xmath534 has pure - imaginary value and is given by @xmath535 it follows that @xmath536 which can be further simplified to @xmath537 the expansion formula of @xmath296 follows by combining the above identity with the expansion @xmath538 [ thm : asymgetainfty ] for fixed @xmath540 and fixed @xmath541 , the green function @xmath296 has the following expansion as @xmath542 : @xmath543 we need some preparations and first consider the asymptotic expansion of the following integral as @xmath542 , @xmath544 where @xmath545 .
we may assume @xmath546 is bounded below , say @xmath547 .
for fixed @xmath541 , define the following functions on the complex plane : @xmath548 and @xmath549 following the proof of lemma [ lem : ksyeta ] , the function @xmath550 can be simplified as @xmath551 we specify the branch of @xmath552 we choose : write @xmath553 then we take @xmath554 and it has the branch cut @xmath555 \cup [ { \frac{1}{2 } } , \infty)$ ] . in particular , when @xmath556 with @xmath557 , we have @xmath558 and @xmath559 it follows that @xmath560 and then @xmath561 [ prop : ietarealintegral ] the integral @xmath562 has the following form @xmath563 we consider the contour integral in figure [ fig : contourieta ] .
the arc @xmath417 has the center @xmath564 with radius @xmath565 and @xmath566 have the center @xmath432 with radius @xmath567 .
since @xmath568 is analytic in the region bounded by the closed contour @xmath412 , the cauchy theorem implies that @xmath569 used in the integral @xmath562 . ] in the following we show that the integral @xmath570 vanishes on both @xmath417 and @xmath571 as @xmath424 and @xmath442 respectively . a similar argument as in the proof of theorem [ thm : greenintegral ] shows that @xmath572 we have @xmath573 on the arcs @xmath487 we have @xmath574 with @xmath575 $ ] on @xmath212 and @xmath576 on @xmath219 , see figure [ fig : thetas ] .
recall @xmath577 with @xmath578 and @xmath579 with @xmath580 .
on @xmath212 we have @xmath581 and @xmath582 where @xmath583 is a small number .
on @xmath219 we have @xmath584 and @xmath585 .
so we have @xmath586 and @xmath587 . ]
next we consider the asymptotic expansion of @xmath550 for large @xmath459 . from the formula of @xmath550 in ( [ eqn : q1f ] ) and the asymptotic expansion of hypergeometric function in ( * ? ? ?
* section 15.12 ) we have @xmath588 where @xmath589 is the modified bessel function and @xmath590 on the arc @xmath212 , we have @xmath591 $ ] and so @xmath592 since @xmath593 it follows that @xmath594 for some constant @xmath595 when @xmath414 is large . note that @xmath596 when @xmath414 is large , so we have @xmath597 note that @xmath598 $ ] and we have assumed that @xmath599 .
so we have @xmath600 and hence @xmath601 on the arc @xmath219 we have different signs in the formulas of @xmath602 and @xmath603 . the argument above carries through and the limit on @xmath219 also vanishes .
this finishes the proof of the claim .
thus , ( [ eqn : ietap ] ) and ( [ eqn : ointpeetaz ] ) imply that @xmath604 note that when @xmath605 we have @xmath606 it follows that @xmath607 this finish the proof of proposition [ prop : ietarealintegral ] .
[ prop : asymietainfty ] for fixed @xmath608 and @xmath609 large we have the asymptotic formula @xmath610 as @xmath542 from proposition [ prop : ietarealintegral ] and the formula of @xmath550 in ( [ eqn : q1f ] ) we have @xmath611 at @xmath612 , we have the following taylor series @xmath613 to apply watson s lemma , see for example @xcite , we only need to show that there exist constants @xmath614 and @xmath615 independent of @xmath261 such that @xmath616 since @xmath617 , we have @xmath618 and the asymptotic expansion of the modified bessel function has a simpler form , i.e. , the @xmath619 term does not appear .
so for large @xmath261 we have @xmath620 for example , we can take @xmath621 and @xmath622 .
now watson s lemma yields @xmath623 which gives the desired asymptotic formula of @xmath562 . from the asymptotic expansion of @xmath562 in proposition [ prop : asymietainfty ]
we show theorem [ thm : asymgetainfty ] . since @xmath624 and @xmath625 as @xmath542
, we have @xmath626 using the asymptotic expansion of @xmath562 in proposition [ prop : asymietainfty ] , we have @xmath627 [ thm : asymgreenrays ] for each @xmath629 , the green function has the following asymptotic expansion along the ray @xmath628 with @xmath630 : @xmath631 we first assume that @xmath541 so that @xmath632 for large @xmath633 . from lemma [ lem : ksyeta ] ,
@xmath634 we consider the real part of the following integral @xmath635 let @xmath636 and @xmath637 for @xmath341 .
so the integral can be written as @xmath638 where @xmath639 denotes the positive real axis .
for @xmath640 , we have the asymptotic expansion @xmath641 we show the above asymptotic expansion by the method of the steepest descent , see for example ( * ? ? ?
write @xmath642 and let @xmath643 with @xmath644 let @xmath219 be half of the hyperbola defined by the equation @xmath645 denote @xmath212 the line segment from @xmath646 to @xmath647 on the @xmath648-axis and @xmath649 the circular arc centered at the origin with radius @xmath414 connecting @xmath219 and the positive @xmath650-axis , see figure [ fig : contourhyperbola ] . ] since @xmath651 is analytic in the region bounded by @xmath650-axis , @xmath212 , @xmath219 and @xmath649 , the cauchy theorem implies that @xmath652 it can be shown that for any fixed @xmath541 the integral along @xmath649 vanishes as @xmath442 .
so we have @xmath653 the integral along @xmath212 can be written as @xmath654 note that @xmath655 is a real - valued function and @xmath656 is also a real - valued function since @xmath657 .
so we have @xmath658 for any @xmath541 .
next we consider the integral along @xmath219 .
we have @xmath659 and @xmath660 from the formula of the principal square root of a complex number @xmath661 we have @xmath662 and @xmath663 along the curve @xmath219 , let @xmath664 and we have @xmath665 .
around @xmath666 we have the following expansion @xmath667 with @xmath668 it follows that @xmath669 and so we have the following asymptotic expansion as @xmath542 : @xmath670 this finishes the proof of the claim . the asymptotic expansion of the green function follows from the one of @xmath671 for @xmath541 .
when @xmath672 we have a similar expansion , except the factor @xmath673 is replaced by @xmath674 so we have finished the proof of theorem [ thm : asymgreenrays ] .
in this section , we first determine the martin compactification @xmath52 of @xmath80 with respect to the operator @xmath0 , see theorem [ thm : martincompsigma ] . then we prove theorem [ thm : uniquenesswpos ] in the introduction at the end of the section . fix a reference point @xmath675 ,
then the martin kernel is given by @xmath676 from the asymptotic expansion of the green function along various paths in the previous section we can determine martin kernel functions of such cases .
for any @xmath677 , denote @xmath678 on the boundary @xmath679 , and @xmath680 while fixing @xmath681 .
recall the short notation of gauss hypergeometric function @xmath682 using theorems [ thm : asymgeta0 ] , [ thm : asymgetainfty ] and [ thm : asymgreenrays ] , we immediately obtain [ prop : martinkernel ] the limits of the martin kernels as @xmath502 diverges to infinity are given by the following cases .
a. for any @xmath677 , when @xmath683 the martin kernel is given by @xmath684 with @xmath685 .
b. when @xmath686 is fixed and @xmath542 , the martin kernel is given by @xmath687 where @xmath688 is a constant such that @xmath689 . c. along the ray @xmath690 with @xmath691 , when @xmath692 the martin kernel is given by @xmath693 where @xmath694 is a constant such that @xmath695 .
the function @xmath696 is the unique ( non - negative ) solution , up to a constant multiple , to the differential equation @xmath697 with @xmath698 $ ] .
it grows like @xmath699 as @xmath40 if @xmath700 or @xmath701 .
when @xmath702 and @xmath701 , the function @xmath703 grows like @xmath704 as @xmath40 .
[ lem : kxixomegalim ] for any fixed @xmath705 , we have the following limits .
@xmath706 the first identity follows from the defining equations of @xmath707 and @xmath708 in proposition [ prop : martinkernel ] .
next we show the second identity .
the third one follows by a similar argument .
we assume that @xmath709 and denote @xmath710 .
let @xmath711 with @xmath712 the martin kernel can be written as @xmath713 denote by @xmath714 since for any @xmath677 , @xmath715 is a positive @xmath34-harmonic function on @xmath1 and has value @xmath110 at @xmath716 , so is @xmath717 . in the following we consider the asymptotic expansion of @xmath718 as @xmath719 .
let @xmath720 and then @xmath721 .
the integral @xmath718 can be written as @xmath722 suppose that @xmath723 has the following expansion at @xmath724 : @xmath725 with @xmath726 for @xmath727 , then we have @xmath728 and the following asymptotic expansion as @xmath729 from watson s lemma @xmath730 it follows that @xmath731 and hence @xmath732 . it can also be shown explicitly that the function @xmath733 solves the differential equation @xmath734 recall the following definition of martin compactification that is equivalent to the one with minimal martin boundary in section 2 , see ( * ? ? ?
* definition 6.2 ) . here
we exclude the case where all positive solutions to @xmath735 are proportional .
note that a topological space is said to be @xmath9-compact if it is the union of countably many compact subspaces .
this property holds for any riemannian manifold . in the notions of martin compactification and martin boundary
, we drop the letter @xmath12 .
[ defn : martincomptaylor ] let @xmath66 be a complete smooth manifold that is @xmath9-compact and @xmath12 be a second order strictly elliptic partial differential operator with smooth coefficients for which @xmath736 . the martin compactification is the compactification @xmath737 such that 1 . the normalized green functions ,
i.e. , the martin kernel functions @xmath738 , extend continuously to @xmath737 for each @xmath739 ; and 2 .
the extended functions separate the points of the ideal boundary @xmath740 .
the ideal boundary of the compactification satisfying condition ( 2 ) in definition [ defn : martincomptaylor ] is the _ minimal martin boundary _ @xmath134 , see ( * ? ? ?
* remark i.7.7 ) .
[ thm : martincompsigma ] the martin compactification of @xmath1 is homeomorphic to the half disk @xmath741 and @xmath742 is the minimal martin boundary .
moreover we have 1 . the real line @xmath743 is identified with the interval @xmath744 of @xmath5 with @xmath745 , 2 .
the vertical half line @xmath746 is identified with the point @xmath41 , 3 .
the asymptotic ray @xmath747 is identified with the semi - circle @xmath748 4 . for any @xmath749 we have @xmath750 denote @xmath751 .
first note that @xmath297 is coercive , see @xcite . since any @xmath752(@xmath677 ) can be approached by the geodesic @xmath753 with @xmath754 $ ] and the curvature of the subset @xmath755 \subset \sigma$ ] is pinched by two negative constants , it follows from ( * ? ? ?
* corollary 16 ) that @xmath752 is a minimal point on the martin boundary and the limit in equation ( [ eqn : kxixomegaylim ] ) for @xmath756 . in the martin compactification
each @xmath752 admits a basis of neighborhoods consisting of geodesic cones that agree with the basis of neighborhood in the topology of @xmath51 .
it follows that the collection of all @xmath752 that is @xmath757 embeds into the martin compactification .
the previous lemma [ lem : kxixomegalim ] shows that martin kernel functions @xmath758 extend continuously to @xmath52 for any @xmath705 , and it is clear to see that extended functions separate the points of @xmath52 . from definition [ defn : martincomptaylor ] @xmath52 is the marin compactification and @xmath759 is the minimal martin boundary . in the following we show equation ( [ eqn : kxixomegaylim ] ) for other points
if @xmath760 is an asymptotic ray @xmath42 with @xmath761 , then we have @xmath40 as @xmath201 approaches @xmath43 .
note that @xmath762 as @xmath40 .
the rest part of the integral with @xmath39 in the numerator of @xmath715 is the fourier coefficient of the following @xmath763-function in @xmath410 : @xmath764
so it converges to zero as @xmath40 and thus we have @xmath765 next we consider the case when @xmath174 and @xmath49 .
we assume that @xmath685 when @xmath766 .
we have the following taylor expansions : @xmath767 and @xmath768 it follows that @xmath769 and hence @xmath715 converges to zero as @xmath770 .
this finishes the proof of theorem [ thm : martincompsigma ] .
it follows from theorem [ thm : geometriccompactification ] that the martin compactification @xmath52 is homeomorphic to the geometric compactification @xmath771 .
the martin kernel function is given by @xmath772 and @xmath773 along the geodesics asymptotic to @xmath774 ( @xmath640 ) and @xmath775 ( @xmath776 ) respectively . from equation ( [ eqn : kxixomegaylim ] ) we know that @xmath777 if the limit exists . otherwise , @xmath715 is the trivial function by the maximal principle which contradicts the fact that @xmath778 .
we finish this section by showing corollary [ cor : dirichlet ] .
we argue by contradiction .
assume that the dirichlet problem at infinity is solvable for the laplace operator @xmath2 .
let @xmath779 be a continuous function on @xmath285 such that @xmath780 on the part @xmath781 and @xmath782 on the part @xmath783 .
then there is a harmonic function @xmath254 on @xmath1 such that the following properties hold : 1 . @xmath784 as @xmath39 approaches @xmath432
; 2 . @xmath785 as @xmath786 approaches infinity along the type ( ii ) geodesic , i.e. , @xmath76 fixed and @xmath40 ; 3 . @xmath787 as @xmath786 approaches infinity along the type ( iv ) geodesic .
it follows that @xmath254 is bounded and positive on @xmath1 by the maximum principle . since @xmath4 is conformal to the half - plane with the hyperbolic metric @xmath788 , @xmath254 is also a positive harmonic function with respect to the metric @xmath788 . since @xmath254 vanishes on the boundary @xmath61 ,
it is a positive constant multiple of @xmath789 , the martin kernel function at @xmath790 , see for example ( * ? ? ?
* remark 4.1 ) .
this contradicts the previous conclusion that @xmath254 is bounded .
[ rem : dirichletuniqueness ] from the proof of corollary [ cor : dirichlet ] , it is easy to see that the dirichlet problem at infinity for the laplace operator @xmath2 has a unique solution only when @xmath791 is constant on the semi - circle @xmath792 .
we assume that @xmath21 is not the trivial solution so that @xmath21 is positive on @xmath80 . from theorem [ thm : martinintegralrep ] , there is a unique borel measure @xmath81 on @xmath36 with @xmath793 such that @xmath794 by proposition [ prop : martinkernel ] and lemma [ lem : kxixomegalim ] , the proof now follows from the following the measure @xmath81 is supported by the one - point set @xmath795 , i.e. , @xmath796 we show the claim in two steps : first @xmath797 by the boundary condition @xmath798 for all @xmath346 , and then @xmath81 is concentrated at @xmath799 by the inequality ( [ eqn : wgrowth ] ) .
for the first step , we follow a similar argument in @xcite .
it is sufficient to show that if a finite borel measure @xmath81 on @xmath25 has @xmath800 , then @xmath801 can not vanish on the @xmath76-axis . when @xmath802 , there is a number @xmath803 such that the @xmath81-measure of the interval @xmath804 $ ] is positive .
let @xmath805 then @xmath806 is non - negative and @xmath34-harmonic on @xmath1 .
it is majorized by @xmath807 , i.e. , @xmath808 on @xmath1 . since @xmath809)>0 $ ]
, @xmath806 is positive on @xmath1 . from theorem [ thm : martincompsigma ]
we have @xmath810 as @xmath811 .
let @xmath812 be a sequence of points such that @xmath813 as @xmath814 . for fixed @xmath815 since @xmath816 is continuous in @xmath633 , it achieves the maximal value on @xmath804 $ ] .
let @xmath817 $ ] with @xmath818}k(x_j , y_j,\omega_{\xi})\ ] ] and so we have @xmath819 it follows from the compactness of @xmath804 $ ] that the limit of @xmath820 is zero as @xmath821 .
so we have @xmath822 as @xmath814 , i.e. , @xmath823 as @xmath811 .
it follows that @xmath824 can not vanish on the @xmath76-axis .
otherwise we would have @xmath825 everywhere that contradicts the fact @xmath826 .
so the function @xmath827 can not vanish on the @xmath76-axis .
therefore , the borel measure @xmath81 in the claim must have @xmath797 because @xmath798 for all @xmath346
. now we proceed to the second step .
since @xmath828 , the integral formula ( [ eqn : wintegral ] ) can be written as @xmath829 inequality ( [ eqn : wgrowth ] ) is equivalent to @xmath830 let @xmath831 then , we have @xmath832 it follows that @xmath833 we may assume that @xmath834 , otherwise @xmath835 defines another finite non - negative borel measure @xmath836 on the semi - circle and the argument below also holds using the measure @xmath836 . from the asymptotic expansion of @xmath696 in remark 6.2 , we have @xmath837
also , for @xmath838 , @xmath839 assume that @xmath81 is not concentrated at @xmath840 .
then there exists some @xmath841 such that @xmath842 with @xmath843 $ ] . from the asymptotic behavior of @xmath844
, there exist @xmath845 and @xmath846 such that when @xmath847 we have @xmath848 - i_{\varepsilon}.\end{aligned}\ ] ] the integral formula ( [ eqn : dwintegral ] ) yields @xmath849 when @xmath847 .
this contradicts inequality ( [ eqn : wgrowthde ] ) for @xmath850 and we finish the proof .
in this section we consider the sturm - liouville problem of the operator @xmath851 on @xmath276 , i.e. , the spectrum of this operator .
it is related to the green s function @xmath155 of the operator @xmath852 in section 4 . for the basics of sturm - liouville problem ,
see for example @xcite and @xcite and the references therein .
recall the positive function @xmath853 and consider the following 2nd order differential operator @xmath854 with the domain @xmath855 here @xmath856 stands for the set of all locally absolutely continuous functions on @xmath276 .
equation @xmath857 has the following two linearly independent solutions @xmath858 since the second function is not square - integrable near @xmath859 and @xmath860 , we have limit - point case for @xmath861 and @xmath862 . in particular , the operator @xmath863 with domain @xmath864 is self - adjoint . the differential equation @xmath865 for any @xmath310 is solved in appendix b and the two linearly independent solutions are denoted by @xmath866 and @xmath867 , see proposition [ prop : wfdmsolutions ] .
note that a number @xmath310 is in the discrete spectrum @xmath868 of @xmath863 if the equation @xmath865 has a solution @xmath869 .
[ thm : adiscretespectrum ] the discrete spectrum @xmath868 of @xmath863 is empty .
since @xmath863 is self - adjoint , the spectrum @xmath870 is in @xmath25 .
we separate our discussion into different intervals of @xmath25 .
when @xmath871 , the two linearly independent solutions of @xmath857 are not square integrable near @xmath872 . in the following
we assume that @xmath873 .
* case 1 * : @xmath874 .
write @xmath489 with @xmath875
. then the linearly independent solutions of @xmath865 are given by @xmath876 and @xmath877 , the complex conjugate .
since @xmath878 we have the following asymptotic expansion as @xmath46 : @xmath879 and so no linear combination of @xmath866 and @xmath867 is square - integrable near @xmath862 .
* case 2 * : @xmath880 with @xmath881 not an integer and @xmath882 .
when @xmath367 is not a half integer , neither @xmath866 nor @xmath867 in proposition [ prop : wfdmsolutions ] is square integrable near @xmath872 .
if @xmath367 is a half integer , then @xmath867 is given by the jacobi polynomial .
the solution @xmath866 is not square integrable near @xmath859 . when @xmath46 , we also have @xmath883 which is not square integrable . * case 3 * : @xmath884 with @xmath885 .
the first solution @xmath886 is not square integrable near @xmath859 . the second solution is given by @xmath887 with @xmath888 .
as @xmath46 , we have @xmath867 approaches @xmath889 which is unbounded by the asymptotic expansion @xmath890 at @xmath891 .
this finishes all the cases and proves theorem [ thm : adiscretespectrum ] . for a complex number @xmath892 with @xmath893 and @xmath894 $ ] , the principal square root is given by @xmath895 .
recall the following functions in ( [ eqn : as ] ) and ( [ eqn : wsy ] ) of theorem [ thm : greenintegral ] : @xmath896 and @xmath897 for @xmath898 .
[ thm : expansions ] for any function @xmath899 we have @xmath900 where the kernel is given by @xmath901 suppose @xmath373 is a complex number on the upper half plane , i.e. , @xmath902 .
denote @xmath903 the principal square root of @xmath904 , see also equation ( [ eqn : alphaphi ] ) .
since @xmath367 is not a real number , by proposition [ prop : wfdmsolutions ] the two linearly independent solutions are given by @xmath905 the wronskian of @xmath337 and @xmath338 is given by @xmath906 let @xmath907 with @xmath908 then @xmath316 is square integrable near @xmath859 . when @xmath76 approaches @xmath909 we have the following expansions , @xmath910 write @xmath911 .
we have @xmath912 and then @xmath913 . it follows that @xmath914 and @xmath915 .
so the real part @xmath916 .
it follows that any square integrable solution near @xmath917 is a multiple of @xmath317 .
we follow the construction in ( * ? ? ?
* section 12 ) ( see also @xcite ) .
take a regular point @xmath918 and consider the fundamental system @xmath919 and @xmath920 with the following boundary conditions : @xmath921
so the matrix @xmath922 is given by @xmath923 and @xmath924 the characteristic matrix @xmath925 and the matrix function @xmath926 are defined for @xmath148 as in ( * ? ? ?
* section 12 ) . if @xmath927 , then @xmath928 is a real number .
so @xmath929 and then @xmath930 are real matrices and thus @xmath931 ( @xmath932 ) . note that the discrete spectrum @xmath868 is empty by theorem [ thm : adiscretespectrum ] .
let @xmath933 then we have @xmath934 and @xmath935 for the fixed @xmath918 , let @xmath936 since @xmath937 , we have @xmath938 denote @xmath939 by @xmath940
. then we have @xmath941 and @xmath942 so the fundamental system is given by @xmath943 the characteristic matrix @xmath944 is given by @xmath945 and then @xmath946 note that @xmath947 , so we have @xmath948 it follows that @xmath949 a straightforward computation shows that the product of the five @xmath950-matrices above equals to @xmath951 so we have @xmath952 now the eigenfunction expansion follows from @xcite or ( * ? ? ?
* theorem 1.13 ) .
in this section we solve equation @xmath865 given in ( [ eqn : adefn ] ) using special functions .
first , let us recall the following special functions : * the hypergeometric function @xmath953 solves the hypergeometric differential equation @xmath954 w'(z ) - ab w(z ) = 0\ ] ] with constants @xmath955 . * the associated legendre functions @xmath956 and
@xmath957 solve the equation @xmath958 w(z ) = 0\ ] ] with constants @xmath345 .
* the jacobi polynomial @xmath959 is a solution of the equation @xmath960 with positive integer @xmath91 and constants @xmath961 .
[ prop : wfdmsolutions ] let @xmath310 .
the general solution to the equation @xmath962 is given by @xmath963 , where @xmath964 are constants , and @xmath337 and @xmath338 are defined for @xmath965 as follows : \(c ) in case ( 1 ) , two linearly fundamental solutions can also be given by associated legendre functions @xmath976 and @xmath977 .
note that in this case @xmath977 is a linear combination of @xmath866 and @xmath867 given in the proposition .
we start with the first case as @xmath880 .
let @xmath978 then equation ( [ eqn : wdd ] ) can be written as @xmath979 that is the hypergeometric de in @xmath980 with @xmath981 , @xmath982 and @xmath983 .
when @xmath367 is not an integer , it has the fundamental solutions @xmath984 with @xmath985 $ ] . since @xmath986 , @xmath987 and @xmath988 are absolutely convergent at @xmath989 , i.e. , @xmath872 . by the transformation formula ( [ eqn : transhy ] ) , we have the fundamental solutions @xmath337 and @xmath338 in case ( 1 ) . when @xmath990 with @xmath991 , let @xmath992 then equation ( [ eqn : wdd ] ) is equivalent to the following legendre differential equation , @xmath993 its fundamental solutions are given by associated legendre functions @xmath994 and @xmath995 with the wronskian @xmath996 so we have @xmath997 which agrees with the first solution in the previous case .
the function @xmath998 can also be represented by hypergeometric function as @xmath999 since @xmath1000 as @xmath1001 , we have @xmath888 .
w. ballmann , _ the martin boundary of certain hadamard manifolds _ , proceedings on analysis and geometry ( russian ) ( novosibirsk akademgorodok , 1999 ) , 3646 , izdat . ross .
inst . mat . ,
novosibirsk , 2000 .
l. a. caffarelli and w. littman , _ representation formulas for solutions to @xmath1004 in @xmath10 _ , studies in partial differential equations , 249263 , maa stud .
, * 23 * , math .
america , washington , d.c . , 1982 .
hua , _ on the theory of functions of several complex variables .
i. a complete orthonormal system in the hyperbolic space of matrices _ ( chinese .
english summary ) , j. chinese math .
soc . , * 2 * ( 1953 ) , 288323 .
hua , _ on the theory of functions of several complex variables .
ii . a complex orthonormal system in the hyperbolic space of lie - hypersphere _ ( chinese .
english summary ) , acta math .
sinica * 5 * ( 1955 ) , 125 .
hua , _ on the theory of functions of several complex variables .
iii . on a complete orthonormal system in the hyperbolic space of symmetric and skew - symmetric matrices _
( chinese . russian summary ) , acta math .
sinica * 5 * ( 1955 ) , 205-242 .
j. c. taylor , _ the martin compactification associated with a second order strictly elliptic partial differential operator on a manifold @xmath66 _ , topics in probability and lie groups : boundary theory , 153202 , crm proc .
lecture notes , * 28 * , amer .
soc . , providence , ri , 2001 .
r. wong , _
asymptotic approximations of integrals _ , corrected reprint of the 1989 original .
classics in applied mathematics , * 34 * society for industrial and applied mathematics ( siam ) , philadelphia , pa , 2001 . | in this paper we consider the martin compactification , associated with the operator @xmath0 , of a complete non - compact surface @xmath1 with negative curvature .
in particular , we investigate positive eigenfunctions with eigenvalue one of the laplace operator @xmath2 of @xmath1 and prove a uniqueness result : such eigenfunctions are unique up to a positive constant multiple if they vanish on the part of the geometric boundary @xmath3 of @xmath1 where the curvature is bounded above by a negative constant , and satisfy some growth estimate on the other part of @xmath3 where the curvature approaches zero .
this uniqueness result plays an essential role in our recent paper @xcite in which we prove an infinitesimal rigidity theorem for deformations of certain three - dimensional collapsed gradient steady ricci soliton with a non - trivial killing vector field . |
reliable and accurate prediction of precipitation is of great importance in agriculture , tourism , aviation and in some other fields of economy as well . in order to represent the uncertainties of forecasts based on observational data and numerical weather prediction ( nwp ) models one can run these models with different initial conditions or change model physics , resulting in a forecast ensemble @xcite . in the last two decades
this approach has became a routinely used technique all over the world and recently all major weather prediction centres have their own operational ensemble prediction systems ( eps ) , e.g. the consortium for small - scale modelling ( cosmo - de ) eps of the german meteorological service ( dwd ; * ? ? ?
* ; * ? ? ?
* ) , the prvision densemble arpege ( pearp ) eps of mteo france @xcite or the eps of the independent intergovernmental european centre for medium - range weather forecasts @xcite . with the help of a forecast ensemble one can estimate the distribution of the predictable weather quantity which opens up the door for probabilistic forecasting @xcite . by post - processing the raw ensemble the most sophisticated probabilistic methods result in full predictive cumulative distribution functions ( cdf ) and correct the possible bias and underdispersion of the original forecasts .
the underdispersive character of the ensemble has been observed with several ensemble prediction systems @xcite and this property also leads to the lack of calibration . using predictive cdfs one can easily get consistent estimators of probabilities of various meteorological events or calculate different prediction intervals .
recently , probably the most widely used ensemble post - processing methods leading to full predictive distributions ( for an overview see e.g. * ? ? ?
* ; * ? ? ?
* ) are the bayesian model averaging ( bma ; * ? ? ?
* ) and the non - homogeneous regression or ensemble model output statistics ( emos ; * ? ? ?
* ) , as they are partially implemented in the ensemblebma and ensemblemos packages of r @xcite .
the bma predictive probability density function ( pdf ) of the future weather quantity is the mixture of individual pdfs corresponding to the ensemble members with mixture weights determined by the relative performance of the ensemble members during a given training period . to model temperature or sea level pressure
a normal mixture seems to be appropriate @xcite , wind speed requires non - negative and skewed component pdfs such as gamma @xcite or truncated normal @xcite distributions , whereas for surface wind direction a von mises distribution @xcite is suggested . however , in some situations bma post - processing might result , for instance , in model overfitting @xcite or over - weighting climatology @xcite .
in contrast to bma , the emos technique uses a single parametric pdf with parameters depending on the ensemble members .
again , for temperature and sea level pressure the emos predictive pdf is normal @xcite , whereas for wind speed truncated normal @xcite , generalized extreme value ( gev ; * ? ? ?
* ) , censored logistic @xcite , truncated logistic , gamma @xcite and log - normal @xcite distributions are suggested .
however , statistical calibration of ensemble forecasts of precipitation is far more difficult than the post - processing of the above quantities . as pointed out by @xcite , precipitation has a discrete - continuous nature with a positive probability of being zero and larger expected precipitation amount results in larger forecast uncertainty .
@xcite introduced a bma model where each individual predictive pdf consists of a discrete component at zero and a gamma distribution modelling the case of positive precipitation amounts .
@xcite uses extends logistic regression to provide full probability distribution forecasts , whereas @xcite suggests an emos model based on a censored gev distribution .
finally , @xcite propose a more complex three step approach where they first fit a censored and shifted gamma ( csg ) distribution model to the climatological distribution of observations , then after adjusting the forecasts to match this climatology derive three ensemble statistics , and with the help of a nonhomogeneus regression model connect these statistics to the csg model .
based on the idea of @xcite we introduce a new emos approach which directly models the distribution of precipitation accumulation with a censored and shifted gamma predictive pdf .
the novel emos approach is applied to 24 hour precipitation accumulation forecasts of the eight - member university of washington mesoscale ensemble ( uwme ; * ? ? ?
* ) and the 11 member operational eps of the hungarian meteorological service ( hms ) called aire limite adaptation dynamique dvelopment international - hungary eps ( aladin - huneps ; * ? ? ?
* ; * ? ? ?
* ) . in these case
studies the performance of the proposed emos model is compared to the forecast skills of the gev emos method of @xcite and to the gamma bma approach of @xcite serving as benchmark models .
as mentioned in the introduction , the emos predictive pdf of a future weather quantity is a single parametric distribution with parameters depending on the ensemble members . due to the special discrete - continuous nature of precipitation
one should think only of non - negative predictive distributions assigning positive mass to the event of zero precipitation . mixing a point mass at zero and a separate non - negative distribution
does the job ( see e.g. the bma model of * ? ? ?
* ) , but left censoring of an appropriate continuous distribution at zero can also be a reasonable choice .
the advantage of the latter approach is that the probability of zero precipitation can directly be derived from the corresponding original ( uncensored ) cumulative distribution function ( cdf ) , so the cases of zero and positive precipitation can be treated together .
the emos model of @xcite utilizes a censored gev distribution with shape parameter ensuring a positive skew and finite mean , whereas our emos approach is based on a csg distribution appearing in the more complex model of @xcite .
consider a gamma distribution @xmath0 with shape @xmath1 and scale @xmath2 having pdf @xmath3 where @xmath4 denotes value of the gamma function at @xmath5 .
a gamma distribution can also be parametrized by its mean @xmath6 and standard deviation @xmath7 using expressions @xmath8 now , let @xmath9 and denote by @xmath10 the cdf of the @xmath0 distribution .
then the shifted gamma distribution left censored at zero ( csg ) @xmath11 with shape @xmath5 , scale @xmath12 and shift @xmath13 can be defined with cdf @xmath14 this distribution assigns mass @xmath15 to the origin and has generalized pdf @xmath16 where @xmath17 denotes the indicator function of the set @xmath18 .
short calculation shows that the mean @xmath19 of @xmath11 equals @xmath20 whereas the @xmath21-quantile @xmath22 ( @xmath23 ) of equals @xmath24 if @xmath25 , and the solution of @xmath26 , otherwise .
now , denote by @xmath27 the ensemble of distinguishable forecasts of precipitation accumulation for a given location and time .
this means that each ensemble member can be identified and tracked , which holds for example for the uwme ( see section [ subs : subs3.1 ] ) or for the cosmo - de eps of the dwd . in the proposed csg emos model the ensemble members are linked to the mean @xmath28 and variance @xmath29 of the underlying gamma distribution via equations @xmath30 where @xmath31 denotes the ensemble mean .
mean parameters @xmath32 and variance parameters @xmath33 of model can be estimated from the training data , consisting of ensemble members and verifying observations from the preceding @xmath34 days , by optimizing an appropriate verification score ( see section [ subs : subs2.2 ] )
. however , most of the currently used epss produce ensembles containing groups of statistically indistinguishable ensemble members which are obtained with the help of random perturbations of the initial conditions .
this is the case for the aladin - huneps ensemble described in section [ subs : subs3.2 ] or for the 51 member ecmwf ensemble .
the existence of several exchangeable groups is also a natural property of some multi - model epss such as the the thorpex interactive grand global ensemble @xcite or the glameps ensemble @xcite .
suppose we have @xmath35 ensemble members divided into @xmath36 exchangeable groups , where the @xmath5th group contains @xmath37 ensemble members , such that @xmath38 .
further , we denote by @xmath39 the @xmath40th member of the @xmath5th group . in this situation ensemble members within a given group
should share the same parameters @xcite resulting in the exchangeable version @xmath41 of model .
note , that the expression of the mean ( or location ) as an affine function of the ensemble is general in emos post - processing ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , whereas the dependence of the variance parameter on the ensemble mean is similar to the expression of the variance in the gamma bma model of @xcite , and it is in line with the relation of forecast uncertainty to the expected precipitation amount mentioned in the introduction .
moreover , practical tests show that , at least for the uwme and aladin - huneps ensemble considered in the case studies of section [ sec : sec5 ] , models and , respectively , significantly outperform the corresponding csg emos models with variance parameters @xmath42 where @xmath43 are the ensemble variance and
the more robust ensemble mean difference @xcite , respectively .
further , compared to the proposed models , natural modifications @xmath44 in the csg emos variance structure do not result in improved forecasts skills .
the main aim of probabilistic forecasting is to access the maximal sharpness of the predictive distribution subject to calibration @xcite .
the latter means a statistical consistency between the predictive distributions and the validating observations , whereas the former refers to the concentration of the predictive distribution .
this goal can be addressed with the help of scoring rules which measure the predictive performance by numerical values assigned to pairs of probabilistic forecasts and observations @xcite . in atmospheric sciences the most popular scoring rules for evaluating predictive distributions are the logarithmic score , i.e. the negative logarithm of the predictive pdf evaluated at the verifying observation @xcite , and the continuous ranked probability score ( crps ; * ? ? ?
* ; * ? ? ?
for a predictive cdf @xmath45 and an observation @xmath46 the crps is defined as @xmath47 where @xmath48 denotes the indicator of a set @xmath49 , while @xmath50 and @xmath51 are independent random variables with cdf @xmath52 and finite first moment .
the crps can be expressed in the same units as the observation and one should also note that both scoring rules are proper @xcite and negatively oriented , that is the smaller the better .
for a csg distribution defined by the crps can be expressed in a closed form , @xcite showed that @xmath53 following the ideas of @xcite and @xcite , the parameters of models ( and as well ) are estimated by minimizing the mean crps of predictive distributions and validating observations corresponding to forecast cases of the training period .
we remark that optimization with respect to the mean logarithmic score , that is , maximum likelihood ( ml ) estimation of parameters , has also been investigated .
obviously , in terms of crps this model can not outperform the one fit via crps minimization , however , in our test cases the ml method results in a reduction of the predictive skill of the csg emos model in terms of almost all verification scores considered , so the corresponding values are not reported .
the eight - member uwme covers the pacific northwest region of north america and operates on a 12 km grid .
the ensemble members are obtained from different runs of the fifth generation pennsylvania state university national center for atmospheric research mesoscale model ( psu - ncar mm5 ; * ? ? ?
* ) with initial and boundary conditions from various weather centres .
we consider 48 h forecasts and corresponding validating observations of 24 h precipitation accumulation for 152 stations in the automated surface observing network @xcite in five us states .
the forecasts are initialized at 0 utc ( 5 pm local time when daylight saving time ( dst ) is in use and 4 pm otherwise ) and we investigate data for calendar year 2008 with additional forecasts and observations from the last three months of 2007 used for parameter estimation . after removing days and locations with missing data
83 stations remain resulting in 20522 forecast cases for 2008 . to 9 cma ) b ) figure [ fig : fig1]a shows the verification rank histogram of the raw ensemble , that is the histogram of ranks of validating observations with respect to the corresponding ensemble forecasts computed for all forecast cases ( see e.g. * ? ? ?
* section 7.7.2 ) , where zero observations are randomized among all zero forecasts .
this histogram is far from the desired uniform distribution as in many cases the ensemble members overestimate the validating observation .
the ensemble range contains the observed precipitation accumulation in @xmath54 of the cases , whereas the nominal coverage of the ensemble equals @xmath55 , i.e @xmath56 .
hence , the uwme is uncalibrated , and would require statistical post - processing to yield an improved forecast probability density function .
the ensemble forecasts produced by the operational aladin - huneps system of the hms are obtained with dynamical downscaling of the global pearp system of mto france by the aladin limited area model with an 8 km horizontal resolution .
the eps covers a large part of continental europe and has 11 ensemble members , 10 exchangeable forecasts from perturbed initial conditions and one control member from the unperturbed analysis @xcite .
the data base at hand contains ensembles of 42 h forecasts ( initialized at 18 utc , i.e. 8 pm local time when dst operates and 7 pm otherwise ) for 24 h precipitation accumulation for 10 major cities in hungary ( miskolc , sopron , szombathely , gyr , budapest , debrecen , nyregyhza , nagykanizsa , pcs , szeged ) together with the corresponding validating observations for the period between 1 october 2010 and 25 march 2011 .
the data set is fairly complete since there are only two dates when three ensemble members are missing for all sites .
these dates are excluded from the analysis .
the verification rank histogram of the raw ensemble , displayed in figure [ fig : fig1]b , shows far better calibration , than that of the uwme .
the coverage of the aladin - huneps ensemble equals @xmath57 , which is very close to the nominal value of @xmath58 ( @xmath59 ) .
as mentioned earlier , the predictive performance of the csg emos model is tested on ensemble forecasts produced by the uwme and aladin - huneps epss , and the results are compared with the fits of the gev emos and gamma bma models investigated by @xcite and @xcite , respectively , and the verification scores of the raw ensemble .
we remark that according to the suggestions of @xcite for estimating the parameters of the gev emos model for a given day , the estimates for the preceding day serve as initial conditions for the box constrained broyden - fletcher - goldfarb - shanno @xcite optimization algorithm . compared with the case of fixed initial conditions this approach results in a slight increase of the forecast skills of the gev emos model , whereas for the csg emos method , at least in our case studies ,
fixed initial conditions are preferred .
further , we consider regional ( or global ) emos approach ( see e.g. * ? ? ?
* ) which is based on ensemble forecasts and validating observations from all available stations during the rolling training period and consequently results in a single universal set of parameters across the entire ensemble domain . to get the first insight about the calibration of emos and bma post - processed forecasts we consider probability integral transform ( pit ) histograms
generally , the pit is the value of the predictive cdf evaluated at the verifying observation @xcite , however , for our discrete - continuous models in the case of zero observed precipitation a random value is chosen uniformly from the interval between zero and the probability of no precipitation @xcite
. obviously , the closer the histogram to the uniform distribution , the better the calibration . in this way
the pit histogram is the continuous counterpart of the verification rank histogram of the raw ensemble and provides a good measure about the possible improvements in calibration .
the predictive performance of probabilistic forecasts is quantified with the help of the mean crps over all forecast cases , where for the raw ensemble the predictive cdf is replaced by the empirical one .
further , as suggested by @xcite , diebold - mariano ( dm ; * ? ? ?
* ) tests are applied for investigating the significance of the differences in scores corresponding to the various post - processing methods .
the dm test takes into account the dependence in the forecasts errors and for this reason it is widely used in econometrics .
besides the crps we also consider brier scores ( bs ; * ? ? ? * section 8.4.2 ) for the dichotomous event that the observed precipitation amount @xmath46 exceeds a given threshold @xmath60 .
for a predictive cdf @xmath45 the probability of this event is @xmath61 , and the corresponding brier score is given by @xmath62 see e.g. @xcite . obviously , the bs is negatively oriented and the crps is the integral of the bss over all possible thresholds . in our case studies we consider 0 mm precipitation , 5 , 15 , 25 , 30 mm and 1 , 5 , 7 , 9 mm threshold values for the uwme and aladin - huneps ensemble , respectively , corresponding approximately to the 45th , 75th , 85th and 90th percentiles of the observed non - zero precipitation accumulations , and compare the mean bss of the pairs of predictive cdfs and verifying observations over all forecast cases .
.5 truecm .@xmath21-values of kolmogorov - smirnov tests for uniformity of pit values for the uwme .
means of @xmath63 random samples of sizes @xmath64 each . [ cols="<,^,^,^",options="header " , ] concerning the two emos approaches , the verification scores of table [ tab : tab6 ] together with the results of the corresponding dm tests for equal predictive performance ( see table [ tab : tab7 ] ) display similar behavior as in the case of the uwme .
there is no significant difference between the mae values of the csg and gev emos methods and the former results in the lowest crps and the sharpest @xmath58 central prediction interval .
further , the emos models significantly outperform both the raw ensemble and the gamma bma approach , despite the raw ensemble is rather well calibrated and has far better predictive skill than the bma calibrated forecast .
note that the large mean crps and coverage of the bma predictive distribution is totally in line with the shape of the corresponding pit histogram of figure [ fig : fig4 ] .
2 mm 2 mm 2 mm the good predictive performance of the aladin - huneps ensemble can also be observed on the large amount of negative skill scores reported in table [ tab : tab8 ] and on the reliability diagrams of figure [ fig : fig5 ] .
similar to the case of the uwme , for 0 mm threshold the gamma bma model has good predictive performance , whereas for higher threshold values it underperforms the csg and gev emos models and the raw ensemble .
however , in connection with the reliability diagrams one should also note that the hectic behavior of the graphs ( compared to the rather smooth diagrams of figure [ fig : fig3 ] ) is a consequence of the shortage of data , as the verification period contains only 394 observations of positive precipitation , which is around one third of the forecast cases . taking into account both the uniformity of the pit values and the verification scores in tables [ tab : tab6 ] and [ tab : tab8 ] it can be said that the proposed csg emos model has the best overall performance in calibration of the raw aladin - huneps ensemble forecasts of precipitation accumulation .
a new emos model for calibrating ensemble forecasts of precipitation accumulation is proposed where the predictive distribution follows a censored and shifted gamma distribution , with mean and variance of the underlying gamma law being affine functions of the raw ensemble and the ensemble mean , respectively .
the csg emos method is tested on ensemble forecasts of 24 h precipitation accumulation of the eight - member university of washington mesoscale ensemble and on the 11 member aladin - huneps ensemble of the hungarian meteorological service .
these ensemble prediction systems differ both in the climate of the covered area and in the generation of the ensemble members . by investigating the uniformity of the pit values of predictive distributions , the mean crps of probabilistic forecasts , the brier scores and reliability diagrams for various thresholds , the mae of median forecasts and the average width and coverage of central prediction intervals corresponding to the nominal coverage , the predictive skill of the new approach is compared with that of the gev emos method @xcite , the gamma bma model @xcite and the raw ensemble . from the results of the presented case studies one can conclude that in terms of calibration of probabilistic and accuracy of point forecasts the proposed csg emos model significantly outperforms both the raw ensemble and the bma model and shows slightly better forecast skill than the gev emos approach .
* acknowledgments . *
sndor baran is supported by the jnos bolyai research scholarship of the hungarian academy of sciences .
dra nemoda partially carried out her research in the framework of the center of excellence of mechatronics and logistics at the university of miskolc .
the authors are indebted to michael scheuerer for his useful suggestions and remarks and for providing the r code for the gev emos model .
the authors further thank the university of washington muri group for providing the uwme data and mihly szcs from the hms for the aladin - huneps data .
99 bao l , gneiting t , raftery ae , grimit ep , guttorp p. 2010 .
bias correction and bayesian model averaging for ensemble forecasts of surface wind direction .
_ monthly weather review _ * 138*:18111821 , doi : 10.1175/2009mwr3138.1 .
baran s. 2014 .
probabilistic wind speed forecasting using bayesian model averaging with truncated normal components .
_ computational statistics and data analysis _ * 75*:227238 , doi : 10.1016/j.csda.2014.02.013 .
baran s , lerch s. 2015 .
log - normal distribution based emos models for probabilistic wind speed forecasting . _ quarterly journal of the royal meteorological society _ * 141*:22892299 , doi : 10.1002/qj.2521 .
baran s , sikolya k , veress l. 2013 . estimating the risk of a down s syndrome term pregnancy using age and serum markers : comparison of various methods .
_ communications in statistics simulation and computation _ * 42*:16541672 , doi : 10.1080/03610918.2012.674596 .
buizza r , houtekamer pl , toth z , pellerin g , wei m , zhu y. 2005 .
a comparison of the ecmwf , msc , and ncep global ensemble prediction systems .
_ monthly weather review _ * 133*:10761097 , doi : 10.1175/mwr2905.1 .
descamps l , labadie c , joly a , bazile e , arbogast p , cbron p. 2014 .
pearp , the mto - france short - range ensemble prediction system .
_ quarterly journal of the royal meteorological society _ * 141*:16711685 , doi : 10.1002/qj.2469 .
fraley c , raftery ae , gneiting t. 2010 .
calibrating multimodel forecast ensembles with exchangeable and missing members using bayesian model averaging .
_ monthly weather review _ * 138*:190202 , doi : 10.1175/2009mwr3046.1 .
friederichs p , thorarinsdottir tl .
forecast verification for extreme value distributions with an application to probabilistic peak wind prediction .
_ environmetrics _ * 23*:579594 , doi : 10.1002/env.2176
. gebhardt c , theis se , paulat m , bouallgue zb .
uncertainties in cosmo - de precipitation forecasts introduced by model perturbations and variation of lateral boundaries .
_ atmospheric research _
* 100*:168177 , doi : 10.1016/j.atmosres.2010.12.008 . gneiting t. 2014 .
calibration of medium - range weather forecasts .
_ ecmwf technical memorandum _ no . 719 .
( available from : http://old.ecmwf.int/publications/library/ecpublications/_pdf/tm/701-800/tm719.pdf . )
[ accessed on 24 november 2015 ] gneiting t , balabdaoui f , raftery ae .
probabilistic forecasts , calibration and sharpness .
_ journal of the royal statistical society : series b _ * 69*:243268 , doi : 10.1111/j.1467 - 9868.2007.00587.x . gneiting t , raftery ae , westveld ah , goldman t. 2005 . calibrated probabilistic forecasting using ensemble model output statistics and minimum crps estimation .
_ monthly weather review _ * 133*:10981118 , doi : 10.1175/mwr2904.1 .
grell ga , dudhia j , stauffer dr .
1995 . a description of the fifth - generation penn state / ncar mesoscale model ( mm5 ) .
technical note ncar / tn-398+str . national center for atmospheric research , boulder .
( available from : http://www2.mmm.ucar.edu/mm5/documents/mm5-desc-doc.html ) [ accessed on 24 november 2015 ] hamill tm .
2007 . comments on `` calibrated surface temperature forecasts from the canadian ensemble prediction system using bayesian model averaging . ''
_ monthly weather review _ * 135*:42264230 , doi : 10.1175/2007mwr1963.1
. hodyss d , satterfield e , mclay j , hamill tm , scheuerer m. 2015 .
inaccuracies with multi - model post - processing methods involving weighted , regression - corrected forecasts .
_ monthly weather review _ doi : 10.1175/mwr - d-15 - 0204.1 .
iversen t , deckmin a , santos c , sattler k , bremnes jb , feddersen h , frogner i - l . 2011 .
evaluation of glameps a proposed multimodel eps for short range forecasting .
_ tellus a _
* 63*:513530 , doi : 10.1111/j.1600 - 0870.2010.00507.x .
messner jw , mayr gj , zeileis a , wilks ds .
heteroscedastic extended logistic regression for postprocessing of ensemble guidance .
_ monthly weather review _ * 142*:448456 , doi : 10.1175/mwr - d-13 - 00271.1 .
scheuerer m. 2014 .
probabilistic quantitative precipitation forecasting using ensemble model output statistics .
_ quarterly journal of the royal meteorological society _ * 149*:10861096 , doi : 10.1002/qj.2183 .
scheuerer m , hamill tm .
statistical post - processing of ensemble precipitation forecasts by fitting censored , shifted gamma distributions _ monthly weather review _ * 143*:45784596 , doi : 10.1175/mwr - d-15 - 0061.1 .
sloughter jm , gneiting t , raftery ae .
probabilistic wind speed forecasting using ensembles and bayesian model averaging .
_ journal of the american statistical association _ * 105*:2537 , doi : 10.1198/jasa.2009.ap08615 .
sloughter jm , raftery ae , gneiting t , fraley c. 2007 .
probabilistic quantitative precipitation forecasting using bayesian model averaging .
_ monthly weather review _ * 135*:32093220 , doi : 10.1175/mwr3441.1
. swinbank r , kyouda m , buchanan p , froude l , hamill tm , hewson td , keller jh , matsueda m , methven j , pappenberger f , scheuerer m , titley ha , wilson l , yamaguchi m. 2015 . the tigge project and its achievements .
_ bulletin of the american meteorological society _ , doi : 10.1175/bams - d-13 - 00191.1 .
thorarinsdottir tl , gneiting t. 2010 .
probabilistic forecasts of wind speed : ensemble model output statistics by using heteroscedastic censored regression .
_ journal of the royal statistical society : series a _ * 173*:371388 , doi : 10.1111/j.1467 - 985x.2009.00616.x .
williams rm , ferro cat , kwasniok f. 2014 .
a comparison of ensemble post - processing methods for extreme events . _ quarterly journal of the royal meteorological society _ * 140*:11121120 , doi : 10.1002/qj.2198 . | recently all major weather prediction centres provide forecast ensembles of different weather quantities which are obtained from multiple runs of numerical weather prediction models with various initial conditions and model parametrizations .
however , ensemble forecasts often show an underdispersive character and may also be biased , so that some post - processing is needed to account for these deficiencies .
probably the most popular modern post - processing techniques are the ensemble model output statistics ( emos ) and the bayesian model averaging ( bma ) which provide estimates of the density of the predictable weather quantity . in the present work an emos method for calibrating ensemble forecasts of precipitation accumulation
is proposed , where the predictive distribution follows a censored and shifted gamma ( csg ) law with parameters depending on the ensemble members .
the csg emos model is tested on ensemble forecasts of 24 h precipitation accumulation of the eight - member university of washington mesoscale ensemble and on the 11 member ensemble produced by the operational limited area model ensemble prediction system of the hungarian meteorological service .
the predictive performance of the new emos approach is compared with the fit of the raw ensemble , the generalized extreme value ( gev ) distribution based emos model and the gamma bma method . according to the results ,
the proposed csg emos model slightly outperforms the gev emos approach in terms of calibration of probabilistic and accuracy of point forecasts and shows significantly better predictive skill that the raw ensemble and the bma model . _
key words : _ continuous ranked probability score , ensemble calibration , ensemble model output statistics , gamma distribution , left censoring . |
the discovery of jet quenching in central @xmath3 collisions at the relativistic heavy - ion collider ( rhic ) at brookhaven national laboratory has provided clear evidence for the formation of strongly interacting dense matter .
the observed jet quenching includes suppression of single hadron spectra at high @xmath0 @xcite , disappearance of back - to - back correlation of high @xmath0 hadrons @xcite and the azimuthal anisotropy of high @xmath0 hadron spectra in non - central @xmath3 collisions @xcite .
the absence of these jet quenching phenomena in @xmath4 collisions @xcite proves that they are indeed due to final state interaction .
detailed analyses @xcite further indicate that they are caused by parton energy loss as predicted @xcite . using the parton energy loss extracted from experimental data
, one can conclude that the initial gluon density in central @xmath3 collisions at @xmath5 gev is about 30 times higher than that in a cold nucleus @xcite .
the observed strong suppression of high @xmath0 hadrons at rhic is in sharp contrast to the results of @xmath6 collisions at the sps energy .
data from the wa98 @xcite experiment show no or little suppression of high @xmath0 ( up to about 4 gev/@xmath7 ) pions @xcite . even if one takes into account the possible uncertainty in the reference @xmath8 data @xcite , the suppression allowed by the data is still significantly less than in @xmath3 collisions at rhic , while the total charged multiplicities or the inferred initial parton densities only differ by a factor of 2 .
this implies that additional physics is at play in the energy dependence of the suppression of single hadron spectra in high - energy heavy - ion collisions
. it could be the energy dependence of the cronin effect due to initial state multiple scattering , the thermalization time and finite lifetime of the dense matter that limits the parton energy loss .
the energy dependence of the hadron suppression at large @xmath0 was predicted in ref .
@xcite and also was recently studied in ref .
@xcite , with parton energy loss proportional to the observed hadron multiplicity . in this brief report
, we explore the sensitivity of the final high @xmath0 hadron suppression to the lifetime of the dense matter as well as the dependence on the colliding energy .
we will study the high @xmath0 hadron suppression at @xmath9 , 62.4 , 200 and 5500 gev .
we focus , in particular , on both neutral pion and charged hadron suppression at @xmath2 gev and study the constraint on the lifetime by the measurement of hadron spectrum suppression .
we will use a lo pqcd model @xcite to calculate the inclusive high-@xmath0 hadron cross section in @xmath10 collisions , @xmath11 where @xmath12 are elementary parton scattering cross sections and @xmath13 is the nuclear thickness function normalized to @xmath14 .
we will use a hard - sphere model of nuclear distribution in this paper .
the @xmath15 factor is used to account for higher order qcd corrections .
the hadron is assumed to have the same rapidity as the parton , @xmath16 and its fractional momentum is defined as @xmath17 .
the parton distributions per nucleon @xmath18 inside the nucleus are assumed to be factorizable into the parton distributions in a free nucleon given by the mrs d@xmath19 parameterization @xcite and the impact - parameter dependent nuclear modification factor given by the new hijing parameterization @xcite .
the initial transverse momentum distribution @xmath20 is assumed to have a gaussian form with a width that includes both an intrinsic part in a nucleon and nuclear broadening .
detailed description of this model and systematic comparisons with experimental data can be found in ref .
@xcite .
the effect of parton energy loss is implemented through an effective modified fragmentation function @xcite , @xmath21 + e^{-\langle\frac{\delta l}{\lambda}\rangle } d^0_{h / c}(z_c , q^2 ) .
\label{modfrag } \end{aligned}\ ] ] this effective form is a good approximation to the actual calculated medium modification in the multiple parton scattering formalism @xcite , given that the actual energy loss should be about 1.6 times of the input value in the above formula .
here @xmath22 , @xmath23 are the rescaled momentum fractions and @xmath24 is the total energy loss during an average number of inelastic scatterings @xmath25 .
the ff s in free space @xmath26 are given by the bbk parameterization @xcite .
in this study , we assume a 1-dimensional expanding medium with a gluon density @xmath27 whose initial distribution is proportional to the transverse profile of participant nucleons .
the total energy loss for a parton propagating this medium is @xmath28 according to recent theoretical studies @xcite , where @xmath29 is the averaged initial gluon density at @xmath30 in a central collision .
@xmath31 is the average parton energy loss over a distance @xmath32 in a 1-d expanding medium with an initial uniform gluon density @xmath29 .
similarly , the average number of scatterings along the path of parton propagation is @xmath33 inclusion of transverse radial expansion generates a faster dilution of the gluon density relative to 1-d expansion , but also results in longer propagation time in medium .
these effects offset each other and the final total energy loss in the cases of 1-d and 3-d expansion is found very similar @xcite . with a hard - sphere nuclear distribution ,
the gluon density profile @xmath27 can be expressed in a simple form and the analytic expressions of @xmath34 and @xmath25 are given in the appendix .
the energy dependence of the energy loss is parameterized as @xmath35 according to the numerical results in ref .
@xcite in which thermal gluon absorption is also taken into account in the calculation of parton energy loss .
fit to the most central @xmath3 collisions at @xmath5 results in @xmath36 gev / fm , @xmath37 gev and @xmath38 fm .
the corresponding energy loss in a static medium with parton density @xmath29 over a distance @xmath32 is @xcite @xmath39 gev / fm .
this is about 30 times higher than the parton energy loss in a cold nucleus @xcite . at different energies , we assume that @xmath40 and @xmath29 are proportional to the measured hadron multiplicity as given in ref . @xcite .
since there is no experimental measurement at the lhc energy yet , we will use the model calculation @xcite to extrapolate and assume @xmath41 . to study the dependence of the high @xmath0 hadron suppression on the colliding energy ,
let us assume first that the lifetime of the dense matter is larger than the system size .
shown in fig .
[ fig1 ] are the the nuclear modification factors @xmath42 for charged hadron ( solid lines ) and neutral pions ( dashed lines ) in central @xmath3 ( @xmath6 at the sps energy ) collisions at different energies , from sps @xmath9 gev to lhc @xmath43 tev . here , @xmath44 is the number of geometrical binary collisions at a given range of impact parameters . at the sps energy ,
the observed nuclear modification factor in central @xmath6 collisions is consistently about 1 due to strong cronin effect via initial multiple parton scattering , leaving not much room for large parton energy loss @xcite .
we shall return to this point later .
in central @xmath3 collisions at rhic , however , strong suppression of high @xmath0 hadrons is observed .
this can be attributed to large parton energy loss that overcomes the modest cronin enhancement as observed in @xmath4 collisions @xcite and gives rise to the large hadron suppression .
the energy - dependence of the parton energy loss in eq .
( [ eq : loss ] ) describes well the flat @xmath0 dependence of the nuclear modification factor @xmath1 for neutral pions at large @xmath0 in @xmath3 collisions at @xmath5 gev , as shown in the figure . for charged hadrons
there are complications from other medium effects as we will discuss later .
the flatness of the modification factor at this energy is more clearly illustrated by the ratio of central to peripheral collisions @xcite .
such a flat @xmath0 dependence is actually a coincidence , as the combined effect of the energy dependence of the parton energy loss in eq .
( [ eq : loss ] ) and the power - law behavior of the initial jet spectra . using the same energy dependence of the parton energy loss but with a reduced amplitude due to smaller initial gluon density at @xmath2 gev
, the nuclear modification factor is found to decrease with @xmath0 and even becomes smaller than the modification factor at @xmath5 gev at high @xmath45 gev/@xmath7 .
this is simply a consequence of the energy dependence of jet spectrum shape .
the initial jet spectra at @xmath46 gev are much steeper than those at 200 gev .
the same amount of energy loss leads to a larger suppression of the final hadron spectra at 62.3 gev than at 200 gev .
as one increases the colliding energy , the power - law spectra for the initial jet production become flatter , and the same parton energy loss will lead to less suppression of the final hadrons . as shown in the same figure ,
the nuclear modification factor at the lhc energy @xmath43 tev is smaller than at 200 gev in the intermediate @xmath0 region , due to larger initial gluon density .
however , the modification factor @xmath1 increases with @xmath0 due to the flatter power - law spectra of jet production at lhc .
the parameterized energy dependence of the parton energy loss is in part due to the detailed balance effect in induced gluon radiation and absorption @xcite .
this effect is most important in the intermediate @xmath0 region . in this region
, one expects the jet fragmentation process to be modified by other non - perturbative processes such as parton recombination or coalescence @xcite .
the observed flavor dependence of the hadron suppression and of the azimuthal anisotropy clearly points to the effect of parton recombination that enhances both baryon and kaon spectra in the presence of dense medium . to include this effect in the current parton model
, we have added a soft component to kaon and baryon ff s that is proportional to the pion ff with a weight @xmath47 $ ] . the functional form and parameters
are adjusted so that @xmath48 at @xmath49 gev/@xmath7 in the most central @xmath3 collisions at @xmath5 gev and approaches its @xmath8 value at @xmath50 gev/@xmath7 .
this gives rise to the splitting of the suppression factor for charged hadrons and @xmath51 in the calculation . because of the steeper power - law spectra of jet production at 62.4 gev
, the effect of the non - perturbative parton recombination persists to higher @xmath0 than in @xmath3 collisions at 200 gev . in this region of @xmath0 ,
the non - perturbative recombination effects dominate the nuclear modification of the charged hadron spectra . as a consequence ,
the @xmath0 dependence of the modification factors @xmath1 at @xmath2 and 200 gev are similar .
they only diverge at high @xmath52 8 gev/@xmath7 where the recombination effects are negligible at both energies .
since the soft components due to parton recombination are not closely related to medium thermalization , they will still contribute to the final hadron spectra in peripheral @xmath10 and @xmath53 collisions .
this will lead to flavor dependence of the cronin effect in @xmath53 collisions @xcite .
in the calculation of the parton energy loss in eqs .
( [ deltae ] ) and ( [ dlamb ] ) , the upper limit of the path integral should be @xmath54 where @xmath55 is the lifetime of the dense matter before breakup , @xmath56 is the distance the parton , produced at @xmath57 , has to travel along @xmath58 at an azimuthal angle @xmath59 relative to the reaction plane in a collision with impact - parameter @xmath60 . according to the analyses of experimental data on high @xmath0 hadron suppression and suppression of away - side correlation in @xmath3 collisions at @xmath5 gev , the extracted energy loss points to an initial gluon density of about 30/@xmath61 at an initial time @xmath62 fm . given the measured transverse energy per charged hadron of 0.8 gev @xcite , this gives a lower bound on the initial energy density of about 25 gev/@xmath61 . in 1-d expansion with the equation of state of an ideal fluid , the energy density decreases with time , @xmath63 . assuming the 1-d hydrodynamics expansion starts at @xmath64 fm/@xmath7 ( free - streaming before that ) , the lifetime of the plasma or the duration for the parton energy loss should be about @xmath65 fm , before the phase transition with a critical energy density @xmath66 gev/@xmath61 .
the early stage of the mixed phase or crossover could also contribute to the jet quenching and thus extend the effective time duration for parton energy loss .
when this time is larger than the average path length , the total parton energy loss is then limited only by the system size . in the previous analysis @xcite ,
such an assumption for central @xmath3 collisions is justified given the high initial energy density . in central @xmath6 collisions at the sps energy @xmath67 ,
the rapidity density of charged particles is about half of that in central @xmath3 collisions at @xmath5 gev @xcite .
the average transverse energy per charged particle is about the same at sps and rhic energy @xcite .
one can then assume the initial energy density at sps to be half of that in central @xmath3 collisions at @xmath5 gev .
the lifetime of the plasma in central @xmath6 collisions at sps , @xmath68 , if it were formed , is then considerable shorter than that at the highest energy of rhic .
the longer thermalization time at sps could make the effective lifetime even shorter . to demonstrate the dependence on the lifetime , we show in fig .
[ fig1 ] the nuclear modification factor of @xmath51 at the sps energy with short lifetime @xmath69 fm/@xmath7 .
the modification factor is quite sensitive to the lifetime @xmath55 .
even if one takes into account the uncertainty in the reference @xmath8 data @xcite , which gives a systematic error of about a factor of 2 , the data still point to a short lived dense system with @xmath70 fm/@xmath7 .
this might also explain the observed elliptic flow @xmath71 at sps that is much smaller than the hydrodynamic limit @xcite .
on the other hand , the longer lifetime of the plasma allows development of a full hydrodynamic flow , giving rise to a large @xmath71 that saturates the hydrodynamic limit @xcite .
to study the sensitivity of the high @xmath0 hadron suppression to the lifetime of the plasma at @xmath2 gev , we show in fig .
[ fig2 ] the nuclear modification factors for both charged hadrons and neutral pions with different values of @xmath55 in @xmath3 collisions at @xmath2 gev .
the hadron suppression in the large @xmath0 region in the most central collisions is very sensitive to the lifetime of the plasma in this calculation . in peripheral collisions , the small size of the dense
medium limits the parton energy loss . as a result
, the hadron suppression is only sensitive to values of @xmath55 that are smaller than the average medium size . in reality
, the values of @xmath55 should decrease from central to peripheral collisions .
the recent experimental results from phobos @xcite on nuclear modification factors for charged hadrons in @xmath3 collisions at 62.4 gev only extend to @xmath72 gev/@xmath7 . in this region ,
the suppression of charged hadron is indeed much smaller than at 200 gev .
however , charged hadrons in this region are also dominated by non - perturbative recombination effects , though our results are still sensitive to the lifetime .
experimental measurements of @xmath51 and high @xmath0 charged hadrons , both are less influenced by the parton recombination effect , should provide more stringent constraints on the lifetime of the dense matter .
within a parton model incorporating medium induced parton energy loss , we have studied in this brief report the suppression of inclusive hadron spectra at high @xmath0 in heavy - ion collisions at different energies .
we found that the @xmath0 dependence of the nuclear modification factor @xmath1 is determined by the energy dependence of the parton energy loss and the power - law behavior of the initial jet spectra . with the onset of parton energy loss and the change of the power - law jet spectra , the @xmath0 dependence of the modification factor changes from monotonic decrease at @xmath2 gev to monotonic increase with @xmath0 at the lhc energy @xmath43 tev .
the flat @xmath0 dependence observed at @xmath5 gev is just a coincidence .
we also studied the sensitivity of the hadron suppression factor to the lifetime of the plasma or the duration of parton energy loss .
we found that the hadron suppression factor at intermediate and large @xmath0 is sensitive to the lifetime if it is comparable or smaller than the system size .
the experimental measurements could provide important constraints .
together with the measurement of elliptic flow , which is also sensitive to the lifetime and thermalization time of the plasma , one can gain additional information on the dynamic evolution of the produced quark - gluon plasma .
one can also calculate the back - to - back dihadron correlation at different energies .
we find that suppression of the back - to - back dihadron correlation at 62.4 gev in central @xmath3 collisions is almost identical to that at 200 gev .
this is partly due to the trigger bias that selects dihadron production close to the surface and results in completely suppression the back - side jets that traverse the whole length of the dense matter .
the suppression due to @xmath73 broadening of initial multiple parton scattering is also independent of the colliding energy .
the author thanks p. jacobs for helpful discussion about this manuscript .
this work was supported by the director , office of energy research , office of high energy and nuclear physics , divisions of nuclear physics , of the u.s .
department of energy under contract no.de-ac03-76sf00098 and de - fg03 - 93er40792 .
in this appendix , we give the basic analytic formula for calculating the path integral in the parton energy loss in eqs .
( [ deltae ] ) and ( [ dlamb ] ) , assuming a hard - sphere nuclear distribution . given two overlapping nuclei as illustrated in fig .
[ fig3 ] , we want to calculate a path integral over the path @xmath74 .
let @xmath75 and @xmath76 be the radial coordinates of the jet production point as measured from the center of the two nuclei . for given @xmath77 and @xmath78 @xmath79 for a hard - sphere distribution , @xmath80 and @xmath81 .
the jet travels at an angle @xmath82 with respect to the impact - parameter @xmath77 ( which determines the reaction plane with the beam direction ) .
@xmath83 is the angle between @xmath77 and @xmath78 .
the distance the jet travels is then the @xmath84 ; @xmath85 where @xmath86 .
these are solutions of @xmath87 assuming that the soft gluon density is proportional to the number of participant nucleons , it is then given by @xmath92,\end{aligned}\ ] ] where @xmath93 and @xmath29 is defined as the averaged gluon density in central collisions ( @xmath94 ) at an initial time @xmath30 : @xmath95 using the nuclear thickness function defined as @xmath96 the gluon density is then @xmath97.\end{aligned}\ ] ] k. adcox _ et al .
_ , phys .
lett . * 88 * , 022301 ( 2002 ) [ arxiv : nucl - ex/0109003 ] . c. adler _ et al .
_ , phys .
lett . * 89 * , 202301 ( 2002 ) . c. adler _ et al .
_ , phys .
lett . * 90 * , 082302 ( 2003 ) . c. adler _ et al .
_ , phys .
lett . * 90 * , 032301 ( 2003 ) .
j. adams _ et al .
_ [ star collaboration ] , phys .
lett . * 91 * , 072304 ( 2003 ) .
x. n. wang , arxiv : nucl - th/0305010 .
i. vitev , in proceedings of 17th international conference on ultrarelativistic heavy - ion collisions , jan .
11 - 17 , oakland , california , eds . h .-
ritter and x .-
wang , arxiv : hep - ph/0403089 .
i. vitev , arxiv : nucl - th/0404052 .
m. m. aggarwal _ et al .
_ [ wa98 collaboration ] , phys .
lett . * 81 * , 4087 ( 1998 ) [ erratum - ibid . *
84 * , 578 ( 2000 ) ] [ arxiv : nucl - ex/9806004 ] ; r. albrecht _ et al . _
[ wa80 collaboration ] , eur .
j. c * 5 * , 255 ( 1998 ) [ arxiv : nucl - ex/9805007 ] .
x. n. wang , phys .
c * 61 * , 064910 ( 2000 ) .
c. a. salgado and u. a. wiedemann , phys .
lett . *
89 * , 092303 ( 2002 ) .
e. wang and x. n. wang , phys .
lett . * 89 * , 162301 ( 2002 ) .
@xcite m. gyulassy , i. vitev , x. n. wang and p. huovinen , phys .
b * 526 * , 301 ( 2002 ) [ arxiv : nucl - th/0109063 ] .
k. adcox _ et al . _
[ phenix collaboration ] , phys .
lett . * 87 * , 052301 ( 2001 ) [ arxiv : nucl - ex/0104015 ] .
h. appelshauser _ et al . _
[ na49 collaboration ] , phys .
lett . * 80 * , 4136 ( 1998 ) [ arxiv : nucl - ex/9711001 ] . | suppression of high @xmath0 hadron spectra in high - energy heavy - ion collisions at different energies is studied within a pqcd parton model incorporating medium induced parton energy loss .
the @xmath0 dependence of the nuclear modification factor @xmath1 is found to depend on both the energy dependence of the parton energy loss and the power - law behavior of the initial jet spectra .
the high @xmath0 hadron suppression at @xmath2 gev and its centrality dependence are studied in detail .
the overall values of the modification factor are found to provide strong constraints on the lifetime of the dense matter . _
pacs numbers : _
12.38.mh , 24.85.+p ; 13.60.-r , 25.75.-q = 0.5 in 2 |
lyapunov analysis is one of the main tools to quantitatively characterize dynamical systems @xcite . in particular , the largest lyapunov exponent ( le ) @xmath3 is a fundamental quantity that allows one to address , at some basic level , the question of the predictability of a system .
a positive @xmath3 implies chaotic behavior : neighboring initial conditions diverge in time at an exponential rate @xmath4 .
actually , the le is an asymptotic magnitude in the sense that it is defined in the infinite - time limit .
however , it has been since long recognized that the infinite - time limit is often irrelevant to address many important questions . therefore ,
the _ finite - time
_ le ( to be defined below ) was introduced as an essential tool to deal with problems such as predictability @xcite and shadowability @xcite of dynamical systems .
correspondingly , the singular vector ( sv ) ( also referred to as the finite - time lyapunov vector ) is defined as the perturbation that will show the maximal expansion after some _ finite _ time @xmath0 .
singular vectors are typically used in operative models for weather prediction @xcite . every day
the ecmwf ( european centre for medium - range weather forecasts ) @xcite computes a set of singular vectors to perform an ensemble forecasting of the european weather @xcite .
also , in a different context , svs have been found to be useful to study advection of passive particles in fluids @xcite .
it is a well - known result @xcite that as the the time horizon @xmath0 increases the sv exponentially approaches the so - called forward lyapunov vector ( flv ) @xcite . for low dimensional systems our understanding in these simple terms might be satisfactory .
however , in extended systems it is clear that spatio - temporal correlations should have an important effect in the way this limit is achieved .
how this convergence of the sv toward the flv occurs and what generic system - independent features ( if any ) exist are important open questions in the case of chaotic extended systems . indeed , this is also of practical and theoretical importance in the field of atmospheric sciences @xcite , because it is directly related with the problem of realistic forecasting within a given finite time horizon @xmath0 . in this paper
we analyze the structure of the sv in extended systems with spatiotemporal chaos .
we show that the sv localizes exponentially in space around one center ( determined by the present state of the system and by the horizon @xmath0 ) .
we find that the localization strength generically scales with the observation time as a power - law @xmath1 .
moreover , we propose a stochastic field - theoretical description of the evolution of the sv , which allows us to give an analytical prediction for the universal exponent @xmath2 and the finite-@xmath0 deviation of the le , in good agreement with simulations in systems of different natures , including coupled - map lattices and differential equations .
numerical results together with general scaling arguments suggest our results are generic for a wide class of systems exhibiting space - time chaos .
for any given dynamical system , infinitesimal perturbations @xmath5 are governed by linear equations ( tangent space ) such that the perturbation after some @xmath0 can be expressed via a linear operator ( resolvent or propagator ) : @xmath6 .
we are interested here in the sv , which is defined as the perturbation in tangent space at time @xmath7 that gets amplified the most at some future time @xmath8 . in a more mathematical form ,
let @xmath9 denotes the sv for a time horizon @xmath0 , then we have @xmath10 where @xmath11 denotes the scalar product .
the `` optimal '' perturbation @xmath12 is the sv and it depends on both the time @xmath7 ( that determines the present state of the system ) , and the optimization time @xmath0 . using the adjoint operator @xmath13 defined by @xmath14 , the problem posed in eq .
( [ eq_p ] ) can be solved by finding the eigenvector of @xmath15 with the largest eigenvalue ( all of them are real and positive ) . for the euclidean scalar product used throughout this paper we have @xmath16 .
the iterative application of the operator @xmath17 to an arbitrary vector ( the power method ) , generically converges to the wanted eigenvector @xmath9 , which satisfies the eigenvalue problem @xmath18 .
then , the finite - time le is typically defined as @xmath19 @xcite , which depends on @xmath7 and on the optimization time @xmath0 .
according to oseledec s theorem @xcite in the limit @xmath20 with probability one @xmath21 converges to a time - independent quantity namely , the largest le of the chaotic attractor @xmath22 . an important observation is in order : @xmath3 is usually obtained from the evolution of a perturbation grown from the past , instead of considering how a perturbation will evolve in the future . thus a perturbation evolved since the remote past is the `` most unstable '' eigenvector of @xmath23 @xcite , which has been called backward lyapunov vector ( blv ) @xcite .
conversely , the flv is eigenvector of @xmath24 . in order to gain some numerical insight into the problem
we have studied the behavior of the sv in two prototypical dynamical systems exhibiting spatiotemporal chaos .
the first example we consider is a cml in a ring , which is discrete in space ( @xmath25 ) and time ( @xmath26 ) : @xmath27 with the logistic map @xmath28 , and coupling strength @xmath29 .
the second model we consider here is an example of a chaotic system described by differential equations .
we study the model proposed by lorenz in 1996 ( l96 ) @xcite as a toy model in the context of weather dynamics .
we consider the variables @xmath30 defined in a ring geometry , and the evolution equations @xmath31 with @xmath32 .
we used the euler method for the integration with @xmath33 .
an important feature of singular and lyapunov vectors is that their amplitudes are much localized in space .
localization of svs was observed long time ago in simple meteorological models @xcite , however , to our knowledge , a description of the localization profile in space or the dynamics does not exists even at a qualitative level .
in contrast , the localization properties of the blv are much better understood .
et al . _
@xcite demonstrated that , in a large class of one - dimensional systems with spatiotemporal chaos , the blv can be mapped into a roughening surface that belongs to the universality class of the kardar - parisi - zhang ( kpz ) equation @xcite .
the solutions of the kpz equation present universal scaling exponents in space and time @xcite . in particular , and as a result
, one can conclude that the blv in one dimension is sub - exponentially localized , with its magnitude decaying in space as @xmath34 from the localization site .
these results apply to the flv as well because the operator @xmath35 has the same coarse - grained long - scale structure than @xmath36 , and thus the flv surface belongs to the kpz universality class too .
the spatial structure of a lyapunov vector is better resolved by making a logarithmic transformation @xcite . in the same spirit , expressing the vector components as @xmath37_{x=1}^{x = l}$ ] we define a sv surface via the hopf - cole transformation : @xmath38 . in this representation
the finite-@xmath0 le becomes the average velocity of the surface : @xmath39 / s_\tau(x , t ) \vert$ ] .
( @xmath40 ) .
curves have been shifted to improve their visibility .
( b ) structure factor for different values of @xmath0 ( averaged over 500 realizations).[fig1],width=302 ] figure [ fig1](a ) shows typical sv surfaces for different values of @xmath0 in the case of the cml model ( [ cml ] ) .
we obtained similar behavior for the l96 model ( [ l96 ] ) .
we observe that for @xmath0 larger than some threshold @xmath41 the corresponding surface has approximately a random - walk profile ; or in terms of the sv itself we have stretched - exponential localization for a a time horizon @xmath42 , as anticipated above .
remarkably , for @xmath43 we find that the height profile of the sv surface is facetted with superimposed fluctuations at small scale .
note that a triangular sv surface corresponds to an exponential localization of the sv .
the triangular pattern progressively dies out for longer times @xmath0 .
this means that the asymptotic convergence to the flv is reached for practical purposes in a finite time : when we look for the perturbation at present time @xmath7 that will exhibit the largest amplification at a future time @xmath8 , there is a characteristic time @xmath41 that separates the genuine finite - time regime from the quasi - infinite - time regime . in the latter regime ( @xmath44 )
the sv has almost collapsed into the flv .
quantitative information about the spatial correlations of the sv can be obtained calculating the structure factor : @xmath45 , where @xmath46 , and the bracket indicates average over realizations .
figure [ fig1](b ) shows the results for different values of the temporal horizon @xmath0 . for large @xmath0
, one recovers the typical spatial correlations corresponding to the flv , @xmath47 , as expected from the previous discussion .
contrastingly , for @xmath48 we find @xmath49 at small wave numbers ( due to the symmetry of the triangular profile only wavenumbers @xmath50 with @xmath51 odd contribute significantly to the structure factor ) . the coarse - grained slope of the triangle @xmath52 , and the surface width @xmath53 ^ 2 $ ] serve to quantify the localization strength .
the following relation holds for small @xmath54 and @xmath43 : @xmath55 for @xmath51 odd .
svs are the asymptotic vectors computed by repetitive application of the operator @xmath56 , to an arbitrary initial vector .
this makes the sv similar , albeit not exactly equivalent , to the blv of a periodic orbit .
in fact , for temporally periodic but spatially chaotic states the evolution in tangent space is mapped exactly onto an one - dimensional localization problem and it can be rigorously proved that all the eigenvectors are exponentially localized @xcite . for the flv to be computed , and because of the @xmath57 limit , only the @xmath35 operator needs to be used @xcite .
however , the structure of the operator @xmath58 is obviously different from that of @xmath35 . in fig .
[ fig2 ] we compare the structure factors obtained with the example of the cml when computing the surfaces associated to the eigenvectors of @xmath59 ( the actual singular vector ) and @xmath13 ( which can be viewed as the blv of an unstable periodic orbit of period @xmath0 ) .
one may see that , in statistical terms , the large length scale behavior of the main eigenvector of @xmath60 is well reproduced using operator @xmath13 alone .
this fact justifies considering a minimal model ( for the long scale structure of the sv ) consisting of a stochastic equation with _ periodic _ noise . and @xmath61 for @xmath62 in the cml model after average over 500 realizations . the long scale structure ( small @xmath54 ) of the eigenvector of @xmath61 is well captured by the eigenvector of @xmath63 alone .
[ fig2],width=264 ] we now write a minimal langevin model to understand in the simplest terms the structure of the sv surface on long wavelengths .
we propose a modification of the kpz equation considering a periodic noise ( pnkpz ) : @xmath64 ^ 2 + \partial_{xx } h_\tau(x , t ) , \label{pnkpz}\ ] ] where one simply assumes @xmath65 to be a random noise with period @xmath0 [ _ i.e. _ @xmath66 and a @xmath67-correlator , @xmath68 , for @xmath69 .
( @xmath40 ) .
curves are arbitrarily shifted to improve their visibility .
( b ) structure factors for different values of @xmath0 ( averages were done over 500 realizations).[fig3],width=302 ] figure [ fig3 ] shows the results of our numerical integration of eq .
( [ pnkpz ] ) ; space and time were discretized ( @xmath70 ) and a noise intensity @xmath71 was used .
other intensities can be used and this , of course , has no effect on the results .
one can see that the solutions of ( [ pnkpz ] ) adopt a triangular pattern , akin to those shown in fig .
[ fig1 ] for the sv surface of the deterministic model .
the triangular structure flattens as @xmath0 increases and eventually disappears above some threshold value @xmath41 . in fig .
[ fig4](a , b ) we plot the surface width @xmath72 as a function of @xmath0 . figure [ fig4](a ) was generated from the sv surfaces of the chaotic systems in eqs .
( [ cml])-([l96 ] ) , while fig .
[ fig4](b ) was obtained from the numerical integration of the stochastic pnkpz equation ( [ pnkpz ] ) . for not too large values of @xmath0
we find a region of scaling where the width is dominated by the triangular structure .
we find @xmath73 the values of @xmath0 with a @xmath74 of confidence are : @xmath75 for the cml , @xmath76 for the l96 model , and @xmath75 for the pnkpz equation .
this strongly suggests that @xmath2 is a universal exponent . .
( a ) svs of cml and l96 models for @xmath77 and @xmath78 , respectively .
( b ) pnkpz for @xmath40 .
( c , d ) finite-@xmath0 deviation of the le .
data of the l96 model have been shifted two decades downwards to avoid overlap of data sets . in all panels symbols mark mean values , and bars indicate the statistical dispersion ( 100 realizations were done for the l96 model and 500 for the other models ) .
[ fig4],width=302 ] in all cases we expect the scaling to break down for large @xmath0 , as can be inferred from the fact that the stochastic model ( [ pnkpz ] ) converges to the kpz problem as @xmath79 .
this translates into the convergence of the sv to the flv as the time horizon @xmath0 tends to infinity .
the existence of the characteristic time @xmath41 means that the flv is virtually reached in a finite time for a finite - size system . for the theoretical analysis to follow
, it is worth to remark that triangular solutions also appear in the case of the quenched columnar kpz ( qckpz ) equation discussed in ref .
@xcite , that corresponds to eq .
( [ pnkpz ] ) but with the periodic noise @xmath80 replaced with a quenched columnar disorder @xmath81 .
note that in the limit @xmath82 of the pnkpz equation [ cf .
( [ pnkpz ] ) ] one recovers the qckpz equation .
however , for a finite @xmath0 correlations spread and one may conjecture that the problem can be seen effectively as a qckpz equation where the disorder @xmath81 has a finite correlation length .
the asymptotic solutions of the qckpz equation have been described as a triangular pattern ( with superimposed random fluctuations ) and exhibit scale - invariant roughness @xmath83 @xcite .
the existence of these two limiting dynamics namely , kpz for @xmath79 and qckpz for @xmath82 describes the behavior of the pnkpz equation and can be exploited to derive an analytical expression for the exponent @xmath2 by resorting to scaling arguments as follows . on the one hand , the width of the pnkpz surface should scale as @xmath84 for @xmath42 , and it should make a crossover to @xmath85 for @xmath86 . therefore , these two limiting behaviors are separated by the crossover time @xmath87 . on the other hand , the characteristic time @xmath41 should also correspond to the typical time that an initially flat kpz surface needs to reach the statistically stationary regime in a system of size @xmath88 , so we have @xmath89 @xcite . equating both expressions we arrive at @xmath90 which is in principle valid in _ any _ dimension and
can be compared with our numerical simulations . inserting the critical exponents in one dimension
@xmath91 and @xmath92 @xcite ; and @xmath93 @xcite we get @xmath94 in good agreement with our simulations ( our direct estimation of @xmath95 for the pnkpz equation ( [ pnkpz ] ) with @xmath96 yields @xmath97 , so @xmath98 ) .
a very interesting outcome of our analysis is the existence of a scaling law for the finite - time le . the average velocity of the pnkpz surface satisfies @xmath99 , where in good approximation the noise term has been assumed to average out within a period .
existing results on qckpz @xcite indicate that there is a complete separation in two independent components such that @xmath100 can be constructed as the sum of a triangular pattern with a tilt @xmath101 and the solutions in @xmath20 . after some algebra , one obtains @xmath102 in fig .
[ fig4](d ) we test the validity of this expression by plotting @xmath103 as a function of @xmath0 .
the dashed line arises inserting in ( [ finite_le ] ) the mean values of @xmath104 depicted in fig .
[ fig4](b ) .
the straight line is computed by inserting in the dominant contribution @xmath105 the fitting power law in fig .
[ fig4](b ) : @xmath106 .
the agreement is , in our opinion , very good taking into account the simplicity of our arguments . for a generic system with spatiotemporal chaos
we may not expect to be able to obtain a formula relating the width of the sv surface and the finite-@xmath0 deviation of the le , but still we may ask whether the @xmath1 dependence is observed . figure [ fig4](c ) indicates that this scaling for the le exists in the same region where @xmath107 , although the result is not conclusive for the l96 model because computer limitations did not allow us to study larger systems . finally , we point out that for @xmath108 the convergence of @xmath109 to @xmath3 becomes faster , precisely as @xmath110 in accordance with the standard asymptotic behavior for low - dimensional systems @xcite .
this means that the spatial degrees of freedom slow down the convergence to the le in extended dynamical systems as compared with the low - dimensional case .
we have shown that for a large family of systems the sv exhibits exponential localization .
we claim that this spatial structure can be described in terms of the solutions of the a stochastic equation , related to kpz equation , in which the noise is periodic or quenched . in terms of the associated surfaces , the sv converges toward the flv following a power law with a universal exponent @xmath111 given by eq .
( [ gamma ] ) .
in particular , for one - dimensional systems one has @xmath112 .
moreover , the same exponent characterizes the finite - time deviation of the le . even for spatio - temporal chaotic systems whose blv ( and flv ) does not belong to the universality class of kpz , as occurs with hamiltonian lattices @xcite
, one may expect a qualitative similarity with the scenario presented in this work , but possibly with different exponents . in forecasting applications @xmath0
is the control parameter to calibrate the ensemble of svs .
our results show how @xmath0 quantifies ( through the universal exponent @xmath2 ) the localization strength , and the exponential growth rate .
financial support from the ministerio de educacin y ciencia ( spain ) under projects fis2006 - 12253-c06 - 04 and cgl2007 - 64387/cli is acknowledged .
d.p . acknowledges the support by csic under the junta de ampliacin de estudios programme ( jae - doc ) . | in a dynamical system the singular vector ( sv ) indicates which perturbation will exhibit maximal growth after a time interval @xmath0 .
we show that in systems with spatiotemporal chaos the sv exponentially localizes in space . under a suitable transformation
, the sv can be described in terms of the kardar - parisi - zhang equation with periodic noise .
a scaling argument allows us to deduce a universal power law @xmath1 for the localization of the sv .
moreover the same exponent @xmath2 characterizes the finite-@xmath0 deviation of the lyapunov exponent in excellent agreement with simulations .
our results may help improving existing forecasting techniques . |
millimeter , sub - millimeter , and centimeter wavelength observations of star - forming sources known as `` hot cores '' reveal a range of complex organic molecules , through the identification of molecular rotational emission spectra ( herbst & van dishoeck 2009 ) .
many of the molecules detected toward hot cores are highly saturated ( i.e. hydrogen - rich ) , and a significant number also contain oxygen and/or nitrogen ; examples include methanol ( ch@xmath4oh ) , formic acid ( hcooh ) , and the ubiquitous methyl formate ( hcooch@xmath4 ) , as well as the recently detected aminoacetonitrile ( nh@xmath0ch@xmath0cn , belloche et al .
2008 ) , which bears both amine and nitrile groups .
the breadth of functional groups borne by molecules detected toward hot cores ( as well as the large number of unidentified emission lines ) suggests that other , as - yet undetected species may also be present , including biologically - significant molecules such as amino acids the building blocks of proteins .
the search for the simplest amino acid , glycine ( nh@xmath0ch@xmath0cooh ) , toward interstellar sources has not so far yielded a detection ( snyder et al .
2005 ) ; recent upper limits toward sgr b2(n ) , a spectrally- and chemically - rich hot - core source in the galactic center , are around 10@xmath5 10@xmath6 @xmath7 for beam - averaged values obtained with the mopra telescope ( cunningham et al .
2007 ) and the atca array ( jones et al .
2007 ) for glycine conformer i. upper limits on the higher - energy conformer ii from the same studies are around 10@xmath8 10@xmath5 @xmath7 .
however , numerous amino acids have been found in meteorites , and glycine was recently detected in a sample taken from comet wild 2 ( elsila et al .
2009 ) ; it has indeed been postulated that terrestrial amino acids may have an interstellar origin ( e.g. ehrenfreund & charnley 2000 ) .
it is likely that amino acids and other biologically - relevant species will , in the coming years , provide plausible targets for searches with the alma telescope , which can provide much greater sensitivity and spatial resolution than was previously available .
it is therefore valuable to extend existing astrochemical models to simulate the putative formation of glycine and related species in hot cores , both to make quantitiative predictions and to assess the viability of observational candidate sources .
current models of hot - core chemistry place great emphasis on the formation of organic molecules within or upon dust - grain ice mantles , as well as in the gas phase .
garrod & herbst ( 2006 , hereafter g&h ) showed that typically - observed molecules such as methyl formate , while difficult to form by gas - phase mechanisms , could be produced in appropriate abundances through grain - surface chemistry at intermediate temperatures , prior to the complete sublimation of the dust - grain ices .
the radicals hco and ch@xmath4 were found to become mobile around 20 40 k , reacting with other radicals associated with formaldehyde and methanol destruction , to produce formic acid , dimethyl ether and methyl formate .
cosmic ray - induced photolysis of the molecular ice mantles was found to be the dominant source of these functional - group radicals .
garrod , widicus weaver & herbst ( 2008 , hereafter gwh ) extended the network to include a much larger number of surface radical species , allowing the study of more complex molecules ; methanol was found to be the key molecule for the supply of the constituent functional groups of many such species .
recent experimental work by berg et al .
( 2009a ) has provided direct evidence that a range of complex organic molecules detected in star - forming regions may be readily produced within uv - irradiated organic ices .
other studies focused specifically on amino - acid formation have elucidated a number of potential chemical pathways for glycine within laboratory ice mixtures .
experiments by bernstein et al .
( 2002 ) and muoz caro et al .
( 2002 ) showed that uv irradiation of interstellar ice analogs containing h@xmath0o , ch@xmath4oh , co , co@xmath0 , nh@xmath4 and/or hcn could produce significant quantities of amino acids , including glycine . sorrell ( 2001 ) had suggested a route involving the reaction of the nh@xmath0 radical with acetic acid ( ch@xmath4cooh ) , itself formed through the addition of ch@xmath4 and hoco , while woon ( 2002 ) suggested that the addition reaction between radicals nh@xmath0ch@xmath0 and hoco could explain the formation of glycine , with these species produced in the ices by successive hydrogenation of hcn and by the addition of co and oh , respectively . experimental work by holtom et al .
( 2005 ) involving the electron bombardment of co@xmath0/nh@xmath0ch@xmath4 ices confirmed the viability of woon s mechanism , while work by elsila et al .
( 2007 ) , extending the investigation of bernstein et al .
( 2002 ) , used isotopic labeling to trace the origins of the constituent atoms in the product molecules .
their findings indicated that woon s mechanism was consistent with the origins of the nitrogen and central carbon in the glycine molecule , while the origin of the acidic carbon was more consistent with the formation of glycine occurring via a nitrile precursor , i.e. amino acetonitrile ( nh@xmath0ch@xmath0cn ) .
however , a strecker - type formation route was found to be of only minor importance . in order to investigate the formation of glycine within dust - grain ices under astrophysical conditions , the extension of the chemical networks of g&h and gwh as well as that of belloche et al .
( 2009 ) , who treated the formation of amino acetonitrile , @xmath9-propyl cyanide , and ethyl formate is continued with the addition of glycine , its amino aldehyde glycinal , and a collection of related species such as propionic acid and propanal .
the new formation mechanisms for these species follow the general treatments of previous networks , which allowed radicals on the grains , frequently produced by uv photodissociation or chemical abstraction processes , to meet via thermal diffusion and thus form more complex structures .
the mechanism suggested by woon ( 2002 ) is included in the new network , as well as several other likely radical - addition mechanisms .
furthermore , a gas - phase mechanism for glycine production is tested , by the inclusion of reactions between protonated hydroxylamine ( nh@xmath4oh@xmath10 ) and acetic acid ( ch@xmath4cooh ) , followed by dissociative electronic recombination , as suggested by blagojevic et al .
all new species in the network are afforded both formation and destruction mechanisms on the grains and in the gas - phase .
a more comprehensive hydrogen - abstraction reaction set is also included in the network . to apply this chemical network to hot - core conditions , a new three - phase astrochemical model , _ magickal _ , is used , which includes a fully - active chemistry within the bulk of the ice mantle ( see section 2.1 ) .
the physico - chemical evolution of the core is traced starting from the free - fall collapse of a cloud , using the dust - temperature / visual - extinction relationship given by g&p , through the subsequent warm - up of the dense core from 8 400 k. in order to simulate the emission from glycine and a selection of other molecules , the results of these models are mapped to existing observationally - determined temperature and density profiles for specific sources .
the equation of radiative transfer is then solved numerically , under the assumption of local thermodynamic equilibrium , along both on- and off - source lines of sight , allowing a determination of line strengths and emission radii .
this emission is then convolved with telescope beams appropriate to current instruments , allowing the detectability of glycine to be assessed .
the nearby hot - core source ngc 6334 irs1 is specifically considered , due both to the availability of physical profiles and molecular observations in the literature , and to its spectroscopic suitability .
comparison is also made with the recent non - detection of hydroxylamine ( nh@xmath0oh ) in another source ( pulliam et al .
2012 ) , with implications for the putative formation of glycine in the gas phase .
section 2 describes the new chemical and physical models ; the results are presented in section 3 .
the results of the inclusion of gas - phase formation mechanisms for glycine are described in section 4 .
the spectroscopic model and results for various molecules are presented in section 5 .
a broad discussion of the results is presented in section 6 ; a list of general conclusions is provided in section 7 .
to model the hot - core chemistry , a new astrochemical code is employed , named _ magickal _
( model for astrophysical gas and ice chemical kinetics and layering ) .
this code uses a rate - equation / modified rate - equation approach to solve the coupled gas - phase , dust grain - surface and ice - mantle chemistry .
the essential functionality of previous approaches ( garrod & pauly , 2011 ) is retained , while a number of additional features are added , as outlined below . the three - phase model used here is described in detail by garrod & pauly ( 2011 ) , and is based on the approach of hasegawa & herbst ( 1993 ) .
in contrast to both of those methods , _ magickal _ simulates an active ice - mantle chemistry in addition to that occurring in the surface layer of the ice or on the dust - grain surface .
previous _ two - phase _ hot - core models , such as those of g&h , gwh , and belloche et al .
( 2009 ) , have implicitly assumed a combined surface / mantle chemistry , using surface kinetics to simulate all dust - grain chemical processes , including those within the bulk ice .
although satisfactory results may be obtained through this simplification , it has become clear through experimental means ( e.g. berg et al . , 2009a ) that the promotion of complex - molecule formation within organic ices by uv photolysis involves sub - surface processes , even in relatively thin ices ( on the order of 10 monolayers ) .
in order to simulate chemical kinetics within the ice mantles , it is here assumed that the mobility of chemical reactants in the bulk ice occurs through a barrier - mediated thermal diffusion process .
other authors have assumed a swapping mechanism to be active in laboratory ice experiments ( berg et al . 2009 ; fayolle et al . 2011 ) , which is also implicitly assumed in this model .
the swapping rate is parameterized in the same way as the surface thermal hopping , i.e. @xmath11\ ] ] where @xmath12 is the characteristic vibrational frequency of an harmonic oscillator , @xmath13 is the dust temperature , and @xmath14 is an energy barrier associated with the swapping of species @xmath15 with an adjacent water molecule , in line with the adoption of surface binding energies appropriate to amorphous water ice , following g&h and gwh .
ratios between the surface diffusion barrier and the desorption energy of any particular species bound to an ice or any other surface are poorly constrained , and various authors have used values in the approximate range of 0.3 0.8 .
garrod & pauly ( 2011 ) found that the production of interstellar co@xmath0 ice could be well reproduced by the adoption of ratios @xmath16:@xmath17 ; here a value of 0.35 is used .
thence , a value of @xmath14:@xmath18 is assumed for the swapping barrier , on the simplistic consideration that a bulk - ice molecule is bound to approximately twice the number of binding partners as a surface molecule . in analogy to the surface chemical reaction rates , the rate of reaction between mantle - species a and b is given by : @xmath19 / n_{\mathrm{m}}\ ] ] where @xmath20 is the population of species @xmath15 in the ice mantle , and @xmath21 is the total population of the ice mantle , including all species .
as discussed by garrod & pauly ( 2011 ) , in the three - phase model , material in the surface layer is transferred to the bulk ice according to the net rate of deposition of material onto the grain from the gas phase , with a similar , alternative transfer occurring in reverse in the case of a net _ desorption _ of mantle species into the gas phase .
these transfers represent the covering or exposure of ice material as the boundary of the ice - surface layer is constantly re - defined within the model , rather than the physical transport of material .
however , the inclusion of bulk diffusion within the ice should allow a true exchange of chemical species between surface and mantle , through the aforementioned swapping mechanism .
this surface
mantle swapping is a pair - wise process , and therefore requires the construction of rates for each chemical species in the mantle and in the surface layer , which must collectively produce no net transfer between the two phases .
this zero net rate would be assured using an explicit pair - wise coupling of every species in the surface with every species in the grain mantle ; however , such an approach is found to be computationally exhaustive , and requires estimates of the pair - wise positions of the swapped species that a rate - based model can not provide , other than through purely statistical considerations . instead ,
the swapping rates from mantle to surface are calculated first , with rates analogous to that of equation ( 2 ) . the total rate in this direction is then matched with an identical total transfer rate from surface to mantle , apportioned to all the surface species according to their fractional contribution to the total surface - layer population .
thus : @xmath22 @xmath23 where @xmath24 is the rate of mantle - to - surface swapping of mantle species @xmath15 , @xmath25 is the rate of surface - to - mantle swapping of surface species @xmath15 , @xmath20 and @xmath26 are the mantle and surface populations , respectively , of species @xmath15 , and @xmath27 and @xmath28 are the total mantle and surface populations . the ratio @xmath29 in equation ( 3 ) is taken as unity when it exceeds this value , representing the case where some surface species reside on the dust - grain surface itself and have no underlying mantle .
because the model does not assign specific positions to each atom or molecule in the mantle , the rate of mantle - to - surface swapping , @xmath24 , is required to be representative not simply of the final diffusion step , but of the entire chain of diffusion steps by which a mantle species at arbitrary distance from the surface layer will reach the surface .
the use of @xmath30 in eq .
( 3 ) as the rate of a typical , or average , step in this process is accurate so long as the final diffusion step from mantle to surface is relatively fast , i.e. that intra - mantle diffusion is the rate - limiting step in the entire journey from mantle to surface . under the assumption that the ice is predominantly water ice ( see above )
, the final surface - to - mantle step will be fast for species whose binding energies are less than or similar to that of water .
for the small minority of species with significantly greater binding energies than this , for which there would thus be an associated endothermicity to the mantle surface swap with a water molecule ( see e.g. berg et al .
2009b ) , the large absolute size of the swapping barrier would render the rate of mantle - to - surface diffusion negligible compared to other processes , such as the mantle - to - surface transfer associated with net surface desorption .
these rates defined in eqs .
( 3 ) and ( 4 ) are also considered in the loss and gain rates used for the modified - rate calculations of garrod ( 2008 ) , which are applied in this model ( using `` method c '' ) .
garrod et al .
( 2009 ) showed that this method produces a very close approximation to the results of exact monte carlo methods , using a large chemical network under a range of physical conditions .
+ the chemical network used in this model is based on that of gwh and the subsequent additions of belloche et al .
( 2009 ) , which featured a treatment for nitriles , including amino acetonitrile ( nh@xmath0ch@xmath0cn ) .
these networks assumed grain - surface radical radical formation mechanisms for newly - added complex organic molecules , as well as destruction mechanisms such as grain - surface and gas - phase photodissociation ( by both the ambient and cosmic ray - induced uv fields ) , gas - phase ion - molecule reactions ( including protonation ) and the related electronic recombinations of the products , as well as accretion from the gas - phase onto the grain surfaces , and thermal and non - thermal desorption from the grains .
the initial elemental abundances used are shown in table [ tab - init ] .
@xcite constructed a network of reactions to treat the formation of _
trans_-methyl formate , the higher energy conformer of the commonly observed _
cis_-methyl formate ; however , that study found no significant effect on the abundances of species other than _ trans_-mf , so the omission of the related chemistry is not expected to have any effect on the models presented here .
following the same methodology , formation and destruction processes are included for glycine and a suite of related molecules , including glycinal , propanal and propionic acid .
radical reactions for a selection of these species are shown in table [ tab - surf ] .
whereas in the previous models , a single complex molecule was typically assumed to be the only possible product of the addition of key grain - surface radicals , here alternative product branches are included , for the case where such a reaction could also result in a pair of stable products , and where the reaction is also exothermic .
thus , for the reaction hco + ch@xmath4o @xmath31 hcooch@xmath4 , which was found by gwh to be the predominant formation route for methyl formate , alternative product branches of [ co + ch@xmath4oh ] and [ h@xmath0co + h@xmath0co ] are now also included .
the branching ratios are in general poorly defined , and are likely to be significantly different between the gas- and solid - phase processes ; statistical ratios are therefore adopted ( e.g. 1:1:1 for hco + ch@xmath4o ) .
in addition , the same such two - product reactions are now also included in the gas - phase chemistry , using measured rates where available ; otherwise , a uniform second - order rate coefficient of @xmath32 @xmath33 s@xmath34 is assumed .
such processes may become important in the late , high - temperature stage of hot - core evolution , during which the evaporation , protonation and subsequent electronic recombination may lead to significant quantities of molecular radicals being present in the gas phase .
however , the single - product branches are omitted from the gas - phase chemistry , because of their negligible rates , due to the absence of a grain surface to carry away excess energy and thereby stabilize the product .
following previous models , grain - surface hydrogen - abstraction reactions are added for reactions between the new complex species and the most significant grain - surface radicals such as h , oh , ch@xmath4 , ch@xmath0oh , ch@xmath4o and nh@xmath0 . table [ tab - act ] shows activation - energy barrier information used in the model , for a selection of ( primarily hydrogen - abstraction ) reactions occurring on the grains . for atomic - hydrogen reactions relating to methanol formation , for which more complex fits to theoretical rates are employed ,
the simpler rectangular - barrier parameter fits are quoted , as for all other barrier - mediated reactions .
the method of hasegawa et al .
( 1992 ) is used to calculate the tunneling and thermal rates associated with these parameters , adopting the faster of the two for any given temperature .
for the majority of reactions , only gas - phase data are available , in which case the activation energy derived from fits to the lowest - temperature data - sets available are adopted for the surface / mantle reactions , assuming a barrier width of 1 . in cases where no data exist ,
the activation energy , @xmath35 , is estimated using the evans - polanyi relationship , which states that the difference in @xmath35 between reactions of a similar type is proportional to the difference between the enthalpy changes associated with each reaction .
parameters for the evans - polanyi relationship are obtained from dean & bozzelli ( 1999 ) , or from fits using similar reactions , indicated in table [ tab - act ] . where the enthalpy change of reaction is unknown , the activation energy of a similar reaction is used directly .
there is evidence that the presence of a surrounding water - ice matrix may alter activation energy barriers for specific product branches ( e.g. goumans 2011 , in the case of formaldehyde - related chemistry ) , or lower the internal barriers to the structural re - arrangement of a product ( e.g. duvernay et al . 2005 , in the case of the isomerization of cyanamide ) .
it is therefore plausible that some of the gas - phase activation energies used in this model over - estimate the barriers in the solid phase ; however , such effects are difficult to quantify without a specific computational or experimental study of each solid - phase reaction and its low - temperature tunneling characteristics . for self - consistency , all reactions of surface radicals with complex species are now assumed to result in hydrogen abstraction , in contrast to gwh , who included a small selection of alternative substitution reactions for aldehyde group - bearing species .
these reactions typically provided only minor destruction or formation routes for the species involved , and their removal has no significant influence over the glycine chemistry described below .
all processes related to grain - surface chemistry outlined above are likewise incorporated into the chemistry of the bulk ice , with identical activation energies ( where present ) , while their reaction kinetics are instead defined by the bulk - ice diffusion parameters discussed in section 2.1 .
photodissociation of ice mantles is also included to all depths , without attenuation , in line with previous models .
this is a reasonable first - order approximation ; using the photon absorption cross - section of amorphous water ice for the first absorption band measured by mason et al .
( 2006 ) , andersson & van dishoeck ( 2008 ) calculate an absorption probability per photon of 0.007 per monolayer of water ice .
the maximum quantity of water ice formed on the grains in this model corresponds to @xmath1114 monolayers ; this would result in a uv field strength at the deepest ice layer of @xmath145% of the value at the surface .
this corresponds to a mean value of 69% of the assumed field strength , for the ice mantle as a whole .
however , the ices formed on interstellar dust - grain surfaces constitute a mixture of various molecular species , and the absorption cross - sections for many of these are currently unknown or poorly defined .
no desorption processes are allowed from the bulk ice ; thus desorption of these species must proceed via a prior transfer of material from mantle to ice surface .
diffusion competition is also included for reactions mediated by activation - energy barriers ( following g&p ) , for both surface and bulk - ice reactions ; this is perhaps the largest departure from the methods of gwh , as this change can produce reaction efficiencies many orders of magnitude different from the simple arrhenius treatment of previous models .
surface desorption processes included are thermal evaporation ( as defined by the binding / desorption energy , @xmath36 ) , reactive desorption as described by garrod et al .
( 2007 ) , and photo - desorption using measured rates determined by oberg et al .
( 2009c , d ) ; following g&p , all surface species are assumed to photodesorb with an efficiency of @xmath37 , in the case where values are not provided by those authors . the indirect mechanism for co@xmath0 formation proposed by g&p is also included for reactions both on the surfaces and within the ice mantles .
the reproduction of the high observed abundance ( @xmath1@xmath38 with respect to h@xmath0 ) of methyl formate ( hcooch@xmath4 ) in hot cores has proved somewhat problematic in previous astrochemical models .
garrod & herbst ( 2006 ) showed that grain - surface formation mechanisms could produce adequate quantities of methyl formate ( mf ) to explain observations ; however , in some cases it was found ( garrod et al .
2008 ; belloche et al .
2009 ) that the evaporation of mf at relatively low temperatures could result ( dependent on the warm - up timescale ) in only a small fraction of this material surviving the transition to the gas - phase .
belloche et al .
( 2009 ) adopted a higher binding energy for mf ( 5200 k vs. 4100 k ) , citing the importance of hydrogen bonding in the estimation of this value ; they found that this value produced results appropriate to observational abundance values for methyl formate .
this value is also adopted here .
table [ tab - bind ] shows a selection of other binding energies used in the present model . as found by gwh and laas et al .
( 2011 ) , the rates and branching ratios of methanol photodissociation in the ice mantles are crucial to the production rates and ratios of various complex organic molecules detected in hot cores , as the ice - mantle methanol provides a large supply of structurally - complex organic material .
the present model uses the methanol branching ratios derived from the experiments of oberg et al .
( 2009a ) for products [ ch@xmath0oh+h]:[ch@xmath4o+h]:[ch@xmath4+oh ] of 5:1:1 . in the absence of other information , these values are applied to both the direct uv photodissociation by the interstellar radiation field , and to photodissociation by the secondary uv photons produced by cosmic rays , the latter of which is most important at the high visual extinctions of the hot - core phase .
these ratios are applied throughout to the ice and gas phases , although , as found by laas et al .
( 2011 ) , the importance of gas - phase methanol photodissociation to complex molecule production is minimal . activation - energy treatments for co and h@xmath0co use best - fit expressions to uni - molecular and bi - molecular rates recently acquired through quantum transition state theory ( goumans , priv .
comm . ; andersson , priv . comm . ) .
hydrogen abstraction from h@xmath0co and ch@xmath4oh are included , mediated by activation energy barriers , while hydrogen abstraction from hco and ch@xmath4o / ch@xmath0oh are also allowed ( producing h@xmath0 + co / h@xmath0co ) , assuming branching ratios of 1:1 with the associated hydrogen - addition channels ( i.e. no activation energy ) .
a number of potential routes have been suggested in the literature for the formation of glycine in interstellar ices , based on laboratory evidence or theoretical considerations .
experiments by bernstein et al .
( 2002 ) and muoz caro et al .
( 2002 ) showed that uv irradiation of interstellar ice analogs containing h@xmath0o , ch@xmath4oh , co , co@xmath0 , nh@xmath4 and/or hcn could produce significant quantities of amino acids , including glycine .
sorrell ( 2001 ) had suggested a route involving the reaction of the nh@xmath0 radical with acetic acid ( ch@xmath4cooh ) , itself formed through the addition of ch@xmath4 and hoco , while woon ( 2002 ) suggested that the addition reaction between radicals nh@xmath0ch@xmath0 and hoco could explain the formation of glycine , with these species produced in the ices by successive hydrogenation of hcn and by the addition of co to oh , respectively . experimental work by holtom et al .
( 2005 ) involving the electron bombardment of co@xmath0/nh@xmath0ch@xmath4 ices confirmed the viability of woon s mechanism .
elsila et al .
( 2007 ) found woon s mechanism to be consistent with the origins of the nitrogen and central carbon in the glycine molecule , while the origin of the acidic carbon was more consistent with the formation of glycine occurring via a nitrile precursor , i.e. amino acetonitrile ( nh@xmath0ch@xmath0cn ) . here ,
we include a number of new radical - addition reactions relating to glycine and similar species .
each hydrogenation step suggested by woon ( 2002 ) is included , although all but the final addition of nh@xmath0ch@xmath0 + hoco was already present in the network .
however , other routes are also included , such as the addition reaction nh@xmath0ch@xmath0 + hco @xmath31 nh@xmath0ch@xmath0cho ( glycinal ) , whose product may ( for example ) be photodissociated to produce the radical nh@xmath0ch@xmath0co , which may further react with oh to produce glycine . there are four final processes that lead to glycine in this network , @xmath39 although there are several ways in which each radical may come to be formed , either through chemical reaction or by photodissociation of a stable molecule .
reaction ( 5 ) is also complemented by exothermic radical - radical reactions with species ( e.g. hco ) that may donate a hydrogen atom to form two stable molecules , as described in section 2.2 .
reaction ( 5 ) typically follows the photodissociation of , or h - abstraction from , glycine itself ; reactions ( 6)-(8 ) are thus the major potential formation routes .
woon suggests the formation of hoco by the reaction of co and oh .
this process is difficult to model accurately , due to the complex nature of the transition states ; the reaction may also lead to the formation of co@xmath0 + h ( see g&p , and references therein ) , or may dissociate back to co + oh , dependent on the rapidity of energy loss to the surface .
as such , the oh + co reaction is modeled separately for each set of products , assuming an activation energy barrier of 2500 k for the hoco branch , following the approach of gwh . in practice , the formation of hoco in the model is dominated by the photodissociation of or chemical hydrogen abstraction from formic acid ( hcooh ) in the ice , which is formed predominantly by hco + oh additions ( see gwh ) .
also included in the model is a selection of species similar to glycine , but with alternative functional groups .
for example , glycinal ( nh@xmath0ch@xmath0cho ) and propionic acid ( c@xmath0h@xmath40cooh ) are modeled , through analogous reactions to those of glycine , while new dust - grain formation routes are added for propionaldehyde ( or propanal , c@xmath0h@xmath40cho ) , which has been detected toward sgr b2(n ) ( hollis et al . 2004 ) .
the inclusion of these other species ensures that the importance of reactions leading to glycine formation is not over - estimated due to the omission of competing reaction pathways . as for other complex species ,
the chemical network includes gas - phase destruction routes for glycine that consist of uv photodissociation , either by the cosmic ray - induced or standard interstellar radiation field , or by reactions with atomic or simple molecular ions .
table [ tab - gas ] shows the newly - included reactions for glycine and its related gas - phase products .
electronic recombination of protonated glycine results in the re - formation of glycine with an efficiency of 5 % , but all the glycine in the gas - phase is originally formed on the grains .
consideration of a potential gas - phase formation route is given in the models of section 5 .
the physical model used here broadly follows that of g&h and later papers , in which a collapse phase is followed by a static warm - up phase . in the first phase , the density increases from @xmath41 to @xmath42 @xmath43 , under the free - fall collapse expression given by nejad et al .
however , in contrast to previous hot - core models , while the gas temperature is held at a constant 10 k , the temperature of the dust varies with visual extinction , according to equation ( 17 ) of g&p , which describes the heating of the dust by the interstellar radiation field , assuming a dust - grain radius of 0.1 @xmath44 m . at the initial visual extinction of @xmath45 ,
the dust temperature is @xmath116 k ; during collapse , this falls to a minimum temperature of 8 k. during the subsequent warm - up phase , the core is heated from 8 400 k , with density held at @xmath42 @xmath43 .
three warm - up timescales are adopted , which are the same as previous models , in so far as they reach 200 k at the same point in time ( table [ tab - mod ] ) . however
, the warm - up is extended beyond this time , until 400 k is reached , so that the later evaporation and gas - phase behavior of more strongly - bound species , such as glycine , may be studied over a longer period . during the warm - up phase , gas and dust temperatures are assumed to be well coupled due to the high density ; the gas kinetic temperature follows that of the dust grains when @xmath4610 k. table [ tab - mod ] shows the details of the three warm - up phase physical models .
table [ tab - ice ] shows the composition of the dust - grain ices at the end of the collapse phase ( corresponding to the starting values for each of the following warm - up models ) , as well as observational values for comparison .
figures 1 , 2 and 3 show the behavior of fractional abundances with respect to total hydrogen ( @xmath47 ) of a selection of simple and complex molecules , for the _ fast _ , _ medium _ and _ slow _ warm - up models , respectively .
peak gas - phase abundances are shown in table [ tab - output ] , along with the corresponding system temperatures for each peak value , which frequently correspond to the temperature of peak evaporation for species formed on the dust grains . in figs
. 1 3 , panel ( a ) shows some of the main ice constituents and a representative ion , hco@xmath10 .
panels ( b ) and ( c ) show some complex organics typically observed toward hot cores .
panel ( d ) shows some commonly observed nitrogen - bearing species as well as the as - yet undetected hydroxylamine ( nh@xmath0oh ) .
panel ( e ) shows a selection of complex species containing a carbonyl group ( c = o ) , including acetone , ( ch@xmath4)@xmath0co .
panel ( f ) shows complex species derived from the larger photo - fragments of methanol .
panel ( g ) shows a selection of complex organics bearing an amine group ( nh@xmath0 ) , while panel ( h ) shows glycine itself ( nh@xmath0ch@xmath0cooh ) along with several similar or related molecules .
not all species shown in table [ tab - output ] are plotted in figs .
the results for a selection of important species and their progenitors are discussed in detail below . due to the large number of molecules included in the model , the results for many molecules are presented in figs .
1 3 and table [ tab - output ] without comment . for a description of the most important reaction mechanisms relating to these species ,
the reader is referred to the analysis presented by garrod et al .
( 2008 ) ; however , some general differences between the current and previous models are discussed in section 6 .
glycinal ( nh@xmath0ch@xmath0cho ) is the amino aldehyde corresponding to the amino acid glycine .
its formation in these chemical models , shown in panel ( h ) of figs . 1
3 , begins at around 25 k , when the radical hco becomes mobile and reacts with ch@xmath0nh@xmath0 .
the latter species is formed by the hydrogenation of ch@xmath0nh , which builds up at around 15 k via the addition of nh and ch@xmath0 radicals . the hco radical is mainly formed by the abstraction of hydrogen from formaldehyde ( h@xmath0co ) by oh , which is a barrierless process ; this occurs until formaldehyde evaporates at around 40 k ( see g&h , gwh , and section 3.2 , below ) .
a later and somewhat stronger glycinal formation route operates at @xmath140 k , when nh@xmath0 radicals add to ch@xmath0cho radicals formed by the photodissociation of acetaldehyde ( ch@xmath4cho ) .
glycinal formation thus occurs at rather lower temperatures than that of glycine ( see below ) .
however , both are at least partially dependent on the earlier formation of acetaldehyde at lower temperatures to provide the ch@xmath0co structure present in both .
peak gas - phase glycinal abundances are a few @xmath48 to @xmath38 with respect to total hydrogen ; it is thus of similar abundance to the more prevalent observed organic molecules in hot cores .
no mm / sub - mm spectral information is apparently available for glycinal at present .
glycine formed on the grains is seen to be released into the gas phase at a temperature of around 200 k , attaining a peak gas - phase abundance of @xmath49 , @xmath50 , and @xmath51 with respect to total hydrogen ( see table [ tab - output ] and figs .
13 panel h ) , for the _ fast _ , _ medium _ , and _ slow _ warm - up timescales , respectively .
the three main production routes included for glycine ( eqs .
8) are all active , and all contribute a proportion of the glycine later released into the gas phase .
significant dust - grain surface / mantle formation of glycine begins at temperatures of 40 50 k , and may continue up to around 120 k. for the _ medium _ warm - up timescale case , at around 40 k the oh radicals in the ice , produced by the photodissociation of water molecules , are sufficiently mobile to find glycinal ( nh@xmath0ch@xmath0cho ) , which is abundant at around @xmath52 with respect to total hydrogen .
the oh radical easily abstracts a hydrogen atom from glycinal ; the activation energy barrier is sufficiently low ( 591 k ) compared to the diffusion barrier for oh that the reaction occurs before oh has an opportunity to diffuse away from its reaction partner .
( this effect indeed confers an efficiency close to unity on most such oh - initiated hydrogen - abstraction reactions ) . the resulting radical , nh@xmath0ch@xmath0co , then reacts with another oh to produce glycine ( eq .
8) . at temperatures of 55 k and above ,
much greater quantities of glycine are formed ; hydrogen abstraction from acetic acid ( ch@xmath4cooh ) by nh@xmath0 radicals within the ice mantles and the subsequent addition of the resultant ch@xmath0cooh radicals to nh@xmath0 produces glycine ( eq .
the nh@xmath0 radicals are primarily formed by hydrogen abstraction from ammonia ( nh@xmath4 ) by oh radicals , which are themselves formed by the photodissociation of water ice . the increased importance to glycine formation of the acetic acid - related path at these temperatures
is caused by the greater abundance of nh@xmath0 radicals that results from the evaporation of hnco molecules from the ices , which up until that time represent a major reaction partner for nh@xmath0 .
the release of hnco into the gas phase may be seen in figs
. 1 3 , panel ( d ) . however , over the 55 75 k range , eq .
( 8) still dominates eq .
( 6 ) by a factor of 2 3 , as the greater availability of nh@xmath0 allows the nh@xmath0ch@xmath0co radical to be formed by hydrogen abstraction from glycinal by nh@xmath0 as well as oh . at around 75 k , an ice _ surface _ ( as opposed to _ mantle _ ) process becomes the strongest contributor to glycine formation ; hydrogen abstraction from ch@xmath4nh@xmath0 ( methylamine ) by oh radicals produces ch@xmath0nh@xmath0 , while abstraction from hcooh ( formic acid ) by oh produces hoco .
these react to produce glycine ( eq .
7 ) on the ice surface , by the route suggested by woon ( 2002 ) .
this mechanism then competes roughly equally with the reaction of ch@xmath0cooh and nh@xmath0 on the ice surface ( eq .
6 ) to be the main producer of glycine , up to @xmath190 k. by this point , most glycine formation is taking place on the surfaces of the ices , rather than within the ice mantles .
this is caused by the increasing rate of mantle - to - surface diffusion among less strongly - bound species , including nh@xmath0 , as the temperature rises , coupled with their commensurately faster reaction rates on the ice surface .
the more strongly - bound products of these reactions , such as glycine itself , are `` swapped '' back into the deeper mantle following their formation , as they are exchanged with more mobile , volatile species . between around 90 120 k , the reaction of nh@xmath0 and ch@xmath0cooh ( eq .
6 ) again takes over as the dominant route to glycine on the grains , as ch@xmath4nh@xmath0 is gradually lost to the gas phase , rendering eq .
( 7 ) far less effective .
the majority of the glycine ultimately released into the gas phase is formed during the 90 100 k period . during this time , eq .
( 6 ) dominates eq .
( 7 ) by a factor of @xmath12 . at around 120 k ( in this and all models ) , ammonia ( nh@xmath4 ) , formic acid ( hcooh ) , and acetic acid ( ch@xmath4cooh )
evaporate fully from the grains , closing down the main glycine formation mechanisms . from this point on
, glycine begins to migrate from the bulk ice mantle into the surface , while suffering a moderate degree of attrition by hydrogen abstraction and photodissociation the products of which typically evaporate .
for the _ fast _ warm - up model , the temperature dependence of the formation mechanisms is broadly similar ; however , the importance of each route is different , due to the differing quantities of precursor molecules available .
most importantly , the abundance of acetic acid ( ch@xmath4cooh ) , shown in panel ( e ) , is around one order of magnitude lower than in the _ medium _ timescale model . as found by garrod et al .
( 2008 ) , in these models acetic acid is formed predominantly by the addition of oh to ch@xmath4co , the product of the abstraction of hydrogen from acetaldehyde ( ch@xmath4cho ) by oh itself .
the _ fast _ warm - up model allows a shorter time period for this process , before acetaldehyde evaporates from the grains , resulting in the lower acetic acid abundance .
this in turn renders the production of glycine via eq .
( 6 ) generally a factor of a few slower than eqs . ( 7 ) & ( 8) , which , overall , provide approximately equal amounts of glycine production in this model . for the _ slow _ warm - up model , the behavior of glycine production is much more similar to that of the _ medium _ case ; however , here the nh@xmath0 + ch@xmath0cooh route is strongly dominant over the 60 120 k range during which glycine production is important .
this is caused by the greater time period for the production of acetic acid .
production via eq .
( 8) is also less significant than in the two shorter - timescale models , due to the lower abundance of glycinal in the ices . in all models , the gas - phase destruction of glycine
is entirely dominated by protonation and dissociative electronic recombination , as is the case for the majority of complex organics ( see also gwh ) ; photodissociation has only a small effect , as the only uv radiation field capable of influencing the chemistry during the warm - up phase is that induced by cosmic rays .
these models produce a low to moderate abundance of glycine in the gas phase , in comparison with other complex species produced in the models .
the peak gas - phase abundance of glycine is also very similar to that of acetic acid in each model , albeit with a different temperature of evaporation .
the question of whether this glycine may be detectable is addressed in section 5 .
amino acetonitrile ( nh@xmath0ch@xmath0cn ) , or aan , shown in panel ( h ) , has been suggested as a precursor to glycine ( elsila et al . 2007 ) .
it is formed in these models with a peak gas - phase fractional abundance of a few @xmath48 to @xmath38 , consistently for each timescale model .
this is around a factor of two higher than the values produced in the model of belloche et al .
( 2009 ) , and around a factor of 10 greater than the value derived from observations toward sgr b2(n ) by belloche et al .
( 2008 ) . in these models
, aan is formed primarily by the addition of nh@xmath0 to the radical ch@xmath0cn , which itself derives from hydrogen abstraction from ch@xmath4cn molecules by nh@xmath0 at around 60 80 k , producing an abundance of around @xmath48 on the grains .
later , at around 90 k , the evaporation of ch@xmath4cn from the grains results in its gas - phase protonation and recombination with electrons , producing ch@xmath0cn with an efficiency of 50 % .
this radical re - accretes onto the grains , to react with nh@xmath0 , which produces an aan fractional abundance of around @xmath38 on the grains .
however , in light of recent measurements of methanol recombination branching ratios ( e.g. geppert et al .
2006 ) , it is unclear whether such a high efficiency of ch@xmath0cn production by gas - phase electronic recombination is realistic .
a value of 5 % , as adopted in these models for the recombination of more complex species , would render the earlier formation of aan by purely grain - surface chemistry the dominant mechanism , and would produce abundances close to the values observed toward sgr b2(n ) .
propionaldehyde ( or propanal , c@xmath0h@xmath40cho ) and propionic acid ( or propanoic acid , c@xmath0h@xmath40cooh ) are structurally similar to glycinal and glycine , and neither have been treated in chemical models previously .
propionaldehyde is formed in great abundance at low temperatures , by the addition of hco and c@xmath0h@xmath40 radicals on the dust - grain ice surfaces ; abundances as high as @xmath53 with respect to total hydrogen are formed in the _ slow _ warm - up model .
production is most rapid at around 30 k , when the evaporation of grain - surface methane ( ch@xmath54 , panel a ) is fastest .
this results in a significant increase in the abundance of gas - phase acetylene ( c@xmath0h@xmath0 ) , which accretes onto the dust grains , where it is rapidly hydrogenated to c@xmath0h@xmath40 , thence reacting to form propionaldehyde .
formation on the ice surface at this time is sufficiently rapid that so - called reactive desorption , whereby the energy released from the exothermic formation of a surface molecule results in desorption ( e.g. garrod et al . 2007 ) , produces a gas - phase peak in the abundance of propionaldehyde of approximately @xmath55 @xmath47 .
this period of carbon - rich gas - phase chemistry associated with methane evaporation is often labeled warm carbon - chain chemistry ( wccc ) , and has been identified observationally by sakai et al .
( 2008 ) ; however , it is unlikely that the enhancement in propionaldehyde seen in the models would be observable at the low abundances predicted by these models .
the large quantity that remains on the grain surfaces later evaporates at around 120 130 k , producing gas - phase peak fractional abundances of around @xmath38 .
this molecule was detected by hollis et al .
( 2004 ) toward sgr b2(n ) ; however , no specific abundance was determined .
propionic acid ( c@xmath0h@xmath40cooh ) is formed on the grains much later than propionaldehyde ; most production occurs from around 70 k onwards , by the addition of oh and c@xmath0h@xmath40co radicals on the ice surface .
the latter species is formed primarily through hydrogen abstraction from propionaldehyde by nh@xmath0 in the ice .
thus , propionic acid production appears to be directly dependent on the prior formation of its associated aldehyde .
methanol ( ch@xmath4oh ) and formaldehyde ( h@xmath0co ) , shown in figs .
13 panel ( b ) , are both present in the dust - grain ice mantles in large quantities , following their formation at low temperature during the collapse phase ( see table [ tab - ice ] ) .
formaldehyde remains the primary source of hco radicals in the ices , due to hydrogen abstraction by oh , until it evaporates at around 40 k , producing a significant gas - phase peak
see g&h and gwh .
this feeds the gas - phase formation of other species , most notably formic acid and methanol .
methanol abundance in the ice is essentially static until it eventually evaporates at around 120 k. but methanol nevertheless provides the majority of organic structure for the formation of more complex molecules , in the form of methyl ( ch@xmath4 ) , methoxy ( ch@xmath4o ) and hydroxymethyl groups ( ch@xmath0oh ) .
these are formed primarily by photodissociation caused by the cosmic ray - induced uv field that penetrates the hot core . above @xmath145 k ,
oh radicals become mobile , and hydrogen abstraction to form ch@xmath0oh becomes the dominant destruction mechanism for methanol , producing around twice as much ch@xmath0oh as the photodissociation route .
the abundance of methanol in the ices is little affected by the warm - up timescales of each model .
the fall in the gas - phase abundances is more significant , but the apparent differences between each timescale model are mainly due to the longer timescales over which the chemistry of the slower warm - up models is simulated , rather than to a difference in the rate of destruction .
the temperature abundance relationships may , however , be important to the interpretation of observationally - determined abundances and rotational temperatures .
the apparently precipitous fall in the gas - phase abundances of various other molecules in the _ slow _ warm - up model is also largely attributable to the longer timescales adopted in that model .
the eventual evaporation of methanol also indirectly produces a high - temperature peak in formaldehyde abundance ; the reaction of methanol with protonated methanol produces protonated dimethyl ether ( ch@xmath4och@xmath54@xmath10 ) , one of whose electronic recombination products is assumed to be formaldehyde in this model .
the branching products of this process are not well constrained , _ pace _ hamberg et al .
( 2010 ) , who indeed do not identify h@xmath0co as a likely product .
the secondary peak in formaldehyde abundance may at least be considered unreliable , due to these uncertainties . the peak abundance of methyl formate ( mf , hcooch@xmath4 ) , shown in figs .
13 panel ( b ) , is in line with that expected from observations ( around @xmath56 ) , except for the _ slow _ warm - up model .
the evaporation of methyl formate prior to that of methanol and various other species leaves it vulnerable to destructive reactions with molecular ions , whose abundance would otherwise be lessened by the presence of other reaction partners .
the long timescale between methyl formate and methanol evaporation in the _ slow _ model therefore leads to significant methyl formate destruction ; a similar result was noted by gwh and belloche et al .
( 2009 ) , in the case of a lower assumed mf binding energy .
methyl formate is formed entirely on the dust grains in this model , by the reaction hco + ch@xmath4o .
production is mainly within the ice mantles , peaking at 25 k , although significant production also occurs on the ice surface .
glycolaldehyde ( ch@xmath0(oh)cho ) , shown in figs .
13 panel ( c ) , is formed similarly to its a structural isomer methyl formate , via hco + ch@xmath0oh .
its gas - phase and ice abundance is greater than that of mf , due to the more rapid production of ch@xmath0oh over ch@xmath4o by a factor of 5 ( for photodissociation ) in this model .
observations of protostellar envelopes , however , do not demonstrate this large abundance of glycolaldehyde ; the recent detection by jrgensen et al .
( 2012 ) toward the class 0 protostellar binary iras 16293 - 2422 , using the alma telescope , indicated a hcooch@xmath4:ch@xmath0(oh)cho ratio of 10 15 ( although the observational glycolaldehyde rotational temperature of 200 300 k _ is _ consistent with the present model results ) .
the discrepancy may be related to the assumed branching ratios of the reactions between hco and either ch@xmath4o or ch@xmath0oh : @xmath57 and @xmath58 ( the enthalpy change associated with each reaction , @xmath59 , is indicated in brackets ) .
the branching of each set of reactions is assumed to be [ 1:1:1 ] in these models .
each branch of equations ( 9 ) and ( 10 ) is exothermic , and branches ( a ) and ( b ) would require minimal internal re - arrangement .
the single - product routes in each case are the most exothermic ; however , in the case of glycolaldehyde formation , the enthalpy change related to reaction branch ( 10b ) is close to that of ( 10a ) . while the assumed efficiency of branch ( c ) ( i.e parity with the other two branches ) may be over - estimated in both equations ( 9 ) and ( 10 )
, the production of co and ch@xmath4oh ( 10b ) may be very competitive with the formation of glycolaldehyde ( 10a ) , while it may be less so in the case of methyl formate .
it thus appears plausible , by this simplistic analysis , that a difference in the efficiencies of formation of mf and glycolaldehyde via their primary grain - surface / ice - mantle formation routes could account for the order - of - magnitude discrepancy in observational abundances .
however , a detailed theoretical determination of the branching behavior of these reactions in the solid phase would be necessary to confirm this theory .
alternative branching ratios in the photodissociation of methanol , to produce a significant excess of ch@xmath4o over ch@xmath0oh is also possible , although the current and only estimate of these values for solid - state methanol , based on experimental rate - fitting , suggests precisely the opposite relationship ( ch@xmath4o : ch@xmath0oh of 1:5 ; berg et al .
. a decrease in the efficiency of glycolaldehyde production as described above would be unlikely to affect significantly the abundances of other _ observed _ species , although the production of postulated molecules related to glycolaldehyde , such as dihydroxyacetone ( hoch@xmath0coch@xmath0oh ) could be affected .
the destruction rates of the hco and ch@xmath0oh radicals would remain the same , while the increased re - formation of co and ch@xmath4oh resulting from the alternative product branch would have only a minor effect on the already large abundances of those species .
the glycine formation routes included in this model are sufficiently independent from the glycolaldehyde chemistry that glycine production would also be unaffected by the adoption of an alternative efficiency for glycolaldehyde formation . similarly , while acetic acid ( ch@xmath4cooh ) is a structural isomer of both glycolaldehyde and methyl formate , its different structure means that its formation routes are not directly related to those of its isomers ; the choice of treatment for glycolaldehyde would there have little effect on its own abundance or , through it , that of glycine .
the chemical behavior of acetic acid is discussed above , in section 3.1.1 .
formic acid ( hcooh ) , shown in figs .
13 panel ( b ) , is formed to a large degree in the gas phase , especially for the _ fast _ and _ medium _ warm - up models ; this occurs when formaldehyde ( h@xmath0co ) evaporates at @xmath140 k , reacting with oh to form hcooh and atomic hydrogen .
the gas - phase formic acid accretes back onto the grains , and does not evaporate until a temperature of around 120 k is reached . a similar or somewhat smaller quantity of the formic acid that ultimately evaporates from the grains is formed at low temperature within the ice mantles by the addition of hco and oh , as hco becomes mobile . as such
, it is unclear whether the formic acid observed toward hot cores originates in the gas phase or on the grains . as discussed above for the case of methyl formate and glycolaldehyde production ,
a plausible alternative product branch for the reaction of hco and oh exists , yielding co and h@xmath0o ( rather than hcooh ) ; indeed , this branch is somewhat more exothermic than the single - product branch .
both branches were included in the model , at a 1:1 efficiency ratio .
the true efficiency for hcooh formation could therefore be twice as large , or significantly smaller than assumed .
the rate of the gas - phase reaction ( oh + h@xmath0co @xmath31 hcooh + h ) is reasonably well known , albeit only around room temperature ( yetter et al .
1989 , and references therein ) ; but the amount of formic acid formed as a result is strongly dependent on the amount of formaldehyde that is released into the gas - phase from the grains . the behavior of the amino aldehyde formamide ( nh@xmath0cho ) , shown in figs .
13 panel ( d ) , is somewhat different to that presented in the models of gwh . in this case , the quantity formed in the ices , via the addition of hco and nh@xmath0 radicals , is somewhat larger than previously found .
however , its peak gas - phase abundance is lower , and falls over time , rather than rising as was found by gwh .
the latter change in behavior is due to the removal of a single gas - phase reaction from the network : nh@xmath0 + h@xmath0co @xmath31 nh@xmath0cho + h. this reaction was present in previous networks with a generic collisional rate coefficient of @xmath60 @xmath61 s@xmath34 , but it is likely to have an activation energy barrier ( based on the barrier of 5.89 kcal / mol ( @xmath13000 k ) calculated by li & lu ( 2002 ) for the more exothermic product branch nh@xmath4 + hco ) . the analogous reaction
oh + h@xmath0co @xmath31 hcooh + h , discussed above , has a measured rate of @xmath62 @xmath61 s@xmath34 , corresponding to a maximum 2% efficiency in the branching ratio compared to the alternative h@xmath0o + hco branch . a theoretical study ( li & lu , 2002 ) of the nh@xmath0 + h@xmath0co @xmath31 nh@xmath4 + hco reaction suggested a rate coefficient of @xmath63 @xmath61 s@xmath34 . for these reasons ,
the nh@xmath0 + h@xmath0co reaction was removed entirely .
while laboratory experiments have shown success in producing glycine and other amino acids by the photolysis of organic ice mixtures ( bernstein et al .
2002 ; muoz caro et al . 2002 ; holtom et al . 2005 ; elsila et al .
2007 ) , gas - phase mechanisms for amino - acid production have also been considered .
blagojevic et al .
( 2003 ) demonstrated that reaction between acetic acid ( ch@xmath4cooh ) and ionized or protonated hydroxylamine ( nh@xmath64oh@xmath10 ) can result in the formation of ionized or protonated glycine ( nh@xmath64ch@xmath0cooh@xmath10 ) and water .
similar schemes involving propionic ( propanoic ) acid were found to produce the equivalent forms of alanine .
the authors estimated rate coefficients of @xmath65 @xmath33 s@xmath34 for each of these processes .
the most probable source of nh@xmath64oh@xmath10 ions in hot cores would be nh@xmath0oh , either by photoionization , electron transfer with atomic ions such as he@xmath10 and c@xmath10 , or protonation by proton donors such as h@xmath4@xmath10 , h@xmath4o@xmath10 or hco@xmath10 , all of which processes are already present in the chemical network .
protonation by ch@xmath40@xmath10 , as suggested by snow et al .
( 2007 ) , is not included ; however , h@xmath4@xmath10 and h@xmath4o@xmath10 are around 2 orders of magnitude more abundant than ch@xmath40@xmath10 in the models , while the protonation rate coefficients in each case are typically very similar .
protonation rates of hydroxylamine by hco@xmath10 , h@xmath4@xmath10 and h@xmath4o@xmath10 of @xmath66 , @xmath67 and @xmath68 @xmath33 s@xmath34 , respectively , are used in the model .
electronic recombination of both ionic and protonated glycine were already added with the rest of the glycine - related chemistry ; ionization and protonation , with subsequent electronic recombination , are the primary destruction routes for glycine in all the models .
glycine is assumed to form in 5 % of recombinations of protonated glycine ( in line with values adopted for other species and with recent experimental determinations for protonated methanol and dimethyl ether ) , with a generic _ total _ recombination rate coefficient of @xmath69 @xmath33 s@xmath34 .
the same rate is assumed for the recombination of ionized glycine ; the ejection of a hydrogen atom is again assumed to occur for 5 % of reactions , producing the radical hnch@xmath0cooh .
this may accrete onto the grains to be directly hydrogenated , or may react with a selection of hydrogen - bearing gas - phase species , such as hco , to produce glycine , as outlined in section 2.2 ( although the influence of these routes on glycine abundance is neglible ) .
various other gas - phase and grain - surface destruction mechanisms are also available for the hnch@xmath0cooh radical , including photodissociation and further ion - molecule destruction processes .
formation of both hydroxylamine ( nh@xmath0oh ) and acetic acid ( ch@xmath4cooh ) occurs on the dust - grains in this model , and each reaches peak gas - phase abundance at around 130 140 k , following evaporation . in order to test the gas - phase production of glycine from these species ,
both of the following reactions are added to the network , assuming a rate constant of @xmath32 @xmath33 s@xmath34 for each : @xmath70 neither of these processes was included in the models already analyzed in section 3 . with the simple addition of reactions ( 11 ) , the results are unchanged ;
the contribution of the gas - phase processes is never remotely significant compared to those on the dust - grains , in spite of the availability of both reactants at similar times / temperatures .
it is noteworthy that the abundances of nh@xmath0oh acheived in the current models are around 3 orders of magnitude lower than those calculated by gwh ; the reason for this is that the major formation route for nh@xmath0oh on the grains is the addition of surface / ice - mantle nh@xmath0 and oh radicals .
the availability of both these reactants is significantly curtailed in the present model by the inclusion of new hydrogen - abstraction reactions for these species ( see table [ tab - act ] ) and by the more accurate consideration of the kinetics of grain - surface reactions that are mediated by activation energy barriers ( see g&p ) ; each reactant is more readily converted back to h@xmath0o or nh@xmath4 before the oh and nh@xmath0 radicals are able to meet . a recent observational search by pulliam et al .
( 2012 ) , using the nrao 12 m telescope , detected no nh@xmath0oh toward a range of previously - observed sources , with fractional abundance upper limits of @xmath71 toward sgr b2(n ) .
the authors noted that emission from a region more compact than 5 arcsec would not be detectable with their instrument , while the model suggests that the nh@xmath0oh would be at least as compact as typical hot - core molecules such as methyl formate or ethanol , based on evaporation temperatures .
nevertheless , the new model results for nh@xmath0oh are more in line with the observational data .
however , a recent laboratory study suggests that the formation of nh@xmath0oh by successive grain - surface hydrogenation of no is extremely efficient ( congiu et al .
2012 ) . while hydrogenation of no and hnoh are already present in the network , the reaction of h with hno has until now been assumed to lead to h - abstraction , with a barrier of 1500 k. to increase the effect of the glycine - forming gas - phase reactions , a parallel branch is added to the surface / ice reaction set : h + hno @xmath31 hnoh .
a barrier of 1500 k is also assumed ; but at low temperatures , the reaction competes effectively with hydrogen diffusion , producing a reaction efficiency that is close to unity . as a result ,
large amounts of nh@xmath0oh are formed , in this case during the cold collapse phase , rather than the warm - up phase .
a total fractional abundance of @xmath72 is present in the ice by the end of collapse , i.e. as the starting point for the warm - up phase ( see table [ tab - ice ] ) .
this agrees well with the cold - cloud models presented by congiu et al .
( 2012 ) . in order to further increase the influence of the gas - phase mechanisms ,
the rate coefficient of reaction ( 11a ) is increased to @xmath48 @xmath33 s@xmath34 , a plausible maximum value for this reaction .
the results of this optimized model are shown in fig .
[ fig - gas ] .
the large quantities of nh@xmath0oh present in the dust - grain ices alters the fractional abundances of various species in the model by a small degree , although with a negligible direct effect on the production of glycine within the ice .
the abundance of glycine is in fact slightly reduced , as a result of the somewhat lower quantity of acetic acid present in the ice .
the destruction of gas - phase acetic acid is rather faster in this model , due to reaction ( 11a ) .
however , when compared to an identical model with reaction ( 11a ) switched off ( not shown ) , the contribution of the gas - phase reaction to the total glycine formed in this model is around 2 % of the total formed on the grains .
some of the glycine that is formed in the gas phase is re - accreted onto the grains , to evaporate later with the rest , while the remainder is destroyed by ion
molecule reactions .
the immediate impact on gas - phase glycine abundance by the newly - included mechanisms may be seen in the small bump at around 130 k ; less than @xmath73 is produced .
it may be noted that , if no glycine - production mechanisms of any kind were active in the ices , and all production thus depended on the gas - phase route , two abundance peaks would likely occur , at 130 k and 200 k , corresponding to peak production and peak evaporation , respectively . on the the basis
that this scenario represents the optimal conditions for gas - phase glycine formation to occur in a hot core ( including the use of a very optimistic reaction rate ) , it appears that the particular gas - phase formation mechanism tested is unlikely to be significant , especially if the mechanisms for formation within the dust - grain ice mantles are in operation as suggested by the model .
the plausibility of such a large hydroxylamine abundance as would be required for significant glycine production is also discussed in section 5.3 .
in order to compare the model results more directly with observations of specific sources , a spectroscopic model is constructed that combines observationally - determined temperature and density profiles with the temperature - dependent chemical outputs of the model , to simulate emission from a selection of key molecules for which observational data exist toward the given source ; this simulated emission is then convolved to allow direct comparison with the observed spectral data .
the warm - up phase of the chemical model produces molecular fractional abundances for pre - defined output times , each corresponding to a temperature .
the resolution of these output times is chosen such that the associated temperature values are separated by less than 1% , over the 8 400 k range .
@xcite provide such physical profiles for the envelopes of 14 nearby high - mass protostellar sources .
using these temperature radius and density radius relations , a radius and density is assigned to each temperature value in the chemical model outputs .
the fractional abundance of each species in the model at each temperature / radius value is multiplied by its local density , to produce an absolute abundance value measured in @xmath43 .
the resulting molecular number - density and temperature profiles are then used to calculate the emission and absorption coefficients at each spherically - symmetric radial position in the core .
partition functions and emission - line data from the jpl line list and the cologne database for molecular spectroscopy are imported into the code as needed , using the ` splatalogue ' online database .
the spectral coefficients may be calculated by this method to incorporate all molecular emission at any given frequency , allowing line blends and line confusion to be taken into account .
gaussian line profiles are assumed , using observationally - determined line widths . a channel width ( frequency - bin size ) much smaller than the line width
is chosen ( in general , 0.1 mhz or @xmath10.08 km / s ) , to ensure that the lines are well resolved . under conditions of local thermodynamic equilibrium , the equation of radiative transfer
may be parameterized thus ( rohlfs & wilson 2006 ) : @xmath74 where @xmath75 is the specific intensity , @xmath76 is the optical depth at frequency @xmath77 , @xmath78 is temperature , @xmath79 is the line - of - sight distance , and @xmath80 is the planck function ( equal to @xmath81/@xmath82 , under lte ) . using the gridded , spherically - symmetric input data , this equation is integrated numerically along lines of sight ( using simpson s rule ) , beginning directly on - source , and moving outward through parallel lines of sight until a user - defined maximum offset is reached .
line - of - sight integrals of the absorption coefficient similarly yield the optical depths .
this procedure results in an intensity map of the hot core for each frequency channel . the final step is to convolve the emission with a gaussian beam centered on the source , using a beam - size appropriate to a chosen instrument observing at a given frequency . the intensity map is numerically integrated with the gaussian function , over the radius of the hot - core envelope , out to a few beam widths .
the technique automatically accounts for beam - dilution , and makes use of the precise emission structure to calculate an intensity value at each frequency .
the resulting simulated spectrum may be directly compared with main - beam temperature ( @xmath83 ) spectra obtained from the telescope for which the convolution has been produced .
the molecular mm / sub - mm line emission from a selection of the sources observed by @xcite was surveyed by @xcite using the james clerk maxwell and iram 30 m telescopes , providing information on dozens on complex organic species .
a member of this subset , the hot - core source ngc 6334 irs1 , is relatively closeby ( 1.7 kpc ) , and exhibits relatively narrow ( @xmath84 km / s ) , strong emission lines ( see van der tak et al . 2000 and bisschop et al . 2007 , and references therein ) .
it is also well situated in the sky ( dec .
( 1950 ) = -35@xmath8544 ) for future observation with the alma telescope ( latitude -23@xmath85 ) . for these reasons ,
it may be identified as a plausible candidate for the future detection of glycine .
the spectroscopic model described above is therefore used to assess the potential for the detection of glycine in this source , on the assumption that its formation and subsequent behavior are well described by the chemical model .
the physical profiles provided by van der tak et al .
( 2000 ) are used . in the case of glycine , only emission lines whose frequencies are within the observable ranges of the jcmt , iram30 m and alma telescopes are modeled .
[ fig - specs ] shows for illustrative purposes the predicted emission from this source over the 241.350 241.425 ghz range , using a channel spacing of 0.488 mhz .
these spectral calculations consider emission from all species in the model for which spectroscopic data exist , using only ground - state transitions .
glycine emission , at 241.373 ghz , is highlighted in red ; glycine conformer i is simulated , but not conformer ii , due to uncertainty in the partition functions .
the upper panel of the figure shows the _ unconvolved _ emission for an on - source line of sight the spectrum may be considered as that expected from a pencil beam of infinite resolution directed at the source .
individual lines are identified in the upper panel ; an identification is made where more than half of the emission within a channel derives from a single molecule .
the emission from glycine is seen to be slightly blended with emission from glycolaldehyde ( ch@xmath0(oh)cho ) in the line wings , but nevertheless shows a clear peak .
the middle panel shows the spectrum after convolution of the emission from the entire source ( same temperature scale ) using a beamwidth of 0.4 arcsec , which encompasses the strongest predicted emission region for glycine ( see below ) and which should be achievable with alma at these frequencies ( although the simulation is for a single - dish telescope , rather than an array ) .
the glycine peak is preserved , with a line strength on the order of 1 k. the lower panel of the figure shows the predicted spectrum ( different temperature scale ) using a beamwidth appropriate to the james clerk maxwell telescope at these frequencies ( @xmath110.4 arcsec ) ; the glycine emission line strength is on the order of 1 mk .
a selection of the strongest emission lines from methanol , methyl formate and glycine have been simulated using the method above .
methanol and methyl formate were both detected by bisschop et al .
( 2007 ) toward this source ; both are typically - observed hot - core species . in these calculations , unlike those illustrated in fig .
[ fig - specs ] , only the emission from an individual molecule is considered at the chosen line frequency ; this enables a more straight - forward comparison with the unblended lines observed by @xcite , and makes the calculations less computationally intensive .
furthermore , while line identifications may be made fairly easily in the unconvolved `` pencil - beam '' simulations , assessing the contributions made by each molecular line to the _ convolved _ emission spectra is much more time - consuming .
the consideration of the detectability of each glycine line in the following analysis therefore does not consider potential line blends simply the expected line strength .
more detailed assessments of the expected line emission from various species will be made in future work .
[ fig - profile ] shows the emission profiles predicted for the line center of a strong methyl formate line ( left panel ) and a strong glycine line of similar frequency ( right panel ; glycine emission is also shown in left panel for direct comparison ) .
the local emission strength ( i.e. assuming a pencil beam ) is shown in terms of brightness temperature for lines of sight targeted on radial displacements from the source position on the sky .
figure [ fig - map ] shows the same profiles mapped onto sky coordinates ; these emission maps are then convolved with the beam to produce simulated detected brightness temperatures . the most significant emission from methyl formate
is seen to emanate from a region of diameter @xmath17000 au , or 4 arcsec .
all potentially - detectable emission from glycine comes from a region @xmath10.8 arcsec in diameter , although the most significant emission region is of diameter @xmath10.4 arcsec , corresponding to a radius of @xmath11000 au . tables [ tab - rt - fast ] , [ tab - rt - mid ] and [ tab - rt - slow ] show details of the predicted emission from each molecule , using model outputs corresponding to the _ fast _ , _ medium _ and _ slow _ warm - up models , respectively .
line frequencies are shown in ghz , along with upper - level energies in k ; in some cases , each ` line ' represents two or more lines of identical or very similar frequencies and/or spectral characteristics .
the column marked ` peak simulated local intensity ' indicates the maximum local emission strength ( assuming a pencil beam ) .
beam widths represent the full - width half - maximum of the gaussian beam used ; the precise beam width appropriate to the jcmt or iram30 m is calculated by interpolation of telescope - specific data available online .
the convolved intensity of the line is the relevant quantity for discerning the detectability of a line ; however , for comparison with the observed lines , the integrated intensity ( in units of k km s@xmath34 ) provides a better test , as this quantity removes the dependence on the average line width ( @xmath86 ) ; this allows all of the emission predicted by the model to be compared against all of the emission detected in the observations .
the final column of these tables shows the integrated intensities measured for the methanol and methyl formate lines by bisschop et al .
( 2007 ) .
it may be seen immediately that the simulated integrated intensities of the methanol lines are an excellent match to the observed values , being generally around a factor of 2 lower , and showing only minor variation across the different warm - up timescale models .
the simulations also provide a good match regardless of the upper - level energies and beam sizes ( which vary according to frequency ) .
the methyl formate emission is much more variable between model timescales ; the _ fast _ model results provide an excellent match to all integrated intensities , falling within a factor less than 1.5 of the observed values .
the _ medium _ timescale model produces values that are around 2 3 times lower than those observed , similar to the agreement shown with the methanol lines .
however , the _ slow _ warm - up timescale results produce integrated intensities only one hundredth part as large as presented in the observations of bisschop et al .
( 2007 ) . in this case ,
methyl formate suffers rapid gas - phase destruction , as described in section 3.2.2 , producing much lower instantaneous fractional abundances and much faster decline over time / temperature .
the two faster warm - up models therefore appear the most favorable in comparison with observations .
the tables show predictions for six of the strongest glycine lines , or suites of lines .
each has an upper - level energy sufficiently low to be well - populated at the 200 + k temperatures at which glycine fractional abundances are strongest in the models .
the predicted intensities between different lines ( within each model ) are very consistent , and peak _ local _ intensities range from around 1 k to 10 k , corresponding to _ fast _ and _ slow _ warm - up timescales , respectively .
it is apparent that even for this very nearby and otherwise favorable hot - core source , the glycine emission strength is predicted to be very moderate ; convolving the emission with a beam appropriate to the jcmt ( @xmath120 22 arcesec ) produces line strengths of no more than 10 mk in the most optimistic model . in the case of
the _ fast _ warm - up model , which reproduces methyl formate emission most accurately , the glycine line strengths are no more than 250 @xmath44k .
convolution of the emission using a beam appropriate to the iram30 m telescope for each line ( @xmath110 12 arcsec ) produces line strengths of around 1 , 3 , and 30 mk , for _ fast _ , _ medium _ , and _ slow _ models respectively .
this suggests that one of the strongest glycine emission lines could plausibly be detected toward the given source using this instrument , assuming that the _ slow _ model values are accurate ; a 3-@xmath87 line detection would require a noise level of around 10 mk ( assuming no line blending , which is not modeled here ) .
however , detection assuming either of the other model values would require impractical signal - to - noise values .
in order to make an approximate estimate of the detectability of glycine using alma , the same single - beam convolution method is used , along with two plausible , fixed beam sizes : a generic 1 arcsec beam , and a 0.4 arcsec beam to match the size of the region of strongest glycine emission ( see fig . [ fig - profile ] ) . the reduced beam - dilution produced in these cases results in much greater line strengths , and suggests that the detection of a strong glycine line with alma is very plausible ( line blending and model inaccuracies notwithstanding ) .
the weakest emission predicted for any of the simulated glycine lines , with a beam width of 1 arcsec , is @xmath187 mk , while at 0.4 arcsec the weakest line has a strength of @xmath1350 mk . under the optimistic _ slow _ warm - up model conditions ,
convolved line strengths are close to 10 k for the strongest lines , using the smaller beam . for comparison
, the alma on - line sensitivity calculator indicates that , with 32 antennae ( cycle 1 ) and a resolution of 0.4 arcsec , alma would achieve an rms of 200 mk per 1 km / s velocity bin in 1 hour .
the line simulations therefore suggest that while detection of glycine using single - dish instruments is unlikely , even for this very favorable source , detection with alma is highly plausible ( ignoring possible line - blending effects ) .
pulliam et al . ( 2012 ) recently looked for hydroxylamine emission toward a selection of sources including sgr b2(n ) , orion kl and w3 irs5 .
they found no nh@xmath0oh emission using data obtained with the nrao12 m telescope , with noise levels of a few mk .
both the gwh model and the enhanced nh@xmath0oh model of section 4.1 exhibit large fractional abundances for this molecule , albeit with different formation mechanisms in each case . in order to test whether the model values are consistent with the non - detections of pulliam et al .
, the spectroscopic model is again applied to the chemical model results . out of the sources observed by both pulliam et al .
( 2012 ) and van der tak et al .
( 2000 ) , the only source present in both datasets is w3 irs5 . for this simulation ,
a generic line width of 9 km / s is assumed , following pulliam et al .
( 2012 ) .
tables [ tab - nh2oh-1 ] and [ tab - nh2oh-2 ] show the results of a selection of nh@xmath0oh emission - line simulations for w3 irs5 , using , respectively , the standard _ medium _ warm - up timescale model results , and the results of the model of sec .
4.1 that employs a gas - phase glycine formation mechanism and a large initial nh@xmath0oh abundance on the grains , produced by the low - temperature formation of nh@xmath0oh on dust grains during the collapse phase .
the results shown in table [ tab - nh2oh-1 ] are all consistent with the non - detection ; pulliam et al .
( 2012 ) find a 1-@xmath87 noise level of 4.3 mk in their w3 irs5 data , while the simulated intensities are of order 10 @xmath44k or less .
however , the model with enhanced nh@xmath0oh formation produces maximum convolved line intensities of around 50 mk , or around 1 order of magnitude greater than the noise level .
since the simulated line strengths scale well with the peak fractional abundance of nh@xmath0oh obtained in the chemical models , the observed values would require a peak gas - phase abundance of @xmath88 for the models to reproduce the non - detection of hydroxylamine , assuming all other parameters in the models are correct .
the implication for the models is that the cold - stage conversion of no / hno to nh@xmath0oh on the grains is too efficient , nominally by a factor of 10 .
this would suggest that the gas - phase formation of significant quantities of glycine is also less probable , in all sources .
the apparent over - estimate of hydroxylamine production could be explained by several possibilities ; one is that the chemical network does not include a sufficent number of competing reaction mechanisms for all of the states of hydrogenation between no and nh@xmath0oh .
another possibility is that the determination that the reaction h + hno @xmath31 h@xmath0no has an insignificant barrier , by congiu et al .
( 2012 ) , is dependent on experimental conditions that are inappropriate to these models , such as the ice composition , which is predominantly water in the ism .
alternatively , the hydroxylamine abundance in the ice may be adversely affected either by photo - destruction or hydrogen abstraction processes within the ice , whose rates or barriers may be inaccurately quantified in the models .
the general behavior of complex molecules as simulated in the new models is broadly similar to that seen in the previous models , such as g&h and gwh ; evaporation of grain - surface molecules into the gas - phase occurs around the same temperatures and in similar quantities .
however , the formation temperatures of molecules on the grains are somewhat different , due both to the distinction between surface and ice - mantle populations and to the difference in diffusion rates in each medium ; hco - related species are formed strongly around 25 k , while ch@xmath4-related species are formed around 20 k both somewhat lower than previous models , though more in line with recent experimental results such as @xcite .
the interaction between the surface and ice - mantle chemistry in the models is complex ; mobile radicals may land on the ice surface from the gas phase or may diffuse out of the mantle , to react to form more complex species , which show a net diffusion back into the mantle as more volatile species move out .
the migration of mantle species to the surface as temperatures increase tends to mitigate the ` trapping ' effect that would be expected using a model in which there were no surface
mantle diffusion .
much complex chemistry also occurs within the mantles themselves , particularly at elevated temperatures .
the importance of mantle versus surface processes is dependent on the source of the reactants , the mobilities of the reactants , the temperature , and the general composition of the surrounding ice ; no _ general _ statement can therefore be made on this point other than that the diffusion within the ice mantles , and between the surface and the bulk ice , appears to play a major part in hot - core chemistry .
( although the importance for specific , individual molecules is discussed in section 3 ) . aside from the effects of the differing treatment of the physical structure of the ices employed in this model , the inclusion of a more comprehensive set of hydrogen - abstraction reactions in the ice chemistry also demonstrates the key role of water and oh in producing other free radicals within the ice that may go on to form more complex organic structures .
the oh radical is highly reactive , and may abstract hydrogen from stable molecules with only a small activation energy typically lower than the barrier to oh diffusion .
thus , hydrogen abstraction by oh may occur with a net rate equal to its rate of diffusion , in the case where the target molecule is less mobile than oh .
the large quantities of water present in the ice makes oh the most commonly produced photodissociation product , caused by the weak , cosmic ray - induced , uv radiation that permeates the hot core .
the complex molecules , such as glycolaldehyde ( ch@xmath0(oh)cho ) , ethanol ( c@xmath0h@xmath40oh ) and methyl formate ( hcooch@xmath4 ) , formed by the addition of hco or ch@xmath4 to the photoproducts of methanol ( ch@xmath4oh ) are produced early , at low temperatures , before oh becomes mobile ( or abundant ) .
their production is therefore reflective of the direct photodissociation of methanol into ch@xmath0oh , ch@xmath4o and ch@xmath4 . while greater temperatures increase oh mobility , below 40 k its primary destruction route ( and the primary formation route for hco ) is the reaction oh + h@xmath0co @xmath31 hco + h@xmath0o , which is assumed to require no activation energy .
the mobility of h@xmath0co itself drives the reaction , and it is not until h@xmath0co evaporates at around 40 k that oh becomes important in providing radicals other than hco . after this , hco becomes less prevalent in the ices , while ch@xmath4 is still produced in some abundance by methanol photodissociation , resulting in a moderate bias toward methyl - group addition over aldehyde production . at this point , ammonia ( nh@xmath4 ) and methanol , as major constituents of the ice , become the dominant reaction partners for oh , producing nh@xmath0 and ch@xmath0oh ( ch@xmath4o is also produced , but the higher activation energy for this process renders it uncompetitive versus the hydroxymethyl channel ) .
it is therefore at temperatures above @xmath140 k that the formation of amino - organics becomes most active , by addition of nh@xmath0 to methanol products , or to more complex radicals .
however , at even higher temperatures , nh@xmath0 also becomes an important source of radicals in itself , as the barriers to hydrogen abstraction by nh@xmath0 are relatively low compared to its diffusion barrier , as is the case for oh . above around 55 k , the photodissociation of water into oh typically results in the subsequent conversion of ammonia to nh@xmath0 , which becomes the ultimate agent of hydrogen abstraction from organic molecules in the ice .
this continues until ammonia , the source of nh@xmath0 radicals , evaporates at @xmath1120 k. the inclusion in the model of ice surface / mantle formation mechanisms for glycine indicates that it may be formed in a number of ways , whose degree of influence depends on the timescale of the hot - core evolution .
the abundance of acetic acid ( ch@xmath4cooh ) is perhaps most important in determining which mechanism is most important . in
the _ slow _ warm - up model , acetic acid abundance is relatively high within the ices , and the abstraction of hydrogen by nh@xmath0 , followed by the addition of another amine group , is the most important glycine - formation route .
the same is true , albeit to a lesser extent , for the _ medium _ warm - up timescale model .
the _ fast _ warm - up model produces less acetic acid , and the addition of ch@xmath0nh@xmath0 and hoco products of oh - induced hydrogen abstraction from methylamine and formic acid , respectively becomes the major formation route in conjunction with oh addition to the nh@xmath0ch@xmath0co radical , which is formed by abstraction from the amino aldehyde glycinal ( nh@xmath0ch@xmath0cho ) .
it is questionable as to whether the high acetic acid abundances produced by the _ slow _ model are supported by observations , indicating both that glycine abundances may be lower than suggested by that model , and that no single ice surface / mantle process of the three that are tested in this work should be strongly dominant .
it may also be noted that , in the case of the ch@xmath0nh@xmath0 + hoco glycine - formation route , a plausible alternative channel , producing ch@xmath4nh@xmath0 + co@xmath0 , may also be important . in this model
a branching ratio of 1:1 is assumed ; however , the accuracy of this assumption can not be ascertained without experimental or computational data for the solid - phase process ( as opposed to the gas - phase equivalent , in which glycine production would be prohibitively disfavored ) .
the effect of a lower efficiency for glycine production through this method would be strongest in the _ fast _ warm - up scenario , although the availability of alternative routes means that glycine abundance is unlikely to fall by more than a factor of 2 as a result ; conversely , a more favorable branching ratio could increase the glycine production by a similar factor under _ fast _ warm - up conditions .
a gas - phase mechanism for the formation of glycine was also tested in this model , consisting of the addition of protonated hydroxylamine ( nh@xmath4oh@xmath10 ) and acetic acid ( ch@xmath4cooh ) , followed by electronic recombination . even assuming the highest plausible rate , with large quantities of hydroxylamine and acetic acid produced on the grains , the amount of glycine produced is only on the order of @xmath55@xmath47 , or 1% of the amount formed on the dust grains .
the limited comparison of simulated nh@xmath0oh emission with observations suggests that the actual fractional abundance may indeed be lower than assumed in that model .
it is possible that the fraction of dissociative recombination of protonated glycine that produce glycine itself may be somewhat higher than assumed ( 5% ) , but it is unlikely to be underestimated by more than a factor of a few . the combination of favorable conditions and rates that would be required therefore suggest that gas - phase production by this method is unlikely to produce significant quantities of glycine .
the possibility , demonstrated in the models , for significant glycine production to occur through several distinct chemical routes suggests that the formation of glycine within the ice mantles is no less plausible than the formation of more commonly observed complex organic molecules .
the peak gas - phase fractional abundances for glycine of @xmath89
@xmath90 produced by the chemical models suggest that this molecule could be reasonably abundant in regions within hot cores .
however , the lte radiative transfer and convolution models suggest that the detection of glycine toward a favorable source may be impossible with single - dish instruments , assuming any but the most optimistic of the chemical models ; this is due largely to the extremely compact emission exhibited by glycine .
the model findings are therefore in agreement with the current lack of a detection of this molecule .
however , it appears that the new alma telescope will have sufficient sensitivity and spatial resolution to easily detect the strongest glycine emission lines , assuming that the emission is not obscured by that from other molecules , and that the source is sufficiently closeby that the highly compact glycine - emission region is well resolved .
detection will likely require the careful choice of sources that are both nearby and that exhibit narrow emission lines .
for this reason , detection toward the well - observed galactic center source sgr b2(n ) may be questionable . while alma should allow the distinction of different structures within that source , which contribute to the considerable line widths ( belloche et al .
2008 ) , it is not clear whether a predicted compact glycine - emission region ( of temperature @xmath91200 k ) could be resolved sufficiently well both to produce high enough line strengths for detection and to filter out contamination from other molecules .
no simulations such as those produced above for other sources are possible , due to the absence of _ both _ a temperature and density profile for sgr b2(n ) in the literature .
the most important factor in the detectibility of glycine toward hot cores is likely to be its binding energy , which determines its evaporation temperature , and thus its radius of peak fractional abundance .
the simulations of glycine emission in ngc 6334 irs1 suggest that the strongest glycine emission should emanate from a region @xmath11000 au in diameter within that source .
this putative emission is predicted to be detectable with alma .
the single - dish convolution of the simulated emission shown in section 5.2 , using a beam width of 0.4 arcsec as appropriate to alma , represents a first approximation to the detectibility of glycine at that resolution ; the use of an interferometer would produce spatial filtering such that the more extended emission of other hot - core molecules could be resolved out , lessening the impact of contamination from other lines on the glycine emission .
the combined chemical and spectroscopic models produce a very good match with observed methanol line emission toward ngc 6334 irs1 .
the match for methyl formate is also good , in the case of the two shorter - timescale models ; warm - up timescales of around 1 myr may not therefore be appropriate to hot cores , although the precise binding energy of methyl formate is important to this determination .
the methyl formate results also indicate that , while other mechanisms may plausibly exist for the formation of methyl formate in hot cores , the addition of methanol photo - fragments within the dust - grain ice mantles , as modeled here , appears adequate to explain observed abundances .
meanwhile , the structural isomer of methyl formate , glycolaldehyde , is almost certainly over - produced , although this may be remedied by a more careful consideration of the product branching ratios of the reaction between hco and ch@xmath0oh radicals in the ices .
branching ratios favoring an alternative [ co + ch@xmath4oh ] channel would seem plausible .
all of the radiative transfer calculations presented here are highly dependent on the observationally - determined temperature density radius profiles provided by van der tak et al .
while the simulations of methanol and methyl formate emission agree remarkably well with recent observational data , the uncertainty in the profiles naturally increases at smaller radii , and the fidelity of a spherically - symmetric model must inevitably break down where outflows or clumpy structure become important .
nevertheless , in the absence of more precise data , perhaps also to be provided by alma , the simple models used here provide a first quantitative estimate of the behavior and detectability of glycine in star - forming regions .
a more comprehensive assessment would also incorporate the analysis of potential line blending , using spectroscopic models that include all species for which data exist , as well as an explicit treatment to simulate interferometric observations where appropriate .
the main conclusions of this work are summarized below : 1 .
the hot - core chemical models predict peak gas - phase glycine abundances of @xmath1@xmath89
@xmath90 @xmath47 ( dependent on warm - up timescale ) .
the peak abundance values are attained at around 200 k , when glycine evaporates from the dust grains .
2 . glycine is found to form almost exclusively within or upon dust - grain ice mantles , beginning at temperatures of @xmath14050 k , and ending at @xmath1120 k. the three main radical radical addition mechanisms investigated appear to have approximately equal influence on the resultant gas - phase quantities .
gas - phase glycine formation involving acetic acid ( ch@xmath4cooh ) and protonated hydroxylamine ( nh@xmath4oh@xmath10 ) is insignificant under even the most optimistic conditions .
4 . related organic species such as propionic acid , propionaldehyde ( propanal ) , and the amino aldehyde glycinal are predicted also to attain significant gas - phase abundances , although their detectibility would be dependent also on their emission radii , via their evaporation temperatures .
the production of glycinal ( nh@xmath0ch@xmath0cho ) and acetic acid ( ch@xmath4cooh ) two key molecules in the formation of glycine in the ice mantles are strongly dependent on the earlier production of acetaldehyde ( ch@xmath4cho ) .
acetaldehyde and acetic acid could provide observational information on potential glycine abundances in the absence of direct detection .
6 . the current lack of a detection of glycine toward hot cores may be explained by its high temperature of evaporation , which results in a small emission radius .
glycine is not expected to be detectable with single - dish instruments over realizable integration times toward any hot - core source .
the spectroscopic model suggests glycine may be detectable using alma at sub - arcsecond resolutions toward bright , nearby sources with relatively narrow emission lines , such as ngc 6334 irs1 .
. observed methyl formate abundances are well reproduced by the two shorter - timescale models , indicating that grain - related formation mechanisms are sufficient ( as suggested by previous models ) , and that ( 8400 k ) warm - up timescales of around 1 myr may be inappropriate for hot cores .
the anomalously high abundance produced by the models for the structural isomer of methyl formate , glyclolaldehyde , may be remedied by the consideration of alternative branches for the hco + ch@xmath0oh reaction in the ice mantles , favoring the formation of co + ch@xmath4oh .
cosmic ray - induced photodissociation of water ice , to produce oh , is crucial to the production of radicals within the bulk ices .
the oh radicals easily abstract hydrogen from other molecules , including ammonia , to produce reactive molecular radicals that may combine to form more complex species . 11 .
nh@xmath0 , formed by hydrogen abstraction from ammonia by oh , also appears to be an important agent of radical propagation within the ice mantles .
this process , and the related production of amino - organics , becomes most significant at temperatures between @xmath140120 k. the observed nitrogen content of hot - core molecules may thus be related to the timescale spent in this temperature range .
much , if not most , of the chemistry that produces complex organic molecules occurs within the ice mantles , while the more rapid diffusion that obtains on the ice surface allows different reaction kinetics to occur concurrently .
thus , the explicit treatment of these two phases , their coupling via thermal diffusion , and the interaction of the dust / ice surface with the gas phase are all essential elements in the chemical model . 13 .
the simple spectroscopic model used here provides a direct comparison between chemical models and observational spectroscopic data for complex molecules in specific sources , an element that has been lacking until now in the study of hot cores .
the author thanks the referee for providing helpful comments , and k. berg for a critical reading of the manuscript and many valuable discussions .
this work is funded by the nasa astrophysics theory program , grant number nnx11ac38 g .
lc h@xmath0 & @xmath92 + h & @xmath93 + he & @xmath94 + c & @xmath95 + n & @xmath96 + o & @xmath97 + s & @xmath98 + na & @xmath99 + mg & @xmath100 + si & @xmath101 + p & @xmath102 + cl & @xmath103 + fe & @xmath102 + lclclcl hoco & + & ch@xmath0nh@xmath0 & @xmath31 & nh@xmath0ch@xmath0cooh + & & & @xmath31 & ch@xmath4nh@xmath0 & + & co@xmath0 + nh@xmath0 & + & ch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh + oh & + & nh@xmath0ch@xmath0co & @xmath31 & nh@xmath0ch@xmath0cooh + h & + & hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh + ch@xmath0oh & + & hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh & + & h@xmath0co + ch@xmath4o & + & hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh & + & h@xmath0co + hco & + & hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh & + & co + hoco & + & hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh & + & co@xmath0 + nh & + & ch@xmath0cooh & @xmath31 & hnch@xmath0cooh + hco & + & ch@xmath0nh@xmath0 & @xmath31 & nh@xmath0ch@xmath0cho + & & & @xmath31 & ch@xmath4nh@xmath0 & + & co + nh@xmath0 & + & ch@xmath0cho & @xmath31 & nh@xmath0ch@xmath0cho + h & + & nh@xmath0ch@xmath0co & @xmath31 & nh@xmath0ch@xmath0cho + ch@xmath0oh & + & nh@xmath0ch@xmath0co & @xmath31 & nh@xmath0ch@xmath0cho & + & h@xmath0co + ch@xmath4o & + & nh@xmath0ch@xmath0co & @xmath31 & nh@xmath0ch@xmath0cho & + & h@xmath0co + hco & + & nh@xmath0ch@xmath0co & @xmath31 & nh@xmath0ch@xmath0cho & + & co + hoco & + & nh@xmath0ch@xmath0co & @xmath31 & nh@xmath0ch@xmath0cho & + & co@xmath0 + h & + & hnch@xmath0cho & @xmath31 & nh@xmath0ch@xmath0cho + ch@xmath0oh & + & hnch@xmath0cho & @xmath31 & nh@xmath0ch@xmath0cho & + & h@xmath0co + ch@xmath4o & + & hnch@xmath0cho & @xmath31 & nh@xmath0ch@xmath0cho & + & h@xmath0co + hco & + & hnch@xmath0cho & @xmath31 & nh@xmath0ch@xmath0cho & + & co + hoco & + & hnch@xmath0cho & @xmath31 & nh@xmath0ch@xmath0cho & + & co@xmath0 + nh & + & ch@xmath0cho & @xmath31 & hnch@xmath0cho + ch@xmath4co & + & oh & @xmath31 & ch@xmath4cooh + ch@xmath4 & + & hoco & @xmath31 & ch@xmath4cooh + h & + & ch@xmath0cooh & @xmath31 & ch@xmath4cooh + ch@xmath0oh & + & ch@xmath0cooh & @xmath31 & ch@xmath4cooh & + & h@xmath0co + ch@xmath4o & + & ch@xmath0cooh & @xmath31 & ch@xmath4cooh & + & h@xmath0co + hco & + & ch@xmath0cooh & @xmath31 & ch@xmath4cooh & + & co + hoco & + & ch@xmath0cooh & @xmath31 & ch@xmath4cooh & + & co@xmath0 + ch@xmath0 & + & hoco & @xmath31 & ch@xmath0cooh + ch@xmath0 & + & ch@xmath0cooh & @xmath31 & c@xmath0h@xmath54cooh + ch@xmath4 & + & ch@xmath0cooh & @xmath31 & c@xmath0h@xmath40cooh + h & + & c@xmath0h@xmath54cooh & @xmath31 & c@xmath0h@xmath40cooh + ch@xmath0oh & + & c@xmath0h@xmath54cooh & @xmath31 & c@xmath0h@xmath40cooh & + & h@xmath0co + ch@xmath4o & + & c@xmath0h@xmath54cooh & @xmath31 & c@xmath0h@xmath40cooh & + & h@xmath0co + hco & + & c@xmath0h@xmath54cooh & @xmath31 & c@xmath0h@xmath40cooh & + & co + hoco & + & c@xmath0h@xmath54cooh & @xmath31 & c@xmath0h@xmath40cooh & + & co@xmath0 + hoco & + & c@xmath0h@xmath40 & @xmath31 & c@xmath0h@xmath40cooh + & & & @xmath31 & c@xmath0h@xmath104 & + & co@xmath0 + oh & + & c@xmath0h@xmath40co & @xmath31 & c@xmath0h@xmath40cooh + ch@xmath4 & + & ch@xmath0cho & @xmath31 & c@xmath0h@xmath40cho + h & + & c@xmath0h@xmath54cho & @xmath31 & c@xmath0h@xmath40cho + ch@xmath0oh & + & c@xmath0h@xmath54cho & @xmath31 & c@xmath0h@xmath40cho & + & h@xmath0co + ch@xmath4o & + & c@xmath0h@xmath54cho & @xmath31 & c@xmath0h@xmath40cho & + & h@xmath0co + hco & + & c@xmath0h@xmath54cho & @xmath31 & c@xmath0h@xmath40cho & + & co + hoco & + & c@xmath0h@xmath54cho & @xmath31 & c@xmath0h@xmath40cho & + & co@xmath0 + ch@xmath0 & + & ch@xmath0cho & @xmath31 & c@xmath0h@xmath54cho + hco & + & c@xmath0h@xmath40 & @xmath31 & c@xmath0h@xmath40cho + h & + & c@xmath0h@xmath40co & @xmath31 & c@xmath0h@xmath40cho + ch@xmath0oh & + & c@xmath0h@xmath40co & @xmath31 & c@xmath0h@xmath40cho & + & h@xmath0co + ch@xmath4o & + & c@xmath0h@xmath40co & @xmath31 & c@xmath0h@xmath40cho & + & h@xmath0co + hco & + & c@xmath0h@xmath40co & @xmath31 & c@xmath0h@xmath40cho & + & co + hoco & + & c@xmath0h@xmath40co & @xmath31 & c@xmath0h@xmath40cho & + & co@xmath0 1 & h + co @xmath31 hco & & & andersson ( priv . comm . ) + 2 & h + h@xmath0co @xmath31 ch@xmath0oh & 4500 & 1.35 & goumans ( priv . comm . )
+ 3 & h + h@xmath0co @xmath31 ch@xmath4o & 2320 & 1.35 & goumans ( priv . comm . ) + 4 & h + h@xmath0co @xmath31 hco + h@xmath0 & 2960 & 1.22 & goumans ( priv . comm . ) + 5 & h + ch@xmath4oh @xmath31 ch@xmath0oh + h@xmath0 & 4380 & 1.0 & goumans ( priv .
comm . ) + 6 & h + ch@xmath4oh @xmath31 ch@xmath4o + h@xmath0 & 6640 & 1.0 & goumans ( priv . comm . )
+ 7 & h + ch@xmath54 @xmath31 ch@xmath4 + h@xmath0 & 5940 & 2.17 & andersson et al .
( 2009 ) ; garrod & pauly ( 2011 ) + 8 & h + ch@xmath4cho @xmath31 h@xmath0 + ch@xmath4co & 2120 & 1.0 & warnatz ( 1984 ) + 9 & h + hcooch@xmath4 @xmath31 h@xmath0 + ch@xmath4oco & 3970 & 1.0 & good & francisco ( 2002 ) + 10 & h + hcooch@xmath4 @xmath31 h@xmath0 + hcooch@xmath0 & 4470 & 1.0 & good & francisco ( 2002 ) + 11 & h + hcooh @xmath31 h@xmath0 + cooh & 6150 & 1.0 & evans - polanyi , # 8/#12 + 12 & h + nh@xmath0cho @xmath31
h@xmath0 + nh@xmath0co & 3500 & 1.0 & syrstad & turecek ( 2001 ) + 13 & h + ch@xmath0(oh)cho @xmath31 h@xmath0 + ch@xmath0(oh)co & 2350 & 1.0 & evans - polanyi , # 8/#12 + 14 & h + hnco @xmath31 nh@xmath0 + co & 2300 & 1.0 & tsang ( 1992 ) + 15 & h + hnco @xmath31 nh@xmath0co & 1390 & 1.0 & nguyen et al .
( 1996 ) + 16 & h@xmath0 + oh @xmath31 h@xmath0o + h & 2100 & 1.0 & baulch et al .
( 1984 ) + 17 & ch@xmath0oh + ch@xmath4cho @xmath31 ch@xmath4oh + ch@xmath4co & 5300 & 1.0 & evans - polanyi + 18 & ch@xmath0oh + co @xmath31 ch@xmath0(oh)co & 2500 & 1.0 & garrod et al . (
2008 ) + 19 & ch@xmath0oh + h@xmath0co @xmath31 ch@xmath4oh + hco & 2940 & 1.0 & tsang ( 1987 ) + 20 & ch@xmath0oh + nh@xmath0cho @xmath31
ch@xmath4oh + nh@xmath0co & 4320 & 1.0 & evans - polanyi + 21 & ch@xmath0oh + ch@xmath0(oh)cho @xmath31 ch@xmath4oh + ch@xmath0(oh)co & 3100 & 1.0 & evans - polanyi + 22 & ch@xmath0oh + hnco @xmath31 ch@xmath4oh + ocn & 1200 & 1.0 & estimate + 23 & ch@xmath0oh + c@xmath0h@xmath40cho @xmath31 ch@xmath4oh + c@xmath0h@xmath40co & 3100 & 1.0 & ( # 21 ) + 24 & ch@xmath0oh + c@xmath4h@xmath54 @xmath31 ch@xmath4oh + c@xmath4h@xmath4 & 4960 & 1.0 & evans - polanyi + 25 & ch@xmath0oh + c@xmath4h@xmath104 @xmath31 ch@xmath4oh + c@xmath4h@xmath40 & 4790 & 1.0 & evans - polanyi + 26 & ch@xmath0oh + ch@xmath0nh @xmath31 ch@xmath4oh + h@xmath0cn & 1200 & 1.0 & ( # 22 ) + 27 & ch@xmath0oh + ch@xmath4nh@xmath0 @xmath31 ch@xmath4oh + ch@xmath0nh@xmath0 & 6780 & 1.0 & evans - polanyi + 28 & ch@xmath0oh + ch@xmath4cn @xmath31 ch@xmath4oh + ch@xmath0cn & 6200 & 1.0 & evans - polanyi + 29 & ch@xmath0oh + c@xmath0h@xmath40cn @xmath31 ch@xmath4oh + c@xmath0h@xmath54cn & 6100 & 1.0 & evans - polanyi + 30 & ch@xmath0oh + c@xmath4h@xmath105cn @xmath31 ch@xmath4oh + c@xmath4h@xmath104cn & 6100 & 1.0 & ( # 29 ) + 31 & ch@xmath0oh + nh@xmath0ch@xmath0cn @xmath31 ch@xmath4oh + nhch@xmath0cn & 6100 & 1.0 & ( # 29 ) + 32 & ch@xmath0oh + nh@xmath0oh @xmath31 ch@xmath4oh + hnoh & 5130 & 1.0 & evans - polanyi + 33 & ch@xmath0oh + n@xmath0h@xmath54 @xmath31 ch@xmath4oh + nh@xmath0nh & 6100 & 1.0 & ( # 29 ) + 34 & ch@xmath0oh + nh@xmath0och@xmath4 @xmath31 ch@xmath4oh + hnoch@xmath4 & 6100 & 1.0 & ( # 29 ) + 35 & ch@xmath0oh + nh@xmath0ch@xmath0oh @xmath31 ch@xmath4oh + hnch@xmath0oh & 6100 & 1.0 & ( # 29 ) + 36 & ch@xmath0oh + nh@xmath0conh@xmath0 @xmath31 ch@xmath4oh + nh@xmath0conh & 6100 & 1.0 & ( # 29 ) + 37 & ch@xmath0oh + nh@xmath0cooh @xmath31 ch@xmath4oh + hncooh & 6100 & 1.0 & ( # 29 ) + 38 & ch@xmath0oh + ch@xmath4oconh@xmath0 @xmath31 ch@xmath4oh + ch@xmath4oconh & 6100 & 1.0 & ( # 29 ) + 39 & ch@xmath0oh + hoch@xmath0conh@xmath0 @xmath31 ch@xmath4oh + hoch@xmath0conh & 6100 & 1.0 & ( # 29 ) + 40 & ch@xmath0oh + nh@xmath0cocho @xmath31 ch@xmath4oh + hncocho & 6100 & 1.0 & ( # 29 ) + 41 & ch@xmath0oh + ch@xmath4conh@xmath0 @xmath31 ch@xmath4oh + ch@xmath4conh & 6100 & 1.0 & ( # 29 ) + 42 & ch@xmath0oh + nh@xmath0ch@xmath0cooh @xmath31 ch@xmath4oh + hnch@xmath0cooh & 6100 & 1.0 & ( # 29 ) + 43 & ch@xmath0oh + c@xmath0h@xmath40cooh @xmath31 ch@xmath4oh + c@xmath0h@xmath54cooh & 6100 & 1.0 & ( # 29 ) + 44 & ch@xmath0oh + nh@xmath0c@xmath0h@xmath40 @xmath31 ch@xmath4oh + nh@xmath0c@xmath0h@xmath54 & 6100 & 1.0 & ( # 29 ) + 45 & ch@xmath0oh + nh@xmath0ch@xmath0cho @xmath31 ch@xmath4oh + nh@xmath0ch@xmath0co & 3100 & 1.0 & ( # 29 ) + 46 & ch@xmath0oh + nh@xmath0ch@xmath0cho @xmath31 ch@xmath4oh + hnch@xmath0cho & 6100 & 1.0 & ( # 29 ) + 47 & ch@xmath0oh + ch@xmath4cooh @xmath31 ch@xmath4oh + ch@xmath0cooh & 6100 & 1.0 & ( # 29 ) + 48 & ch@xmath4 + ch@xmath4cho @xmath31 ch@xmath54 + ch@xmath4co & 1720 & 1.0 & evans - polanyi + 49 & ch@xmath4 + ch@xmath4oh @xmath31 ch@xmath54 + ch@xmath0oh & 3600 & 1.0 & tsang ( 1987 ) + 50 & ch@xmath4 + ch@xmath4oh @xmath31 ch@xmath54 + ch@xmath4o & 3480 & 1.0 & tsang ( 1987 ) + 51 & ch@xmath4 + co @xmath31 ch@xmath4co & 2500 & 1.0 & garrod et al .
( 2008 ) + 52 & ch@xmath4 + h@xmath0co @xmath31 ch@xmath54 + hco & 4440 & 1.0 & baulch et al .
( 1992 ) + 53 & ch@xmath4 + hcooch@xmath4 @xmath31 ch@xmath54 + ch@xmath4oco & 5040 & 1.0 & good & francisco ( 2002 ) + 54 & ch@xmath4 + hcooch@xmath4 @xmath31 ch@xmath54 + hcooch@xmath0 & 5800 & 1.0 & good & francisco ( 2002 ) + 55 & ch@xmath4 + hcooh @xmath31 ch@xmath54 + cooh & 4550 & 1.0 & evans - polanyi + 56 & ch@xmath4 + nh@xmath0cho @xmath31 ch@xmath54 + nh@xmath0co & 3520 & 1.0 & boden & back ( 1970 ) + 57 & ch@xmath4 + ch@xmath0(oh)cho @xmath31 ch@xmath54 + ch@xmath0(oh)co & 1880 & 1.0 & evans - polanyi + 58 & ch@xmath4 + hnco @xmath31 ch@xmath54 + ocn & 1200 & 1.0 & estimate + 59 & ch@xmath4o + ch@xmath4cho @xmath31 ch@xmath4oh + ch@xmath4co & 1360 & 1.0 & evans - polanyi + 60 & ch@xmath4o + co @xmath31 ch@xmath4oco & 2500 & 1.0 & garrod et al .
( 2008 ) + 61 & ch@xmath4o + h@xmath0co @xmath31 ch@xmath4oh + hco & 1500 & 1.0 & tsang & hampson ( 1986 ) + 62 & ch@xmath4o + hcooch@xmath4 @xmath31 ch@xmath4oh + ch@xmath4oco & 3200 & 1.0 & evans - polanyi + 63 & ch@xmath4o + hcooch@xmath4 @xmath31 ch@xmath4oh + hcooch@xmath0 & 4120 & 1.0 & thynne & gray ( 1963 ) + 64 & ch@xmath4o + hcooh @xmath31 ch@xmath4oh + cooh & 3170 & 1.0 & evans - polanyi + 65 & ch@xmath4o + nh@xmath0cho @xmath31 ch@xmath4oh + nh@xmath0co & 1980 & 1.0 & evans - polanyi + 66 & ch@xmath4o + ch@xmath0(oh)cho @xmath31 ch@xmath4oh + ch@xmath0(oh)co & 1470 & 1.0 & evans - polanyi + 67 & ch@xmath4o + hnco @xmath31 ch@xmath4oh + ocn & 1200 & 1.0 & estimate + 68 & ch@xmath4o + c@xmath0h@xmath40cho @xmath31 ch@xmath4oh + c@xmath0h@xmath40co & 1470 & 1.0 & ( # 66 ) + 69 & ch@xmath4o + c@xmath4h@xmath54 @xmath31 ch@xmath4oh + c@xmath4h@xmath4 & 1100 & 1.0 & evans - polanyi + 70 & ch@xmath4o + c@xmath4h@xmath104 @xmath31 ch@xmath4oh + c@xmath4h@xmath40 & 965 & 1.0 & evans - polanyi + 71 & ch@xmath4o + ch@xmath0nh @xmath31 ch@xmath4oh + h@xmath0cn & 1200 & 1.0 & ( # 67 ) + 72 & ch@xmath4o + ch@xmath4nh@xmath0 @xmath31 ch@xmath4oh + ch@xmath0nh@xmath0 & 2530 & 1.0 & evans - polanyi + 73 & ch@xmath4o + ch@xmath4cn @xmath31 ch@xmath4oh + ch@xmath0cn & 2070 & 1.0 & evans - polanyi + 74 & ch@xmath4o + c@xmath0h@xmath40cn @xmath31 ch@xmath4oh + c@xmath0h@xmath54cn & 2000 & 1.0 & evans - polanyi + 75 & ch@xmath4o + c@xmath4h@xmath105cn @xmath31 ch@xmath4oh + c@xmath4h@xmath104cn & 2000 & 1.0 & ( # 74 ) + 76 & ch@xmath4o + nh@xmath0ch@xmath0cn @xmath31 ch@xmath4oh + nhch@xmath0cn & 2000 & 1.0 & ( # 74 ) + 77 & ch@xmath4o + nh@xmath0oh @xmath31 ch@xmath4oh + hnoh & 2000 & 1.0 & ( # 74 ) + 78 & ch@xmath4o + n@xmath0h@xmath54 @xmath31 ch@xmath4oh + nh@xmath0nh & 2000 & 1.0 & ( # 74 ) + 79 & ch@xmath4o + nh@xmath0och@xmath4 @xmath31 ch@xmath4oh + hnoch@xmath4 & 2000 & 1.0 & ( # 74 ) + 80 & ch@xmath4o + nh@xmath0ch@xmath0oh @xmath31 ch@xmath4oh + hnch@xmath0oh & 2000 & 1.0 & ( # 74 ) + 81 & ch@xmath4o + nh@xmath0conh@xmath0 @xmath31 ch@xmath4oh + nh@xmath0conh & 2000 & 1.0 & ( # 74 ) + 82 & ch@xmath4o + nh@xmath0cooh @xmath31 ch@xmath4oh + hncooh & 2000 & 1.0 & ( # 74 ) + 83 & ch@xmath4o + ch@xmath4oconh@xmath0 @xmath31 ch@xmath4oh + ch@xmath4oconh & 2000 & 1.0 & ( # 74 ) + 84 & ch@xmath4o + hoch@xmath0conh@xmath0 @xmath31 ch@xmath4oh + hoch@xmath0conh & 2000 & 1.0 & ( # 74 ) + 85 & ch@xmath4o + nh@xmath0cocho @xmath31 ch@xmath4oh + hncocho & 2000 & 1.0 & ( # 74 ) + 86 & ch@xmath4o + ch@xmath4conh@xmath0 @xmath31 ch@xmath4oh + ch@xmath4conh & 2000 & 1.0 & ( # 74 ) + 87 & ch@xmath4o + nh@xmath0ch@xmath0cooh @xmath31 ch@xmath4oh + hnch@xmath0cooh & 2000 & 1.0 & ( # 74 ) + 88 & ch@xmath4o + c@xmath0h@xmath40cooh @xmath31 ch@xmath4oh + c@xmath0h@xmath54cooh & 2000 & 1.0 & ( # 74 ) + 89 & ch@xmath4o + nh@xmath0c@xmath0h@xmath40 @xmath31 ch@xmath4oh + nh@xmath0c@xmath0h@xmath54 & 2000 & 1.0 & ( # 74 ) + 90 & ch@xmath4o + nh@xmath0ch@xmath0cho @xmath31 ch@xmath4oh + nh@xmath0ch@xmath0co & 1470 & 1.0 & ( # 66 ) + 91 & ch@xmath4o + nh@xmath0ch@xmath0cho @xmath31 ch@xmath4oh + hnch@xmath0cho & 2000 & 1.0 & ( # 74 ) + 92 & ch@xmath4o + ch@xmath4cooh @xmath31 ch@xmath4oh + ch@xmath0cooh & 2000 & 1.0 & ( # 74 ) + 93 & nh@xmath0 + ch@xmath4cho @xmath31 nh@xmath4 + ch@xmath4co & 1250 & 1.0 & hack et al .
( 1986 ) + 94 & nh@xmath0 + ch@xmath4oh @xmath31 nh@xmath4 + ch@xmath0oh & 3280 & 1.0 & evans - polanyi + 95 & nh@xmath0 + ch@xmath4oh @xmath31 nh@xmath4 + ch@xmath4o & 3290 & 1.0 & evans - polanyi + 96 & nh@xmath0 + co @xmath31 nh@xmath0co & 2500 & 1.0 & garrod et al .
( 2008 ) + 97 & nh@xmath0 + h@xmath0co @xmath31 nh@xmath4 + hco & 2360 & 1.0 & evans - polanyi + 98 & nh@xmath0 + hcooch@xmath4 @xmath31 nh@xmath4 + ch@xmath4oco & 3360 & 1.0 & evans - polanyi + 99 & nh@xmath0 + hcooch@xmath4 @xmath31 nh@xmath4 + hcooch@xmath0 & 2690 & 1.0 & ( # 113 ) + 100 & nh@xmath0 + hcooh @xmath31 nh@xmath4 + cooh & 3340 & 1.0 & evans - polanyi + 101 & nh@xmath0 + nh@xmath0cho @xmath31 nh@xmath4 + nh@xmath0co & 2670 & 1.0 & evans - polanyi + 102 & nh@xmath0 + ch@xmath0(oh)cho @xmath31 nh@xmath4 + ch@xmath0(oh)co & 2400 & 1.0 & evans - polanyi + 103 & nh@xmath0 + hnco @xmath31
nh@xmath4 + ocn & 1200 & 1.0 & tomeczek & gradon ( 2003 ) + 104 & nh@xmath0 + ch@xmath54 @xmath31 nh@xmath4 + ch@xmath4 & 4680 & 1.0 & moller & wagner ( 1984 ) + 105 & nh@xmath0 + c@xmath0h@xmath40cho @xmath31 nh@xmath4 + c@xmath0h@xmath40co & 2400 & 1.0 & ( # 102 ) + 106 & nh@xmath0 + c@xmath0h@xmath104 @xmath31 nh@xmath4 + c@xmath0h@xmath40 & 5340 & 1.0 & ehbrecht et al .
( 1987 ) + 107 & nh@xmath0 + c@xmath4h@xmath54 @xmath31 nh@xmath4 + c@xmath4h@xmath4 & 2200 & 1.0 & evans - polanyi + 108 & nh@xmath0 + c@xmath4h@xmath104 @xmath31 nh@xmath4 + c@xmath4h@xmath40 & 2160 & 1.0 & khe & lesclaux ( 1979 ) + 109 & nh@xmath0 + c@xmath4h8 @xmath31 nh@xmath4 + c@xmath4h@xmath105 & 4730 & 1.0 & ehbrecht et al .
( 1987 ) + 110 & nh@xmath0 + ch@xmath0nh @xmath31 nh@xmath4 + h@xmath0cn & 1200 & 1.0 & ( # 103 ) + 111 & nh@xmath0 + ch@xmath4nh@xmath0 @xmath31 nh@xmath4 + ch@xmath0nh@xmath0 & 2990 & 1.0 & evans - polanyi + 112 & nh@xmath0 + ch@xmath4cn @xmath31 nh@xmath4 + ch@xmath0cn & 2730 & 1.0 & evans - polanyi + 113 & nh@xmath0 + c@xmath0h@xmath40cn @xmath31 nh@xmath4 + c@xmath0h@xmath54cn & 2690 & 1.0 & evans - polanyi + 114 & nh@xmath0 + c@xmath4h@xmath105cn @xmath31 nh@xmath4 + c@xmath4h@xmath104cn & 2690 & 1.0 & ( # 113 ) + 115 & nh@xmath0 + nh@xmath0ch@xmath0cn @xmath31 nh@xmath4 + nhch@xmath0cn & 2690 & 1.0 & ( # 113 ) + 116 & nh@xmath0 + nh@xmath0oh @xmath31
nh@xmath4 + hnoh & 2690 & 1.0 & ( # 113 ) + 117 & nh@xmath0 + n@xmath0h@xmath54 @xmath31 nh@xmath4 + nh@xmath0nh & 2690 & 1.0 & ( # 113 ) + 118 & nh@xmath0 + nh@xmath0och@xmath4 @xmath31 nh@xmath4 + hnoch@xmath4 & 2690 & 1.0 & ( # 113 ) + 119 & nh@xmath0 + nh@xmath0ch@xmath0oh @xmath31 nh@xmath4 + hnch@xmath0oh & 2690 & 1.0 & ( # 113 ) + 120 & nh@xmath0 + nh@xmath0conh@xmath0 @xmath31 nh@xmath4 + nh@xmath0conh & 2690 & 1.0 & ( # 113 ) + 121 & nh@xmath0 + nh@xmath0cooh @xmath31 nh@xmath4 + hncooh & 2690 & 1.0 & ( # 113 ) + 122 & nh@xmath0 + ch@xmath4oconh@xmath0 @xmath31 nh@xmath4 + ch@xmath4oconh & 2690 & 1.0 & ( # 113 ) + 123 & nh@xmath0 + hoch@xmath0conh@xmath0 @xmath31 nh@xmath4 + hoch@xmath0conh & 2690 & 1.0 & ( # 113 ) + 124 & nh@xmath0 + nh@xmath0cocho @xmath31 nh@xmath4 + hncocho & 2690 & 1.0 & ( # 113 ) + 125 & nh@xmath0 + ch@xmath4conh@xmath0 @xmath31 nh@xmath4 + ch@xmath4conh & 2690 & 1.0 & ( # 113 ) + 126 & nh@xmath0 + nh@xmath0ch@xmath0cooh @xmath31 nh@xmath4 + hnch@xmath0cooh & 2690 & 1.0 & ( # 113 ) + 127 & nh@xmath0 + c@xmath0h@xmath40cooh @xmath31 nh@xmath4 + c@xmath0h@xmath54cooh & 2690 & 1.0 & ( # 113 ) + 128 & nh@xmath0 + nh@xmath0c@xmath0h@xmath40 @xmath31 nh@xmath4 + nh@xmath0c@xmath0h@xmath54 & 2690 & 1.0 & ( # 113 ) + 129 & nh@xmath0 + nh@xmath0ch@xmath0cho @xmath31 nh@xmath4 + nh@xmath0ch@xmath0co & 2400 & 1.0 & ( # 102 ) + 130 & nh@xmath0 + nh@xmath0ch@xmath0cho @xmath31 nh@xmath4 + hnch@xmath0cho & 2690 & 1.0 & ( # 113 ) + 131 & nh@xmath0 + ch@xmath4cooh @xmath31 nh@xmath4 + ch@xmath0cooh & 2690 & 1.0 & ( # 113 ) + 132 & nh@xmath0 + c@xmath0h@xmath40oh @xmath31 nh@xmath4 + c@xmath0h@xmath40o & 4080 & 1.0 & evans - polanyi + 133 & o + co @xmath31 co@xmath0 & 1000 & 1.27 & goumans ( tunnelling ) + 134 & oh + ch@xmath4oh @xmath31 ch@xmath0oh + h@xmath0o & 359 & 1.0 & atkinson et al .
( 2001 ) + 135 & oh + ch@xmath4oh @xmath31 ch@xmath4o + h@xmath0o & 852 & 1.0 & warnatz ( 1984 ) + 136 & oh + ch@xmath4cho @xmath31 h@xmath0o + ch@xmath4co & 600 & 1.0 & estimate + 137 & oh + co @xmath31 co@xmath0 + h & 80 & 1.0 & ruffle & herbst ( 2001 ) + 138 & oh + co @xmath31 cooh & 2500 & 1.0 & garrod et al .
( 2008 ) + 139 & oh + hcooch@xmath4 @xmath31 h@xmath0o + ch@xmath4oco & 600 & 1.0 & estimate + 140 & oh + hcooch@xmath4 @xmath31 h@xmath0o + hcooch@xmath0 & 1000 & 1.0 & estimate + 141 & oh + hcooh @xmath31 h@xmath0o + cooh & 1000 & 1.0 & estimate + 142 & oh + nh@xmath0cho @xmath31 h@xmath0o + nh@xmath0co & 600 & 1.0 & estimate + 143 & oh + nh@xmath0cho @xmath31 h@xmath0o + hncho & 591 & 1.0 & evans - polanyi + 144 & oh + ch@xmath0(oh)cho @xmath31 h@xmath0o + ch@xmath0(oh)co & 600 & 1.0 & estimate + 145 & oh + hnco @xmath31 h@xmath0o + ocn & 1290 & 1.0 & tsang ( 1992 ) + 146 & oh + ch@xmath54 @xmath31 h@xmath0o + ch@xmath4 & 1780 & 1.0 & demore et al .
( 1997 ) + 147 & oh + nh@xmath4 @xmath31 h@xmath0o + nh@xmath0 & 925 & 1.0 & atkinson et al .
( 2004 ) + 148 & oh + c@xmath0h@xmath0 @xmath31 h@xmath0o + c@xmath0h & 5280 & 1.0 & liu et al .
( 1988 ) + 149 & oh + c@xmath0h@xmath54 @xmath31 h@xmath0o + c@xmath0h@xmath4 & 2990 & 1.0 & baulch et al .
( 1992 ) + 150 & oh + c@xmath0h@xmath40cho @xmath31 h@xmath0o + c@xmath0h@xmath40co & 600 & 1.0 & estimate + 151 & oh + c@xmath0h@xmath104 @xmath31 h@xmath0o + c@xmath0h@xmath40 & 1000 & 1.0 & atkinson et al .
( 2001 ) + 152 & oh + c@xmath4h@xmath54 @xmath31 h@xmath0o + c@xmath4h@xmath4 & 650 & 1.0 & warnatz ( 1984 ) + 153 &
oh + c@xmath4h@xmath104 @xmath31 h@xmath0o + c@xmath4h@xmath40 & 730 & 1.0 & tsang ( 1991 ) + 154 & oh + c@xmath4h8 @xmath31 h@xmath0o + c@xmath4h@xmath105 & 1310 & 1.0 & hu et al .
( 1997 ) + 155 & oh + ch@xmath0nh @xmath31 h@xmath0o + h@xmath0cn & 591 & 1.0 & ( # 143 ) + 156 & oh + ch@xmath4nh@xmath0 @xmath31 h@xmath0o + ch@xmath0nh@xmath0 & 1000 & 1.0 & estimate + 157 & oh + ch@xmath4cn @xmath31 h@xmath0o + ch@xmath0cn & 1030 & 1.0 & kurylo & knable ( 1984 ) + 158 & oh + c@xmath0h@xmath4cn @xmath31 h@xmath0o + c@xmath0h@xmath0cn & 4000 & 1.0 & evans - polanyi + 159 & oh + c@xmath0h@xmath40cn @xmath31 h@xmath0o + c@xmath0h@xmath54cn & 1000 & 1.0 & estimate + 160 & oh + c@xmath4h@xmath105cn @xmath31 h@xmath0o + c@xmath4h@xmath104cn & 1000 & 1.0 & estimate + 161 & oh + nh@xmath0ch@xmath0cn @xmath31 h@xmath0o + nhch@xmath0cn & 591 & 1.0 & ( # 143 ) + 162 &
oh + nh@xmath0oh @xmath31 h@xmath0o + hnoh & 2140 & 1.0 & espinosa - garcia et al .
( 1993 ) + 163 & oh + n@xmath0h@xmath54 @xmath31 h@xmath0o + nh@xmath0nh & 591 & 1.0 & ( # 143 ) + 164 &
oh + nh@xmath0och@xmath4 @xmath31 h@xmath0o + hnoch@xmath4 & 591 & 1.0 & ( # 143 ) + 165 & oh + nh@xmath0ch@xmath0oh @xmath31 h@xmath0o + hnch@xmath0oh & 591 & 1.0 & ( # 143 ) + 166 & oh + nh@xmath0conh@xmath0 @xmath31 h@xmath0o + nh@xmath0conh & 591 & 1.0 & ( # 143 ) + 167 & oh + nh@xmath0cooh @xmath31 h@xmath0o + hncooh & 591 & 1.0 & ( # 143 ) + 168 & oh + ch@xmath4oconh@xmath0 @xmath31 h@xmath0o + ch@xmath4oconh & 591 & 1.0 & ( # 143 ) + 169 & oh + hoch@xmath0conh@xmath0 @xmath31 h@xmath0o + hoch@xmath0conh & 591 & 1.0 & ( # 143 ) + 170 & oh + nh@xmath0cocho @xmath31 h@xmath0o + hncocho & 591 & 1.0 & ( # 143 ) + 171 & oh + ch@xmath4conh@xmath0 @xmath31 h@xmath0o + ch@xmath4conh & 591 & 1.0 & ( # 143 ) + 172 & oh + nh@xmath0ch@xmath0cooh @xmath31 h@xmath0o + hnch@xmath0cooh & 591 & 1.0 & ( # 143 ) + 173 &
oh + ch@xmath4ch@xmath0cooh @xmath31 h@xmath0o + c@xmath0h@xmath54cooh & 1000 & 1.0 & estimate + 174 & oh + nh@xmath0c@xmath0h@xmath40 @xmath31 h@xmath0o + nh@xmath0c@xmath0h@xmath54 & 1000 & 1.0 & estimate + 175 & oh + nh@xmath0ch@xmath0cho @xmath31 h@xmath0o + nh@xmath0ch@xmath0co & 600 & 1.0 & ( # 136 ) + 176 & oh + nh@xmath0ch@xmath0cho @xmath31 h@xmath0o + hnch@xmath0cho & 591 & 1.0 & ( # 143 ) + 177 & oh + ch@xmath4cooh @xmath31 h@xmath0o + ch@xmath0cooh & 1000 & 1.0 & estimate + 178 & oh + c@xmath0h@xmath40oh @xmath31 h@xmath0o + c@xmath0h@xmath40o & 1510 & 1.0 & evans - polanyi lrrr h & 450 & 315 & 158 + h@xmath0 & 430 & 301 & 150 + oh & 2850 & 1995 & 998 + h@xmath0o & 5700 & 3990 & 1995 + ch@xmath0 & 1050 & 735 & 368 + ch@xmath4 & 1175 & 823 & 411 + ch@xmath54 & 1300 & 910 & 455 + nh & 2380 & 1666 & 833 + nh@xmath0 & 3960 & 2772 & 1386 + nh@xmath4 & 5530 & 3871 & 1936 + co & 1150 & 805 & 403 + co@xmath0 & 2575 & 1803 & 901 + hco & 1600 & 1120 & 560 + h@xmath0co & 2050 & 1435 & 718 + ch@xmath4o & 2500 & 1750 & 875 + ch@xmath0oh & 5080 & 3556 & 1778 + ch@xmath4oh & 5530 & 3871 & 1936 + ch@xmath0co & 2200 & 1540 & 770 + ch@xmath4co & 2325 & 1628 & 814 + ch@xmath4cho & 2775 & 1943 & 971 + hoco & 5120 & 3584 & 1792 + hcooh & 5570 & 3899 & 1950 + ch@xmath4och@xmath4 & 3675 & 2573 & 1286 + c@xmath0h@xmath40oh & 6260 & 4382 & 2191 + hcooch@xmath4 & 5200 & 3640 & 1820 + ch@xmath0(oh)cho & 6680 & 4676 & 2338 + ch@xmath4cooh & 6300 & 4410 & 2205 + ( ch@xmath4o)@xmath0 & 5000 & 3500 & 1750 + ch@xmath4och@xmath0oh & 7580 & 5306 & 2653 + ( ch@xmath0oh)@xmath0 & 10200 & 7140 & 3570 + hcooc@xmath0h@xmath40 & 6250 & 4375 & 2188 + c@xmath0h@xmath40cooh & 9060 & 6342 & 3171 + c@xmath0h@xmath40cho & 5540 & 3878 & 1939 + ch@xmath4cn & 4680 & 3276 & 1638 + c@xmath0h@xmath4cn & 4640 & 3248 & 1624 + c@xmath0h@xmath40cn & 5540 & 3878 & 1939 + c@xmath4h@xmath105cn & 7240 & 5068 & 2534 + ch@xmath0nh@xmath0 & 4680 & 3276 & 1638 + ch@xmath4nh@xmath0 & 5130 & 3591 & 1796 + nh@xmath0cho & 5560 & 3892 & 1946 + nh@xmath0oh & 6810 & 4767 &
2384 + ( nh@xmath0)@xmath0 & 7910 & 5537 & 2769 + nh@xmath0och@xmath4 & 6460 & 4522 & 2261 + nh@xmath0ch@xmath0oh & 9040 & 6328 & 3164 + nh@xmath0ch@xmath0cn & 5480 & 3836 & 1918 + nh@xmath0ch@xmath0cooh & 10100 & 7070 & 3535 + nh@xmath0c@xmath0h@xmath40 & 7890 & 5523 & 2762 + nh@xmath0ch@xmath0cho & 6610 & 4627 & 2314 + ( nh@xmath0)@xmath0co & 7910 & 5537 & 2769 + nh@xmath0cooh & 9080 & 6356 & 3178 + nh@xmath0cocho & 6710 & 4697 & 2349 + ch@xmath4oconh@xmath0 & 7610 & 5327 & 2664 + ch@xmath0(oh)conh@xmath0 & 10190 & 7133 & 3567 + ch@xmath4conh@xmath0 & 6280 & 4396 & 2198 + ch@xmath4ocooh & 7620 & 5334 & 2667 + ch@xmath0(oh)cooh & 10200 & 7140 & 3570 + ch@xmath4ooh & 5350 & 3745 & 1873 + ch@xmath0(oh)@xmath0 & 7930 & 5551 &
2776 + ( ch@xmath4)@xmath0co & 3500 & 2450 & 1225 + ch@xmath4ococh@xmath4 & 4830 & 3381 & 1691 + ch@xmath4coch@xmath0oh & 7410 & 5187 & 2594 + ( ch@xmath4)@xmath0co & 6150 & 4305 & 2153 + ch@xmath4ococh@xmath0oh & 8730 & 6111 & 3056 + ( ch@xmath0oh)@xmath0co & 11300 & 7910 & 3955 lcll nh@xmath0ch@xmath0cooh + @xmath106 ( cr ) & @xmath31 & nh@xmath0ch@xmath0co + oh & @xmath107 + nh@xmath0ch@xmath0cooh + @xmath106 ( cr ) & @xmath31 & ch@xmath0nh@xmath0 + hoco & @xmath108 + nh@xmath0ch@xmath0cooh + @xmath106 ( cr ) & @xmath31 & ch@xmath0cooh + nh@xmath0 & @xmath109 + nh@xmath0ch@xmath0cooh + @xmath106 ( cr ) & @xmath31 & hnch@xmath0cooh + h & @xmath110 + nh@xmath0ch@xmath0cooh + @xmath106 & @xmath31 & nh@xmath0ch@xmath0co + oh & @xmath111s@xmath112 + nh@xmath0ch@xmath0cooh + @xmath106 & @xmath31 & ch@xmath0nh@xmath0 + hoco & @xmath113s@xmath112 + nh@xmath0ch@xmath0cooh + @xmath106 & @xmath31 & ch@xmath0cooh + nh@xmath0 & @xmath114s@xmath112 + nh@xmath0ch@xmath0cooh + @xmath106 & @xmath31 & hnch@xmath0cooh + h & @xmath111s@xmath112 + c@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh@xmath10 + c & @xmath115@xmath116s@xmath34 + h@xmath4@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh@xmath0@xmath10 + h@xmath0 & @xmath117@xmath116s@xmath34 + h@xmath4o@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh@xmath0@xmath10 + h@xmath0o & @xmath118@xmath116s@xmath34 + hco@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh@xmath0@xmath10 + co & @xmath119@xmath116s@xmath34 + he@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & ch@xmath0cooh@xmath10 + nh@xmath0 + he & @xmath120@xmath116s@xmath34 + he@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0co + oh@xmath10 + he & @xmath120@xmath116s@xmath34 + he@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & ch@xmath4nh@xmath10 + hoco + he & @xmath120@xmath116s@xmath34 + he@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & ch@xmath0cooh + nh@xmath0@xmath10 + he & @xmath120@xmath116s@xmath34 + he@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0co@xmath10 + oh + he & @xmath120@xmath116s@xmath34 + he@xmath10 + nh@xmath0ch@xmath0cooh & @xmath31 & ch@xmath0nh@xmath0 + cooh@xmath10 + he & @xmath120@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath0@xmath10 + e@xmath121 & @xmath31 & nh@xmath0 + ch@xmath0 + co@xmath0 + h@xmath0 & @xmath122@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath0@xmath10 + e@xmath121 & @xmath31 & nh@xmath0 + ch@xmath0 + hoco + h & @xmath123@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath0@xmath10 + e@xmath121 & @xmath31 & ch@xmath0nh@xmath0 + co + oh + h & @xmath123@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath0@xmath10 + e@xmath121 & @xmath31 & ch@xmath0nh@xmath0 + co@xmath0 + h@xmath0 & @xmath123@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath0@xmath10 + e@xmath121 & @xmath31 & ch@xmath0nh@xmath0 + hoco + h & @xmath123@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath0@xmath10 + e@xmath121 & @xmath31 & ch@xmath0cooh + nh@xmath0 + h & @xmath123@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath0@xmath10 + e@xmath121 & @xmath31 & nh@xmath0ch@xmath0co + h@xmath0o & @xmath124@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath0@xmath10 + e@xmath121 & @xmath31 & ch@xmath0nh@xmath0 + hcooh & @xmath124@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath0@xmath10 + e@xmath121 & @xmath31 & nh@xmath0ch@xmath0cooh + h & @xmath124@xmath116s@xmath34 + c@xmath10 + hnch@xmath0cooh & @xmath31 & hnch@xmath0cooh@xmath10 + c & @xmath115@xmath116s@xmath34 + h@xmath4@xmath10 + hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh@xmath10 + h@xmath0 & @xmath117@xmath116s@xmath34 + h@xmath4o@xmath10 + hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh@xmath10 + h@xmath0o & @xmath118@xmath116s@xmath34 + hco@xmath10 + hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh@xmath10 + co & @xmath125@xmath116s@xmath34 + he@xmath10 + hnch@xmath0cooh & @xmath31 & ch@xmath0nh + cooh@xmath10 + he & @xmath126@xmath116s@xmath34 + he@xmath10 + hnch@xmath0cooh & @xmath31 & ch@xmath0cooh + nh@xmath10 + he & @xmath126@xmath116s@xmath34 + he@xmath10 + hnch@xmath0cooh & @xmath31 & ch@xmath0nh@xmath10 + hoco + he & @xmath126@xmath116s@xmath34 + he@xmath10 + hnch@xmath0cooh & @xmath31 & ch@xmath0cooh@xmath10 + nh + he & @xmath126@xmath116s@xmath34 + hnch@xmath0cooh + @xmath106 ( cr ) & @xmath31 & ch@xmath0nh + hoco & @xmath108 + hnch@xmath0cooh + @xmath106 ( cr ) & @xmath31 & ch@xmath0cooh + nh & @xmath109 + hnch@xmath0cooh + @xmath106 & @xmath31 & ch@xmath0nh + hoco & @xmath113s@xmath112 + hnch@xmath0cooh + @xmath106 & @xmath31 & ch@xmath0cooh + nh & @xmath127s@xmath112 + h + hnch@xmath0cooh & @xmath31 & ch@xmath0nh + hcooh & @xmath128@xmath116s@xmath34 + ch@xmath0oh + hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh + h@xmath0co & @xmath128@xmath116s@xmath34 + ch@xmath4o + hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh + h@xmath0co & @xmath128@xmath116s@xmath34 + hoco + hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh + co@xmath0 & @xmath128@xmath116s@xmath34 + hco + hnch@xmath0cooh & @xmath31 & nh@xmath0ch@xmath0cooh + co & @xmath128@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath10 + e@xmath121 & @xmath31 & nh@xmath0 + ch@xmath0 + co@xmath0 + h & @xmath129@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath10 + e@xmath121 & @xmath31 & nh@xmath0 + ch@xmath0 + co + oh & @xmath123@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath10 + e@xmath121 & @xmath31 & ch@xmath0nh@xmath0 + co@xmath0 + h & @xmath123@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath10 + e@xmath121 & @xmath31 & ch@xmath0nh@xmath0 + co + oh & @xmath123@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath10 + e@xmath121 & @xmath31 & nh@xmath0co + ch@xmath0 + oh & @xmath123@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath10 + e@xmath121 & @xmath31 & ch@xmath0nh@xmath0 + hoco & @xmath124@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath10 + e@xmath121 & @xmath31 & ch@xmath0cooh + nh@xmath0 & @xmath124@xmath116s@xmath34 + nh@xmath0ch@xmath0cooh@xmath10 + e@xmath121 & @xmath31 & hnch@xmath0cooh + h & @xmath124@xmath116s@xmath34 + nh@xmath0oh@xmath10 + ch@xmath4cooh & @xmath31 & nh@xmath0ch@xmath0cooh@xmath10 + h@xmath0o & @xmath128@xmath116s@xmath34 + nh@xmath4oh@xmath10 + ch@xmath4cooh & @xmath31 & nh@xmath0ch@xmath0cooh@xmath0@xmath10 + h@xmath0o & @xmath130@xmath116s@xmath34 lcccccc co & 8 & 16 & @xmath13712 & 29 & 13 & 57 + co@xmath0 & 13 & 22 & 14 & 29 & 13 & 18 + ch@xmath54 & 1.5 & 2 & 2 & 5 & 2 & 1.9 + h@xmath0co & 6 & 4 & @xmath1373 & @xmath1382 & @xmath1382 & 1.6 + ch@xmath4oh & 18 & 5 & @xmath1374 & 3 & 4 & 6.9 + nh@xmath4 & 15 & 13 & 2030 & 5 & 5 & 18 + nh@xmath0oh & & & & & & 0 / 0.55 + lcccccccccccc h@xmath0o & & 1.6e-4 & & 392 & & 1.6e-4 & & 395 & & 1.6e-4 & & 398 + co & & 9.5e-5 & & 400 & & 1.0e-4 & & 391 & & 1.2e-4 & & 398 + co@xmath0 & & 2.9e-5 & & 58 & & 3.0e-5 & & 56 & & 3.5e-5 & & 55 + ch@xmath54 & & 4.4e-6 & & 400 & & 4.4e-6 & & 398 & & 3.1e-6 & & 27 + hco@xmath10 & & 7.1e-10 & & 36 & & 7.7e-10 & & 96 & & 8.7e-10 & & 56 + h@xmath0co & & 4.6e-6 & & 43 & & 3.7e-6 & & 42 & & 5.5e-7 & & 39 + ch@xmath4oh & & 1.1e-5 & & 123 & & 9.1e-6 & & 122 & & 3.8e-6 & & 121 + nh@xmath4 & & 2.7e-5 & & 124 & & 2.7e-5 & & 122 & & 2.7e-5 & & 122 + h@xmath0o@xmath0 & & 9.0e-7 & & 127 & & 7.6e-7 & & 123 & & 3.7e-7 & & 122 + oh & & 4.7e-8 & & 70 & & 1.4e-7 & & 50 & & 2.9e-8 & & 40 + o@xmath0 & & 2.2e-7 & & 10 & & 5.4e-7 & & 109 & & 5.5e-6 & & 115 + o@xmath4 & & 8.8e-9 & & 10 & & 8.8e-9 & & 10 & & 8.8e-9 & & 10 + no & & 2.2e-6 & & 97 & & 2.0e-6 & & 54 & & 1.5e-6 & & 48 + hno & & 6.1e-7 & & 42 & & 4.0e-7 & & 41 & & 1.7e-7 & & 39 + hnco & & 8.5e-9 & & 400 & & 1.2e-7 & & 238 & & 1.1e-8 & & 222 + ocn & & 2.0e-8 & & 108 & & 9.4e-8 & & 59 & & 8.9e-9 & & 40 + so & & 5.4e-9 & & 400 & & 2.9e-8 & & 395 & & 3.8e-8 & & 400 + so@xmath0 & & 7.4e-8 & & 76 & & 7.6e-8 & & 81 & & 7.8e-8 & & 127 + c@xmath0h & & 1.4e-9 & & 27 & & 1.9e-9 & & 26 & & 2.5e-9 & & 26 + c@xmath0h@xmath0 & & 4.5e-8 & & 400 & & 6.9e-8 & & 53 & & 7.4e-8 & & 205 + c@xmath0h@xmath54 & & 2.5e-8 & & 71 & & 1.8e-8 & & 398 & & 1.2e-7 & & 156 + c@xmath0h@xmath104 & & 4.6e-8 & & 94 & & 1.6e-7 & & 93 & & 9.2e-7 & & 90 + c@xmath4h@xmath104 & & 1.7e-9 & & 400 & & 1.4e-8 & & 398 & & 6.6e-8 & & 285 + c@xmath4h@xmath139 & & 1.1e-7 & & 132 & & 1.6e-7 & & 130 & & 3.4e-7 & & 129 + cn & & 2.6e-9 & & 103 & & 1.2e-8 & & 46 & & 4.3e-8 & & 41 + hcn & & 3.9e-7 & & 338 & & 3.3e-7 & & 398 & & 8.2e-7 & & 400 + hnc & & 1.5e-7 & & 400 & & 1.9e-7 & & 398 & & 7.1e-7 & & 400 + h@xmath0cn & & 2.1e-10 & & 65 & & 1.6e-11 & & 47 & & 1.1e-11 & & 27 + ch@xmath4cn & & 4.9e-9 & & 400 & & 2.1e-9 & & 92 & & 1.3e-9 & & 90 + c@xmath0h@xmath4cn & & 1.1e-8 & & 108 & & 5.6e-9 & & 398 & & 5.6e-9 & & 202 + c@xmath0h@xmath40cn & & 5.7e-9 & & 123 & & 7.7e-8 & & 122 & & 3.0e-8 & & 122 + c@xmath4h@xmath105cn & & 6.0e-9 & & 153 & & 1.0e-8 & & 149 & & 8.4e-10 & & 144 + ch@xmath0co & & 9.0e-8 & & 78 & & 1.9e-8 & & 77 & & 1.1e-8 & & 203 + ch@xmath4cho & & 3.1e-9 & & 400 & & 9.4e-9 & & 337 & & 1.4e-8 & & 154 + c@xmath0h@xmath40cho & & 6.9e-9 & & 123 & & 1.2e-8 & & 120 & & 2.6e-8 & & 116 + ch@xmath0nh & & 1.1e-8 & & 400 & & 2.4e-8 & & 398 & & 1.5e-8 & & 161 + ch@xmath4nh@xmath0 & & 8.0e-8 & & 114 & & 3.6e-8 & & 105 & & 6.8e-9 & & 100 + nh@xmath0cho & & 3.9e-7 & & 123 & & 1.5e-7 & & 122 & & 2.9e-9 & & 117 + nh@xmath0cn & & 2.3e-8 & & 267 & & 3.6e-8 & & 205 & & 3.3e-8 & & 195 + nh@xmath0oh & & 1.6e-10 & & 145 & & 6.0e-10 & & 141 & & 1.7e-9 & & 135 + c@xmath0h@xmath40oh & & 5.9e-8 & & 135 & & 5.8e-8 & & 131 & & 4.9e-8 & & 127 + ch@xmath4och@xmath4 & & 4.8e-8 & & 400 & & 6.3e-8 & & 301 & & 2.8e-8 & & 143 + hcooch@xmath4 & & 9.2e-8 & & 120 & & 3.7e-8 & & 111 & & 3.1e-9 & & 103 + hcooc@xmath0h@xmath40 & & 3.4e-9 & & 136 & & 4.1e-10 & & 131 & & 4.1e-12 & & 122 + hcooh & & 4.8e-8 & & 123 & & 4.0e-8 & & 122 & & 4.3e-8 & & 121 + ch@xmath4cooh & & 1.0e-10 & & 134 & & 5.0e-10 & & 131 & & 7.2e-9 & & 130 + c@xmath0h@xmath40cooh & & 1.1e-9 & & 195 & & 2.7e-10 & & 190 & & 3.3e-9 & & 183 + ch@xmath4cocho & & 3.4e-8 & & 81 & & 3.0e-8 & & 76 & & 3.5e-9 & & 74 + ( ch@xmath4)@xmath0co & & 1.5e-9 & & 72 & & 1.0e-8 & & 68 & & 1.9e-9 & & 66 + ch@xmath4conh@xmath0 & & 2.8e-8 & & 134 & & 2.2e-8 & & 130 & & 4.2e-8 & & 125 + ch@xmath4ococh@xmath4 & & 1.2e-9 & & 102 & & 7.9e-11 & & 94 & & 1.0e-10 & & 92 + ch@xmath0(oh)coch@xmath4 & & 5.8e-8 & & 159 & & 6.8e-9 & & 154 & & 2.5e-12 & & 147 + hocooh & & 4.2e-13 & & 173 & & 1.1e-11 & & 167 & & 1.1e-10 & & 161 + hcococho & & 1.2e-8 & & 91 & & 5.7e-9 & & 85 & & 2.7e-9 & & 82 + ohccho & & 6.3e-7 & & 69 & & 3.3e-7 & & 64 & & 4.3e-7 & & 62 + ohccooh & & 5.9e-11 & & 147 & & 1.2e-7 & & 142 & & 2.2e-9 & & 138 + nh@xmath0cocho & & 6.6e-8 & & 145 & & 2.6e-8 & & 141 & & 3.4e-8 & & 136 + nh@xmath0cooh & & 1.8e-10 & & 193 & & 1.2e-9 & & 188 & & 9.8e-10 & & 182 + ( nh@xmath0)@xmath0co & & 7.1e-8 & & 169 & & 1.2e-7 & & 163 & & 3.2e-8 & & 157 + nh@xmath0ch@xmath0cho & & 7.5e-9 & & 141 & & 1.2e-8 & & 136 & & 2.9e-9 & & 134 + nh@xmath0ch@xmath0cn & & 4.3e-9 & & 123 & & 1.2e-8 & & 122 & & 4.7e-9 & & 118 + nh@xmath0ch@xmath0cooh & & 8.4e-11 & & 216 & & 4.6e-10 & & 207 & & 8.1e-9 & & 203 + nh@xmath0c@xmath0h@xmath40 & & 6.0e-9 & & 170 & & 1.0e-9 & & 164 & & 6.9e-10 & & 160 + ( nh@xmath0)@xmath0 & & 1.6e-8 & & 172 & & 3.2e-8 & & 165 & & 3.3e-8 & & 160 + ch@xmath4onh@xmath0 & & 2.2e-8 & & 140 & & 1.5e-8 & & 136 & & 4.1e-9 & & 131 + ch@xmath0(oh)nh@xmath0 & & 5.5e-8 & & 193 & & 8.5e-8 & & 188 & & 7.6e-8 & & 182 + ch@xmath4ococho & & 4.5e-11 & & 118 & & 5.1e-13 & & 128 & & 1.6e-12 & & 105 + ch@xmath4ocooh & & 1.5e-12 & & 164 & & 3.8e-12 & & 160 & & 1.2e-15 & & 151 + ch@xmath4oconh@xmath0 & & 3.2e-9 & & 163 & & 3.2e-9 & & 158 & & 1.0e-9 & & 153 + ( ch@xmath4o)@xmath0co & & 3.4e-12 & & 137 & & 3.4e-12 & & 135 & & 1.8e-13 & & 127 + ch@xmath4ococh@xmath0oh & & 1.8e-9 & & 187 & & 4.3e-10 & & 180 & & 1.0e-14 & & 125 + ch@xmath4ooh & & 7.7e-13 & & 123 & & 3.9e-13 & & 120 & & 9.2e-14 & & 112 + ( ch@xmath4o)@xmath0 & & 3.2e-9 & & 107 & & 3.2e-11 & & 99 & & 1.7e-13 & & 96 + ch@xmath4och@xmath0oh & & 1.5e-8 & & 166 & & 1.4e-9 & & 160 & & 1.1e-13 & & 151 + ch@xmath0(oh)cho & & 2.9e-7 & & 143 & & 1.1e-7 & & 141 & & 1.9e-8 & & 147 + ch@xmath0(oh)cocho & & 3.3e-8 & & 240 & & 5.0e-8 & & 209 & & 1.1e-8 & & 195 + ch@xmath0(oh)cooh & & 4.5e-10 & & 216 & & 6.6e-11 & & 202 & & 4.1e-15 & & 134 + ch@xmath0(oh)conh@xmath0 & & 1.3e-7 & & 215 & & 2.3e-7 & & 207 & & 4.9e-8 & & 203 + ( ch@xmath0oh)@xmath0co & & 4.8e-9 & & 222 & & 2.2e-10 & & 203 & & 5.8e-13 & & 191 + ch@xmath0(oh)@xmath0 & & 1.0e-9 & & 169 & & 7.6e-10 & & 164 & & 5.6e-14 & & 154 + ( ch@xmath0oh)@xmath0 & & 3.7e-9 & & 206 & & 1.1e-9 & & 198 & & 4.1e-12 & & 158 + lccccccc methanol & 220.079 & 96.61 & 113 & 21.9@xmath140 & 1.69 & 9.03 & 18 + ( ch@xmath4oh ) & 230.027 & 39.83 & 118 & 21.0@xmath140 & 1.69 & 9.00 & 17 + & 234.683 & 60.92 & 116 & 20.7@xmath140 & 1.76 & 9.40 & 18 + & 338.125 & 78.08 & 107 & 14.4@xmath140 & 4.24 & 22.6 & 56 + & 338.345 & 70.55 & 106 & 14.4@xmath140 & 4.30 & 23.0 & 71 + & 340.141 & 44.67 & 120 & 14.3@xmath140 & 3.34 & 17.8 & 27 + & 356.007 & 295.3 & 107 & 13.9@xmath140 & 4.14 & 22.1 & 29 + methyl formate & 220.190 & 103.1 & 246 & 21.9@xmath140 & 0.897 & 4.78 & 6.4 + ( hcooch@xmath4 ) & 221.979 & 154.7 & 248 & 21.7@xmath140 & 0.980 & 5.23 & 5.7 + & 226.719 & 120.2 & 247 & 21.3@xmath140 & 1.00 & 5.33 & 14 + & 338.356 & 267.2 & 272 & 14.4@xmath140 & 1.53 & 8.14 & 8.8 + & 338.396 & 257.8 & 268 & 14.4@xmath140 & 1.63 & 8.70 & 11 + & 346.660 & 320.4 & 286 & 14.0@xmath140 & 1.20 & 6.39 & 7.5 + & 346.675 & 320.4 & 255 & 14.0@xmath140 & 1.86 & 9.90 & 11 + & 348.050 & 266.1 & 266 & 14.0@xmath140 & 1.76 & 9.37 & 12 + glycine & 218.105 & 201.2 & 1.33 & 22.0@xmath140 & 2.01e-4 & 0.00107 & + ( nh@xmath0ch@xmath0cooh ) & & & & 11.4@xmath141 & 7.46e-4 & 0.00398 & + & & & & 1.00@xmath142 & 0.0869 & 0.464 & + & & & & 0.400@xmath142 & 0.345 & 1.84 & + & 223.923 & 211.9 & 1.36 & 21.5@xmath140 & 2.12e-4 & 0.00113 & + & & & & 11.2@xmath141 & 7.89e-4 & 0.00421 & + & & & & 1.00@xmath142 & 0.0878 & 0.468 & + & & & & 0.400@xmath142 & 0.349 & 1.86 & + & 229.740 & 223.0 & 1.39 & 21.0@xmath140 & 2.24e-4 & 0.00120 & + & & & & 10.9@xmath141 & 8.35e-4 & 0.00445 & + & & & & 1.00@xmath142 & 0.0885 & 0.472 & + & & & & 0.400@xmath142 & 0.35 & 1.88 & + & 235.557 & 234.3 & 1.42 & 20.7@xmath140 & 2.33e-4 & 0.00124 & + & & & & 10.6@xmath141 & 8.82e-4 & 0.00470 & + & & & & 1.00@xmath142 & 0.0891 & 0.475 & + & & & & 0.400@xmath142 & 0.356 & 1.90 & + & 241.373 & 245.9 & 1.44 & 20.3@xmath140 & 2.42e-4 & 0.00129 & + & & & & 10.4@xmath141 & 9.31e-4 & 0.00496 & + & & & & 1.00@xmath142 & 0.0893 & 0.477 & + & & & & 0.400@xmath142 & 0.359 & 1.91 & + & 247.190 & 257.7 & 1.47 & 20.0@xmath140 & 2.51e-4 & 0.00134 & + & & & & 10.1@xmath141 & 9.81e-4 & 0.00523 & + & & & & 1.00@xmath142 & 0.0895 & 0.477 & + & & & & 0.400@xmath142 & 0.360 & 1.92 & + lccccccc methanol & 220.079 & 96.61 & 112 & 21.9@xmath140 & 1.71 & 9.14 & 18 + ( ch@xmath4oh ) & 230.027 & 39.83 & 118 & 21.0@xmath140 & 1.72 & 9.19 & 17 + & 234.683 & 60.92 & 116 & 20.7@xmath140 & 1.80 & 9.59 & 18 + & 338.125 & 78.08 & 106 & 14.4@xmath140 & 4.17 & 22.2 & 56 + & 338.345 & 70.55 & 106 & 14.4@xmath140 & 4.21 & 22.5 & 71 + & 340.141 & 44.67 & 122 & 14.3@xmath140 & 3.44 & 18.4 & 27 + & 356.007 & 295.3 & 106 & 13.9@xmath140 & 4.25 & 22.7 & 29 + methyl formate & 220.190 & 103.1 & 133 & 21.9@xmath140 & 0.458 & 2.44 & 6.4 + ( hcooch@xmath4 ) & 221.979 & 154.7 & 166 & 21.7@xmath140 & 0.514 & 2.74 & 5.7 + & 226.719 & 120.2 & 148 & 21.3@xmath140 & 0.520 & 2.77 & 14 + & 338.356 & 267.2 & 165 & 14.4@xmath140 & 0.727 & 3.88 & 8.8 + & 338.396 & 257.8 & 170 & 14.4@xmath140 & 0.789 & 4.21 & 11 + & 346.660 & 320.4 & 148 & 14.0@xmath140 & 0.534 & 2.85 & 7.5 + & 346.675 & 320.4 & 212 & 14.0@xmath140 & 0.926 & 4.94 & 11 + & 348.050 & 266.1 & 176 & 14.0@xmath140 & 0.854 & 4.55 & 12 + glycine & 218.105 & 201.2 & 2.81 & 22.0@xmath140 & 7.67e-4 & 0.00409 & + ( nh@xmath0ch@xmath0cooh ) & & & & 11.4@xmath141 & 0.00285 & 0.0152 & + & & & & 1.00@xmath142 & 0.321 & 1.71 & + & & & & 0.400@xmath142 & 1.11 & 5.94 & + & 223.923 & 211.9 & 2.87 & 21.5@xmath140 & 8.08e-4 & 0.00431 & + & & & & 11.2@xmath141 & 0.00301 & 0.0160 & + & & & & 1.00@xmath142 & 0.323 & 1.72 & + & & & & 0.400@xmath142 & 2.10 & 11.2 & + & 229.740 & 223.0 & 2.93 & 21.0@xmath140 & 8.52e-4 & 0.00454 & + & & & & 10.9@xmath141 & 0.00317 & 0.0169 & + & & & & 1.00@xmath142 & 0.325 & 1.73 & + & & & & 0.400@xmath142 & 1.13 & 6.05 & + & 235.557 & 234.3 & 2.98 & 20.7@xmath140 & 8.83e-4 & 0.00471 & + & & & & 10.6@xmath141 & 0.00334 & 0.0178 & + & & & & 1.00@xmath142 & 0.326 & 1.74 & + & & & & 0.400@xmath142 & 1.14 & 6.08 & + & 241.373 & 245.9 & 3.02 & 20.3@xmath140 & 9.14e-4 & 0.00487 & + & & & & 10.4@xmath141 & 0.00351 & 0.0187 & + & & & & 1.00@xmath142 & 0.326 & 1.74 & + & & & & 0.400@xmath142 & 1.14 & 6.10 & + & 247.190 & 257.7 & 3.07 & 20.0@xmath140 & 9.44e-4 & 0.00503 & + & & & & 10.1@xmath141 & 0.00369 & 0.0197 & + & & & & 1.00@xmath142 & 0.325 & 1.73 & + & & & & 0.400@xmath142 & 1.14 & 6.10 & + lccccccc methanol & 220.079 & 96.61 & 117 & 21.9@xmath140 & 1.65 & 8.82 & 18 + ( ch@xmath4oh ) & 230.027 & 39.83 & 127 & 21.0@xmath140 & 1.63 & 8.67 & 17 + & 234.683 & 60.92 & 126 & 20.7@xmath140 & 1.73 & 9.25 & 18 + & 338.125 & 78.08 & 107 & 14.4@xmath140 & 3.97 & 21.2 & 56 + & 338.345 & 70.55 & 107 & 14.4@xmath140 & 4.00 & 21.3 & 71 + & 340.141 & 44.67 & 116 & 14.3@xmath140 & 2.99 & 15.9 & 27 + & 356.007 & 295.3 & 108 & 13.9@xmath140 & 4.16 & 22.2 & 29 + methyl formate & 220.190 & 103.1 & 0.773 & 21.9@xmath140 & 0.0152 & 0.0813 & 6.4 + ( hcooch@xmath4 ) & 221.979 & 154.7 & 0.792 & 21.7@xmath140 & 0.0161 & 0.0856 & 5.7 + & 226.719 & 120.2 & 0.818 & 21.3@xmath140 & 0.0171 & 0.0910 & 14 + & 338.356 & 267.2 & 0.372 & 14.4@xmath140 & 0.0174 & 0.0929 & 8.8 + & 338.396 & 257.8 & 0.416 & 14.4@xmath140 & 0.0194 & 0.104 & 11 + & 346.660 & 320.4 & 0.220 & 14.0@xmath140 & 0.0111 & 0.0593 & 7.5 + & 346.675 & 320.4 & 0.410 & 14.0@xmath140 & 0.0207 & 0.111 & 11 + & 348.050 & 266.1 & 0.417 & 14.0@xmath140 & 0.0208 & 0.111 & 12 + glycine & 218.105 & 201.2 & 10.2 & 22.0@xmath140 & 0.00804 & 0.0429 & + ( nh@xmath0ch@xmath0cooh ) & & & & 11.4@xmath141 & 0.0299 & 0.159 & + & & & & 1.00@xmath142 & 3.20 & 17.0 & + & & & & 0.400@xmath142 & 8.76 & 46.7 & + & 223.923 & 211.9 & 10.2 & 21.5@xmath140 & 0.00844 & 0.0450 & + & & & & 11.2@xmath141 & 0.0314 & 0.167 & + & & & & 1.00@xmath142 & 3.20 & 17.1 & + & & & & 0.400@xmath142 & 8.80 & 46.9 & + & 229.740 & 223.0 & 10.2 & 21.0@xmath140 & 0.00886 & 0.0472 & + & & & & 10.9@xmath141 & 0.0329 & 0.176 & + & & & & 1.00@xmath142 & 3.21 & 17.1 & + & & & & 0.400@xmath142 & 8.82 & 47.0 & + & 235.557 & 234.3 & 10.2 & 20.7@xmath140 & 0.00914 & 0.0488 & + & & & & 10.6@xmath141 & 0.0345 & 0.184 & + & & & & 1.00@xmath142 & 3.20 & 17.1 & + & & & & 0.400@xmath142 & 8.81 & 47.0 & + & 241.373 & 245.9 & 10.2 & 20.3@xmath140 & 0.00941 & 0.0502 & + & & & & 10.4@xmath141 & 0.0361 & 0.193 & + & & & & 1.00@xmath142 & 3.18 & 17.0 & + & & & & 0.400@xmath142 & 8.78 & 46.8 & + & 247.190 & 257.7 & 10.1 & 20.0@xmath140 & 0.00968 & 0.0516 & + & & & & 10.1@xmath141 & 0.0378 & 0.202 & + & & & & 1.00@xmath142 & 3.16 & 16.9 & + & & & & 0.400@xmath142 & 8.73 & 46.6 & + lcccccc hydroxylamine & 151.102 & 46.33 & 0.0354 & 41.5 & 4.37e-5 & 4.20e-4 + ( nh@xmath0oh ) & 151.118 & 14.51 & 0.0391 & 41.5 & 5.01e-5 & 4.81e-4 + & 164.341 & 116.7 & 8.05e-4 & 38.2 & 1.08e-6 & 1.03e-5 + & 164.627 & 94.92 & 8.10e-4 & 38.1 & 1.12e-6 & 1.07e-5 + & 164.883 & 75.60 & 7.93e-4 & 38.0 & 1.13e-6 & 1.08e-5 lcccccc hydroxylamine & 151.102 & 46.33 & 34.7 & 41.5 & 0.0424 & 0.407 + ( nh@xmath0oh ) & 151.118 & 14.51 & 37.7 & 41.5 & 0.0494 & 0.475 + & 164.341 & 116.7 & 0.883 & 38.2 & 0.00111 & 0.0107 + & 164.627 & 94.92 & 0.885 & 38.1 & 0.00115 & 0.0111 + & 164.883 & 75.60 & 0.865 & 38.0 & 0.00117 & 0.0112 atkinson , r. , baulch , d. l. , cox , r. a. , crowley , j. n. , hampson , r. f. , jr .
, kerr , j. a. , rossi , m. j. & troe , j. 2001 , _ summary of evaluated kinetic and photochemical data for atmospheric chemistry _ ,
iupac subcommittee on gas kinetic data evaluation for atmospheric chemistry
web version demore , w. b. , sander , s. p. , golden , d. m. , hampson , r. f. , kurylo , m. j. , howard , c. j. , ravishankara , a. r. , kolb , c. e. & molina , m. j. 1997 , _ chemical kinetics and photochemical data for use in stratospheric modeling .
evaluation number 12 _ , jpl publication 97 - 4 , 1 gibb , e. l. , whittet , d. c. b. , schutte , w. a. , boogert , a. c. a. , chiar , j. e. , ehrenfreund , p. , gerakines , p. a. , keane , j. v. , tielens , a. g. g. m. , van dishoeck , e. f. , & kerkhof , o. 2000 , , 536 , 347 mason , n. j. , dawes , a. , holtom , p. d. , mukerji , r. j. , davis , m. p. , sivaraman , b. , kaiser , r. i. , hoffmann , s. v. & shaw , d. a. 2006 , faraday discuss .
133 , chemical evolution of the universe ( cambridge : rsc pub . ) , 311 | a new chemical model is presented that simulates fully - coupled gas - phase , grain - surface and bulk - ice chemistry in hot cores .
glycine ( nh@xmath0ch@xmath0cooh ) , the simplest amino acid , and related molecules such as glycinal , propionic acid and propanal , are included in the chemical network .
glycine is found to form in moderate abundance within and upon dust - grain ices via three radical - addition mechanisms , with no single mechanism strongly dominant .
glycine production in the ice occurs over temperatures @xmath140120 k. peak gas - phase glycine fractional abundances lie in the range @xmath2@xmath3 , occuring at @xmath1200 k , the evaporation temperature of glycine . a gas - phase mechanism for glycine production is tested and found insignificant , even under optimal conditions .
a new spectroscopic radiative - transfer model is used , allowing the translation and comparison of the chemical - model results with observations of specific sources .
comparison with the nearby hot - core source ngc 6334 irs1 shows excellent agreement with integrated line intensities of observed species , including methyl formate .
the results for glycine are consistent with the current lack of a detection of this molecule toward other sources ; the high evaporation temperature of glycine renders the emission region extremely compact .
glycine detection with alma is predicted to be highly plausible , for bright , nearby sources with narrow emission lines .
photodissociation of water and subsequent hydrogen - abstraction from organic molecules by oh , and nh@xmath0 , are crucial to the build - up of complex organic species in the ice .
the inclusion of alternative branches within the network of radical - addition reactions appears important to the abundances of hot - core molecules ; less favorable branching ratios may remedy the anomalously high abundance of glycolaldehyde predicted by this and previous models . |
this paper considers the stability of toroidal magnetic fields in rotating radiation zones of stars and focuses on the destabilizing effect of finite thermal diffusion and the stabilizing effect of differential rotation .
the equations for the linear stability of toroidal magnetic fields under the influences of basic rotation and gravitational buoyancy are solved for two different latitudinal profiles of the toroidal field , i.e. with the two possible symmetries with respect to the equator .
our equations are global in both horizontal dimensions but they are local in radius , i.e. the short - wave approximation @xmath6 is used ( @xmath7 is the radial wave number ) . we know that a toroidal magnetic field @xmath8 which fulfills the condition @xmath9 is unstable against nonaxisymmetric disturbances in an ideal and incompressible medium ( tayler 1957 , 1973 , vandakurov 1972 , acheson 1978 ) . here @xmath10 is the distance from the axis where @xmath11 .
this ` tayler instability ' ( ti ) is suppressed by rigid rotation unless the magnetic field is strong enough to fulfill the condition @xmath12 ( pitts and tayler 1985 ) with the magnetic frequency @xmath13 and the rotational frequency @xmath14 of the star . for a simple spherical model with @xmath15 const ( one magnetic belt with maximum in the equatorial plane ) the dashed line in figure [ f0 ]
demonstrates this situation where the solar value of the magnetic prandtl number @xmath16 of @xmath17 was used . here
@xmath18 and @xmath19 are the microscopic values of the viscosity and magnetic diffusivity .
the growth rate @xmath20 at the vertical axis of the plot is normalized with the basic rotation so that the result of the calculation is @xmath21 the instability thus exists only for subalfvnic rotation and it is obviously very fast , i.e. @xmath22 . due to the assumed isothermal state of the medium no buoyancy - term exists in this calculation .
generally , in stably - stratified plasma the real buoyancy is ` negative ' and should stabilize the magnetic instability .
this is indeed the case .
the solid line in figure [ f0 ] results from a model with adiabatic density fluctuations in boussinesq approximation , i.e. for vanishing thermal diffusivity @xmath23 .
the resulting field strength for onset of instability slightly exceeds its value without buoyancy .
surprisingly , the stabilization of the ti by ` negative buoyancy ' is a small effect .
the opposite case with @xmath24 provides results identical with those for the incompressible model ( dashed line ) .
no temperature fluctuations can develop for @xmath25 so that the stabilizing effect of buoyancy does not apply .
after inspection of the extremes @xmath26 and @xmath27 one could believe that the inclusion of the thermal equation into the instability theory is not significant .
we know , however , that the heat conductivity in the stellar radiation zones is the most effective dissipation process , far from the adiabatic limit , i.e. @xmath28 the prandtl number of the gas is of order @xmath29 which yields a roberts number @xmath30 of @xmath31 .
note that liquid metals of mhd laboratory experiments have @xmath32 .
the nonadiabatic models with small prandtl number ( see equation ( [ eq5 ] ) ) will lead to a surprising result .
the buoyancy does _ not _ suppress the instability any longer but it creates another instability which exists for much weaker magnetic fields .
acheson ( 1978 ) found that the roberts number q enters the stability equations in a rather complicated manner so that one term acts stabilizing while another one acts destabilizing .
`` it is natural , therefore , to expect quite complicated changes in the stability properties of the system ... '' .
all terms , however , vanish when @xmath33
. we shall find that indeed finite thermal conductivities lead to a destabilization of magnetic fields with @xmath1 much smaller than @xmath2 ; the growth rates of this doubly - diffusive instability , however , are small .
neither time stepping codes nor technical experiments can find such slow instabilities with growth rates of order @xmath34 .
of course , such timescales are still very short compared to stellar evolutionary times but the physical relevance of such instabilities is still in question ( cally 2003 ) .
critical for the astrophysical meaning of nonaxisymmetric magnetic instabilities is also the differential rotation .
one expects that the nonuniform rotation acts against the excitation of nonaxisymmetric modes if they are already excited for rigid rotation . in the solar tachocline the rotation law is per definition nonuniform .
the differential rotation itself is needed for the production of the strong toroidal fields leading to the tachocline phenomenon .
a possible magnetic instability of these toroidal fields ( `` tayler - spruit dynamo '' ) should basically be stabilized by the nonuniformity of the rotation .
one has thus to check how important the differential rotation is to stabilize / destabilize the toroidal magnetic fields .
there are indications that the influence of the differential rotation strongly depends on the symmetry type of the toroidal field patterns with respect to the equator . in the present paper only fields with antisymmetry with respect to the equator are considered as subject to differential rotation .
the equations for small disturbances ( @xmath35 ) of background magnetic ( @xmath36 ) and velocity ( @xmath37 ) fields are @xmath38 for the velocity fluctuations @xmath39 , @xmath40 for the magnetic fluctuations @xmath41 and @xmath42 for the entropy fluctuations
@xmath43 related to the density fluctuations by @xmath44 .
the basic flow is a rotation with uniform angular velocity @xmath2 and the mean magnetic field @xmath45 has only a toroidal component @xmath8 .
the mathematical method for solving the equation system ( [ 5])([7 ] ) has been described earlier ( kitchatinov and rdiger 2008 ) . in this previous paper
the instability has been considered for the case of rigid rotation and for given heat conductivity . in the present paper the _
destabilizing _ role of the heat conductivity is demonstrated and first results for the _ stabilizing _ role of differential rotation are given .
the equations are global in horizontal dimensions but local in radius .
the radial scale of disturbances is assumed as short and their dependence on radius @xmath46 is taken in the form of fourier modes @xmath47 .
there are only symmetry conditions along the polar axes which will be fulfilled by the series expansions after legendre polynomials .
only the modes with @xmath0 are considered . then at the axes
the radial components of flow and field must vanish and also the @xmath48-derivatives of the horizontal components ( see elstner _
_ 1990 , gilman and fox 1997 ) . for given radial wave number and for given field amplitude ( in units of the basic rotation velocity ) the resulting eigenfrequency is computed including growth rate ( imaginary part of @xmath49 ) of the instability and drift rate of the eigensolutions ( real part of @xmath49 ) .
the key parameter for the effect of the stable stratification is @xmath50 where @xmath51 is the buoyancy frequency @xmath52 in stellar radiation zones it is @xmath53 .
for the most unstable modes we find @xmath54 so that the radial scale of the modes is indeed small , i.e. @xmath6 .
our equations include finite diffusion via @xmath55 with @xmath19 and @xmath18 as the magnetic resistivity and the kinematic viscosity .
the thermal conductivity @xmath23 enters the equations in the normalized form @xmath56 which is the free parameter in the following discussion .
the present article focuses on the effect of thermal diffusivity from the following reason .
the current - driven tayler instability requires radial displacements .
the instability does not exist in the 2d case of purely horizontal disturbances ( see dicke 1979 ) . in a stably stratified radiation zone
the radial displacements are opposed by buoyancy .
finite thermal diffusivity reduces the buoyancy and thus supports the instability .
the relevant parameter is the ratio @xmath57 @xmath58 of the frequency @xmath59 , with which the thermal diffusion destroys the buoyancy , to the characteristic frequency @xmath60 of gravity waves .
when @xmath61 and @xmath62 are simultaneously varied for constant @xmath57-parameter ( [ 3 ] ) the results change little .
the toroidal field profile of the model is parameterized as @xmath63 where now @xmath64 is the _ amplitude _ of the magnetic alfvn frequency .
computations were made for @xmath65 so that the toroidal field is symmetric ( or quadrupolar ) with respect to the equator .
the other model with @xmath66 represents a ( dipolar ) background field antisymmetric relative to the equator or in other words with two belts of opposite signs .
the diffusion parameters @xmath67 and @xmath68 characteristic for the upper radiation zone of the sun are kept fixed and @xmath61 is varied to study the effect of thermal diffusion . the normalized wavelengths ( [ 1 ] )
were also kept constant .
the constant values were @xmath69 for quadrupolar background field ( @xmath65)and @xmath70 for dipolar fields ( @xmath66 ) .
the choice is motivated by the finding that for this values of @xmath62 the marginal field strengths for onset of the instability are minimized .
the values of @xmath62 corresponding to maximum growth rates of supercritical excitations change with the external parameters but only slightly . some results of this paper were obtained for the instability modes of symmetry type s1 , the equatorially symmetric excitations with azimuthal wave number @xmath0 .
for @xmath71 and for the given prandtl number ( [ eq5 ] ) one finds @xmath72 as the solar value .
computing the critical magnetic frequencies under the presence of differential rotation the wave numbers have been varied as long as the minimum eigenvalues have been found ( see below ) .
the linear code is able to calculate the eigenfrequencies for the realistic small magnetic prandtl number @xmath73 so far the best nonlinear mhd codes reach values down to 0.001 @xcite .
in the following the stability of the magnetic background field ( [ 4 ] ) is probed for nonaxisymmetric disturbances with the azimuthal wave number @xmath0 and with a fixed radial scale . by use of the method described by kitchatinov and rdiger ( 2008 ) the complex equation system ( [ 5])([7 ] )
is numerically solved to find the growth rate of a possible instability .
generally , the results do not change if @xmath74 is replaced by @xmath75 .
indeed , for purely toroidal magnetic fields there is no handedness in the system .
the addition of even a weak poloidal field would break the symmetry between @xmath74 and @xmath75 .
we start with @xmath65 as the most simple case .
as the fields are symmetric with respect to the equator this model resembles stellar models with toroidal fields of quadrupolar parity .
the growth rates of the instability for various values of the thermal diffusivity parameter @xmath57 are shown in figure .
all the lines converge for sufficiently strong fields , @xmath76 , showing that then the instability is insensitive to diffusion .
the growth rates @xmath20 for strong fields can be estimated as @xmath77 @xcite .
the magnetic instability in adiabatic fluids , i.e. with @xmath78 , is suppressed by the rotation , @xcite .
but for finite thermal diffusion also weak fields under the presence of superalfvnic rotation , @xmath79 , are unstable though with small growth rates .
the growth rates initially increase with increasing @xmath57 to saturate for about @xmath80 ( figure [ f1 ] , left ) . due to its very small growth rates
the instability of weak fields ( @xmath81 ) for both small and very large @xmath57 can not be detected with time stepping codes ( see braithwaite 2006 ) . in the wide range @xmath82
also a weak - field instability is faintly sensitive to thermal diffusivity .
the growth rate runs as @xmath83 only for very large @xmath57 when the buoyancy is totally suppressed the results for unstratified fluids with their strong rotational suppression of the instability are reproduced ( figure [ f1 ] , right ) .
the case of one magnetic belt , @xmath84 , is exceptional @xcite .
when the ratio of rotational velocity to alfvn velocity is uniform then there is no preferred latitude for the instability to start .
finite thermal diffusion promotes the instability as the effect of diffusion depends on the latitudinal scale of the disturbances .
the isothermal limit @xmath85 represents also the case of unstratified fluid and the instability is totally suppressed for fast rotation .
we find a different situation for dipolar background fields ( @xmath66 ) with two belts of opposite polarity as resulting from the interaction of differential rotation and dipolar poloidal fields .
figure [ f2 ] shows the results . in the above model of quadrupolar fields ,
the vertical lines of figure [ f0 ] for adiabatic ( @xmath86 ) and isothermal ( @xmath27 ) disturbances are rather parallel and close together .
now the ( dashed ) line for @xmath87 ( or , what is the same , for unstratified fluids ) and the vertical line for @xmath86 differ completely .
the dashed line is close below the line for @xmath57= 1000 for which the maximum growth rates appear .
it also shows the typical behavior ( [ 42 ] ) for subequipartition fields ( @xmath81 ) .
consequently , one finds the relation ( [ 42 ] ) true for all nonadiabatic disturbances in fluids with @xmath88 .
hence , if the very small growth rates are acceptable , then already toroidal fields with @xmath89 become unstable if the @xmath57 is not too small .
for the sun the maximum strength of stable fields is about 500 gauss ( cf .
spruit 1999 ) .
note that the growth rates for the solar value of @xmath90 differ slightly from the growth rates for isothermal case .
for the stellar magnetic fields with dipolar parity , @xmath91 already represents the situation for @xmath27 .
the inclusion of the buoyancy is thus not even necessary .
this statement does not hold for the above model of quadrupolar field geometry .
a very new feature of the considered two - belts geometry are resonances .
the lines with fixed but small @xmath57 show two peaks at magnetic field amplitudes slightly below the equipartition level of @xmath92 .
the resonant eigenmodes do not appear in the one - belt model and they show a more detailed fine - structure than the nonresonant ones ( figure [ f3 ] ) . as a doubling of the resolution does not change the results the numerically detected resonances seem to be real .
note that all the unstable modes are global in horizontal dimensions .
the resonant nature of the peaks in figure [ f2 ] is illustrated by figure [ f4 ] showing the drift velocities for a small @xmath93 .
generally , the unstable modes drift against the direction of the global rotation . note that the drift rates of the resonant modes are zero .
these modes corotate with the fluid . for strong fields with @xmath94 the instability pattern does not follow the basic rotation .
the normalized drift rates approach the value of -1 which means that the modes are resting in the inertial frame of reference .
test calculations showed that this result also holds for @xmath95 which for the interior of hot stars is not unrealistic due to the large radiative viscosity .
the nonaxisymmetric field pattern produced by instability of a strong field seems to rest in the laboratory system .
if only such a magnetic pattern is observed on a star then it seems to rest or to exhibit an extremely slow rotation .
the steepness of the drift rate profile of figure [ f4 ] suggests that the transition between the nondrifting solution ( where the difference between the rotation of the star and the rotation of the magnetic field is very small ) and the drifting solution ( where the magnetic pattern rotation disappear ) is very sharp .
if ap stars are assumed as stars with unstable toroidal background fields ( which themselves are invisible ) then two groups among them should exist depending on the ratio of @xmath64 to @xmath2 .
the magnetic field pattern of the group with @xmath81 rotates slightly slower than the star but the magnetic field pattern of the group with @xmath96 should rotate extremely slow .
the instability for strong fields is very fast .
the growth times for this case after the equation ( [ eq3 ] ) are shorter than the rotation period . on the other hand , after ( [ 42 ] ) the instability of so weak fields that @xmath81 is much slower .
it is , however , hardly controlled by the thermal diffusion .
the instability of weak fields in stellar radiation zones should thus not be too sensitive to the chemical details .
sofar we have assumed the stellar rotation as rigid .
this is only true if the star is old enough .
a possible differential rotation produces a strong toroidal field from the original fossil poloidal field .
the resulting maxwell stress suppresses the differential rotation producing an almost rigid rotation after the alfvn travel time estimated for the poloidal field which for hot stars with fields of order mgauss lasts longer than 10 myr .
hence , for young stars the instability of the field pattern must be considered under the presence of differential rotation .
as the current - driven instability is basically nonaxisymmetric one must expect the action of the differential rotation as stabilizing so that a possible instability might occur only after the alfvn travel time .
we have worked with the simple rotation law @xmath97 which in its simplified local formulation reads as @xmath98 .
equation ( [ 45 ] ) describes a rotation law with cylindric isolines . for negative @xmath99
the rotation rate decreases outwards and v.v .
the stabilizing effect of the differential rotation should _ not _ depend on the sign of @xmath99 . here
we only consider the magnetic field geometry ( [ 4 ] ) for dipolar field structures , i.e. with antisymmetry of the fields with respect to the equator .
figure [ f7 ] gives the results for @xmath72 .
the critical magnetic frequencies are optimized by choice of the wave numbers .
we find the stabilization by differential rotation as highly effective .
already very small shear values lead to an increase of the critical magnetic field by a factor of ( say ) five .
the influences of both the sign and the real value of @xmath99 are small .
the figure [ f7 ] suggests a very efficient stabilization of the toroidal fields by any kind of differential rotation if the field is antisymmetric with respect to the equator .
the same might _ not _ be true for other field geometries ( see rdiger and schultz 2010 ) .
hence , we find the toroidal magnetic fields in stellar tachoclines with their strong differential rotation ( if due to a fossil poloidal dipolar field ) much more stable than they are in the rigidly rotating cores of stars .
the stabilizing action of differential rotation does hardly depend on the form of the rotation law .
if the star rotates nonhomogeneously then higher amplitudes of the induced toroidal fields remain stable .
the results of this section suggest the importance of further studies of the interaction of magnetic fields and differential rotation .
if the fields are produced by a dynamo mechanism then the magnetic field geometry can easily differ ( like in galaxies ) from that considered in the present paper .
we have shown that for realistic values of the heat diffusivity the growth rates of the kink - type instability in stellar radiation zones do hardly differ from the growth rates obtained for fully incompressible models without buoyancy .
there are , however , strong differences for other types of the magnetic geometry .
cylindric models with uniform alfvn frequency do not completely cover the situation for spherical models with toroidal fields of dipolar parity .
for stellar applications the main results for the instability of such fields are given in figure [ f6 ] .
note how well the approximation without gravity @xmath100 and/or @xmath24 ( dashed line ) works in comparison to the ` exact ' profile for @xmath101 ( solid line ) . after the figure [ f6 ] ( left )
three groups of hot stars can be distinguished in dependence on the amplitude of their toroidal fields .
the toroidal magnetic field of the first group fulfills the relation @xmath102 so that it remains stable .
if the field amplitude exceeds the lower limit ( or the rotation is slow enough ) then there are two possibilities .
if it is not too strong , i.e. @xmath81 , then it becomes unstable with very small rates of growth and azimuthal drift .
if strong enough , the poloidal component of the resulting nonaxisymmetric field should be observable .
the critical alfvn velocity for a typical hot star is about 10 km / s corresponding to a magnetic field of order 10@xmath103 gauss .
if only 1% of the magnetic energy move to the poloidal perturbation , the amplitude of poloidal field is about 10@xmath104 gauss ( see gellert _
this value is even larger than the observed fields of ap stars .
the third group of stars fulfills the condition @xmath96 .
they are unstable with very short growth times .
their drift rates , however , approach the value of -1 so that the magnetic patterns may only show a very slow global rotation .
there are indeed examples among the group of the ap stars with rotation periods of several years .
the bright star @xmath20 equ has a rotation period longer than 70 yr . in the light of the presented theory the basic fact that the ap stars are slow rotators compared with the normal
a stars mainly means that slow rotation is less stabilizing for the toroidal magnetic fields . with other words ,
the condition @xmath105 for instability is more easily fulfilled for slow rotators .
it is thus understandable with our results that for slow rotation weaker toroidal fields become unstable and also the resulting amplitude of the @xmath0 mode is smaller for slow rotation than for fast rotation which indeed is observed @xcite .
the condition @xmath105 for instability could easily be the counterpart of the lower limit of about 300 gauss found by aurire _
( 2007 ) for magnetic fields of ap stars .
the existence of rather strong toroidal magnetic fields within stellar radiation zones can be thought of as the outcome of the differential rotation and a weak fossil poloidal field . for the above calculations we have assumed that the differential rotation only exists before the magnetic instability develops .
the reason is that ( any form of ) differential rotation stabilizes the instability of fields with equatorial antisymmetry .
from this point of view all the ap stars are considered as ( slow ) rigid rotators .
a complete explanation of the ap star magnetism still meets open questions .
so the axis of the magnetic field pattern is obviously _ not _ orthogonal to the axis of rotation ( oetken 1977 ) .
the obliquity of the field , i.e. the ratio of nonaxisymmetric and axisymmetric field parts , depends on the rotation rate : it is maximum for large @xmath2 ( landstreet and mathys 2000 ) .
the instability of a single @xmath0 mode can not explain this finding .
it is also known that the magnetic ap stars do _ not _ exist close to the zams , they are concentrated toward the center of the main - sequence .
the earliest observed evolution time across the main - sequence of a magnetic ap and bp star is about 20 myr , no one younger magnetic star has been observed ( hubrig _ et al . _ 2000 ) .
there are also possibilities to explain these empirical findings but those are beyond the scope of the present paper .
aurire , m. , wade , g.a .
, silvester , j. , _ et al . _ , weak magnetic fields in ap / bp stars .
evidence for a dipole field lower limit and a tentative interpretation of the magnetic dichotomy .
astrophys . _ , 2007 , * 475 * , 10531065 .
hubrig , s. , north , p. and schller , m. , evolution of magnetic fields in stars across the upper main sequence : ii .
observed distribution of the magnetic field geometry .
nachr . _ , 2007 , * 328 * , 475490 . | the stability of toroidal magnetic fields in rotating radiative stellar zones is studied for realistic values of both the prandtl numbers .
the two considered models for the magnetic geometry represent fields with odd and even symmetry with respect to the equator . in the linear theory in boussinesq approximation
the resulting complex eigenfrequency ( including growth rate and drift rate ) are calculated for a given radial wave number of a nonaxisymmetric perturbation with @xmath0 .
the ratio of the alfvn frequency , @xmath1 , to the rate of the basic rotation , @xmath2 , controls the eigenfrequency of the solution . for strong fields with @xmath3
the solutions do not feel the thermal diffusion .
the growth rate runs with @xmath1 and the drift rate is close to @xmath4 so that the magnetic pattern will rest in the laboratory system . for weaker fields with @xmath5
the growth rate strongly depends on the thermal conductivity . for fields with dipolar parity and for typical values of the heat conductivity
the resulting very small growth rates are almost identical with those for vanishing gravity . for fields with dipolar symmetry
the differential rotation of any stellar radiative zone ( like the solar tachocline ) is shown as basically stabilizing the instability independent of the sign of the shear .
finally , the current - driven kink - type instability of a toroidal background field is proposed as a model for the magnetism of ap stars .
the recent observation of a lower magnetic field treshold of about 300 gauss for ap stars is understood as corresponding to the minimum magnetic field producing the instability .
stellar magnetic fields ; stellar rotation ; differential rotation ; solar tachocline |
the maximum luminosity of type ia supernovae ( sne ia ) is a key ingredient of their analysis and is also essential for their use as distance indicators in cosmology ( @xcite ; @xcite ; @xcite ) . for the nearby sne
ia with cepheid - determined distances , the overall dispersion in the peak magnitude measurements is rather small , about 0.5 mag . in @xmath10 and @xmath5 ( @xcite ; @xcite ; @xcite ) .
when the sample is enlarged to include more distant sne ia , there are several subluminous events that broaden the variation to about 1 mag . in @xmath10 ( @xcite ;
@xcite ; @xcite ; @xcite ) , but the bulk of the sne ia sample have peak brightnesses within a 0.5 mag . range in @xmath10 and @xmath5 .
an interesting feature of sne ia is that the luminosity is set not by the explosion , for which the deposited energy goes into expansion , but rather by the decay of @xmath0 and @xmath11 that are formed during the nucleosynthesis @xcite . at the time of peak luminosity , @xmath11 has not yet decayed , and hence the peak luminosity is a measurement of the amount of @xmath0 synthesized during the explosion .
this amount presumably depends on the progenitor and on the details of the explosion .
in particular , it has long been known that the amount of @xmath0 synthesized depends in part on the asymmetry between neutrons and protons in the progenitor white dwarf ( @xcite ; @xcite ) . in this letter
we explore how the intrinsic variation in the initial cno abundances of the progenitors translates into a variation in the mass of @xmath0 synthesized ( @xmath12 ) and hence in the peak luminosity of sne ia .
we show analytically , using mass and charge conservation , that @xmath12 depends _ linearly _ on the initial metallicity @xmath13 , and that the observed variation in the @xmath13 of field dwarfs , from to 3 times solar , leads to a @xmath14 variation in @xmath12 , with most of the effect occurring at @xmath15 .
this is the expected variation if sne ia progenitors uniformly represent the range of cno abundances observed in stars today .
our conclusion rests on ( 1 ) considerations of nuclear equilibrium during the explosion , and ( 2 ) the observed scatter in @xmath13 of stars within the galactic disk .
the dependence of the ejected @xmath0 mass on the progenitor s initial @xmath13 has been investigated previously , both via the evolution of the progenitor ( @xcite ; @xcite ; @xcite , hereafter d01 ) and the explosion itself ( @xcite ; @xcite ; d01 ) . although these one - dimensional simulations are sophisticated in their treatment of the flame microphysics and the nuclear burning , we desire to elucidate features that are robust to any complicated hydrodynamics .
our demonstration that the mass of nickel depends linearly on the initial @xmath13 of the progenitor serves as a check and stimulus for future numerical studies and provides insight into possible evolutionary effects . in [ sec : variation - ni - mass ]
we demonstrate the linear dependence of @xmath12 on the progenitor s @xmath13 , and compare this with detailed nucleosynthesis calculations and previous numerical work .
we speculate on the source of differences between different calculations , and then discuss the intrinsic scatter in cno abundances in the ism ( [ sec : scatter - metallicity ] ) and the implications ( [ sec : implications ] ) of our results .
nearly all one - dimensional chandrasekhar mass models of sne ia produce most of their @xmath0 in a burn to nuclear statistical equilibrium ( nse ) between the mass shells 0.2@xmath1@xmath2 and 0.8@xmath1@xmath2(@xcite ; @xcite ; @xcite ; @xcite ) . in this region , unlike in the innermost 0.2@xmath1@xmath2 @xcite
, weak interactions operate on timescales longer than the time for the thermonuclear burning front to disrupt the white dwarf . following this rapid burn to nse ,
most of the mass is in the iron - peak nuclei @xmath0 , @xmath16 , and @xmath17 .
first consider the case when @xmath0 and @xmath16 are the only two competing species .
the addition of @xmath17 will be considered next .
mass and charge conservation , @xmath18 imply that the mass fraction of @xmath0 depends linearly on @xmath3 , @xmath19 where isotope @xmath20 has @xmath21 protons , @xmath22 nucleons ( protons + neutrons ) , and a mass fraction @xmath23 .
the aggregate ensemble has a proton to nucleon ratio of @xmath3 .
most of a main - sequence star s initial @xmath13 comes from the cno and @xmath24 nuclei inherited from its ambient interstellar medium .
the slowest step in the hydrogen burning cno cycle is proton capture onto @xmath25 .
this results in all the cno catalysts piling up into @xmath25 when hydrogen burning on the main sequence is completed . during helium burning the reactions @xmath26
convert all of the @xmath25 into @xmath8 .
thus , the mass fraction of @xmath8 in the carbon - oxygen white dwarf remnant is @xmath27 where the mass fractions refer to the original distribution of the star ( prior to main - sequence burning ) . for a uniform distribution of @xmath8and @xmath24 throughout the star , eq .
( [ eq : mqconserv ] ) gives the initial @xmath3 of the white dwarf @xmath28 .
\label{eq : ye_initial}\ ] ] according to the argument at the beginning of this section , this @xmath3 is fixed in the region where the @xmath0 is created .
substituting eqs .
( [ eq : ye_initial ] ) and ( [ eq : x22 ] ) into eq .
( [ eq : nse1 ] ) leads to a linear expression for the mass fraction of @xmath0 in an nse distribution in terms of the main - sequence star s initial @xmath13 @xmath29 \nonumber \\ & = & 1 - 0.057 \frac{z}{{\ensuremath{z_\odot}}}. \label{eq : x56}\end{aligned}\ ] ] the average peak @xmath10 and @xmath5 magnitudes of nearby sne ia@xcite strongly imply that a fiducial sne ia produces @xmath30 of @xmath0 .
taking eq .
( [ eq : x56 ] ) to represent the mass fraction of @xmath0 relative to this fiducial mass gives @xmath31 here we assume that @xmath3 is uniform throughout the star , and that all material within the @xmath0-producing mass shell passes through nse with normal freeze - out .
if @xmath17 is also present then an additional saha - like equation for the chemical potentials @xcite is required .
consider the chemical equation @xmath32 . both @xmath17 and @xmath16 carry two extra neutrons
; this fixes the stoichiometric ratios @xmath33 and @xmath34 . as a result ,
the equation for chemical equilibrium is of the form @xmath35^{1/14 } $ ] .
here @xmath36 contains the mass excesses and phase space factors common to the chemical potential of the three species . expanding about @xmath37 , @xmath38 , and @xmath39 , we find that @xmath40 for this ratio , @xmath41 , the relationship between @xmath13 and @xmath42 becomes slightly shallower than eq .
( [ eq : x56 ] ) , @xmath43 .
similar considerations show that @xmath24 is not present unless @xmath3 is much smaller than the case considered
. equation ( [ eq : fe54-ni58 ] ) does not hold if an alpha - rich freeze - out occurs , in which case @xmath44 reactions convert @xmath17 to @xmath16 .
hence the ratio @xmath45 is sensitive to the flame speed .
the @xmath0 mass is affected only slightly , however , as the charge - to - mass of @xmath16 ( which sets the slope in eq .
[ [ eq : nse1 ] ] ) is @xmath46 and differs by only 0.3% from the charge - to - mass of @xmath17 , @xmath47 .
our main result , eq . ( [ eq : x56 ] ) , is thus relatively insensitive to the precise details of the explosion , so long as nse favors a @xmath0-dominated `` iron '' peak .
the simple linear relation between @xmath3 and the mass fraction of @xmath0 in eq . ( [ eq : x56 ] ) is robust , as it only relies on basic properties of nse . as a test of this analytical result , we calculated the mass of @xmath0 ejected by w7-like models ( @xcite ; @xcite ; @xcite ) by integrating a 510 isotope nuclear reaction network over the thermodynamical trajectories for a wide range of @xmath13 .
the results are shown in fig .
[ fig : mni56 ] with the short - dashed curve .
we also plot the linear relation ( eq .
[ [ eq : m56 ] ] ; _ solid curve _ ) , with @xmath48 adjusted to give the @xmath8 abundance used by w7 , @xmath49 .
most of the differences between eq .
( [ eq : m56 ] ) and the detailed w7-like models is attributable to our assumption that all of the @xmath0 comes from the 0.2 to 0.8@xmath1@xmath2 region in the white dwarf , with an additional small correction for our neglect of weak interactions that slightly decrease @xmath3 .
the difference between the slope given by the detailed w7-like models and the analytical model is less than 5% .
previous investigations ( @xcite ; d01 ) into the effect of a varying @xmath50 do not agree on the mass of @xmath0 produced .
@xcite found that varying the metallicity from @xmath51 to @xmath52 produced only an @xmath53 variation in the @xmath0 mass ejected , in sharp contrast to our results as well the results of @xcite , who found that increasing @xmath13 from zero to solar in fast - deflagration models leads to an @xmath54 decrease in the @xmath0 mass ejected .
both of these works relied on post - processing the thermodynamical trajectories , as we have done .
recently , d01 investigated the range @xmath55 to @xmath56 .
of the one - dimensional calculations just mentioned , only this one accounts for the effect of @xmath3 on the energy generation rate .
the long - dashed curve in fig .
[ fig : mni56 ] shows the d01 results for their 1.5@xmath1@xmath2progenitors ; other progenitor masses in their survey display the same trend with @xmath13 . our analytical result and
post - processed w7-like models essentially agree with the findings of d01 over the range of metallicities common to all three calculations . as is evident in fig .
[ fig : mni56 ] , the largest variation in the mass of @xmath0 occurs at @xmath15 .
we note that the calculations of d01 evolve a main - sequence star into the sne ia progenitor , whereas our calculation and those of @xcite start from a given white dwarf configuration . as a caveat
, we note that our post - processing of the w7 thermodynamic trajectories is not completely self - consistent .
the reason is that the temperature and density profiles of the w7 were calculated using the energy released by burning matter of solar @xmath13 . while @xmath13 is likely to influence the flame propagation ( via the change in the rate of energy production ; * ? ? ?
* ; * ? ? ?
* ) , in the mass range under consideration si - burning is complete , so our assumption of nse still holds .
indeed , @xcite found that the mass of @xmath0 synthesized was also rather insensitive to the details of the flame microphysics .
it is also possible the density , @xmath57 , at which a transition from deflagration to detonation occurs will influence the amount of @xmath0produced ( * ? ? ?
* and references therein )
. fits to observations seem to require , however , that @xmath58 which corresponds to a lagrangian mass coordinate of @xmath59 1.0@xmath1@xmath2 , which is exterior to the main @xmath0-producing layers .
further numerical studies should elucidate the source of the discrepancy .
in particular they should determine if other effects , such as a systematic change in the mass coordinates where most of the @xmath0 is produced , offsets or enhances the variation in @xmath12 that we find .
the demonstrated linear dependence between @xmath12 and @xmath13 will make it easier to untangle competing physical effects in a fully self - consistent calculation .
stellar abundance determinations are discussed in terms of an elemental abundance relative to iron , @xmath60 $ ] , as a function of the iron to hydrogen ratio @xmath61 $ ] , primarily because @xmath61 $ ] is relatively easy to measure in stars .
the @xmath61 $ ] ratio represents a chronometer in that the accumulation of iron in the ism increases monotonically with time @xcite .
calibration of @xmath61 $ ] as a function of time forms the basis of the age - metallicity relationship .
the @xmath13 of local field stars rapidly increased about 1013@xmath1@xmath62@xmath63 ago during formation of the galaxy s disk and then increased much more gradually over the last @xmath64 ( @xcite ; @xcite ; @xcite ) . more importantly for our purposes
, however , is the relatively large scatter in stellar metallicities , @xmath65}}\sim 0.5 $ ] , at any given age .
@xcite constructed an age - metallicity diagram for 5828 dwarf and sub - dwarf stars from the hipparcos catalog using evolutionary tracks to derive ages and strmgren photometry to derive metallicities .
they concluded that the age - metallicity diagram is well - populated at all ages , that old , metal - rich stars do exist , and that the scatter in @xmath13 at any given age is larger than the observational errors .
other surveys of stellar metallicities ( @xcite ; @xcite ) are in good agreement with these trends . the most abundant elements in the galaxy after h and he are cno .
both @xmath66 $ ] and @xmath67 $ ] in halo and disk dwarfs are observed to be roughly solar and constant ( @xcite ; @xcite ; @xcite ) .
the @xmath68 $ ] ratio is larger at low metallicities
oxygen being the dominant element ejected by sne ii and
then slowly decreases due to variations in mass and @xmath13 ( @xcite ; @xcite ; @xcite ) . within these general trends is a relatively large scatter , @xmath69}}\sim \delta{\ensuremath{[\mathrm{n / fe}]}}\sim \delta{\ensuremath{[\mathrm{o / fe}]}}\sim { \ensuremath{0.5{{\ensuremath{\,}}}\mathrm{dex}}}$ ] , at any given @xmath61 $ ] . according to the simple analytical relation ( eq . [ [ eq : x56 ] ] ) and the detailed w7-like models ( fig .
[ fig : mni56 ] ) , a scatter of a factor of 3 about the mean in the initial metallicity ( @xmath70 ) leads to a variation of about 25% ( 0.13@xmath1@xmath2 ) in the mass of @xmath0 ejected by sne ia if the @xmath8 and @xmath24 are uniformly distributed within the white dwarf .
the minimum peak brightness variations caused by this variation in @xmath0 mass are @xmath71 @xcite .
thus , the amplitude of this effect can not account for all of the observed variation in peak luminosity of local sne ia ( 0.5 magnitude in @xmath10 and @xmath5 ) .
the observed scatter in peak brightnesses may be even larger , as cepheid - based distances to the host galaxies of peculiar events such as sub - luminous sn 1991bg or brighter - than - normal sn 1991 t havent been measured yet ( @xcite ; @xcite ; @xcite ) .
there is evidence for a larger scatter when more distant supernovae are included @xcite .
it would take a scatter of about a factor of seven ( @xmath65}}\sim { \ensuremath{0.8{{\ensuremath{\,}}}\mathrm{dex}}}$ ] ) in the initial @xmath13 to account for a factor of two variation in the @xmath0 mass and peak luminosity .
using the properties of nse , we find that where weak interactions are unimportant the mass fraction of @xmath0 produced depends linearly upon the initial metallicity of the white dwarf progenitor .
this result is robust : so long as the region that reaches nse does so on a timescale over which @xmath3 is nearly constant , then the mass of @xmath0 produced is largely independent of the detailed physics of the flame front propagation .
this fact offers a check on sophisticated numerical calculations and could be exploited to understand better the disagreements between different codes .
the variation from to 3 times solar metallicity observed in dwarf stars in the galactic disk implies that sne ia should have a minimum variation of 25% in the ejected @xmath0 mass .
this has implications for the brightness variations observed in both near and distant sne ia and for galactic chemical evolution .
the most distant sne ia observed today have redshifts @xmath72 .
this corresponds to a lookback time of about 47@xmath1@xmath62@xmath63 depending on the cosmological model , and implies a mean @xmath61 $ ] ratio between -0.1 and -0.3 .
there is , however , still a scatter of @xmath65}}\sim 0.5 $ ] at these mean @xmath61$]ratios ( see the discussion in 3 ) .
we therefore expect that the variation in peak luminosities of sne ia in spiral galaxies at @xmath72 will show the same minimum variation of @xmath71 in peak luminosity as nearby sne ia .
this variation is superposed on other evolutionary effects , such as that from a reduction in the c / o ratio ( @xcite ; d01 ) . for combustion to nse with no freeze - out
( see 2 ) , the @xmath0 mass is anticorrelated with the mass of @xmath17 and @xmath16 ejected , and subluminous sne ia will tend , therefore , to have larger @xmath73 ratios than brighter ones . because sne ia produce @xmath74 of the galactic iron - peak nuclei of iron - group nuclei while type ii sne produce @xmath75 , and the ratio of thermonuclear to core collapse events is about 0.150.27 in the galaxy @xcite . ] , the isotopic ratios among the iron group in sne ia ejecta should not exceed the solar ratios by about a factor of two@xcite .
there are several uncertainties with the @xmath73 ratio .
first , some of the @xmath17 is produced in the core where weak interactions are important for determining the final @xmath3 ( although this may be alleviated by the overturning of matter in the core from rayleigh - taylor instabilities ; see @xcite for a recent calculation ) . at high densities
these reaction rates are sensitive to the input nuclear physics ( @xcite ; @xcite ) .
isotopic measurements of cosmic rays , such as with _ ulysses _ ( @xcite ; @xcite ) and _ ace
_ @xcite can probe the evolution of @xmath73 over the past @xmath76 .
since most of the @xmath0 created in one - dimensional sne ia models lies in the 0.2@[email protected]@xmath1@xmath2 mass shell , our assumption of a uniform @xmath8 distribution deserves close scrutiny .
there has long been speculation about sedimentation of @xmath8 and its effect on the cooling of isolated white dwarfs ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
if , for example , all of the @xmath8 from the outermost 0.6@xmath1@xmath2 were to settle into the shell between 0.2@xmath1@xmath2 and 0.8@xmath1@xmath2 , then the effective @xmath3 in this shell would double .
indeed , @xcite noted that the production of @xmath17 was an indirect test of the sedimentation of @xmath8 . for nearby sne ia , which presumably sample a range of progenitor ages ,
the variability in @xmath0 production would increase to 50% , or about @xmath77 .
since it takes about 7@xmath1@xmath62@xmath63 for @xmath8 at the surface to fall through the outer @xmath78 of a @xmath79 co white dwarf ( @xcite ) , the effect of sedimentation will be diminished for those sne ia at @xmath80 .
we thank franziska brachwitz and friedel thielemann for providing the initial w7 thermodynamic trajectories .
we also thank the referees for their critical reading , which greatly improved the manuscript .
this work is supported by the department of energy under grant no .
b341495 to the center for astrophysical thermonuclear flashes and grant no .
de - fg02 - 91er40606 in nuclear physics and astrophysics at the university of chicago . | we explore the idea that the observed variations in the peak luminosities of type ia supernovae originate in part from a scatter in metallicity of the main - sequence stars that become white dwarfs .
previous , numerical , studies have not self - consistently explored metallicities greater than solar .
one - dimensional chandrasekhar mass models of sne ia produce most of their @xmath0 in a burn to nuclear statistical equilibrium between the mass shells 0.2@xmath1@xmath2 and 0.8@xmath1@xmath2 , for which the electron to nucleon ratio @xmath3 is constant during the burn .
we show analytically that , under these conditions , charge and mass conservation constrain the mass of @xmath0 produced to depend _ linearly _ on the original metallicity of the white dwarf progenitor .
detailed post - processing of w7-like models confirms this linear dependence .
the effect that we have identified is most evident at metallicities larger than solar , and is in agreement with previous self - consistent calculations over the metallicity range common to both calculations .
the observed scatter in the metallicity ( @xmath43@xmath4 ) of the solar neighborhood is enough to induce a 25% variation in the mass of @xmath0 ejected by type ia supernova .
this is sufficient to vary the peak @xmath5-band brightness by @xmath6 .
this scatter in metallicity is present out to the limiting redshifts of current observations ( @xmath7 ) .
sedimentation of @xmath8 can possibly amplify the variation in @xmath0 mass to @xmath9 .
further numerical studies can determine if other metallicity - induced effects , such as a change in the mass of the @xmath0-producing region , offset or enhance the variation we identify . |
consider a memoryless channel @xmath0 with input @xmath1 , output @xmath2 , and state @xmath3 .
we assume that the channel state @xmath3 , distributed according to @xmath4 , is provided to the decoder , and a noisy state observation @xmath5 , generated by @xmath3 through side channel @xmath6 , is available causally at the encoder . here
@xmath1 , @xmath2 , @xmath3 , and @xmath5 are defined over finite alphabets @xmath7 , @xmath8 , @xmath9 , and @xmath10 , respectively . in this
setting ( see fig . [
fig : model ] ) , shannon s remarkable result @xcite ( see also ( * ? ? ?
* eq . ( 3 ) ) and ( * ? ? ?
7.2 ) ) implies that the channel capacity is given by @xmath11 the auxiliary random variable @xmath12 is defined over alphabet @xmath13 with @xmath14 , whose joint distribution with @xmath15 factors as @xmath16 where @xmath17 is the indicator function , and @xmath18 , @xmath19 , are @xmath20 different mappings from @xmath10 to @xmath7 . without loss of generality ,
we set @xmath21 , @xmath22 , @xmath23 , and order the mappings @xmath18 , @xmath19 , in such a way that the first @xmath24 mappings are @xmath25 moreover , we assume that @xmath26 .
the capacity formula ( [ eq : noisystate ] ) can be simplified in the following two special cases . specifically , when there is no encoder side information , the channel capacity reduces to (
* eq . ( 7.2 ) ) @xmath27 where @xmath28 ; on the other hand , when perfect state information is available at the encoder ( as well as the decoder ) , the channel capacity becomes ( * ? ? ?
* eq . ( 7.3 ) ) @xmath29 where @xmath30 . for comparison ,
consider the following similarly defined quantity @xmath31 where the joint distribution of @xmath32 is also given by ( [ eq : jointdistribution ] ) .
we shall refer to @xmath33 as the generalized probing capacity . by the functional representation lemma @xcite ( see also ( * ? ? ?
* lemma 1 ) ) , @xmath33 can be defined equivalently as @xmath34 where @xmath35 clearly , @xmath36 moreover , we have @xmath37 if @xmath3 and @xmath5 are independent ( i.e. , @xmath38 ) , and @xmath39 if @xmath3 is a deterministic function of @xmath5 ( i.e. , @xmath40 ) . to elucidate the operational meaning of @xmath33 and its connection with @xmath41 , it is instructive to consider the special case where @xmath6 is a binary erasure channel with erasure probability @xmath42 ( denoted by @xmath43 ) , which corresponds to the probing channel setup studied in @xcite .
the probing channel model is essentially the same as the one in fig .
[ fig : model ] except that , in fig .
[ fig : model ] , the encoder ( which , with high probability , observes approximately @xmath44 state symbols out of the whole state sequence of length @xmath45 when @xmath45 is large enough ) has no control of the exact positions of these @xmath44 symbols whereas , in the probing channel model , the encoder has the freedom to specify the positions of these @xmath44 symbols according to the message to be sent .
it is shown in @xcite that this additional freedom increases the achievable rate from @xmath46 to @xmath47 .
now consider an example ( see also fig .
[ fig : szchannel ] ) where @xmath48 for this example , it can be verified that @xmath49 note that @xmath50 is strictly greater than @xmath51 unless @xmath52 or @xmath53 .
it follows by ( [ eq : endpoint1 ] ) and ( [ eq : endpoint2 ] ) that @xmath54 to gain a better understanding , we plot @xmath46 and @xmath47 against @xmath42 for @xmath55 $ ] in fig .
[ fig : plot1 ] .
it turns out that , somewhat counterintuitively , @xmath46 coincides with @xmath51 way before @xmath42 reaches 1 .
that is to say , when @xmath42 is above a certain threshold strictly less than 1 , the noisy state observation @xmath5 is useless and can be ignored ( as far as the channel capacity is concerned ) . on the the hand
, it can be seen that @xmath47 is equal to @xmath50 for a large range of @xmath42 strictly greater than 0 .
hence , in terms of the probing capacity , the noisy state observation can be as good as the perfect one . as shown in fig .
[ fig : plot2 ] , the same phenomena arise if we choose @xmath6 to be a binary symmetric channel with crossover probability @xmath56 ( denoted by @xmath57 ) . the contributions of the present work are summarized in the following theorems , which indicate that the aforedescribed surprising phenomena can in fact be observed for all binary - input channels .
[ thm : theorem1 ] for any binary - input channel @xmath0 , state distribution @xmath4 , and side channel @xmath6 , @xmath58 if @xmath59 , where @xmath60 .
[ thm : theorem2 ] for any binary - input channel @xmath0 , state distribution @xmath4 , and side channel @xmath6 , @xmath61 if @xmath62 , where @xmath60 .
on the surface these two results may look rather similar .
one might even suspect the existence of a certain duality between them .
however , it will be seen that the underlying reasons are actually quite different .
the proof of theorem [ thm : theorem1 ] hinges upon , among other things , a perturbation analysis .
in contrast , theorem [ thm : theorem2 ] is essentially a manifestation of an induced markov structure . the conditions in theorem [ thm : theorem1 ] and
theorem [ thm : theorem2 ] are stated in terms of bounds on @xmath63 and @xmath64 ; as a consequence , they depend inevitably on @xmath4 . as shown by theorem [ thm : theorem1variant ] in section [ sec : proof1 ] and theorem [ thm : theorem2variant ] in section [ sec : proof2 ] , it is in fact possible to establish these two results under more general conditions on @xmath6 that are universal for all binary - input channels and state distributions .
the rest of this paper is organized as follows .
we present the proofs of theorems [ thm : theorem1 ] and [ thm : theorem2 ] in sections [ sec : proof1 ] and [ sec : proof2 ] , respectively .
the validity of these two results under various modified conditions is discussed in section [ sec : discussion ] .
section [ sec : conclusion ] contains some concluding remarks . throughout this paper ,
all logarithms are base-@xmath65 .
first consider the special case where @xmath6 is a generalized erasure channel ( with erasure probability @xmath55 $ ] ) defined as @xmath66 [ lem : core ] given any binary - input channel @xmath0 and state distribution @xmath4 , @xmath67 for @xmath68 $ ] .
_ remark : _ lemma [ lem : core ] provides a universal upper bound on the erasure probability threshold above which the encoder side information is useless . the actual threshold , however , depends on @xmath0 and @xmath4 ( see section [ subsec : extensionthm1 ] for a detailed analysis ) . as indicated by ( [ eq : noisystate ] ) , the capacity of the channel model in fig .
[ fig : model ] ( i.e. , @xmath41 ) is equal to that of channel @xmath69 , where @xmath70 according to ( * ? ? ?
4.5.1 ) , @xmath71 is a capacity - achieving input distribution of channel @xmath69 ( i.e. , @xmath72 is a maximizer of the optimization problem in ( [ eq : noisystate ] ) ) if and only if there exists some number @xmath73 such that @xmath74 furthermore , the number @xmath73 is equal to @xmath41 . in view of ( [ eq : orderpsi ] ) ,
we have @xmath75 let @xmath76 be a capacity - achieving input distribution of channel @xmath77 ( i.e , @xmath76 is a maximizer of the optimization problem in ( [ eq : nostate ] ) ) .
define @xmath78 it is clear that @xmath79 if and only if @xmath80 is a capacity - achieving input distribution of channel @xmath69 .
now consider the special case where @xmath6 is a generalized erasure channel with erasure probability @xmath42 , and define @xmath81 to stress the dependence of @xmath82 on @xmath71 , @xmath42 , and @xmath83 .
it can be verified that @xmath84 where @xmath85 since @xmath86 , there is no loss of generality in assuming that ( * ? ? ?
2 ) @xmath87 to the end of proving lemma [ lem : core ] , it suffices to show that , for @xmath68 $ ] , @xmath88 clearly , @xmath80 is a capacity - achieving input distribution of channel @xmath69 when @xmath89 .
therefore , we have ) is in fact an equality . ]
@xmath90 note that @xmath91 , u\in\mathcal{u},\label{eq : combine3}\end{aligned}\ ] ] where ( [ eq : subp ] ) is due to ( [ eq : subh2 ] ) , and ( [ eq : combine3 ] ) is due to ( [ eq : orderpsi ] ) and ( [ eq : uconstruction ] ) .
moreover , @xmath92 , u\in\mathcal{u}.\label{eq : newsub}\end{aligned}\ ] ] define @xmath93 in light of ( [ eq : combine3 ] ) , @xmath94 , u\in\mathcal{g}_{\delta}.\label{eq : invariant}\end{aligned}\ ] ] for any @xmath95 , there must exist some @xmath96 and @xmath97 such that @xmath98 ; furthermore , since @xmath86 , we have @xmath99 where @xmath100 continuing from ( [ eq : newsub ] ) , @xmath101 , u\in\mathcal{u},\label{eq : comb4}\end{aligned}\ ] ] where ( [ eq : invokeposinput ] ) is due to ( [ eq : posinput ] ) , and ( [ eq : twoimply ] ) is due to ( [ eq : pos ] ) and ( [ eq : neg ] ) . combining ( [ eq : twoactiveinputs ] ) , ( [ eq : ineqtoeq ] ) , ( [ eq : invariant ] ) , ( [ eq : comb4 ] ) , and the fact @xmath102 yields the desired result .
recall @xcite that @xmath103 ( with input alphabet @xmath9 and output alphabet @xmath104 ) is said to be a stochastically degraded version of @xmath105 ( with input alphabet @xmath9 and output alphabet @xmath106 ) if there exists @xmath107 satisfying @xmath108 we can write ( [ eq : equivalent ] ) equivalently as @xmath109 by viewing @xmath103 , @xmath105 , and @xmath107 as probability transition matrices
. the following result is obvious and its proof is omitted .
[ lem : degraded ] if @xmath103 is a stochastically degraded version of @xmath105 , then @xmath110 next we extend lemma [ lem : core ] to the general case by characterizing the condition under which @xmath6 is a stochastically degraded version of @xmath111 . [ lem : erasure ] @xmath6 is a stochastically degraded version of @xmath111 if and only if @xmath112 the problem boils down to finding a necessary and sufficient condition for the existence of @xmath113 such that @xmath114 it suffices to consider the case @xmath115 since lemma [ lem : erasure ] is trivially true when @xmath89 .
note that @xmath116 combining ( [ eq : cond1 ] ) and ( [ eq : cond2 ] ) gives @xmath117 in light of ( [ eq : inview ] ) , @xmath118 it can be readily seen that the existence of conditional distribution @xmath113 satisfying ( [ eq : cond1 ] ) is equivalent to the existence of probability vector @xmath119 satisfying ( [ eq : ns2 ] ) .
clearly , ( [ eq : ns ] ) is a necessary and sufficient condition for the existence of such @xmath119 .
[ thm : theorem1variant ] for any binary - input channel @xmath0 , state distribution @xmath4 , and side channel @xmath6 , @xmath58 if @xmath120 in view of lemmas [ lem : core ] , [ lem : degraded ] , and [ lem : erasure ] , we have @xmath121 if ( [ eq : translate ] ) is satisfied . combining ( [ eq : comb1 ] ) and
( [ eq : comb2 ] ) completes the proof of theorem [ thm : theorem1variant ] .
now we proceed to prove theorem [ thm : theorem1 ] by translating ( [ eq : translate ] ) ( which is a condition on @xmath6 that is universal for all binary input channels and state distributions ) to an upper bound on @xmath63 .
this upper bound , however , depends inevitably on the state distribution . for any @xmath6 violating ( [ eq : translate ] ) ( i.e , @xmath122 ) , we have @xmath123 where ( [ eq : pinsker ] ) is due to pinsker s inequality @xcite , and @xmath124 is a minimizer of @xmath125 , @xmath126 . as a consequence , ( [ eq : translate ] )
must hold if @xmath59 .
this completes the proof of theorem [ thm : theorem1 ] .
first consider the special case where @xmath6 is a generalized symmetric channel ( with crossover probability @xmath127 $ ] ) defined as @xmath128 [ lem : nssym ] @xmath129 if and only if @xmath130 for some @xmath131 , where @xmath132 denotes the set of maximizers of the optimization problem in ( [ eq : perfectstate ] ) , and @xmath133 .
clearly , @xmath129 if and only if there exists @xmath131 that is a stochastically degraded version of @xmath134 . when @xmath135 , ( [ eq : nssym ] ) is equivalent to the desired condition that @xmath136 needs to be independent of @xmath3
. when @xmath137 , @xmath134 is invertible and @xmath138 the problem boils down to finding a necessary and sufficient condition under which @xmath139 is a valid probability transition matrix ( i.e. , all entries are non - negative and the sum of each row vector is equal to 1 ) .
note that @xmath140 moreover , all entries of @xmath139 are non - negative if and only if @xmath141 which is equivalent to ( [ eq : nssym ] ) . the following result is obvious and its proof is omitted .
[ lem : degraded2 ] if @xmath103 is a stochastically degraded version of @xmath105 , then @xmath142 [ lem : mmatrix ] @xmath134 is a stochastically degraded version of @xmath6 if @xmath143 where @xmath144 is the maximum likelihood estimate of @xmath3 based on @xmath5 , and @xmath145 .
the case @xmath135 is trivial .
when @xmath137 , @xmath134 is invertible and @xmath146 is given by ( [ eq : inverse ] ) .
it can be shown ( see the derivation of ( [ eq : identity ] ) ) that the sum of each row of @xmath147 is equal to 1 ; moreover , the off - diagonal entries of @xmath147 are non - positive if and only if @xmath148 which is equivalent to ( [ eq : suff ] ) . therefore , ( [ eq : suff ] ) ensures that @xmath147 is a non - singular @xmath149-matrix , which in turn ensures that @xmath150 exists and is a non - negative matrix @xcite .
hence , if ( [ eq : suff ] ) is satisfied , then @xmath150 is a valid probability transition matrix ( the requirement that the entries in each row of @xmath150 add up to 1 is automatically satisfied ) , which implies that @xmath134 is a stochastically degraded version of @xmath151 ( and consequently a stochastically degraded version of @xmath6 ) .
[ thm : theorem2variant ] for any binary - input channel @xmath0 , state distribution @xmath4 , and side channel @xmath6 , @xmath61 if @xmath152 where @xmath144 is the maximum likelihood estimate of @xmath3 based on @xmath5 .
since @xmath86 , it follows from ( * ? ? ? * th .
2 ) that there exists @xmath131 satisfying @xmath153 for such @xmath154 , @xmath155 in view of of lemmas [ lem : nssym ] , [ lem : degraded2 ] , and [ lem : mmatrix ] , we have @xmath156 if ( [ eq : sidechannelcondition ] ) is satisfied . combining ( [ eq : comb1 ] ) and ( [ eq : comb3 ] ) completes the proof of theorem [ thm : theorem2variant ] .
now we are in a position to prove theorem [ thm : theorem2 ] .
let @xmath144 and @xmath157 denote respectively the maximum likelihood estimate and the maximum _ a posteriori _ estimate of @xmath3 based on @xmath5 . according to ( * ? ? ?
11 ) , @xmath158 it can be verified that @xmath159 substituting ( [ eq : hvineqnew ] ) into ( [ eq : offdiagnew ] ) yields @xmath160 note that @xmath161 indeed , ( [ eq : twocases ] ) is trivially true when @xmath162 ; moreover , when @xmath163 , @xmath164 where ( [ eq : h1 ] ) and ( [ eq : h2 ] ) are due to ( [ eq : tbinvokenew ] ) . in view of theorem [ thm : theorem2variant ] , it suffices to have @xmath165 note that ( [ eq : hbar ] ) is not satisfied when @xmath162 since its left - hand side is equal to 1 whereas its right - hand side is strictly less than 1 ( @xmath162 implies @xmath166 ) .
when @xmath163 , we can rewrite ( [ eq : hbar ] ) as implies @xmath163 when @xmath166 .
the case @xmath167 is trivial since @xmath168 can only take the value 0 . ]
@xmath169 which is exactly the desired result .
this completes the proof of theorem [ thm : theorem2 ] . in appendix
[ app : alternative ] , we give an alternative proof of theorem [ thm : theorem2 ] with a different threshold on @xmath64 .
it is interesting to know to what extent theorem [ thm : theorem1 ] can be extended beyond the binary - input case .
this subsection is largely devoted to answering this question . for any @xmath0 and @xmath4 , define @xmath170:c(p_{y|x , s},p_s , p_{\tilde{s}_{ge}^{(\epsilon)}|s})\\ & \hspace{2.1in}=\underline{c}(p_{y|x , s},p_s)\},\\ & \underline{q}(p_{y|x , s},p_s)=\min\{q\in[0,\frac{1}{|\mathcal{s}|}]:c(p_{y|x , s},p_s , p_{\tilde{s}_{gs}^{(q)}|s})\\ & \hspace{2.25in}=\underline{c}(p_{y|x , s},p_s)\}.\end{aligned}\ ] ] [ prop : ns1 ] 1 .
there exists @xmath171 such that @xmath79 for all @xmath6 satisfying @xmath172 if and only if @xmath173 .
2 . @xmath173 if and only if @xmath174 where @xmath175 and @xmath176 are defined in ( [ eq : defdelta ] ) and ( [ eq : defutilde ] ) , respectively , @xmath76 is an arbitrary maximizer of the optimization problem in ( [ eq : nostate ] ) , and @xmath177 _ remark : _ all maximizers of the optimization problem in ( [ eq : nostate ] ) give rise to the same @xmath178 , @xmath96 , @xmath97 ( * ? ? ?
the first statement can be easily extracted from the proof of theorem [ thm : theorem1 ] .
now we proceed to prove the second statement .
first recall the definitions of @xmath179 and @xmath80 in ( [ eq : defd ] ) and ( [ eq : uconstruction ] ) , respectively . since @xmath80 is a capacity - achieving input distribution of channel @xmath69 when @xmath89 , we must have @xmath180 which , together with the fact @xmath181 , implies @xmath182 it can be verified that @xmath183 , u\in\mathcal{g}_{\delta}.\label{eq : xx3}\end{aligned}\ ] ] moreover , in view of ( [ eq : newsub ] ) , we can write ( [ eq : thm1ns ] ) equivalently as @xmath184 according to ( [ eq : xx1])([eq : xx4 ] ) , there exists @xmath185 such that @xmath186 for @xmath187 . in light of ( [ eq : intersect1 ] ) and the fact @xmath188 , we have @xmath189 combining ( [ eq : kkt1 ] ) , ( [ eq : kkt2 ] ) , and ( [ eq : intersect2 ] ) proves the if " part of the second statement .
next we turn to the only if " part of the second statement . assuming the existence of @xmath185 such that @xmath190 for @xmath187 ( or equivalently @xmath80 is a capacity - achieving input distribution of channel @xmath69 for @xmath187 )
, we must have @xmath191 it can be verified that @xmath192 , u\in\mathcal{u}\backslash\mathcal{g}_{\delta}.\label{eq : secondderivative}\end{aligned}\ ] ] moreover , by the definition of @xmath193 , @xmath194 note that ( [ eq : lower ] ) , ( [ eq : secondderivative ] ) , and ( [ eq : boundary ] ) hold simultaneously for @xmath195 , from which ( [ eq : thm1ns ] ) ( or equivalently ( [ eq : xx4 ] ) ) can be readily deduced .
this completes the proof of proposition [ prop : ns1 ] .
as shown by the following example , the necessary and sufficient condition ( [ eq : thm1ns ] ) is not always satisfied when @xmath196 .
let @xmath197 for this example , it can be verified that @xmath198 and @xmath199 where @xmath200 is given by @xmath201 , @xmath202 , and @xmath203 ; indeed , fig .
[ fig : thm1nonbinary ] shows that @xmath204 for @xmath115 .
the proof of proposition [ prop : ns1 ] in fact suggests a strategy for computing @xmath205 .
let @xmath76 be an arbitrary maximizer of the optimization problem in ( [ eq : nostate ] ) and define @xmath80 according to ( [ eq : uconstruction ] ) .
note that * @xmath206 for @xmath19 ( see ( [ eq : xx1 ] ) and ( [ eq : xx2 ] ) ) , * @xmath207 does not depend on @xmath42 for @xmath208 ( see ( [ eq : xx3 ] ) ) , * @xmath207 is a strictly convex function of @xmath42 for @xmath95 ( see ( [ eq : secondderivative ] ) ) .
hence , for each @xmath19 , there are three mutually exclusive cases . 1 .
@xmath209 : we have @xmath210 for @xmath211 $ ] , where @xmath212 .
2 . @xmath213 and @xmath214 ( this case can arise only when @xmath196 ) : we have @xmath215 for @xmath216 , where @xmath217 .
3 . otherwise : we have @xmath218 for @xmath216 and @xmath210 for @xmath211 $ ] , where @xmath219 is the unique solution of @xmath220 for @xmath221 .
it can be readily shown that @xmath222 we can compute @xmath223 in a similar way . define @xmath224 where @xmath225 with @xmath226 again , let @xmath80 be defined depends on @xmath10 .
in particular , @xmath227 when @xmath6 is a generalized symmetric channel whereas @xmath228 when @xmath6 is a generalized erasure channel . ]
according to ( [ eq : uconstruction ] ) .
it can be verified that @xmath229 , u\in\mathcal{u},\\ & \frac{\partial}{\partial q}d_{gs}(p_{\hat{u}},q , u)\\ & = \sum\limits_{y\in\mathcal{y},s\in\mathcal{s } } p_s(s)\delta(u , y , s)\\ & \hspace{0.5in}\times\log\frac{p_{y|x , s}(y|\psi(u , s),s)+q\omega(u , y , s)}{\sum_{x\in\mathcal{x}}p_{\hat{x}}(x)p_{y|x , s}(y|x , s)},\\ & \hspace{2 in } q\in[0,\frac{1}{|\mathcal{s}| } ] , u\in\mathcal{u},\\ & \frac{\partial^2}{\partial q^2}d_{gs}(p_{\hat{u}},q , u)\\ & = \sum\limits_{y\in\mathcal{y},s\in\mathcal{s}}\frac{p_s(s)\delta^2(u , y , s)}{p_{y|x , s}(y|\psi(u , s),s)+q\omega(u , y , s)}>0,\\ & \hspace{2in}q\in[0,\frac{1}{|\mathcal{s}| } ] , u\in\mathcal{u}\backslash\mathcal{g}_{\omega},\end{aligned}\ ] ] where @xmath230 clearly , * @xmath231 for @xmath19 , * @xmath232 does not depend on @xmath56 for @xmath233 , * @xmath232 is a strictly convex function of @xmath56 for @xmath234 . hence , for each @xmath19 , there are also three mutually exclusive cases .
1 . @xmath235 : we have @xmath236 for @xmath237 $ ] , where @xmath238 .
2 . @xmath239 and @xmath240 ( this case can arise only when @xmath196 ) : we have @xmath241 for @xmath242 , where @xmath243 .
3 . otherwise : we have @xmath244 for @xmath242 and @xmath236 for @xmath245 $ ] , where @xmath246 is the unique solution of @xmath247 for @xmath248 .
it can be readily shown that @xmath249 for @xmath0 and @xmath4 illustrated in fig .
[ fig : szchannel ] ( see also ( [ eq : para1 ] ) and ( [ eq : para2 ] ) ) , we show in appendix [ app : underline ] that @xmath250 where @xmath251 is the unique solution of @xmath252 for @xmath221 , and @xmath253 is the unique solution of @xmath254 for @xmath255
. setting @xmath256 gives @xmath257 ( cf .
[ fig : plot1 ] ) and @xmath258 ( cf . fig .
[ fig : plot2 ] )
. we shall extend theorem [ thm : theorem2 ] in a similar fashion .
for any @xmath0 and @xmath4 , define @xmath259:c'(p_{y|x , s},p_s , p_{\tilde{s}_{ge}^{(\epsilon)}|s})\\ & \hspace{2.2in}=\overline{c}(p_{y|x , s},p_s)\},\\ & \overline{q}(p_{y|x , s},p_s)=\max\{q\in[0,\frac{1}{|\mathcal{s}|}]:c'(p_{y|x , s},p_s , p_{\tilde{s}_{gs}^{(q)}|s})\\ & \hspace{2.3in}=\overline{c}(p_{y|x , s},p_s)\}.\end{aligned}\ ] ] [ prop : ns2 ] 1 .
there exists @xmath260 such that @xmath261 for all @xmath6 satisfying @xmath262 if and only if @xmath263 .
2 . @xmath263 if and only if there exists @xmath131 such that @xmath264 the first statement can be easily extracted from the proof of theorem [ thm : theorem2 ] .
the second statement is a consequence of lemma [ lem : nssym ] . as shown by the following example , the necessary and sufficient condition ( [ eq : constantset ] ) is not always satisfied when @xmath196 .
let @xmath265 for this example , it can be verified that the maximizer of the optimization problem in ( [ eq : perfectstate ] ) , denoted by @xmath154 , is unique and @xmath266 indeed , fig .
[ fig : thm2nonbinary ] shows that @xmath267 for @xmath268 $ ] . in view of lemmas [ lem : erasure ] and [ lem : nssym ]
, we have @xmath269 note that @xmath132 does not depend on @xmath4 ( under the assumption @xmath270 ) ; as a consequence , @xmath271 and @xmath272 do not depend on @xmath4 either . for @xmath0 and @xmath4
illustrated in fig .
[ fig : szchannel ] ( see also ( [ eq : para1 ] ) and ( [ eq : para2 ] ) ) , we show in appendix [ app : overline ] that @xmath273 setting @xmath256 gives @xmath274 ( cf . fig . [
fig : plot1 ] ) and @xmath275 ( cf .
[ fig : plot2 ] ) . in this subsection
, we shall examine the following two implicit conditions in theorem [ thm : theorem1 ] : 1 .
perfect state information at the decoder , 2 .
causal noisy state observation at the encoder .
if no state information is available at the decoder , then the channel capacity is given by @xmath276 where the joint distribution of @xmath32 is given by ( [ eq : jointdistribution ] ) .
furthermore , if there is also no state information available at the encoder , then the channel capacity becomes @xmath277 where @xmath28 .
define @xmath278:\tilde{c}(p_{y|x , s},p_s , p_{\tilde{s}_{ge}^{(\epsilon)}|s})\\ & \hspace{2.1in}=\underline{\tilde{c}}(p_{y|x , s},p_s)\}.\end{aligned}\ ] ] the proof of the following result is similar to that of proposition [ prop : ns1 ] and is omitted .
[ prop : ns3 ] 1 .
there exists @xmath279 such that @xmath280 for all @xmath6 satisfying @xmath281 if and only if @xmath282 .
@xmath282 if and only if @xmath283 where @xmath175 is defined in ( [ eq : defdelta ] ) , @xmath76 is an arbitrary maximizer of the optimization problem in ( [ eq : noedstate ] ) , and @xmath284 as shown by the following example , the necessary and sufficient condition ( [ eq : prop3ns ] ) is not always satisfied even when @xmath86 .
let @xmath285 where @xmath286 is the modulo-2 addition .
it can be verified that ( [ eq : prop3ns ] ) is not satisfied for this example ; indeed , fig .
[ fig : nsdecoder ] indicates that @xmath287 here we give an alternative way to prove ( [ eq : example ] ) .
write @xmath288 , where @xmath5 and @xmath289 are two mutually independent bernoulli random variables with @xmath290,\\ & p_{\delta}(1)=\frac{\mu-\nu}{1 - 2\nu}.\end{aligned}\ ] ] it is clear that @xmath291.\label{eq : ex1}\end{aligned}\ ] ] in light of lemma [ lem : erasure ] , @xmath6 is a stochastically degraded version of @xmath43 and consequently @xmath292 if @xmath293 . combining ( [ eq : ex1 ] ) and ( [ eq : ex2 ] ) proves ( [ eq : example ] ) .
now we proceed to examine the second implicit condition .
if the noisy state observation is available non - causally at the encoder , the gelfand - pinsker theorem @xcite ( see also ( * ? ? ?
7.3 ) ) indicates that the channel capacity is given by @xmath294 where the joint distribution of @xmath32 factors as @xmath295 it turns out that @xmath296 is bounded between @xmath41 and @xmath33 , i.e. , @xmath297 indeed , the first inequality is obvious , and the second one holds because @xmath298 in fig .
[ fig : gp ] we plot @xmath299 against @xmath42 for @xmath55 $ ] , where @xmath0 and @xmath4 are given by ( [ eq : para1 ] ) with @xmath256 and ( [ eq : para2 ] ) , respectively ; it can be seen that @xmath299 is strictly greater than @xmath51 except when @xmath89 .
so the causality condition on the noisy state observation at the encoder is not superfluous for theorem [ thm : theorem1 ] .
we have shown that the capacity of binary - input channels is very sensitive " to the quality of the encoder side information whereas the generalized probing capacity is very robust " . here
the words sensitive " and robust " should not be understood in a quantitative sense .
indeed , it is known @xcite that , when @xmath86 , the ratio of @xmath51 to @xmath50 is at least 0.942 and the difference between these two quantities is at most @xmath3000.011 bit ; in other words , the gain that can be obtained by exploiting the encoder side information ( or the loss that can be incurred by ignoring the encoder side information ) is very limited anyway .
binary signalling is widely used , especially in wideband communications .
so our work might have some practical relevance .
however , great caution should be exercised in interpreting theorems [ thm : theorem1 ] and [ thm : theorem2 ] .
specifically , both results rely on the assumption that the channel state takes values from a finite set and @xmath301 is not essential ] , which is not necessarily satisfied in reality ; moreover , the freedom of power control in real communication systems is not captured by our results .
nevertheless , our work can be viewed as an initial step towards a better understanding of the fundamental performance limits of communication systems where the transmitter side information and the receiver side information are not deterministically related .
finally , it is worth mentioning that our results might have their counterparts in source coding .
we shall show that , for any binary - input channel @xmath0 , state distribution @xmath4 , and side channel @xmath6 , @xmath61 if @xmath302 [ lem : newthreshold ] @xmath154 is a stochastically degraded version of @xmath6 if @xmath303 where @xmath304 let @xmath144 denote the maximum likelihood estimate of @xmath3 based on @xmath5 .
it suffices to show that @xmath151 is invertible and @xmath305 is a valid probability transition matrix if ( [ eq : anotherbound ] ) is satisfied . [ cols="^,^,^,^",options="header " , ] [ lem : aninequality ] for @xmath306 , @xmath307 we have @xmath308 which , together with the fact @xmath309 , implies the desired result .
when @xmath52 or @xmath53 , we have @xmath310 , which implies @xmath311 . when @xmath306 , the maximizer of the optimization problem in ( [ eq : nostate ] ) , denoted by @xmath76 , is unique and is given by @xmath312 now consider @xmath313 specified by table [ tab1 ] .
it can be verified that @xmath314 moreover , @xmath315 where ( [ eq : invokeineq1 ] ) and ( [ eq : invokeineq2 ] ) follow from lemma [ lem : aninequality ] .
therefore , we have @xmath316 which , together with ( [ eq : maxepsilon ] ) , proves ( [ eq : proofinappb1 ] ) for @xmath306 .
next consider @xmath313 specified by table [ tab2 ] .
it can be verified that @xmath317 moreover , @xmath318 where ( [ eq : invokeineq1again ] ) and ( [ eq : invokeineq2again ] ) follow from lemma [ lem : aninequality ] .
therefore , we have @xmath319 which , together with ( [ eq : maxq ] ) , proves ( [ eq : proofinappb2 ] ) for @xmath306 .
when @xmath52 or @xmath53 , we have @xmath310 , which implies @xmath320 and @xmath321 .
when @xmath306 , the maximizer of the optimization problem in ( [ eq : perfectstate ] ) , denoted by @xmath154 , is unique and is given by @xmath322 in view of ( [ eq : invewepsilon ] ) and ( [ eq : inviewq ] ) , it suffices to show that @xmath323 indeed , for @xmath306 , @xmath324 and the last inequality is true according to lemma [ lem : aninequality ] .
the authors wish to thank the associate editor and the anonymous reviewer for their valuable comments and suggestions .
j. wang , j. chen , l. zhao , p. cuff , and h. permuter , on the role of the refinement layer in multiple description coding and scalable coding , " _ ieee trans .
inf . theory _ ,
3 , pp . 14431456 , mar . | for any binary - input channel with perfect state information at the decoder , if the mutual information between the noisy state observation at the encoder and the true channel state is below a positive threshold determined solely by the state distribution , then the capacity is the same as that with no encoder side information .
a complementary phenomenon is revealed for the generalized probing capacity .
extensions beyond binary - input channels are developed .
= [ draw , -stealth , semithick ] = [ draw , rectangle , text width=7em , text centered , minimum height=12 mm , node distance=8em , semithick ] = [ draw , rectangle , text width=4em , text centered , minimum height=40 mm , node distance=8em , semithick ] binary - input , channel capacity , erasure channel , probing capacity , state information , stochastically degraded . |
in @xcite , the authors introduce a model for stochastic gene expression to study the heterogeneity of cell populations .
they assume that the cells , or for example the product of some gene , can be in two distinct states or colonies .
let @xmath0 and @xmath1 be the sizes of these colonies , which are here considered as birth with migration processes .
we assume that the birth rates are either @xmath2 or @xmath3 with @xmath4 , and that the associated migration rates @xmath5 and @xmath6 are such that @xmath7 , that is cells located in the colony having the smaller birth rate migrate at a higher rate to the colony with the higher birth rate than the other way round .
if the birth and migration rates are assigned once and for all to a corresponding colony ( e.g. @xmath3 and @xmath6 to @xmath8 , and @xmath2 and @xmath5 to @xmath9 ) , then the mean sizes @xmath10 and @xmath11 satisfy the pair of differential equations ( see @xcite or @xcite ) @xmath12 according to @xcite , we say that cells of the first colony represented by @xmath0 are _ unfit _ ( they have the lower birth rates ) , and conversely that cells of the second colony represented by @xmath1 are _ fit_. the proportion of fit cells in the global population , @xmath13 , @xmath14 , satisfies the non - linear differential equation @xmath15 then , as @xmath16 , @xmath17 , where @xmath18 follows directly from ( [ fit ] ) , see section [ s.stationary ] .
this describes the equilibrium value of the proportion of fit cells in a non - changing environment .
fixing the values of the parameters @xmath6 , @xmath3 and @xmath2 , we can ask for the value of @xmath19 which maximizes the proportion of fit cells , i.e. the equilibrium value of @xmath20 : the optimal strategy is to keep all the fit cells in the fit state , that is to set their migration rate to zero , @xmath21 .
this leads to @xmath22 , and thus the optimal solution would be a homogeneous population .
observations reveal however that most cell populations are not homogeneous ; to explain this , the authors of @xcite propose to introduce a small modification in the model by allowing environmental changes ( for related questions in this context , see e.g. @xcite ) , and show through monte - carlo simulations that the homogeneous solution @xmath21 is then not always optimal .
the idea in their model is to allow the birth and migration rates to switch at random times from one colony to the other , so that cells in the fit colony become unfit and vice versa .
if for example an environmental change occurs at some random time @xmath23 ( @xmath24 ) , then the function @xmath25 representing the proportion of fit cells solves ( [ fit ] ) up to time @xmath26 , and just after @xmath26 , say at time @xmath27 , the fit cells corresponding to @xmath1 become unfit and vice versa .
the proportion of fit cells @xmath25 is then switched to @xmath28 .
after @xmath26 , the random process @xmath29 solves ( [ fit ] ) with initial data @xmath30 at time @xmath27 , until a new environmental change occurs , say at time @xmath31 .
there is a new switch , and the process is again solution of ( [ fit ] ) , until a new event occurs and so on . in @xcite ,
the fluctuations of the environment are modeled using a renewal process ; the instants @xmath32 , @xmath33 , are such that the sequence of random variables @xmath34 given by @xmath35 , @xmath36 , is i.i.d .
distributed according to some law @xmath37 on @xmath38 .
the authors then use monte - carlo simulations to estimate the limiting value of the time averages along trajectories of the process @xmath25 , of the form @xmath39 this limiting average value is denoted by @xmath40 to express its dependency on the migration rate @xmath41 , when all the remaining parameters are fixed .
their simulations indicate that there is a range of parameters ( @xmath6 not too large ) such that @xmath42 which means that heterogeneous populations are more adapted than homogeneous ones in a switching environment . in this paper
, we study mathematically the limiting behavior of the stochastic process @xmath25 and the associated time average @xmath43 by giving its stationary measure , and we provide mathematical formulas and numerical solutions , which might be of interest in practical laboratory experiments ( see e.g. @xcite ) .
our technique uses the process @xmath44 , @xmath45 , which is such that @xmath46 , for some mapping @xmath47 ( see definition [ markovchain ] ) .
@xmath48 is a stochastic recursive markov chain , and @xmath43 can be expressed as an additive functional of the trajectory of @xmath49 . in section [ s.stationary ] , we recall a theorem from @xcite on the convergence of stochastic recursive chains , which applies in this setting .
we give conditions ensuring the existence and uniqueness of a stationary measure @xmath50 , as well as geometric ergodicity . in section [ stationarity ]
, we consider the case where @xmath37 is exponential of parameter @xmath51 , and show that @xmath50 has a @xmath52 density @xmath53 with respect to lebesgue measure .
we furthermore prove in theorem [ ergodic ] that a multiple @xmath54 of @xmath53 solves a second order differential equation with weak singularities .
proposition [ diffsolutions ] provides series expansions for @xmath53 , which are necessary to derive properties of @xmath53 near the singularities . in section [ s.numerical ] ,
we show numerical solutions , using the series expansions of proposition [ diffsolutions ] to start the numerical integration .
we provide an example where @xmath55 , which shows that it can be better to allow fit cells to migrate to the unfit state than to conserve all the fit cells in the fit state in such a switching environment .
this is a regime where it is suitable for the colonies to anticipate bad hypothetical future events .
we first give some basic results for the differential equation ( [ fit ] ) . the right hand side of ( [ fit ] ) can be factored into @xmath56 , where @xmath57 , @xmath58 , and @xmath59 .
then @xmath60 implies that @xmath61 and that the derivative @xmath62 is positive when @xmath25 is in the interval @xmath63 , negative in @xmath64 $ ] , and it vanishes for @xmath65 .
it is not hard to check that any realization of the trajectory @xmath29 , with initial data @xmath66 will remain forever in @xmath67 , and that any trajectory starting in the interval @xmath68\setminus i$ ] will enter @xmath67 after an almost surely finite time .
( however , @xmath69 implies @xmath70 . )
we thus restrict our study to the interval @xmath67 .
given @xmath71 , we define the mapping @xmath72 such that @xmath47 is the value of the solution of ( [ fit ] ) at time @xmath73 when starting at @xmath74 at time @xmath75 . using separation of variables for ( [ fit ] )
, we obtain the relation @xmath76 where we set @xmath77 . given @xmath78 , let @xmath79 denote the time interval the orbit of the dynamical system ( [ fit ] ) needs to join @xmath80 and @xmath81 , @xmath82 , when starting at time @xmath83 at @xmath80 .
then @xmath84 [ markovchain ] given @xmath85 , consider the markov chain with values in @xmath67 defined by @xmath86 where the sequence of random variables @xmath87 is i.i.d .
distributed according to some law @xmath37 on @xmath38 .
this markov chain describes the evolution of @xmath88 , at the instants just before the switches , with @xmath89 .
we first recall and adapt results of @xcite on the convergence of such markov chains , also called _ stochastic recursive chains _ , see e.g. @xcite .
the general setting is described by a complete separable metric space @xmath90 , the set of values taken by the markov chain , a family of mappings @xmath91 , indexed by parameters @xmath92 living in some parameter space @xmath93 , and a probability measure @xmath37 on @xmath93 . given an i.i.d .
sequence of random elements @xmath94 , @xmath95 , of law @xmath37 , we can consider the markov chain @xmath96 given by @xmath97 .
the following theorem gives conditions for the existence and uniqueness of a stationary measure ( theorem 1.1 of @xcite ) . in what follows
, @xmath98 denotes the law of the markov chain @xmath99 and @xmath100 $ ] is the prokhorov metric , see below .
[ diacconv ] assume that the family of functions @xmath101 , @xmath102 is lipschitz with @xmath103 @xmath104 . assume furthermore that @xmath105 for some @xmath106 , and that @xmath107 then * the markov chain has a unique stationary distribution @xmath50 , * @xmath100{\leqslant}a_x r^n$ ] , for constants @xmath108 and @xmath109 with @xmath110 and @xmath111 ; this bound holds for all times @xmath112 and all starting positions @xmath113 , * the constant @xmath109 does not depend on @xmath112 or @xmath113 ; the constant @xmath108 does not depend on @xmath112 , and @xmath114 , where @xmath115 . in our setting ,
@xmath116 is given by @xmath67 and the parameter set @xmath117 is just @xmath38 .
the prokhorov distance @xmath118 $ ] is the infimum of the @xmath119 such that @xmath120 where @xmath121 runs over the borel sets of @xmath67 and , for given @xmath122 , @xmath123 denotes the set of points of @xmath67 whose distance from @xmath121 is less than @xmath124 ( see section 5.1 of @xcite ) .
condition ( [ contraction ] ) means that the functions @xmath101 are contractions in the average .
we first express this condition in our setting : for @xmath125 and @xmath126 , the mapping @xmath127 is given explicitly by @xmath128 setting @xmath129 , we obtain [ lip ] for all @xmath71 , @xmath130 @xmath131 if @xmath37 is exponential of parameter @xmath51 , and @xmath132 , then the conditions given in ( [ trivcondi ] ) are satisfied , and @xmath133 where we set @xmath134 . condition ( [ contraction ] ) is thus satisfied if @xmath135 when @xmath136 , the integral @xmath137 is the lerch phi function @xmath138 , with @xmath139 and @xmath140 , and is also equal to gauss s hypergeometric function @xmath141 ( see e.g. @xcite , chap .
1.11 ) .
* proof * : taking the derivative of ( [ explicitphi ] ) with respect to @xmath80 , we obtain @xmath142 and thus @xmath143 as required .
the expression for @xmath144 follows from direct computation .
assume that @xmath37 is exponential of parameter @xmath51 .
we will see in the sequel that the stationary measure @xmath50 has , under some conditions , a density @xmath145 such that with @xmath146 , where @xmath132 , the function @xmath147 satisfies the differential equation @xmath148 where @xmath149 , @xmath150 , @xmath151 and @xmath152 the following proposition will therefore be useful : [ diffsolutions ] the solutions of the second order homogeneous linear differential equation ( [ diff ] ) are analytic on the interval @xmath153 .
two fundamental solutions @xmath154 , @xmath155 are * @xmath156 , where @xmath157 is analytic on @xmath158 for some @xmath159 and with @xmath160 . * @xmath161 + with @xmath162 analytic on @xmath163 for some @xmath159 , @xmath164 and @xmath165 .
another set of two fundamental solutions @xmath166 , @xmath167 is * @xmath168 , where @xmath169 is analytic on @xmath170 for some @xmath159 and with @xmath171 . *
@xmath172 + with @xmath173 analytic on @xmath170 for some @xmath159 , @xmath174 and @xmath175 . in the appendix
, we show this result for completeness , and also how these fundamental solutions can be computed by series expansion about @xmath176 and @xmath177 respectively .
[ ergodic ] assume that @xmath178 where @xmath134 .
then the markov chain @xmath179 from definition [ markovchain ] , with initial data @xmath180 has a unique stationary distribution @xmath50 of @xmath181 density @xmath182 here , @xmath183 , where @xmath132 , @xmath157 is the analytic function on @xmath158 with @xmath184 , such that @xmath156 is a solution of the differential equation ( [ diff ] ) . in the neighborhood of @xmath185 ,
this solution is such that @xmath186 .
finally , the behavior of the density @xmath53 near @xmath177 is given by @xmath187 , and thus converges when @xmath188 and diverges toward @xmath189 when @xmath190 .
let @xmath191 and @xmath192 be defined on @xmath67 .
then @xmath193 with @xmath194 relation ( [ moment ] ) will be useful when considering time averages for monte - carlo simulations , see section [ average ] .
* proof * : the existence and uniqueness of the stationary measure follows from theorem [ diacconv ] and lemma [ lip ] .
let @xmath9 be a random variable of law @xmath50 , and let @xmath195 be exponential of parameter @xmath51 , independent of @xmath9 . in the stationary regime , @xmath196 .
let @xmath197 .
then @xmath198 where @xmath199 denotes the indicator function . for given @xmath200
, the time variable @xmath73 is restricted to the interval @xmath201 , see ( [ timeinterval ] ) .
thus @xmath202 for given @xmath73 in this interval , the set of @xmath203 with @xmath204 is given by @xmath205 it follows that @xmath206 , where we set @xmath207 , with @xmath208 .
we make the change of variable @xmath209 with @xmath210 then @xmath211 this is a fixed point equation for the distribution function @xmath212 .
we use it for proving that the probability measure @xmath50 has a @xmath52 density on the interval @xmath67 .
first notice that @xmath212 is monotonically increasing and integrable on @xmath67 .
the above relation then shows that @xmath212 is continuous on @xmath67 . using again this argument recursively
, one sees that @xmath212 is the integral of a continuous function and is therefore differentiable , with a continuous derivative .
the @xmath181 differentiability is obtained by iterating this argument .
let @xmath53 be the @xmath181 density of @xmath50 with respect to lebesgue measure .
our strategy runs as follows : we use the fixed point relation to show that a multiple @xmath54 of @xmath53 satisfies a second order differential equation , which has only weak singularities , and then deduce properties of @xmath53 with the help of proposition [ diffsolutions ] . for given @xmath203 ,
the time variable @xmath73 is restricted to the interval @xmath213 where @xmath214 ( see [ timeinterval ] ) .
it follows that @xmath215 with @xmath216 where we set @xmath146 . using @xmath214 and setting @xmath147
, one gets @xmath217 where @xmath218 is such that @xmath219 , and @xmath220 . taking the derivative gives @xmath221 or @xmath222 taking a second derivative then gives @xmath223 and simplifying the terms leads to ( [ diff ] ) .
we see that @xmath224 , as @xmath225 .
the exponents associated with the fundamental solutions are @xmath226 or @xmath227 in the neighborhood of @xmath228 and @xmath229 or @xmath230 near @xmath185 .
assume first that @xmath231 .
we first check the behavior of @xmath54 in a neighborhood of @xmath232 .
set @xmath233 , @xmath234 , with @xmath235 .
proposition [ diffsolutions ] shows that @xmath54 is a linear combination @xmath236 , for constants @xmath237 , @xmath238 .
similarly , @xmath239 , for real constants @xmath240 and @xmath241 . as @xmath242
, the right hand side of ( [ relc ] ) behaves like @xmath243 .
suppose that @xmath244 .
then @xmath245 , and ( [ relc ] ) ca nt be satisfied .
one must thus have @xmath246 , so that @xmath247 .
when @xmath248 , ( [ relc ] ) implies that @xmath249 .
it follows that , for arbitrary @xmath250 , @xmath251 , and that @xmath252 , @xmath242 , as required .
the corresponding result for @xmath53 follows .
suppose that @xmath253 .
the right hand side of ( [ relc ] ) behaves like @xmath254 with @xmath255 as @xmath242 , and @xmath256 behaves like @xmath257 one has @xmath258 , @xmath242 , when @xmath244 and @xmath259 , when @xmath246 .
( [ relc ] ) shows that necessarily @xmath246 .
suppose that @xmath260 .
then one must have @xmath261 , implying the existence of the limit @xmath262 .
when @xmath248 , @xmath249 , @xmath263 , and @xmath264 , as required .
finally , we check the identity ( [ moment ] ) .
first @xmath265 follows from the behavior of the density @xmath53 at the boundaries of @xmath67 , as described above .
next , @xmath266 where @xmath267 is such that @xmath268 where we use ( [ deriv ] ) .
it follows that @xmath269 proving ( [ moment ] ) since @xmath270 . [ limitinglaw ]
assume that condition ( [ condi ] ) holds .
as @xmath271 , the law of the stochastic process @xmath25 , @xmath14 , @xmath272 , converges toward the stationary measure @xmath50 of density @xmath53 of the markov chain @xmath179 . * proof * : given @xmath71 , let @xmath273 be the last renewal time before @xmath73 , and set @xmath274 .
when the length of the overlapping random interval is exponential , @xmath275 is also exponential . in the stationary regime , or equivalently for large @xmath73
, one has the identity in law @xmath276 , where @xmath8 is distributed according to @xmath50 , and the result follows .
when the conclusions of theorem [ diacconv ] hold , the chain @xmath179 has a unique stationary probability measure @xmath50 , and @xmath277 converges a.s . toward the expectation of @xmath278 under @xmath50 , for any function @xmath278 in @xmath279 , ( see e.g. @xcite ) . in @xcite ,
the authors use monte - carlo methods based on the process @xmath25 , @xmath14 , to estimate the mean fitness by considering the time average @xmath280 where @xmath281 is a fixed number of renewal periods .
[ formula ] let @xmath282 . given a realization @xmath283 of the renewal process
, we have @xmath284 * proof * : consider the integrals @xmath285 where @xmath286 and @xmath287 . the value of @xmath288
, @xmath289 is given implicitly by ( [ basicrelation ] ) ; therefore @xmath290 where we set @xmath291 , and thus , after a longer but not difficult computation , one obtains @xmath292 and the result follows , since @xmath293 [ convergencetimeaverage ] suppose that @xmath37 is exponential of parameter @xmath51 , and assume ( [ condi ] ) .
let @xmath191 and @xmath192 be defined on @xmath67 .
then @xmath294 * proof * : from equation ( [ timeaverageb ] ) , we obtain @xmath295 as @xmath296 is a renewal process with exponential inter arrival times of parameter @xmath297 , it follows that @xmath298 converges a.s . toward @xmath299
next , @xmath300 follows from the behavior of the density @xmath53 at the boundaries of @xmath67 , as described in theorem [ ergodic ] . from proposition [ diffsolutions ] and theorem [ ergodic ] ,
the behavior of @xmath53 in the neighborhood of @xmath228 is given by @xmath301 where @xmath302 and by @xmath303 in the neighborhood of @xmath185 , where @xmath304 .
the markov chain @xmath305 is geometrically ergodic , and thus the last term converges a.s . toward @xmath306 .
we finally check that @xmath307 converges a.s . toward 0 .
given @xmath308 , consider the probability @xmath309 where @xmath310 . using the behavior of @xmath53 in the neighborhood of @xmath185 , one gets that @xmath311 , for some positive constant @xmath312 .
let @xmath313 , and let @xmath314 be the prokhorov distance defined in ( [ prok ] ) . if @xmath315 , then @xmath316 . if @xmath317 , one has @xmath318 , and it follows that @xmath319 for some positive constant @xmath320 .
theorem [ diacconv ] gives that @xmath321 for some bounded function @xmath322 and a positive number @xmath323 .
the result then follows from the borel - cantelli lemma .
the last identity is ( [ moment ] ) of theorem [ ergodic ] .
we now compute the density @xmath53 given in theorem [ ergodic ] numerically .
to do so , we solve the differential equation ( [ diff ] ) numerically , starting in the neighborhood of the singular point @xmath324 .
proposition [ diffsolutions ] and theorem [ ergodic ] show that @xmath325 , and that the first derivative of @xmath53 behaves like @xmath326 , which goes to @xmath189 when @xmath190 .
we start the numerical solution at the point @xmath327 , where @xmath328 is small , and use the initial conditions @xmath329 and @xmath330 from the series expansions given in proposition [ diffsolutions ] . in addition
, we use the numerical integration procedure to compute the integral to scale the density @xmath53 , by adding an additional ordinary differential equation to ( [ diff ] ) .
we show in figures [ fig12 ] to [ fig0 ] the results obtained for five different sets of parameters . on the left
when @xmath331 , @xmath332 , @xmath333 , @xmath334 , and on the right when @xmath335 , @xmath332 , @xmath336 , @xmath334.,title="fig:",scaledwidth=49.0% ] on the left when @xmath331 , @xmath332 , @xmath333 , @xmath334 , and on the right when @xmath335 , @xmath332 , @xmath336 , @xmath334.,title="fig:",scaledwidth=49.0% ] on the left when @xmath337 , @xmath338 , @xmath336 , @xmath21 .
@xmath339 , and on the right when @xmath337 , @xmath338 , @xmath336 , @xmath334 .
clearly the average fitness is larger when @xmath334 than when @xmath21.,title="fig:",scaledwidth=49.0% ] on the left when @xmath337 , @xmath338 , @xmath336 , @xmath21 .
@xmath339 , and on the right when @xmath337 , @xmath338 , @xmath336 , @xmath334 .
clearly the average fitness is larger when @xmath334 than when @xmath21.,title="fig:",scaledwidth=49.0% ] when @xmath341 , @xmath342 , @xmath343 , @xmath344 , a case where @xmath345.,scaledwidth=49.0% ] in all the figures , we show the computed solution @xmath54 of the differential equation ( [ diff ] ) in dashed , the computed density @xmath53 as a solid line , and the results of a monte - carlo simulation with 100000 samples as circles . the density from the theory and the monte - carlo simulations agree very well .
it is interesting to see in figures [ fig12 ] and [ fig34 ] the variety of densities that can be generated by this simple model .
figure [ fig34 ] contains a case where increasing @xmath5 increases the overall fitness of the population .
figure [ fig0 ] finally shows a case where @xmath345 .
we note that the numerical integration out of the singularity can be challenging .
in particular , for the first case in figure [ fig12 ] , the standard ode45 from matlab needed very small absolute tolerances to succeed with the integration for @xmath346 .
a more robust method turned out to be dopri853 , see @xcite .
in this appendix we show for completeness the proof of proposition [ diffsolutions ] and describe a method how to solve the differential equation ( [ diff ] ) ( see also @xcite , pp .
317 - 321 ) .
this equation is of the form @xmath347 where the functions @xmath348 and @xmath349 are meromorphic in the complex plane with four poles of order one at @xmath350 .
the solutions are therefore analytic in the open disc of radius @xmath351 centered at @xmath352 .
we look for real solutions in the interval @xmath153 . in order to simplify calculations
, we use the variable transformation @xmath353 and set @xmath354 . with this transformation , the differential equation ( [ diff ] )
becomes @xmath355 where @xmath80 and @xmath356 have four poles of order one at the points @xmath357 with @xmath358 : @xmath359 we can therefore rewrite this equation as @xmath360 where @xmath361 and @xmath362 are analytic in the disc of radius @xmath363 centered at 0 : @xmath364 multiplying the equation ( [ gendiff ] ) by @xmath365 we get an equivalent equation which can be written as @xmath366 where @xmath367 denotes differentiation and @xmath368 multiplication by a function @xmath369 . looking for solutions of the form
@xmath370 we may identify the function @xmath371 with the infinite row @xmath372=[1,w_1,w_2,w_3,\ldots]$ ] and write ( [ opdiff ] ) in matrix form : @xmath373[w]^{\rm t}=0.\ ] ] if we write @xmath374 as @xmath375 , we get the lower triangular matrix @xmath376=\left[\begin{array}{ccccc } \rho(\rho+\alpha_0 - 1)+\beta_0&0&\ldots\\ \rho \alpha_1+\beta_1&(\rho+1)(\rho+\alpha_0)+\beta_0&0&\ldots\\ \rho \alpha_2+\beta_2&(\rho+1)\alpha_1 + \beta_1&(\rho+2)(\rho+1+\alpha_0)+\beta_0&0&\ldots\\ \vdots&\vdots&\vdots&\vdots&\ddots \end{array}\right].\ ] ] a solution @xmath372=[1,w_1,w_2,\ldots]$ ] of the linear system ( [ lindiff ] ) exists if and only if @xmath377 .
this is the so - called indicial equation for @xmath378 . from now on we
shall no longer treat the general case but only the case corresponding to our differential equation ( [ newdiff ] ) . in this case @xmath379 and @xmath380 .
so the indicial equation is @xmath381 and yields the two characteristic exponents @xmath302 and @xmath382 .
we shall write @xmath383 instead of @xmath384 . for @xmath385 , the solution @xmath386=[1,w_1^{(1)},w_2^{(1)},\ldots]$ ]
may be calculated by the recursion scheme @xmath387 with these coefficients @xmath388 , the function @xmath389 is a solution of ( [ newdiff ] ) . from the general theory of linear differential equations in the complex plane
it follows that @xmath390 is analytic in the disc of radius @xmath352 centered at @xmath352 , but the power series for @xmath391 might have a convergence radius @xmath392 .
if @xmath227 is not an integer , another solution @xmath393 , linearly independent of @xmath394 , can be obtained in the same way from @xmath395 .
if , however , @xmath227 is an integer , the corresponding matrix has the entry @xmath396 for @xmath397 , and we look in this case for a solution @xmath393 of the form @xmath398 . as @xmath390 is a solution , the terms in @xmath399 containing @xmath400 cancel and the function @xmath401 must satisfy the equation @xmath402 identifying @xmath403 with the infinite row @xmath404=[1,w_1^{(2)},w_2^{(2)},\ldots]$ ] , we can write this in matrix form @xmath405[w^{(2)}]^{\rm t}=-c[v_1,v_2,\ldots]^{\rm t}.\ ] ] for the right - hand side one checks easily that @xmath406 for @xmath407 and @xmath408 .
therefore we can resolve the inhomogeneous linear system ( [ lindiff2 ] ) in the following way : 1 .
we determine @xmath409 for @xmath410 in the same way as @xmath411 .
2 . we set @xmath412 and determine the constant @xmath121 by the equation @xmath413 .
3 . we determine the coefficients @xmath414 for @xmath415 by the recursion formula @xmath416 we shall not go into further details , for example present concrete formulas expressing the @xmath417 by the @xmath388 , because we do nt really need the solution @xmath418 of ( [ gendiff ] ) in our case , as we have shown in the proof of theorem [ ergodic ] . in order to find fundamental solutions near the singularity @xmath177 , we can apply the same method once more , but using the variable transformation @xmath421 one easily checks that in this case the indicial equation is @xmath422 and that therefore the two characteristic exponents at @xmath177 are @xmath423 we obtain thus the second fundamental system of solutions @xmath166 and @xmath167 . 3 ( 1998 ) .
_ ergodicity and stability of stochastic processes _ , wiley series in probability and statistics , new - york .
( 1960 ) . the strong law of large numbers for a class of markov chains .
_ , * 31 , * 801803 . iterated random functions .
_ siam review _ , * 41 , * 4576 . _
higher transcendental functions . _
bateman manuscript project .
mcgraw - hill ( 2000 ) _ solving ordinary differential equations i , nonstiff problems _ , springer series in computational mathematics .
second edition .
_ analysis fr physiker und ingenieure _ , springer verlag .
( 1973 ) birth , death and migration processes .
_ biometrika _ , * 59 , * 4969 .
( 1991 ) _ modelling biological populations in space and time _ , cambridge studies in mathematical biology .
cambridge university press .
( 2004 ) stochastic gene expression in fluctuating environments _ genetics _ , * 167 , * 523530 . | we study a stochastic model proposed recently in the genetic literature to explain the heterogeneity of cell populations or of gene products .
cells are located in two colonies , whose sizes fluctuate as birth with migration processes in switching environment .
we prove that there is a range of parameters where heterogeneity induces a larger mean fitness .
_ keywords : _ gene expression , recursive chain , ergodic , stationary measure |
klein @xcite considered the reflection and transmission of electrons of energy @xmath1 incident on the potential step @xmath2 for @xmath3 in the one - dimensional time - independent dirac equation which can be written @xcite in terms of the usual pauli matrices as @xmath4 he found that the reflection and transmission coefficients @xmath5 when the step height @xmath6 were given by @xmath7 where @xmath8 is the kinematic factor @xmath9 @xmath10 is the momentum of the incident electron and @xmath11 is the momentum of the transmitted particle for @xmath12 provided @xmath13 is negative to be negative so that the group velocity of the particle for @xmath14 was positive . ] , @xmath15 and @xmath16 are both positive and satisfy @xmath17 telegdi @xcite has recently reviewed the history of the klein paradox , concentrating on the inaccuracies which followed from choosing the wrong sign of @xmath18 nowadays the paradox is taken to lie in the result that as @xmath19 for fixed @xmath20 tends to a non - zero limit .
the physical essence of the klein paradox thus lies in the prediction that fermions can pass through large repulsive potentials without exponential damping .
this we call klein tunnelling
. there have been various attempts over the last seventy years to explain the klein paradox . the preferred explanation given by telegdi and in textbooks
@xcite is due to hansen and ravndal @xcite : the left - right asymmetric step emits electron - positron pairs just as a constant electric field does @xcite and it is the inability of an observer to distinguish between this emission and the static scattering properties of the step which gives rise to the paradox .
@xmath21 measures the positrons emitted by the step while @xmath15 is not unity because the pauli principle causes destructive interference between the emitted electron and the reflected electron . in the absence of particle production
@xmath22would indeed be @xmath23 and @xmath21 would be @xmath24 these claims are investigated in this paper .
we begin from a viewpoint which sees no connection between left - right asymmetry and quantum tunnelling , nor indeed any necessary connection between either of these and particle production .
we will seek to show that klein tunnelling is characteristic of relativistic wave equations as such , because negative energy states really do represent physical particles .
our approach in this respect is similar to that of jensen et al @xcite .
we therefore investigate a class of problems in which klein tunnelling takes place even in the absence of particle production .
klein was unfortunate in that he considered a pathological problem which gives misleading results : the potentials we consider can be very similar to a klein step but are better defined and both their time - independent and emission properties can be found simply using eq.([dir ] ) in the presence of a four - potential @xmath25 .
particle emission from strong potentials is well - understood in terms of supercritical positron production @xcite .
a full field theoretic treatment is given in our earlier paper @xcite but provided we restrict ourselves to potentials which are switched on adiabatically we manage here with first quantisation ( except in the appendix where we derive the relation between the transmission coefficient @xmath21 and the current emitted by the step ) .
finally we look at klein tunnelling in the context of the coulomb potential .
consider the square barrier @xmath26 instead of the klein step . for @xmath27 this potential
may be expected to give similar results to those found by klein .
it does , but the results are perhaps surprising .
it is easy to show that the reflection and transmission coefficients for the square barrier are given by @xcite @xmath28 note that klein tunnelling is enhanced for a barrier : if @xmath29 corresponding to @xmath30 the electron passes right through the barrier with no reflection : this is called a transmission resonance @xcite . as @xmath31 becomes very large for fixed @xmath32 and @xmath33 ,
we can average over the phase angle @xmath34 to find the limit @xmath35 it may seem unphysical that @xmath36 and @xmath37 are not the same as @xmath15 and @xmath21 but it is not : it is well known in electromagnetic wave theory @xcite that reflection off a transparent barrier of large but finite width ( with 2 sides ) is different from reflection off a transparent step ( with 1 side ) . @xmath36 and
@xmath38 thus involve klein tunnelling but they arise from a more physical problem than the klein step .
the zero of potential is properly defined for a barrier whereas it is arbitrary for a step and the energy spectrum of a barrier is easily calculable .
emission from a barrier or well is described by supercriticality : the condition when the ground state energy of the system overlaps with the continuum ( @xmath39 for a barrier ; @xmath40 for a well ) and so any connection between particle emission and the time - independent scattering coefficients @xmath41 and @xmath42 can be investigated .
we discussed this topic in our previous paper @xcite which we refer to as cdi .
we quickly review the argument .
spontaneous fermionic emission is a non - static process and in the case of a seemingly static potential , it is necessary to ask how the potential was switched on from zero .
we follow cdi in turning on the potential adiabatically . the bound state spectrum for the well @xmath43 is easily obtained @xcite : there are even and odd solutions given by the equations @xmath44 @xmath45 where now the well momentum is given by @xmath46 .
we have changed the sign of @xmath33 so that it is now attractive to electrons rather than positrons in order to conform with other authors who have studied supercritical positron emission rather than electron emission .
the simplest case to discuss is a very narrow deep well @xmath47 which is the limit of a square well with @xmath48 .
the bound states are then given for even @xmath49 and odd ( @xmath50 wave functions by @xmath51 when the potential is initially turned on and @xmath52 is small there is one bound state just below @xmath39 . as @xmath52 increases , @xmath1 decreases and at @xmath53 @xmath54 reaches zero . for @xmath55 , @xmath1 becomes negative and if the well were originally vacant , we now have the absence of a negative energy state which we must interpret as the presence of a ( bound ) positron according to dirac s hole theory .
we can use the anticommutation relations for the electron field @xcite to write the charge @xmath56 as @xmath57 where @xmath58 is a c - number called the vacuum charge @xcite@xcite and @xmath59 is the particle ( or normal - ordered ) charge .
the total charge @xmath56 is always conserved .
for @xmath52 just larger than @xmath60 , @xmath61 because of the presence of the positron and so the vacuum charge @xmath58 must now equal @xmath62 to conserve charge@xmath63 as the potential is increased further , @xmath52 will reach @xmath64 here @xmath40 which is the condition for supercriticality : the bound positron reaches the continuum and becomes free .
this is the well - known scenario of spontaneous positron production first discussed @xcite@xcite over 25 years ago .
note that at supercriticality @xmath65 , the even bound state disappears and the first odd state appears .
we can continue to increase @xmath52 and count positrons : the total number of positrons produced for a given @xmath52 is the number of times @xmath1 has crossed @xmath66 that is @xmath67 \label{charge0}\ ] ] where @xmath68 $ ] denotes the integer part of @xmath69 the more interesting quantity for us is the number of supercritical positrons @xmath70 : the number of states which have crossed @xmath40 .
this is given by @xmath71 \label{super}\ ] ] note that for any @xmath52 there is at most one bound positron state .
we consider the general case of @xmath6 and then look in particular at the case @xmath27 most closely corresponding to the klein step .
we must find first the condition for supercriticality of the potential @xmath33 , and then the number of bound and supercritical positrons produced for a given @xmath72 from eq ( [ even ] ) we see that the ground state becomes supercritical when @xmath73 and therefore @xmath74 from eq ( [ odd ] ) the first odd state becomes supercritical when @xmath75 and @xmath76 clearly the supercritical potential corresponding to the nth positron is @xmath77 it follows from eq ( [ nth ] ) that @xmath78 is an accumulation point of supercritical states as @xmath79 .
furthermore it is a threshold : a potential @xmath33 is subcritical if @xmath80 .
it is not difficult to show for a given @xmath6 that the number of supercritical positrons is given by @xmath81 \label{super2}\ ] ] the corresponding value of the total positron charge @xmath59 can be shown u@xmath82 g eqs ( [ even],[odd ] ) to satisfy @xmath83\leq q_p \label{charge}\ ] ] so for large @xmath31 we have the estimates @xmath84 @xmath85 now we can build up an overall picture of the wide square well @xmath86 . when @xmath33 is turned on from zero in the vacuum state an enormous number of bound states is produced .
as @xmath33 crosses @xmath87 a very large number @xmath88 of these states cross @xmath89 and become bound positrons . as @xmath33 crosses @xmath90
a large number @xmath70 of bound states become supercritical together .
this therefore gives rise to a positively charged current flowing from the well .
but in this case , unlike that of the klein step , the charge in the well is finite and therefore the particle emission process has a finite lifetime .
nevertheless , for @xmath91 large enough the transient positron current for a wide barrier is approximately constant in time for a considerable time as we shall see in the next section .
we now restrict ourselves to the case @xmath92 with @xmath93 .
this is not necessary but it avoids having to calculate the dynamics of positron emission while the potential is still increasing beyond the critical value .
we can assume all the positrons are produced almost instantaneously as the potential passes through @xmath94 it also means that the kinematics are non - relativistic .
hence for a sufficiently wide
well so that @xmath95 is large , @xmath96 .
the well momentum of the nth supercritical positron is still given by eq ( [ quant ] ) @xmath97 which corresponds to an emitted positron energy @xmath98 .
note that the emitted energies have discrete values although for @xmath31 large , they are closely spaced .
the lifetime @xmath99 of the supercritical well is given by the time for the slowest positron to get out of the well .
the slowest positron is the deepest lying state with @xmath100 and momentum @xmath101 .
hence @xmath102 so the lifetime is finite but scales as @xmath103 . but
a large number of positrons will have escaped well before @xmath99 .
there are @xmath70 supercritical positrons initially and their average momentum @xmath104 corresponds to @xmath105 ; hence @xmath106 which is independent of @xmath31 .
thus a transient current of positrons is produced which is effectively constant in time for a long time of order @xmath107 @xmath108 .
we thus see that the square well ( or barrier ) for @xmath31 sufficiently large behaves just like the klein step : it emits a seemingly constant current with a seemingly continuous energy spectrum .
but initially the current must build up from zero and eventually must return to zero .
so the well / barrier is a time - dependent physical entity with a finite but long lifetime for emission of supercritical positrons or electrons . in the appendix
we show that if we assume that the transient current emitted is constant in time ( which it is not ) , then it is possible to obtain a relationship between this current and the transmission coefficient just as for the klein step .
note again that the transmission resonances of the time - independent scattering problem coincide with the energies of particles emitted by the well or barrier .
it is therefore tempting to use the pauli principle to explain the connection . following hansen and ravndal
, we could say that @xmath41 must be zero at the resonance energy because the electron state is already filled by the emitted electron with that energy .
but it is easy to show that the reflection coefficient is zero for bosons as well as fermions of that energy , and no pauli principle can work in that case .
furthermore emission ceases after time @xmath99 whereas @xmath109 for times @xmath110 .
it follows that we must conclude that klein tunnelling is a physical phenomenon in its own right , independent of any emission process . in our next section
we consider further ways to investigate this tunnelling theoretically and experimentally .
it is clear from eq ( [ barr ] ) that while the reflection coefficient @xmath41 can not be @xmath111 , neither is the transmission coefficient @xmath42 exponentially small for energies @xmath3 , even though the scattering is classically forbidden .
the simplest way to understand this is to consider the negative energy states under the potential barrier as corresponding to physical particles which can carry energy in exactly the same way that positrons are described by negative energy states which can carry energy .
it follows from eq ( [ rs / ts ] ) that @xmath15 and @xmath21 correspond to reflection and transmission coefficients in transparent media with differing refractive indices : thus @xmath8 is nothing more than an effective fermionic refractive index corresponding to the differing velocities of propagation by particles in the presence and absence of the potential . on this basis
, tuning the momentum @xmath13 to obtain a transmission resonance for scattering off a square barrier is nothing more than finding the frequency for which a given slab of refractive material is tranparent .
this is not a new idea . in jensen s words
a potential hill of sufficient height acts as a fabry - perot etalon for electrons , being completely transparent for some wavelengths , partly or completely reflecting for others @xcite .
we can now look in more detail at klein tunnelling : both in terms of our model square well / barrier problem and at the analogous coulomb problem .
the interesting region is where the potential is strong but subcritical so that emission dynamics play no role and sensible time independent scattering parameters can be defined . for electrons scattering off the square barrier
@xmath112 we would thus require @xmath113 @xmath114 together with @xmath6 so that positrons can propagate under the barrier .
for the corresponding square well @xmath115 there are negative energy bound states @xmath116 provided that @xmath117 @xmath118 [ cf .
eq.([charge ] ) ] .
so when the potential well is deep enough , it will in fact bind positrons .
correspondingly , a high barrier will bind electrons .
it is thus not surprising that electrons can tunnel through the barrier for strong subcritical potentials since they are attracted by those potential barriers .
another way of seeing this phenomenon is by using the concept of effective potential @xmath119 which is the potential which can be used in a schrodinger equation to simulate the properties of a relativistic wave equation . for a potential @xmath25 introduced as the time - component of a four - vector into a relativistic wave equation ( klein - gordon or dirac ) ,
it is easy to see that @xmath120 hence as the energy @xmath1 changes sign , the effective potential can change from repulsive to attractive . for the coulomb barrier , anchishkin @xcite has already suggested looking at scattering of @xmath121 off heavy nuclei to see if there was any experimental evidence for tunnelling : the klein - gordon equation like the dirac equation has negative energy solutions , so similar arguments apply .
his calculations show a 20% enhancement in @xmath122 scattering compared with non - relativistic expectations
. the analogous process for fermions would be positron - heavy nucleus scattering . for a positron of initial energy @xmath1 incident on a heavy nucleus of charge @xmath123
the classical turning point @xmath124 so it would be interesting to measure the wave function at the origin for positron scattering off a heavy nucleus compared with the wave function at the origin for electron scattering off the nucleus at the same energy in order to demonstrate tunnelling . provided that @xmath125 and normal coulomb wave functions should be accurate enough for the calculation of the ratio @xmath126 of wave functions at the origin .
the wave function at the origin for a dirac particle is singular but the ratio is finite .
the exact continuum wave functions for a dirac particle in a coulomb potential are discussed by rose@xcite .
we shall just write down his results : if the particles are non - relativistic then @xmath127 where p and @xmath1 are the particle momenta and energies and this is ofcourse exponentially small as @xmath128 .
but if the particles are relativistic @xmath129 where @xmath130 is a ratio of complex gamma - functions and is approximately unity for large @xmath0 .
so @xmath131 for @xmath132which is not specially small although it still decreases exponentially with @xmath133 it should be possible to measure @xmath134 by , for example , internal electron - positron pair production in positron - nucleus scattering versus pair production in electron - nucleus scattering .
alternatively if the positrons were longitudinally polarised any resultant asymmetry in the scattering cross section would be due to weak processes .
these could only occur at a distance @xmath135 which is effectively at the origin for atomic systems .
so the asymmetry for polarised positron scattering could be compared with that for polarised electron scattering .
the ratios of the asymmetries would give @xmath134 directly .
but while experiments for @xmath136 should show tunnelling it will not yet amount to klein tunnelling .
for that we require @xmath0 large enough so that bound positron states are present .
this means that @xmath0 must be below its supercritical value @xmath137 of around @xmath138 but large enough for the @xmath139 state to have @xmath140 .
the calculations of references @xcite and @xcite which depend on particular models of the nuclear charge distribution give this region as @xmath141 which unfortunately will be difficult to demonstrate experimentally
. nevertheless , the theory seems to be clear : in this subcritical region positrons should no longer obey a tunnelling relation which decreases exponentially with z such as that of eq .
( [ rel ] ) .
instead the coulomb barrier should become more transparent as @xmath123 increases , at least for some energies .
the work of jensen and his colleagues @xcite shows that a transmission resonance ( i.e. maximal transmission ) for positron scattering on a coulomb potential may well occur at @xmath142 although the onset of supercriticality implies that time independent scattering quantities will then no longer be well - defined .
further detailed calculations are needed to clarify the situation for positron scattering off nuclei with @xmath0 just below @xmath137 .
it is also important to see if the square well / barrier relationship we illustrate above that transmission resonances at barriers occur when the corresponding well becomes supercritical [ cf .
eqs ( [ quant ] ) and ( [ even ] ] can be generalised to arbitrary potentials .
we hope that this discussion has demonstrated that the klein step is pathological and therefore a misleading guide to the underlying physics .
it represents a limit in which time - dependent emission processes become time - independent and therefore a relationship between the emitted current and the transmission coefficient exists , as we show in the appendix . in general no such relationship exists .
the underlying physics discovered by klein in his solution of the dirac equation is not particle emission but tunnelling by means of the negative energy states which are characteristic of relativistic wave equations , similar to interband tunnelling in semiconductors @xcite .
it is time to finally bring this 70 year old puzzle to a conclusion .
we would like to thank a. anselm , g barton , j d bjorken , l b okun , r laughlin , g e volovik and d waxman for advice and help .
consider the potential step @xmath143 for @xmath6 of section i. we will show that the expectation value of the current in the vacuum state in the presence of the step is non - zero . the derivation hinges on a careful definition of the vacuum state .
scattering is usually described by a solution describing a wave incident ( say from the left ) plus a reflected wave ( from the right ) plus a transmitted wave ( to the right ) . it is convenient here to use waves of different form either describing a wave ( subscript @xmath145 incident from the left with no reflected wave or describing a wave ( subscript @xmath41 ) incident from the right with no reflected wave .
positive and negative energy wavefunctions will be denoted by @xmath146 and @xmath147 respectively .
it is clear that the nontrivial result we are seeking arises from the overlap of the negative energy continuum @xmath148 on the right with the positive energy continuum @xmath149 on the left .
we are thus concerned with wavefunctions with energies in the range @xmath150 the expressions for @xmath151 in this energy range are given below .
[ [ the - definition - of - the - vacuum - and - the - vacuum - expectation - value - of - the - current . ] ] the definition of the vacuum and the vacuum expectation value of the current . ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ with @xmath162 given by the hermitean conjugate expansion.we must now determine the appropriate vacuum state in the presence of the step .
states described by wavefunctions @xmath163 and @xmath164 correspond to ( positive energy ) electrons and positrons respectively coming from the left . hence with respect to an observer to the left ( of the step ) such states should be absent from the vacuum state , so the wavefunctions that play the crucial role in the klein problem belong to the set @xmath166 for @xmath150 for an observer to the right these states are positive energy positrons and hence they should be filled in the vacuum state , i.e. substituting ( [ exp ] ) in ( [ vev ] ) and noticing that all terms involving @xmath173 and @xmath174 can be dropped since the corresponding energies lie outside the interesting range @xmath175 we end up with @xmath176 @xmath177 if we use the anticommutation relations and ( [ a1 ] ) .
the third term yields @xmath178 using ( [ vac ] ) and the fourth term vanishes using the anticommutation relations ( i.e. the exclusion principle ; the state @xmath179 already contains an electron in the state @xmath180 hence we get zero when we operate on it with @xmath181 ) .
one energy integration is performed immediately using the @xmath182 function .
the final result is where the energy integration is over the klein range .
this is the result first obtained by hansen and ravndal @xcite linking the pair production current with the transmission coefficient .
but the fact that these quantities are linked does not represent a causal relationship between them .
this is clear if we attempt to repeat this analysis for a square barrier for @xmath6 and @xmath27 .
it is easily shown by the same method that provided that averages over phase angles are understood , where now the current is measured by an observer situated at a point @xmath185 the result of eq .
( [ cur ] ) of course can only be true in the approximation where the transient current is taken to be independent of time .
note also that whereas for a klein step we describe the particle emission in terms of electron - positron pair production while we say that a supercritical well will spontaneously emit positrons .
these terms are relative : an observer outside a supercritical well would see a positron current going to the right while an observer measuring the same phenomenon but inside the well would talk of an electron current going to the left . a a grib , s g mamayev and v m mostepanenko , vacuum effects in strong fields , friedmann , st petersburg , 1994 ; " w greiner , b muller and j rafelski ,
quantum electrodynamics of strong fields , springer - verlag , berlin , 1985 . | the klein paradox is reassessed by considering the properties of a finite square well or barrier in the dirac equation .
it is shown that spontaneous positron emission occurs for a well if the potential is strong enough . the vacuum charge and lifetime of the well
are estimated .
if the well is wide enough , a seemingly constant current is emitted .
these phenomena are transient whereas the tunnelling first calculated by klein is time - independent .
klein tunnelling is a property of relativistic wave equations , not necessarily connected to particle emission .
the coulomb potential is investigated in this context : it is shown that a heavy nucleus of sufficiently large @xmath0 will bind positrons .
correspondingly , it is expected that as @xmath0 increases the coulomb barrier will become increasingly transparent to positrons .
this is an example of klein tunnelling . |
multi - component kaup - boussinesq(kb ) equations can be obtained from the lax operator @xmath3 where @xmath4 , @xmath5 are the multi - kb fields @xcite-@xcite . here
@xmath6 is a positive integer .
+ the multi system of kb equation is given as @xmath7 where @xmath8 and @xmath9 .
this system in ( [ eqn1 ] ) was shown to be also a degenerate kdv system of rank one @xcite , @xcite .
this system admits also recursion operator for all values of @xmath10 . in this work
we shall investigate the traveling wave solutions of these coupled equations . for this purpose
we start with the case @xmath11 . to find such solutions we use time and space translation symmetries of the coupled system .
the kb equation for @xmath11 is @xmath12 in @xcite the inverse problem of the above system was studied and soliton solutions which decay asymptotically were found .
the @xmath13 solution found in that work corresponds to the interaction of two solitary waves .
it was also mentioned in @xcite that there is no solution in the form of traveling wave .
here in this work we prove that there exists no asymptotically vanishing traveling wave solutions of system of equations for @xmath11 .
this is consistent with the observation of @xcite .
we show that this is also valid for @xmath2 .
we claim it to be true for all even positive integers .
we show that it is possible to find solitary wave solutions of ( [ l=2 ] ) which asymptotically decay to non - zero constants . furthermore in addition to the solitary wave solutions of ( [ l=2 ] ) we find all traveling wave solutions which are expressible in terms of jacobi elliptic functions . traveling wave solutions of a system of equations can be obtained if the equations possess time and space translation symmetries .
such symmetries exist in our case . hence letting @xmath14 where @xmath15 is a constant(the speed of the wave ) and @xmath16 , and @xmath17 from the first equation of ( [ l=2 ] ) we have @xmath18 which gives @xmath19 where @xmath20 is an integration constant . using @xmath21 in the second equation of ( [ l=2 ] ) yields @xmath22 integrating above equation
once we obtain @xmath23 by using @xmath24 as an integrating factor , we can integrate once more .
finally we get @xmath25 where @xmath26 are constants .
these constants can be determined from the initial conditions @xmath27 , @xmath28 , @xmath29 and @xmath30 . if @xmath31 has zeros , these zeros are related to these initial conditions . for
asymptotically decaying solutions of @xmath0 kb equations @xmath32 , @xmath24 , @xmath33 and @xmath34 go to zero as @xmath35 . here in this work
we shall find all possible solutions @xmath32 of ( [ f(f ) ] ) .
given a solution @xmath32 one can find the corresponding solution @xmath21 from ( [ functiong ] ) .
in @xcite and @xcite , a kb like system @xmath36 was considered .
traveling wave solutions of this system satisfy a differential equation like ( [ f(f ) ] ) but the corresponding polynomial @xmath37 is asymptotically positive definite .
this means that the above kb like system possesses asymptotically decaying traveling wave solutions . in @xcite and @xcite some solitary wave solutions were found .
since the fourth degree polynomial arising in traveling wave solutions of the system ( [ kb2 ] ) is different than the one given in ( [ f(f ) ] ) .
then the behavior of solutions here in this work and in @xcite , @xcite are different . in @xcite a modified version of the system ( [ kb2 ] ) ,
i.e. @xmath38 was considered , where @xmath39 is a parameter which controls the dispersion effects .
the upper sign is for the case when the gravity force dominates over the capillary one , and the lower sign is for the opposite case when capillary dominates over the gravity .
the traveling wave solutions of the above system ( [ kb4 ] ) were considered in @xcite .
the equation ( [ f(f ) ] ) becomes now @xmath40 . in both cases solitary wave solutions ( dark and bright solitons )
were found in @xcite .
the lower case ( negative sign ) resembles to our case .
hence our solution in section @xmath41 can be considered as a dark soliton in the sense of @xcite .
this is the solution corresponding one double and two simple zeros of the polynomial @xmath31 .
we have all other solutions corresponding to different combinations of the zeros of @xmath31 in sections @xmath42 , @xmath43 , and @xmath44 .
the layout of our paper is as follows : in section @xmath45 , we study the behavior of the solutions in the neighborhood of the zeros of @xmath31 and discuss all possible cases .
we find all solitary wave solutions of the system ( [ l=2 ] ) in section @xmath42 .
these correspond to one double and two simple zeros of @xmath31 , and one triple and one simple zeros of @xmath31 . in section @xmath43
, we find all elliptic type of solutions starting from very special ones to the most general elliptic type of solutions .
these solutions are given in terms of the zeros of the function @xmath31 . in section @xmath44
, we discuss @xmath1 and @xmath2 cases . in section @xmath46 , we give the graphs of the solutions corresponding to all cases considered in the text .
there is no real asymptotically vanishing traveling wave solution of the equation ( [ l=2 ] ) in the form @xmath16 and @xmath17 , where @xmath47 .
* proof . *
if we apply the boundary conditions @xmath48 as @xmath35 which describe the solitary wave , we get @xmath49 . hence we end up with @xmath50 clearly , we do not have a real solution @xmath32 .
@xmath51 now we will deal with the equation ( [ f(f ) ] ) . in order to have real solutions
, @xmath52 must take values so that the following inequality holds : @xmath53 here we will analyze the zeros of @xmath31 . *
( i ) * if @xmath54 is a _ simple zero _ of @xmath31 we have @xmath55 .
taylor expansion of @xmath31 gives @xmath56 from here we get @xmath57 and @xmath58 .
hence we can write the function @xmath59 as @xmath60 thus , in the neighborhood of @xmath61 , the function @xmath59 has local minimum or maximum as @xmath62 is positive or negative respectively since @xmath58 . + * ( ii ) * if @xmath54 is a _
double zero _ of @xmath31 we have @xmath63 .
taylor expansion of @xmath31 gives @xmath64 to have real solution @xmath32 , we should have @xmath65 . from the equality ( [ doublezero ] ) we get @xmath66 which gives @xmath67 where @xmath68 is a constant . hence @xmath69 as @xmath70 .
the solution @xmath32 can have only one peak and the wave extends from @xmath71 to @xmath72 . *
( iii ) * if @xmath54 is a _ triple zero _ of @xmath31 we have @xmath73 . taylor expansion of @xmath31 gives @xmath74 this is valid only if both signs of @xmath75 and @xmath76 are same i.e. we have the following two possibilities to have real solution @xmath32 : + @xmath77 @xmath78 and @xmath79 + @xmath80 @xmath81 and @xmath82 + let us analyze these cases . if @xmath78 and @xmath83 then we have @xmath84 which gives @xmath85 where @xmath86 is a constant .
thus @xmath69 as @xmath35 if @xmath83 .
let @xmath81 and @xmath87 hold . in this case , @xmath88 and @xmath89 , @xmath90 .
then @xmath91 which yields @xmath92 where @xmath93 is a constant .
thus @xmath69 as @xmath35 if @xmath94 .
* ( iv ) * if @xmath54 is a _
quadruple zero _ of @xmath31 then there is only one possibility @xmath95 . it is clear that this case does not give a real solution except when @xmath96 . here
we present the sketches of the graphs of @xmath31 .
real solutions @xmath97 occur in the shaded regions .
now we analyze all possible cases about the zeros of @xmath31 and above graphs . *
( 1 ) no real zero .
* if there is no real zeros of @xmath31 then @xmath98 .
hence there is no real solution of ( [ f(f ) ] ) in that case . *
( 2 ) two simple real zeros . *
if there is a simple zero @xmath99 of @xmath31 , since the order of @xmath31 is four , there should be another simple zero @xmath100 of @xmath31 .
the corresponding graph to this case is given in @xmath101 . here
, the real solution occurs when @xmath32 is between two simple zeros @xmath99 and @xmath100 . at @xmath99 , @xmath102
so graph of the function @xmath32 is concave up at @xmath103 . at @xmath100 , @xmath104
hence graph of the function @xmath32 is concave up at @xmath105 .
thus it is clear that the solution is periodic . *
( 3 ) one double zero .
* if there is only one double zero @xmath99 then @xmath106 where @xmath107 has no real zero .
this means @xmath108 which yields that @xmath109 .
then @xmath110 hence there is no real solution in that case except when @xmath96 .
similarly , in the case when @xmath31 has two double zeros @xmath99 and @xmath100 , no real solutions exist since @xmath111 except when @xmath96 or @xmath112 . *
( 4 ) one double and two simple zeros . *
the corresponding graphs for this case are @xmath113 , @xmath114 and @xmath115 . in @xmath113 and @xmath114 , there are two simple zeros @xmath99 and @xmath116 and one double zero @xmath100 .
we have @xmath117 in @xmath113 and in the graph @xmath114 , @xmath118 . in both cases ,
the real solution occurs when @xmath32 is between two simple zeros @xmath99 and @xmath116 . at @xmath99 , @xmath102
so graph of the function @xmath32 is concave up at @xmath103 . at @xmath116 , @xmath119
hence graph of the function @xmath32 is concave up at @xmath120 .
it is clear that the solution is periodic in this case . in @xmath115 , different than the graphs @xmath113 and @xmath114 we have @xmath121 .
the real solution occurs when @xmath32 stays between @xmath99 and @xmath100 or @xmath100 and @xmath116 . at @xmath99 , @xmath102
hence graph of the function @xmath32 is concave up at @xmath103 . at double zero @xmath100 ,
@xmath122 as @xmath123 . hence we have a solitary wave solution with amplitude @xmath124 .
similarly at @xmath116 , @xmath125 , hence graph of the function @xmath32 is concave down at @xmath120 .
therefore , we also have a solitary wave solution with amplitude @xmath126 .
explicit solitary wave solution for this case can be found in the next section . * ( 5 ) one triple and one simple zero .
* for this case , we can analyze the graphs @xmath127 and @xmath128 . in @xmath127 ,
@xmath99 is simple and @xmath100 is triple zeros of @xmath31 .
we see that @xmath129 hence graph of the function @xmath32 is concave up at @xmath103 . from the case ( iii ) , we know that @xmath122 as @xmath123 for @xmath130 and @xmath131 .
hence we have solitary wave solution with amplitude @xmath132 .
similarly , in @xmath128 we have one triple zero @xmath99 and one simple zero @xmath100 .
for triple zero @xmath99 we have @xmath69 as @xmath123 for @xmath133 and @xmath134 .
for simple zero we have @xmath135 therefore graph of the function @xmath32 is concave down at @xmath105 . clearly , we have a solitary wave solution with amplitude @xmath136 .
explicit solitary wave solution for this case can be found in the next section . * ( 6 ) four different simple zeros .
* the corresponding graph for this case is given in @xmath137 .
here , there are four simple zeros @xmath138 . for @xmath99 and @xmath116
, we have @xmath129 and @xmath139 thus graph of the function @xmath32 is concave up at @xmath103 and @xmath120 . for @xmath100 and @xmath140 , we have @xmath141 and @xmath142 so graph of the function @xmath32 is concave down at @xmath105 and @xmath143 . obviously , the solution is periodic . as a summary we have the following results . by solution below , we mean non - constant solutions .
[ pro22 ] equation ( [ f(f ) ] ) has no real solutions when the function f(f ) has one the following properties : ( i ) it has no real zeros , ( ii ) it has only two real zeros , ( iii ) it has only one double zero , ( iv ) it has only two double zeros and ( v ) it has a quartic zero
. equation ( [ f(f ) ] ) admits solitary wave solutions when the function f(f ) admits ( i ) one double and two simple zeros and ( ii ) one triple and one simple zeros . from the proposition [ pro22 ]
we can conclude that the function @xmath31 must have four zeros , @xmath144 the constants @xmath26 can be expressed in terms of the zeros of @xmath31 : @xmath145 in the next section we shall find the solitary wave solutions mentioned in the above proposition which correspond to special cases of the zeros @xmath146 .
let @xmath99 and @xmath116 be simple zeros and @xmath100 be a double zero of @xmath31 .
thus we have @xmath147 let @xmath148 and so @xmath149 , where @xmath150 and @xmath151 , where @xmath152 .
hence the above equation becomes @xmath153 using the substitution @xmath154 @xmath155 after some arrangements we have @xmath156 ^ 2-\frac{1}{4}\big(\frac{1}{u_1}-\frac{1}{u_3}\big)^2\big\}.\ ] ] using the trigonometric substitution @xmath157 the equation ( [ doublesimplesimple ] ) becomes @xmath158 note that in the case when @xmath31 has two simple zeros and one double zero , the solitary wave solution occurs only when we have @xmath121 and this makes @xmath159 or @xmath160 .
so from the above equation we get @xmath161 which yields @xmath162 where @xmath163 is an integration constant .
hence the solution @xmath32 is @xmath164 where @xmath165 and @xmath166 .
it is clear that @xmath122 as @xmath35 .
note that when @xmath167 which means @xmath118 or @xmath117 we have the following solution which is not a solitary wave solution : @xmath168 with the same @xmath169 and @xmath170 stated above .
let @xmath99 be simple and @xmath100 be triple zeros of @xmath31 .
hence @xmath171 the relations between the zeros of @xmath31 and the parameters are @xmath172 let us solve the equation ( [ triplesimple ] ) .
let @xmath173 so above equation becomes @xmath174 we have @xmath175 by making the substitution @xmath176 , the above equality can be solved as @xmath177 where @xmath163 is an integration constant .
hence we find @xmath178 and inserting @xmath179 and @xmath180 we get the solution @xmath181 it is clear that @xmath69 as @xmath182 . here
we will analyze the solution ( [ onedoubletwosimple ] ) which corresponds to the case when @xmath31 has one double zero @xmath100 and two simple zeros @xmath99 and @xmath116 . *
( a ) * when @xmath183 , the solution ( [ onedoubletwosimple ] ) reduces to @xmath184 * ( b ) * when @xmath185 , the solution ( [ onedoubletwosimple ] ) reduces to @xmath186[limitingb ] which can be converted to @xmath187 where @xmath165 and @xmath166 . *
( c ) * when @xmath188 , then the solution ( [ onedoubletwosimple ] ) reduces to @xmath189 which can also be written as @xmath190 * ( d ) * when @xmath191 or @xmath192 , then the case turns to the case when @xmath31 has one triple zero and one simple zero . if @xmath191 , the solution ( [ onedoubletwosimple ] ) reduces to @xmath193 and if @xmath192 , the solution ( [ onedoubletwosimple ] ) reduces to @xmath194
in this section we will find exact solutions of ( [ l=2 ] ) by using the jacobi elliptic functions .
let us give the list of the jacobi elliptic functions and first order differential equations satisfied by them .
@xmath195,\label{eq : jacobitnu}\\ & & \quad \displaystyle y=\frac{1}{\mathrm{sn}v } \quad ( y')^2=(y^2 - 1)(y^2-k^2),\label{eq : jacobioneoversnu}\\ & & \quad \displaystyle y=\frac{1}{\mathrm{cn}v } \quad ( y')^2=(y^2 - 1)[(1-k^2)y^2+k^2],\label{eq : jacobioneovercnu}\\ & & \quad y=\mathrm{dn}v\mathrm{tn}v \quad ( y')^2=(1+y^2)^2 - 4k^2y^2,\label{eq : jacobidnutnu}\end{aligned}\ ] ] and for the squares of these functions we have cubic equations @xmath196,\label{eq
: jacobisquaretn}\\ & & \quad y=\frac{1}{\mathrm{cn}^2v } \quad ( y')^2=4y(y-1)[(1-k^2)y+k^2],\label{eq : jacobisquareoneovercn}\\ & & \quad \displaystyle y=\frac{1}{\mathrm{sn}^2v } \quad ( y')^2=4y(y-1)[y - k^2],\label{eq : jacobisquareoneoversn}\\ & & \quad y=\mathrm{dn}^2v\mathrm{tn}^2v \quad ( y')^2=4y[(1+y)^2 - 4k^2y].\label{eq : jacobisquaredntn}\end{aligned}\ ] ] we will also make analysis at the limiting points @xmath197 and @xmath198 .
remind that @xmath199 for some special values of @xmath200 , we have solutions of ( [ l=2 ] ) in terms of jacobi elliptic functions . here
we will present two such types of solutions .
* case 1.solutions of the form @xmath201 * here we shall find the solutions of ( [ l=2 ] ) having the form @xmath201 , where @xmath202 are constants , @xmath47 and @xmath203 is one of the jacobi elliptic functions . when we use this form in ( [ f(f ) ] ) we get the following equation : @xmath204 since the parameters are real , we have @xmath205 .
hence the coefficient of the term @xmath206 is negative .
thus there are two possibilities : @xmath207 which corresponds to jacobi elliptic function @xmath208 and @xmath209 corresponding to @xmath210 . comparing the differential equations for @xmath208 and @xmath210 with ( [ exactsolncaseone ] )
, we note that the coefficients of the terms @xmath211 and @xmath203 should be zero . that gives @xmath212,\end{aligned}\ ] ] where @xmath20 is given in ( [ parameters ] ) .
note that the equality @xmath213 yields a relation between the zeros of @xmath31 : @xmath214 the equation is simplified as @xmath215 where @xmath216 * 1.a * * cn solution * + let @xmath217 with @xmath47 where the function @xmath203 satisfies the first order differential equation .
hence when we compare the coefficients of ( [ exactsolncaseone ] ) and , we get @xmath218 note that there is also one more condition that should be satisfied given in ( [ relationzeros ] ) .
assume that @xmath219 or @xmath220 .
in addition , without loss of generality let @xmath221 .
then we have @xmath222 hence the solution is @xmath223+\frac{f_1+f_3}{2}.\ ] ] let us check the limiting points . for @xmath197 , we have @xmath224 and the solution becomes @xmath225 . for @xmath198
, we get the relation @xmath226 hence the solution is @xmath227+\frac{f_1+f_3}{2}.\ ] ] * 1.bdn solution * + let @xmath228 with @xmath47 where the function satisfies the differential equation .
if we compare the coefficients of ( [ exactsolncaseone ] ) and , we get @xmath229 if we assume that @xmath220 and @xmath221 then we have @xmath230 hence the solution is @xmath231+\frac{f_1+f_3}{2}.\ ] ] for @xmath197 , we have either @xmath232 or @xmath233 . for both cases , depending on the sign of @xmath234 , we have constant solutions : @xmath225 or @xmath235 . for @xmath198 , from ( [ parameterscase1b ] )
we get the relation @xmath236 .
thus the corresponding solution is @xmath237+\frac{f_1+f_3}{2}.\ ] ] * case 2.solutions of the form @xmath238 * + here we shall find solutions of ( [ l=2 ] ) having the form @xmath238 , where @xmath239 are constants and @xmath47 .
if we use this form in the equation ( [ f(f ) ] ) we get the following equation : @xmath240 where @xmath241 . as we did in the previous case we shall again use jacobi elliptic functions ( [ eq : jacobisnu])-([eq : jacobioneoversnu ] ) and study the special cases for @xmath197 and @xmath198 .
the differential equations satisfied by these elliptic functions do not have terms with @xmath211 and @xmath203 .
hence the coefficients of @xmath211 and @xmath203 should be zero in ( [ exactsolncasetwo ] ) .
let also @xmath242 and @xmath243 , @xmath244 .
then we get @xmath245\\ d_2&=&-4d_3a=\frac{1}{2}af_1f_2f_3f_4\\ a&=&\frac{f_1f_2f_4+f_1f_2f_3+f_2f_3f_4+f_1f_3f_4}{4f_1f_2f_3f_4}\end{aligned}\ ] ] with a relation between the zeros of @xmath31 : @xmath246 hence is simplified as @xmath247 where @xmath248 now let us study the elliptic functions satisfying ( [ simplifiedexactsolncasetwo ] ) .
* 2.asn solution * + let @xmath249 with @xmath47 where the function @xmath203 satisfies the first order differential equation . then when we compare the coefficients of ( [ simplifiedexactsolncasetwo ] ) and , we get @xmath250 by the relation ( [ case2relationonzeros ] )
let us take @xmath251 .
hence ( [ parameterscase2a ] ) becomes @xmath252 and we have four choices for the value @xmath253 : @xmath254 , @xmath255 .
taking @xmath256 yields @xmath257 hence the solution is @xmath258},\ ] ] where @xmath259 let us study the limiting cases .
for @xmath197 , there are two possibilities : @xmath188 or @xmath224 . if @xmath188 then @xmath260 , @xmath261 and the corresponding solution is @xmath262}.\ ] ] if @xmath224 then @xmath263 and @xmath264 so we have constant solution @xmath225 . for @xmath198 then from ( [ case2aparameterswithzeros ] ) we have @xmath265 it is not possible to have @xmath266 or @xmath267 because of the definition of @xmath253 .
if @xmath232 or @xmath233 we have @xmath268 .
hence we do not have real solution for @xmath198 .
* 2.bcn solution * + let @xmath217 with @xmath47 where the function @xmath203 satisfies the first order differential equation .
if we compare the coefficients of ( [ simplifiedexactsolncasetwo ] ) and , we get @xmath269 by ( [ case2relationonzeros ] ) let us take @xmath251 . since we have the same relation for @xmath253 as in the case @xmath270 , we may also take @xmath256 .
hence ( [ parameterscase2b ] ) becomes @xmath271 thus the solution is @xmath272},\ ] ] where @xmath259 for @xmath197 , there are two possibilities : @xmath188 or @xmath224 . if @xmath188 then @xmath260 , @xmath273 and the corresponding solution is @xmath274}.\ ] ] if @xmath224 then @xmath263 and @xmath264 so we have a constant solution @xmath225 . for @xmath198
, we have the following relation from ( [ case2bcnsystem ] ) : @xmath275 hence the solution is @xmath276},\ ] ] where @xmath277 .
* 2.cdn solution * + let @xmath228 with @xmath47 where the function @xmath203 satisfies the first order differential equation .
when we compare the coefficients of ( [ simplifiedexactsolncasetwo ] ) and , we get @xmath278 same as before let us take @xmath251 and @xmath256 . hence ( [ parameterscase2c ] ) becomes @xmath279 thus the solution is @xmath280},\ ] ] where @xmath281 for @xmath197 , there are four possibilities : @xmath266 , @xmath267 , @xmath232 or @xmath233 .
we can not have @xmath266 or @xmath267 because of the definition of @xmath253 . if @xmath232 or @xmath233 , the solution is @xmath235 . for @xmath198
, we have @xmath185 .
so the corresponding solution is @xmath282 } , \quad f_1f_3>0,\ ] ] where @xmath277 .
* 2.dtn solution * + let @xmath283 with @xmath47 where the function @xmath203 satisfies the first order differential equation . hence
when we compare the coefficients of ( [ simplifiedexactsolncasetwo ] ) and , we get @xmath284 let us take @xmath251 . here
we notice that third equality of ( [ parameterscase2d ] ) reveals that @xmath253 is not real for any values of @xmath285 .
hence for all values of @xmath286 $ ] we do not have real solution .
* 2.e1/sn solution * + let @xmath287 with @xmath47 where the function @xmath203 satisfies the first order differential equation . hence when we compare the coefficients of ( [ simplifiedexactsolncasetwo ] ) and , we get @xmath288 if we take @xmath251 and @xmath256 , ( [ parameterscase2e ] ) becomes @xmath289 the corresponding solution is @xmath290}{(f_1+f_3)\mathrm{sn}[\beta(x - ct)]+(f_1-f_3)},\ ] ] where @xmath259 for @xmath197 , we have the relation @xmath291 and the solution becomes @xmath292}{(f_1+f_3)\sin\big[\frac{f_1-f_3}{f_1+f_3}\sqrt{-f_1f_3}(x - ct)\big]\pm(f_1-f_3)},\quad -f_1f_3>0,\ ] ] where @xmath277 . for @xmath198
then from ( [ case2eparameterswithzeros ] ) we have @xmath265 it is not possible to have @xmath266 or @xmath267 because of the definition of @xmath253 .
if @xmath232 or @xmath233 we have @xmath293 .
hence we do not have real solution for @xmath198 .
* 2.f1/cn solution * + let @xmath294 with @xmath47 where the function @xmath203 satisfies the first order differential equation . if we compare the coefficients of ( [ simplifiedexactsolncasetwo ] ) and , we get @xmath295 since we take @xmath251 and @xmath256 , ( [ parameterscase2f ] ) becomes @xmath296 ^ 2}{4f_1f_3(f_1-f_2)(f_2-f_3)}.\ ] ] the corresponding solution is @xmath297}{(f_1+f_3)\mathrm{cn}[\beta(x - ct)]+(f_1-f_3)},\ ] ] where @xmath259 for @xmath197
, we have the relation @xmath291 and the solution becomes @xmath298}{(f_1+f_3)\cos\big[\frac{f_1-f_3}{f_1+f_3}\sqrt{-f_1f_3}(x - ct)\big]+(f_1-f_3)},\quad -f_1f_3>0,\ ] ] where @xmath277 . the case for @xmath198
gives the condition @xmath299 to be satisfied .
hence we have two possibilities : @xmath188 or @xmath224 . if @xmath188 then @xmath300 , @xmath301 and the corresponding solution is @xmath302}{(f_1+f_3)\mathrm{sech}[\sqrt{-f_1f_3}(x+\frac{f_1+f_3}{4}t)]+(f_1-f_3)},\quad -f_1f_3>0.\ ] ] if @xmath224 then @xmath263 and @xmath264 so we have a constant solution @xmath225 .
* 2.g * * dntn solution * + let @xmath303 with @xmath47 where the function @xmath203 satisfies the first order differential equation .
hence when we compare the coefficients of ( [ simplifiedexactsolncasetwo ] ) and , we get @xmath304 let us take @xmath251 .
the third equality above gives four choices for @xmath253 : @xmath305 to have real solutions , the parameters must be real .
hence from the expressions for @xmath253 we have either @xmath306 or @xmath307 .
if the first one is true then @xmath308 if the second one is true then @xmath309 from the equality for @xmath310 in ( [ firstgroup2 g ] ) we get @xmath311 this gives that @xmath312 .
we know that for the parameter @xmath310 of jacobi elliptic functions we have @xmath313 . additionally , at the limiting points @xmath197 and @xmath198 it yields that @xmath31 has two double zeros that is the case which does not give real solution as we stated in section @xmath314 .
we also have the similar result for ( [ secondgroup2 g ] ) .
hence we do not have real solutions for all @xmath286 $ ] .
when @xmath31 has one double @xmath100 and two simple zeros @xmath99 and @xmath116 we have the following system of equations : @xmath315\nonumber \\
8d_2&=&2f_1f_2f_3+f_2 ^ 2(f_1+f_3)\nonumber \\
8d_3&=&-f_1f_2 ^ 2f_3.\end{aligned}\ ] ] the exact solutions in terms of the jacobi elliptic functions take the following forms . *
( i ) * in case @xmath316 and case @xmath317 , we have @xmath213 . using this in ( [ systemofeqn ] )
we obtain that either @xmath236 or @xmath224 .
the second one is not allowed due to the discussion in the section @xmath318 . by using ( [ systemofeqn ] )
, the first one leads to @xmath319 . in this case
the solution is given in ( [ case1asoln ] ) and ( [ case1bsoln ] ) which are compatible with the limiting solutions discussed in section @xmath44 , part @xmath320 . *
( ii ) * in case @xmath321 and case @xmath322 , we have @xmath323 . from the first equation of ( [ systemofeqn ] ) we have @xmath324
. then this implies @xmath325 .
this constraint gives @xmath185 which yields @xmath319 . in this case the solutions are given in ( [ case2bsoln ] ) and ( [ case2csoln ] ) which are compatible with the limiting solutions discussed in section @xmath44 , part @xmath253 . *
( iii ) * if @xmath188 then @xmath326 hence @xmath327 which leads to @xmath319 . in this case
the solution is ( [ case2fsoln ] ) given in case @xmath328 which are compatible with the limiting solutions discussed in section @xmath44 , part @xmath15 .
here we shall deal with the most general form of solutions @xmath329 when we insert this form into the equation @xmath31 we get @xmath330 where @xmath331 we have four arbitrary constants @xmath26 in the differential equation ( [ f(f ) ] ) . in ( [ generalsoln ] )
we have effectively four independent parameters . by choosing these constants properly
we get several solutions in terms of elliptic functions .
we can analyze these solutions in two groups : * i ) * if @xmath31 has zeros then we can make the coefficients of @xmath206 to vanish by taking @xmath332 .
this means that @xmath333 is a zero of @xmath31 .
in addition to that choosing the constant @xmath334 yields that @xmath335 where @xmath336 .
this also means that @xmath336 is another zero of @xmath31 .
note that @xmath337 since @xmath338 . then the equation ( [ generalsoln ] ) takes the form where the square of elliptic functions and their inverses given in ( [ eq : jacobisquaresn])-([eq : jacobisquaredntn ] ) satisfy . by making substitution @xmath336 and @xmath333 ,
the equation ( [ generalsoln ] ) becomes @xmath339 where @xmath340 if @xmath320 and @xmath253 are the zeros of @xmath31 , then @xmath341 and we do not have the terms with @xmath206 and the constant term in ( [ generalsolncasei ] ) .
for instance , let @xmath342 and @xmath343 , then we can write @xmath344 such that @xmath345 are zeros of @xmath31 .
let us write @xmath346 , @xmath347 and @xmath348 in terms of the zeros of the function @xmath31 by the help of ( [ parameters ] ) .
@xmath349 now we give all solutions of ( [ l=2 ] ) of the form ( [ generalformofsoln ] ) .
let @xmath350 with @xmath47 where the function @xmath203 satisfies the first order differential equation ( [ eq : jacobisquaresn ] ) .
hence when we compare the coefficients of ( [ generalsolncasei ] ) and ( [ eq : jacobisquaresn ] ) , we get @xmath351 here without loosing any generality we take @xmath352 . * ( 1 ) * let * @xmath342 , @xmath343*. for this choice we have @xmath353 and hence the solution with the initial condition @xmath354 is @xmath355 since @xmath320 and @xmath253 are any zeros of @xmath31 we have other choices of these parameters . *
( 2 ) * let * @xmath356 , @xmath357*. for this choice we have @xmath358 and hence the solution with the initial condition @xmath359 is @xmath360 * ( 3 ) * let * @xmath361 , @xmath362*. for this choice we have @xmath363 and hence the solution with the initial condition @xmath364 is @xmath365 * ( 4 ) * let * @xmath366*. for this choice we have @xmath367 and hence the solution with the initial condition @xmath368 is @xmath369 similarly , we can also find other type of solutions including square of jacobi elliptic functions and inverses of them .
but they are equivalent because of the relations @xmath370 and @xmath371 .
hence the solutions given in @xmath372 , @xmath373 , @xmath374 and @xmath375 are all the most general solutions of ( [ l=2 ] ) depending upon the initial conditions .
* ii ) * another choice is taking @xmath376 so that @xmath377
. then the equation ( [ generalsoln ] ) takes the form where elliptic functions and their inverses given in ( [ eq : jacobisnu])-([eq : jacobioneovercnu ] ) satisfy .
note that if we take @xmath378 to make @xmath379 we have @xmath380 , @xmath213 and the solution becomes @xmath381 which we have already studied in section @xmath382 , case @xmath383 .
if @xmath384 we can use inverse of jacobi elliptic functions for @xmath203 and then the case turns to @xmath378 case . when we take @xmath385 , to make @xmath379 we have @xmath386 and @xmath387 and the solution becomes @xmath388 that is the case we have already studied in section @xmath382 , case @xmath45 . in the next section we mention about the system ( [ eqn1 ] )
when @xmath1 and @xmath2 .
* 1 ) * the degenerate coupled kdv equation for @xmath1 is @xmath389 here we will show that unlike the case @xmath11 , we have real traveling wave solution with asymptotically vanishing boundary condition in @xmath1 case .
let @xmath16 , @xmath17 , and @xmath390 , where @xmath47 . from the first equation of ( [ l=3 ] ) we have @xmath18 which gives @xmath391 where @xmath20 is an integration constant .
using @xmath21 in the second equation of yields @xmath392 integrating above equation once we get @xmath393 where @xmath394 is an integration constant . using @xmath395 in the third equation of yields @xmath396 integrating above equation
once we obtain @xmath397 by using @xmath24 as an integrating factor , we integrate once more . finally , we get @xmath398 where @xmath399 are constants . if we apply the boundary conditions @xmath400 as @xmath35 we get @xmath401 .
hence we have @xmath402 by using trigonometric substitution @xmath403 and making the cancelations , above equality becomes @xmath404 making the substitution @xmath405 gives @xmath406 which is solved as @xmath407 where @xmath408 is an integration constant .
note that @xmath409 . when the solution @xmath410 , @xmath411 is either @xmath383 or @xmath412 .
insert the expression for @xmath411 into the above equation so we get the relation defining the solution @xmath32 , @xmath413 hence we have asymptotically vanishing real traveling solution for @xmath1 .
we expect that this is true for all odd @xmath10 .
* 2 ) * now let us analyze @xmath2 case . the degenerate coupled kdv equation for @xmath2 is @xmath414 there is no real asymptotically vanishing traveling wave solution of the equation ( [ l=4 ] ) in the form @xmath16 , @xmath17 , @xmath390 and @xmath415 , where @xmath47 .
* let @xmath16 , @xmath17 , @xmath390 and @xmath415 , where @xmath47 . from the first equation of ( [ l=4 ] ) we have @xmath18 which gives @xmath391 where @xmath20 is an integration constant . using @xmath21 in the second equation of yields @xmath392 integrating above equation once we have @xmath393 where @xmath394 is an integration constant . using @xmath395 in the third equation of yields @xmath416 integrating this equation
once gives @xmath417 where @xmath418 is an integration constant .
using @xmath419 in the fourth equation of gives @xmath420 integrating the above equation once we get @xmath421 where @xmath422 is an integration constant . by using @xmath24 as an integrating factor
, we integrate once more . finally , we get @xmath423 where @xmath424 is an integration constant . if we apply the boundary conditions @xmath32 , @xmath24 , @xmath33 , @xmath425 , @xmath34 , @xmath426 , @xmath427 , @xmath428 , @xmath429 , @xmath430 @xmath431 @xmath432 as @xmath433 , we get @xmath434 .
hence the above equation becomes @xmath435 obviously , there is no real traveling wave solution of the case @xmath2 with asymptotically vanishing boundary conditions .
* conjecture : * for all even @xmath10 , since we have the following equality @xmath436 the degenerate coupled kdv equation ( [ eqn1 ] ) does not have real traveling wave solution with asymptotically vanishing boundary conditions .
here we give the graphs of exact solutions to see the behavior of the solutions . * case @xmath316 and case @xmath317 for @xmath198 : * + according to the conditions on parameters , they are chosen as @xmath437 hence the solution becomes @xmath438 and the graph of this function is -@xmath317 ( @xmath198 ) ] note that by the choice of the parameters of this case the equation ( [ f(f ) ] ) becomes @xmath439 the numerical values of the zeros of @xmath31 are such that the graph corresponds to the exact solitary wave solution given in section @xmath440 , part @xmath101 .
* case @xmath316 for different values of @xmath285 : * + here to see the behavior of the solution by the change of the value of @xmath285 we give the following graph : ( different values of @xmath285 ) ] * case @xmath317 for @xmath441 : * + the parameters are chosen as @xmath442 the solution is @xmath443 and the graph of this function is ( @xmath441 ) ] note that by the choice of the parameters of this case the equation ( [ f(f ) ] ) becomes @xmath444 since @xmath31 has four different simple zeros , we expect periodic solution as in the graph . * * case @xmath270 for @xmath197:**the parameters are @xmath445 hence the solution becomes @xmath446 and the graph of this function is ( @xmath197 ) ] note that by the choice of the parameters of this case the equation ( [ f(f ) ] ) becomes @xmath447 here the function @xmath31 has one double zero @xmath188 and two simple zeros @xmath448 and @xmath449 so @xmath118 . as it is stated in section @xmath314 , part @xmath450 we have periodic solution which can also be seen in the above graph . * * case @xmath321 for @xmath197:**the parameters are chosen as @xmath451 hence the solution becomes @xmath452 and the graph of this function is ( @xmath197 ) ] note that by the choice of the parameters of this case the equation ( [ f(f ) ] ) becomes @xmath453 the function @xmath31 has one double zero @xmath454 and two simple zeros @xmath455 and @xmath456 so @xmath117 . as it is given in section @xmath314 , part @xmath450 ,
the solution is periodic , which can be easily seen in the graph . *
* case @xmath321 and case @xmath322 for @xmath198:**the parameters are chosen as @xmath457 hence the solution becomes @xmath458 and the graph of this function is -@xmath322 ( @xmath198 ) ] note that by the choice of the parameters of this case the equation ( [ f(f ) ] ) becomes @xmath459 the numerical values of the zeros of @xmath31 are such that the graph corresponds to the exact solitary wave solution given in section @xmath440 , part @xmath101 . * * case @xmath460 for @xmath197:**the parameters are chosen as @xmath461 hence the solution becomes @xmath462 and the graph of this function is ( @xmath197 ) ] note that by the choice of the parameters of this case the equation ( [ f(f ) ] ) becomes @xmath463 here the function @xmath31 has one double zero @xmath464 and two simple zeros @xmath465 and @xmath466 so @xmath118 . as it is noted in section @xmath314 , part @xmath450 we have periodic solution which can be seen in the graph . * * case @xmath328 for @xmath197:**the parameters are chosen as @xmath467 hence the solution becomes @xmath468 and the graph of this function is ( @xmath197 ) ] note that by the choice of the parameters of this case the equation ( [ f(f ) ] ) becomes @xmath463 the zeros of the function @xmath31 are same as in the previous case .
so the graph fits to the fact given in section @xmath314 , part @xmath450 . * * case @xmath328 for @xmath198:**the parameters are chosen as @xmath469 hence the solution becomes @xmath470 and the graph of this function is ( @xmath198 ) ] note that by the choice of the parameters of this case the equation ( [ f(f ) ] ) becomes @xmath471 the numerical values of the zeros of @xmath31 are such that the graph corresponds to the exact solitary wave solution given in section @xmath440 , part @xmath101 .
we have studied symmetry reduced ( traveling waves ) equations of the kaup - boussinesq ( kb ) type of coupled degenerate kdv equations for @xmath11 .
the reduced equation turns out to be such that the square of the derivative of the dependent variable is equal to a fourth degree polynomial of the dependent variable .
there are four arbitrary constants in the polynomial function .
we have investigated all possible cases and gave all solitary wave solutions which rapidly decay to some constants of the @xmath0 kb equations .
there are periodic solutions of this set of coupled kdv equations in terms of the jacobi elliptic functions .
we first introduced special solutions of this type where the zeros of @xmath31 satisfy certain constraints . if we remove these constraints among the zeros we obtained the most general solution in terms of the elliptic functions of kb system under the assumed symmetry .
there are four different such solutions which differ by the initial values at the origin . for illustration
we have given the graphs of some interesting solutions .
we have also initiated the work on the cases for @xmath1 and @xmath2 .
we have given some results concerning these cases .
a detailed study of the traveling wave solutions of the cases @xmath1 and @xmath2 will be communicated later .
this work is partially supported by the scientific and technological research council of turkey ( tbitak ) .
references a.p .
fordy , p.p .
kulish , _ nonlinear schrdinger equations and simple lie algebras _ , commun . math.phys . * 89 * , 427 ( 1983 ) .
fordy , _ derivative nonlinear schrdinger equations and hermitian symmetric spaces _ , j. phys .
a * 17 * , 1235 ( 1984 ) .
svinolupov , _ jordan algebras and generalized korteweg - de vries equations _ , teo
fiz . * 87 * , 391 - 403 ( 1991 ) .
svinolupov , _ jordan algebras and integrable systems _ , funct
* 27 * , 257 - 265 ( 1993 ) .
schrdinger spectral problems with energy - dependent potentials as sources of nonlinear hamiltonian evolution equations _ , j. math
* 21 * , 2342 - 2349 ( 1980 ) , m. antonowicz , a.p .
fordy , _ a family of completely integrable multi - hamiltonian systems _ , phys .
a * 122 * , 95 - 99 , ( 1987 ) .
m. antonowicz , a.p .
fordy , _ coupled kdv equations with multi - hamiltonian structures _ , physica d * 28 * , 345 - 357 ( 1987 ) .
m. antonowicz , a.p .
fordy , _ factorization of energy - dependent schr@xmath472dinger operators : miura maps and modified systems _ ,
phys . * 124 * , 465 - 486 ( 1989 ) .
m. grses , a. karasu , _ degenerate svinolupov sytems _ , phys .
a * 214 * , 21 - 26 ( 1996 ) .
m. grses , a. karasu , _ integrable coupled kdv systems _ , j. math .
* 39 * , 2103 - 2111 ( 1998 ) .
m. grses , a. karasu , and v. sokolov , _ on construction of recursion operators from lax representation _ , j. math .
phys . * 40 * , 6473 - 6490 ( 1999 ) .
m. grses , _ integrable hierarchy of degenerate coupled kdv equations _
, in progress .
kaup , _ a higher - order water - wave equation and the method for solving it _ , progr .
* 54 * , 396 - 408 ( 1975 ) .
el , r.h.j .
grimshaw , and m.v .
pavlov , _ integrable shallow - water equations and undular bores _ , stud .
appl . math . * 106 * , 157 - 186 ( 2001 ) .
el , r.h.j .
grimshaw , and a.m. kamchatnov , _ wave breaking and the generation of undular bores in an integrable water wave system _ , stud .
* 114 * , 395 - 411 ( 2005 ) .
a.m. kamchatnov , r.a .
kraenkel , and b.a .
umrabov , _ asymptotic soliton train solutions of kaup - boussinesq equations _ ,
wave motion * 38 * , 355 - 365 ( 2003 ) .
ivanov , t. lyons , _ integrable models for shallow water with energy dependent spectral problems _
, arxiv:1211.5567 . | we give a detailed study of the traveling wave solutions of @xmath0 kaup - boussinesq type of coupled kdv equations .
depending upon the zeros of a fourth degree polynomial , we have cases where there exist no nontrivial real solutions , cases where asymptotically decaying to a constant solitary wave solutions , and cases where there are periodic solutions .
all such possible solutions are given explicitly in the form of jacobi elliptic functions .
graphs of some exact solutions in solitary wave and periodic shapes are exhibited .
extension of our study to the cases @xmath1 and @xmath2 are also mentioned . |
modern , highly specialised experimental systems often require extensive original software development .
this is true for all tasks from direct hardware control through to data reduction .
whilst individual research groups often approach such software development from an isolated perspective , the proliferation of digital hardware and wide - area networking makes global cooperative software development far more attractive , provided suitable common ground exists .
both cooperative software development and the adoption of standard packages provide a number of distinct advantages to the research community .
for instance , as less effort is wasted unnecessarily duplicating the work of others , cooperative development can lead to more efficient allocation of resources .
in addition , supporting the requirements of a larger user community promotes the development of basic , general purpose routines that may be used in a wider variety of situations .
these influences result in more modular and extensible software .
+ however , it should be noted that a greater level of care and cooperation is required in collaborative software development , especially the open source approach advocated in this paper .
for example , in contrast to most commercial software , `` black box '' modularity is undesirable in scientific analysis , especially when the application of certain algorithms requires experienced judgment .
open , well - documented code provides researchers with an accurate understanding of third - party analytical tools .
therefore , contributing developers must be willing to put their code in the public domain , making it freely available for non - commercial use by any other academic organisation .
although this facilitates the exchange of ideas , it also raises the issue of potential loss of intellectual property , which might discourage some authors . + it is also the case that collaborative development tends to become de - centralised , especially when multiple developers have the ability to commit fundamental changes to the code .
effective communication between the core developers becomes essential to the smooth running of the project , necessitating greater attention to version control , the maintenance of stable releases , and the development of extensive and concise documentation .
also , when a wider user community is affected by modifications to the software , exhaustive methods must be employed to ensure the validity of changes and the integrity of the system as a whole .
although each of these issues tend to increase the workload of the collaborative developer , a much larger body of users will benefit from the effort .
the global pulsar community is ideally suited to adopt a collaborative approach to software development .
it consists of a relatively small number of locally centralised groups that deal with different telescopes and instruments , leading to several parallel but incompatible software development paths .
as each path tends to be built around a highly specific data storage format , cross examination of data and algorithms is problematic .
in addition , because such software is generally designed for a limited purpose , it is often difficult to extend its functionality without introducing obfuscated code .
this is especially true when the program develops in an experimental fashion , as is often the case with scientific applications . in order to avoid future inflexibility , sufficient time and care
must be invested during the planning stage , calling on input from both experienced software developers and pulsar astronomers .
radio pulsars are broadband point sources of highly polarised emission that exhibit rapid pulsations with a characteristic period anywhere between one millisecond and ten seconds .
they are thought to be rotating neutron stars with a strong dipolar magnetic field whose axis is not aligned with the rotation axis of the star ( gold 1968 ) .
intense beams of emission originate at the magnetic poles , which sweep across the sky with each rotation of the star and produce the pulsed radio signal observed at earth .
+ the characteristic _ signature _ of any radio pulsar is its integrated polarimetric pulse profile , given by the observed stokes parameters averaged ( folded ) as a function of pulse longitude over several thousand individual pulses ( helfand , manchester , & taylor 1975 ) . under the influence of electrons in the interstellar medium ( taylor & cordes 1993 ) ,
this pulsed signal is broadened by dispersive frequency smearing , which must be corrected in order to infer the shape of the characteristic profile at the source .
this is normally done by dividing the observed bandwidth into narrow frequency channels , which are appropriately delayed relative to each other before summing the detected flux densities in each channel . however , as the dispersion measure may vary with time or may not be known with sufficient accuracy at the time of the observation , it is often necessary to store the individual pulse profiles observed in each frequency channel . + additionally , it is possible to create a mean pulse profile only if a suitably accurate model of the pulsar s spin frequency and phase is available .
the apparent pulse period is affected by a number of phenomena , including the spin - down , timing noise , and/or glitches intrinsic to the pulsar , variations in the interstellar dispersion , and doppler effects introduced by the relative motions of the earth and pulsar .
inaccuracies in the model that describes these effects introduce phase errors that accumulate with time and cause the integrated profile to become smeared .
therefore , it is often beneficial to store multiple , shorter integrations of the mean pulse profile instead of a single , long integration .
furthermore , when a pulsar is bright enough , a great deal of additional information about the characteristics of the pulsar emission can be obtained by recording and analysing each individual pulse .
therefore , a useful pulsar data format must be able to represent pulse profiles observed over multiple epochs of arbitrary length .
+ in summary , pulsar observations generally consist of a four - dimensional array of data indexed by polarization component , pulse phase , frequency , and epoch .
software support for sensible groupings in other dimensions , such as orbital phase , is also highly desirable .
in addition , data from a number of telescopes can be combined to increase sensitivity and contribute to the eventual detection of new phenomena , such as the cosmic background of stochastic gravitational radiation ( e.g. hellings & downs 1983 ; stinebring et al .
therefore , the data storage format should have a flexible structure that provides efficient access to key parameters , removed from any considerations of individual instruments or signal processing schemes .
pulsars are observed for a variety of reasons , from studying the nature of their structure and emission mechanism ( dodson , mcculloch , & lewis 2002 ) to utilising them as highly stable clocks and astrophysical probes ( taylor & weisberg 1982 ) .
consequently , the same pulsar observation can be used in a number of different contexts , one focusing on the variation of polarization with frequency , another measuring general relativistic effects on pulse times of arrival , etc .
nevertheless , our experience has shown that there exist many common tasks associated with pulsar data analysis that can be standardised within a modern open source development environment . + as a demonstration of the types of operations performed on pulsar data , consider the specific example of pulse time of arrival calculation . in order to increase the signal - to - noise ratio ( s / n ) of each observation , data are often integrated ( scrunched ) by several factors in one or more of the available dimensions . each resultant profile
is then cross - correlated with a high s / n standard profile known as a template , yielding an estimate of the longitudinal offset between the two .
this offset is added to the reference epoch associated with a fiducial point in the observed pulse profile , yielding an arrival time in the reference frame of the observatory , which is later converted into a barycentric arrival time using a solar system ephemeris .
this data reduction operation involves a number of typical tasks , including loading the arrays of numbers that represent the folded profiles and computing sums , products , rotations , weighted averages , and correlations of these arrays ; sometimes in the fourier domain .
most of these various operations must be performed in a manner consistent with the observational parameters , taking into account dispersive delays , observation time stamps and relative weightings of different frequency channels , for example . at each step
, the software must also ensure that all parameters are updated accordingly .
it should be noted that the pulsar data under consideration represents a point near the end of the typical pulsar data reduction chain .
the software presented in this paper is not intended for the direct handling of radio data such as that recorded by baseband systems , or for the purposes of performing computationally expensive offline searching , although some support for the storage of such data is provided in psrfits .
the code is also not designed to perform any phase - coherent dispersion removal or formation of filter - bank data ; these techniques are treated as separate computational tasks .
code for such data reduction is also available from the repository at the swinburne centre for astrophysics and supercomputing under the umbrella name of baseband dsp , a general library for use in digital signal processing .
+ the psrchive and psrfits schemes were designed from the beginning to form an object - oriented framework into which existing algorithms and data structures could be transplanted . by introducing layers of abstraction between the various levels of responsibility , the design remains both flexible and extensible .
for example , different telescopes and instruments require the storage of different types of information , including configuration parameters , observatory and instrumental status information , and other site - specific data . because there is no way of knowing exactly what future systems might include , both psrchive and psrfits implement a generalised scheme for incorporating arbitrarily complex data extensions , as described in sections [ sec : abstraction ] and [ sec : psrfits_defn ] . + in addition , a basic framework of crucial parameters common to all pulsar observations and a wide variety of fundamental data reduction algorithms , such as those described in section [ sec : processing ] ,
have been implemented .
each of these basic data structures and reduction operations may be used in the composition of more complex scientific processing algorithms . by virtue of continued development amongst the authors , the psrchive library includes an extensive array of such high - level algorithms for use in the calibration , visualisation , and analysis of pulsar data ; these can be used immediately on any of the supported file formats .
+ psrchive and psrfits were developed in parallel and are presented in the hope that they will promote increased data portability .
the psrfits file format also serves as an example of how to incorporate other , pre - existing file formats into the new scheme , as described in section [ sec : plugin ] . after two years of development ,
the code is now ready for formal release to the wider pulsar community . in the following sections
, we describe the implementation of the new schemes and outline the specific advantages that they offer .
the modularity and extensibility required of our new scheme suggested an object - oriented approach .
since much of the existing swinburne analysis code had already been written in both the c and c++ programming languages , it seemed a natural step to progress in c++ .
the concepts of object classes and inheritance provided and enforced by the syntax of this language offer a sound foundation on which to develop .
in particular , object - oriented design has aided in the realization of simultaneous support of multiple file formats .
we are aware that a majority of pulsar research groups prefer to write a more procedural style of code , using fortran or c. however , we feel that the benefits of an object - oriented approach to data processing significantly outweigh the potential learning curve involved in becoming proficient with c++ .
the required functionality of psrchive is built around a core framework of c++ object classes .
the fundamental unit of all pulsar observations is the individual pulse * profile * , a one - dimensional array of floating point numbers , indexed by pulse phase .
the * integration * is a two dimensional vector of * profile * instances , indexed by frequency and polarisation , as measured over a particular epoch . in turn , the * archive * is a one dimensional vector of * integration * instances , indexed in one of a number of possible ways ( normally time ) . each of these classes implement a wide range of basic data manipulation and processing operations . + in the language of c++ , we define the namespace * pulsar * , which contains the three _ base classes _ : * pulsar::archive * , * pulsar::integration * , and * pulsar::profile*. in addition , there are other object classes in the * pulsar * namespace that deal with specific tasks related to pulsar data analysis . for example , the * pulsar::calibration * class employs various mathematical models of the instrumental response to calibrate polarimetric observations ( van straten 2004 ) . the three base classes implement a wide variety of basic algorithms , known as _ methods _ , that are commonly used in pulsar data analysis .
however , they do not require knowledge of any specific details related to system architecture , enabling their use as templates upon which to base lower - level development .
these templates define the minimum set of parameters , known as _ attributes _ , required to implement the data analysis methods , including observational parameters such as the name of the source , centre frequency , bandwidth , etc . at the level of the * pulsar::archive * and * pulsar::integration * base classes , nothing is known about how data are stored on permanent media or in computer memory .
+ the necessary task of translating between the two realms is performed by _ derived classes _ that _ inherit _ the base classes . in order to inherit a base class ,
it is necessary for the derived class to provide access to the required attributes and to implement the methods used to read and write the data stored on disk .
therefore , for each specific file format represented in the psrchive scheme , there corresponds a derived class that inherits * pulsar::archive*. the syntax for the data access and file input / output methods is defined by the base class and enforced by the c++ compiler , allowing all derived classes to be treated as equal .
therefore , high - level code can be written in the language of the base class definition without the need for considering the implementation details of the derived classes .
this abstraction , which is crucial to the flexibility of the psrchive scheme , is demonstrated by the unified modeling language ( uml ) class diagram shown in figure [ fig : class_diagram ] . in order to take full advantage of this level of data abstraction ,
the psrchive scheme makes use of dynamic shared object libraries , or _
plug - ins_. these libraries are compiled using special options that allow them to be linked into a program at run - time .
perhaps the best known example of such a system is the plug - in scheme used to add functionality to many web browsers . within psrchive , the machine code that defines a * pulsar::archive*-derived class
is stored inside a single plug - in file .
the plug - in files corresponding to different file formats are held in a central location that is scanned on the first attempt to load pulsar data .
the available plug - ins determine which file formats are understood by providing a test routine that returns true if a file on disk is of its own format .
in this way , psrchive applications can quickly scan a given data file , select the appropriate derived class and load the pulsar data .
this ensures that , as the number of supported file formats grows , the size and complexity of any given application program remains the same .
we encourage all interested research groups that currently maintain separate data formats to gain experience with the psrchive scheme by developing their own file format plug - ins . by making the plug - in code publicly available
, research groups will be able to exchange data already stored using different file formats .
+ in order to accommodate the stringent reliability requirements of observatory online processing and monitoring systems , we also offer the option of compiling all psrchive applications using a static linkage scheme .
this makes the binary executables larger on disk but removes the possibility of problems arising should a shared object file be accidentally deleted or recompiled .
either option can be selected by simply editing one line in the makefile scheme .
one of the motivating factors behind the development of the psrchive scheme was the alleviation of problems associated with highly specific and non - portable data storage formats .
this effort has highlighted several compelling reasons for the pulsar community to move towards a more modular and standardised storage format .
for instance , the existence of a standard file format would significantly decrease the amount of effort required to integrate and test new instrumentation .
historically , file formats have accreted features as they became desirable or necessary .
given the wealth of past experience available , it seems a logical step to define a new format that encompasses a wide range of features from the beginning and is written in a modular way so as to enable rapid , backwards - compatible upgrades .
indeed , one particular standard storage format has already won wide acclaim within the astronomical community ; the flexible image transport system ( fits ) ( hanisch et al . 2001 ) has been in widespread use for approximately three decades and has evolved into a highly adaptable data storage scheme .
the format has been placed under the administration of the iau fits working group and a wide array of software is available for fits file manipulation .
the nasa high energy astrophysics science archive research centre provides useful libraries and applications for manipulation and interrogation of fits - based files .
for example , the program * fv * has made the process of testing and debugging the relevant psrchive software much more straight forward .
+ in accordance with fits standards , a psrfits file consists of a primary header - data unit ( hdu ) followed by a series of extension hdus .
the primary hdu contains basic information such as telescope identification and location , observation start time etc .
extension hdus , formatted as binary tables , contain specific information related to the observation such as the pulsar ephemeris , calibration data , and the pulsar data .
although psrfits is primarily designed to store folded or single - pulse profile data , it can also accommodate continuous time series data .
a useful feature of the standard fits input / output routines is that new hdus and header parameters may be added transparently if they are unknown to the reading program , they are ignored .
furthermore , unused hdus need not be written , even though they are present in the definition .
this feature allows , for example , a user group to add information particular to a certain instrument without compromising use of the definition by other groups .
a novel feature of the psrfits definition is the inclusion of hdus containing ` history ' information .
for example , the first line of processing history hdu contains information about the data acquisition program and the initial structure of the file .
subsequent lines record details of modifications to the structure or data ( for example , partial or complete de - dispersion or interference excision )
. a permanent record of the steps that have been applied during data reduction has proven to be of great value when later assessing the quality and validity of observational data .
the current version of the psrfits definition file is available on the atnf web pages .
table [ tb : psrfits ] describes the header - data units included in the current definition ( version 1.18 ) .
+ in addition to the main and processing history hdus , a number of optional hdus have been defined for general use with a variety of instrumentation .
these enable the storage of important status and diagnostic information about the observation , and demonstrate the modularity and extensibility of the psrfits file format .
the physical parameters stored in the ephemeris history hdu are based on the pulsar timing program , * * tempo**. from the ephemeris parameters are derived the polynomial coefficients ( polyco ) used to predict the apparent pulsar period and phase at the epoch of the observation ; these coefficients are stored in the polyco history hdu . as improved physical parameters become available , the data may be reprocessed , leading to new rows in the ephemeris and polyco history tables .
the calibration and feed cross - coupling hdus are designed to work with the routines in the * pulsar::calibration * class . owing to the intrinsic modularity of fits , these additional hdus are all optional ; in fact , it is not even strictly necessary to include any integration data in a psrfits file .
for example , the polarimetric calibration modeling program creates a file containing only the feed cross - coupling , injected calibration polarisation , and flux calibration hdus .
this modularity is similar to that made available through the use of votable xml standards and it is likely that psrfits could in future be incorporated into the international virtual observatory system with a minimum of effort .
llhdu title & + + main header & observer , telescope and receiver information , + & source name and observation date and time + + processing history & date , program and details of data acquisition + & and each subsequent processing step + + digitiser statistics & digitiser mode and count statistics + + digitiser counts & digitiser mode and count rate distribution + + original bandpass & observed bandpass in each polarisation + & averaged over the observation + + coherent de - dispersion & parameters for coherent de - dispersion of + & baseband data + + ephemeris history & pulsar parameters used to create or modify + & profile data + + polyco history & elements of the polyco file used to predict + & the apparent pulsar period + + flux calibration & system temperature and injected noise + & calibration data as a function of frequency + & across the bandpass + + injected calibration polarisation & apparent polarisation of the injected + & noise calibration signal as a function + & of frequency + + feed cross - coupling & parameters of feed cross - coupling as a + & function of frequency + + integration data & pulse profiles or fast - sampled data as a function + & of time , frequency and polarisation + +
the psrchive scheme includes an extensive set of pre - written application programs that can be used to manipulate pulsar data in various ways .
these include both command line tools and graphical user interfaces built using trolltech s qt , a c++ toolkit for multi - platform gui and application development .
table [ tb : psrapps ] presents a list of applications included in the package at the time of publication , with a brief description of each .
1.0 @ll application & + + * pav * & archive data visualisation . based on the pgplot graphics + & subroutine library with a simple command line interface + + * vap * & archive header examination , allowing multiple user + & selectable header parameters to be printed as ascii to + & the terminal + + * pam * & archive manipulation , compression and processing + + * pat * & pulse profile arrival time calculation , based on cross + & correlation with a standard template profile + + * pas * & standard profile phase alignment , for timing with multiple + & standard template profiles + + * paz * &
radio frequency interference mitigation tool including + & manual and automated channel zapping and sub - integration + & removal + + * pac * & archive polarimetric and flux calibration tool based on a + & user - selectable set of advanced algorithms + + * pcm * & polarimetric calibration modeling , creates instrumental + & response transformations for use with * pac * + + * psrgui * & interactive point - and - click data visualisation with a qt + & graphical interface + + * psradd * & combination of multiple archives for formation of high + & s / n profiles + + * rhythm * & a graphical interface for pulse arrival time fitting based + & on * tempo * + + readers may note that the modular philosophy at the heart of psrchive extends all the way through to the user level applications .
each program tends to be small and focused on a specific task , be it data compression , timing , rfi mitigation , etc .
this greatly simplifies development and maintenance compared to having one monolithic program with multiple purposes .
psrchive was designed to provide users with more than just a set of pre - made applications .
the classes , libraries and examples provided are intended to simplify the task of building new processing tools . to a large extent , developers who build on the psrchive scheme do not have to directly manipulate the arrays of pulse profile amplitudes .
instead , member functions of the various classes can be called to perform basic operations like baseline removal and phase rotation .
this has the dual benefit of labour saving both in the initial development phase and in the debugging phase , as both the authors and other users have already verified and tested the provided routines . in case direct access to the profile amplitudes
is required , we also provide interface functions that return c style arrays . in the experience of the authors , the extra layer of abstraction provided by the psrchive scheme can cut down the time between program concept and full implementation to a matter of hours .
new applications can be built with only a few lines of code .
for example , to remove the system noise floor , compress all frequency channels and output the processed archive : .... # include " pulsar / archive.h " int main ( ) { pulsar::archive * arch = 0 ; arch = pulsar::archive::load("filename " ) ; arch->remove_baseline ( ) ; arch->fscrunch ( ) ; arch->unload ( ) ; } .... this simple program defines a pointer to a * pulsar::archive * and calls the generic * pulsar::archive::load * routine , which takes a filename argument , applies a number of tests to the file on disk ( depending on the available plug - ins ) and decides whether or not it understands the particular format .
if so , it summons the appropriate derived class to read the data from disk .
once the data have been loaded , the * pulsar::archive::remove_baseline * function is called .
.... void pulsar::archive::remove_baseline ( float phase , float width ) { try { if ( phase = = -1.0 ) phase = find_min_phase ( width ) ; for ( unsigned isub=0 ; isub < get_nsubint ( ) ; isub++ ) get_integration(isub ) - > remove_baseline ( phase , width ) ; } catch ( error & error ) { throw error + = " pulsar::archive::remove_baseline " ; } } .... the * pulsar::archive::remove_baseline * function takes two arguments : the phase and width of the off - pulse baseline .
both arguments are assigned default values in the archive.h header file ; if phase is left unspecified , then the off - pulse baseline phase will be found using the * pulsar::archive::find_min_phase * method ; if width is unspecified , then a default value will be used .
the * pulsar::archive::remove_baseline * method makes multiple calls to the * pulsar::integration::remove_baseline * routine , which performs the actual modification of amplitudes as follows : .... void pulsar::integration::remove_baseline ( float phase , float width ) { if ( pulsar::integration::verbose ) cerr < < " pulsar::integration::remove_baseline entered " < < endl ; try { if ( phase = = -1.0 ) phase = find_min_phase ( width ) ; vector < float > phases ; dispersive_phases ( this , phases ) ; for ( unsigned ichan=0 ; ichan < get_nchan ( ) ; ichan++ ) { float chanphase = phase + phases[ichan ] ; for ( unsigned ipol=0 ; ipol < get_npol ( ) ; ipol++ ) * ( profiles[ipol][ichan ] ) -= profiles[ipol][ichan ] - > mean ( chanphase , width ) ; } } catch ( error & error ) { throw error + = " integration::remove_baseline " ; } } .... this nested structure reduces the length of high - level routines , allowing actual computations to be done at the level of abstraction that best suits the task .
likewise , the * pulsar::integration::remove_baseline * routine calls various member functions of both the * pulsar::integration * and * pulsar::profile * classes , computing the pulse phase at which the minimum baseline level occurs in the total intensity of the entire band .
adjustments for dispersive delays in each channel are performed and the mean level at this phase is individually removed from each * pulsar::profile * stored in the * pulsar::integration*. developers should also note the extensive use of * try / catch * blocks and a specifically designed * error * class that carries descriptive information about any exceptions thrown back to the calling procedure .
psrchive is freely available to the worldwide academic community .
it is held in a repository at swinburne university of technology in melbourne , australia and may be accessed via the concurrent versions system . as it is distributed as source code ,
some experience with programming and compilation is necessary ; however , installation can be done in a fairly simple step - by - step manner thanks to the standard makefile scheme included with the package .
the code is compatible with all versions of the gnu compiler collection between 2.95 and 3.2.2 and is routinely tested on both the linux and solaris operating systems .
every effort will be made to ensure compatibility with future gcc releases .
+ the psrchive scheme makes use of several external libraries , including the starlink project slalib package .
it also requires at least one external fft library and includes wrappers that provide compatibility with either fftw 2.1.5 ( available under the gpl ) or intel mkl ( commercially available from intel ) .
the pgplot graphics subroutine library is also an integral part of the scheme . + full documentation including instructions for download and installation are available online by following the menu options at the swinburne centre for astrophysics and supercomputing web site .
read - only access to the repository is granted upon receipt by the developers of a secure shell v2.0 public key that is used to allow remote entry to the server .
write permissions to the repository require a computing account with the swinburne centre for astrophysics and supercomputing .
please direct all enquiries regarding access to * [email protected]*. psrchive reference documentation is maintained online .
in addition to the online descriptions , each command line application has a * -h * option that displays a quick summary of how to use the program .
the library of c++ classes is extensively documented using the doxygen system ; the source code contains tagged comments from which the online manual is automatically generated .
this manual is intended as a reference to programmers as it primarily describes the member functions available in each class and the syntax of their arguments .
although we provide no official support for the software , we are willing to assist with psrchive related problems as time permits .
general queries regarding installation or operation can be addressed to * [email protected]*. we also provide a mechanism for reporting serious bugs via an online interface known as yaq , which can be found at the swinburne pulsar group web site .
the task of organising astronomical data into a logical format lends itself surprisingly well to the object - oriented programming paradigm .
the combination of psrchive and psrfits provides a powerful , ready - to - use pulsar data archival and reduction system that can be rapidly adapted to new instruments .
we hope that the ready availability of an open source data reduction framework will facilitate large scale collaborative projects , such as an extended pulsar timing array ( foster and backer 1990 ) .
therefore , we encourage both scientists and engineers involved with pulsar data acquisition and reduction to consider taking advantage of these packages .
we have benefited greatly from the advice and assistance of many colleagues in developing the scheme described . in particular , we thank matthew bailes and others in the swinburne pulsar group and nina wang at the australia telescope national facility .
thanks also to ben stappers , russell edwards and george hobbs for constructive feedback during development .
we would also like to thank the referee for insightful comments which have led to important improvements to this manuscript .
dodson , r. g. , mcculloch , p. m. & lewis , d. r. 2002 , apj , 564 , 85 | a new set of software applications and libraries for use in the archival and analysis of pulsar astronomical data is introduced .
known collectively as the psrchive scheme , the code was developed in parallel with a new data storage format called psrfits , which is based on the flexible image transport system ( fits ) .
both of these projects utilise a modular , object - oriented design philosophy .
psrchive is an open source development environment that incorporates an extensive range of c++ object classes and pre - built command line and graphical utilities .
these deal transparently and simultaneously with multiple data storage formats , thereby enhancing data portability and facilitating the adoption of the psrfits file format . here
, data are stored in a series of modular header - data units that provide flexibility and scope for future expansion . as it is based on fits , various standard libraries and applications
may be used for data input / output and visualisation . both psrchive and psrfits
are made publicly available to the academic community in the hope that this will promote their widespread use and acceptance .
= 2em = 15.5 cm = 22.6 cm = -1.0 cm = 0.5 cm = 0.5 cm @xmath0 swinburne centre for astrophysics & supercomputing , + mail # 31 , po box 218 , hawthorn , vic 3122 , australia + [email protected] + @xmath1 australia telescope national facility , + po box 76 , epping , nsw 1710 , australia , + [email protected] + @xmath2 netherlands foundation for research in astronomy , + po box 2 , 7990 aa dwingeloo , the netherlands , + [email protected] + * keywords : pulsars : general , methods : data analysis * |
in the supplementary material , we derive the tight - binding model from the continuous model @xmath43 in eq . ( 3 ) of the main text and compare their spectra to verity the tight - binding model s reliability .
let us first focus on the discretization of @xmath27 , which can be written as in the second quantization language @xmath187where @xmath188 is restricted to the @xmath28 plane and @xmath189^{t}$ ] with @xmath190 [ @xmath191 annihilating ( creating ) an atom with spin @xmath50 ( @xmath192 ) located at @xmath193 .
they satisfy the anti - commutation or commutation relation @xmath194_{\pm } = \delta _ { \sigma \sigma ^{\prime } } \delta ( \mathbf{r}-\mathbf{r}^{\prime } ) $ ] for fermionic atoms ( @xmath195 ) or bosonic atoms ( @xmath196 ) , respectively .
we expand the field operator using local wannier functions @xmath197where @xmath198 is the wannier function for @xmath199 located at the site @xmath200 for the @xmath104-th band , and @xmath201 ( @xmath202 ) annihilates ( creates ) an atom located at the state @xmath198 with spin @xmath50 .
let us focus on the physics in the lowest band and thus assume @xmath203 , thereby simplifying the above expression @xmath204where @xmath205 with @xmath206 being the wannier function along @xmath15 .
we note that although this is an approximation , it proves to be qualitatively correct and we will verify it by comparing the spectra obtained by solving the continuous model and the tight - binding one . using this expansion , we obtain the following tight - binding model without @xmath18 @xmath207where @xmath208 with @xmath209 denote the hopping amplitudes defined as @xmath210 w_{j_{\nu } + 1}^\nu(r_\nu ) .
\label{et}\]]we approximately derive the tight - binding term contributed by the spin - dependent lattice as follows @xmath211where @xmath212 ) plane for @xmath213 in ( a ) , ( b ) , ( e ) and ( f ) , and in the ( @xmath214 ) plane for @xmath215 in ( c ) and ( d ) .
( b ) and ( f ) are obtained by diagonalizing the tight - binding model while other spectra by solving the continuous model .
( a - d ) correspond to the model with @xmath216 , @xmath217 , and ( e ) and ( f ) to the simplified model with @xmath218 and @xmath219 . in ( d ) , @xmath220 , while in other figures , @xmath127 .
the parameters for the continuous model are @xmath221 and @xmath222 ; the parameters for the tight - binding model are @xmath223 and @xmath224 . for the spectra of the continuous model , we have shifted the energy at the touching point to the energy at the same point of the spectra of the tight - binding model.,width=384 ] employing the condition
@xmath225 given that one of the optical wells is located at @xmath226 when @xmath199 , we get @xmath227where the last two relations are obtained because @xmath228 and @xmath229 . therefore ,
when we only consider the nearest - neighbor hopping , we obtain the following spin - orbit coupling term of the tight - binding model : @xmath230+h.c.\\ & = & t_{so}\sum_{j_x , j_y}(-1)^{j_x+j_y}\left[(\hat{c}_{j_x , j_y,\uparrow } ^{\dagger } \hat{c}_{j_x+1,j_y,\downarrow } -\hat{c}_{j_x , j_y,\uparrow } ^{\dagger } \hat{c}_{j_x-1,j_y,\downarrow } ) \right .
\nonumber \\ & & \left.+i(\hat{c}_{j_x , j_y,\uparrow } ^{\dagger } \hat{c}_{j_x , j_y+1,\downarrow } -\hat{c}_{j_x , j_y,\uparrow } ^{\dagger } \hat{c}_{j_x , j_y-1,\downarrow } ) \right]+h.c.,\end{aligned}\]]where we have assumed @xmath231 and @xmath232 in the last equation so that @xmath233 in 3d , after replacing @xmath234 with @xmath235 and including the zeeman field term , we obtain @xmath236 in eq.(5 ) in the main text . to verify the reliability of the tight - binding model , in fig . [ sifig1 ] , we compare the energy spectra of the tight - binding model with that of the continuous model , which are numerically calculated using fourier series expansion of a bloch function . for the model with @xmath216 and @xmath217 ,
the doubly degenerate spectra in fig .
[ sifig1](a ) and that in fig .
[ sifig1](b ) qualitatively agree with each other and both exhibit a touching point . in fig .
1(c ) and ( d ) of the main text , we plot the spectra of the tight - binding model in the @xmath237 plane , and here we plot those of the continuous model in ( c ) and ( d ) , which also qualitatively agree with those in the main text figure . for the simplified model with @xmath218 and @xmath219 , the spectra in fig .
[ sifig1](e ) and ( f ) are also qualitatively consistent with each other ; in both figures , a four - fold degenerate touching point splits into two doubly degenerate touching points .
we note that for the simplified model , we have added an additional term @xmath238 in the tight - binding model since the system breaks is but still respects ptrs ; @xmath239 is evaluated by the position of weyl points in the spectrum of the continuous model . here
, we have utilized the relation @xmath240 in the case that @xmath231 and @xmath232 , which can be justified from the perspective of a symmetry .
it is easy to confirm that the 2d continuous hamiltonian respects a @xmath241 symmetry , i.e. , @xmath242 , where @xmath243 with @xmath244 ; the representation of this symmetry in the momentum space corresponds to @xmath245 [ @xmath246 with @xmath247 , indicating that the additional term must take the form of @xmath248 . | weyl points can be divided into two classes : type - i and type - ii , which are separated by a topological lifshitz - type quantum phase transition .
we propose a scheme to realize both type - i and type - ii weyl points in the single - particle spectra of ultracold atomic gases by introducing an experimentally feasible configuration for three - dimensional spin - orbit coupling .
we find three weyl points in the resultant hamiltonian , one of which is a four - fold degenerate touching point consisting of two weyl points carrying the same topological charge .
all these weyl points can transition from type - i to type - ii via tuning of a convenient experimental knob .
the proposed system provides an ideal platform to study different types of weyl points and directly probe the lifshitz phase transition between them .
weyl fermions @xcite were initially predicted to describe one - half of a massless dirac fermion in particle physics .
although neutrinos were previously conjectured to be weyl fermions , they may have masses and are therefore not weyl particles . recently
, weyl fermions were theoretically predicted and experimentally observed as the gapless touching points in the band structure of a class of three - dimensional ( 3d ) solid - state materials called weyl semimetals murakami2007njp , wan2011prb , vafek2014,lu2015,xu2015,lv2015 .
near such a weyl point , the bands can be described by weyl hamiltonian and the dispersion is linear along all three momentum dimensions .
a weyl point can also be viewed as a magnetic monopole in 3d momentum space carrying an integer topological charge ( i.e. , chern number ) @xcite .
apart from the conventional type - i weyl point , a new type of weyl points ( called type - ii ) have been proposed to exist in type - ii weyl semimetals soluyanov2015nature and quasiparticle spectra of spin - orbit coupled superfluids @xcite .
different from a single point fermi surface of a type - i weyl point , the fermi surface of a type - ii weyl point consists of particle and hole pockets along with a touching point yong2015prl , soluyanov2015nature , and type - i and type - ii weyl points are separated by a topological lifshitz - type quantum phase transition where the dimension of the fermi surface changes @xcite .
because of their distinct features such as open fermi surface yong2015prl , soluyanov2015nature , bergholtz2015prl , finite density of states @xcite , chiral anomaly soluyanov2015nature , yang2016,udagawa2016 and other peculiar properties volovik2016,yang2016,udagawa2016,tiwari2016,beenakker2016 , type - ii weyl points have attracted significant attention in condensed matter communities @xcite .
remarkable experimental progress has been reported recently on type - ii weyl semimetals and their properties @xcite , however , it remains a significant challenge in solid - state materials to observe the characteristic lifshitz phase transition between type - i and type - ii weyl points .
recent experiments on one - dimensional and two - dimensional spin - orbit couplings @xcite in ultracold atomic gases have paved another way for discovering novel quantum states with topological properties @xcite .
although type - i and type - ii weyl points were proposed in quasiparticle spectra of spin - orbit coupled fermi superfluids @xcite , realization of such superfluids is still a big challenge with current experimental technology @xcite .
a more feasible experimental scheme is to realize weyl points in the single - particle spectra without the need of low - temperature fermi superfluids .
some proposals have been made concerning type - i weyl points @xcite , but there is no scheme yet to realize type - ii weyl points in cold atoms in the single - particle spectra . in this paper , we propose a scheme to realize both type - i and type - ii weyl points in the single - particle spectra with cold atoms in a 3d optical lattice .
we find a laser - atom coupling configuration readily achievable with the current experimental technology that realizes the required spin - orbit coupling . due to its controllability
, this system offers a unique opportunity to observe the characteristic lifshitz - type topological quantum phase transition between type - i and type - ii weyl points through continuous tuning of a convenient experimental knob .
in addition to two doubly degenerate weyl points , we also find a four - fold degenerate touching point in the single - particle spectra of this system protected by pseudo - time - reversal ( ptrs ) , inversion ( is ) , and combined rotational symmetries , which is comprised of two weyl points with the same topological charge .
we start by briefly reviewing the concept of type - i and type - ii weyl points that are described by an effective hamiltonian @xmath0 @xcite , where @xmath1 ( @xmath2 ) denote momenta ( pauli spins ) and @xmath3 are real parameters .
its energy spectrum is given by @xmath4 with @xmath5 labeling particle and hole bands .
when @xmath6 , the energy of the particle ( hole ) band is positive ( negative ) , except at the touching point where the energy vanishes .
this touching point is dubbed type - i weyl point .
when @xmath7 , in certain regions , the energy goes negative for the particle band and positive for the hole band , leading to the open fermi surface besides a touching point @xmath8 .
this structure is dubbed type - ii weyl point . to realize type - i and type - ii weyl points with cold atoms , we consider the following hamiltonian that describes atoms in an optical lattice @xmath9where @xmath10 is the momentum operator , @xmath11 is the mass of atoms , @xmath12 ( @xmath13 )
denote the strength of the optical lattices with the period being @xmath14 along the @xmath15 direction , @xmath16 is the zeeman field , @xmath17 are pauli matrices for spins , and @xmath18 is a laser - induced spin - orbit coupling term taking the form @xmath19with @xmath20 , @xmath21 , and @xmath22 proportional to the laser strength .
later , we will describe the laser configuration that directly realizes the hamiltonian ( 1 ) , in particular the spin - orbit coupling term @xmath18 .
after a unitary transformation , the hamiltonian @xmath23 is equivalent to @xmath24where @xmath25 , @xmath26 , and the 2d hamiltonian @xmath27 in the @xmath28 plane is expressed as @xmath29 \notag \\ & + & \left [ \omega _ { so}(m_{x}+im_{y})|\uparrow \rangle \left\langle \downarrow \right\vert + h.c.\right ] .\end{aligned}\ ] ]
plane , where each unit cell is made up of a and b sites and @xmath30 with @xmath31 and @xmath32 are unit vectors .
( b ) the first brillouin zone denoted by the dashed box . @xmath33 and @xmath34 are reciprocal unit vectors .
green [ @xmath35 and red solid [ @xmath36 circles denote the position where weyl points exist in the momentum space without zeeman fields @xmath37 . here
, @xmath38 .
( c)(d ) spectra near a type - i and type - ii weyl point with respect to @xmath39 for @xmath40 ( see the spectra of the continuous model in supplementary materials ) . in ( d ) , the gray plane ( i.e. , @xmath41 with @xmath42 being the energy at the touching point ) intersects both the particle and hole bands in addition to the touching point , implying the weyl point is type - ii.,width=326 ] to see how the weyl points appear in this model , we discretize @xmath27 and study its physics in the tight - binding model ( see supplementary materials for details of discretization )
. the tight - binding form of @xmath43 can be written as [ let us first neglect @xmath44 term and focus on type - i weyl points ] @xmath45where the hopping term is @xmath46 with @xmath47 ( @xmath48 ) being the operator annihilating ( creating ) an atom localized at @xmath49 in the @xmath28 plane with spin @xmath50 and momentum @xmath51 along the @xmath52 direction ; the zeeman field term is @xmath53 ; and the position dependent spin - orbit coupling term is given by @xmath54.$ ] because of the existence of the position dependent spin - orbit coupling , we need to choose a unit cell consisting of two sites : a and b [ as shown in fig . [ fig1](a ) ] and we write the hamiltonian in the momentum space in the new basis @xmath55 with @xmath56 @xmath57where @xmath58 , @xmath59 and @xmath60 ; @xmath61 are pauli matrices acting on a , b sublattices . in the @xmath62 plane ,
@xmath63 ( @xmath64 ) respects is : @xmath65 and in the absence of @xmath66 , ptrs : @xmath67 with @xmath68 and @xmath69 being the complex conjugate operator .
these two symmetries guarantee that the spectrum in this specific plane ( @xmath70 ) is at least doubly degenerate , implying that the gapless point , if exists , is four - fold degenerate .
we note that in the continuous model is corresponds to @xmath71 with the inversion center at @xmath72 and ptrs corresponds to @xmath73 with @xmath74 and @xmath75 .
specifically , the eigenvalues of @xmath76 read @xmath77 with @xmath78 , which is consistent with our symmetry analysis that the energy band at each @xmath79 is doubly degenerate without @xmath66 .
clearly , when @xmath80 and @xmath81 , there emerge touching points as shown in fig .
[ fig1](c ) .
the first condition requires that @xmath82 is equal to @xmath83 or @xmath84 . in the former case ,
the touching point is located at @xmath85 if @xmath86 ( thus @xmath87 at this point ) as a result of a combined rotational symmetry ( i.e. , @xmath88 where @xmath89 with @xmath90 and @xmath91 being the four - fold rotational operator along @xmath52 when @xmath92 and @xmath93 ) readily achievable in experiments ; in this plane , both ptrs and is are preserved and the touching point is therefore four - fold degenerate ( also seen from the eigenvalues ) .
the dispersion is linear along all three momenta directions as visually shown in fig .
[ fig1](c ) . generally , such a four - fold degenerate touching point is a dirac point incorporating two weyl points with the opposite chern number once a system respects these two symmetries .
however , in our case , they can be simultaneously satisfied only in the plane @xmath70 .
in fact , the touching point comprises two weyl points with the same chern number . to demonstrate this ,
we write down the effective hamiltonian in the vicinity of this touching point , which , after a unitary transformation , reads @xmath94where @xmath95 , @xmath96 , and @xmath97 ; @xmath98 is a @xmath99 identity matrix ; the momenta @xmath100 are measured with respect to the weyl point .
it is clear that the two weyl points have the same chirality . to characterize the topological charge of a weyl point ,
we define the first chern number @xmath101where the surface @xmath102 encloses the weyl point , and @xmath103 is the berry curvature @xcite for the @xmath104-th band with @xmath105 being its wave function . for our parameters ,
this yields @xmath106 for the above discussed weyl point , where each spinor contributes @xmath107 .
in the latter case where @xmath108 , there are two touching points , respectively located at @xmath109 with @xmath110 .
they are only doubly degenerate ( owing to the breaking of ptrs ) and the dispersion is linear in all three momentum directions .
our calculation demonstrates that their chern numbers are equal to @xmath111 .
we note that because of the existence of the spin - independent energy @xmath112 , the energy at these two weyl points is equal only when @xmath113 . to illustrate that these type - i weyl points can transition to the type - ii weyl points ,
let us include @xmath44 and expand the hamiltonian near a weyl point , e.g. , @xmath114 , @xmath115with @xmath116 .
clearly , when @xmath117 ( i.e. , @xmath118 ) , at certain regions , the energy of both particle bands goes negative while that of both hole bands goes positive [ as visually displayed in fig .
[ fig1](d ) ] , indicating that the weyl point becomes type - ii @xcite .
similarly , the type - i weyl point at @xmath119 becomes type - ii when @xmath120 or @xmath121 , and the point at @xmath122 becomes type - ii when @xmath123 or @xmath124 .
from the overlap and distinct constraint under which type - ii weyl points appear at different positions , we see that the system can be explicitly categorized into three groups : all weyl points are type - i , all are type - ii , and partial type - i and partial type - ii . in cold atoms , we can control the two - photon detuning ( i.e. , @xmath16 ) to continuously tune their transitions . for @xmath125 under the open boundary condition along the @xmath126 direction .
in ( a ) , @xmath127 and the weyl points are all type - i ; in ( b ) , @xmath128 and a weyl point on the left side is type - ii and all others are type - i .
the black lines denote the surface states.,width=326 ] generally , in a system with weyl points , there emerge surface states connecting two weyl points with opposite chern numbers ( called fermi arch ) in the presence of an edge . to see the surface states in our system ,
we calculate the energy spectra under the condition of open boundaries along @xmath126 . in fig .
[ fig2 ] , we plot the energy spectra with respect to @xmath129 for @xmath125 when @xmath127 with type - i weyl points [ in ( a ) ] and @xmath128 with a type - ii weyl point [ in ( b ) ] .
it shows that in both cases there exist surface states ( denoted by the black lines ) connecting the weyl point in the middle with those on two sides .
the spectrum in fig . [ fig2](b ) also illustrates the characteristic of the type - ii weyl point located at @xmath130 that all the particle and hole bands near the weyl point at each @xmath131 are positive or negative with respect to @xmath132 , the energy at the weyl point . besides the model that we have discussed ,
if we choose a simplified scheme with @xmath133 and @xmath134 , we can still obtain both type - i and type - ii weyl points .
however , while it still respects ptrs , it breaks is , splitting the four - fold degenerate weyl point into two doubly degenerate ones ( see supplementary materials ) . to satisfy these symmetry requirements ,
we need to add an additional term into the model ( [ htbk ] ) : @xmath135 which respects ptrs but breaks is .
if @xmath92 and @xmath93 , we have @xmath136 due to a symmetry requirement ( see supplementary materials ) , and the spectrum is @xmath137 with @xmath138 , and weyl points emerge at the position @xmath139=[\lambda
\delta \theta ,
\pi + \lambda \delta \theta , -2m\pi h_{z}/(\hbar ^{2}k_{l{z}}^{2})]$ ] ( or @xmath140 $ ] ) with @xmath141 .
moreover , when @xmath142 ( @xmath143 ) , the former ( latter ) wey points become type - ii .
to realize the hamiltonian ( [ hc ] ) , we propose an experimental scheme ( as shown in fig .
[ fig3 ] ) that is based on a modification of the experimental configuration in ref .
@xcite .
we consider two hyperfine states of alkali atoms such as @xmath144k and @xmath145rb and employ two independent raman processes to create the spin - dependent optical lattices .
each raman process involves two linearly polarized blue - detuned raman laser beams with the polarization being along @xmath146 ( parallel to the magnetic field ) and @xmath126 , respectively .
each pair of raman laser beams is characterized by a pair of rabi frequencies @xmath147 $ ] and @xmath148 $ ] , respectively .
they form a standing wave along @xmath126 or @xmath146 but remain a plane wave along @xmath52 .
since the laser beam @xmath149 ( @xmath150 ) is obtained by reflecting the beam @xmath151 ( @xmath152 ) by mirrors , they possess the same frequency @xmath153 ( @xmath154 ) , and no phase locking is required @xcite .
the laser beam @xmath152 with a different frequency @xmath155 is generated by applying an acoustic - optical modulator on a beam split from the first laser beam .
each pair of raman laser beams couple two hyperfine states independently , leading to the spin - orbit coupling term @xmath18 with @xmath156 .
moreover , owing to the stark effects , these lasers also create optical lattices along the @xmath126 and @xmath146 directions : @xmath157 and @xmath158 with @xmath159 .
the zeeman field @xmath16 is generated by the two - photon detuning @xmath160 through @xmath161 as shown in fig .
[ fig3 ] . in comparison
, a similar laser configuration can also realize the simplified spin - orbit coupling scheme that we mentioned before . in this case
, we further simplify the laser configuration [ as shown in fig .
[ fig3](c ) ] so that the rabi frequencies take the form @xmath162 and @xmath163 , corresponding to plane waves for the second set of laser beams . )
[ ( b ) ] and its simplified one [ ( c ) ] .
@xmath164 ( @xmath152 ) possess the same frequency as @xmath165 ( @xmath166 ) , which is obtained by reflecting the laser beam @xmath167 ( @xmath168 ) by mirrors .
the magnetic field is along the @xmath146 direction , and @xmath169 is the two - photon detuning .
the double arrows denote the orientation of linear polarization of laser beams.,width=326 ] we may choose either @xmath144k ( fermions ) or @xmath145rb ( bosons ) atoms for observation of weyl points in experiments . here
we take @xmath144k as an example . with a blue - detuned laser beam at wavelength @xmath170 nm ( corresponding to @xmath171 thz ) @xcite , the recoil energy @xmath172 khz .
if we choose a geometry with @xmath173 and @xmath174 with @xmath175 , @xmath176 ghz , @xmath177 , we have @xmath178 and @xmath179 . with these experimental parameters ,
the tight - binding parameters are @xmath180 , and @xmath181 .
@xmath160 can be readily tuned from zero and when @xmath160 crosses @xmath182 , we will observe a lifshitz - type quantum phase transition from type - i to type - ii weyl points . to detect the weyl points of fermionic atoms and their lifshitz phase transition , one can measure their linear spectra along all three momenta directions by momentum resolved radio - frequency spectroscopy , which has been utilized for observation of a 2d dirac cone in spin - orbit - coupled atomic gases @xcite .
the lifshitz transition is directly reflected by a sharp change of the dispersion of particle or hole bands along @xmath131 near a weyl point so that the slopes of their spectra have the same sign . for bosonic atoms such as @xmath145rb , although the fermi surface does not exist , there is still the band structure with a touching point .
one may consider driving a bec across a weyl point by a constant force @xmath183 and measuring the landau - zener tunneling probability esslinger2012nature , lim2012prl , law2015,xiongjun2016 , which is @xmath184 with @xmath185 being the gap between the considered particle and hole bands and @xmath186 being the velocity of the bec @xcite .
therefore , the gap closing at a weyl point is signalled by a peak of the landau - zener tunneling probability .
when the bec bypasses a type - i ( type - ii ) weyl point , a finite fraction of atoms remains in the hole band and these atoms move in the opposite ( same ) direction along @xmath52 compared with those tunneling into a higher band .
this different behavior can be utilized to measure the lifshitz transition . in summary
, we have proposed a scheme well based on the current experimental technology to realize both type - i and type - ii weyl points in the single - particle spectra with cold atoms in an optical lattice .
the proposed system offers a unique opportunity to observe and study the topological lifshitz - type quantum phase transition from type - i to type - ii weyl points by continuously tuning one of the experimental knobs .
we thank s. wang for helpful discussions .
this work was supported by the arl , the iarpa logiq program , and the afosr muri program . |
over the last few years , we have witnessed renewed interest in the spectral line polarization produced by the presence of population imbalances and quantum interferences ( or coherences ) among the magnetic sublevels of atomic energy levels ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
this so - called atomic level polarization is typically produced by anisotropic radiation pumping processes , which are particularly efficient in the outer layers of stellar atmospheres , where the depolarizing effect of isotropic collisions tends to be negligible .
particular attention has been given to coherences between pairs of magnetic sublevels pertaining to each particular @xmath0-level of the atomic model under consideration ( @xmath0 being the level s total angular momentum ) .
the sensitivity of these coherences to the hanle effect produces changes in the linear polarization of the emergent spectral line radiation , which can be exploited for detecting magnetic fields that are too weak and/or too tangled so as to produce measurable zeeman polarization signals . in general , quantum interferences
are also present between pairs of magnetic sublevels pertaining to different @xmath0-levels .
it is known that this kind of coherences may produce observable effects in the wings of some spectral lines .
perhaps the most illustrative solar example is the @xmath9 pattern observed by @xcite across the ca ii h and k lines ( see also * ? ? ?
* ) , which shows a positive signal at the core of the k line ( @xmath10 and @xmath11 ) , a sign reversal between the two lines , and zero polarization at the center of the h line ( @xmath10 and @xmath12 ) . that peculiar @xmath9 pattern could be explained by @xcite using a simple theoretical model that accounts for the possibility of quantum interferences between the two upper @xmath0-levels of the ca ii resonance lines .
a rigorous theoretical framework for describing the spectral line polarization produced by radiatively induced population imbalances and quantum coherences , in the presence of arbitrary magnetic fields , is the density - matrix theory described in the monograph `` polarization in spectral lines '' ( landi deglinnocenti & landolfi 2004 , hereafter ll04 ) .
this theory is based on the hypothesis that the pumping radiation field has no spectral structure over frequency intervals larger than the frequency separation between the interfering levels ( flat - spectrum approximation ) .
because of this approximation , the theory is very suitable for treating spectral lines that can be described under the hypothesis of complete frequency redistribution ( crd ) , while it can not account for the effects of partial redistribution in frequency ( prd ) . despite of this limitation
, this theory represents the most robust quantum approach to the physics of polarization developed so far .
in particular , it accounts for the role of quantum interferences in a very general , self - consistent way ( e.g. , the review by * ? ? ?
as a matter of fact , although the ca ii h and k lines are 35 apart ( the flat - spectrum approximation thus appearing rather restrictive ) , the theory explains very well the sign reversal observed in the blue wing of the h line in terms of coherences between the two upper levels of these lines ( see ll04 ) .
theoretical approaches aimed at including prd effects in the presence of @xmath0-state interferences can be found in @xcite and @xcite .
it is known that interferences between pairs of magnetic sublevels pertaining to different @xmath0-levels are the smaller the larger the energy separation between them .
when this separation is large , their effects are negligible in the core of the lines , becoming more important in the far wings ( see * ? ? ?
* ll04 ) . on the other hand ,
far from line center the line emissivity is very low , and the presence of the continuum generally masks the effects of such interferences . because of these basic arguments , interferences between different @xmath0-levels are usually neglected when investigating the scattering polarization properties of many spectral lines , with a considerable simplification of the problem .
however , this kind of qualitative considerations should not be applied as rigorous rules , as testified by the ca ii h and k lines and by several other signals of the linearly - polarized solar limb spectrum ( or second solar spectrum ) which show the signatures of this kind of interferences ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
as far as the solar ca ii h and k lines are considered , it is now clear that the reason why quantum interferences between such significantly separated lines produce observable effects is because these strong lines have very extended wings , so that the continuum is not immediately reached when moving away from line center , and because the large optical depth of the solar atmosphere compensates for the low line emissivity at the wavelengths where that observational signature appears ( cf .
* ll04 ) . in this paper , we present a systematic theoretical investigation of the role of quantum interferences between different @xmath0-levels
. we aim at clarifying their observable effects in various interesting multiplets and at providing a series of general criteria for establishing under which circumstances their effects are expected to be observable , and when , on the contrary , the modeling of spectropolarimetric observations can be safely carried out ignoring their contribution .
the investigation is carried out within the framework of the above - mentioned quantum theory of polarization .
after briefly introducing the density - matrix formalism ( section 2 ) , and presenting the scattering polarization model that is applied for our investigation ( section 3 ) , we start focusing the attention on the @xmath2 multiplet , the simplest one where interferences between different @xmath0-levels occur ( section 4 ) . we show that although the large energy separation generally present between different @xmath0-levels makes this kind of interferences usually very small , as far as fractional polarization signals ( i.e. , ratios , such as @xmath9 ) are considered , their effects are not necessarily negligible .
indeed , when only line processes are considered ( no continuum ) , the fractional polarization patterns obtained taking into account and neglecting these interferences coincide within a small spectral interval around the center of the lines , but are found to be very different at all other wavelengths , even when the @xmath0-levels are very separated from each other ( see section 4.2 ) .
particular attention is given to the analysis of the effects of the doppler broadening of the lines , a physical aspect that is found to play an important role for establishing where and when such interferences can be neglected .
as previously mentioned , what makes the effects of interferences between different @xmath0-levels vanish in the observed polarization patterns is the continuum , which starts dominating over the line emissivity moving from the line center to the wings . in section 4.3 ,
we show that the amount of continuum needed to mask the signatures of such interferences is the larger the smaller the separation between the @xmath0-levels .
the effects of interferences are thus expected to be observable either when the interfering lines have very extended wings , so that the continuum is not immediately reached between them ( as in the case of the ca ii h and k lines ) , or when the separation between the @xmath0-levels is sufficiently small but , as we will see , still large when compared to the doppler width of the corresponding spectral lines . in the first part of the paper , we analyze the scattering polarization profiles of hypothetical multiplets characterized by different values of the energy separation between the interfering @xmath0-levels , of the doppler width , and of the continuum intensity .
the results of this analysis , and their consequences for practical applications , are then investigated in detail on various multiplets of particular interest for the diagnostics of the magnetism of the solar atmosphere ( sections 4.5 , 5 , 6 , and 7 ) .
the effects due to the presence of magnetic fields of various intensities and configurations are also investigated ( section 8) .
a convenient way to describe the populations of the various magnetic sublevels of an atomic system , as well as the quantum interferences between pairs of them , is through the matrix elements of the density operator .
the density operator is a very useful theoretical tool for describing any physical system which is in a statistical mixture of states ( e.g. , * ? ? ?
if @xmath13 , @xmath14 , , @xmath15 , , are the probabilities for a given physical system of being in the dynamical states represented by the vectors @xmath16 , @xmath17 , , @xmath18 , , respectively , the corresponding density operator is defined by @xmath19 as any other quantum operator , also the density operator is completely specified once its matrix elements , evaluated on a given basis of the hilbert space associated with the physical system , are known .
the matrix elements of the density operator contain all the accessible information about the system . on the basis of the atomic energy eigenvectors @xmath20 ,
the matrix elements of the density operator are given by @xmath21 the diagonal elements represent the populations of the various energy levels , while the off - diagonal elements represent the quantum interferences between pairs of them .
on this basis , the statistical equilibrium equations ( sees ) for the atomic density matrix are given by ( see equation ( 6.62 ) of ll04 ) @xmath22 \nonumber \\ & & - \sum_{m^{\prime \prime } } \big [ \rho_{mm^{\prime \prime } } \,r_{e } ( m^{\prime \prime},m , m^{\prime } ) + \rho_{m^{\prime \prime}m^{\prime } } \,r_{e } ( m , m^{\prime},m^{\prime \prime } ) \big ] \nonumber \\ & & - \sum_{m^{\prime \prime } } \big [ \rho_{mm^{\prime \prime } } \,r_{s } ( m^{\prime \prime},m , m^{\prime } ) + \rho_{m^{\prime \prime}m^{\prime } } \,r_{s } ( m , m^{\prime},m^{\prime \prime } ) \big ] \;\ ; .\end{aligned}\ ] ] these equations describe the transfer ( @xmath23 rates ) and relaxation ( @xmath24 rates ) of populations and coherences due to absorption ( index @xmath25 ) , spontaneous emission ( index @xmath26 ) , and stimulated emission ( index @xmath27 ) processes .
they also describe the effects due to the presence of a magnetic field . a complete derivation of these equations , as well as the explicit expression of the various rates can be found in ll04 . here
we focus our attention on the first term in the right - hand side of equations ( [ eq : see ] ) . this term , proportional to the bohr frequency @xmath28 , with @xmath29 the energy of level @xmath18 and @xmath30 the planck constant , is zero both for populations
( i.e. , for the diagonal elements @xmath31 ) and for coherences between degenerate levels , while it produces a relaxation of coherences between pairs of non - degenerate levels .
the coherence @xmath32 is thus the smaller , the larger the energy separation between the levels @xmath33 and @xmath34 . modifying the frequency separations @xmath35 among the various magnetic sublevels , a magnetic field modulates , through this term , the corresponding coherences @xmath32 , and consequently the polarization of the emitted radiation .
this is the basic physical mechanism at the origin of the so - called hanle effect . finally , it is important to recall that equations ( [ eq : see ] ) are valid under the so - called flat - spectrum approximation .
this approximation requires the pumping radiation field to be spectrally flat over frequency intervals larger than the bohr frequency relating pairs of levels between which quantum interferences are considered , and larger than the inverse lifetime of the same levels . for an atomic system devoid of hyperfine structure ( hfs ) ,
the matrix elements of the density operator are often defined on the basis of the eigenvectors of the angular momentum , with @xmath0 and @xmath36 the quantum numbers associated with the total angular momentum and to its projection along the quantization axis , and @xmath8 a set of inner quantum numbers associated with the atomic hamiltonian coupling scheme , @xmath8 could represent , for example , the set of quantum numbers ( @xmath37 , @xmath38 , @xmath27 ) which describe the electronic configuration , the total orbital angular momentum , and the total electronic spin . ] . on this basis , the matrix elements of the density operator are @xmath39 the diagonal elements represent the populations of the various magnetic sublevels , while the off - diagonal elements the quantum interferences between pairs of them .
an atomic model accounting for coherences between pairs of magnetic sublevels pertaining either to the same @xmath0-level or to different @xmath0-levels within the same term is generally referred to as `` multi - term atom '' ( see section 7.5 of ll04 ) . in this case
, the flat - spectrum approximation requires the radiation field incident on the atom to be flat over frequency intervals larger than the frequency separation among the @xmath0-levels belonging to each term .
this implies that the incident radiation field must be flat across the frequency interval covered by the transitions of a given multiplet .
when coherences between different @xmath0-levels are neglected , the corresponding atomic model is generally referred to as `` multi - level atom '' ( see section 7.1 of ll04 ) . in this case , the flat - spectrum approximation is less restrictive , since it requires the radiation field to be constant across frequency intervals larger than the zeeman splitting among the various magnetic sublevels pertaining to each @xmath0-level , and larger than their inverse lifetimes . in a multi - level atom , the incident radiation field can thus vary across the frequency interval spanned by the transitions of a given multiplet .
the sees equations and the expressions of the radiative transfer coefficients for a multi - term and a multi - level atom can be found in chapter 7 of ll04 .
in order to analyze the effects of interferences between different @xmath0-levels , we consider an atomic system composed of two terms ( each being characterized by the total orbital angular momentum @xmath38 , by the total electronic spin @xmath27 , and by given fine - structure ( fs ) splittings of the various @xmath0-levels ) , and we compare the polarization patterns of the radiation emitted across the transitions of the corresponding multiplet as obtained both taking into account and neglecting such coherences ( i.e. , as calculated within the framework of a two - term atomic model , which accounts for such interferences , and within the framework of the corresponding multi - level atomic model , which neglects them ) .
although in the multi - level atom case the pumping radiation field can be different at the frequencies of the various transitions of the multiplet , in order to analyze the net effect of the coherences between different @xmath0-levels , we consider the same incident field , flat across the whole frequency interval of the multiplet , both in the case of the two - term atom , and in the case of the corresponding multi - level atom .
interferences between different @xmath0-levels have a double role : on one hand they enter the sees and therefore , in principle , their presence may modify populations and interferences within the same @xmath0-level with respect to the case in which they are neglected ; on the other hand they contribute ( exactly as populations and interferences within the same @xmath0-level do ) to the radiative transfer coefficients .
we focus our attention on the radiation scattered at @xmath40 by a plane - parallel slab of plasma illuminated by the solar continuum radiation field . in this simple scenario ,
the fractional polarization of the scattered radiation can be calculated through the approximate formula ( see * ? ? ?
* ) @xmath41 where @xmath42 and @xmath43 ( @xmath44 ) are the line emission and absorption coefficients , respectively , in the four stokes parameters , while @xmath45 and @xmath46 are the continuum intensity emission and absorption coefficients , respectively .
the incident continuum radiation is assumed to have axial symmetry around the normal to the slab ( the local vertical ) , to be unpolarized , and to be flat over the frequency intervals covered by the multiplets that will be considered in this investigation . the reference direction for positive @xmath47
is assumed perpendicular to the scattering plane .
the doppler width used in the calculation of the line emission and absorption coefficients is derived from the temperature and microturbulent velocity of a solar model atmosphere at the height where the line - center optical depth is unity , for an observation at @xmath48 .
the semi - empirical atmospheric model @xmath49 of @xcite has been used .
the first step is to write down and solve the sees for the given incident ( pumping ) continuum radiation field .
we describe the incident radiation through the radiation field tensor @xmath50 ( see equation ( 5.157 ) of ll04 for its definition ) , taking the quantization axis along the local vertical .
due to the cylindrical symmetry of the incident continuum radiation around this direction , only two components of the radiation field tensor are non - zero : @xmath51 and @xmath52 .
the former describes the average intensity of the radiation field over all the directions of propagation , while the latter gives a measure of the anisotropy of the radiation field . under such circumstances
, it is customary to describe the radiation field through two equivalent dimensionless quantities , the mean number of photons per mode ( @xmath53 ) and the anisotropy factor ( @xmath54 ) , defined by ( ll04 ) @xmath55 in the absence of magnetic fields and of other mechanisms able to break the cylindrical symmetry of the problem , if the quantization axis is taken along the symmetry axis ( as we are assuming here ) , the only interferences that can be excited are those between pairs of magnetic sublevels pertaining to different @xmath0-levels and characterized by the same value of the magnetic quantum number @xmath36 .
these are exactly the interferences that we are going to investigate in the following sections .
once the sees are solved , and the density - matrix elements are known , we calculate the line emission and absorption coefficients appearing in equation ( [ eq : emer - pol ] ) , according to the expressions given in ll04 ( chapter 7 )
. the continuum emission coefficient @xmath45 will be a free parameter of the problem , used to analyze the effect of the continuum in masking the observational signatures of the coherences under investigation .
the value of the continuum absorption coefficient is calculated here through the equation @xmath56 where @xmath57 is the planck function at the wavelength of the transition under investigation and at the temperature chosen for the calculation of the doppler width of each line , as previously discussed .
we start our investigation considering the @xmath2 multiplet , the simplest one where interferences between different @xmath0-levels occur , and one of the most interesting given the large number of strong spectral lines belonging to this multiplet that are observed on the sun ( ca ii h and k , mg ii h and k , na i d@xmath5 and d@xmath6 , ly@xmath8 , etc . ) .
this multiplet consists of the following two transitions : @xmath58 ( transition of h or d@xmath5 type , in the following referred to as transition 1 ) and @xmath59 ( transition of k or d@xmath6 type , in the following referred to as transition 2 ) .
we want to investigate the effects of the quantum interferences between the upper levels @xmath60 and @xmath61 .
since the common lower level of the transitions of this multiplet , having @xmath62 , can not be polarized by the unpolarized incident radiation field , the sees considerably simplify with respect to the general case .
moreover , @xmath63 so that the second term in the right - hand side of equation ( [ eq : emer - pol ] ) , which describes the effects of dichroism , does not bring any contribution to the emergent polarization .
if stimulation effects are neglected ( which is a good approximation in the solar atmosphere ) , and no magnetic fields are considered , it is possible to find an analytical solution of the sees , and rather simple analytical expressions for the ratio @xmath64 , both in the case in which the interferences between the two upper @xmath0-levels are taken into account , and in the case in which they are neglected ( see equations ( [ eq : eps3lp ] ) and ( [ eq : eps2tp ] ) in the appendix ) .
the @xmath65 profiles obtained taking into account ( solid line ) and neglecting ( dashed line ) interferences between the upper @xmath0-levels , assuming @xmath66 and a wavelength separation between the two components @xmath67 , with @xmath68 the doppler width of the two lines , are plotted in panel @xmath69 of figure [ fig : pq - gen ] . as seen in the figure ,
the two profiles coincide in the core of the two lines , while they are very different at all the other wavelengths ( cf .
the most remarkable difference is the sign reversal between the two lines shown by the @xmath65 profile calculated taking into account interferences .
this particular signature of the quantum interferences between the two upper @xmath0-levels has been clearly observed between the h and k lines of ca ii ( see * ? ? ?
the negligible contribution of interferences between different @xmath0-levels in the center of the single lines of an arbitrary multiplet has already been discussed in section 10.17 of ll04 , where an analytical expression for the value of @xmath65 in the core of the lines is derived ( cf .
equation ( [ eq : pq-2tcore ] ) ) . as discussed in ll04 , under the hypotheses previously introduced , the asymptotic value of @xmath65 for an arbitrary two - term atom is equal to the ( constant ) @xmath65 value of a two - level atom with @xmath70 and @xmath71 .
the analytical expression of @xmath65 for a two - level atom can be found in the appendix ( equation ( [ eq : pq-2l ] ) ) . as shown by equations ( [ eq : eps3lp ] ) and ( [ eq : eps2tp ] ) ,
the quantity @xmath65 , as calculated both taking into account and neglecting interferences between different @xmath0-levels , depends on the value of @xmath54 but not on the value of @xmath53 .
, as implicit in equations ( [ eq : eps3lp ] ) and ( [ eq : eps2tp ] ) which are derived neglecting stimulation effects . ] concerning the dependence on the anisotropy factor , it should be observed that for small values of @xmath54 the second term in the denominator of equations ( [ eq : eps3lp ] ) and ( [ eq : eps2tp ] ) can be neglected with respect to the first one , so that @xmath54 represents just a scaling factor of the whole profiles .
this property can be clearly appreciated from figure [ fig : w - dep ] , where the absolute value of the ratio between the values of @xmath65 ( calculated taking into account interferences ) at line center of transition 2 and at the wavelength position of the ( negative ) minimum , is plotted as a function of the anisotropy factor .
the value of @xmath65 at line center of transition 2 is given by equation ( [ eq : pq-2tcore ] ) , while it is not possible to find a simple analytical expression for the amplitude of the negative dip . as it can be observed from the figure , the above - mentioned ratio is constant and very close to unity for values of @xmath54 smaller than 0.1 . for these values , which are those that are commonly encountered in the solar atmosphere ,
the anisotropy factor is just a scaling factor of the whole @xmath65 profile .
numerical calculations show that the position of the minimum does not depend on the particular value of @xmath54 .
if not explicitly stated , all the results presented throughout this investigation are obtained assuming @xmath72 and @xmath73 . as shown in panels @xmath7 and @xmath74 of figure [ fig : pq - gen ] , at the wavelength positions of the two transitions , the @xmath65 profiles obtained taking into account and neglecting interferences coincide and are constant over a wavelength interval of about five doppler widths . the mathematical and physical origin of these `` plateaux '' is discussed in the appendix , where it is shown that their boundaries are determined by the wavelength positions at which the behavior of the corresponding voigt profiles changes from gaussian to lorentzian .
since the extension of the plateaux does not show significant variations with the wavelength separation between the two components of the multiplet , their presence only marginally affects the @xmath65 profile when the separation between the two components is much larger than the doppler width of the two lines ( see panel @xmath69 of figures [ fig : pq - gen ] and [ fig : dl - dep ] ) , while it starts modifying the overall shape of the @xmath65 profiles when the distance between the two components is of the same order of magnitude as the doppler width of the lines ( see panel @xmath7 of figure [ fig : dl - dep ] ) . when the wavelength separation between the two components is equal or smaller than the extension of the two plateaux , these start merging , so that between the two lines all the signatures of the interferences between the upper @xmath0-levels ( such as the negativity previously discussed ) disappear ( see panels @xmath74 and @xmath75 of figure [ fig : dl - dep ] ) . since the width of each plateau is about five doppler widths , it is important to observe that their merging , and therefore the disappearance of the effects of the interferences , starts when the intensity profiles of the two lines are still separated from each other ( see panel @xmath74 of figure [ fig : dl - dep ] ) by an optically thin slab . ] .
the effects of interferences become appreciable in the line - core when the separation between the two components is of the same order of magnitude as the width of the two interfering @xmath0-levels . indeed , in the limiting case in which the separation between the two levels goes to zero , the effects ( and in particular the depolarizing effect ) of the fine structure have to disappear , and the polarization pattern of a two - level atom with @xmath70 and @xmath71 must be recovered ( because of the principle of spectroscopic stability ) .
this effect is shown in panels @xmath76 and @xmath77 of figure [ fig : dl - dep ] , where a frequency separation between the two components of the same order of magnitude as the einstein coefficient for spontaneous emission has been considered .
it must be observed that the fs splitting of the various @xmath0-levels is always much larger ( also in the case of hydrogen ) than the natural width of the levels .
for example , in the case of the ly@xmath8 line , the separation between the two fs components is of the order of 10@xmath78 s@xmath79 , while the einstein coefficient for spontaneous emission is equal to .
however , we have to remind that collisions ( here neglected ) may have an important role on these phenomena ( both for their broadening and depolarizing effect ) .
another interesting result to be pointed out is that when the separation between the two components is smaller than the doppler width of the two lines , so that the two plateaux completely merge , the fs depolarizing effect takes place on a wavelength interval of about five doppler widths ( the width of a single plateau ) , independently of the actual separation between the two components .
we now analyze how the fractional polarization profiles previously obtained taking into account only line processes are modified when the contribution of an unpolarized continuum is added according to equation ( [ eq : emer - pol ] ) .
as shown by the various panels of figure [ fig : cont - dep ] , the main effect of the continuum is to make the fractional polarization vanish as one moves from the core of the lines toward the wings .
although this overall effect makes the patterns obtained taking into account and neglecting interferences coincide , if the continuum is not too strong , clear signatures of the interferences between different @xmath0-levels can still be noticed .
let us consider first the case in which the wavelength separation between the two transitions is much larger than the doppler width ( figure [ fig : cont - dep ] , panels @xmath69 and @xmath7 ) .
as far as the profile calculated taking into account interferences is concerned , we observe that if the continuum is not sufficiently strong , the maximum value of the polarization is not found at the center of transition 2 , but at slightly shorter wavelengths .
this circumstance can be seen in the observations of the ca ii h and k system ( see * ? ? ?
* ; * ? ? ?
the amount of continuum needed in order to make the maximum signal falling at the center of transition 2 is the larger the smaller the wavelength separation between the two components .
the presence of the continuum reduces the negativity of the @xmath9 pattern between the two transitions , it moves the negative minimum toward the wavelength position of transition 1 , but it does not affect the wavelength position of the sign reversals . moreover , it produces a positive peak in the red wing of transition 1 , so that an antisymmetrical profile across this component appears .
this latter effect , due to interferences between the different @xmath0-levels of the upper term , can be clearly observed across the na i d@xmath5 line ( see * ? ? ?
as the value of the continuum is increased , the amplitude of this antisymmetrical pattern is decreased , and the overall profile becomes more similar to the one obtained neglecting interferences .
the amount of continuum needed for this signature to be cancelled out is the larger the smaller the wavelength separation between the two components ( this explains why this structure can be observed across the d@xmath5 line of na i , but not across the corresponding ba ii line at 4934 ) .
we consider now the opposite case in which the various components are very close to each other with respect to the doppler width of the single lines ( figure [ fig : cont - dep ] , panels @xmath74 and @xmath75 ) . in this case , as previously discussed , interferences between different @xmath0-levels do not produce any observable effect across a spectral interval of about five doppler widths centered at the wavelength of the two transitions , irrespectively of the continuum value . just outside this interval ,
on the other hand , the @xmath9 profile obtained taking into account interferences shows , if the continuum is not too strong , two peaks .
these @xmath9 peaks disappear as the value of the continuum is increased .
it should be emphasized that the physical origin of this two - peak structure , which is obtained only when interferences are taken into account , lies in the way the continuum is included in the slab model we are considering here .
this does not mean that it is an artifact , but it is clear that its reliability is intimately related to the suitability of the model that we are using .
we finally observe that for the same value of the continuum , the @xmath9 profiles obtained taking into account interferences always show more extended wings than the corresponding profiles obtained neglecting them .
all the results shown in this investigation are based on the slab model described in section [ sect : model ] , which allows us to study the polarization properties of the scattered radiation through the approximate analytical expression of equation ( [ eq : emer - pol ] ) .
this allows us to analyze in great detail the atomic aspects of the problem , without introducing the complications due to radiative transfer processes in stratified model atmospheres ) describes the possibility of dichroism ( i.e. , selective absorption of polarization components ) , which is absent in optically thin media . ] .
the analysis of the effects of radiative transfer on the polarization signatures produced by interferences between different @xmath0-levels is beyond the scope of this paper .
nevertheless , we can make some brief , qualitative considerations on this topic .
the main consequence of radiative transfer is that , depending on its frequency across the spectral line profile , the scattered radiation comes from different atmospheric heights , characterized by different values of the anisotropy factor of the local radiation field .
the fractional polarization patterns previously obtained using a single value of the anisotropy factor will thus be modified by radiative transfer effects , through a modulation on the different values of @xmath54 .
such modifications will take place mainly between the center and the wings of the single spectral lines .
very likely , when radiative transfer effects in solar atmospheric models are considered , the amplitude and shape of the @xmath9 profile around the line center will be different from those shown in the plots of figure [ fig : cont - dep ] .
for example , we believe that the flat plateaux previously discussed will not be obtained , or will be different . on the other hand , the results of the analysis of the spectral intervals where the effects of interferences are negligible , and where the fs depolarization takes place , should remain valid , as well as the qualitative shapes of the signatures of interferences , as far as these appear in the far wings of the lines , where assuming a constant value of @xmath54 is quite a good approximation . finally , we emphasize that , despite of its simplicity , the slab model of section 3 has already been applied with success for interpreting several peculiarities observed in scattering polarization signals ( e.g. , * ? ? ?
* ) , thus showing that it represents a suitable approximation , at least for the modeling and understanding of the physical mechanisms producing polarization in the solar atmosphere .
we consider now various @xmath2 multiplets of particular interest . if not explicitly stated , we keep assuming @xmath72 and @xmath73 .
figure [ fig : bana ] shows the fractional linear polarization patterns obtained across the d@xmath5 and d@xmath6 lines of ba ii- and d@xmath6-type lines ( e.g. , * ? ? ?
* ; * ? ? ?
* ) due to the similarities of the @xmath9 scattering polarization signals that they produce at the limb with the corresponding signals produced by the d@xmath5 and d@xmath6 lines of na i. ] , the h and k lines of ca ii , the h and k lines of mg ii , and the d@xmath5 and d@xmath6 lines of na i , taking into account the interferences between the two upper @xmath0-levels .
although some of these elements ( ba , mg and na ) have isotopes with non - zero nuclear spin , hfs has been neglected in this investigation .
for all these ions , the wavelength separation between the two lines of the multiplet is much larger than their doppler width , so that , depending on the continuum emissivity , patterns similar to those shown in panels @xmath69 and @xmath7 of figure [ fig : cont - dep ] are obtained . comparing the profiles of the various ions plotted in figure [ fig : bana ] ,
it can be clearly observed how the amount of continuum needed for masking the signatures of interferences is the larger the smaller the separation between the two lines .
for example , in the case of barium , a continuum @xmath80 is already sufficient to completely cancel out the sign reversal between the two lines , and the ensuing antisymmetric pattern across the d@xmath5 line .
these signatures , on the other hand , are still observable in calcium , magnesium and sodium , if the same continuum is considered . from figure
[ fig : bana ] , it can also be observed that for the wavelength separation of the magnesium ( and sodium ) lines , the amount of continuum needed in order to obtain a given ratio between the amplitude of the negative minimum ( in absolute value ) and the amplitude of the polarization peak of the k ( or d@xmath6 ) line is more than 10 times stronger than in the case of the calcium lines . in agreement with these considerations ,
the above - mentioned signatures of interferences between different @xmath0-levels can be clearly observed in calcium and sodium , but not in barium .
unfortunately , there are no high sensitivity observations available as far as the h and k lines of mg ii are concerned .
the fact that these two lines are very close to each other ( their separation is similar to that between the sodium d - lines ) , and with very extended wings ( so that , as in the case of calcium , the continuum is not completely reached between them ) , suggests that these signatures should be observable also across these lines . on the other hand , at the wavelength position of these lines the continuum is very strong , and has a very high degree of linear polarization , whose effect should be carefully taken into account .
we note that no sign reversal was found in the pioneering observations of these @xmath9 signals performed by @xcite using the ultraviolet spectrometer and polarimeter on board of the _ solar maximum mission _ satellite , which however was not designed for detecting weak scattering polarization signals . as a last example of @xmath2 multiplet
, we consider the h i ly@xmath81 line ( see figure [ fig : ly - alpha ] ) . since the separation between the two components is in this case much smaller than the doppler width of the line , the profiles that are obtained for different values of the continuum are very similar to those shown in panels @xmath74 and @xmath75 of figure [ fig : cont - dep ] .
as previously pointed out , in this case interferences do not produce any effect in the core of the line , across a wavelength interval of about five doppler widths .
it can be noticed that this is the wavelength interval over which the depolarizing effect of fine structure takes place .
signatures of interferences between the upper @xmath0-levels , consisting of two peaks appearing in the wings of the line , might be produced if the continuum is sufficiently weak ( of the order of 10@xmath82 or smaller ) .
however , we already pointed out that these features in the calculated @xmath9 profile might be due to the simple slab model that we are using and that might not be obtained once radiative transfer effects are properly taken into account .
the analysis carried out in this section on particular multiplets , characterized by different values of the wavelength separation between the two components , and of the doppler widths , provides complete information on the role of interferences on @xmath2 doublets .
the extension of these results to other interesting doublets of this kind should not present particular difficulties .
in this section , we investigate the effects of interferences between different @xmath0-levels on the scattering polarization patterns produced across @xmath3 multiplets .
the possibility for the ( common ) lower level of the transitions of this multiplet to carry atomic polarization ( @xmath83 ) allows us to investigate the effects of dichroism on the polarization signatures produced by interferences .
the analysis will be carried out on the cr i triplet at 5207 and on the o i triplet at 7773 .
the cr i inverted - levels increases with the value of @xmath0 , while it is said to be inverted in the opposite case . ]
triplet at 5207 is composed by the following transitions : @xmath84 ( line 1 at 5205.50 ) , @xmath85 ( line 2 at 5206.04 ) , and @xmath86 ( line 3 at 5208.42 ) .
the hfs of the only stable isotope of chromium with non - zero nuclear spin ( @xmath87cr , abundance 9.5% ) is neglected in the present investigation . as in the previous sections , we consider the radiation scattered at 90@xmath1 by a slab of solar atmospheric plasma illuminated by the solar continuum radiation field ( @xmath72 , @xmath73 ) . as it can be observed in panel @xmath69 of figure [ fig : cr ] , the @xmath65 profile obtained taking into account interferences between different @xmath0-levels shows in this multiplet two sign reversals , one between lines 1 and 2 , and a larger one between lines 2 and 3 .
the presence of a small plateau can be observed in the core of the three lines ; here , as in the case of the @xmath2 multiplet , interferences between different @xmath0-levels do not produce any observable effect .
note that the profiles of panel @xmath69 of figure [ fig : cr ] have been obtained neglecting lower - level polarization and stimulation effects : the value of @xmath65 in the core of the lines is thus given by equation ( [ eq : pq-2tcore ] ) .
the effect of the continuum in masking the signatures of interferences is similar to the case of the @xmath2 multiplet .
as shown in panel @xmath7 of figure [ fig : cr ] , the continuum makes the polarization vanish in the far blue wing of line 1 and in the far red wing of line 3 , it reduces the amplitude of the negative patterns between the lines , and it produces an antisymmetrical profile across transition 1 ( @xmath88 ) . this last polarization signature , as well as a small `` sign reversal '' ( with respect to the continuum polarization level ) between lines 2 and 3 , can be observed in gandorfer s ( 2000 ) atlas of the second solar spectrum
. the lower level of this triplet , having @xmath83 , can carry atomic polarization .
the first thing that has to be pointed out is the appreciable feedback that the presence of lower - level polarization has on the atomic polarization of the upper levels .
as shown in panels @xmath74 and @xmath75 of figure [ fig : cr ] , the value of @xmath9 ( as calculated still neglecting the contribution of dichroism , given by the second term in the right - hand side of equation ( [ eq : emer - pol ] ) ) is modified by the presence of lower - level polarization in the core of the three lines ( it is increased in lines 1 and 3 , and decreased in line 2 ) .
as far as the profiles obtained taking into account interferences between the upper @xmath0-levels are concerned ( see panel @xmath74 ) , we observe that lower - level polarization slightly modifies the negative pattern between lines 2 and 3 , as well as the antisymmetrical profile across line 1 . interferences between different @xmath0-levels do not produce any effect in the core of the lines also when lower - level polarization is taken into account . when lower - level polarization is taken into account , @xmath89 is non - zero , and also the second term in the right - hand side of equation ( [ eq : emer - pol ] ) ( i.e. , dichroism ) brings a contribution to the polarization of the emergent radiation .
the effects of dichroism on the scattering polarization pattern of this multiplet can be clearly observed in figure [ fig : cr - dichr ] . as shown in panels @xmath69 and @xmath7 of this figure , the presence of dichroism modifies the value of @xmath9 in the core of the three lines ( note in particular that the polarization in the core of line 1 becomes negative ) , as well as the signatures of interferences between lines 2 and 3 ( the wavelength positions of the sign reversals are changed ) , and across line 1 ( the whole polarization pattern across this line is negative when dichroism is included , both when interferences are taken into account and when they are neglected ) . in panels @xmath69 and @xmath7 of figure [ fig : cr - dichr ] , the value of @xmath90 has been calculated according to equation ( [ eq : etaic ] ) , as explained in section [ sect : model ] .
however , if we consider this physical quantity as a free parameter , we observe that the signatures of interferences between different @xmath0-levels are sensitive to its value .
this can be observed in panels @xmath74 and @xmath75 of figure [ fig : cr - dichr ] . if the value of @xmath91 is increased ( with respect to the one calculated according to equation [ eq : etaic ] )
, the dip produced by interferences in the red wing of line 1 ( between lines 1 and 2 ) , which is still observable in panels @xmath69 and @xmath7 , gradually decreases until disappearing , while a positive bump appears in the blue wing of this line . a profile with a sign reversal ( qualitatively similar to the one observed by * ?
* ) is thus recovered in line 1 . however
, contrary to the profiles plotted in panel @xmath74 of figure [ fig : cr ] , the negative minimum is now in the core of the line and not in its red wing .
indeed , the underlying physics is now the following : the positive peak in the blue wing is still due to interferences between different @xmath0-levels , while the negative minimum in the core of line 1 is produced by dichroism ( see panel @xmath75 of figure [ fig : cr - dichr ] ) . the o i triplet at 7773 is composed by the following transitions @xmath84 ( line 1 at 7775.39 ) , @xmath85 ( line 2 at 7774.16 ) , and @xmath86 ( line 3 at 7771.94 ) . from an atomic point of view
, this multiplet is very similar to the cr i triplet at 5207 previously investigated . in particular ,
the fs splitting of the upper term , and therefore the wavelength separations among the various lines , is very similar in the two multiplets .
the only difference is that this is a regular triplet ( i.e. , not inverted , like the chromium one ) , so that the wavelength order of the transitions is the opposite with respect to the previous case . because of these similarities , not only the @xmath65 patterns calculated taking into account and neglecting interferences are the same as in the cr i triplet ( compare panel @xmath69 of figures [ fig : o ] and [ fig : cr ] ) but also the @xmath9 profiles obtained including the same amount of continuum are very similar in the two cases ( compare panel @xmath7 of figures [ fig : o ] and [ fig : cr ] ) .
the signatures of interferences between different @xmath0-levels are thus the same as in the cr i triplet at 5207 .
if a continuum @xmath92 stronger than in the case of chromium is considered ( which seems to be a more suitable choice for these lines as they are much weaker than the corresponding chromium lines in the solar atmosphere ) , the signatures of interferences are , as expected , strongly reduced ( see panels @xmath74 and @xmath75 of figure [ fig : o ] ) .
it should be noticed that the antisymmetric profile which is obtained across transition 1 when interferences between different @xmath0-levels and lower - level polarization are taken into account , and when the continuum is not too strong ( see the dotted profile in panel @xmath74 of figure [ fig : o ] ) is very similar to the one observed by @xcite outside the solar limb ( see their figure 5 ) . on the other hand , the profiles plotted in panel @xmath75 of figure [ fig : o ] , obtained assuming a continuum sufficiently strong in order to cancel out all the observational signatures of interferences , are in good qualitative agreement with the off - limb observation presented in figure 5 ( top - right panel ) of @xcite .
this illustrates the sensitivity of the theoretical polarization profiles to the parameters of the model . also
the role of lower - level polarization is the same as in the case of chromium , both concerning its feedback on the atomic polarization of the upper levels ( see panels @xmath74 and @xmath75 of figure [ fig : o ] ) , as well as concerning the ensuing contribution of dichroism ( see panels @xmath69 and @xmath7 of figure [ fig : o - dichr ] , and note the negative @xmath9 signal present in the core of line 1 ) .
if the same , large value of @xmath93 as in panel @xmath75 of figure [ fig : o ] is considered , the signatures of interferences disappear also in the presence of dichroism ( see panel @xmath69 of figure [ fig : o - dichr ] ) . in particular
, we note that although a large value of @xmath91 ( larger than the one that would be obtained through equation ( [ eq : etaic ] ) ) is considered , no positive bump is obtained in the red wing of line 1 ( unlike the cr i case shown in panel @xmath74 of figure [ fig : cr - dichr ] which , being an inverted multiplet , shows such a positive bump in the blue wing of transition 1 ) .
the profiles obtained taking into account dichroism are very similar to the ones calculated by @xcite and show a very good agreement with the close to the limb , on - disk observation presented in figure 4 of @xcite .
note that it is natural that off - limb and inside - limb observations may show different polarization features since the effects of dichroism are enhanced for on - disk observations .
the multiplets considered in the previous sections allowed us to analyze the effects of interferences between different upper @xmath0-levels . in order to investigate the role of interferences between different @xmath0-levels of the lower term
, we consider now the @xmath4 triplet of mg i at 5178 .
this triplet is composed by the following transitions : @xmath84 ( @xmath94 line at 5183.60 ) , @xmath95 ( @xmath96 line at 5172.68 ) , and @xmath97 ( @xmath98 line at 5167.32 ) . we first note that under the hypothesis that the incident field is flat across the whole multiplet , if lower - term polarization is neglected , the emission coefficient @xmath99 is identically zero both for the two - term atom and for the corresponding four - level atom symbol @xmath100 in equations ( [ eq : eps2 t ] ) and ( [ eq : wk - gen ] ) , which is zero for @xmath101 and @xmath102 . ] .
we emphasize that this is a consequence of the flat - spectrum approximation required for the theory of the two - term atom to hold .
if we describe this triplet within the framework of a multi - level atom approach , so that we are allowed to consider a pumping field which varies among the various transitions , the @xmath99 coefficient will be in general different from zero , also under the hypothesis of unpolarized lower levels ( see * ? ? ?
nevertheless , @xcite showed that the presence of a given amount of atomic polarization in the lower levels of this triplet is required in order to explain the observations presented in @xcite , which show @xmath9 signals of the same amplitude in all the three lines .
although lower - term polarization plays an important role in the mg i @xmath7-lines , interferences between different lower @xmath0-levels are found to produce negligible effects on the @xmath103 pattern of this multiplet ( see panel @xmath69 of figure [ fig : mgi ] ) .
this result implies that the atomic polarization of the upper level is practically unaffected by the presence of this kind of interferences in the sees ( as already pointed out in section 10.21 of ll04 ) .
interestingly , interferences between different lower @xmath0-levels are found to play a negligible role also on the ratio @xmath104 ( see panel @xmath74 of figure [ fig : mgi ] ) .
detailed analytical calculations performed in section 10.21 of ll04 on the simpler @xmath105 multiplet show that interferences between different @xmath0-levels of the lower term are the smaller the larger the ratio @xmath106 between the fs splitting of the lower term and its natural width ( given by @xmath107).-levels vanish , and the levels become completely uncorrelated ( like in a multi - level atom ) .
vice versa , when this ratio goes to zero , the lower @xmath0-levels are degenerate , and for the principle of spectroscopic stability the lower term has to behave like a @xmath0-level with @xmath108 . ] indeed , such a negligible effect of these interferences on the @xmath109 profile is due to the low value of @xmath53 that we are considering ( @xmath110 ) . as shown in panel @xmath75 of figure [ fig : mgi ] ,
if larger values of @xmath53 ( of the order of 0.1 or larger ) are considered , clear differences can be observed between the @xmath109 profiles obtained taking into account and neglecting interferences between lower @xmath0-levels values are typical of masers . ] .
such differences are larger in the wings of the lines , while they disappear in the cores , where the effects of these interferences remain negligible also when large values of @xmath53 are considered . as discussed in ll04 for the @xmath105 multiplet , when lower - term polarization is taken into account , not only the interferences between different lower @xmath0-levels result to be sensitive to the value of @xmath106 , but also the interferences between magnetic sublevels pertaining to the same @xmath0-level , as well as the populations of the various magnetic sublevels .
this explains the variation with @xmath53 of the @xmath109 profiles calculated neglecting the interferences between lower @xmath0-levels .
it can be noticed that the sensitivity of these latter profiles to the value of @xmath53 is limited to the cores of the lines ( the asymptotic values do not change significantly with @xmath53 ) , while a variation with @xmath53 of the overall pattern ( from the core to the far wings ) takes place when interferences between different @xmath0-levels of the lower term are taken into account .
the effect of interferences between lower @xmath0-levels on the @xmath111 profiles is found to be negligible also when large values of @xmath53 are considered ( see panel @xmath7 of figure [ fig : mgi ] ) .
indeed , these profiles are much more sensitive to the value of @xmath53 in the core of the lines than in the wings .
as expected from the previous discussion , interferences between lower @xmath0-levels do not produce any observable effect on the @xmath9 profile calculated including the contribution of the continuum , taking into account dichroism , and assuming @xmath73 ( see panel @xmath69 of figure [ fig : mgi - b ] ) .
note that within the modeling assumptions here considered ( and in particular the flat - spectrum approximation ) , a negative signal is obtained in the @xmath98 line ( for understanding why the observations of @xcite show positive signals in the three mg i @xmath7-lines , see @xcite and references therein ) .
the @xmath9 profiles obtained for larger values of @xmath53 are shown in panel @xmath7 of figure [ fig : mgi - b ] .
as it can be observed , the @xmath9 pattern calculated through equation ( [ eq : emer - pol ] ) is quite sensitive to the value of @xmath53 in the wings of the lines , while the line center polarization does not show any variation with @xmath53 .
we conclude pointing out that for the values of the continuum considered in figure [ fig : mgi - b ] , no differences can be observed between the @xmath9 profiles obtained taking into account and neglecting interferences between different lower @xmath0-levels , also when high values of @xmath53 are assumed .
here we investigate the role of interferences between different @xmath0-levels on the scattering polarization profile of the h@xmath81 line .
since this line is composed by seven fs components , belonging to three different multiplets , a multi - term atomic model must be applied for the analysis of interferences in this line .
one of the lower terms is the upper term of the strong ly@xmath8 line , while one of the upper terms is also the upper term of the ly@xmath37 line .
the inclusion of these lines in the atomic model is required for a correct analysis of the scattering polarization properties of h@xmath8 ( e.g. , * ? ? ?
the grotrian diagrams of the multi - level and multi - term atomic models considered are shown in figure [ fig : grotrian ] .
we keep assuming @xmath73 and @xmath72 for all the lines considered in the model ( the impact of interferences between different @xmath0-levels does not depend critically on the relative values of these quantities in the various lines ) .
since the wavelength separation among the various components is much smaller than the doppler width of the line , interferences between different @xmath0-levels do not produce any observable signature in the core of the line , across the whole wavelength interval over which the fs depolarization takes place ( see left panel of figure [ fig : h - alpha ] ) .
in agreement with the previous results , the width of this wavelength interval is about 5 times the doppler width of the line .
the polarization present in the @xmath112 lower term ( the upper term of ly@xmath8 ) has been taken into account .
however , its effect on the atomic polarization of the upper levels / terms through the sees is found to be negligible . also in this case
, the @xmath9 profile calculated taking into account interferences shows , for the same value of the continuum , slightly more extended wings than the profile calculated neglecting them ( see right panel of figure [ fig : h - alpha ] ) .
finally , it must be pointed out that the line - core asymmetry shown by the @xmath9 profile of the right panel of figure [ fig : h - alpha ] is not present when radiative transfer effects are fully taken into account in given semi - empirical models of the solar atmosphere ( see * ? ? ?
* ) , although an asymmetric @xmath9 profile similar to that observed by @xcite can be produced in the presence of magnetic field gradients ( see * ? ? ?
* ; * ? ? ?
we investigate now how the fractional polarization patterns described in the previous sections are modified by the presence of a magnetic field . before carrying out this analysis
, we recall that in the multi - level atom approximation , any kind of correlation between different @xmath0-levels is neglected by definition .
this implies that when this approach is applied in the presence of a magnetic field , it must be always assumed that the field is sufficiently weak for the zeeman effect regime to hold ( the splitting of the magnetic sublevels must be much smaller than the separation among the various @xmath0-levels ) .
this limitation is not required as far as the multi - term atom approximation is considered . in this latter case ,
magnetic fields going from the zeeman effect regime to the complete paschen - back effect regime can be considered . as we will see ( and as it is discussed in detail in ll04 ) , in the incomplete paschen - back effect regime , interferences between different @xmath0-levels are at the origin of interesting phenomena ( e.g. , `` level - crossing '' and `` anti - level - crossing '' effects ) , which may leave their signatures in the observed polarization profiles .
we start considering the @xmath113 doublet of mg ii at 2800 .
the energy separation between the two @xmath0-levels of the upper term is sufficiently large for the zeeman effect regime to hold , at least for the magnetic field intensities of the solar atmospheric plasma . in the zeeman effect regime ,
the magnetic sublevels pertaining to the same @xmath0-level split monotonically with the magnetic field strength .
because of the relaxation term proportional to @xmath35 appearing in the sees ( recall equations ( [ eq : see ] ) in section [ sect : theory ] , and the discussion therein ) , this splitting causes a decrease of the interferences between different magnetic sublevels pertaining to the same @xmath0-level ) . ] , which in turn produces a decrease of the linear polarization degree of the emitted radiation with respect to the non magnetic case .
this is the hanle effect for quantum interferences within the same @xmath0-level . in the mg ii doublet under investigation
, this effect can be clearly observed in the core of the k line ( see panel @xmath69 of figure [ fig : mg_mag ] ) . in this regime , on the other hand
, the variation of the energy separation between magnetic sublevels pertaining to different @xmath0-levels is extremely small , so that interferences between different @xmath0-levels do not show significant variations with respect to the unmagnetized case .
the hanle effect has therefore the same impact on the linear polarization patterns calculated taking into account and neglecting interferences , as shown in panel @xmath7 of figure [ fig : mg_mag ] .
as it can be clearly observed in the same figure , the hanle effect takes place only in the core of the lines , right in the wavelength interval where the contribution of interferences between different @xmath0-levels is negligible .
since the hanle effect takes place only in the core of the lines , if the same continuum as in the unmagnetized case is considered , a two - peak @xmath9 profile is obtained , the central dip being produced by the hanle depolarization ( see panel @xmath74 of figure [ fig : mg_mag ] ) .
this structure is gradually lost as the continuum intensity is increased . on the other hand ,
if a continuum sufficiently strong to cancel out this two - peak structure is considered , also the signatures of interferences between different @xmath0-levels ( such as the antisymmetric pattern across the h line ) result to be strongly reduced ( see panel @xmath75 of figure [ fig : mg_mag ] ) . as in section [ sect : continuum ] , we emphasize that the physical origin of this two - peak structure lies in the way the continuum is included in the slab model . in the presence of a magnetic field with a longitudinal component , an appreciable @xmath114 profile is obtained in the k line .
this signal , due to the hanle effect , appears in the core of the k line , over the same wavelength interval where the hanle depolarization of the @xmath111 profile takes place .
this is also the wavelength interval ( of about five doppler widths ) where interferences between different @xmath0-levels do not produce any appreciable effect on the @xmath111 pattern .
indeed , the profiles obtained taking into account and neglecting such interferences can not be distinguished ( see panel @xmath69 of figure [ fig : mg_mag_uv ] ) .
we note that a very weak antisymmetrical @xmath114 profile ( of the order of @xmath115 , not visible in the figure ) is obtained in the core of the h line when interferences are taken into account .
this signal , too weak to be observable in practice , and of purely academic importance , is probably due to the small variation of the interferences between different @xmath0-levels produced by the magnetic field in the zeeman effect regime . as previously pointed out , this kind of effect ( i.e. , hanle effect for interferences between different @xmath0-levels ) is generally negligible for magnetic fields in this regime , since the variation of the energy separation between the interfering magnetic sublevels ( pertaining to different @xmath0-levels ) is very small .
these effects become more important as the paschen - back effect regime is reached . in the presence of the same field , also a @xmath116 profile due to the zeeman effect is obtained ( see panel @xmath7 of figure [ fig : mg_mag_uv ] ) .
there are no differences between the @xmath116 profiles calculated taking into account and neglecting interferences .
the mg ii h and k lines are an example of a doublet in which the separation between the two lines is much larger than their doppler width .
the results obtained in this section can be extended to any other @xmath2 multiplet in which the two components are sufficiently separated from each other .
the results can also be extended to different multiplets , provided that the various lines are well separated , that lower - term polarization is absent or negligible ( in this multiplet it is zero by definition ) , and that the same hypotheses here considered can be made . in this section ,
we investigate the effects of a magnetic field on the antisymmetric interference pattern characterizing the emergent @xmath9 profile across the @xmath117 transition of the @xmath2 multiplets .
no hanle depolarization takes place in this line , since its line - core polarization is identically zero already in the absence of magnetic fields ( see , for example , the previous analysis of the mg ii doublet at 2800 ) . nevertheless , it is interesting to observe how the interference pattern is modified by the presence of a magnetic field strong enough for the transverse zeeman effect to be appreciable . as shown in figure [ fig : d1-mag ] ,
when the typical signatures of the transverse zeeman effect superimpose on the interference pattern , a peculiar profile , qualitatively similar to the one observed by @xcite in the na i d@xmath5 line , is obtained ( see also the theoretical investigations by trujillo bueno et al .
2002 and casini & manso sainz 2005 , which took hfs into account ) .
the exact shape of the resulting profile depends on the relative amplitude and width of the antisymmetric interference pattern , and of the transverse zeeman effect pattern ( compare the three panels of figure [ fig : d1-mag ] ) .
the shape of the interference pattern is controlled by the continuum intensity ( recall the discussion in section [ sect : continuum ] ) , while the shape of the zeeman pattern by the magnetic field strength .
as expected , the intensity of the magnetic field required for the zeeman pattern to be appreciable on the overall profile is the higher the shorter the wavelength ( compare the three panels of figure [ fig : d1-mag ] ) . in order to investigate the effects of a magnetic field on a multiplet in which the separation among the various components is much smaller than their doppler width , we consider now the ly@xmath81 line .
as discussed in section [ sect : applications ] , interferences between different @xmath0-levels do not produce any observable signature in the core of this line .
also in this case , the impact of the hanle effect due to a magnetic field with strength in the zeeman effect regime is the same on the polarization profiles obtained taking into account and neglecting interferences ( see figure [ fig : ly - alpha_mag ] ) . in the ly@xmath81 line , on the other hand , the separation between the two upper @xmath0-levels is sufficiently small for the incomplete paschen - back effect regime to be reached for magnetic fields of the order of 1 kg . in this regime ,
the variation of the energy separation between pairs of magnetic sublevels pertaining to different @xmath0-levels is no longer negligible . in particular , when the separation between two magnetic sublevels is of the same order of magnitude as their natural width ( i.e. , when the two sublevels overlap ) , the first term in the right - hand side of equations ( [ eq : see ] ) produces a significant modification of the corresponding coherence ( generally an increase in absolute value ) , which may produce observable effects in the polarization of the emergent radiation .
it can be demonstrated that while pairs of magnetic sublevels with @xmath118 can approach and cross each other ( note that when this happens the relaxation term is exactly zero ) , pairs of magnetic sublevels with @xmath119 can approach but never cross .
the terminology of ` level - crossing ' and ` anti - level - crossing ' ( see * ? ? ?
* ) effects is often used to indicate the hanle effect produced by these particular behaviors of the magnetic sublevels in the incomplete paschen - back effect regime .
a detailed description of these phenomena can be found in ll04 . here
we want to focus the attention on the following aspect : the possibility , when interferences between different @xmath0-levels are considered , to have hanle effect also in the presence of a `` vertical '' magnetic field .
as discussed in section [ sect : model ] , in a medium with cylindrical symmetry around a given direction ( e.g. , the vertical ) , if the quantization axis is taken along this direction , it can be shown that no quantum interferences can be induced between pairs of magnetic sublevels pertaining to the same @xmath0-level .
such interferences , which are the only ones to be accounted for in a multi - level atom , also remain zero in the presence of a magnetic field directed along the symmetry axis of the incident radiation , since this does not break the symmetry of the problem .
since the magnetic field does not modify the populations ( note that under the same hypotheses population imbalances can be induced among the various magnetic sublevels ) , it follows that , as known , for the multi - level atom model there is no hanle effect in the presence of a vertical magnetic field .
this is no more true when interferences between different @xmath0-levels are considered .
as previously shown , in a medium with cylindrical symmetry , interferences between magnetic sublevels with the same value of @xmath36 , pertaining to different @xmath0-levels , are actually non - zero .
these interferences are modified by a vertical magnetic field , in the incomplete paschen - back effect regime , with appreciable effects on the polarization of the scattered radiation .
this particular example of hanle effect for interferences between different @xmath0-levels can be clearly observed in figure [ fig : ly - alpha_mag ] : in the presence of a vertical magnetic field of about 2 kg ( thus in the incomplete paschen - back effect regime ) , the amplitude of the scattering polarization signal is increased with respect to the zero - field case .
this enhancement of the polarization due to a vertical magnetic field was already pointed out by @xcite for the na i d@xmath6 line , and by @xcite for the ba ii d@xmath6 line .
however , it should be observed that in these latter cases the effect is due to interferences between different hfs @xmath120-levels , and not between different @xmath0-levels as in the ly@xmath81 line ( the physical mechanism is exactly the same ) .
it is clear that this effect is of more practical interest when hfs is present , since magnetic fields relatively weak are sufficient for the incomplete paschen - back effect regime of hfs to be reached .
we analyze here the effect of a magnetic field on the mg i triplet at 5178 .
as discussed in section [ sect : mgi ] , the polarization of the radiation emitted in these lines is very sensitive to the atomic polarization of the lower levels .
the so - called lower - level hanle effect can thus be clearly observed in this triplet .
if the lower level is metastable ( as in this case ) , the lower - level hanle effect is sensitive to rather weak magnetic fields . in these lines , assuming @xmath121 , it can already be appreciated for magnetic field intensities of the order of a few mg ( see figure [ fig : wing - he ] ) , while for a magnetic field of 1 g it is already saturated . also in the presence of magnetic fields
, interferences between different lower @xmath0-levels do not produce any observable signature on the @xmath9 profile of the emergent radiation , and they can be safely neglected .
an interesting property of the lower - level hanle effect is that it does not vanish in the wings of the lines , but it depolarizes the whole pattern ( see panel @xmath69 of figure [ fig : wing - he ] ) . as a consequence , if the same continuum as in the unmagnetized case is considered , no double - peak structures are obtained as a result of the hanle depolarization ( see panel @xmath7 of figure [ fig : wing - he ] ) .
we recall that the property of the ( upper - level ) hanle effect to vanish in the wings of the lines is strictly verified under the assumptions of unpolarized lower level , no stimulation effects , and no elastic collision ( see section 10.4 of ll04 for an analytical proof ) .
it is clear that this property remains valid whenever stimulation effects are very weak ( like in the solar atmosphere ) , and whenever the influence of lower - level polarization and collisions on the polarization properties of the line is negligible .
an analysis of the influence of elastic collisions on the hanle effect can be found in section 10.6 of ll04 .
a detailed analysis of the physical conditions under which `` wing - hanle - effect '' can be observed will be presented in a forthcoming paper .
we have investigated the effects of quantum interferences between different @xmath0-levels on the linear polarization pattern of the radiation scattered at 90@xmath1 by a slab of stellar atmospheric plasma .
the investigation has been carried out within the framework of the quantum theory of polarization presented in ll04 , which is based on the flat - spectrum approximation .
we started focusing our attention on the @xmath2 doublet .
we analyzed the effects of the interferences between the two upper @xmath0-levels as a function of the wavelength separation ( @xmath122 ) between the two fs components , and assuming a finite doppler width ( @xmath68 ) for the two spectral lines , in the absence of the continuum .
the most important results of this part of our study , carried out for the unmagnetized case , and neglecting the effects of collisions and stimulated emission , are the following . 1
. the effects of interferences between different @xmath0-levels are negligible in the core of the two lines , while they become important moving from the center to the wings ( see panel @xmath69 of figure [ fig : pq - gen ] ) .
if the separation @xmath122 between the two lines is much larger than their doppler width , the shape of the overall interference pattern ( shown in panel @xmath69 of figure [ fig : pq - gen ] ) does not depend on the particular value of @xmath122 .
2 . in the core of the two lines ,
the fractional polarization profiles obtained taking into account and neglecting such interferences show flat `` plateaux '' of about @xmath123 , across which they perfectly coincide ( see panels @xmath7 and @xmath74 of figure [ fig : pq - gen ] ) .
when @xmath124 ( so that the two plateaux merge ) , all the signatures due to interferences disappear between the two - lines ( note that this merging starts when the intensity profiles are still well separated from each other ; see panel @xmath74 of figure [ fig : dl - dep ] ) .
when @xmath125 ( so that the two plateaux completely merge ) , the fine structure depolarization takes place on a wavelength interval of about @xmath123 ( the width of a single plateau ) , irrespectively of the actual separation between the two components ( see panels @xmath75 , @xmath76 , and @xmath77 of figure [ fig : dl - dep ] ) .
signatures of interferences become appreciable in the line - core when the separation between the interfering @xmath0-levels is of the same order of magnitude as their natural width ( see panels @xmath76 and @xmath77 of figure [ fig : dl - dep ] ; note that this circumstance is not met by any real multiplet ) .
although we believe that the flat plateaux that appear in the core of the lines may not be obtained once radiative transfer effects in realistic solar model atmospheres are taken into account , the fact that across the corresponding wavelength intervals the effects of interferences are negligible is a physical result that is expected to remain valid also when full radiative transfer is properly considered . a detailed analysis of the physical and mathematical origin of these intervals , and of their width ( of the order of @xmath123 ) ,
has thus been carried out ( see the appendix ) .
we then analyzed how the signatures of interferences are masked when the contribution of an unpolarized continuum is taken into account , finding the following main results . 1 .
the amount of the continuum emissivity needed to mask the signatures of interferences is the larger the smaller the wavelength separation between the interfering lines ( see figures [ fig : cont - dep ] and [ fig : bana ] ) .
2 . in the case of the @xmath2 multiplet ,
the effect of the continuum is to reduce the amplitude of the negative minimum between the two lines , to move it toward the @xmath126 transition , and to produce an antisymmetrical pattern across this transition ( see figure [ fig : cont - dep ] ) .
for a given value of the continuum emissivity , the amplitude of the above - mentioned antisymmetrical pattern is the smaller the larger the separation between the two components of the multiplet ( see figure [ fig : bana ] ) .
the above - mentioned results are not all limited to the @xmath2 doublet , but they can be generalized to any other multiplet .
we carried out an analysis of the signatures produced by interferences between different @xmath0-levels on the following multiplets : * ba ii @xmath2 doublet ( 4554 and 4934 resonance lines ) ; * ca ii @xmath2 doublet ( h and k lines ) ; * mg ii @xmath2 doublet ( h and k lines ) ; * na i @xmath2 doublet ( d@xmath5 and d@xmath6 lines ) ; * h i @xmath2 doublet ( ly@xmath8 ) ; * cr
i @xmath3 triplet at 5207 ; * o i @xmath3 triplet at 7773 ; * mg i @xmath4 triplet ( b@xmath5 , b@xmath6 and b@xmath127 lines ) ; * h i h@xmath8 ( line composed of seven fs components belonging to three different multiplets ) . the analysis of the cr i and o i triplets allowed us to investigate in some details the combined effect on the emergent scattering polarization profiles of interferences between different @xmath0-levels , lower - level polarization ( see figures [ fig : cr ] and [ fig : o ] ) , and dichroism ( see figures [ fig : cr - dichr ] and [ fig : o - dichr ] ) , for different values of the background continuum . the analysis of the mg i b - lines allowed us to investigate the role of interferences between different @xmath0-levels of the lower term .
we found that for the typical solar values of @xmath53 ( the mean number of photons per mode of the pumping radiation field ) , their effect on the scattering polarization profiles is completely negligible ( see panels @xmath69 and @xmath74 of figure [ fig : mgi ] , and panel @xmath69 of figure [ fig : mgi - b ] ) .
values of @xmath53 of the order of 0.1 or larger are needed in order to observe their signatures on the ratio @xmath128 , their effect being in any case negligible in the core of the lines ( see panel @xmath75 of figure [ fig : mgi ] ) .
the atomic polarization of the upper levels is practically unaffected by the presence of these interferences in the sees , also for high values of @xmath53 ( see panel @xmath7 of figure [ fig : mgi ] ) .
the signatures of interferences between different lower @xmath0-levels ( appreciable only when high values of @xmath53 are considered ) are in any case strongly masked by the presence of the continuum .
our calculations also showed an appreciable sensitivity of the @xmath9 profile of the emergent radiation to the value of @xmath53 ( see panel @xmath7 of figure [ fig : mgi - b ] ) .
we finally investigated whether or not , and to which extent , the signatures due to interferences between different @xmath0-levels are modified by the presence of a magnetic field .
the results can be summarized as follows . 1 . in the zeeman effect regime ,
the influence of the magnetic field on the values of the interferences between different @xmath0-levels is extremely small .
the hanle effect thus leaves the same signatures on the fractional polarization patterns calculated taking into account and neglecting interferences between different @xmath0-levels .
as shown by the profiles calculated for the mg ii h and k lines , as far as the @xmath111 pattern is concerned , the hanle effect takes place in the core of the k line , right in the wavelength interval where the effect of interferences is negligible ( see figure [ fig : mg_mag ] ) .
also the @xmath114 signal , produced by the hanle effect in the same line in the presence of a longitudinal field , appears across this wavelength interval ( see panel @xmath69 of figure [ fig : mg_mag_uv ] ) .
2 . in the incomplete paschen - back effect regime ,
interferences between different @xmath0-levels are significantly modified by the magnetic field , thus producing observable effects on the emergent radiation .
the physical mechanisms at the origin of these effects ( e.g. , level - crossings , anti - level - crossings , alignment - to - orientation conversion mechanism ) are described in detail in ll04 . in this paper
, we focused our attention on the possibility to have hanle effect also in the presence of `` vertical '' fields when interferences between different @xmath0-levels are taken into account ( see figure [ fig : ly - alpha_mag ] ) .
we should remember that the paschen - back effect regime is more typical for hfs multiplets .
even for the ly@xmath8 line considered here , for which the separation between the two interfering @xmath0-levels is particularly small if compared to that of other fs multiplets , fields of about 2 kg are needed for producing this kind of effects . in summary ,
when the energy separation between pairs of magnetic sublevels pertaining to different @xmath0-levels is much larger than their width ( which is always the case in the absence of magnetic fields or in the presence of weak magnetic fields in the zeeman effect regime ) , the contribution of the interferences between such magnetic sublevels modifies the fractional linear polarization pattern in the wings of the lines ( outside a @xmath129 wavelength interval around the line center ) .
however , at these wavelengths the line contribution to the total emissivity and absorptivity is no longer dominating with respect to that of the continuum .
the presence of the continuum strongly masks the effects of interferences between different @xmath0-levels , which indeed are observable only in rather strong spectral lines ( e.g. , ca ii h and k , na i d@xmath5 and d@xmath6 ) . because of the crucial role of the continuum , it is not possible to establish a simple quantitative criterion for deciding whether or not interferences between different @xmath0-levels
are expected to produce observable effects .
on the other hand , in the presence of a magnetic field sufficiently intense for the paschen - back effect regime to be reached ( or , in other words , when the energy separation between pairs of magnetic sublevels pertaining to different @xmath0-levels is no longer negligible with respect to their width ) , interferences between different @xmath0-levels may modify the amplitude of the fractional polarization pattern in the core of the lines through the mechanisms described in detail in sections 10.18 and 10.20 of ll04 .
we conclude pointing out that all the results presented in this paper on the effects of interferences between different @xmath0-levels can be generalized to the case of interferences between different hfs @xmath120-levels .
the only remarkable difference for practical applications is the fact that the paschen - back effect regime for hfs is generally reached for magnetic fields much weaker than those needed for reaching the paschen - back effect regime for fs . financial support by the spanish ministry of science and innovation through project aya2010 - 18029 ( solar magnetism and astrophysical spectropolarimetry )
is gratefully acknowledged .
in this appendix , we first recall some equations and analytical results derived in ll04 that have been mentioned in the text . these equations will be the starting point for the following derivation of a series of analytical expressions that will allow us to get more insights on some results presented in this paper on the @xmath2 multiplets .
let us consider a two - term atom , the upper and lower terms being characterized by the spin @xmath27 and by the total orbital angular momentum @xmath38 .
each term is composed of @xmath130 fs @xmath0-levels , each having @xmath131 magnetic sublevels @xmath36 , for a total of @xmath132 sublevels . as shown in section 10.16 of ll04 , under the simplifying hypotheses of unpolarized lower term , no stimulation effects , no collisions , no magnetic field , and in the flat - spectrum approximation
, it is possible to find an analytical solution of the sees for the multipole moments of the density matrix of the upper term ( see equation ( 10.126 ) of ll04 ) .
if this solution is substituted into the expression of the emission coefficient of a two - term atom ( see equation ( 7.47e ) of ll04 ) , one obtains ( see equation ( 10.129 ) of ll04 ) @xmath133 where @xmath134 is the number density of atoms in the lower term , @xmath135 and @xmath136 are the einstein coefficients for spontaneous emission and for absorption , respectively , between the two terms , @xmath137 is a geometrical tensor that depends on the direction of the emitted radiation @xmath138 , and on the reference direction for positive @xmath47 ( see section 5.11 of ll04 and equations therein for its explicit expression ) , @xmath139 is the bohr frequency between levels @xmath140 and @xmath141 , and @xmath142 is the radiation field tensor introduced in section [ sect : model ] , describing the incident ( pumping ) radiation field .
we recall that because of the flat - spectrum approximation it is sufficient to calculate it at a single frequency @xmath143 within the frequency interval covered by the multiplet .
the profile @xmath144 is given by @xmath145 where @xmath146 the voigt function @xmath147 and the associated dispersion profile @xmath148 are functions of the reduced frequency @xmath149 , with @xmath150 the doppler width in frequency units and @xmath151 the bohr frequency between levels @xmath152 and @xmath153 , and of the damping constant @xmath154 .
the broadening constant @xmath155 is given by @xmath156 where @xmath157 and @xmath158 are the inverse lifetimes of the levels @xmath152 and @xmath153 , respectively .
the contribution of interferences between different @xmath0-levels to the emission coefficient is described by the terms with @xmath159 in equation ( [ eq : eps2 t ] ) .
if these terms are neglected , recalling the relation @xmath160 and introducing the quantity ( see equation ( 10.147 ) of ll04 ) @xmath161 equation ( [ eq : eps2 t ] ) reduces to @xmath162_{\rm no\,int . } & \!\!\ ! = \!\!\ !
& \frac{h \nu_0}{4 \pi } \ ; \mathbb{n}_{\ell } \ ; \frac{1}{2s+1 } \ ; \frac{1}{2l_{\ell}+1 } \ ; \sum_{j_{\ell } j_u } \ , ( 2j_{\ell}+1 ) \ ; b(j_{\ell } \rightarrow j_u ) \ ; \phi(\nu_{j_u , j_{\ell}}-\nu ) \\
\nonumber & & \sum_{kq } \ , w_k(l_{\ell } l_u s , j_{\ell } j_u ) \ ; ( -1)^q \ ; \mathcal{t}^k_q(i,\mathbf{\omega } ) \ ; j^k_{-q}(\nu_0 ) \;\ ; .\end{aligned}\ ] ] this is the expression of the emission coefficient for the corresponding multi - level atom , under the same hypotheses required for equation ( [ eq : eps2 t ] ) to hold .
indeed , it could be obtained starting from the sees and from the expression of the emission coefficient of a multi - level atom , under the hypotheses of no magnetic field , no collisions , no stimulation effects , and assuming that the incident radiation field is spectrally flat across all the transitions considered and that all the magnetic sublevels of the lower @xmath0-levels are equally populated ( hypothesis equivalent to that of the unpolarized lower term ) . in the particular case of @xmath163 ,
the multiplet reduces to a single line with @xmath164 and @xmath108 , the symbol @xmath165 reduces to ( see section 10.17 of ll04 ) @xmath166 and the expression of the emission coefficient of a two - level atom is recovered ( see equation ( 10.16 ) of ll04 ) : @xmath167_{\rm two - lev } = \frac{h \nu_0}{4 \pi } \ ; \mathbb{n}_{\ell } \ ; b(j_{\ell } \rightarrow j_u ) \ ; \phi(\nu_{j_u , j_{\ell}}-\nu ) \ ; \sum_{kq } \
, w_k(j_{\ell } , j_u ) \;(-1)^q \ ; \mathcal{t}^k_q(i,\mathbf{\omega } ) \ ; j^k_{-q}(\nu_0)\;\ ; .\ ] ] the quantity @xmath165 is thus a sort of generalization of @xmath168 that takes into account ( though under the flat - spectrum approximation ) the effects due to the presence of the various components of a multiplet on the polarization properties of a given transition . besides the trivial case of @xmath163
, it can be shown that the symbol @xmath165 coincides with the symbol @xmath168 in all the @xmath169 multiplets ( all the multiplets with @xmath170 , @xmath171 , and any value of the spin ) .
we consider an unmagnetized plane - parallel atmosphere , we take a cartesian reference system with the @xmath172-axis ( quantization axis ) directed along the local vertical and focus our attention on the radiation scattered at 90@xmath1 by an optically thin slab of solar plasma . choosing the reference direction for positive @xmath47 parallel to the atmosphere , and recalling that ,
because of the symmetry of the problem , the only non - zero components of the radiation field tensor are @xmath51 and @xmath52 , the geometrical tensors @xmath137 that enter the previous expressions of the emission coefficients assume the values @xmath173 substituting these values into equation ( [ eq : eps2l ] ) , we can easily obtain the analytical expression of the fractional polarization pattern @xmath174 for the radiation scattered at 90@xmath1 by a two - level atom : @xmath175_{\rm two - lev } = \frac{3w_2(j_{\ell},j_u ) } { \displaystyle \frac{4}{w } - w_2(j_{\ell},j_u ) } \;\ ; , \label{eq : pq-2l}\ ] ] where @xmath54 is the anisotropy factor defined in equation ( [ eq : w - barn ] ) .
as discussed in ll04 , in a two - term atom , when the separation among the various lines of the multiplet is much larger than their natural width , the emission coefficients in the neighborhood of a single line with @xmath176 and @xmath177 can be evaluated by restricting the summation over @xmath140 , @xmath141 , and @xmath178 in equation ( [ eq : eps2 t ] ) to the values @xmath179 and @xmath177 ( i.e. , neglecting , in particular , the terms corresponding to the interferences between different @xmath0-levels ) . from equation ( [ eq : epsml ] ) , we can thus obtain the analytical expression of the fractional polarization of the radiation scattered by a two - term atom in the core of the various lines of the corresponding multiplet . for 90@xmath1 scattering
we obtain @xmath180_{\rm two - term } = \frac{3w_2(l_{\ell } l_u s , j_{\ell } j_u ) } { \displaystyle \frac{4}{w } - w_2(l_{\ell } l_u s , j_{\ell } j_u ) } \;\ ; . \label{eq : pq-2tcore}\ ] ] we now focus our attention on the @xmath181 multiplet . taking into account that for this multiplet @xmath182 and that @xmath183 ( transition 1 ) , while @xmath184 ( transition 2 ) , starting from equations ( [ eq : eps2 t ] ) and ( [ eq : epsml ] ) , it is possible to find rather compact analytical expressions for the fractional polarization pattern @xmath185 of the radiation scattered at 90@xmath1 , both neglecting or taking into account interferences . using the more compact notation @xmath186 and @xmath187 for the profiles defined in equation ( [ eq : profiles ] ) , with @xmath188 the frequency of transition @xmath189 , after some algebra ( involving the calculation of several 6-@xmath189 symbols , and the use of equation ( [ eq : beinst ] ) ) , one obtains the following expressions : + @xmath175_{\rm no\,int . }
= \frac{3/2 \ , \phi_2 } { \displaystyle \frac{4}{w } \ , \bigg ( \frac{1}{2 } \phi_1 + \phi_2 \bigg ) - \frac{1}{2}\,\phi_2 } \;\ ; , \label{eq : eps3lp}\ ] ] @xmath175_{\rm int . }
= \frac{3/2 \ , ( \phi_2 + \chi ) } { \displaystyle \frac{4}{w } \ , \bigg ( \frac{1}{2 } \phi_1 + \phi_2 \bigg ) -\frac{1}{2 } \bigg ( \phi_2 + \chi \bigg ) } \;\ ; , \label{eq : eps2tp}\ ] ] where @xmath190 with @xmath191 .
the quantity @xmath192 is the frequency separation between the two components of the multiplet .
as it can be observed comparing equations ( [ eq : eps3lp ] ) and ( [ eq : eps2tp ] ) , the contribution of interferences between different @xmath0-levels is fully described by the quantity @xmath193 . in section [
sect : dw - effect ] , we observed that in the neighborhood of the single transitions , the @xmath185 profiles calculated taking into account and neglecting interferences coincide and are constant over a wavelength interval of about five doppler widths .
the analytical expressions of @xmath65 given by equations ( [ eq : eps3lp ] ) and ( [ eq : eps2tp ] ) allow to analyze in detail the origin of this behavior . as far as @xmath194_{\rm no\,int.}$ ] is concerned , equation ( [ eq : eps3lp ] ) can be rewritten as @xmath175_{\rm no\,int . }
= \frac{3/2 } { \displaystyle \frac{4}{w } \ , \bigg(1+\frac{1}{2 } \frac{\phi_1}{\phi_2 } \bigg ) - \frac{1}{2 } } \;\ ; .\ ] ] as shown in panel @xmath69 of figure [ fig : voigt - vc ] , the ratio @xmath195 is much smaller than unity for frequencies close to transition 2 , while it is extremely large for frequencies close to transition 1 . in the core of the two transitions ,
we thus obtain @xmath196_{\rm no\,int . }
= \frac{3/2 } { \displaystyle \frac{4}{w } - \frac{1}{2 } } \;\ ; , \;\;\;\;\;\;\ ; \big [ p_q(\nu \approx \nu_{01 } ) \big]_{\rm no\,int . }
= 0 \;\ ; .\ ] ] these expressions coincide with those valid for a two - level atom ( see equation ( [ eq : pq-2l ] ) ) .
if the two lines are very separated from each other , the boundaries of the intervals over which the ratio @xmath195 is very small or very large ( and thus the fractional polarization @xmath194_{\rm no \ , int.}$ ] is practically constant ) are determined by the points where the shape of the voigt profile changes from gaussian to lorentzian ( see panel @xmath69 of figure [ fig : voigt - vc ] ) .
: ratio between the voigt profile of transition 1 ( @xmath197 ) , and the voigt profile of transition 2 ( @xmath198 ) , plotted as a function of the reduced wavelength @xmath199 , with @xmath68 the doppler width of the two lines , and @xmath200 the wavelength corresponding to the energy difference between the centers of gravity of the two terms .
the separation between the two components is @xmath201 .
the damping parameter is @xmath202 .
panel @xmath7 : plot of @xmath203 ( see the text ) as a function of log(@xmath69 ) . ]
exploiting a series of properties of the functions @xmath147 and @xmath148 discussed in section 5.4 of ll04 , it can be shown that an asymptotic expansion of the voigt function in power series of @xmath69 is given by @xmath204 the value @xmath203 for which the behavior of the voigt function changes from gaussian to lorentzian can thus be evaluated through the equality @xmath205 it can be verified that the solution of this transcendent equation is given by the following recursive expression ( e. landi deglinnocenti , private communication ) : @xmath206 with @xmath207 .
the value of @xmath203 as a function of log(@xmath69 ) is shown in panel @xmath7 of figure [ fig : voigt - vc ] .
we see that @xmath203 varies almost linearly with log(@xmath69 ) , going from a value of 3.844 for @xmath208 to a value of 2.673 for @xmath209 , in agreement with the extension ( of about five doppler widths ) of the plateaux observed in the plots of section [ sect : dw - effect ] . :
plot of the interference term @xmath210 as a function of the reduced wavelength , with @xmath68 the doppler width of the two lines , and @xmath200 the wavelength corresponding to the energy difference between the centers of gravity of the two terms . as in panel @xmath69 of figure [ fig : voigt - vc ] , the separation between the two components is @xmath211 , while the damping parameter is @xmath202 .
panel @xmath7 : ratio between the dispersion profile of line 2 ( @xmath212 ) and the voigt profile of line 2 ( @xmath198 ) .
the inner panel shows in more details the behavior of this ratio around the wavelength position of transition 2 .
panel @xmath74 : same as panel @xmath7 , with the ratio @xmath213 multiplied by @xmath214 .
panel @xmath75 : plot of the first derivative of the ratio @xmath213 . ]
the situation is similar as far as the @xmath194_{\rm int.}$ ] profile is concerned .
noticing that for small values of @xmath54 the second term in the denominator of equation ( [ eq : eps2tp ] ) is much smaller than the first one , this equation can be simplified as @xmath175_{\rm int . }
= \frac{3w}{8 } \left [ \frac{\phi_2}{\frac{1}{2 } \phi_1 + \phi_2 } + \frac{\chi}{\frac{1}{2 } \phi_1 + \phi_2 } \right ] \;\ ; .\ ] ] the contribution of interferences is described by the term @xmath215 ( see panel @xmath69 of figure [ fig : chi - voigt ] ) , the remaining part of the expression being the same as in the case without interferences . in order to analyze the behavior of this term around transition 2 , recalling equation ( [ eq : chi ] ) , we rewrite it as @xmath216 \;\ ; .
\label{eq : interf}\ ] ] noticing that when the two transition are sufficiently separated from each other , as we are assuming here , @xmath197 and @xmath217 are practically constant close to transition 2 , it is clear that both the ratios @xmath195 ( as observed in panel @xmath69 of figure [ fig : voigt - vc ] ) and @xmath218 are practically zero in the frequency interval over which @xmath198 has a gaussian behavior . more complex is the behavior of @xmath213 .
this ratio is exactly zero for @xmath219 but , as shown in panel @xmath7 of figure [ fig : chi - voigt ] , it immediately increases ( in absolute value ) moving away from the line center , assuming values of the order of unity already for @xmath220 .
the first derivative of this ratio shows an abrupt increase at the frequencies where the behavior of the voigt function @xmath198 changes from gaussian to lorentzian , becoming practically constant for larger values of @xmath152 ( see panel @xmath75 of figure [ fig : chi - voigt ] ) .
however , as shown in panel @xmath74 of figure [ fig : chi - voigt ] , the contribution of this term is in any case negligible within the interval where the voigt function @xmath198 has a gaussian behavior , when the multiplicative factor @xmath221 ( which is extremely small when the separation between the two components is much larger than the einstein coefficient ) is taken into account . noticing that the multiplicative factor @xmath222 makes the contribution of the first term in the square bracket of equation ( [ eq : interf ] ) negligible within the interval over which @xmath198 has a gaussian behavior
, it follows that interferences bring a negligible contribution on this spectral interval around transition 2 . with analogous considerations
, it can be shown that interferences bring a negligible contribution around transition 1 , within the interval over which the voigt profile @xmath197 has a gaussian behavior .
belluzzi , l. 2011 , in asp conf .
437 , solar polarization 6 , theoretical approaches to the physics of spectral line polarization , ed .
kuhn , s.v .
berdyugina , d.m .
harrington , s. keil , h. lin , t. rimmele , & j. trujillo bueno ( san francisco , ca : asp ) , 29 belluzzi , l. , trujillo bueno , j. , & landi deglinnocenti , e. 2007 , , 666 , 588 belluzzi , l. , landi deglinnocenti , e. , & trujillo bueno , j. 2009 , , 705 , 218 bommier , v. 1980 , , 87 , 109 casini , r. , & manso sainz , r. 2005 , , 624 , 1025 casini , r. , & landi deglinnocenti , e. 2007 , in plasma polarization spectroscopy , ed .
t. fujimoto & a. iwamae ( berlin : springer ) , 249 fano , u. 1957 , rev .
phys . , 29 , 74 fontenla , j.m . ,
avrett , e.h . , & loeser , r. 1993 , , 406 , 319 gandorfer , a. 2000 , the second solar spectrum : a high resolution polarimetric survey of scattering polarization at the solar limb in graphical representation .
i : 4625 @xmath223 to 6995 @xmath223 ( zrich : vdf eth ) gandorfer , a. 2002 , the second solar spectrum : a high resolution polarimetric survey of scattering polarization at the solar limb in graphical representation . vol .
ii : 3910 @xmath223 to 4630 @xmath223 ( zrich : vdf eth ) gandorfer , a. 2005 , the second solar spectrum : a high resolution polarimetric survey of scattering polarization at the solar limb in graphical representation . vol .
iii : 3160 @xmath223 to 3920 @xmath223 ( zrich : vdf eth ) henze , w. , & stenflo , j.o .
1987 , , 111 , 243 keller , c.u .
, & sheeley , n.r.,jr . 1999 , in astrophysics and space science library 243 , solar polarization , scattering polarization in the chromosphere , ed .
nagendra & j.o .
stenflo ( boston , mass .
: kluwer ) , 17 landi deglinnocenti , e. , & landolfi , m. 2004 , ` polarization in spectral lines ' ( dordrecht : kluwer ) ( ll04 ) landi deglinnocenti , e. , landi deglinnocenti , m. , & landolfi , m. 1997 , in science with thmis , density matrix and polarized radiative transfer : a heuristic approach to the problem of frequency redistribution , ed .
n. mein , & s. sahal - brchot ( paris : obs .
paris - meudon ) , 59 manso sainz , r. 2011 , in asp conf .
437 , solar polarization 6 , elementary theory of scattering polarization in complex atoms , ed .
kuhn , s.v .
berdyugina , d.m .
harrington , s. keil , h. lin , t. rimmele , & j. trujillo bueno ( san francisco , ca : asp ) , 19 sheeley , n.r . , jr . , & keller , c.u . 2003 , , 594 , 1085 smitha , h.n . ,
sampoorna , m. , nagendra , k.n . , & stenflo , j.o .
2011 , , 733 , 4 stenflo , j.o . 1980 , , 84 , 68 stenflo , j.o .
2009 , in iau symp .
259 , cosmic magnetic fields : from planets , to stars and galaxies , the second solar spectrum and the hidden magnetism , ed .
strassmeier , a.g .
kosovichev & j.e .
beckman ( cambridge : cambridge univ .
press ) , 211 stenflo , j.o .
, & keller , c.u .
1997 , , 321 , 924 stenflo , j.o . ,
baur , t.g .
, & elmore , d.f .
1980 , , 84 , 60 stenflo , j.o .
, gandorfer , a. , & keller , c.u .
2000 , , 355 , 781 tpn , j. , & trujillo bueno , j. 2010 , , 711 , l133 tpn , j. , & trujillo bueno , j. 2011 , , 732 , 80 trujillo bueno , j. 1999 , in astrophysics and space science library 243 , solar polarization , towards the modelling of the second solar spectrum , ed .
nagendra & j.o .
stenflo ( boston , mass .
: kluwer ) , 73 trujillo bueno , j. 2001 , in asp conf .
236 , advanced solar polarimetry : theory , observations and instrumentation , atomic polarization and the hanle effect , ed . m. sigwarth ( san francisco , ca : asp ) , 161 trujillo bueno , j. 2003 , in asp conf . ser .
307 , solar polarization 3 , new diagnostic windows on the weak magnetism of the solar atmosphere , ed .
j. trujillo bueno & j. snchez almeida ( san francisco , ca : asp ) , 407 trujillo bueno , j. 2009 , in asp conf .
405 , solar polarization 5 , the magnetic sensitivity of the second solar spectrum , ed .
s. berdyugina , k.n .
nagendra & r. ramelli ( san francisco , ca : asp ) , 65 trujillo bueno , j. , casini , r. , landolfi , m. & landi deglinnocenti , e. 2002 , , 566 , l53 | the spectral line polarization produced by optically pumped atoms contains a wealth of information on the thermal and magnetic structure of a variety of astrophysical plasmas , including that of the solar atmosphere .
a correct decoding of such information from the observed stokes profiles requires a clear understanding of the effects that radiatively induced quantum interferences ( or coherences ) between pairs of magnetic sublevels produce on these observables , in the absence and in the presence of magnetic fields of arbitrary strength . here
we present a detailed theoretical investigation on the role of coherences between pairs of sublevels pertaining to different fine - structure @xmath0-levels , clarifying when they can be neglected for facilitating the modeling of the linear polarization produced by scattering processes in spectral lines . to this end
, we apply the quantum theory of spectral line polarization and calculate the linear polarization patterns of the radiation scattered at 90@xmath1 by a slab of stellar atmospheric plasma , taking into account and neglecting the above - mentioned quantum interferences .
particular attention is given to the @xmath2 , @xmath3 , and @xmath4 multiplets .
we point out the observational signatures of this kind of interferences and analyze their sensitivity to the energy separation between the interfering levels , to the amount of emissivity in the background continuum radiation , to lower - level polarization , and to the presence of a magnetic field . some interesting applications to the following spectral lines are also presented : ca ii h and k , mg ii h and k , na i d@xmath5 and d@xmath6 , the ba ii 4554 and 4934 resonance lines , the cr i triplet at 5207 , the o i triplet at 7773 , the mg i @xmath7-lines , and the h@xmath8 and ly@xmath8 lines of h i. |
phase - space methods were introduced in the very early times of quantum mechanics to avoid some of the troubles arising in the abstract hilbert - space formulation .
the pioneering works of weyl @xcite , wigner @xcite , and moyal @xcite paved thus the way to formally representing quantum mechanics as a statistical theory on phase space @xcite . in this approach one looks for a mapping relating operators ( in hilbert space ) to functions ( in phase space ) . in particular ,
quasiprobability distributions are the functions connected with the density operator @xcite .
usually , the relevant theoretical tools are illustrated with continuous variables ( such as cartesian position and momentum ) , for which much physical knowledge has been inferred .
likewise , the increasing role of qudits in modern quantum information has fuelled a lot of interest in consistently extending these achievements to the case of a discrete phase space @xcite .
however , other physical problems call for different topologies of the phase space @xcite . specifically , in this paper we focus on the discrete cylinder @xmath0 ( @xmath1 denotes here the unit circle and @xmath2 the integers ) , which is associated to the canonical pair angle - angular momentum . a number of systems , such as molecular rotations , electron wave packets , hall fluids , and light fields , to cite only a few examples , can be described in terms of this geometry @xcite . in quantum optics , it is of primary importance to deal with the orbital angular momentum of twisted photons @xcite , which have been proposed for numerous applications @xcite .
the proper definition of angles in quantum mechanics has a long history and requires more care than perhaps might be expected @xcite .
since the conjugate angular momentum has an unbounded spectrum ( that includes positive and negative integers ) , it is , in principle , possible to introduce a _
bona fide _ angle operator .
periodicity , however , brings out subtleties that have triggered long and heated discussions .
we think that a phase - space treatment of this variable may shed light on the origin of all those problems .
a pioneer attempt in that direction was made by mukunda @xcite , who for the first time constructed a wigner function on the discrete cylinder .
this work was subsequently reelaborated and developed in a variety of directions by other authors @xcite .
coherent states for this pair have also attracted much attention @xcite .
these accomplishments , as well as some other related questions , such as the associated minimum uncertainty states , are reviewed in great detail by kastrup @xcite . in spite of this considerable progress
, there is still the need , in our view , for a comprehensive geometrical approach to this problem , which allows us to integrate in a structured way all the previous knowledge , while providing the right tools to attack recent and attractive problems , such as , e.g. , the tomography of these systems @xcite .
the purpose of this paper is precisely to provide such a description .
in this section we recall the structures needed to set up a phase - space description of cartesian quantum mechanics .
this will facilitate comparison with the cylinder later on . for simplicity , we choose one degree of freedom , so the associated phase space is the plane @xmath3 .
the relevant observables are the hermitian coordinate and momentum operators @xmath4 and @xmath5 , with canonical commutation relation ( with @xmath6 throughout ) @xmath7 = i \ , { \hat { { \leavevmode\hbox{\small1\normalsize\kern-.33em1 } } } } \ , , \ ] ] so that they are the generators of the heisenberg - weyl algebra @xcite .
ubiquitous and profound , this algebra has become the hallmark of noncommutativity in quantum theory . to avoid technical problems with the unboundedness of @xmath4 and @xmath5 , it is convenient to work with their unitary counterparts @xmath8 whose action in the bases of eigenvectors of position and momentum is @xmath9 so they represent displacements along the corresponding coordinate axes .
the commutation relations are then expressed in the weyl form @xcite @xmath10 their infinitesimal version immediately gives ( [ eq : hwcom ] ) , but ( [ eq : weyl ] ) is more useful in many instances . in terms of @xmath11 and
@xmath12 a general displacement operator can be introduced as @xmath13 \
, , \ ] ] with the parameters @xmath14 labeling phase - space points .
these operators form a complete trace - orthonormal set ( in the continuum sense ) in the space of operators acting on @xmath15 ( the hilbert space of square integrable functions on @xmath16 ) : @xmath17 = 2 \pi \ , \delta ( q - q^\prime ) \delta ( p - p^\prime ) \ , .\ ] ] note that @xmath18 , while @xmath19 .
in addition , they obey the simple composition law @xmath20 \ , { \hat { d } } ( q + q^{\prime } , p + p^{\prime } ) \ , .\ ] ] we can also work with the fourier transform of @xmath21 @xmath22 \ , { \hat { d } } ( q^\prime , p^\prime ) \ , dq^\prime dp^\prime \ , , \ ] ] which is called a stratonovich - weyl quantizer @xcite .
one can check that the operators @xmath23 are also a complete trace - orthonormal set that transforms properly under displacements @xmath24 where @xmath25 and @xmath26 is the parity operator .
let @xmath27 be an arbitrary ( hilbert - schmidt ) operator acting on @xmath15 . using the stratonovich - weyl quantizer we can associate to @xmath27 a tempered distribution @xmath28 on @xmath29 representing the action of the corresponding dynamical variable in phase space .
in fact , this is known as the wigner - weyl map and reads as @xmath30 \ , .\ ] ] the function @xmath31 is the symbol of the operator @xmath27 .
conversely , we can reconstruct the operator from its symbol through @xmath32 the symbols form an associative algebra endowed with a noncommutative ( star ) product inherited by the operator product .
let the functions @xmath33 and @xmath34 correspond to the operators @xmath27 and @xmath35 , respectively .
if we denote by @xmath36 the function corresponding to @xmath37 , some simple manipulations lead us to @xcite @xmath38 \
, b(q + q^{\prime \prime } , p + p^{\prime \prime } ) \ , dq^{\prime } dp^{\prime } dq^{\prime \prime } dp^{\prime \prime } \ , .\ ] ] expanding the functions @xmath39 and @xmath40 in a formal taylor series at the point @xmath41 we obtain @xmath42 where @xmath43 is the poisson operator @xmath44 in terms of which we can define the moyal bracket @xcite @xmath45 which is the phase - space counterpart of the commutator in the hilbert - space formulation .
consequently , the time evolution of @xmath33 can be expressed as @xmath46 where @xmath47 is the symbol of the hamiltonian . in this context
, the wigner function is nothing but the symbol of the density matrix @xmath48 .
therefore , @xmath49 \ , , & \nonumber \\ & & \label{eq : hwwignerdef } \\ & { \hat { \varrho } } = \displaystyle \frac{1}{(2\pi)^{2 } } \int_{\mathbb{r}^{2 } } { \hat { w}}(q , p ) w_{{\hat { \varrho}}}(q , p ) \ , dq dp \ , . & \nonumber\end{aligned}\ ] ] for a pure state @xmath50 , it can be represented as @xmath51 which is , perhaps , the most convenient form for actual calculations . according to ( [ eq : evoleq ] ) ,
the liouville - von neumann equation in phase space is @xmath52 when applying this equation in practical cases , one needs the following mapping for the action of the basic variables @xmath4 and @xmath5 on @xmath48 : @xmath53 the wigner function defined in ( [ eq : hwwignerdef ] ) fulfills all the basic properties required for any good probabilistic description .
first , due to the hermiticity of @xmath54 , it is real for hermitian operators .
second , on integrating @xmath55 over the lines @xmath56 , the probability distributions of the rotated quadratures @xmath57 are reproduced @xmath58 in particular , the probability distributions for the canonical variables can be obtained as the marginals @xmath59 third , @xmath60 is translationally covariant , which means that for the displaced state @xmath61 @xmath62 , one has @xmath63 so that it follows displacements rigidly without changing its form , reflecting the fact that physics should not depend on a certain choice of the origin .
the same holds true for any linear canonical transformation .
finally , the overlap of two density operators is proportional to the integral of the associated wigner functions : @xmath64 this property ( known as traciality ) offers practical advantages , since it allows one to predict the statistics of any outcome , once the wigner function of the measured state is known .
however , the wigner function can take on negative values , a property which distinguishes it from a true probability distribution .
indeed , this negativity is associated with the existence of quantum interference , which itself may be identified as a signal of nonclassical behavior @xcite .
the characterization of quantum states that are classical , in the sense of giving rise to nonnegative wigner functions , is a topic of undoubted interest . among pure states
, it was proven by hudson @xcite that the only states that have nonnegative wigner functions are gaussian states @xcite .
coherent states are closely linked with the notion of gaussian states .
the displacements constitute a basic ingredient for their definition : indeed , if we choose a fixed normalized reference state @xmath65 , we have @xcite @xmath66 so they are parametrized by phase - space points .
these states have a number of remarkable properties inherited from those of @xmath67 .
in particular , @xmath21 transforms any coherent state in another coherent state : @xmath68 \ , | q + q^{\prime } , p + p^{\prime } \rangle \ , .
\label{eq : comcoh}\ ] ] the standard choice for the fiducial vector @xmath69 is the vacuum @xmath70 .
this has quite a number of relevant properties , but the one we want to stress for what follows is that @xmath71 is an eigenstate of the fourier transform ( as they are all the fock states ) @xcite .
in fact , over this apparently trivial property rests a huge amount of physical knowledge .
so , @xmath72 is taken as the gaussian @xmath73 in appropriate units .
in addition , this wave function , as any coherent state , represents a minimum uncertainty state , namely @xmath74 where @xmath75 and @xmath76 are the corresponding variances . as a last illustration of the role played by the displacement operators , we note that their completeness allows us to write for any observable @xmath27 @xmath77 \ , { \hat { d } } ( q , p ) \ , dq dp \ , .\ ] ] evaluating the trace in terms of the set of eigenvectors of the rotated quadratures @xmath78 we obtain @xcite @xmath79 where @xmath80 is the kernel @xmath81 the tomograms @xmath82 are measured in a homodyne detector .
this shows that these rotated quadratures provide a complete quorum to reconstruct the density operator and hence the wigner function .
in this section we trace the changes required when we replace the cartesian position by a periodic angular position @xmath83 . for definiteness , we take the window @xmath84 in further considerations . the canonical conjugate variable , denoted now as @xmath85 , is the component of the angular momentum along the axis orthogonal to the rotation plane .
while classically a point particle is necessarily located at a single value of the angle @xmath86 , the corresponding quantum wave function is an object extended around @xmath1 and can be directly affected by the nontrivial topology .
one may be tempted to think that angular position should stand in the same relationship to angular momentum as ordinary position stands to linear momentum .
this would prompt to use the commutation relation @xmath87 = i \ , { \hat { { \leavevmode\hbox{\small1\normalsize\kern-.33em1}}}}$ ] and interpret the angle operator as multiplication by @xmath86 , while @xmath88 is the differential operator @xmath89 .
nevertheless , the use of this operator may entail many pitfalls for the unwary and needs a very subtle analysis @xcite .
as stressed in the introduction , a long discussion about the properties of this angle operator turns out to be unnecessary to settle a phase - space description .
indeed , let us denote by @xmath90 the basis of angular - momentum eigenstates . in principle , the corresponding eigenvalues can be arbitrary .
however , when some invariance principles hold @xcite , we can restrict our attention to integer values . in this case
, the states @xmath91 constitute a basis of states with well - defined angle , so that @xmath92 , where @xmath93 represents the periodic delta function ( or dirac comb ) of period @xmath94 @xcite , and they allow for a resolution of the identity . denoting the wave function components in both bases as @xmath95 and @xmath96
, we have @xmath97 which translates the fourier relationship inherent to any canonical pair .
to proceed further , we restrict ourselves to the exponentiated version of the canonical pair @xmath98 which , in addition , are experimentally feasible operations @xcite .
we have the action @xmath99 so that they represent displacements along the coordinate axes of the cylinder @xmath0 . throughout
, the angle addition and subtraction must be understood modulo @xmath94 . in dealing with this pair ,
it is customary also to take as the fundamental angular variable the complex exponential of the angle @xmath100 instead of the angle itself . the weyl form for this pair reads as @xmath101 and following the ideas of section [ sec : pscv ] , the displacement operators can be introduced as @xmath102 \ , { \hat { v } } ( \ell ) { \hat { u } } ( \phi ) \ , , \ ] ] where @xmath103 is a phase . the reasonable condition @xmath104 imposes @xmath105 = \exp ( -i \ell \phi ) \ , .\ ] ] we can not rewrite now equation ( [ eq : displace1 ] ) as an entangled exponential . these displacement operators form a complete trace - orthonormal set @xmath106 = 2 \pi \ , \delta_{\ell \ell^\prime } \ , \delta_{2\pi}(\phi - \phi^\prime ) \ , , \ ] ] whose resemblance with the relation ( [ eq : hwdisportho ] ) is evident .
one can also introduce the stratonovich - weyl quantizer as @xmath107 \
, { \hat { d } } ( \ell^\prime , \phi^\prime ) \ , d\phi^\prime \ , , \ ] ]
so that it fulfills the covariance condition @xmath108 note carefully that @xmath109 can not be identified now with the parity operator on the cylinder @xmath110 as it is the case for continuous variables .
this is due to the fact that @xmath111 do not constitute an operator basis .
one gets a basis if one supplements @xmath112 by @xmath113 ( or @xmath114 ) , @xmath115 being the exponential of the angle ( [ eq : defe ] ) .
the symbol of an operator @xmath27 turns out to be @xmath116 \
, , \ ] ] while the inversion is given by @xmath117 as explained by plebaski and coworkers @xcite , this wigner - weyl correspondence on the cylinder directly distinguishes between the `` classical '' phase space @xmath118 from the `` quantum '' phase space @xmath119 : only the former admits a formalism based on the use of symbols .
we can also look for the form of the star product . in close analogy with equation
, we have the integral form @xmath120 \ , b(\ell + \ell^{\prime \prime } , \phi + \phi^{\prime \prime } /2 ) \
, d\phi^{\prime } d\phi^{\prime \prime } \ , .\ ] ] expanding the functions @xmath39 and @xmath40 in a formal taylor series at @xmath121 , this result can also be rewritten in a `` differential '' form [ compare with equation ( [ eq : stardiff ] ) ] @xmath122 where @xmath123 is the counterpart of the poisson operator ( [ eq : poiss ] ) on the cylinder @xmath124 here @xmath125 is some sort of a continuous interpolation between discrete translations , and is defined through @xmath126 \,\tilde f(\phi ) \ , d\phi \ , , \ ] ] @xmath127 being the fourier transform of the function @xmath128 . with this star product
, we can introduce a moyal bracket as in equation ( [ eq : moyalbrac ] ) and the time evolution of dynamical variables is also given by ( [ eq : evoleq ] ) . to conclude , some remarks concerning the phase @xmath129 in ( [ eq : displace1 ] ) are in order .
although many of the results in this paper are independent of @xmath130 , a judicious choice of this factor may greatly facilitate calculations .
the periodicity in @xmath86 requires @xmath131 [ apart from ( [ eq : phaseconditionalpha ] ) ] . obviously , there are many admissible functions . for example , quadratic definitions like @xmath132 = \exp[-i \ell ( \ell + 1 ) \phi /2]$ ] or @xmath133 $ ] are legitimate , but we discard them because they introduce a lot of difficulties when computing practical wigner functions .
another possibility is @xmath134 = \exp ( -i \ell \phi /2 ) \ ,
i \phi \ , [ 1 - ( -1)^\ell ] / 4 \ } \ , , \ ] ] leading to the wigner kernel @xmath135 which , as discussed before , contains the complete set @xmath112 and @xmath113 .
in addition , this option brings the advantage of very easy calculations , especially for states given in the angular momentum representation .
however , in this case the term @xmath136 in eq .
( [ eq : alternativekernel ] ) is not parity invariant .
since the phase space should not have a `` preferred direction '' , we will also discard this definition .
we finally fix the phase @xmath137 as @xmath138 = \exp ( -i \ell \phi /2 ) \
, , \ ] ] which may lead to slightly more involved calculations than ( [ eq : alternativephases ] ) , but has the clear advantage of a parity - invariant kernel .
we stress that this definition of @xmath137 is related to the choice of the @xmath94 window for @xmath86 , so it is exclusively valid for @xmath139 $ ] .
since the wigner function is the symbol associated with the density matrix , we have the mapping @xmath140 \ , , & \nonumber \\ & & \label{eq : wigfundef1 } \\ & \displaystyle { \hat { \varrho } } = 2\pi { \sum_{{\ell } \in \mathbb{z } } } \int_{- \pi}^{\pi } { \hat { w}}(\ell,\phi ) \ , w_{{\hat { \varrho}}}(\ell,\phi ) \ , d\phi \ , .
& \nonumber\end{aligned}\ ] ] to obtain an operational form of this wigner function , we note that with the phase definition ( [ eq : alphaexplicit ] ) , the quantizer kernel ( [ eq : wigkerdef1 ] ) becomes @xmath141 \exp[i ( \ell \phi^{\prime } - \ell^{\prime } \phi ) ] \ , |\ell^{\prime \prime } \rangle \langle \ell^{\prime \prime } - \ell^{\prime}| \ , d\phi^\prime \ , .
\end{aligned}\ ] ] because of the contribution with a half - integer exponent , to sum over @xmath142 one has to split this sum in two different parts according @xmath142 is either even or odd .
after some manipulations , one gets @xmath143 \ , \langle\ell + \ell^{\prime \prime}-\ell^{\prime}| { \hat { \varrho } } |\ell + \ell^{\prime \prime}+\ell^{\prime}+1\rangle \ , .
\end{aligned}\ ] ] this expression contains exclusively angular - momentum matrix elements of the density operator and it is involved in practical computations due to the second line , which presents a logarithmically slow convergence due to the fact that for odd @xmath144 , the phase ( [ eq : alphaexplicit ] ) as a function of @xmath86 has a discontinuity at @xmath145 .
incidentally , this second line would be greatly simplified by employing the alternative kernel ( [ eq : alternativekernel ] ) .
sometimes , it is preferable to work in the angle representation , which can be obtained by introducing a resolution of the unity in equation ( [ eq : wigkerdef1 ] ) . in this way , we get @xmath146 which coincides with the form introduced in the pioneering work by mukunda @xcite .
this wigner function reproduces the proper marginal distributions @xmath147 and it is explicitly covariant under displacements on the cylinder .
this confirms the good properties of this approach .
note , in closing , that the time evolution of this wigner function is given by a liouville - von neumann equation analogous to ( [ eq : evoleqw ] ) .
now , in practical calculations we shall need the following map , which can be easily checked @xmath148 again , we have taken as the fundamental angular variable the complex exponential of the angle instead of the angle itself .
to further appreciate these ideas , we present a few relevant examples . for an angular momentum eigenstate @xmath149 and for an angle eigenstate @xmath150 , one has @xmath151 in the first case , the wigner function is flat in @xmath86 and the integral over the whole phase space gives the unity , reflecting the normalization of @xmath152 , while in the second , it is flat in the conjugate variable @xmath144 , and thus , the integral over the whole phase space diverges , which is a consequence of the fact that the state @xmath153 is not normalizable .
next , we consider coherent states
. the standard approach due to perelomov @xcite does not work for the cylinder .
however , much as in ( [ eq : defcs ] ) , we can introduce them as @xmath154 the choice of the fiducial vector is not so obvious .
we take @xcite @xmath155 where @xmath156 denotes the third jacobi theta function @xcite , which incidentally plays the role of the gaussian in circular statistics . in this way ,
the states @xmath157 inherit the properties of @xmath158 and they turn out to be equivalent to the ones defined in @xcite by an eigenvalue equation or in @xcite via a zak transform .
the wigner function for the coherent state @xmath159 splits as @xmath160 . the `` even ''
part turns out to be @xmath161 this seems a sensible result , since it is a discrete gaussian in the variable @xmath144 , and the equivalent for a gaussian for the continuous angle @xmath86 it .
however , the `` odd '' contribution spoils this simple picture : @xmath162}{2\pi^2 \vartheta_3 \left ( 0 \big | \frac{1}{e } \right ) } \vartheta_3\left ( \phi- \phi_0 + i/2 \big|\frac{1}{e } \right ) \ , { \sum_{{\ell^\prime } \in \mathbb{z } } } ( -1)^{\ell^\prime - \ell + \ell_0 } \frac{\exp(-\ell^\prime { } ^2 - \ell^\prime)}{\ell^\prime + \ell_0 - \ell + 1/2 } \ , .
\label{eq : examplewoddcoh}\ ] ] and @xmath163 .
the cylinder extends vertically from @xmath164 to @xmath165 .
the two corresponding marginal distributions are shown . in the right panel
we show the unwrapped version of the wigner function . ] in figure [ fig : coherentstandard ] , the wigner function for the coherent state @xmath166 is plotted .
a pronounced peak at @xmath167 for @xmath168 and slightly smaller ones for @xmath169 can be observed .
the associated marginal distributions are also plotted .
they are strictly positive , as correspond to true probability distributions .
for quantitative comparisons , however , sometimes it may be convenient to `` cut '' this cylindrical plot along a line @xmath86=constant and unwrap it .
this is also shown in figure [ fig : coherentstandard ]
. a closer look at this figure reveals also a remarkable fact : for values close to @xmath170 and @xmath171 , the wigner function takes negative values .
actually , a numeric analysis suggests the existence of negativities close to @xmath145 for any odd value of @xmath144 . as our last example
, we address the superposition @xmath172 of two angular - momentum eigenstates with a relative phase @xmath173 .
the wigner function splits again ; now the `` even '' part reads as @xmath174 \ } \ , .\ ] ] for the `` odd '' part , the diagonal contributions vanish , and one has @xmath175 \ , \frac{(-1)^{\ell+(\ell_1+\ell_2 -1)/2}}{\ell_1 + \ell_2 - 2\ell } \delta_{\ell_1 + \ell_2 = \mathrm{odd } } \ , , \label{eq : examplessuperodd2}\ ] ] where @xmath176 indicates that the sum is nonzero only when @xmath177 is odd . in consequence ,
when @xmath178 is odd , the interference term contains contributions for any @xmath144 , damped as @xmath179 .
when @xmath180 is an even number , the contribution ( [ eq : examplessuperodd2 ] ) vanishes and we have three contributions : two flat slices coming from the states @xmath181 and @xmath182 and an interference term located at @xmath183 . with @xmath184 for @xmath164 to @xmath165 ( left ) and @xmath185 for @xmath164 to @xmath186 .
] these features are nicely illustrated in figure [ fig : eventornillo ] .
the state @xmath187 is plotted for @xmath188 and @xmath189 and and @xmath188 and @xmath190 . changing the relative phase @xmath191 results in a global rotation of the cylinder . in can be observed in that the two rings at @xmath192 and @xmath193 ( as opposed to the rings at @xmath194 ) , are not flat in @xmath86 , but show a weak dependence on the angle due to the odd contributions added to the flat kronecker deltas . in view of these results , one can wonder what are the pure state for which the wigner function is nonnegative .
the answer was found quite recently @xcite : the wigner function @xmath195 is nonnegative if and only if @xmath196 is an angular momentum eigenstate @xmath152 .
the proof is rather technical and we skip it . the important point we wish to stress is that while for the continuous case the notions of coherent states , gaussian wave packets , and states with nonnegative wigner functions are completely equivalent , special care must be paid in extending these ideas to other physical systems , like angle - angular momentum , since they lose their equivalence . to round up this section , we illustrate the benefits of our phase - space approach by dealing with the evolution of the `` quantum '' pendulum @xmath197 which has been proposed as a good candidate to describe the evolution of the wave function of josephson junctions @xcite .
the corresponding weyl symbol is @xmath198 using the relations , we get the evolution equation for the wigner function @xmath199 w ( \ell , \phi ) \ , .\ ] ] we can directly check that @xmath200 , so ( [ eq : wevolutionexact ] ) becomes @xmath201 \right ) \ , .\ ] ] we are interested in the semiclassical evolution , i.e. , for states whose angular momentum components are sufficiently concentrated around a certain value @xmath202 with @xmath203 .
thus , we may view @xmath204 as a derivative respect to @xmath144 , so we finally get @xmath205 whose formal solution is @xmath206 in terms of the `` classical '' trajectories @xmath207 this constitutes a nice solution for an involved problem . to complete our theory we propose a reconstruction scheme for observables in this cylindrical phase space . the equivalent version of ( [ eq : dtom ] ) reads now , particularized for the density operator , @xmath208 where @xmath209/(2\pi)$ ] . in terms of @xmath210 ,
the wigner function is @xmath211 \ , \varrho ( \ell^\prime , \phi^\prime ) \ ,
d\phi^\prime \ , .\ ] ] a reconstruction of this wigner function is thus tantamount to finding the coefficients @xmath212 . for @xmath213 , these coefficients @xmath214
are simply the fourier transform of the angular - momentum spectrum . for @xmath215
, we assume that we are able to experimentally determine @xmath216 transformations on the input state followed by angular projections , that is , @xmath217 one can check that @xmath218 so the measurement of @xmath219 allows for the determination @xmath210 and hence the full reconstruction of the wigner function via equation ( [ eq : wig11 ] ) . as a rather simple yet illustrative example ,
let us note that for the vortex state @xmath149 , @xmath216 is diagonal , so the tomograms @xmath220 are independent of @xmath86 and @xmath144 and all of them equal to @xmath221 . performing the integration
we obtain precisely the wigner function in equation ( [ eq : exampleoamstaleell ] ) .
the feasibility of the proposed scheme relies on two crucial points .
first , the implementation of the @xmath216 transformation , which corresponds to a free rotor .
this has been used to describe the evolution in a variety of situations .
second , we need to assess the measurement of the angular spectrum , which can be done only approximately . though the implementation of this scheme may differ depending on the system under considerations , our formulation provides a common theoretical framework on the hilbert space generated by the action of angle and angular momentum .
a experimental demonstration of the method in terms of optical beams has been recently established @xcite . for continuous variables ,
it is possible to make a direct sampling of the wigner function by a scheme that determines the parity @xcite .
this is an unexplored territory for the cylinder that is worth investigating .
in summary , we have shown how to extend in a consistent way all the techniques developed for a continuous - variable phase space to the case of angle and angular momentum .
while we have not left aside the mathematical details , our main emphasis has been on presenting a simple and useful toolkit that any practitioner in the field should master . in our view , far from being an academic curiosity , the ideas expressed here have a wide range of potential applications in numerous hot topics .
this paper has greatly benefited from the criticism and advice of prof .
the work was supported by the spanish research directorate ( grants fis2005 - 06714 and fis2008 - 04356 ) , the ucm - bcsh program ( grant gr-920992 ) , the mexican conacyt ( grant 106525 ) , the czech ministry of education ( project msm6198959213 ) , and the czech ministry of industry and trade ( project fr - ti1/364 ) .
82 natexlab#1#1[2]#2 , , , , . , ( ) . ,
( ) . , , ( ) . , ( ) . , ( ) . , , ( ) . , , ( ) . , , ( ) . , , ( ) . , ( ) . , , ( ) . , , , , ( ) . , ( ) . , , , , .
, , ( eds . ) , , , , . , ( ) . , , ( ) . , , ( ) . , ( ) . , , , ( ) . , ( ) . , , , ( ) . , ( ) . , , , ( ) . , , , , ( ) . , , , , ( ) . , , ( ) . , ( ) . , ( ) . , , , ( ) . , , , , , , . ( ed . ) , , , , . , , , ( ) . , , , ( ) . , , , ( ) . , , , , , , , , ( ) . , ,
( ) . , , , , , , . , , ( ) . ,
( ) . , , , , , ( ) . ,
( ) . , , ,
, ( ) . , , ( ) . , , ( )
, , , ( ) . , , ( ) .
, , , , ( ) . , , , , , ( ) . , ( ) .
, , , , , ( ) . , , , , , .
, , , , , .
, , ( ) . ,
( ) . , , , , . ,
( ) . , ( eds . ) , , volume of _ _ , , , . , , ( ) . , , , , . , , ph.d . thesis ,
university of utrecht , .
, , , edition , .
, , , , , ( ) .
, , , , , ( ) . , , , , ( ) .
, , , , , ( ) . , , , ( ) .
, , , , , , ( ) . , , , ( ) .
, , , , , , , , ( ) . | a comprehensive theory of the weyl - wigner formalism for the canonical pair angle - angular momentum is presented .
special attention is paid to the problems linked to rotational periodicity and angular - momentum discreteness . |
the study of permutations by descent sets is a natural generalization of the study of permutations by number of descents : the study of eulerian numbers .
let @xmath0 be the symmetric group on @xmath2 elements .
we think of permutations in @xmath3 as bijections @xmath4 \to [ n],\ ] ] where @xmath5 $ ] denotes the set @xmath6 . for any permutation @xmath7
, we say @xmath8 has a _ descent _ in position @xmath9 if @xmath10 .
define the set @xmath11 and let @xmath12 denote the number of elements in @xmath13 .
we call @xmath13 the _ descent set _ of @xmath8 , and @xmath12 the _ descent number _ of @xmath8 . for example , the permutation @xmath14 has descent set @xmath15 and descent number 2
. the number of permutations of @xmath5 $ ] with descent number @xmath16 is denoted by the eulerian number @xmath17 , and we recall that the eulerian polynomial is defined as @xmath18 for each subset @xmath19 of @xmath20 $ ] , let @xmath21 denote the sum , in the group algebra @xmath22 $ ] , of all permutations with descent set @xmath19 .
louis solomon @xcite showed that the linear span of the @xmath21 forms a subalgebra of the group algebra , called the _ descent algebra_. more generally , he showed that one can define a descent algebra for any coxeter group . for now consider the descent algebra of the symmetric group .
this descent algebra was studied in great detail by adriano garsia and christophe reutenauer @xcite .
jean - louis loday @xcite proved the existence of a commutative , semisimple subalgebra of solomon s descent algebra .
sometimes called the
eulerian subalgebra , " it is defined as follows . for @xmath23 , let @xmath24 be the sum of all permutations in @xmath0 with descent number @xmath25 . then the eulerian descent algebra is the linear span of the @xmath26 .
define @xmath27 which we refer to as the structure polynomial . "
notice that the structure polynomial is a polynomial in @xmath28 , with coefficients in @xmath29 $ ] , of degree @xmath2 and with no constant term .
( in other words , it has exactly as many nonzero terms as there are possible descent numbers . )
the eulerian subalgebra is described by the following multiplication of structure polynomials : [ thm : ges ] as polynomials in @xmath28 and @xmath30 with coefficients in the group algebra , we have @xmath31 define elements @xmath32 in the group algebra by @xmath33 . by examining the coefficients of @xmath34 in ,
it is clear that the @xmath32 are orthogonal idempotents : @xmath35 and @xmath36 if @xmath37 .
this formulation is essentially an unsigned version of loday s thorme 1.7 ( see @xcite ) .
in particular , for @xmath38 , we have @xmath39 , where loday defines @xmath40 and @xmath41 means that we forget the signs on all the summands . as an interesting aside
, we remark that theorem [ thm : ges ] has connections to the well - known card shuffling analysis of dave bayer and persi diaconis @xcite .
if we define @xmath42 then @xmath43 is the generating function for the probability distribution for one shuffle of a deck of @xmath2 cards ( as defined in @xcite ) .
we use the multiplication rule given by theorem [ thm : ges ] to compute the probability distribution after @xmath44 ( independent ) shuffles of the deck ( it is not hard to show that @xmath45 obeys the same multiplication rule as @xmath46 ) : @xmath47 which gives theorem 1 from @xcite . theorem [ thm : ges ] can be proved in several ways , but we will focus on one that employs richard stanley s theory of @xmath1-partitions .
more specifically , the approach taken in this paper follows from work of ira gessel
theorem [ thm : ges ] is in fact an easy corollary of theorem 11 from @xcite .
later we will give a proof of theorem [ thm : ges ] that derives from gessel s work and then extend this method to prove the main results of this paper .
while equivalent results can be found elsewhere in the literature , the @xmath1-partition approach is self - contained and rather elementary .
further , it often yields @xmath48-analogs for formulas like quite naturally .
now we introduce another descent algebra based on a modified definition of descent . for a permutation @xmath49
we define a _ cyclic descent _ at position @xmath9 if @xmath10 , or if @xmath50 and @xmath51 .
define @xmath52 to be the set of cyclic descent positions of @xmath8 , called the _
cyclic descent set_. let the _
cyclic descent number _ , @xmath53 ,
be the number of cyclic descents .
the number of cyclic descents is between 1 and @xmath54 .
one can quickly observe that a permutation @xmath8 has the same number of cyclic descents as both @xmath55 and @xmath56 for @xmath57 , where @xmath58 is the @xmath2-cycle @xmath59 .
define the _ cyclic eulerian polynomial _ to be @xmath60 where @xmath61 is the number of permutations with cyclic descent number @xmath16 .
we can make the following proposition ( observed by jason fulman in @xcite , corollary 1 ) .
[ prp : eul ] the cyclic eulerian polynomial is expressible in terms of the ordinary eulerian polynomial : @xmath62 we will compare the coefficient of @xmath63 on each side of the equation to show @xmath64 .
let @xmath65 be any permutation of @xmath20 $ ] such that @xmath66 .
let @xmath67 be the permutation defined by @xmath68 for @xmath69 and @xmath70 .
then we have @xmath71 and @xmath72 .
let @xmath73 , the set consisting of all @xmath2 cyclic permutations of @xmath74 .
every permutation in the set has exactly @xmath75 cyclic descents .
there is a bijection between permutations of @xmath76 and such subsets of @xmath3 given by the map @xmath77 and so the proposition follows .
let @xmath78 be the sum in the group algebra of all those permutations with @xmath9 cyclic descents .
then we define the cyclic structure polynomial @xmath79 notice that @xmath80 is a polynomial of degree @xmath54 with no constant term , giving as many nonzero terms as possible cyclic descent numbers .
paola cellini studied cyclic descents more generally in the papers @xcite , @xcite , and @xcite .
though her approach is quite different from the one taken in this paper , from her work and loday s we can derive the following theorem .
[ thmcyc ] as polynomials in @xmath28 and @xmath30 with coefficients in the group algebra of the symmetric group , we have @xmath81 now if we define elements @xmath82 by @xmath83 , we see that @xmath84 and @xmath85 if @xmath37 . therefore the elements @xmath82 are orthogonal idempotents .
fulman relates cyclic descents to card shuffling ( now we consider cuts as well as shuffles of the deck ) ; theorem 2 of @xcite is closely related to theorem [ thmcyc ]
. we will prove theorem [ thmcyc ] in section [ sec : cycprf ] using formula .
that the multiplication for the cyclic structure polynomials is so similar to that of the ordinary structure polynomials is no accident .
indeed , it will be clear from the proof of theorem [ thmcyc ] that the map @xmath86 where @xmath87 is as in the proof of proposition [ prp : eul ] , gives an isomorphism of algebras between the ordinary eulerian descent algebra of @xmath88 $ ] and the cyclic eulerian descent algebra of @xmath89 $ ] .
let @xmath90 $ ] denote the set @xmath91 .
let @xmath92 denote the hyperoctahedral group , the group of all bijections @xmath93
\to \pm[n]$ ] with the property that @xmath94 , for @xmath95 . since the elements of the hyperoctahedral group are uniquely determined by where they map @xmath96
, we can think of them as signed permutations . for a signed permutation @xmath97 we will write @xmath98 . in moving from the symmetric group to the hyperoctahedral group
, we define the _ descent set _ @xmath13 of a signed permutation @xmath99 to be the set of all @xmath100 such that @xmath101 , where we always take @xmath102 .
the _ descent number _ of @xmath8 is again denoted @xmath12 and is equal to the cardinality of @xmath13 . as a simple example , the signed permutation @xmath103 has descent set @xmath104 and descent number @xmath105 .
there is an eulerian subalgebra for the hyperoctahedral group , previously studied by cellini @xcite , @xcite , francois and nantel bergeron @xcite , @xcite , n. bergeron @xcite , and fulman @xcite . for @xmath106 let @xmath24 be the sum of all permutations in @xmath92 with @xmath25 descents .
define the type b structure polynomial @xmath107 chak - on chow @xcite was able to use the theory of @xmath1-partitions to prove the following .
[ thm : cho ] as polynomials in @xmath28 and @xmath30 with coefficients in the group algebra of @xmath108 , we have @xmath109 or upon substituting @xmath110 @xmath111 then @xmath112 we therefore have orthogonal idempotents @xmath32 defined by @xmath113 ( notice that shifting the polynomial by 1/2 gives a nontrivial constant term ) .
just as theorem [ thm : ges ] was equivalent to thorme 1.7 of @xcite , theorem [ thm : cho ] is equivalent to theorem 4.1 from @xcite .
further , both fulman @xcite and bergeron and bergeron @xcite made connections to signed " card shuffling ( flip a card face up or down to change sign ) . for a permutation @xmath97
, we define a _ type b cyclic descent _ , or
_ augmented descent _ at position @xmath9 if @xmath10 or if @xmath114 and @xmath115 .
if we consider that signed permutations always begin with 0 , then augmented descents are the natural generalization of type a cyclic descents .
the set of all augmented descent positions is denoted @xmath116 , the _ augmented descent set_. it is the ordinary descent set of @xmath8 along with @xmath2 if @xmath117 .
the _ augmented descent number _ , @xmath118 , is the number of augmented descents . with these definitions ,
@xmath103 has augmented descent set @xmath119 and augmented descent number @xmath120 . note that while @xmath121 , @xmath122 , and @xmath123 .
denote the number of signed permutations in @xmath108 with @xmath16 augmented descents by @xmath124 and define the _ augmented eulerian polynomial _ as @xmath125 in section [ sec : def ] we prove the following relation using the theory of @xmath1-partitions .
[ prp : augeul ] the number of signed permutations with @xmath126 augmented descents is @xmath127 times the number of unsigned permutations with @xmath9 descents , @xmath128 . in other words , @xmath129 .
define the type b cyclic structure polynomial as @xmath130 where @xmath131 is the sum of all signed permutations with @xmath9 augmented descents .
[ conj1 ] as polynomials in @xmath28 and @xmath30 with coefficients in the group algebra of the hyperoctahedral group we have @xmath132 or upon substituting @xmath133 @xmath134 then @xmath135 we get orthogonal idempotents @xmath136 defined by @xmath137
. we will prove theorem [ conj1 ] in section [ sec : prf ] using modified types of @xmath1-partitions called _ augmented _ @xmath1-partitions .
also in section [ sec : prf ] we prove the following theorem for which we have no type a equivalent .
[ thm : hypermod ] as polynomials in @xmath28 and @xmath30 with coefficients in the group algebra of the hyperoctahedral group we have @xmath138 or upon substituting @xmath133 @xmath111 then @xmath139 this formula tells us that @xmath140 and that @xmath141 if @xmath37 .
in other words , the augmented eulerian descent algebra is an ideal in the subalgebra of the group algebra formed by the span of the @xmath142 and the @xmath136 .
this relationship shows up again in the case of peak algebras of type a , and forms the basis of future work as discussed in section [ sec : fut ] .
see the paper of marcelo aguiar , n. bergeron , and kathryn nyman @xcite for more .
we also point out here that when taken together , theorems [ thm : cho ] , [ conj1 ] , and [ thm : hypermod ] , imply cellini s theorem a from @xcite .
indeed , for positive integers @xmath16 , cellini s elements @xmath143 are equivalent to @xmath144 , and the @xmath145 are equivalent to @xmath146 . here , as in section [ sec : sym ] , @xmath147 and @xmath148 are the same as @xmath149 and @xmath150 except that we replace the coefficient of @xmath8 with the coefficient of @xmath151 .
see also lemma 4 of @xcite .
throughout this paper we will use @xmath1 to denote a partially ordered set , or poset . the partial order on @xmath1
is denoted @xmath152 , or simply @xmath153 when the meaning is clear .
our posets will always be finite and for a poset of @xmath2 elements , the elements will be distinctly labeled by the numbers @xmath96 .
[ def : p ] let @xmath154 be a countable , totally ordered set . for a given poset @xmath1 , a _
_ @xmath1-partition _ _ if @xmath155
. however , in @xcite the maps of definition [ def : p ] are called ( perhaps misleadingly ) _ reverse _ @xmath1-partitions . ]
is a function @xmath156 \to x$ ] such that : * @xmath157 if @xmath158 * @xmath159 if @xmath158 and @xmath160 in @xmath161 for our purposes we usually think of @xmath162 as a subset of the positive integers .
let @xmath163 denote the set of all @xmath1-partitions .
when @xmath162 has finite cardinality @xmath16 , then the number of @xmath1-partitions is finite . in this case , define the _ order polynomial _
, denoted @xmath164 , to be the number of @xmath1-partitions
@xmath165 \to x$ ]
. we will consider any permutation @xmath49 to be a poset with the total order @xmath166 .
for example , the permutation @xmath167 has @xmath168 as a poset . with this convention , the set of all @xmath8-partitions is easily characterized .
the set @xmath169 is the set of all functions @xmath156
\to x$ ] such that ( if we take @xmath162 to be the positive integers ) @xmath170 and whenever @xmath171 , then @xmath172 , @xmath173 .
the set of all @xmath8-partitions where @xmath174 is all maps @xmath175 such that @xmath176 .
[ h ] .,title="fig : " ] for a poset @xmath1 of order @xmath2 , let @xmath177 denote the set of all permutations of @xmath5 $ ] which extend @xmath1 to a total order .
this set is sometimes called the set of linear extensions " of @xmath1 . for example
let @xmath1 be the poset defined by @xmath178 . in linearizing " @xmath1 we form a total order by retaining all the relations of @xmath1 but introducing new relations so that any element is comparable to any other . in this case ,
1 and 2 are not comparable , so we have exactly two ways of linearizing @xmath1 : @xmath179 and @xmath180 .
these correspond to the permutations @xmath181 and @xmath182 .
let us make the following observation .
[ ob1 ] a permutation @xmath8 is in @xmath177 if and only if @xmath183 implies @xmath184 . in other words , if @xmath9 is below " @xmath185 in the hasse diagram of the poset @xmath1 , it had better be below @xmath185 in any linear extension of the poset .
we also now prove what is sometimes called the fundamental theorem ( or lemma ) of @xmath1-partitions .
the set of all @xmath1-partitions of a poset @xmath1 is the disjoint union of the set of @xmath8-partitions of all linear extensions @xmath8 of @xmath1 : @xmath186 the proof follows from induction on the number of incomparable pairs of elements of @xmath1 .
if there are no incomparable pairs , then @xmath1 has a total order and already represents a permutation .
suppose @xmath9 and @xmath185 are incomparable in @xmath1 .
let @xmath187 be the poset formed from @xmath1 by introducing the relation @xmath188 .
then it is clear that @xmath189 .
we continue to split these posets ( each with strictly fewer incomparable pairs ) until we have a collection of totally ordered chains corresponding to distinct linear extensions of @xmath1 .
@xmath190 the fundamental theorem tells us that in order to study @xmath1-partitions for a given poset , we can focus on the @xmath8-partitions for its linear extensions a more straightforward task . in particular , counting @xmath8-partitions is not too difficult .
notice that for any permutation @xmath8 and any positive integer @xmath16 , @xmath191 where @xmath192 denotes the multi - choose " function the number of ways to choose @xmath193 objects from a set of @xmath194 objects with repetitions .
another interpretation of @xmath192 is the number of integer solutions to the set of inequalities @xmath195 with this in mind , @xmath196 is the same as the number of solutions to @xmath197 better still , we can say it is the number of solutions ( though not in general the same _ set _ of solutions ) to @xmath198 for example , the number of solutions to @xmath199 is the same as the number of solutions to @xmath200 or the solutions to @xmath201 . in general ,
if we take @xmath202 it is clear that @xmath175 is a @xmath8-partition if and only if @xmath203 is a solution to , which gives @xmath204 before moving on , let us point out that in order to prove that the formulas in this paper hold as polynomials in @xmath28 and @xmath30 , it will suffice to prove that they hold for all pairs of positive integers .
it is not hard to prove this fact , and we use it in each of the proofs presented in this paper
. we will now prove theorem [ thm : ges ] using the theory of @xmath1-partitions .
if we write out @xmath205 using the definition , we have @xmath206 if we equate the coefficients of @xmath8 we have @xmath207 clearly , if formula holds for all @xmath8 , then formula is true
. let us interpret the left hand side of this equation .
let @xmath208 , and @xmath209 be positive integers .
then the left hand side of equation ( [ eqst ] ) is just the order polynomial @xmath210 . to compute this order polynomial we need to count the number of @xmath8-partitions
@xmath165 \to x$ ] , where @xmath162 is some totally ordered set with @xmath211 elements .
but instead of using @xmath212 $ ] as our image set , we will use a different totally ordered set of the same cardinality .
let us count the @xmath8-partitions @xmath165\to [ l]\times[k]$ ] .
this is equal to the number of solutions to @xmath213 here we take the _ lexicographic ordering _ on pairs of integers .
specifically , @xmath214 if @xmath215 or else if @xmath216 and @xmath217 . to get the result we desire , we will sort the set of all solutions to into distinct cases indexed by subsets @xmath218 $ ] .
the sorting depends on @xmath8 and proceeds as follows .
let @xmath219 be any solution to .
for any @xmath173 , if @xmath220 , then @xmath221 , which falls into one of two mutually exclusive cases : @xmath222 if @xmath171 , then @xmath223 , which means either : @xmath224 also mutually exclusive . define @xmath225\setminus \operatorname{des}(\pi ) \,\mid\ , j_s > j_{s+1 } \ }
\cup \ { s \in \operatorname{des}(\pi ) \,\mid\ , j_s \geq j_{s+1 } \}$ ] .
then @xmath226 is the set of all @xmath227 such that either or holds for @xmath228 .
notice that in both cases , @xmath229 .
now for any @xmath218 $ ] , let @xmath230 be the set of all solutions @xmath228 to satisfying @xmath231 .
we have split the solutions of into @xmath232 distinct cases indexed by all the different subsets @xmath19 of @xmath20 $ ] .
. then we want to count the number of solutions to @xmath234 which splits into four distinct cases : @xmath235 we now want to count all the solutions contained in each of these cases and add them up . for a fixed subset @xmath19 we will use the theory of @xmath1-partitions to count the number of solutions for the set of inequalities first for the @xmath236 s and then for the @xmath237 s .
multiplying will give us the number of solutions in @xmath230 ; we do the same for the remaining subsets and sum to obtain the final result . for @xmath238 in the example above , we would count first the number of integer solutions to @xmath239 , with @xmath240 , and then we multiply this number by the number of solutions to @xmath241 to obtain the cardinality of @xmath242 .
we will now carry out the computation in general . for any particular @xmath243 $ ] , form the poset @xmath244 of the elements @xmath96 by @xmath245 if @xmath246 , @xmath247 if @xmath248 .
zig - zag " poset of @xmath2 elements labeled consecutively by @xmath249 , with downward zigs corresponding to the elements of @xmath19 . for example , if @xmath250 for @xmath251 , then @xmath244 has @xmath252 .
[ h ] for @xmath253$].,title="fig : " ] for any solution in @xmath230 , let @xmath156 \to [ k]$ ] be defined by @xmath254 for @xmath255 .
we will show that @xmath175 is a @xmath244-partition . if @xmath245 and @xmath220 in @xmath161 , then tells us that @xmath256 .
if @xmath245 and @xmath171 in @xmath161 , then tells us that @xmath257 . if @xmath258 and @xmath220 in @xmath161 , then gives us that @xmath259 . if @xmath247 and @xmath260 in @xmath161 , then gives us that @xmath261 . in other words , we have verified that @xmath175 is a @xmath244-partition .
so for any particular solution in @xmath230 , the @xmath236 s can be thought of as a @xmath244-partition .
conversely , any @xmath244-partition @xmath175 gives a solution in @xmath230 since if @xmath262 , then @xmath263 if and only if @xmath264 and @xmath229 for all @xmath265 .
we can therefore turn our attention to counting @xmath244-partitions .
let @xmath266 .
then for any @xmath267-partition @xmath175 , we get a chain @xmath268 with @xmath269 if @xmath270 .
the number of solutions to this set of inequalities is @xmath271 recall by observation [ ob1 ] that @xmath272 if @xmath245 , i.e. , if @xmath246 . if @xmath247 then @xmath273 and @xmath274 .
we get that @xmath275 if and only if @xmath276 .
set @xmath277 .
the number of solutions to @xmath278 is given by @xmath279 now for a given @xmath19 , the number of solutions in @xmath230 is @xmath280 summing over all subsets @xmath218 $ ] , we can write the number of all solutions to as @xmath281 and so we have derived formula .
we now have a taste of how @xmath1-partitions can be used .
we are ready to go on and prove theorem [ thmcyc ] .
if we write out the definition for @xmath80 in the statement of theorem [ thmcyc ] , multiply both sides by @xmath282 , and equate coefficients , we have for any @xmath49 , @xmath283 for some @xmath9 , we can write @xmath284 where @xmath58 is the @xmath2-cycle @xmath285 and @xmath286 . observe that @xmath287 .
form the permutation @xmath288 by @xmath289 , @xmath290 .
then we can see that @xmath291 .
we have @xmath292 now we can apply equation to give us @xmath293 for each pair of permutations @xmath294 such that @xmath295 , define the permutations @xmath296 as follows . for @xmath290 ,
let @xmath297 and @xmath298 .
let @xmath299 and @xmath300 .
then by construction we have @xmath301 and a quick observation tells us that @xmath302 and @xmath303 . on the other hand , from any pair of permutations @xmath304 such that @xmath301 , @xmath299 , we can construct a pair of permutations @xmath305 such that @xmath306 by reversing the process .
observe now that if @xmath307 and @xmath301 , then @xmath300 .
therefore we have that is equal to @xmath308 where the last two formulas hold for any @xmath309 $ ] .
notice that the number of cyclic descents of @xmath310 is still the same as the number of cyclic descents of @xmath311 .
we take the sum over all @xmath312 , yielding @xmath313 as desired .
here we present the definitions and basic results we will need to prove the remaining theorems .
proofs of some of these basic facts are identical to the proofs of analogous statements for ordinary @xmath1-partitions and may be omitted .
it bears mentioning that the following definitions for type b posets and type b @xmath1-partitions , though taken from chow @xcite , derive from earlier work by victor reiner @xcite . in @xcite
, reiner extends this concept to any coxeter group . a _
@xmath92 poset _ is a poset @xmath1 whose elements are @xmath314 such that if @xmath158 then @xmath315 .
note that if we are given a poset with @xmath316 elements labeled by @xmath317 where @xmath318 or @xmath319 , then we can extend it to a @xmath92 poset of @xmath320 elements .
[ h ] posets.,title="fig : " ] let @xmath321 be a countable , totally ordered set with total order @xmath322 then define @xmath323 to be the set @xmath324 with total order @xmath325 [ def : typebppart ] for any @xmath92 poset @xmath1 , a _
@xmath1-partition of type b _ is a function @xmath326 \to \pm x$ ] such that : * @xmath157 if @xmath158 * @xmath159 if @xmath158 and @xmath160 in @xmath161 * @xmath327 note that type b @xmath1-partitions differ from ordinary @xmath1-partitions only in the addition of the property @xmath327 .
let @xmath163 denote the set of all type b @xmath1-partitions .
we usually think of @xmath162 as a subset of the nonnegative integers , and when @xmath162 has finite cardinality @xmath328 , then the _ type b order polynomial _
, denoted @xmath164 , is the number of @xmath1-partitions @xmath329 \to \pm x$ ]
. we can think of any signed permutation @xmath97 as a @xmath92 poset with the total order @xmath166 , @xmath330 .
for example , the signed permutation @xmath103 has @xmath331 as a poset .
note that @xmath169 is the set of all functions
@xmath332 \to \pm x$ ] such that for @xmath333 , @xmath334 and @xmath335 where @xmath172 whenever @xmath171 , @xmath336 .
for example , the type b @xmath8-partitions where @xmath337 are all maps @xmath175 such that @xmath338 . as before , [ ob2 ] for a type b poset @xmath1 ,
a signed permutation @xmath8 is in @xmath177 if and only if @xmath183 implies @xmath184 .
we have a fundamental theorem for @xmath1-partitions of type b. the set of all type b @xmath1-partitions of a @xmath92 poset @xmath1 is the disjoint union of the set of @xmath8-partitions of all linear extensions @xmath8 of @xmath1 : @xmath339 @xmath190 [ h ] poset @xmath1.,title="fig : " ] similarly to the type a case , it is easy to compute the order polynomial @xmath340 for any permutation @xmath341 .
any @xmath8-partition @xmath332 \to \pm[k]$ ] is determined by where we map @xmath342 since @xmath327 ( and so @xmath343 ) . to count them
we can take @xmath344 , and look at the number of integer solutions to the set of inequalities @xmath345 this number is the same as the number of solutions to @xmath346 which we know to be @xmath347 .
we have @xmath348 now we have the tools to prove theorem [ thm : cho ]
. we will omit the details , since they are essentially the same as for theorem [ thm : ges ] .
the main difference is that we want to count @xmath8-partitions
@xmath332 \to \pm[l]\times\pm[k]$ ] .
we notice that because of the property @xmath334 , this is just like counting all @xmath156 \to \{0,1,\ldots , l\}\times \{-k,\ldots,-1,0,1,\ldots , k\}$ ] where for @xmath349 , @xmath350 with @xmath351 in the lexicographic order .
[ h ] with @xmath351.,title="fig : " ] this choice of image set @xmath162 has @xmath352 elements , and so for each @xmath8 we can count all these maps with @xmath353 .
we use similar arguments to those of theorem [ thm : ges ] for splitting the lexicographic solutions to @xmath354 once we have properly grouped the set of solutions it is not much more work to obtain the crucial formula : @xmath355 we now give the definition of an augmented @xmath1-partition and basic tools related to their study .
let @xmath356 be a countable , totally ordered set with a maximal element @xmath357 .
the total ordering on @xmath162 is given by @xmath358 define @xmath323 to be @xmath359 with the total order @xmath360 [ def : augp ] for any @xmath92 poset @xmath1 , an _ augmented @xmath1-partition _ is a function @xmath326 \to \pm x$ ] such that : * @xmath157 if @xmath158 * @xmath159 if @xmath158 and @xmath160 in @xmath161 * @xmath327 * if @xmath361 in @xmath161 , then @xmath362 .
note that augmented @xmath1-partitions differ from @xmath1-partitions of type b only in the addition of maximal and minimal elements of the image set @xmath323 and in the last criterion .
let @xmath363 denote the set of all augmented @xmath1-partitions .
when @xmath162 has finite cardinality @xmath328 ( and so @xmath323 has cardinality @xmath364 ) , then the _ augmented order polynomial _ , denoted @xmath365 , is the number of augmented @xmath1-partitions . for any signed permutation @xmath341 , note that @xmath366 is the set of all functions @xmath332
\to \pm x$ ] such that for @xmath333 , @xmath334 and @xmath367 and @xmath172 whenever @xmath171 .
in addition , we have @xmath368 whenever @xmath117 .
the set of all augmented @xmath8-partitions where @xmath337 is all maps @xmath175 such that @xmath369 .
we have a fundamental theorem for augmented @xmath1-partitions .
the set of all augmented @xmath1-partitions of a @xmath92 poset @xmath1 is the disjoint union of the set of @xmath8-partitions of all linear extensions @xmath8 of @xmath1 : @xmath370 @xmath371 as in our previous cases , it is fairly easy to compute the augmented order polynomial for totally ordered chains .
the number of augmented @xmath8-permutations @xmath332 \to \pm[k]$ ] is equal to the number of integer solutions to the set of inequalities @xmath372 therefore , just as with other types of order polynomials , we can express the augmented order polynomial as a binomial coefficient , @xmath373 to give one example of the usefulness of augmented @xmath1-partitions , we conclude this subsection with proof of proposition [ prp : augeul ] , claiming @xmath129 . from the general theory of @xmath1-partitions in stanley s book @xcite
, we have @xmath374 let @xmath1 be an antichain that is , a poset with no relations of @xmath2 elements .
then @xmath375 since each of the @xmath2 elements of @xmath1 is free to be mapped to any of @xmath16 places .
furthermore , @xmath376 , so we get the following equation , @xmath377 now let @xmath1 be the poset given by an antichain of @xmath320 elements labeled @xmath378 .
the number of augmented @xmath1-partitions @xmath332 \to \pm[k]$ ] is determined by the choices for @xmath379 , which can take any of the values in @xmath380 .
therefore @xmath381 . for @xmath108 posets @xmath1 , it is not difficult to show that we have the identity @xmath382 for our antichain we have @xmath383 , and therefore @xmath384 so the proposition is proved .
the following proofs will follow the same basic structure as the proof of theorem [ thm : ges ] , but with some important changes in detail . in both cases
we will rely on a slightly different total ordering on the integer points @xmath385 , where @xmath9 and @xmath185 are bounded both above and below .
let us now define the _ augmented lexicographic order_. [ h ] consider all points @xmath385 with @xmath386 , @xmath387 .
we have @xmath214 if @xmath388 or else if @xmath216 and @xmath217 as before , except in the important special case that follows .
we now say @xmath389 in one of two situations .
either @xmath390 or @xmath391 if we have @xmath392 , @xmath393 , then in augmented lexicographic order , the first few points @xmath394 are : @xmath395 to be more precise , what we have done is to form equivalence classes of points and to introduce a total order on these equivalence classes . if @xmath396 , then the class represented by @xmath385 is just the point itself . otherwise , the classes consist of the two points @xmath397 and @xmath398 .
when we write @xmath389 , what we mean is that the two points are in the same equivalence class . in the proofs that follow , it will be important to remember the original points as well as the equivalence classes to which they belong .
this special ordering will be very apparent in deriving the @xmath48-analogs of theorem [ conj1 ] and theorem [ thm : hypermod ]
. we will now prove theorem [ conj1 ] .
as before , we equate coefficients and prove that a simpler formula , @xmath399 holds for any @xmath97 .
we recognize the left - hand side of equation as @xmath400 , so we want to count augmented @xmath1-partitions @xmath329\to \pm x$ ] , where @xmath162 is a totally ordered set of order @xmath401 .
we interpret this as the number of solutions , in the augmented lexicographic ordering , to @xmath402 where we have * @xmath403 , * @xmath404 if @xmath405 , * @xmath406 if @xmath407 , and * @xmath408 if @xmath409 .
let us clarify .
there are @xmath410 points @xmath385 with @xmath411 and @xmath412 , not including the points @xmath413 with @xmath414 , or the points @xmath415 with @xmath416 . under the augmented lexicographic
ordering , @xmath417 of these points are identified : points of the form @xmath418 , for @xmath419 .
any particular @xmath420 can only occupy one of @xmath397 or @xmath398 , but not both .
so there are truly @xmath401 distinct classes in which the @xmath2 points can fall .
this confirms our interpretation of the order polynomial .
now as before , we will split the solutions to the inequalities into distinct cases .
let @xmath421 , @xmath422 , @xmath423 , and @xmath424 .
let @xmath219 be any solution to .
if @xmath220 , then @xmath425 , which falls into one of two mutually exclusive cases : @xmath426 if @xmath171 , then @xmath223 , which we split as : @xmath427 also mutually exclusive .
define @xmath428 .
then @xmath226 is the set of all @xmath227 such that either or holds for @xmath228 .
now for any @xmath429 , let @xmath230 be the set of all solutions @xmath228 to satisfying @xmath231 .
we have split the solutions of into @xmath430 distinct cases indexed by all the different subsets @xmath19 of @xmath431 .
however , @xmath432 is empty , since @xmath433 yields @xmath434 as discussed before , the augmented descent set of a signed permutation is never empty , so we would get @xmath435 , a contradiction . at the other extreme ,
the set @xmath436 has no solutions either .
here we get @xmath437 and consequently @xmath438 but @xmath116 can not equal @xmath431 , so we get the contradiction @xmath439 .
now let @xmath19 be any nonempty , proper subset of @xmath431 .
form the poset @xmath244 by @xmath247 if @xmath440 , @xmath245 otherwise .
the poset @xmath244 looks like a zig - zag , labeled consecutively by @xmath441 with downward zigs corresponding to the elements of @xmath19 . because @xmath19 is neither empty nor full , we never have @xmath442 , so @xmath244 is a well - defined , nontrivial type b poset .
for a given @xmath443 , let @xmath332 \to \pm [ k]$ ] be defined by @xmath254 and @xmath334 for @xmath95 .
we will show that @xmath175 is an augmented @xmath244 partition . if @xmath245 and @xmath220 in @xmath161 , then tells us that @xmath444 . if @xmath245 and @xmath260 in @xmath161 , then tells us that @xmath257 . if @xmath258 and @xmath220 in @xmath161 , then gives us that @xmath259 . if @xmath247 and @xmath445 in @xmath161 , then gives us that @xmath446 . since we required that @xmath404 if @xmath447 and @xmath406 if @xmath407 , we have that for any particular solution in @xmath230 , the @xmath236 s can be thought of as an augmented @xmath244-partition .
conversely , any augmented @xmath244-partition @xmath175 gives a solution in @xmath230 since if @xmath262 , then @xmath263 if and only if @xmath448 and @xmath229 for all @xmath265 .
we can therefore turn our attention to counting augmented @xmath244-partitions .
let @xmath266 .
then we get for any @xmath267-partition @xmath175 , @xmath449 and @xmath269 whenever @xmath450 , where we take @xmath451 .
the number of solutions to this set of inequalities is @xmath452 recall by observation [ ob2 ] that @xmath272 if @xmath245 , i.e. , if @xmath453 . if @xmath247 then @xmath273 and @xmath248 .
we get that @xmath454 if and only if @xmath276 .
set @xmath277 .
the number of solutions to @xmath455 is given by @xmath456 now for a given @xmath19 , the number of solutions to is @xmath457 summing over all subsets @xmath429 , we can write the number of all solutions to as @xmath458 and so the theorem is proved .
the proof of theorem [ thm : hypermod ] proceeds nearly identically , so we will omit unimportant details in the proof below .
we equate coefficients and prove that @xmath459 holds for any @xmath97 .
we recognize the left - hand side of equation as @xmath460 , so we want to count augmented @xmath1-partitions @xmath329\to \pm x$ ] , where @xmath162 is a totally ordered set of order @xmath461 .
we interpret this as the number of solutions , in the augmented lexicographic ordering , to @xmath462 where we have * @xmath403 , * @xmath404 if @xmath405 , * @xmath406 if @xmath407 , and * @xmath408 if @xmath409 . with these restrictions , we split the solutions to by our prior rules .
let @xmath463 be any particular solution . if @xmath220 , then @xmath221 , which falls into one of two mutually exclusive cases : @xmath464 if @xmath171 , then @xmath223 , giving : @xmath465 also mutually exclusive . with @xmath466
, there is only one case , depending on @xmath8 . if @xmath117 , then @xmath467 and @xmath468 and @xmath469 .
similarly , if @xmath470 , then @xmath471 and we have @xmath472 and @xmath473 . define @xmath226 and @xmath230 as before .
we get @xmath474 mutually exclusive sets @xmath230 indexed by subsets @xmath475 ( these subsets will correspond to ordinary descent sets ) . now for any @xmath476 , define the @xmath92 poset @xmath244 to be the poset given by @xmath258 if @xmath248 , and @xmath477 if @xmath453 , for @xmath336 .
we form a zig - zag poset labeled consecutively by @xmath478 . for a given solution @xmath443 ,
let @xmath332 \to \pm[k]$ ] be defined by @xmath254 for @xmath479 , with @xmath334 .
it is not too difficult to check that @xmath175 is an augmented @xmath244-partition , and that any augmented @xmath480-partition corresponds to a solution in @xmath230 .
let @xmath481 .
then for any @xmath267-partition @xmath175 we get @xmath482 with @xmath269 whenever @xmath450 .
the number of solutions to this set of inequalities is @xmath452 we see that for @xmath483 , @xmath272 if @xmath245 , i.e. , if @xmath453 .
also , if @xmath247 then @xmath273 and @xmath248 .
this time we get that @xmath275 , an ordinary descent set , if and only if @xmath484 .
set @xmath485 .
the number of solutions to @xmath486 is given by @xmath487 we take the sum over all subsets @xmath19 to show the number of solutions to is @xmath488 and the theorem is proved .
in this section we give some @xmath48-analogs of our previous theorems .
first we define the @xmath48-variant of the ( type a ) order polynomial , or @xmath48-order polynomial .- analog of the order polynomial is not a polynomial in @xmath16
. however , we will refer to it as the @xmath48-order polynomial , " even if it might be more appropriate to call it the
@xmath48-analog of the order polynomial . " ] let @xmath1 be a poset with @xmath2 elements , and take @xmath489 .
then , @xmath490 let @xmath491 and define the _ @xmath48-binomial coefficient _
@xmath492 in the natural way : @xmath493 another way to interpret the @xmath48-binomial coefficient is as the coefficient of @xmath494 in @xmath495 where @xmath28 and @xmath30 @xmath48-commute " via the relation @xmath496 .
this interpretation is good for some purposes , but we will use a different point of view .
we will define the @xmath48-multi - choose function @xmath497 as the following : @xmath498 where @xmath499 is the totally ordered chain @xmath500 , the identity permutation in @xmath501 . in this case
, the @xmath48-order polynomial counts certain integer partitions ( so there is at least some overlap between @xmath1-partitions and the more familiar integer partitions ) . specifically , the coefficient of @xmath502 is the number of integer partitions of @xmath503 with at most @xmath193 parts , each of size at most @xmath504.-analog of his shuffling results , though he never states @xmath48-analogs of his main theorems .
see @xcite , definitions 3 and 5 . ]
now we will obtain formulas for @xmath48-order polynomials for all permutations . for any permutation @xmath505
, the @xmath48-order polynomial may be expressed as @xmath506 when we computed the ordinary order polynomial in section [ sec : def ] we only cared about the _ number _ of solutions , rather than the _ set _ of solutions to the inequalities @xmath507 since we only cared how many solutions there were and not what the solutions were , we could count solutions to a system where all the inequalities were weak .
we will still follow the same basic procedure , but as we manipulate our system of inequalities we need to keep track of how we modify the set of solutions .
the @xmath48-order polynomial will be seen to be simply a power of @xmath48 ( depending on @xmath8 ) times a @xmath48-binomial coefficient .
we can form a new system of inequalities that has the same number of solutions as , but in which every inequality is weak : @xmath508 the bijection between these sets of solutions is given by @xmath509 where @xmath510 is the number of descents to the left of @xmath227 .
therefore the @xmath48-order polynomial is given by @xmath511 the sum of all @xmath510 can be expressed as @xmath512 , which is sometimes referred to as the _ comajor index _ , denoted @xmath513 .. indeed , had we made our @xmath1-partitions order reversing , as in stanley s original definition of a @xmath1-partition , we would have gotten @xmath514 rather than @xmath515 above . ]
the rest of the sum is now recognizable as a @xmath48-binomial coefficient . in summary
, we have @xmath516 now we will give a @xmath48-analog of theorem [ thm : ges ] .
we would like to have a @xmath48-analog for theorem [ thmcyc ] as well , but if such a theorem exists , it does not seem to be immediate given the method of proof in this paper .
recall that the proof of theorem [ thmcyc ] hinged on the observation that cyclically permuting a permutation @xmath505 leaves the number of cyclic descents unchanged : @xmath517 for any @xmath518 .
unfortunately , cyclically permuting @xmath8 has a more subtle effect on the comajor index . moving on
, we define the @xmath48-version of the structure polynomial , @xmath519 [ t ] [ thm : qsymmetric ] as polynomials in @xmath28 and @xmath30 ( and @xmath48 ) with coefficients in the group algebra we have @xmath520 the proof will follow nearly identical lines of reasoning as in the ordinary ( @xmath521 ) case . here
we sketch the proof with emphasis on the major differences .
again , we will decompose the coefficient of @xmath8 : @xmath522 by , we have that the coefficient of @xmath8 is the order polynomial @xmath523 so we will examine the @xmath8-partitions @xmath156\to \{0,1,\ldots , l-1\}\times \{0,1,\ldots , k-1\}$ ] .
notice that we are still mapping into a set with @xmath211 elements . as before we impose the lexicographic ordering on this image set . to ensure that we keep the proper powers of @xmath48
, we think of the order polynomial now as:@xmath524 we have given each point @xmath385 the weight @xmath525 so that the weight corresponds to the position of the point in the lexicographic ordering on @xmath526 .
we now proceed exactly as in the proof of theorem [ thm : ges ] .
@xmath527}\left(\sum_{\substack { 0\leq i_1 \leq \cdots \leq
i_n \leq l-1\\ s\in i \rightarrow i_s < i_s+1 } } q^{ki_s } \right ) \left(\sum_{\sigma \in \mathcal{l}(p_i ) } \omega_{\sigma}(q;k ) \right)\\ & = & \sum_{\sigma\tau = \pi } \omega_{\sigma}(q;k ) \omega_{\tau}(q^{k};l),\end{aligned}\ ] ] as desired . in order to give a @xmath48-analog of theorem [ thm : cho ] , we define the @xmath48-order polynomial for a signed permutation @xmath341 as @xmath528 now we define the type b @xmath48-structure polynomial , @xmath529 [ thm : qhyper ] the following relation holds as polynomials in @xmath28 and @xmath30 ( and @xmath48 ) with coefficients in the group algebra of the hyperoctahedral group : @xmath530 or upon substituting @xmath110 @xmath111 then @xmath531 we will omit most of the details , but the crucial step is to keep the proper exponent on @xmath48 .
we give to each point @xmath385 the weight @xmath532 so that the weight corresponds to the position of the point in the lexicographic order on the set @xmath533 .
the proof is outlined in two steps below . for any @xmath8 and any pair of positive integers @xmath534 , @xmath535 the augmented version of the @xmath48-order polynomial gives us @xmath48-analogs for theorem [ conj1 ] and theorem [ thm : hypermod ] . for a signed permutation @xmath341
we have @xmath536 where if @xmath510 is the number of descents of @xmath8 to the left of @xmath227 , then the _ augmented comajor index _
, @xmath537 , is the sum over all @xmath227 of the numbers @xmath510 .
define @xmath538 [ thm : qaug ] as polynomials in @xmath28 and @xmath30 ( and @xmath48 ) with coefficients in the group algebra of the hyperoctahedral group we have @xmath539 or upon substituting @xmath133 @xmath134 then @xmath540 [ h ] here the key is to give the integer pairs @xmath385 the proper weight in the augmented lexicographic ordering .
if we take @xmath541 as the weight of the point @xmath385 then we get that the points @xmath397 and @xmath398 have the same weight .
so the weight does indeed correspond to the position of @xmath385 in the augmented lexicographic ordering .
everything else follows as in the proof of theorem [ conj1 ] . for any @xmath341 and any pair of positive integers @xmath534 , @xmath542 [ thm : qhypermod ] as polynomials in @xmath28 and @xmath30 ( and @xmath48 ) with coefficients in the group algebra of the hyperoctahedral group we have @xmath543 upon substituting @xmath133 @xmath111 then @xmath544
because we exploit the augmented lexicographic order in the proof of theorem [ thm : hypermod ] ( the @xmath545 case ) , we will use the same weighting scheme as in the proof of theorem [ thm : qaug ] for the points @xmath385 .
we have for any @xmath341 and any pair of positive integers @xmath534 , @xmath546
we hope to be able to apply the method of @xmath1-partitions to study descent algebras more generally . in his thesis work , vic reiner @xcite defined @xmath1-partitions purely in terms of the root system of a coxeter group
. his definition may provide a way to obtain a theorem that generalizes theorems [ thm : ges ] and [ thm : cho ] to any coxeter group .
another avenue that we have already begun to investigate is the study of _ peak algebras_. a _ peak _ of a permutation @xmath505 is a position @xmath9 such that @xmath547 .
kathryn nyman @xcite showed that the span of sums of permutations with the same set of peaks forms a subalgebra of the group algebra .
later , aguiar , bergeron , and nyman @xcite explored many properties of this and related peak algebras for the symmetric group , including connections to descent algebras of types b and d. a key tool in @xcite was john stembridge s _ enriched _
@xmath1-partitions @xcite . in forthcoming work
, we use stembridge s tool to study subalgebras formed by the span of sums of permutations with the same _ number _ of peaks , obtaining results similar to the main theorems of this paper .
a meta - question for this area of research is : what is so special about descents ( or peaks ) ? why should this way of grouping permutations give rise to a subalgebra at all ? in terms of permutation patterns ,
a descent is an occurrence of the pattern 21 , and a peak is an occurrence of the pattern 132 or 231 . what distinguishes descents and peaks from , say , grouping permutations according to instances of 1324 ( which does nt give a subalgebra ) ?
m. aguiar , n. bergeron , and k. nyman , _ the peak algebra and the descent algebras of types @xmath548 and @xmath549 _ , transactions of the american mathematical society * 356 * ( 2004 ) , 27812824 .
d. bayer and p. diaconis , _ trailing the dovetail shuffle to its lair _ , annals of applied probability * 2 * ( 1992 ) , 294313 .
n. bergeron , _ a decomposition of the descent algebra of the hyperoctahedral group .
journal of algebra * 148 * ( 1992 ) , 98122 .
f. bergeron and n. bergeron , _ a decomposition of the descent algebra of the hyperoctahedral group .
i. _ journal of algebra * 148 * ( 1992 ) , 8697 .
f. bergeron and n. bergeron , _ orthogonal idempotents in the descent algebra of @xmath550 and applications _ , journal of pure and applied algebra * 79 * ( 1992 ) , 109129 .
p. cellini , _ a general commutative descent algebra _ , journal of algebra * 175 * ( 1995 ) , 9901014 .
p. cellini , _ a general commutative descent algebra .
the case @xmath551 _ , journal of algebra * 175 * ( 1995 ) , 10151026 .
p. cellini , _ cyclic eulerian elements _ , european journal of combinatorics * 19 * ( 1998 ) , 545552 .
chow , _ noncommutative symmetric functions of type @xmath552 _ , ph.d . thesis ( 2001 ) .
j. fulman , _ affine shuffles , shuffles with cuts , the whitehouse model , and patience sorting _ , journal of algebra * 231 * ( 2000 ) , 614639 .
j. fulman , _ applications of the brauer complex : card shuffling , permutation statistics , and dynamical systems _ , journal of algebra * 243 * ( 2001 ) , 96122 .
a. garsia and c. reutenauer , _ a decomposition of solomon s descent algebra _ ,
advances in mathematics * 77 * ( 1989 ) , 189262 .
i. gessel , personal communication .
i. gessel , _
multipartite @xmath1-partitions and inner products of skew schur functions _ , contemporary mathematics * 34 * ( 1984 ) , 289317 .
loday , _ operations sur lhomologie cyclique des algbres commutatives _ , inventiones mathematicae * 96 * ( 1989 ) , 205230 .
k. nyman , _ the peak algebra of the symmetric group _ , journal of algebraic combinatorics * 17 * ( 2003 ) , 309322 . v. reiner ,
_ quotients of coxeter complexes and @xmath1-partitions _ , memoirs of the american mathematical society * 95 * ( 1992 ) .
v. reiner , _ signed posets _ , journal of combinatorial theory series a * 62 * ( 1993 ) , 324360 .
l. solomon , _ a mackey formula in the group ring of a finite coxeter group _ , journal of algebra * 41 * ( 1976 ) , 255264 .
r. stanley , _ enumerative combinatorics , volume i _ , cambridge university press , 1997 .
j. stembridge , _ enriched @xmath553-partitions _ , transactions of the american mathematical society * 349 * ( 1997 ) , 763788 . | louis solomon showed that the group algebra of the symmetric group @xmath0 has a subalgebra called the descent algebra , generated by sums of permutations with a given descent set .
in fact , he showed that every coxeter group has something that can be called a descent algebra .
there is also a commutative , semisimple subalgebra of solomon s descent algebra generated by sums of permutations with the same number of descents : an eulerian " descent algebra . for any coxeter group that is also a weyl group , paola cellini proved the existence of a different eulerian subalgebra based on a modified definition of descent .
we derive the existence of cellini s subalgebra for the case of the symmetric group and of the hyperoctahedral group using a variation on richard stanley s theory of @xmath1-partitions . |
the study at the @xmath67 collider hera of exclusive diffractive processes in the presence of a hard scale has provided insight into the parton dynamics of the diffractive exchange .
the four - momentum squared transferred at the proton vertex , @xmath4 , provides a relevant scale to investigate the application of perturbative quantum chromodynamics ( pqcd ) for @xmath68 @xcite . in this letter , the first measurement at large @xmath4 ( @xmath69 gev@xmath70 ) of diffractive photon - proton scattering , @xmath71 , where @xmath1 is the proton dissociative system , is presented .
the measurement is performed at hera by studying the reaction @xmath72 in the photoproduction regime with a large rapidity gap between the final state photon and the proton dissociative system @xmath1 ( as illustrated in figure [ fig : photondiag]a ) .
the centre of mass energy of the system formed by the exchanged photon and proton is in the range @xmath73 .
this process constitutes an extension of deeply virtual compton scattering @xcite into the large @xmath7 and small @xmath74 regime .
diffractive photon scattering can be modelled in the proton rest frame by the fluctuation of the incoming photon into a @xmath75 pair at a long distance from the proton target .
the @xmath75 pair is then involved in a hard interaction with the proton via the exchange of a colour singlet state . in the leading logarithmic approximation ( lla )
, the colour singlet exchange is modelled by the effective exchange of a gluon ladder ( figure [ fig : photondiag]b ) . for sufficiently low values of bjorken @xmath76 ( i.e. large values of @xmath3 ) ,
the bfkl @xcite approach is expected to be appropriate for describing the gluon ladder . in the lla bfkl model the gluon ladder couples to a single parton ( predominantly a gluon ) within the proton .
the cross section depends therefore linearly on the parton distribution in the proton . due to the quasi - real nature of the incoming photon ( @xmath2 )
, the transverse momentum of the final state photon , @xmath77 , is entirely transferred by the gluon ladder to the parton in the proton .
the separation in rapidity between the parton scattered by the gluon ladder and the final state photon is given by @xmath78 , where @xmath79 is the invariant mass of the system formed by the incoming photon and the struck parton .
the proton remnant and the struck parton hadronise through fragmentation processes to form the hadronic system @xmath1 . assuming parton - hadron duality , hadrons originating from the struck parton correspond to the particles with the largest transverse momenta and hence are the closest in rapidity to the scattered photon .
the present analysis complements the measurements of exclusive production of @xmath80 and @xmath81 mesons at large @xmath7 @xcite .
the measured @xmath3 and @xmath4 dependences of their cross sections were found to be in agreement with lla bfkl based calculations @xcite . for the process studied here , theoretical calculations are simplified by the absence of a vector meson wave function : the only non - perturbative part of the calculation is the parton distribution functions of the proton . however , the cross section is suppressed relative to that of vector meson production by the electromagnetic coupling of the @xmath82 pair to the final state photon , making the measurement more challenging .
( 120,50 ) ( 1,-2 ) ( 27,20)@xmath83 ( 4.5,44.0)@xmath84 ( 41.0,46.0)@xmath85 ( 67.0,40.5)@xmath86 ( 24.5,30.5)@xmath87 ( 12.0,14.0)@xmath88 ( 71.5,8.5)@xmath89 ( 66.0,25.0)@xmath90 ( 61.0,26.5)(0,10.0)(5,0)(0,-10.0 ) ( 90,0 ) ( 35,-2)@xmath91 ( 120,-2)@xmath92 in the following , measurements of the photon - proton cross sections are presented as a function of @xmath3 and differentially in @xmath7 and are compared to predictions based on lla bfkl calculations @xcite .
a detailed description of the h1 detector can be found elsewhere @xcite .
the following briefly describes the detector components relevant to this analysis .
a right handed coordinate system is defined with the origin at the nominal @xmath67 interaction vertex , such that the positive @xmath93 axis ( forward direction ) corresponds to the direction of the outgoing proton beam .
the polar angle @xmath94 and transverse momentum @xmath95 are defined with respect to the @xmath93 axis .
the pseudorapidity is defined as @xmath96 . a liquid argon ( lar ) calorimeter covers the polar angle range @xmath97 with full azimuthal coverage .
the lar calorimeter consists of both an electromagnetic section and a hadronic section .
the energy resolution for single particles measured in a test beam is @xmath98 for electrons and @xmath99 for hadrons@xcite .
the polar angle region @xmath100 is covered by a lead - scintillating fibre calorimeter , the spacal @xcite , with both electromagnetic and hadronic sections .
the spacal has an energy resolution for electromagnetic showers of @xmath101 .
charged particles are detected in the polar angle ranges @xmath102 by the central track detector ( ctd ) and @xmath103 by the forward track detector ( ftd ) .
the ctd comprises two large cylindrical jet drift chambers , providing precise track measurements in the @xmath104 plane , supplemented by two @xmath93-chambers and two multi - wire proportional chambers arranged concentrically around the beam - line .
the ctd is complemented by a silicon vertex detector @xcite covering the range @xmath105 .
the ftd consists of layers of planar and radial drift chambers to provide measurement of the @xmath94 and @xmath106 angles of forward tracks , respectively .
the trackers and the calorimeters are operated within a solenoidal magnetic field of @xmath107 t. the luminosity is determined from the rate of bethe - heitler events , @xmath108 , measured using a erenkov crystal calorimeter , the photon detector ( pd ) , situated near the hera beam pipes at @xmath109 m. a second erenkov crystal calorimeter , the electron tagger ( et ) , with an energy resolution of @xmath110 , is located at @xmath111 m. the et measures the energy of the positron when scattered through an angle of less than @xmath112 mrad in the energy range @xmath113 gev .
the detection of the scattered positron in the et ensures that the exchanged photon is quasi - real with @xmath114 .
following the notation introduced in figure [ fig : photondiag]a , the scattering process , @xmath115 , is described by the usual deep inelastic scattering ( dis ) kinematic variables : @xmath116 where @xmath117 , @xmath118 , @xmath119 and @xmath120 are the four - momentum of the incoming positron , the incoming proton , the scattered positron and the exchanged photon , respectively .
the variable @xmath121 is the virtuality of the exchanged photon and @xmath122 is the inelasticity of the @xmath67 interaction , corresponding to the relative energy loss of the scattered positron in the proton rest frame .
the @xmath67 centre of mass energy squared is given by @xmath123 and the @xmath124 centre of mass energy squared is @xmath125 .
in addition , the longitudinal momentum fraction of the diffractive exchange ( called the pomeron in the regge model @xcite ) with respect to the proton is defined as : @xmath126 where @xmath127 is the four - momentum of the @xmath1 system .
the elasticity of the @xmath124 interaction , which can be seen as the fractional energy of the exchanged photon transferred to the final state photon , is given by @xmath128 , where : @xmath129 @xmath130 being the four - momentum of the final state photon .
finally the square of the four - momentum transfer across the diffractive exchange is given by : @xmath131 the data used in this analysis were collected with the h1 detector during the @xmath132 running period , when positrons of energy @xmath133 collided with protons of energy @xmath134 in the hera accelerator .
the data sample corresponds to an integrated luminosity of @xmath135 pb@xmath136 .
more details on the present analysis can be found in @xcite .
the data were recorded using a combination of two triggers . both triggers select an electromagnetically interacting particle in the spacal which corresponds to the scattered photon candidate .
one of the triggers requires in addition an energy deposit in the et , corresponding to the scattered positron .
the effective trigger efficiency , including time dependent downscale factors , amounts to approximately @xmath137 .
the reconstruction of the kinematic quantities used in the following are approximations valid in the limit of small scattering angles of the positron and small transverse momentum of the @xmath1 system compared to its longitudinal momentum .
the quantity @xmath122 , and hence @xmath3 , is calculated from the scattered positron energy , @xmath138 , measured in the et using the relation @xmath139 .
the relative resolution of @xmath3 is @xmath140 . to avoid regions of low et acceptance , the energy of the scattered positron is limited to @xmath141 gev , corresponding to @xmath142 .
in addition , to suppress backgrounds from processes occurring in coincidence with bethe - heitler events , it is required that no energy deposits above the noise threshold are measured in the pd .
photon candidates are selected from energy clusters with small radii detected in the electromagnetic section of the spacal .
if significant energy is measured behind the cluster in the hadronic section of the spacal , the event is rejected .
events with more than one cluster above the noise level in the spacal are also rejected . to reduce the background from charged particles ,
the cluster of the photon candidate must have no associated track in the ctd .
the photon candidates are furthermore required to have an energy @xmath143 and a transverse momentum @xmath144 .
the hadronic final state @xmath1 is reconstructed using a combination of tracking and calorimetric information .
the difference between the total energy @xmath145 and the longitudinal component of the total momentum @xmath146 , as calculated from the scattered positron , the final state photon and the hadronic system @xmath1 , is restricted to @xmath147 gev .
this requirement suppresses non-@xmath67 induced background and background due to the overlap during the same bunch crossing of a bethe - heitler event with a dis event . for a
fully contained @xmath67 event the relation @xmath148 gev holds . in the case that charged particles from the final state
are detected by the tracking detector , allowing the primary vertex to be reconstructed , the @xmath93 coordinate of the vertex has to satisfy @xmath149 cm . for @xmath150 of the selected events , all charged particles of the final state lie outside the detector acceptance and no event vertex
is reconstructed . in this case , the time averaged vertex position is used for the kinematic reconstruction . the kinematic variable @xmath7 is reconstructed as : @xmath151 with a relative resolution of @xmath152 .
the longitudinal momentum fraction of the diffractive exchange with respect to the proton is reconstructed using the formula : @xmath153 where @xmath154 is the sum of the energy and longitudinal momentum of the final state photon .
the event inelasticity of the @xmath124 interaction is reconstructed as : @xmath155 where the summation is performed on all detected hadronic final state particles in the event , i.e. all measured particles except the scattered positron and the final state photon .
this reconstruction method has the advantage that the loss of particles along the forward beam pipe only has a marginal effect on the reconstruction of @xmath156 .
diffractive events are selected by requiring that @xmath157 .
the cut value ensures a large pseudorapidity gap , @xmath90 , between the photon and the proton dissociative system @xmath1 , since @xmath158 . finally , to reduce contamination of the signal by non - diffractive background
, it is required that the difference in pseudorapidity between the photon and the closest final state hadron satisfies @xmath159 .
this rapidity gap is inferred from the absence of activity in the relevant detector parts , i.e. absence of any track or a cluster of energy deposits above the noise threshold of @xmath160 mev in the lar calorimeter .
mev per cell and the average number of cells per cluster is typically @xmath161 . ] after applying all selection criteria , 240 events remain in the data sample .
monte carlo ( mc ) simulations are used to correct the data for effects of detector acceptance and efficiency , to estimate the background and to compare model predictions to the data .
all generated mc events are passed through the full geant@xcite based simulation of the h1 detector and are reconstructed using the same analysis chain as is used for the data .
the herwig 6.4 @xcite mc event generator is used to simulate the diffractive high @xmath7 photon scattering using the lla bfkl model @xcite . at leading logarithmic accuracy
there are two independent free parameters in the bfkl calculation : the value of the strong coupling @xmath162 and the scale , @xmath163 , which defines the leading logarithms in the expansion of the bfkl amplitude , @xmath164 . in exclusive production of vector mesons ,
the scale parameter @xmath163 is related to the vector meson mass . in the case of diffractive photon scattering , the unknown scale results in the absence of a prediction for the normalisation of the cross section @xcite . in the calculations considered here @xcite , the running of @xmath162 as a function of the scale
is ignored . in order to distinguish the @xmath162 parameter of the lla bfkl model from the usual strong coupling constant this parameter
will henceforth be referred to as @xmath165 .
the choice of @xmath166 is used for the simulation in this letter . in the asymptotic approximation of the calculations @xcite
, the w distribution follows a power - law : @xmath167 where the exponent , also called the pomeron intercept , is given directly by the choice of @xmath165 with @xmath168 .
using the herwig simulation this approximation is found to be justified given the current experimental statistical precision .
the lla bfkl model predicts an approximate power - law behaviour for the @xmath4 dependence of the cross section of the form @xmath169 , where @xmath170 , also predicted by the model , depends only on the parton density functions ( pdfs ) of the proton and the value of @xmath165 . running coupling effects ,
not considered here , would in principle allow @xmath170 to depend on @xmath4 .
the grv94 pdfs @xcite are used for the proton pdfs in the herwig prediction . a comparison with the cteq5 @xcite and mrst
pdfs @xcite shows little dependence of the herwig prediction on the input proton pdfs . in order to describe the data ,
the @xmath4 dependence of the diffractive photon scattering simulated using herwig ( predicting a value of @xmath171 for @xmath172 ) was weighted by a factor @xmath173 , i.e.the @xmath7 power is modified from @xmath174 to @xmath175 .
this weighted herwig prediction is used to correct the data for resolution and acceptance effects .
possible sources of background are estimated using mc simulations .
the background from inclusive diffractive photoproduction events ( @xmath176 , where the two hadronic final states are separated by a rapidity gap ) is simulated using the phojet mc event generator @xcite .
this background contributes when a single electromagnetic particle fakes the photon candidate in the spacal .
it is estimated to amount to @xmath177 of the measured cross section .
the background from electron pair production ( @xmath178 ) is modelled using the grape event generator @xcite .
this process contributes to the selection if one lepton is detected in the et , a second lepton fakes the photon within the spacal and the remaining lepton escapes detection .
this background contributes @xmath140 of the measured cross section . in order to investigate the background from high @xmath7 diffractive exclusive @xmath179 production , where the @xmath179 decays through the @xmath180 or @xmath181 channel ,
a sample was generated using the diffvm monte carlo generator @xcite . the contribution from this background process
is found to be negligible .
the background from dis events , in which the scattered positron fakes the photon candidate and an overlapping photoproduction or bethe - heitler event gives a positron detected in the et , has been studied and was also found to be negligible . in figure
[ fig : control ] the data , corrected for trigger efficiency , are compared to the monte carlo simulations .
the background predictions from grape and phojet are normalised to the integrated luminosity of the data sample .
the weighted herwig prediction is normalised to the number of events obtained after the subtraction of the predicted background from the data .
a good description of the data by the combined monte carlo simulations is observed .
the @xmath182 differential cross sections are defined using the formula : @xmath183 where @xmath184 is the number of observed events corrected for trigger efficiency , @xmath185 the expected contribution from background events as estimated using the phojet and grape monte carlos simulations , @xmath186 the integrated luminosity , @xmath187 the signal acceptance and @xmath188 and @xmath189 the bin widths in @xmath3 and @xmath4 , respectively .
the acceptance , estimated using the weighted herwig simulation , is the ratio of the number of events accepted after reconstruction to the number of events generated in the defined phase space on hadron level .
it accounts for all detector effects , including bin - to - bin migrations , as well as geometrical acceptance and detector efficiencies .
qed radiative correction effects are estimated to be less than @xmath190% @xcite and are neglected .
the @xmath71 differential cross section is then extracted from the @xmath67 cross section using : @xmath191 where the photon flux , @xmath192 , is integrated over the range @xmath193 gev@xmath70 according to the modified weizscker - williams approximation @xcite .
the @xmath124 cross section is obtained by modelling @xmath194 as a power - law in @xmath3 , whose parameters are iteratively adjusted to reproduce the measured @xmath3 dependence of the @xmath67 cross section .
the differential @xmath124 cross section in @xmath7 is then extracted from the @xmath67 cross section by correcting for the effect of the photon flux over the visible @xmath3 range ( @xmath195 gev ) .
the @xmath124 cross section as a function of @xmath3 is obtained by first integrating equation ( [ eq : flux ] ) over the @xmath7 range , and then correcting for the effect of the photon flux in each bin in @xmath3 .
more details on the extraction procedure of the @xmath124 cross section can be found in @xcite . the systematic error on the measurement stems from experimental uncertainties and from model dependences .
they are calculated using the weighted herwig simulation of the signal process .
the sources of systematic error are listed below . for each of them the typical effect on the cross section measurement is indicated .
the experimental systematic errors are : * the energy scale uncertainty of @xmath196 for an electromagnetic cluster measured by the spacal gives errors in the range of @xmath197 to @xmath140 . * the uncertainty of @xmath198 mrad for the measurement of the photon candidate polar angle results in an error of up to @xmath177 .
* the hadronic final state energy scale uncertainty of @xmath199 leads to an error of less than @xmath200 . * the energy scale uncertainty of @xmath201 of the et produces an error ranging from @xmath202 in the highest @xmath7 bin to @xmath203 in the lowest @xmath3 bin . *
the uncertainty of @xmath204 on the noise threshold from the calorimeters gives an error varying from @xmath205 at low @xmath7 to @xmath152 at the highest @xmath7 value .
* the luminosity is measured with an accuracy of @xmath201 which enters as an overall normalisation uncertainty .
the systematic errors due to uncertainty of model parameters are : * the uncertainty due to the @xmath206dependence , estimated by weighting the @xmath206distribution by the form @xmath207 , yields an error varying from @xmath177 in the central @xmath7 bin up to @xmath208 in the lowest @xmath3 bin .
this weight would correspond to changing @xmath165 to the values @xmath209 and @xmath210 . *
the uncertainty due to the @xmath7 dependence , estimated by weighting the @xmath7 distribution by the form @xmath211 , leads to an error ranging from @xmath200 to @xmath140 . * the uncertainty in the modelling of the proton remnant system @xmath1 , estimated by weighting the @xmath212 distribution by the form @xmath213 , results in a typical error of @xmath202 to @xmath140 .
* the uncertainty of @xmath214 assumed on the normalisation of the subtracted inclusive diffractive background ( from the phojet monte carlo simulation ) leads to an error of approximately @xmath177 in the highest @xmath3 bins decreasing to @xmath202 in the highest @xmath7 bin . *
the propagation of the uncertainty on the power - law parameter @xmath215 in the @xmath124 cross section extraction procedure leads to an error of @xmath140 on @xmath216 , independent of @xmath7 , and is below 1% on @xmath194 . the uncertainty on the phojet mc normalisation and the model uncertainties from the unknown @xmath206 , @xmath7 and @xmath212 dependences are estimated from data comparisons and defined by the range where the model describes the data .
each source of systematic error is varied in the weighted herwig monte carlo within its uncertainty . in each measurement bin , the corresponding deviation of the normalised cross section from the central value is taken as systematic error . the total systematic error is obtained by adding the individual contributions in quadrature , on a bin - by - bin basis . the largest systematic error on the cross sections comes from the uncertainty on the @xmath206and @xmath212 dependences in the mc simulation .
the total systematic error on the @xmath3 dependence of the cross section varies from @xmath203 in the central bins to @xmath217 in the lowest bin .
the systematic error on the @xmath7 dependence of the cross section varies from @xmath218 in the lowest bin to @xmath219 in the highest bin .
an additional global uncertainty of 4% arises from the @xmath124 cross section extraction procedure .
the total systematic errors are comparable to or smaller than the statistical errors .
the cross sections measured for the domain @xmath5 , @xmath220 gev@xmath70 , @xmath157 and @xmath221 are presented in table [ csvalues ] .
the @xmath0 cross section as a function of @xmath3 is shown in figure [ fig : sigmaw ] .
a power - law dependence of the form @xmath222 is fitted to the measured cross section .
the fit yields @xmath223 with @xmath224 .
the contributions from the systematic errors are calculated by shifting the data points according to each source of uncertainty , taking correlations into account , and repeating the fit .
the errors are then added in quadrature to obtain the total systematic error .
the steep rise of the cross section with @xmath3 is usually interpreted as an indication of the presence of a hard sub - process in the diffractive interaction and of the applicability of perturbative qcd .
the present @xmath215 value , measured at an average @xmath7 value of @xmath225 gev@xmath70 , is compatible with that measured by h1 in diffractive @xmath81 photoproduction of @xmath226 @xcite at an average @xmath7 of @xmath227 gev@xmath70 .
the pomeron intercepts associated with these @xmath215 values correspond to the strongest energy dependences measured in diffractive processes .
@xmath3 & @xmath3 range & @xmath228 & @xmath192 & @xmath229 + [ gev ] & [ gev ] & @xmath230pb / gev@xmath231 $ ] & [ gev@xmath136 ] & [ nb ] + @xmath233 & @xmath234 & @xmath235 & @xmath236 & @xmath237 + @xmath238 & @xmath239 & @xmath240 & @xmath241 & @xmath242 + @xmath243 & @xmath244 & @xmath245 & @xmath246 & @xmath247 + @xmath248 & @xmath249 & @xmath250 & @xmath251 & @xmath252 + @xmath253 & @xmath253 range & d@xmath254 & @xmath255 & d@xmath256d@xmath253 + [ gev@xmath70 ] & [ gev@xmath70 ] & @xmath230pb / gev@xmath258 $ ] & [ gev@xmath136 ] & [ pb / gev@xmath70 ] + @xmath259 & @xmath260 & @xmath261 & @xmath262 & @xmath263 + @xmath264 & @xmath265 & @xmath266 & @xmath262 & @xmath267 + @xmath268 & @xmath269 & @xmath270 & @xmath262 & @xmath271 + the @xmath124 cross section differential in @xmath7 , at @xmath272 gev , is shown in figure [ fig : sigmat ] .
figure [ fig : sigmat ] also shows the comparison of the cross section to a fit of the form @xmath273 .
the fit result is @xmath274 with @xmath275 .
the @xmath7 dependence is harder than that measured by h1 in the diffractive photoproduction of @xmath81 mesons at large @xmath253 @xcite corresponding to @xmath276 . in figures [ fig : sigmaw ] and [ fig : sigmat ] the measured cross sections
are compared to predictions of the lla bfkl model , using the herwig monte carlo , as described in section [ sec : mc ] , with no @xmath7 weighting .
the predictions are normalised to the integrated measured cross section , as the normalisation is not predicted by the lla bfkl calculation @xcite .
the @xmath3 dependence of the cross section is well described by the lla bfkl prediction , as shown in figure [ fig : sigmaw ] .
the sensitivity of the measurement to the free parameter @xmath165 in the theoretical prediction is illustrated in figures [ fig : sigmaw ] and [ fig : sigmat ] .
using @xmath277 , the measured @xmath3 dependence leads to @xmath278 .
predictions are shown in figure [ fig : sigmaw ] for the values @xmath279 and @xmath280 . the lla bfkl curve corresponding to @xmath281 coincides with the solid fit line .
previous measurements of diffractive scattering at hera are well described by bfkl predictions with @xmath165 values similar to the value measured in this letter .
zeus measurements of exclusive @xmath80 and @xmath81 production at large @xmath253 are best described with the value of @xmath282 @xcite .
the h1 measurement of high @xmath7 @xmath283 production @xcite is compatible with @xmath282 and the h1 measurement of high @xmath7 @xmath81 production @xcite is described using @xmath284 . the lla bfkl prediction with @xmath166 gives a good description of the double dissociation process with a rapidity gap between jets measured by h1 @xcite .
events with rapidity gaps between jets were also measured by zeus @xcite and are found to be compatible with a model which uses @xmath285 .
note , however , that for these measurements the hard scale , corresponding to the jet transverse momentum squared , is of the order of or larger than @xmath286 gev@xmath70 .
as shown in figure [ fig : sigmat ] , the lla bfkl calculation for @xmath287 and @xmath280 , all of which give a reasonable description of the @xmath3 dependence , predict steeper @xmath7 distributions than is measured in the data .
the same effect can not be established for the exclusive @xmath283 measurement @xcite , where the measured @xmath4 range is limited to @xmath288 gev@xmath70 , although an underestimate of the cross section was observed at the largest values of @xmath7 .
the present situation is in contrast with the analysis of @xmath81 production @xcite , where the @xmath7 dependence was found to be well described by the lla bfkl prediction over a similar range in @xmath4 .
using the h1 detector , diffractive photon scattering , @xmath289 , where the final state photon carries a large transverse momentum and is well separated from the proton dissociative system @xmath1 , is measured for the first time at hera .
the measurement provides a unique test of the underlying qcd dynamics of the diffractive exchange .
cross sections are presented as a function of @xmath3 and differentially in @xmath253 .
a fit of the form @xmath290 performed on the cross section gives @xmath291 .
this strong energy dependence is compatible with that measured for the exclusive diffractive @xmath81 production at high @xmath7 .
a fit of the form @xmath292 yields @xmath293 , corresponding to a harder @xmath7 dependence of the cross section than measured for high @xmath7 diffractive @xmath81 production .
the measured cross sections are compared to the predictions of an lla bfkl model .
a good description of the @xmath3 dependence of the cross section is found , whereas the lla bfkl model predicts a @xmath7 dependence that is too soft and hence unable to describe the data .
we are grateful to the hera machine group whose outstanding efforts have made this experiment possible .
we thank the engineers and technicians for their work in constructing and maintaining the h1 detector , our funding agencies for financial support , the desy technical staff for continual assistance and the desy directorate for support and for the hospitality which they extend to the non desy members of the collaboration .
we are also grateful to j.r .
forshaw and m. diehl for fruitful discussions .
j. r. forshaw and p. j. sutton , eur .
j. c * 1 * ( 1998 ) 285 [ hep - ph/9703225 ] .
f. d. aaron _ et al . _
[ h1 collaboration ] , phys .
b * 659 * ( 2008 ) 796 [ arxiv:0709.4114 [ hep - ex ] ] .
e. a. kuraev , l. n. lipatov and v. s. fadin , sov . phys .
jetp * 45 * ( 1977 ) 199 [ zh . eksp .
* 72 * ( 1977 ) 377 ] ; + i. i. balitsky and l. n. lipatov , sov . j. nucl
. phys .
* 28 * ( 1978 )
822 [ yad . fiz .
* 28 * ( 1978 ) 1597 ] ; + l. n. lipatov , sov .
jetp * 63 * ( 1986 ) 904 [ zh .
. teor .
fiz . * 90 * ( 1986 ) 1536 ] .
s. chekanov _ et al . _
[ zeus collaboration ] , eur .
j. c * 26 * ( 2003 ) 389 [ hep - ex/0205081 ] .
a. aktas _ et al .
_ [ h1 collaboration ] , phys .
b * 638 * ( 2006 ) 422 [ hep - ex/0603038 ] .
a. aktas _ et al .
_ [ h1 collaboration ] , phys .
b * 568 * ( 2003 ) 205 [ hep - ex/0306013 ] .
s. chekanov _ et al . _
[ zeus collaboration ] , eur .
j. c * 24 * ( 2002 ) 345 [ hep - ex/0201043 ] .
i. f. ginzburg and d. y. ivanov , phys .
d * 54 * ( 1996 ) 5523 [ hep - ph/9604437 ] .
i. f. ginzburg , s. l. panfil and v. g. serbo , nucl .
b * 284 * ( 1987 ) 685 .
d. y. ivanov and m. wsthoff , eur .
j. c * 8 * ( 1999 ) 107 [ hep - ph/9808455 ] .
n. g. evanson and j. r. forshaw , phys .
d * 60 * ( 1999 ) 034016 [ hep - ph/9902481 ] .
b. e. cox and j. r. forshaw , j. phys .
g * 26 * ( 2000 ) 702 [ hep - ph/9912486 ] .
i. abt _ et al . _
[ h1 collaboration ] , nucl .
instrum .
meth . a * 386 * ( 1997 ) 310 and 348 .
b. andrieu _ et al . _
[ h1 calorimeter group ] , nucl .
instrum .
meth . a * 350 * ( 1994 ) 57 .
r. d. appuhn _ et al .
_ [ h1 spacal group collaboration ] , nucl .
instrum .
. a * 386 * ( 1997 ) 397 .
d. pitzl _ et al .
_ , nucl .
instrum .
a * 454 * ( 2000 ) 334 [ hep - ex/0002044 ] .
e. feinberg and i. pomeranchuk , suppl .
nuovo cimento * 3 * ( 1956 ) 652 ; + v. gribov , jetp lett . * 41 * ( 1961 ) 667 .
t. hreus , ph.d .
thesis , universit libre de bruxelles and pavol jozef afrik university ( 2008 ) , available at http://www-h1.desy.de/psfiles/theses/h1th-495.pdf ; + m. beckingham , ph.d .
thesis , `` diffractive photoproduction of high @xmath294 photons at hera '' , university of manchester ( 2003 ) , available at http://www-h1.desy.de/psfiles/theses/h1th-434.ps .
r. brun _ et al .
_ , `` geant3 , '' cern - dd / ee/84 - 1
. g. marchesini _ et al .
_ , comput .
. commun .
* 67 * ( 1992 ) 465 .
b. cox , j. r. forshaw and l. lnnblad , jhep * 9910 * ( 1999 ) 023 [ hep - ph/9908464 ] .
m. glck , e. reya and a. vogt , z. phys .
c * 67 * ( 1995 ) 433 .
h. l. lai _ et al . _
[ cteq collaboration ] , eur .
j. c * 12 * ( 2000 ) 375 [ hep - ph/9903282 ] .
a. d. martin , r. g. roberts , w. j. stirling and r. s. thorne , phys .
b * 531 * ( 2002 ) 216 [ hep - ph/0201127 ] .
f. w. bopp , r. engel and j. ranft , `` rapidity gaps and the phojet monte carlo , '' hep - ph/9803437 .
t. abe , `` grape - dilepton ( version 1.1 ) : a generator for dilepton production in e p collisions , '' comput .
commun . * 136 * ( 2001 ) 126 [ hep - ph/0012029 ] . b. list and a. mastroberardino , `` diffvm : a monte carlo generator for diffractive processes in e p scattering '' , prepared for _ workshop on monte carlo generators for hera physics _ ,
hamburg , germany , 27 - 30 apr 1998 .
s. aid _ et al .
_ [ h1 collaboration ] , z. phys .
c * 69 * ( 1995 ) 27 [ hep - ex/9509001 ] . c. f. von weizscker , z. phys .
* 88 * ( 1934 ) 612 ; + e. j. williams , phys .
rev . * 45 * ( 1934 ) 729 ; + s. frixione _ et al .
_ , phys .
b * 319 * ( 1993 ) 339 [ hep - ph/9310350 ] . c. adloff
_ et al . _
[ h1 collaboration ] , eur .
j. c * 24 * ( 2002 ) 517 [ hep - ex/0203011 ] .
s. chekanov _ et al . _
[ zeus collaboration ] , eur .
j. c * 50 * ( 2007 ) 283 - 297 [ hep - ex/0612008 ] . | the first measurement of diffractive scattering of quasi - real photons with large momentum transfer @xmath0 , where @xmath1 is the proton dissociative system , is made using the h1 detector at hera .
the measurement is performed for initial photon virtualities @xmath2 .
single differential cross sections are measured as a function of @xmath3 , the incident photon - proton centre of mass energy , and @xmath4 , the square of the four - momentum transferred at the proton vertex , in the range @xmath5 and @xmath6 .
the @xmath3 dependence is well described by a model based on perturbative qcd using a leading logarithmic approximation of the bfkl evolution . the measured @xmath7 dependence is harder than that predicted by the model and those observed in exclusive vector meson production .
# 1#2#3#4#1 * # 2 * ( # 3 ) # 4 desy 08 - 077 issn 0418 - 9833 + december 2008 + * measurement of diffractive scattering of photons with large momentum transfer at hera * h1 collaboration accepted by physics letters b f.d .
aaron@xmath8 , c. alexa@xmath9 , v. andreev@xmath10 , b. antunovic@xmath11 , s. aplin@xmath11 , a. asmone@xmath12 , a. astvatsatourov@xmath13 , a. bacchetta@xmath11 , s. backovic@xmath14 , a. baghdasaryan@xmath15 , p. baranov@xmath16 , e. barrelet@xmath17 , w. bartel@xmath11 , m. beckingham@xmath11 , k. begzsuren@xmath18 , o. behnke@xmath19 , a. belousov@xmath10 , n. berger@xmath20 , j.c .
bizot@xmath21 , m .- o .
boenig@xmath22 , v. boudry@xmath23 , i. bozovic - jelisavcic@xmath24 , j. bracinik@xmath25 , g. brandt@xmath11 , m. brinkmann@xmath11 , v. brisson@xmath21 , d. bruncko@xmath26 , a. bunyatyan@xmath27 , g. buschhorn@xmath28 , l. bystritskaya@xmath29 , a.j .
campbell@xmath11 , k.b .
cantun avila@xmath30 , f. cassol - brunner@xmath31 , k. cerny@xmath32 , v. cerny@xmath33 , v. chekelian@xmath28 , a. cholewa@xmath11 , j.g .
contreras@xmath30 , j.a .
coughlan@xmath34 , g. cozzika@xmath35 , j. cvach@xmath36 , j.b .
dainton@xmath37 , k. daum@xmath38 , m. dek@xmath11 , y. de boer@xmath11 , b. delcourt@xmath21 , m. del degan@xmath20 , j. delvax@xmath13 , a. de roeck@xmath39 , e.a .
de wolf@xmath13 , c. diaconu@xmath31 , v. dodonov@xmath40 , a. dossanov@xmath28 , a. dubak@xmath41 , g. eckerlin@xmath11 , v. efremenko@xmath29 , s. egli@xmath42 , a. eliseev@xmath10 , e. elsen@xmath11 , s. essenov@xmath29 , a. falkiewicz@xmath43 , p.j.w .
faulkner@xmath25 , l. favart@xmath13 , a. fedotov@xmath29 , r. felst@xmath11 , j. feltesse@xmath44 , j. ferencei@xmath26 , l. finke@xmath11 , m. fleischer@xmath11 , a. fomenko@xmath10 , e. gabathuler@xmath37 , j. gayler@xmath11 , s. ghazaryan@xmath15 , a. glazov@xmath11 , i. glushkov@xmath45 , l. goerlich@xmath43 , m. goettlich@xmath46 , n. gogitidze@xmath10 , m. gouzevitch@xmath23 , c. grab@xmath20 , t. greenshaw@xmath37 , b.r .
grell@xmath11 , g. grindhammer@xmath28 , s. habib@xmath47 , d. haidt@xmath11 , m. hansson@xmath48 , c. helebrant@xmath11 , r.c.w .
henderson@xmath49 , h. henschel@xmath45 , g. herrera@xmath50 , m. hildebrandt@xmath42 , k.h .
hiller@xmath45 , d. hoffmann@xmath31 , r. horisberger@xmath42 , a. hovhannisyan@xmath15 , t. hreus@xmath51 , m. jacquet@xmath21 , m.e .
janssen@xmath11 , x. janssen@xmath13 , v. jemanov@xmath46 , l. jnsson@xmath48 , d.p .
johnson@xmath52 , a.w .
jung@xmath53 , h. jung@xmath11 , m. kapichine@xmath54 , j. katzy@xmath11 , i.r .
kenyon@xmath25 , c. kiesling@xmath28 , m. klein@xmath37 , c. kleinwort@xmath11 , t. klimkovich@xmath11 , t. kluge@xmath37 , a. knutsson@xmath11 , r. kogler@xmath28 , v. korbel@xmath11 , p. kostka@xmath45 , m. kraemer@xmath11 , k. krastev@xmath11 , j. kretzschmar@xmath37 , a. kropivnitskaya@xmath29 , k. krger@xmath53 , k. kutak@xmath11 , m.p.j .
landon@xmath55 , w. lange@xmath45 , g. latovika - medin@xmath14 , p. laycock@xmath37 , a. lebedev@xmath10 , g. leibenguth@xmath20 , v. lendermann@xmath53 , s. levonian@xmath11 , g. li@xmath21 , k. lipka@xmath46 , a. liptaj@xmath28 , b. list@xmath46 , j. list@xmath11 , n. loktionova@xmath10 , r. lopez - fernandez@xmath50 , v. lubimov@xmath29 , a .-
i .
lucaci - timoce@xmath11 , l. lytkin@xmath40 , a. makankine@xmath54 , e. malinovski@xmath10 , p. marage@xmath13 , ll .
marti@xmath11 , h .- u .
martyn@xmath56 , s.j .
maxfield@xmath37 , a. mehta@xmath37 , k. meier@xmath53 , a.b .
meyer@xmath11 , h. meyer@xmath11 , h. meyer@xmath57 , j. meyer@xmath11 , v. michels@xmath11 , s. mikocki@xmath43 , i. milcewicz - mika@xmath43 , f. moreau@xmath23 , a. morozov@xmath54 , j.v .
morris@xmath34 , m.u .
mozer@xmath13 , m. mudrinic@xmath24 , k. mller@xmath58 , p. murn@xmath59 , k. nankov@xmath60 , b. naroska@xmath61 , th .
naumann@xmath45 , p.r .
newman@xmath25 , c. niebuhr@xmath11 , a. nikiforov@xmath11 , g. nowak@xmath43 , k. nowak@xmath58 , m. nozicka@xmath11 , b. olivier@xmath28 , j.e .
olsson@xmath11 , s. osman@xmath48 , d. ozerov@xmath29 , v. palichik@xmath54 , i. panagoulias@xmath62@xmath63 , m. pandurovic@xmath24 , th .
papadopoulou@xmath62@xmath63 , c. pascaud@xmath21 , g.d .
patel@xmath37 , o. pejchal@xmath32 , h. peng@xmath11 , e. perez@xmath64 , a. petrukhin@xmath29 , i. picuric@xmath14 , s. piec@xmath45 , d. pitzl@xmath11 , r. plaakyt@xmath11 , r. polifka@xmath32 , b. povh@xmath40 , t. preda@xmath9 , v. radescu@xmath11 , a.j .
rahmat@xmath37 , n. raicevic@xmath14 , a. raspiareza@xmath28 , t. ravdandorj@xmath18 , p. reimer@xmath36 , e. rizvi@xmath55 , p. robmann@xmath58 , b. roland@xmath13 , r. roosen@xmath13 , a. rostovtsev@xmath29 , m. rotaru@xmath9 , j.e .
ruiz tabasco@xmath30 , z. rurikova@xmath11 , s. rusakov@xmath10 , d. salek@xmath32 , f. salvaire@xmath11 , d.p.c .
sankey@xmath34 , m. sauter@xmath20 , e. sauvan@xmath31 , s. schmidt@xmath11 , s. schmitt@xmath11 , c. schmitz@xmath58 , l. schoeffel@xmath35 , a. schning@xmath65 , h .- c .
schultz - coulon@xmath53 , f. sefkow@xmath11 , r.n .
shaw - west@xmath25 , i. sheviakov@xmath10 , l.n .
shtarkov@xmath10 , s. shushkevich@xmath28 , t. sloan@xmath49 , i. smiljanic@xmath24 , p. smirnov@xmath10 , y. soloviev@xmath10 , p. sopicki@xmath43 , d. south@xmath22 , v. spaskov@xmath54 , a. specka@xmath23 , z. staykova@xmath11 , m. steder@xmath11 , b. stella@xmath12 , u. straumann@xmath58 , d. sunar@xmath13 , t. sykora@xmath13 , v. tchoulakov@xmath54 , g. thompson@xmath55 , p.d .
thompson@xmath25 , t. toll@xmath11 , f. tomasz@xmath26 , t.h .
tran@xmath21 , d. traynor@xmath55 , t.n .
trinh@xmath31 , p. trul@xmath58 , i. tsakov@xmath60 , b. tseepeldorj@xmath66 , i. tsurin@xmath45 , j. turnau@xmath43 , e. tzamariudaki@xmath28 , k. urban@xmath53 , a. valkrov@xmath32 , c. valle@xmath31 , p. van mechelen@xmath13 , a. vargas trevino@xmath11 , y. vazdik@xmath10 , s. vinokurova@xmath11 , v. volchinski@xmath15 , d. wegener@xmath22 , m. wessels@xmath11 , ch .
wissing@xmath11 , e. wnsch@xmath11 , v. yeganov@xmath15 , j. ek@xmath32 , j. zlek@xmath36 , z. zhang@xmath21 , a. zhelezov@xmath29 , a. zhokin@xmath29 , y.c .
zhu@xmath11 , t. zimmermann@xmath20 , h. zohrabyan@xmath15 , and f. zomer@xmath21 |
electronic structure of correlated metal has a significant temperature dependence .
this can be seen in the momentum - resolved spectra , defined as @xmath70 in lda+dmft calculations , in which @xmath71 , @xmath72 , @xmath73 is the embedding operator and @xmath74 is the impurity self energy in orbital space . as mentioned in the main text , the quasiparticles ( qps ) are well defined and meaningful for the transport properties even when the scattering rate in the self energy is large .
the qp dispersion @xmath75 is defined as the solution to the following equation : @xmath76)=0 .
\label{qp}\ ] ] the spectra @xmath77 and the qp dispersion @xmath78 for two different temperature are depicted in fig .
[ spectra_dmft_v2o3](a)(b ) .
we find that increasing temperature broadens the spectra @xmath77 significantly due to the temperature dependence of qp scattering rate .
the temperature also modifies the qp dispersion in two aspects : with increasing temperature the qp bands near the fermi level are less renormalized in accordance with the temperature dependence of qp weight shown in the main text , and the qp bands are shifted accordingly .
both effects affect the transport properties .
the shifts of the quasiparticles can be described with the effective chemical potential @xmath79 , which in general has an orbital index due to orbital differentiation .
[ spectra_dmft_v2o3](c ) shows the temperature dependence of @xmath80 and @xmath81 .
the shifts of qp bands result in a change of fermi surface upon increasing temperature .
we note that our predicted qp bands with full electron calculations are different from those calculated with a downfolded hubbard model @xcite , where the orbital polarization was found to be very large and only a single fermi surface remains .
this is because the crystal field splitting of 2o3 is mainly hybridization driven and thus better described with @xmath82-@xmath35 like model rather than the hubbard model .
this prediction is to be verified by further detailed experiments especially with angular - resolved photoemission spectroscopy measurements .
( a ) and @xmath83 ( b ) .
the corresponding qp dispersion @xmath78 computed using eqn.[qp ] is shown with dots .
the effective chemical potentials for @xmath60 and @xmath38 orbitals are shown in ( c ) .
, title="fig : " ] ( a ) and @xmath83 ( b ) .
the corresponding qp dispersion @xmath78 computed using eqn.[qp ] is shown with dots .
the effective chemical potentials for @xmath60 and @xmath38 orbitals are shown in ( c ) .
, title="fig : " ] ( a ) and @xmath83 ( b ) .
the corresponding qp dispersion @xmath78 computed using eqn.[qp ] is shown with dots .
the effective chemical potentials for @xmath60 and @xmath38 orbitals are shown in ( c ) .
, title="fig : " ] 49ifxundefined [ 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 * * , ( ) link:\doibase 10.1103/physrev.134.a566 [ * * , ( ) ] @noop _ _ @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.80.4775 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.086401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.036401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.75.105 [ * * , ( ) ] link:\doibase 10.1126/science.1088386 [ * * , ( ) ] link:\doibase 10.1038/ncomms1109 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.047401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.205113 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.056401 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.83.471 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.165115 [ * * , ( ) ] link:\doibase 10.1002/pssb.200642097 [ * * , ( ) ] in link:\doibase 10.1007/978 - 1 - 4020 - 3463 - 3_7 [ _ _ ] , , vol . , ( , ) pp .
@noop ( ) , link:\doibase 10.1103/physrevlett.23.1384 [ * * , ( ) ] link:\doibase 10.1103/physrevb.7.1920 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.70.1039 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.90.186403 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.066805 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.075117 [ * * , ( ) ] link:\doibase 10.1103/physrevb.7.326 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.27.941 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.78.865 [ * * , ( ) ] link:\doibase 10.1080/00018730701619647 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.86.5345 [ * * , ( ) ] link:\doibase 10.1103/physrevb.73.045109 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.085127 [ * * , ( ) ] link:\doibase 10.1002/pssb.201248476 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.155121 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.195107 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/physrevb.75.155113 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.97.076405 [ * * , ( ) ] link:\doibase 10.1103/physrevb.61.11506 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.113107 [ * * , ( ) ] link:\doibase 10.1103/physrevb.45.8209 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.3451462 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.176401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.075125 [ * * , ( ) ] link:\doibase 10.1038/nphys1343 [ * * , ( ) ] link:\doibase 10.1073/pnas.1218846110 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.104509 [ * * , ( ) ] | we use low energy optical spectroscopy and first principles lda+dmft calculations to test the hypothesis that the anomalous transport properties of strongly correlated metals originate in the strong temperature dependence of their underlying resilient quasiparticles .
we express the resistivity in terms of an effective plasma frequency @xmath0 and an effective scattering rate @xmath1 .
we show that in the archetypal correlated material 2o3 , @xmath0 increases with increasing temperature , while the plasma frequency from partial sum rule exhibits the opposite trend . @xmath2 has a more pronounced temperature dependence than the scattering rate obtained from the extended drude analysis .
the theoretical calculations of these quantities are in quantitative agreement with experiment .
we conjecture that these are robust properties of all strongly correlated metals , and test it by carrying out a similar analysis on thin film on substrate .
understanding the transport properties in metallic states of strongly correlated materials is a long - standing challenge in condensed matter physics .
many correlated metals are not canonical landau fermi liquids ( lfl ) as their resistivities do not follow the @xmath3 law in a broad temperature range .
fermi liquid behavior emerges only below a very low temperature scale , @xmath4 , which can be vanishingly small or hidden by the onset of some form of long range order . above @xmath4 ,
the resistivity usually rises smoothly and eventually exceeds the mott - ioffe - regel limit , entering the so - called bad metal " regime@xcite with no clear sign of saturation @xcite . as stressed in ref .
an interpretation of the transport properties in terms of quasiparticles ( qps ) is problematic when the mean free path is comparable with the de broglie wavelength of the carriers and describing the charge transport above @xmath5 is an important challenge for the theory of strongly correlated materials .
it was shown in the context of the interacting electron phonon system , that the qp picture is actually valid in regimes that fall outside the lfl hypothesis@xcite .
there are peaks in the spectral functions which define renormalized qps even though the qp scattering rate is comparable to the qp energy .
the transport properties can be formulated in terms of a transport boltzman kinetic equation for the qp distribution function , which has precisely the form proposed by landau . solving the transport equation ,
the dc conductivity can be expressed as @xmath6 in analogy with drude formula .
the effective transport scattering rate @xmath7 characterizes the decay of the current carried by qps due to collisions involving umklapp effects , and @xmath0 is the low energy effective plasma frequency of qps and can be expressed in terms of qp velocities and the landau parameters@xcite .
the temperature dependence of the transport coefficients beyond the scope of lfl and many salient features seen in correlated oxides , such as their low coherence scale , non saturating resistivities and anomalous transfer of spectral weight , are described well in studies of doped hubbard model within the framework of dynamical field mean theory ( dmft ) ( for early reviews of this topic see @xcite )
. a complete understanding of the transport anomalies has been reached recently@xcite . as in the prange - kadanoff theory@xcite ,
the qps are resilient surviving in a broad region above @xmath4 @xcite and a quantum kinetic equation provides a quantitative description of the transport@xcite . while in the electron - phonon coupled system treated in ref .
the fermi liquid parameters such as the quasiparticle velocities and therefore @xmath0 are temperature independent , they are strongly temperature dependent in the doped mott insulator within dmft due to changes in the fermi surface at high temperatures @xcite and a strong temperature dependence of the effective mass at intermediate temperatures @xcite .
this strong temperature dependence of @xmath0 hides the more conventional temperature dependence of @xmath8 in the resistivity , which is quadratic in a broad region of temperatures and has saturating behavior at high temperatures @xcite .
strong temperature dependence in the qp electronic structure with the resulting temperature dependence of @xmath0 and @xmath8 thus provides a simple scenario to describe the anomalous transport of correlated metals . in this letter , we provide experimental and theoretical evidences that this picture holds beyond the dmft treatment of simplified hubbard model , and is indeed relevant to real materials .
we focus on 2o3 .
this archetypal correlated material provided the first experimental corroboration of the validity of the dmft picture of mott transition@xcite and is still a subject of intense experimental studies@xcite .
we propose formulas to extract the effective plasma frequency @xmath0 and effective scattering rate @xmath7 from optical conductivity and show that they display the predicted temperature dependence .
we contrast their temperature dependence to that of the plasma frequency and scattering rate extracted from the standard extended drude analysis . in correlated systems the optical conductivity is usually parametrized with the so - called extended drude analysis in terms of two frequency dependent quantities , the scattering rate @xmath9 and the mass enhancement @xmath10 @xcite , @xmath11 the plasma frequency @xmath12 is obtained with the partial sum rule @xmath13 and depends on the cutoff @xmath14 chosen so as to exclude interband transitions .
to test the theory , instead we focus on quantities that have a simple qp interpretation , namely @xmath15 and @xmath16 , from low frequency optical conductivity extracted as follows , @xmath17 when a direct determination of the imaginary part of optical conductivity ( as for example in ellipsometry measurements ) is not available , they can be extracted from @xmath18 only , using @xmath19 comparing with extended drude analysis , we have @xmath20 , @xmath21 .
thus this analysis is related to extended drude analysis , but free of partial sum rule .
similar low frequency analysis has been used in previous works@xcite , however the temperature dependence of @xmath0 was not the focus of those studies .
we apply the proposed analysis to 2o3 , a prototype of metal insulator transition ( mit ) @xcite .
pure 2o3 is a paramagnetic metal ( pm ) at ambient condition .
it enters antiferromagnetic insulating state ( afi ) below @xmath22 with a concomitant structural transition , and the afi can be quenched by ti - doping or pressure .
the pm can be turned into the paramagnetic insulator ( pi ) by slight cr - doping , which induces a first order isostructural transition with a small change in @xmath23 ratio , indicating a typical band - controlled mit scenario@xcite .
this first order transition ends at a second order critical point at temperature around 400@xmath24 @xcite .
the pm phase exhibits significant signatures of correlations , for instance , a pronounced qp peak and a broad lower hubbard band were revealed in photoemission spectroscopy measurements @xcite .
the pm phase is a fermi liquid at low temperature when afi state is suppressed @xcite . and ( b ) @xmath25 of 2o3 at different temperature is taken from ref .
@xcite , where dashed lines indicate data at @xmath26 very close to mit .
( c ) @xmath27 and ( d ) @xmath7 of 2o3 are extracted according to eqn .
[ eq : lowfrequencydrude ] . for comparison ,
@xmath28 and @xmath29 extracted from the extended drude analysis are also shown .
dashed lines are guides for the eyes by fitting @xmath16 ( @xmath28 , excluding @xmath26 ) and @xmath7 ( @xmath29 ) to linear and parabolic functions respectively.,title="fig : " ] and ( b ) @xmath25 of 2o3 at different temperature is taken from ref .
@xcite , where dashed lines indicate data at @xmath26 very close to mit .
( c ) @xmath27 and ( d ) @xmath7 of 2o3 are extracted according to eqn .
[ eq : lowfrequencydrude ] . for comparison ,
@xmath28 and @xmath29 extracted from the extended drude analysis are also shown .
dashed lines are guides for the eyes by fitting @xmath16 ( @xmath28 , excluding @xmath26 ) and @xmath7 ( @xmath29 ) to linear and parabolic functions respectively.,title="fig : " ] and ( b ) @xmath25 of 2o3 at different temperature is taken from ref .
@xcite , where dashed lines indicate data at @xmath26 very close to mit .
( c ) @xmath27 and ( d ) @xmath7 of 2o3 are extracted according to eqn .
[ eq : lowfrequencydrude ] . for comparison ,
@xmath28 and @xmath29 extracted from the extended drude analysis are also shown .
dashed lines are guides for the eyes by fitting @xmath16 ( @xmath28 , excluding @xmath26 ) and @xmath7 ( @xmath29 ) to linear and parabolic functions respectively.,title="fig : " ] and ( b ) @xmath25 of 2o3 at different temperature is taken from ref .
@xcite , where dashed lines indicate data at @xmath26 very close to mit .
( c ) @xmath27 and ( d ) @xmath7 of 2o3 are extracted according to eqn .
[ eq : lowfrequencydrude ] . for comparison ,
@xmath28 and @xmath29 extracted from the extended drude analysis are also shown .
dashed lines are guides for the eyes by fitting @xmath16 ( @xmath28 , excluding @xmath26 ) and @xmath7 ( @xmath29 ) to linear and parabolic functions respectively.,title="fig : " ] fig [ wsq_v2o3](a)(b ) shows the measured optical conductivity @xmath30 of pure 2o3 in pm phase@xcite .
pronounced drude peaks show up even when the resistivity is high ( of the order of @xmath31 ) and does not follow @xmath3-law@xcite .
the drude peak diminishes gradually upon increasing temperature , except at the lowest temperature where the transport is probably affected by the precursor of ordered phase . @xmath0 and @xmath8 extracted according to eqn .
[ eq : lowfrequencydrude ] are shown in fig [ wsq_v2o3](c)(d ) .
we find that @xmath16 increases with increasing temperature .
this is in contrast with the drude plasma frequency square @xmath32 obtained by the partial sum rule with a cut off @xmath33 , which slightly decreases@xcite except at the lowest temperature . @xmath8 increases with increasing temperature and
has the same trend as the scattering rate extracted with extended drude analysis at zero frequency @xmath29 , but with a much stronger temperature dependence .
the experimental data is consistent with an @xmath16 which has a term linear and a @xmath7 which is quadratic in temperature , revealing a fermi liquid behavior that is hidden in @xmath34 .
the analysis of the experimental data , thus corroborates the main qualitative predictions of the dmft description of transport properties in simple model hamiltonian@xcite .
we now argue that realistic lda+dmft @xcite calculations describe well the optical properties as well as the extracted quantities @xmath0 and @xmath8 , hence a local approximation , which ignores vertex corrections , is sufficiently accurate to capture the experimental trends .
lda+dmft investigations on 2o3 by several groups have successfully described the properties of this material near the mit @xcite .
the correlation in 2o3 is due to the partially filled narrow @xmath35-orbitals with a nominal occupancy @xmath36 .
the two electrons mainly populate the @xmath37 and @xmath38 states of vanadium due to surrounding oxygen octahedron with trigonal distortion .
we perform the lda+dmft calculations with an implementation as described in ref .
@xcite , which is based on wien2k package@xcite .
we use projectors within a large ( 20@xmath39 ) energy window constructing local orbitals .
this study thus includes explicitly the oxygen orbitals hybridizing with the d orbitals .
the interaction is applied to @xmath37 and @xmath38 orbitals only . to solve the impurity problem , we use continuous - time quantum monte - carlo method with hybridization expansion .
the brillouin zone integration is performed with a regular 12x12x12 mesh .
the muffin - tin radius is 1.95 and 1.73 bohr radius for v and o respectively .
the structure is taken from pure 2o3 at room temperature and only paramagnetic state is considered .
the coulomb interaction @xmath40 and the hund s coupling @xmath41 are set to @xmath42 and @xmath43 respectively .
they are consistent with the ones used in previous studies @xcite , but @xmath40 is slightly larger .
this is because in previous studies the relevant correlated orbitals are more extended due to downfolding or projection onto a small energy window and hence experience a reduced repulsion . with these parameters
the calculated total spectra is consistent with experiment photoemission spectroscopy measurements @xcite .
the occupancies of @xmath37 and @xmath38 orbitals at @xmath44 are 1.60 and 0.50 respectively , in good agreement with x - ray absorption spectroscopy @xcite measurements and previous lda+dmft calculations @xcite .
these parameters place 2o3 on the metallic side but close to the mit in the phase diagram ( temperature versus interaction strength ) with the temperature at critical endpoint of the first order mit close to the experimental findings @xcite . and compared to those extracted from experimental data .
dashed lines are guides to the eyes by fitting @xmath16 and @xmath7 to linear and parabolic functions respectively.,title="fig : " ] and compared to those extracted from experimental data . dashed lines are guides to the eyes by fitting @xmath16 and @xmath7 to linear and parabolic functions respectively.,title="fig : " ] and compared to those extracted from experimental data .
dashed lines are guides to the eyes by fitting @xmath16 and @xmath7 to linear and parabolic functions respectively.,title="fig : " ] we calculate the optical conductivity in a broad temperature range as shown in fig .
[ dmftoptic_v2o3](a ) . the main feature of the experimental optical conductivity , the drude peak and the shoulder structure at about @xmath45 as well as their scale , are reasonably reproduced in our calculations .
the drude peak is gradually diminished and merges with the shoulder structure at around @xmath46 , in agreement with experiments @xcite .
therefore lda+dmft calculation provides a satisfactory description of the optical properties of 2o3 . from the optical conductivity @xmath16 and @xmath7
are extracted using eqn .
[ eq : lowfrequencydrude ] .
as shown in fig [ dmftoptic_v2o3](b)(c ) , they agree reasonably well with those extracted from experimental data .
in particular the same trends found with the experimental data , thus the main charactersistics of the `` hidden '' fermi liquid behavior , show up more clearly in the broad temperature range studied in our calculations : @xmath16 appears linear and a @xmath7 appears quadratic versus temperature .
therefore the proposed analysis of both the experimental data and the first principle calculations reveals significant temperature dependence of qps in terms of @xmath16 and an extended quadratic temperature dependence of @xmath7 , but not of @xmath29 . to further understand the observations ,
let us recall the qp interpretation of low frequency optical conductivity in the dmft treatment of the doped single band hubbard model .
in this case , @xmath47 , in which @xmath48 and @xmath49 are the qp weight and life time , @xmath50 is the transport function @xmath51 and @xmath52 is the effective chemical potential of qps@xcite . applying the analysis in eqn.[eq : lowfrequencydrude ] , we have @xmath53 and @xmath54 , therefore in this simple model , @xmath16 and @xmath8 directly relate to the qp weight and life time . in situations where @xmath55 varies little with temperature ,
the observations above imply a strong temperature dependence of @xmath56 and @xmath57 .
we emphasize that although a strong dependence of scattering rate is generally expected in a fermi liquid , the temperature dependence of qp weight is not , but it was observed in model studies@xcite . we then extract from our calculated self energies , the qp weight ( which is the inverse of the mass enhancement within single site dmft ) defined as @xmath58 and the scattering rate defined as @xmath59 .
the qp weight and the qp scattering rate are shown in fig .
[ dmft_v2o3](a)(b ) .
there is orbital differentiation between @xmath60 and @xmath38orbitals as pointed out in earlier lda+dmft studies @xcite
. both orbital self energies exhibit the same trends observed in the studies of model hamiltonians : @xmath56 increases almost linearly with temperature and @xmath57 is nearly quadratic in temperature for each orbital in the temperature range considered .
note that @xmath60 orbitals have a much larger spectra weight at the fermi level than the @xmath38 orbital thus dominate the transport .
this pronounced temperature dependence is consistent with that of @xmath16 and @xmath7 extracted from optical conductivity . therefore the properties of the underlying qps , especially the temperature dependence of the qp weight and the qp scattering rate , are captured in our analysis on optical conductivities .
we note that in addition the temperature dependence of underlying qps manifest itself in the temperature dependence of the effective chemical potentials of qps ( see online supplementary ) , which also contribute to the the temperature dependence of @xmath16 . and ( b ) effective scattering rate @xmath61 of 2o3 extracted from lda+dmft self energies for @xmath60 and @xmath38 orbitals .
dashed lines are guides for the eyes by fitting @xmath56 and @xmath62 to linear and parabolic functions respectively , title="fig : " ] and ( b ) effective scattering rate @xmath61 of 2o3 extracted from lda+dmft self energies for @xmath60 and @xmath38 orbitals . dashed lines are guides for the eyes by fitting @xmath56 and @xmath62 to linear and parabolic functions respectively , title="fig : " ] we expect that this picture of anomalous transport in correlated materials is not limited to v2o3 and is in fact generally applicable to various strongly correlated metals . to check the validity of this general conjecture we apply the same analysis to experimental data of ( nno ) film on ( lao ) substrate .
nno is another typical correlated material exhibiting temperature - driven mit@xcite . while deposited as film on lao substrate , the mit can be quenched so that it remains metallic down to very low temperature @xcite .
high quality optical conductivities of nno film are taken from ref as shown in fig [ wsq_nno](a)(b ) .
we note that the resistivity is not @xmath3-like except possibly at the lowest temperature @xmath63 @xcite .
we perform the same analysis as above in 2o3 .
@xmath16 and @xmath7 are shown in fig .
[ wsq_nno](c)(d ) in comparison with @xmath28 and @xmath29 obtained by extended drude analysis with a cutoff of @xmath64 .
again we have the same features as in 2o3 : @xmath16 is linear in temperature and has the opposite trend with @xmath32 , while @xmath7 has a more pronounced quadratic behavior in a wide temperature range well above @xmath4 . and ( b ) @xmath25 of nno film on lao substrate at different temperature is taken from ref .
@xcite , from which ( c ) @xmath27 and ( d ) @xmath7 are extracted according to eqn.[eq : lowfrequencydrude ] . for comparison ,
@xmath28 and @xmath29 extracted from the extended drude analysis are shown .
dashed lines are guides for the eyes by fitting @xmath16 ( @xmath28 excluding @xmath65 ) and @xmath7 ( @xmath66 to linear and parabolic functions respectively.,title="fig : " ] and ( b ) @xmath25 of nno film on lao substrate at different temperature is taken from ref .
@xcite , from which ( c ) @xmath27 and ( d ) @xmath7 are extracted according to eqn.[eq : lowfrequencydrude ] . for comparison ,
@xmath28 and @xmath29 extracted from the extended drude analysis are shown .
dashed lines are guides for the eyes by fitting @xmath16 ( @xmath28 excluding @xmath65 ) and @xmath7 ( @xmath66 to linear and parabolic functions respectively.,title="fig : " ] and ( b ) @xmath25 of nno film on lao substrate at different temperature is taken from ref .
@xcite , from which ( c ) @xmath27 and ( d ) @xmath7 are extracted according to eqn.[eq : lowfrequencydrude ] . for comparison ,
@xmath28 and @xmath29 extracted from the extended drude analysis are shown .
dashed lines are guides for the eyes by fitting @xmath16 ( @xmath28 excluding @xmath65 ) and @xmath7 ( @xmath66 to linear and parabolic functions respectively.,title="fig : " ] and ( b ) @xmath25 of nno film on lao substrate at different temperature is taken from ref .
@xcite , from which ( c ) @xmath27 and ( d ) @xmath7 are extracted according to eqn.[eq : lowfrequencydrude ] . for comparison ,
@xmath28 and @xmath29 extracted from the extended drude analysis are shown .
dashed lines are guides for the eyes by fitting @xmath16 ( @xmath28 excluding @xmath65 ) and @xmath7 ( @xmath66 to linear and parabolic functions respectively.,title="fig : " ] in conclusion , in this letter we point out that the anomalous transport properties observed in many transition metal oxides arise from a temperature dependent @xmath16 and @xmath7 .
we establish that by analyzing both the experimental and the theoretical data .
this scenario calls for further investigations in other compounds , starting from systems where there are already preliminary indications that it applies , for example , caru@xmath67 where a similar low energy analysis was performed @xcite . in many other systems , such as nickelate and pnictides ,
a temperature dependent @xmath68 is seen in the extended drude analysis @xcite .
thus an extraction of the low energy effective plasma frequency @xmath16 as outlined in this paper , and a comparison with @xmath32 from the partial sum rule would be illuminating .
recent optical spectroscopy studies of underdoped cuprates @xcite at low temperatures revealed temperature independent @xmath16 and a quadratic temperature dependence in @xmath69 , consistent with the earlier theoretical predictions of cluster dmft @xcite
. it would be interesting to extend the measurements to higher temperatures where deviations from canonical fermi liquid are expected .
finally high resolution studies using spectroscopies such as arpes and stm in quasiparticle interference mode would be very useful to separate the various contributions to the temperature dependence of @xmath16 by probing directly the electronic structure .
we acknowledge very useful discussions with a. georges and p. armitage .
this work was supported by nsf dmr-1308141 ( x. d. and g.k ) , nsf dmr 0746395 ( k. h. ) and doe - bes ( a. s. and d. b. ) . |
the formation of a flattened disk surrounding a young stellar object is a natural outcome of a rotating and collapsing molecular core .
a core with initial angular momentum leads to a flattened structure as material is accreted toward the protostar while conserving angular momentum .
one of the puzzles in star formation is the so called `` angular momentum problem '' .
as matter is accreted toward the star , it has to shed significant angular momentum in order for the accretion to continue .
disks , observed in protostellar systems of a wide range of masses ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , may play a pivotal role in the mass growth of protostars by extracting excess angular momentum in the infalling matter through winds and outflows along the polar direction @xcite . in theoretical models of star formation ,
magneto hydrodynamic winds extract angular momentum from the disk @xcite . as a consequence
, matter can be accreted onto the forming star .
while theories on protostellar wind do not agree on the location at which the wind is launched , the role it plays in shedding angular momentum is a common feature in all theoretical models . despite the theoretical consensus ,
observational confirmation of wind angular momentum has been elusive for decades .
high spatial resolution differential spectroscopy of optical jets in t tauri stars showed early promise in revealing transverse velocity gradients indicative of jet rotation ( e.g. * ? ? ? * ; * ? ? ?
* ) . however , many of the candidates were dismissed because of a lack of kinematic consistency . in the 2000s ,
the search for jet rotation was extended to molecular jets and outflows with the advance in millimeter and sub - millimeter wavelength interferometry @xcite . using high j transitions of sio @xcite that trace material in the primary jet / wind launched from the disk ,
@xcite reported velocity gradients in the hh211 sio jet and constrained the jet launching radius to within 0.05 au of the star .
however , subsequent observations at a better angular resolution did not confirm these results @xcite .
the challenge of detecting rotation in the optical and molecular jets lies in the fact that the spectral line emission from the jet is broad ( 20 to 100 -1 ) due to turbulence and other motions , making it difficult to discern any velocity gradients that are much smaller in magnitude .
in addition , asymmetries in shocks can produce velocity gradients that mimic rotation .
more recently , velocity gradients were reported in co outflows ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , demonstrating another promising avenue in the search for jet rotation .
however , co emission may not trace the primary wind due to its low critical density , and is often mixed with the surrounding molecular gas accelerated by the wind . hence
, its interpretation can be complicated by the distribution of the circumstellar material .
a more complete review on the subject can be found in @xcite and @xcite .
the hydrogen recombination masers around the young star mwc349a offer a unique opportunity for studying the structure and dynamics in a disk and wind at an angular resolution unattainable using thermal emission .
the target of interest , mwc349a , is in a binary system located at a distance from the sun of 1.2 kpc @xcite .
the object is classified as a herbig b[e ] star surrounded by ionized gas seen in hydrogen recombination line emission @xcite as well as continuum emission at centimeter wavelengths @xcite .
its companion , mwc349b , is offset by @xmath2 .
a remarkable and peculiar property of mwc349a is that a subset of the @xmath0 transition of the hydrogen recombination lines are masing , making it one of a small number of hydrogen recombination maser sources known .
the maser emission was first identified in the h29@xmath0 , h30@xmath0 , and h31@xmath0 transitions @xcite .
additional maser transitions were discovered subsequently at sub - millimeter wavelengths @xcite , and later at infrared wavelengths @xcite .
it appears that the @xmath0 transitions from @xmath3 all exhibit maser action @xcite , while at n @xmath4 , the recombination spectra all exhibit a single feature of thermal like gaussian profiles .
high angular resolution observations of the h30@xmath0 line revealed that the emission occurs at a collection of discrete locations from a disk structure @xcite .
however , subsequent observations of the h30@xmath0 transition by @xcite found that only those closer to the systemic velocity arise from the disk ; masers at more extreme velocities from the systemic velocity are formed in a wind . using a non - lte radiative transfer calculation
, @xcite modeled the h30@xmath0 masers and constrained the disk to a nearly edge - on geometry with an inclination angle of @xmath5 from the line of sight . here we present simultaneous observations of the h30@xmath0 and h26@xmath0 masers with the submillimeter array ( sma ) in its very extended configuration .
the high signal - to - noise data , coupled with an angular resolution as fine as @xmath6 , yield relative astrometry for masers better than 2 milli - arcsecond ( mas ) precision for masers stronger than 20 jy . in addition , the first high resolution h26@xmath0 transition images reveal that the lower quantum number masers are excited in the inner part of the disk as well as in the denser part of the wind than the h30@xmath0 line masers .
we found that masers in the disk are amplified in a thin annulus at a fixed radius for each hydrogen recombination transition .
furthermore , the maser velocity structure in the wind demonstrates keplerian - like rotation , which implies that a substantial fraction of the angular momentum is extracted by the wind .
the paper is organized as follows : in section 2 , we present the observational setup and the details of data processing . in section 3
, we present the observational results . in section 4
, we discuss the implication of the results .
a brief conclusion is provided in section 5 .
observations of mwc349a were carried out on 2012 october 12 using the sma @xcite in the very extended configuration .
six antennas were operational during the observations , providing projected baselines from 84 to 509 m. observations made use of both 230 ghz and 400 ghz receivers simultaneously , targeting the h@xmath7 and h@xmath8 lines at rest frequencies of 231.9009 ghz and 353.6228 ghz , respectively .
the digital correlator was configured to a resolution of 256 channels per 104-mhz window for three contiguous windows close to the recombination lines . for this configuration , the channel spacing is 0.41 mhz , corresponding to 0.50 and 0.35 -1 at 232 and 354 ghz , respectively .
the remaining windows in the 2-ghz band were set to 32 channels per 104-mhz band .
the zenith opacity reported from the cso @xmath9-meter was 0.1 at 225 ghz .
the double side band system temperatures varied from 80 to 150 k for the 230 ghz receivers , and 230 to 600 k for the 400 ghz receivers .
the atmospheric phases were stable during the start of the track , but deteriorated after 6 hours into the observations .
the phase center for mwc349a during the observations was @xmath10=@xmath11 , @xmath12=@xmath13 .
qso j2015 + 371 , at an angular distance of @xmath14 from mwc349a , was observed periodically to monitor gain variations during the course of the observations .
in addition , callisto was observed for flux calibration and 3c84 for passband calibration . since accurate passband calibration is essential for determining the astrometry of maser spots across the spectral line , we first observed 3c84 for 2 hours in the dual receiver mode with the lower - frequency if tuned to the h30@xmath0 line and the high - frequency if tuned to the h26@xmath0 line .
additional passband data were taken for the high - frequency if by rerouting the lower - frequency signal through the high - frequency if path to characterize the passband response .
this approach enabled us to obtain higher signal - to - noise passband data for the h26@xmath0 line since 3c84 is brighter and system temperatures are lower in the 230 ghz band .
we performed data calibration using the idl subset mir software package .
the absolute flux was bootstrapped to callisto , which yields a flux density of 2.3 jy at 230 ghz and 1.6 jy at 345 ghz for qso j2015 + 371 .
the continuum flux densities derived for mwc349a are 1.7 jy at 230 ghz and 2.4 jy at 345 ghz , in agreement with the model of an ionized wind source ( e.g. , * ? ? ?
the time - dependent gain calibration is achieved through the nearby qso j2015 + 371 .
for passband calibration of the 3 windows with 256 channels each , we averaged every 16 channels ( equivalent to 6.5 mhz ) of the 3c84 data before solving for antenna based passband solutions .
the smoothing reduces the noise contribution from the passband solution to the spectral line data of mwc349a ( see section 3 ) . in this approach ,
passband variations on scales less than 6.5 mhz are smoothed out .
we believe that there is no systematic phase structure on frequency scales smaller than 6.5 mhz .
figure 1 presents visibility phase of the baseline involving antennas 5 and 6 for the h30@xmath0 and h26@xmath0 lines .
as shown in the figure , visibility phases in the velocity range from 25 to 42 -1 have no apparent phase structure within 16 channels . after applying passband solutions ,
the phase residual in the higher resolution spectral channels of 3c84 is 2@xmath15 . for the lower spectral resolution windows ,
no average was performed when deriving the passband solution
. the calibrated visibility data of mwc349a were exported to the miriad format .
since the continuum emission of mwc349a is unresolved ( expected to be @xmath16 at 232 ghz and @xmath17 at 354 ghz according to the model of * ? ? ?
* ) in the very extended array configuration , we performed self calibration on the continuum data using a time interval of 30 sec to further remove phase fluctuations at shorter time scales .
the self calibrated spectral line visibility data were then fourier transformed and the images were cleaned . for the h26@xmath0 line , the continuum data did not have sufficient signal - to - noise ratios for self calibration at time intervals of several minutes .
hence , we resorted to using the strongest maser feature at @xmath18 -1 for self calibration .
since the self calibration of the h30@xmath0 and h26@xmath0 lines adopted different references , images of the h26@xmath0 and h30@xmath0 emission are not registered in position .
the synthesized beam is @xmath19 for the h26@xmath0 line and @xmath20 for the h30@xmath0 line .
the @xmath21 rms noise in the self calibrated images is 80 mjy and 110 mjy per 0.5 -1 channel in the h30@xmath0 and h26@xmath0 lines , respectively .
figure 2 presents the spectra of the h30@xmath0 and h26@xmath0 lines .
the spectra are color coded by the velocity of the channel .
both recombination line spectra show two emission peaks near @xmath22 and 31 -1 , respectively .
the redshifted peak is a factor of 2 stronger than the blueshifted peak . in between the two peaks
lies a flat portion of the spectrum detected at a level of more than 5 jy .
maser spots arise from compact regions that are not spatially resolved by the sma observations at resolutions of @xmath23 . if visibility phases are well determined across the spectral line , centroid fitting to the maser emission yields astrometry with precisions better than the spatial resolution .
as shown in figure 1 , over the spectral window of 100 -1 , the visibility phase varies from @xmath24 at @xmath25 -1 to @xmath26 at 35 -1 for the h30@xmath0 line , indicating a systematic spatial shift in the maser position .
the ability to measure the position is limited by the phase noise in the spectrum , which in turn is related to the signal - to - noise ratio ( snr ) in the data . the position error ,
@xmath27 is related to the size of the synthesized beam and the snr through the approximate relation @xmath28 @xcite .
the synthesized beam size at 354 ghz for the h26@xmath0 line observations was @xmath19 .
we calculated a noise value of 110 mjy by measuring the standard deviation in a 0.5 -1 wide channel that does not contain significant maser emission . for h26@xmath0 masers with flux intensities of 40 jy ,
the corresponding position error was 0.8 mas .
the synthesized beam at 232 ghz for the h30@xmath0 observations was @xmath29 .
the @xmath21 rms in the channel of width 0.5 -1 was 80 mjy , which gave a position error of 0.6 mas for maser features of 40 jy .
in addition to statistical errors from centroid fitting , the phase noise in the passband data introduces additional error to the maser position through passband calibrations .
the passband calibrator 3c84 had a flux density of 9.5 jy at the 230 ghz band during the time of the observations .
if one ignores the slight difference in the on - source time between the 3c84 and mwc349a observations , the noise contribution from the passband is equivalent to that of a flux density of 9.5 jy .
we improve the snr in the passband solution by a factor of 4 by applying a smoothing of 16 channels when deriving antenna based passband solutions .
after the channel averaging , the phase noise in 3c84 is reduced to @xmath30 = 2@xmath15 .
this phase value is equivalent to a position error , @xmath31 of 2.5 mas and 1.8 mas for h30@xmath0 and h26@xmath0 , respectively .
the total error @xmath32 in each maser position is given by @xmath33 .
figure 3 shows the spatial distribution of the h26@xmath0 and h30@xmath0 spots determined from the centroiding method .
here we only plot the data in the velocity range of @xmath34 to 47 -1 because of signal - to - noise considerations .
the error bar corresponds to the @xmath35 error given by @xmath32 .
masers from @xmath36 to 25 -1 for both transitions are mostly distributed in a linear structure in the southeast - northwest direction .
the least - squares fit to the maser distribution within this velocity range yields a position angle of @xmath37 for the h30@xmath0 emission and @xmath38 for the h26@xmath0 emission .
the position angles of the two maser distributions are consistent within the statistical error in the fit .
the position angle of the radio continuum emission is @xmath39 @xcite .
the position angle of the maser structure is consistent with the orientation of the disk inferred from the centimeter - wavelength continuum and the previous high resolution data of h30@xmath0 obtained from the sma and pdbi @xcite .
as shown in figure 3 , the maser emission exhibits a velocity gradient from southeast to northwest .
masers are red shifted toward the southeast and are blue shifted toward the northwest .
the sense of velocity gradients is the same for the two lines .
the spatial extent of the h26@xmath0 maser spots in the disk is slightly smaller than that of the h30@xmath0 maser spots in the disk .
this implies that h26@xmath0 masers arise from zones closer to the center of the disk than the corresponding h30@xmath0 maser spots .
in addition to masers distributed along the linear structure , those with velocities greater than about 20 -1 from the systemic velocity are distributed off the disk plane .
this is most apparent in the h26@xmath0 masers at velocities from 30 to 60 -1 and @xmath25 to @xmath40 -1 .
it has been suggested that these masers arise from the ionized wind of gas lifted off the disk plane @xcite . to compare the spatial distribution and kinematics between the two maser transitions , their relative positions must be aligned to an accuracy of 2 mas , the uncertainty limited by the passband calibration . as seen in figure 3
, there is a relative spatial shift between the h30@xmath0 and h26@xmath0 masers due to the self calibration that referenced different sources ( see section 2 ) .
we explored several avenues to align the two maser distributions .
the first approach is to align the 230 ghz and 345 ghz continuum emission .
since the 230ghz and 345 ghz continuum emission arise from the free - free emission of the ionized gas and are spatially unresolved in the sma observations , one expects that they share the same position .
we fit the two continuum images .
the fit to the 230 ghz continuum data yields an offset ( @xmath41 ) = ( @xmath42 ) , as expected from the data after self calibration using a point source model at the phase center . for the 345 ghz continuum ,
reliable fits were not attainable at all due to residual phase variations despite the self calibration .
therefore , we resorted to the second approach by using the strongest blueshifted and redshifted features in the h30@xmath0 and h26@xmath0 masers in the disk .
this approach assumes that two maser transitions in the disk share a common geometric center ( see discussions in section 4.1 ) .
we use strongest masers in the disk for alignment since they have the most reliable positions due to their high snr .
in addition , they arise from the tangential points in the disk ( see discussions in section 4.1 ) . the centers derived this way are ( @xmath43 ) = ( @xmath44 ) for the h30@xmath0 masers and ( @xmath45 ) for the h26@xmath0 masers , respectively .
these centers are marked by the large triangles in figure 3 .
to further analyze the kinematics , we present rotation curves for both maser transitions in figure 4 .
a rotation curve is the line - of - sight velocity against projected distance @xmath46 , defined as the separation between masers and the center of the disk projected on the plane of the sky .
the rotation curves of the h30@xmath0 and h26@xmath0 masers exhibit a linear relation in the position - velocity diagram between @xmath36 to 25 -1 .
maser velocities outside of this range decrease at increasing @xmath46 .
more significantly , the linear portion of the rotation curve shows different slopes between the maser transitions , with the h26@xmath0 masers exhibiting a steeper gradient than those of the h30@xmath0 .
based on the spatial distribution and the kinematics shown in figures 3 and 4 , we divide the masers in two groups : group _ i _ are those close to the cloud systemic velocity from @xmath36 to 25 -1 ; group _ ii _ are those at the blueshifted velocities of @xmath47 @xmath48 -1 and at the redshifted velocities of @xmath49 30 -1 .
the group _ i _
maser features between @xmath36 to 25 -1 are distributed in a linear strip in the southeast - northwest direction .
they consist of emission at the flat part of the spectra as well as those with increasing intensities toward the blueshifted and redshifted velocities ( see figure 2 ) .
in addition , these masers lie in the linear portion of the rotation curve .
this group of masers likely arises from the disk at a fixed radius of @xmath50 as proposed by @xcite based on the study of the h30@xmath0 masers .
the position angle of the maser disk is perpendicular to the position angle of the wind in @xcite and @xcite , respectively .
the h26@xmath0 group _ i _
masers are distributed in a linear extent smaller than that of the h30@xmath0 masers .
in addition , the former maser transition exhibits a larger velocity gradient .
both findings indicate that h26@xmath0 masers are formed inside the radius of the h30@xmath0 masers .
this is consistent with the fact that the formation of the h26@xmath0 masers requires higher densities @xcite , a condition that is met in a centrally condensed disk .
these linear relations suggest a simple model in figure 5 : an edge - on disk with masers in each transition distributed in a thin annulus at a fixed radius from @xmath51 to @xmath52 . here
@xmath53 is the angle between the line of sight and the vector connecting the star and the maser .
it is conceivable that masers populate the entire annulus .
however , the continuum emission from the disk within the annulus can be optically thick @xcite .
thus , it attenuates the maser emission in the section of the annulus from @xmath54 to @xmath55 in an edge - on disk . with this geometry , the line - of - sight velocity with respect to the systemic velocity , @xmath56 , is given by @xmath57 where @xmath58 is the circular velocity of the masers .
such a geometry leads to @xmath59 for keplerian motion . here
@xmath60 is the enclosed mass , and the projected distance @xmath61 .
this relation explains the constant velocity gradient in figure 4 for masers between @xmath36 to 25 -1 .
masers with close to the systemic velocity lie in the section closer to the line of sight between the observer and the star ( with small angles of @xmath53 ) .
masers at the more blueshifted and redshifted ends in this group are close to the tangent point in the annulus ( with @xmath62 ) . the velocity gradient of the group _
i _ masers in the disk can constrain the enclosed mass within the orbit
. a least - squares fit to the rotation curves yields a velocity gradient @xmath63 of @xmath64 -1 per arcsecond , and a systemic velocity @xmath65 of @xmath66 -1 for the h30@xmath0 masers , and @xmath67 -1 per arcsecond and a systemic velocity of @xmath68 -1 for the h26@xmath0 masers , respectively .
the systemic velocities of the two maser transitions are consistent within the error bar , and are close to the mean velocities between the blueshifted and redshifted maser peaks @xcite .
we adopt the average velocity of 7.2 -1 as the systemic velocity of the system .
[ 2 ] shows that the ratio of annular radii of the maser disks at 232 and 354 ghz is @xmath69 .
this is one of the more robust quantitative results of this study . for discussion purposes we can convert this ratio into a power law form , @xmath70
this dependence follows the prediction of non - lte maser emission in an environment where the density decreases with radius .
@xcite made detailed maser models for mwc349 based on the population departure coefficient calculations of @xcite .
they found that the maser emission from transitions with upper quantum number @xmath71 is fairly sensitive to density @xmath72 and follows the relation @xmath73 .
@xcite followed a somewhat different methodology and determined that @xmath74 .
if we adopt the scaling law @xmath75 , we can use our measurement to derive a power law dependence for @xmath72 as follows . since the recombination line frequency scales as @xmath76 ,
our results indicate that the radius - frequency relation @xmath77 becomes @xmath78 or @xmath79 .
hence our result is consistent with non - lte maser theory and a radially declining density profile which is often found in thin accretion disks ( e.g. , * ? ? ?
we now turn our attention to the estimation of the central stellar mass of mwc349a from the aligned rotation curves shown in figure 6 . in this figure
the systemic velocity of 7.2 -1 has been removed and the fits to the linear parts of the rotation curve shown .
we believe that these rotation curves trace the emission to the limiting azimuth angles from @xmath80 to @xmath52 for the reasons described in section 4.2 .
however the exact separation between the wind and disk masers is not sharply defined .
we judge that a mass of 10 , which is appropriate for a maximum projected diameter of 51 and 42 mas ( see figure 6 ) for the h30@xmath0 and h26@xmath0 emission is an appropriate determination .
it is possible that the masers with velocities up to 19 and 21 -1 should be included in the disk , in which case the central mass would be 15 .
this mass range of 10 to 15 is significantly smaller than previous estimates of 25 to 35 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) which do not make a distinction between the disk masers that are in true keplerian motion , and wind masers that trace gas not gravitationally bound to the star .
similar to the group _
masers , the group _ ii _ masers present distinct spatial and kinematic characteristics as well .
maser emission at @xmath49 30 -1 is found in a linear structure off the plane of the disk with a noticeable spatial offset between the two transitions . both h30@xmath0 and h26@xmath0 maser features are distributed in an approximately linear structure extending 0.02@xmath81 to the northwest at a position angle of @xmath82 and @xmath83 , respectively ( see figure 3 ) .
the maser features at blueshifted @xmath47 @xmath84 -1 are also offset from the disk plane for the h26@xmath0 line , although the extent is relatively smaller than the redshifted masers .
furthermore , these masers are located outside of the linear part of the rotation curves ( see figure 6 ) .
these properties suggest that group _ ii _ masers arise from the wind as proposed by @xcite .
the strongest masers in this group are likely excited along the tangent point in the wind bubble where a longer path of amplification is achieved ( see discussion in section 4.2 ) . the fact that these masers do not follow a keplerian rotation is consistent with the notion that they are not dynamically bound to the star .
hydrogen recombination maser emission in mwc349a exhibits a double - peaked profile .
assuming that masers are excited in a disk with an inner hole , @xcite reproduced the double - peaked profile when the inner radius is greater than 70% of the outer radius .
our observations confirm that masers from the same transition are indeed excited in a thin annulus of a fixed radius .
this geometry shown in figure 5 can explain why the group _ i _ maser emission from the @xmath54 points in the disk is much stronger than that from near @xmath51 .
our analysis is based on a simplistic model rather than on a detailed radiative transfer calculation in @xcite .
the velocity along the line of sight near @xmath54 is given by @xmath85 where @xmath86 is @xmath87 .
note that the velocity gradient is only zero at @xmath88 .
we define a coherent amplification bandwidth @xmath89 , which gives the azimuthal angle range contributing to the maser line formation of @xmath90 the coherent amplification length is then @xmath91 , or @xmath92 for the h30@xmath0 transition we have @xmath58 = 17.2 -1 at @xmath50 = 25 mas ( 30 au ) . if we adopt @xmath93 = 0.5 -1 then @xmath30 = 0.48 radians or 27@xmath15 .
the coherent amplification length is therefore 12 mas ( 14 au ) . at @xmath51
the velocity gradient is zero for all radii , but the population inversion is highly sensitive to the density which decreases monotonically with @xmath50 . since our model for the h26@xmath0 emission has @xmath50 = 21 mas ( 25 au ) , we assume that the peak shifts by about 1.0 mas per quantum number , @xmath94 ( 25 - 21)/4 = 1.0 mas ( 1.2 au ) .
we further assume that the coherent amplification length is approximately equal to twice this value or 2 mas ( 2.4 au ) the amplification length ratio , @xmath95 , is 12/2 or about 6 . depending on details of the maser model the maser intensity scales exponentially with @xmath96 for unsaturated amplification , @xmath97 for saturated amplification in a filamentary structure , and @xmath96 for saturated amplification in a one dimensional geometry @xcite .
for these three cases the intensity ratio are expected to be 403 ( @xmath98 ) , 216 ( @xmath99 ) , and 6 , respectively .
the observed intensity ratio for our experiment is about 16 and 8 for the redshifted and blueshifted lines , respectively .
historically , these ratios vary over the range 5 - 30 @xcite .
this range fits within the range of our simple model , perhaps favoring the one - dimensional saturated case .
this situation differs from the case of water megamasers where the annular emission region is thick and masers near the systemic velocity dominate because of a larger coherent length ( e.g. * ? ? ?
* ) . while the spatial characteristics of the masers in the disk seem to be fairly stable over the time scale of several years based on the disk maser distributions observed in 2004 , 2005 and 2012 ( * ?
* ; * ? ? ?
* and this paper ) , the masers in the wind seem to be more variable and sensitive to excitation conditions .
we note that the redshifted wind features near 30 -1 extend below the disk in the h30@xmath0 image ( see figure 3 ) , while they extend above the disk in the h26@xmath0 image .
this suggests that maser amplification may be sensitive to density and other excitation conditions .
in addition , in the 2005 data by @xcite the h30@xmath0 feature extends above the disk , suggesting temporal variability .
we note that the intensity profile monitored over many years by @xcite shows considerable variation , which indicates changing excitation conditions .
the angular momentum in stellar systems like the sun is four orders of magnitude lower than that in dense cores .
therefore , angular momentum has to be redistributed during the collapse of dense cores and formation of protostars . while the formation of binaries and multiples can be effective in redistributing angular momentum ( e.g. * ? ? ?
* ) , disks must shed sufficient angular momentum through the wind and the jet in order for the gas to be accreted onto the protostar .
the wind is launched via a magneto hydrodynamical process from the disk and then interacts with the ambient material to give rise to outflows seen in spectral lines and continuum in the optical , infrared and radio wavelengths .
the two models on protostellar outflows , i.e. the x - wind @xcite and the disk wind @xcite , differ significantly in zones where the wind is launched .
the x - wind model has the wind launched in the co - rotation radius between the protostar and the disk at a radius of @xmath100 au , while the disk wind model features the launch radius of several au . since the corresponding angular scales of the jet launching radii are @xmath101 mas even for the nearby star formation regions such as the taurus molecular cloud , they are beyond the sensitivity / resolution of many telescope facilities .
observations of jet rotation that occurs at much larger spatial scales offer the added benefit of constraining the jet launching zones and distinguishing the two outflow models . in an attempt to confirm that protostellar jets carry out angular momentum , considerable observational effort has been devoted to searching for rotation in protostellar jets and outflows using thermal emission @xcite .
masers seen in jets and outflows complement the thermal probes thanks to their brightness that enables much higher angular resolution observations .
@xcite presented multi - epoch observations of an h@xmath102o maser jet in the massive protostar s235ab - mir using the vlbi network vera .
they found a velocity gradient and proper motion in masers that are consistent with jet rotation .
the hydrogen recombination masers in mwc349a offer the direct probe of the ionized wind and its kinematics .
it appears that the group _ ii _ masers with close to @xmath48 and 30 -1 have projected line - of - sight velocities similar to a keplerian relation .
masers with much less than @xmath48 -1 or much greater than 30 -1 are fainter , thus do not have sufficient snr to determine their positions accurately .
the radio continuum observations of mwc349a yield electron densities in the wind , thus providing a direct measure of the mass loss in the ionized gas .
@xcite carried out centimeter - wavelength continuum observations and estimated a mass loss of @xmath103 yr@xmath104 .
stars in the protostellar phase of their evolution are expected to experience much higher rates of accretion and mass loss than stars in the more evolved phase of their evolution . for an order of magnitude estimate , we assume a constant mass loss rate over @xmath105 yrs , the accretion time scale of a massive protostar @xcite
. then , the lower limit of the total mass loss over the lifetime of its accretion amounts to 2.5 .
the present disk mass of mwc349a is constrained to @xmath106 based on the near ir observations @xcite .
it is reasonable to expect that the star is surrounded by a more massive disk during the protostellar stage @xcite .
nevertheless , the lower limit of the total mass loss during the entire accretion history is a significant fraction of the disk mass .
therefore , the wind removes significant angular momentum during the accretion phase of the star formation .
the spatial distribution of the maser spots also indicates that the wind is launched at a radius as far out as 25 au , much larger than models of the x - wind @xcite or the disk wind @xcite . since the maser emission may not fully trace the distribution of the mass , it is not certain how the material is distributed in the wind .
while the masers are distributed over a radius of 25 au , it remains possible that the majority of the mass ejection occurs at much smaller radii close to the ( proto)star .
observations of recombination masers of lower quantum numbers will help define the wind structure in the higher density regime .
the high signal - to - noise ratio data obtained with the sma at sub - arcsecond resolutions provide precise relative astrometry of h30@xmath0 and h26@xmath0 masers in mwc349a .
we found \(1 ) two groups of masers : group _ i _ masers with close to the systemic velocity from @xmath36 to 25 -1 lie in a disk ; and group _ ii _ masers outside of the velocity range are distributed off the disk in a wind ; \(2 ) the h26@xmath0 masers are distributed in a region inside that of the h30@xmath0 masers , indicating that the former are excited at higher density regimes in the disk and the wind .
the kinematics of masers in the disk indicate that they form in a thin annulus of a fixed radius for each maser transition ; \(3 ) the kinematics in the masers reveal a wind in the young stellar system that rotates in the same sense as the disk rotation .
mwc349a offers strong observational support that the wind extracts angular momentum from the disk . in doing so
, it facilitates gas accretion toward the star through the disk .
we thank k. young for his help in setting up sma observations and j. weintroub for discussions during the data analysis .
data presented in this paper were taken during the astronomy 191 course in experimental astrophysics at harvard university through the director discretionary time .
we thank r. blundell for his support of the project .
we appreciate the comments of the anonymous referee , which helped improve the clarity of the paper . and h26@xmath0 lines for a baseline of 420 m. the phase variations across the spectral window are due to the differences in position of the maser spots .
the nominal phase sensitivity is 1.2@xmath15 mas@xmath104 and 1.7@xmath15 mas@xmath104 at the frequencies of the h30@xmath0 and h26@xmath0 transitions , respectively.,width=566 ] and h30@xmath0 masers on the disk .
the red crosses represent the hydrogen recombination masers .
the central star marks the position of the young stellar object .
the h26@xmath0 masers are located at the inner part of the disk of a radius of @xmath107 ( 25 au ) , whereas the h30@xmath0 masers are located at the outer part of the disk of a radius of @xmath108 ( 30 au ) .
additional masers in the wind are not depicted here . ,
width=566 ] ( * red squares * ) and h26@xmath0 ( * blue triangles * ) maser emission .
the projected maser distance @xmath46 is computed as @xmath109 with respect to the geometric center marked in figure 3 .
@xmath56 is the maser line - of - sight velocity relative to the systemic velocity of 7.2 -1 .
the red and blue lines represent the least - squares fits to the rotation curves of the h30@xmath0 and h26@xmath0 masers , respectively .
the black dash - dotted lines and dashed lines represent keplerian rotation with a central mass of 10 and 15 , respectively . , width=491 ] | disk winds are thought to play a critical role in star birth .
as winds extract excess angular momentum from accretion disks , matter in the disk can be transported inward to the star to fuel mass growth . however , the observational evidence of wind carrying angular momentum has been very limited .
we present submillimeter array ( sma ) observations of the young star mwc349a in the h26@xmath0 and h30@xmath0 recombination lines .
the high signal - to - noise ratios made possible by the maser emission process allow us to constrain the relative astrometry of the maser spots to a milli - arcsecond precision .
previous observations of the h30@xmath0 line with the sma and the plateau de bure interferometer ( pdbi ) showed that masers are distributed in the disk and wind . our new high resolution observations of the h26@xmath0 line reveal differences in spatial distribution from that of the h30@xmath0 line .
h26@xmath0 line masers in the disk are excited in a thin annulus with a radius of about 25 au , while the h30@xmath0 line masers are formed in a slightly larger annulus with a radius of 30 au .
this is consistent with expectations for maser excitation in the presence of an electron density variation of approximately @xmath1 .
in addition , the h30@xmath0 and h26@xmath0 line masers arise from different parts in the wind .
this difference is also expected from the maser theory .
the wind component of both masers exhibits line - of - sight velocities that closely follow a keplerian law .
this result provides strong evidence that the disk wind extracts significant angular momentum and thereby facilitating mass accretion in the young star . |
the determination of the elements of the cabibbo - kobayashi - maskawa ( ckm ) matrix is one of the most important issues of quark flavor physics .
the precise determination of @xmath15 and @xmath16 elements has principal meaning , since they are solely responsible for the origin of cp violation in the ckm version of cp - violation within the standard model ( sm ) .
furthermore , the accurate knowledge of these matrix elements can be useful in relating them to the fermion masses and also in searches for hints of new physics beyond the sm .
therefore , strategies for the accurate determination of @xmath15 and @xmath16 are urgently required . in the existing literature
, we can find proposals of different methods for precise determination of @xmath16 and @xmath15 from inclusive and exclusive , semileptonic and non leptonic decays of @xmath17 meson ( see @xcite for a recent review ) .
the quantity @xmath18 has been historically measured by looking at the endpoint of the inclusive lepton spectrum in semileptonic @xmath17 decays , or from the exclusive semileptonic decays @xmath19 .
it has been suggested that the measurements of hadronic invariant mass spectrum @xcite as well as hadronic energy spectrum @xcite in the inclusive @xmath20 decays can be useful in extracting @xmath21 with better theoretical understanding .
the measurement of the ratio @xmath22 from the differential decay widths of the processes @xmath23 and @xmath24 by using @xmath25-flavor symmetry and heavy quark symmetry has also been proposed @xcite . there has also been recent theoretical progress on the exclusive @xmath26 semileptonic decay form factors using hqet - based scaling laws to extrapolate the form factors from semileptonic @xmath27 meson decays @xcite .
the element @xmath15 can be extracted indirectly from @xmath28 mixing . however , in @xmath28 mixing the large uncertainty of hadronic matrix elements prevents one from extracting @xmath15 with good accuracy .
a better extraction of @xmath29 can be made if @xmath30 mixing is measured as well , since the ratio @xmath31 can be determined much better .
another method to determine @xmath32 comes from the analysis of the invariant dilepton mass distributions of @xmath33 decays @xcite .
an interesting strategy for measuring @xmath34 was proposed in @xcite , which uses isospin symmetry to relate the decay @xmath35 to the well measured decay @xmath36 . in this work
we propose a new method to determine the ratio @xmath37 from an analysis of exclusive @xmath38 decays , where @xmath39 means pseudoscalar @xmath40 and vector @xmath41 mesons .
the inclusive @xmath42 decay is theoretically very clean because of the absence of any long distance effects and very small qcd corrections @xmath43 @xcite , and is therefore practically free from the scale ( @xmath44 ) dependence .
however , in spite of such theoretical advantages , it would be very difficult to detect this inclusive decay in experiments because the final state contains two missing neutrinos and ( many ) hadrons .
this paper is organized as follows . in section 2
we give the necessary theoretical framework to describe @xmath45 decays . in section 3
we study the ratios of branching fractions @xmath46 we also study the @xmath47 dependence of the differential decay rate of @xmath48 , and the ratio of the differential decay rates @xmath49 as well as @xmath50 section 4 is devoted to a discussion of our results and conclusion .
in the standard model ( sm ) , the process @xmath45 is described at quark level by the @xmath52 transition , and receives contributions from @xmath53-penguin and box diagrams , where dominant contributions come from intermediate top quarks .
the effective hamiltonian responsible for @xmath52 decays is described by only one wilson coefficient , namely @xmath54 , and its explicit form is @xmath55 where @xmath56 is the fermi constant , @xmath57 is the fine structure constant ( at the @xmath53 mass scale ) , and @xmath58 are elements of the ckm matrix . in eq . ( 1 ) , the wilson coefficient @xmath54 has the following form , including @xmath59 corrections : @xmath60 where @xmath61 in eq .
( 3 ) , @xmath62\ ] ] is the inami - lim function @xcite , and @xmath63 where @xmath64 is the spence function , and @xmath65 here @xmath44 describes the scale dependence when leading qcd corrections are taken into account .
the term @xmath66 is calculated in ref .
the presence of only one operator in the effective hamiltonian makes the process @xmath67 very attractive , because the estimated theoretical uncertainty is related only to the value of the wilson coefficient @xmath54 ( @xmath68 the uncertainty due to the top quark mass ) , contrary to the @xmath69 decay , where the uncertainties are described by three independent wilson coefficients , @xmath70 , @xmath71 and @xmath72 .
another favorable property of this decay is the absence of any long distance effects , which make the @xmath73 process considerably more complicated . in spite of such theoretical advantages , in practice the inclusive channel
@xmath74 would be very difficult to detect in experiments . only exclusive channels , namely @xmath75 ,
may be studied experimentally . at this point
we consider the problem of computing the matrix elements of the effective hamiltonian ( 1 ) between @xmath17 and @xmath39 states .
this problem is related to the non - perturbative sector of qcd , and it can be solved only by using non - perturbative methods .
the matrix element @xmath76 has been investigated through different approaches , such as chiral perturbation theory @xcite , three point qcd sum rules @xcite , relativistic quark model by the light front formalism @xcite , effective heavy quark theory @xcite , light - cone qcd sum rules @xcite-@xcite , @xmath77 .
the hadronic matrix elements for @xmath78 ( @xmath79 is a pseudoscalar meson , @xmath80 or @xmath81 ) decays can be parametrized in terms of the form - factors @xmath82 and @xmath83 in the following way ; @xmath84 where @xmath85 and @xmath86 . for @xmath87 (
@xmath88 is the vector @xmath89 or @xmath90 meson ) decays , the hadronic matrix element can be written in terms of five form - factors : @xmath91 \label{(5)}\end{aligned}\ ] ] with condition @xmath92 note that after using the equations of motion the form - factor @xmath93 can be written as a linear combination of the form - factors @xmath94 and @xmath95 ( for more details see the first reference in @xcite ) : @xmath96 . \label{(7)}\ ] ] in eq .
( 5 ) , @xmath97 and @xmath98 are the polarization 4vector , 4momentum and mass of the vector particle , respectively . using eqs . ( 1 ) , ( 4 ) and ( 5 ) , and after performing summation over vector meson polarization and taking into account the number of light neutrinos @xmath99 , we have : @xmath100 and @xmath101 \bigg ) .. \nonumber \end{aligned}\ ] ] in eqs .
( 8) and ( 9 ) , @xmath102 is the usual triangle function @xmath103 similarly , calculations for the @xmath104 decay lead to the following results : @xmath105 and @xmath106 \bigg ) .
\label{(10 ) } \end{aligned}\ ] ]
in deriving eqs.(8)(11 ) , we set the masses of @xmath107 and @xmath108 equal and the electron mass is neglected . using isospin symmetry
the branching ratio for @xmath109 can be related to that for @xmath110 .
it is clear that their ratio is independent of form - factors , @xmath68 free of hadronic long distance uncertainties in the limit @xmath111 .
corrections to the strict isospin symmetry , which come from phase space factors due to the difference of masses of @xmath112 and @xmath113 , isospin violation in the @xmath114 form - factors and electromagnetic radiative corrections to the @xmath115 transition , are all small . in the following discussions we shall neglect these small isospin violation effects . also note that these corrections for @xmath116 transition have been calculated in @xcite and found to be small , @xmath117 now we relate the branching ratio @xmath118 with @xmath119 . from eqs .
( 9 ) and ( 11 ) , we have @xmath120 here the numerical factor 6 comes from the number of light neutrinos , and isospin symmetry relation between the form - factors of @xmath121 and @xmath122 . in eq .
( 12 ) , we also put @xmath123 . from eq .
( 12 ) , we get @xmath124 where @xmath125 the second relation in ( 13 ) holds only for the ckm version of cp - violation within the sm . from eq . ( 13 ) , we can see that measurements of the ratio of the branching fractions allow to determine the ratio of @xmath126 and @xmath127 .
up to now and @xmath126 from @xmath128 decays . ] , various methods for measuring each angle separately have been proposed , @xmath129 , the angle @xmath130 will be measured from @xmath131 decay with high accuracy , and angle @xmath132 is from the charged @xmath17 decay @xmath133 with larger uncertainty . as follows from eq .
( 13 ) , one can measure the angle @xmath132 with small theoretical uncertainty , if @xmath134 is measured independently with high accuracy .
the following relations will also be useful for extracting the phase angle @xmath132 precisely : @xmath135 @xmath136 @xmath137 and @xmath138 in derivation ( 14)-(17 ) , we assumed that the mass of charged and neutral final states mesons are equal .
now we consider the differential decay widths , @xmath139 .
for the hadronic form - factors we have used the results of the works @xcite-@xcite , @xmath68 the monopole type form - factors based on light cone qcd sum rules .
the values of the form - factors at @xmath140 are ( see also ref .
@xcite ) : @xmath141 note that all errors , which come from the uncertainties of the @xmath142 quark mass , the borel parameter variation , wave functions , non - inclusion of higher twists and radiative corrections , are added in quadrature . in fig . 1 , we present the differential decay width @xmath143 as a function of the normalized momentum transfer square , @xmath144 . in fig .
2 , we show the @xmath47 dependence of the ratio of the differential decay rates @xmath145 and @xmath48 , normalized to @xmath146 . in these figures ,
dotted and dash - dotted curves correspond to the cases when the uncertainty is added and subtracted from the central values of all form - factors , respectively . for the central solid curve we use the central values of form - factors .
we note that the errors in the differential decay width of fig . 1 due to the form - factors uncertainties are about @xmath147 . however , the errors in the ratio of fig .
2 are reduced to about @xmath148 .
we conclude that even though the errors from uncertainties of the form - factors for each channel are substantial , those in the corresponding ratio are comparatively small , and that for precise determination of the elements of the ckm matrix the investigation of the corresponding ratio is very suitable .
we also note that the uncertainties for our main results , eqs .
( 12)(16 ) , where we only assume flavor @xmath149 ( isospin ) , should be even much smaller than that shown in fig .
2 , since there we had to assume flavor @xmath25 symmetry . for completeness
we present the integrated value for the branching fractions of @xmath150 and @xmath151 as well as the value of the ratio @xmath152 and @xmath153 : @xmath154 the values of the main input parameters , which appear in the expressions for the decay widths are @xmath155 for the @xmath17 meson life time , we take @xmath156 @xcite .
we proposed a new method for the precise determination of @xmath3 from the ratios of the branching fractions @xmath157 as is well known , each partial decay width depends very strongly on hadronic form - factors .
however , as also shown in eqs .
( 9)(13 ) , these ratios , @xmath158 , are free of any hadronic uncertainties , if small isospin breaking effects are neglected .
measurements of @xmath159 allow to determine @xmath3 with little theoretical error , which equals @xmath160 for the ckm version of cp - violation within the standard model .
therefore , @xmath159 measures a relation between two different phases angles , which can be measured separately by experiments .
we also found that each exclusive channel @xmath161 has rather large theoretical uncertainties due to the unknown hadronic form - factors , as shown in fig .
1 . in order to reduce these uncertainties we have considered the ratio of the corresponding exclusive channels , @xmath162 @xmath163 , as shown in fig .
2 . a few words about experimental statistics for detecting the @xmath164 decay follow :
future symmetric and asymmetric @xmath17 factories should produce much more than @xmath165 @xmath166 mesons by the year 2010 . with @xmath167
@xmath17 mesons effectively reconstructed , the number of expected events for @xmath168 channel is @xmath169 and the statistically estimated error for @xmath170 decay is approximately @xmath171 we argue that within the next decade the decay channel @xmath168 has a good chance for being detected in future @xmath17 factories . note that the inclusive channels @xmath172 are also free of any theoretical uncertainties .
however , measuring inclusive channels in experiments would be very difficult because of the two missing neutrinos and ( many ) hadrons . for completeness , we give here the summarized results for the inclusive decays in the lowest order : @xmath173 @xmath174 and @xmath175 in derivation of eq .
( 20 ) we have neglected the charm quark mass . +
* acknowledgements * + one of the authors ( t.m.a . )
sincerely thanks mustafa savci for helpful discussions and for his assistance in numerical calculations .
we thank m. drees for careful reading of the manuscript and his valuable comments .
the work of csk was supported in part by the ctp of snu , in part by the bsri program bsri-97 - 2425 , in part by non - directed - research - fund of 1997 , in part by yonsei university faculty research fund of 1997 , and in part by the kosef - dfg , project no . 96 - 0702 - 01 - 01 - 2 .
v. barger , c.s .
kim and r.j.n .
phillips , phys . lett . *
b235 * , 187 ( 1990 ) ; @xmath176 * b251 * ( 1990 ) 629 ; c.s .
kim , d.s .
hwang , p. ko and w. namgung , nucl .
b ( proc . suppl . ) * 37a * ( 1994 ) 69 ; phys . rev .
* d50 * ( 1994 ) 5762 .
kim , hep - ph/9605201 ( phys . lett . * b * , in press ) ; c.s .
kim , hep - ph/9701316 , talk given at 4th kek topical conference on flavor physics , tsukuba , japan , ( oct 1996 ) ; a.f .
falk , z. ligeti and m.b .
wise , phys . lett .
* b406 * ( 1997 ) 225 ; i. bigi , r.d . dikeman and n. uraltsev , hep - ph/9706520 , presented at 3rd babar physics workshop , orsay , france ( jun 1997 ) ; p. ball , hep - ph/9709407 ,
talk given at 7th international symposium on heavy flavor physics , santa barbara , ca ( jul 1997 ) .
kim , t. morozumi and a.i .
sanda , hep - ph/9708299 ( phys .
d * , in press ) ; c.s .
kim , t. morozumi and a.i .
sanda , hep - ph/9706380 , talk given at 2nd conference on b physics and cp violation , honolulu , hi ( mar 1997 ) . | we propose a new method for precise determination of @xmath0 from the ratios of branching ratios @xmath1 and @xmath2 .
these ratios depend only on the ratio of the cabibbo - kobayashi - maskawa ( ckm ) elements @xmath0 with little theoretical uncertainty , when very small isospin breaking effects are neglected .
as is well known , @xmath3 equals to @xmath4 for the ckm version of cp - violation within the standard model .
we also give in detail analytical and numerical results on the differential decay width @xmath5 and the ratio of the differential rates @xmath6 as well as @xmath7 and @xmath8 .
-1.68 cm 6.5 in 9.5 in -0.3 cm -0.3 cm hep - ph/9710428 + snutp 97 - 124 + yums 97 - 027 + * measuring @xmath9 + through @xmath10 decays * + * t. m. aliev @xmath11 * and * c. s. kim @xmath12 * + @xmath13 physics department , girne american university , + girne , cyprus + @xmath14 physics department , yonsei university , seoul 120 - 749 , korea + ( ) |
the electroweak sector of the standard model ( sm ) predicts new physics at sub - tev scales to unitarize @xmath3-scattering . with a higgs boson
, the hierarchy problem suggests additional new physics near the tev scale to stabilize the electroweak scale , and the minimal supersymmetric standard model ( mssm ) is one of the leading candidates of such new physics . for reviews of the mssm , see , for examples , drees @xcite , martin @xcite , dine @xcite , and peskin @xcite . in the mssm
, the mass of the lightest @xmath1-even higgs boson is bounded at tree - level by @xmath2 , because the tree - level quartic couplings are parameterized by gauge couplings , and such a light higgs boson is ruled out by the cern lep searches of the sm higgs boson @xcite@xcite@xcite@xcite@xcite that impose @xmath4 at one - loop level , however , there can be large radiative corrections due to heavy scalar tops ( @xmath5 and @xmath6 , superpartners of the top - quark ) and/or a large coupling of the trilinear interaction @xmath7 @xcite@xcite@xcite@xcite@xcite@xcite@xcite@xcite@xcite@xcite .
while such radiative corrections can be large enough to satisfy the lep bounds , they also contribute to the quadratic term of the higgs potential , leading to the `` little hierarchy problem '' .
the mssm also suffers from a @xmath8-problem in that its lone dimensionful susy - invariant parameter , @xmath8 , is phenomenologically required to be of order 100 gev , while its natural scale can in principle be much larger .
the next - to - minimal supersymmetric standard model ( nmssm ) solves the @xmath8-problem and alleviates the little hierarchy problem @xcite by extending the mssm with a singlet chiral superfield @xmath9 . for reviews of the nmssm ,
see balazs et al .
@xcite and references therein .
the higgs couplings with @xmath9 lead to additional contributions to the quartic couplings in the higgs potential , while the @xmath8-term is dynamically generated from the vacuum expectation value ( vev ) of the scalar component of @xmath9 . with these additional contributions to the quartic couplings , the mass of the lightest @xmath1-even higgs boson may be larger than @xmath2 at tree - level , and the nmssm can satisfy the lep bounds on the higgs mass with lighter s - tops compared to the mssm @xcite@xcite@xcite@xcite@xcite@xcite@xcite . in this paper , we extend the mssm with a hypercharge - neutral , @xmath10-triplet chiral superfield @xmath11 and name the model triplet - extended supersymmetric standard model ( tessm ) .
extensions of this type have been studied extensively by espinosa and quiros @xcite@xcite , felix - beltran @xcite , setzer and spinner @xcite , and diaz - cruz et al .
while this model does not solve the @xmath8-problem , it is an interesting alternative to the nmssm , as an economical extension of the mssm , because it can also achieve a mass of the lightest @xmath1-even higgs boson that is larger than @xmath2 at tree level .
furthermore , compared to the mssm and the nmssm , we expect there to be more radiative corrections to the mass of the lightest @xmath1-even higgs boson due to the additional states in the triplet . to the extent that these triplet - induced radiative corrections are significant
, we may further alleviate the little hierarchy problem .
unfortunately , in both the nmssm and the tessm , the respective singlet - induced and triplet - induced radiative corrections are typically small when we demand perturbativity at the scale of grand unified theory ( gut ) near @xmath12 gev .
this is because perturbativity at the gut scale imposes the bound @xmath13 at the weak scale , where @xmath14 is respectively the higgs - singlet - higgs and the higgs - triplet - higgs coupling in the superpotential of the nmssm and tessm . in both models , while the tree - level mass of the lightest @xmath1-even higgs boson can be as large as 100 gev , the @xmath15 radiative corrections are not large enough to lift the higgs mass over the lep bounds .
on the other hand , in the tessm , when we have @xmath16 ( so that @xmath14 is comparable with the top yukawa coupling)at the weak scale , we find the tree - level mass of the lightest @xmath1-even higgs boson to be close to the lep bound and the @xmath15 radiative corrections alone can easily lift the higgs mass over the lep bound even with small susy - breaking in the triplet sector . as the small susy - breaking in the triplet sector
translate into small fine - tuning , we can solve the little hierarchy problem at the expense of giving up perturbativity at the gut scale . without demanding perturbativity at the gut scale
, we also expect the nmssm to be a solution to the little hierarchy problem , with the mass of the lightest @xmath1-even higgs boson that satisfies the lep bounds without significant contribution from the top s - top sector . however , as an alternative to the nmssm and a reasonably economical extension of the mssm , the tessm and its phenomenology are interesting in their own right .
for example , as we show in this paper , the mssm limit of the tessm is achieved with @xmath17 , where @xmath18 is the susy - invariant mass of the triplet , keeping @xmath14 fixed , whereas in the nmssm one requires @xmath19 to achieve the mssm limit . as another example , even though the sub - tev , electrically - neutral component of the triplet acquires a vev , we can still satisfy the precision electroweak constraints without the extreme fine - tuning noted in the triplet - extended sm @xcite .
moreover , there may be other considerations that motivate extending the mssm by a triplet instead of a singlet .
for example , in obtaining neutrino masses through the type - ii @xcite and type - iii seesaw mechanisms , the sm is commonly extended with higgs triplets . though the higgs triplets may have nonzero hypercharge , hypercharge - neutral triplets are often present when the models are supersymmetrized and embedded in a unified gauge group @xcite@xcite@xcite .
we organize our paper as follows . in sec .
[ sec : tessm ] , we lay out the superpotential and the lagrangian of the tessm , compare it to the nmssm , and discuss constraints on its parameter space from electroweak precision tests and the requirement of perturbativity at the gut scale . in sec .
[ sec : higgsmass ] , we numerically evaluate the mass of the lightest , @xmath1-even higgs boson to one - loop , and show that we can satisfy the lep2 bounds without the contributions from the top s - top sector when @xmath14 is comparable with the top yukawa coupling .
we also discuss the gluon - gluon fusion production and diphoton decay of the lightest , @xmath1-even higgs boson in sec .
[ sec : higgsmass ] .
our discussions of the gluon - gluon fusion production rely only on the existence of light s - tops and the minimal color sector of the mssm , and are therefore applicable to any extensions of the mssm that solves the little hierarchy problem without invoking additional colored states . in sec .
[ sec : finetuning ] , we estimate two sources of fine - tuning in this model , and find that we can achieve a small fine - tuning of about one part in 30 in the higgs sector . finally , we conclude with sec . [
sec : conclusion ] that summarizes our results .
we extend the mssm with a hypercharge - neutral , @xmath10-triplet @xmath20 with the superpotential @xmath21 where @xmath22 are the higgs doublets of the mssm , @xmath23 is a lagrange - multiplier determined from the potential , and @xmath24 is the mssm superpotential sans the @xmath8-term @xmath25 note that , since @xmath11 is a chiral superfield , its scalar component necessarily contains a _ complex _ @xmath10-triplet , whereas in non - susy extensions of the sm @xcite@xcite@xcite@xcite@xcite@xcite , we can extend the sm with a real @xmath10-triplet . in components
, we have the fields @xmath26 and the superpotential ( sans the sm yukawa couplings ) @xmath27 the factor of 2 in front of @xmath14 in eq . [ eq : model ] gives us a coefficient of unity for the term @xmath28 as with the case of the nmssm when @xmath29 is replaced by a singlet @xmath9 , and facilitates direct comparisons between tessm and nmssm .
we can achieve gauge coupling unification at @xmath30 by including additional chiral superfields with quantum numbers @xmath31 where the first and second entries inside the parenthesis denote , respectively , the representations under the color @xmath32 and weak @xmath33 gauge groups , and the subscripts denote the charge under hypercharge @xmath34 gauge group .
this added content can have both susy - invariant and susy - breaking masses sufficiently large ( say 2 tev s ) so that they decouple from the electroweak scale physics , while still allowing for gauge coupling unification .
the added matter content ( triplet plus those in eq .
[ eq : addedx ] ) does not constitute a complete multiplet of su(5 ) , but can form a complete multiplet of trinification group @xmath35 @xcite@xcite@xcite .
in addition to the mssm soft susy - breaking parameters , we also have soft terms involving @xmath11 @xmath36 for simplicity , we assume that all couplings and masses in the superpotential are real . the tree - level potential involving the @xmath37-neutral higgs doublets and triplet is then @xmath38 where @xmath39 compared with the higgs potential in the mssm , we have an enhancement in the quartic coupling of the form @xmath40 and this in principle allows for a tree - level mass eigenvalue larger than @xmath2 after electroweak symmetry breaking ( ewsb ) .
this is similar to the case in the nmssm , where such quartic couplings are also generated from a superpotential of the form @xmath41 as with the nmssm , where the requirement of perturbativity at the gut scale limits @xmath13 at the tev - scale , the tessm also has a bound @xmath42 at the tev - scale while still preserving perturbativity at the gut scale .
though the bounds are similar , the details of obtaining such bounds are different and may be important for further model - building where perturbativity at the gut scale is imposed .
we elaborate briefly on some key differences between the two models from the relevant renormalization group equations ( rges ) given by @xmath43 and note the following points : * in the nmssm , there are two possible landau poles in @xmath14 and @xmath44 .
the rge of @xmath44 is such that @xmath44 always increases when evolving to higher energies , and @xmath44 feeds into the evolution of @xmath14 . in the tessm
, there is no such contribution because tr@xmath45 , but there are now additional contributions to the @xmath46 coefficient in @xmath47 in the tessm . * in the tessm , @xmath47 has a larger coefficient for the negative contribution of the form @xmath48 because @xmath11 is charged under @xmath10 . as the coupling @xmath49 is non - asymptotically - free in the tessm ( also in the nmssm ) , this gives a stronger suppression at higher energies and could potentially delay the appearance of the @xmath14 landau pole .
* the coupling @xmath49 also flows to larger values in the tessm than in the nmssm because of the added matter content .
this again gives a suppression at higher energies and may delay the @xmath14 landau pole .
( this can be achieved in the nmssm with added matter content , for example , in the nmssm with gauge - mediated susy - breaking . )
thorough studies on the upper bounds of @xmath14 in tessm and its extensions would require examining fixed points from the rges , and we leave these investigations for the future . for our work , it suffices to note that perturbativity at the gut scale imposes @xmath42 at the weak scale , so that @xmath14 is of similar strength to the weak gauge coupling . as such ,
while the tree - level mass of the lightest @xmath1-even higgs boson can be 100 gev ( as we later show ) , we expect the @xmath15 radiative corrections to the mass of the lightest @xmath1-even higgs boson to be insufficient to lift the higgs mass above the lep bounds .
however , motivated by solving the hierarchy problem , we take the view point that the landau pole we encounter at a higher scale ( around @xmath50 gev ) is rescued by some other new physics and analyze the higgs spectra and the phenomenology for larger values of @xmath14 .
we take values of @xmath14 comparable to the top yukawa coupling , so that the tev scale physics is still perturbative . as @xmath14 is now near unity at the tev - scale in the tessm and
we expect there to be more @xmath15 radiative corrections to the mass of the lightest @xmath1-even higgs boson compared to the nmssm , it is worthwhile to investigate these radiative corrections in detail . in extensions of the nmssm , such as fat higgs models @xcite@xcite@xcite@xcite
, @xmath14 can achieve much larger values and give rise to a very large mass for the lightest @xmath1-even higgs boson .
the model - building techniques of fat higgs models can also be applied to the tessm , but in this work we focus on the tessm as a simple extension of the mssm and an alternative to the nmssm without imposing the constraint of perturbativity at the gut scale . in the nmssm , the @xmath8-term and the masses of the singlet(ino )
are related by @xmath51 and the mssm limit is @xmath52 while keeping @xmath8 fixed . keeping @xmath44 perturbative in the mssm limit
then gives @xmath19 .
as @xmath14 is the only coupling between the singlet and the mssm sector , the mssm limit of @xmath19 with fixed @xmath8 decouples the singlet . in the tessm , the mssm limit is achieved with @xmath53 , holding the values of all other masses and couplings at the weak scale fixed , and in particular we do not need @xmath19 to achieve the mssm limit .
the decoupling of the additional contribution to the higgs quartic coupling in eq .
[ eq : mainq ] is accomplished by the effective operator obtained by integrating out the heavy triplet fields when @xmath54 . setting @xmath55 for simplicity , the equation of motion for @xmath29 , among other terms , has contributions of the form @xmath56 and this induces a contribution in the effective lagrangian @xmath57 that cancels the @xmath58 contribution to the quartic in the higgs potential when @xmath59 . in terms of feynman diagrams ,
this effective operator arises from the diagrams such as the one shown in fig .
[ fig : effop ] with the amplitude ( in the limit of large @xmath59 ) @xmath60 where @xmath61 is the scale of external momenta of the higgs bosons . in the limit @xmath62 , this gives the canceling contribution to the higgs potential , and the resulting theory is the mssm .
contributions in the higgs potential when the triplet field , @xmath11 , decouples .
, width=192 ] when we do not explicitly integrate out the heavy triplet sector , the full mass matrices ( involving both higgs doublets and the triplet ) provide a seesaw - like mechanism in the limit of @xmath17 that seesaws away any @xmath14 dependence in the higgs doublets sector . we will demonstrate this in the next section when we compute the mass of the lightest @xmath1-even higgs boson .
as we are assuming real couplings and masses for simplicity , there is no mixing between the real and imaginary components of the complex scalar fields @xmath63 and @xmath29 and it is convenient to separate them into real and imaginary parts @xmath64 where we have also shifted the real components ( @xmath65 ) to the physical modes ( @xmath66 ) by the respective vacuum expectation values ( @xmath67 ) . prior to ewsb ,
all the vevs vanish and the real components of the higgses have the mass matrix ( in the basis @xmath68 ) @xmath69 the corresponding mass matrix for the imaginary components can be obtained from eq . [
eq : v0realmatrix ] by changing the signs of @xmath70 and @xmath71 in the ( 1,2 ) , ( 2,1 ) , and ( 3,3 ) elements . in the mssm , the conditions for successful ewsb breaking are that ( i ) the ( top - left 2@xmath722 ) mass matrix in eq .
[ eq : v0realmatrix ] has one negative eigenvalue , and ( ii ) the potential is bounded from below along the @xmath73-flat direction @xmath74 . in the tessm model , the first condition gives us the same condition as the mssm @xmath75 while the second condition is automatically satisfied by the presence of the quartic coupling @xmath76 .
however , the minimization conditions now demand @xmath77 where we have defined @xmath78 so that the gauge bosons receive masses of @xmath79 where @xmath49 and @xmath80 are respectively the gauge couplings of the @xmath10 and @xmath34 groups .
we have also abbreviated for convenience the trigonometric functions @xmath81 while the condition of successful ewsb in eq .
[ eq : ewconstrt ] gives a constraint on the parameters , electroweak precision tests offer a much more stringent constraint .
the induced vev @xmath82 contributes to the oblique parameter @xmath83 because it contributes to the mass of the charged gauge bosons @xmath84 , but not to that of the neutral gauge boson @xmath85 .
we find the oblique contribution to be @xmath86 the oblique correction due to the triplet vanishes in the limit of @xmath53 holding all other parameters fixed , as expected .
however , even if @xmath18 is of the same order of @xmath8 and @xmath87 , @xmath88 can be small due to a partial cancellation between @xmath8 and @xmath89 .
we impose the constraint that @xmath90 , which in turn translates to an upper bound on @xmath82 @xmath91 and provides the main constraint on the parameters @xmath14 and @xmath18 . in fig .
[ fig : tplot1 ] , we plot the allowed regions on @xmath92 plane with @xmath93 , for various values of @xmath94 and @xmath8 . for small values of @xmath94 and @xmath8 , @xmath95 is only viable with either small @xmath14 or a cancellation in the numerator of eq .
[ eq : obliquetformula ] . in fig .
[ fig : tplot2 ] , we plot the allowed region in @xmath96 plane for @xmath97 , @xmath93 , for various values of @xmath18 .
as expected , for larger values of @xmath18 , there is a thicker band on the @xmath96 plane that is allowed .
we will quantify the degree of fine - tuning in the cancellation for allowed @xmath95 in sec .
[ sec : finetuning ] . for
now , we may estimate the fine - tuning along the ideas of athron et al .
for example , with the parameters that require a fine - tuning in @xmath18 @xmath98 we have viable @xmath88 in the regions @xmath99 for @xmath18 below 3.0 tev , we would typically have unacceptably large @xmath88 that violates precision electroweak constraints except in the a small region of @xmath18 between 250 gev and 375 gev because of cancellations in the numerator of eq .
[ eq : obliquetformula ] .
if we sample @xmath18 at random in the range between 0 and 3.0 tev , the only region with viable @xmath88 is only 125(=375 - 250 ) gev wide , and we can thus estimate the fine - tuning as @xmath100 so a cancellation of 1 part in 24 is required to have small @xmath95 for the parameters in eq .
[ eq : texample ] .
( in gray ) on @xmath101 plane for various values of @xmath94 and @xmath8 as indicated in each plot . , width=480 ] ( in gray ) on @xmath102 plane for various values of @xmath18 as indicated in each plot . , width=480 ]
after ewsb , the neutralino ( @xmath103 ) and chargino ( @xmath104 ) mass matrices are now extended with the triplet sector .
the mass matrix for the neutralinos in the basis ( @xmath105 , @xmath106 , @xmath107 @xmath108 , @xmath109 ) is given by @xmath110 where @xmath111 and @xmath112 are respectively the susy - breaking bino and wino masses , and @xmath113 is as defined in eq .
[ eq : mueff ] . for the charginos , in the basis @xmath114
, the chargino mass matrix appears in the lagrangian as @xmath115 where @xmath116
the lightest @xmath1-even higgs boson in tessm is a linear combination of the @xmath1-even components of the higgs doublets @xmath22 and the neutral component of the triplet @xmath29 .
after ewsb , the squared - mass matrix for the neutral scalar has the entries @xmath117 where @xmath82 should be considered as a function of the input parameters via the minimization condition in eq .
[ eq : vtcond ] and we define @xmath118 as in the case of the mssm @xmath119 as in the case of the nmssm , the lightest mass - squared eigenvalue is bounded by the lightest eigenvalue of top - left 2@xmath722 block of the mass matrix , @xmath120 in fig .
[ fig : treeplot ] , we plot this tree - level upper bound as a function of @xmath94 for @xmath121 , 0.8 , and 0.9 . for @xmath97 , it is possible to obtain a tree - level higgs mass larger than 100 gev for @xmath122 , and even satisfy the lep2 bounds at tree - level for small @xmath123 .
-even higgs boson as given by eq .
[ eq : treemassbound ] as a function of @xmath94 for @xmath121 ( lowest line ) , 0.8 ( middle line ) , and 0.9 ( top line ) , width=408 ] we can also see the mssm limit in the mass matrix when the triplet decouples with fixed @xmath14 . in the limit @xmath53 , keeping all other parameters fixed , we have @xmath124 and the mass matrix has the form @xmath125 where @xmath126 is the 2@xmath722 mssm mass matrix for the @xmath1-even higgs bosons , and @xmath127 integrating out the third row and column of the mass matrix in eq .
[ eq : ma - approx ] , the effective top - left 2@xmath722 sub - matrix becomes @xmath128 and we recover the mssm limit as @xmath53 . in fig .
[ fig : deplot ] , we show this decoupling behavior by plotting the tree - level mass of the lightest higgs boson as a function of @xmath18 for various values of @xmath94 , and see that , for @xmath129 gev , we recover the mssm results .
-even higgs boson as a function @xmath18 for @xmath130 ( left ) and @xmath97 ( right ) .
the other parameters are kept fixed as @xmath131 , @xmath132 gev , and @xmath133 gev .
the three curves have values of @xmath94 of 10 ( solid ) , 5 ( dashed ) , and 3 ( dotted ) . for each case , we see decoupling in large @xmath18 , and the limiting value agrees with the mssm result .
, title="fig : " ] -even higgs boson as a function @xmath18 for @xmath130 ( left ) and @xmath97 ( right ) .
the other parameters are kept fixed as @xmath131 , @xmath132 gev , and @xmath133 gev .
the three curves have values of @xmath94 of 10 ( solid ) , 5 ( dashed ) , and 3 ( dotted ) . for each case , we see decoupling in large @xmath18 , and the limiting value agrees with the mssm result .
, title="fig : " ] in this subsection , we numerically evaluate the mass of the lightest @xmath1-even higgs boson at tree - level . with the minimization conditions
, we can take as input parameters @xmath134 and fix @xmath135 , @xmath136 and @xmath82 by solving the minimization conditions with the experimental inputs of @xmath137 gev and the gauge couplings @xmath138 , @xmath139 ( this fixes @xmath140 gev ) .
we discard sets of input parameters that give large @xmath82 inconsistent with electroweak constraints . for all our numerical studies ,
we analyze two the cases of @xmath130 and @xmath97 , and scan the parameter space in the range @xmath141 the range of @xmath94 is chosen so that the bottom yukawa coupling is smaller than the top yukawa coupling , and we can neglect the bottom yukawa coupling when we study the production and decay properties of the lightest @xmath1-even higgs boson . since solutions to the minimization conditions only guarantee an extremum , we only keep solutions that give a local minimum by checking that all scalar masses are positive at the desired vev .
we also discard any points that give unacceptably large @xmath142 or contain charged scalar particles lighter than 100 gev . for the range of parameters listed in eq .
[ eq : scannedrange ] , we show the mass of the lightest @xmath1-even boson as a function of @xmath94 in fig . [
fig : ntreeplot ] . in fig .
[ fig : ntreeplot ] , we also plot the upper bound of the tree - level higgs mass given in eq .
[ eq : treemassbound ] .
the plots show that we can indeed achieve large ( greater than @xmath2 ) tree - level higgs mass with large @xmath14 , and we can even satisfy lep2 bounds at tree - level for small values of @xmath94 ( @xmath123 ) when @xmath97 .
-even higgs boson as a function @xmath94 scanned over the parameter space as listed in eq .
[ eq : scannedrange ] .
the plot on the left ( right ) has @xmath130 ( @xmath97 ) , and the top curve indicates the upper bound as computed from eq .
[ eq : treemassbound ] .
, title="fig : " ] -even higgs boson as a function @xmath94 scanned over the parameter space as listed in eq .
[ eq : scannedrange ] . the plot on the left ( right ) has @xmath130 ( @xmath97 ) , and the top curve indicates the upper bound as computed from eq .
[ eq : treemassbound ] .
, title="fig : " ] since the lightest @xmath1-even higgs boson is a linear combination of @xmath66 for @xmath143 , we will construct the coleman - weinberg ( cw ) potential @xcite only for the fields @xmath65 , and extract the corrections to @xmath144 from the cw potential .
furthermore , we will make the two following assumptions : * we assume that both the s - top masses are close to the top - quark mass , and the famous @xmath145 contributions in the mssm are small .
in other words , we only consider the corrections from the higgs boson , neutralino , and chargino sectors .
these contributions are dominated by the coupling @xmath14 and the susy - breaking parameters in the triplet sector .
our results will show that these contributions are sufficient to satisfy the lep2 bounds on the higgs mass , and we do not need large contributions from the top s - top sector as in the case of the mssm . * in the neutralino and chargino mass matrices , we ignore the mixing induced by gauge interactions .
this removes dependencies on the bino and wino susy - breaking masses in our analysis as we do not include the bino and wino states , and we expect their contributions to be small when @xmath146 .
( if we include the bino and wino states , we would also have to include the corresponding superpartners in the @xmath147 and @xmath85 gauge bosons . ) the coleman - weinberg potential is given by @xmath148,\end{aligned}\ ] ] where @xmath149 are field - dependent mass matrices in which the fields are not replaced with their vev s , @xmath150 is the renormalization scale , and the supertrace includes a factor of @xmath151 so that fermions contribute oppositely to bosons , and the spin degrees of freedoms are appropriately summed over .
since here we are only interested in the cw potential of the fields @xmath65 that always appear in the combination @xmath152 , the field - dependent matrices for the charginos and neutralinos are simply those in eqs .
[ eq : nmatrix ] and [ eq : cmatrix ] with the vevs @xmath67 replaced by the corresponding fields @xmath65 . for the scalars ,
the naive replace - vev - by - field method fails and we need to distinguish between the contributions from the minimization conditions and those from the replacement of the fields with their corresponding vevs .
for example , while the ( 11)-element of the mass matrix of the @xmath1-even neutral higgs boson is @xmath153 it is incorrect to have the field - dependence @xmath154 because some of the vev - dependence in eq .
[ eq : cwex01 ] comes from the minimization conditions eq .
[ eq : min - mhu ] . the correct field - dependent ( 11)-element of the @xmath1-even
neutral higgs boson is @xmath155 where @xmath156 is related to the vev s ( but not the fields ) through the minimization condition in eq .
[ eq : min - mhu ] . in appendix
[ app : field - dep - masses ] , we give the field - dependent mass matrices used in the calculation of the coleman - weinberg potential .
since the analytical results for the mass eigenvalues of the field - dependent matrices are complicated , we will compute the higgs mass numerically .
the one - loop mass matrix can be extracted from the coleman - weinberg potential by numerically evaluating the derivatives of the mass eigenvalues with respect to the fields about the vevs @xcite ( dropping the pre - factor from the supertrace for convenience ) @xmath157 where the second term in eq .
[ eq : cwex02 ] takes into account the shift in the minimization conditions , and @xmath158 is the set of mass eigenvalues that enter the coleman - weinberg potential .
our set of @xmath158 includes the eigenvalues of the mass matrices of the @xmath1-even , @xmath1-odd , and charged higgs bosons , as well as the neutralinos and charginos mass matrices .
these field - dependent mass matrices are given in appendix [ app : field - dep - masses ] .
-even higgs boson , including one - loop contributions from the triplet sector , as a function @xmath94 , scanned over the parameter space as listed in eq .
[ eq : scannedrange ] .
the plot on the left has @xmath130 , and the plot on the right has @xmath97 .
the input parameters of the individual points are the same as those that give rise to the points shown in the corresponding plot of fig .
[ fig : ntreeplot ] . , title="fig : " ] -even higgs boson , including one - loop contributions from the triplet sector , as a function @xmath94 , scanned over the parameter space as listed in eq .
[ eq : scannedrange ] .
the plot on the left has @xmath130 , and the plot on the right has @xmath97 .
the input parameters of the individual points are the same as those that give rise to the points shown in the corresponding plot of fig .
[ fig : ntreeplot ] . , title="fig : " ] -even higgs boson with and without the one - loop contribution from the triplet sector for the points shown in the corresponding plots of figs .
[ fig : nloopplot ] and [ fig : ntreeplot ] .
the plot on the left has @xmath130 , and the plot on the right has @xmath97 . , title="fig : " ] -even higgs boson with and without the one - loop contribution from the triplet sector for the points shown in the corresponding plots of figs .
[ fig : nloopplot ] and [ fig : ntreeplot ] .
the plot on the left has @xmath130 , and the plot on the right has @xmath97 .
, title="fig : " ] we numerically compute the mass of the lightest @xmath1-even higgs boson to one - loop using the coleman - weinberg potential for the parameter space in eq .
[ eq : scannedrange ] . for the same input parameters that give rise to the tree - level results shown in fig .
[ fig : ntreeplot ] , we show the corresponding higgs mass computed to one loop in fig .
[ fig : nloopplot ] , and the difference between the loop - level and tree - level masses in fig .
[ fig : ndiffplot ] .
we use the value of @xmath18 as the renormalization scale @xmath150 that enters the coleman - weinberg potential . from these plots
, we see that the triplet sector can give a large contribution to the mass of the lightest @xmath1-even higgs boson , and we can satisfy the lep bounds without the s - top contributions for all values of @xmath94 in our scanned range .
.sample higgs spectra .
all dimensionful parameters are in units of gev , except for @xmath159 and @xmath70 , which are in units of ( gev)@xmath160 .
the definitions of fine - tuning @xmath161 , @xmath162 , and @xmath163 are given in , respectively , eqs .
[ eq : higgsfinetune ] , [ eq : tripletfinetune ] , and [ eq : tripletfinetune2 ] .
the value of @xmath161 indicates the percent change in @xmath164 induced by a 1% change in @xmath156 at a fundamental scale of susy - breaking , and the value of @xmath162 indicates the percent change in @xmath88 induced by a 1% change in @xmath18 .
the measure @xmath165 is only applicable to points 1 and 2 , and shows that there is a cancellation of one part in 23 to give a viable value of @xmath88 .
points 1 and 2 show examples of input parameters that give a viable higgs mass with a small fine - tuning in the electroweak sector .
points 3 and 4 differ only in @xmath14 and are samples that give large higgs masses of about 120 gev ( for @xmath166 ) and about 135 gev ( for @xmath167 ) .
points 5 and 6 have large @xmath168 and @xmath169 at tree - level , but there are large radiative corrections to have viable higgs masses at one - loop . [ cols="^,^,^,^,^,^,^",options="header " , ] in table [ tb : samplepoints ] , we give some sample points in our scan .
points 1 and 2 are sample points that have small fine - tuning ( as will be defined later in eq .
[ eq : higgsfinetune ] ) .
the tessm can achieve small fine - tuning because the higgs mass can be large at tree - level and does not require large contributions from radiative corrections .
points 3 and 4 are samples of the points with the largest higgs masses ( and therefore fine - tuning ) in our scanned range of parameter space , as evident in the values of susy - breaking parameters @xmath159 , @xmath87 , and @xmath70 being near the boundary of the scanned range .
points 5 and 6 are samples of points having a large @xmath168 , where the tree - level higgs mass is only slightly larger than @xmath2 , and there is a significant one - loop contribution , and , correspondingly , large fine - tuning . with large @xmath14 in both the tessm and the nmssm
, we do not require heavy s - tops .
in these cases , the gluon - gluon fusion production of the lightest @xmath1-even higgs boson , @xmath170 , and its diphoton partial decay width , @xmath171 , can be very different from the mssm because these processes involve s - top loops . in this section
we perform a simplified analysis showing that in the tessm there may be a gluo - philic higgs boson whose gluon - gluon fusion production cross section can be larger than that of the sm by a factor of 1.8 . as stated in the introduction ,
our discussions of the gluon - gluon fusion production rely only on the existence of light s - tops and the minimal color sector of the mssm , and are therefore applicable to any extensions of the mssm that solves the little hierarchy problem without invoking additional colored states . for the diphoton partial decay width
, there are several sources of suppression , and we may have a partial decay width that is about 0.8 times that in the sm . of course , at the lhc the relevant quantity is the product of the gluon - gluon fusion production cross section and the diphoton branching ratio @xmath172 and a more complete analysis would have to take into account the effects of light s - tops to all the decay channels as well as the large , higher - loop corrections from qcd and large couplings .
we leave the complete analysis of the higgs production and decay for future work .
the well - known formula for the decay width of a real scalar decaying into two photons can be found in gunion et al .
this formula is also presented in the appendix [ app : di - photon - width ] . in the sm , ignoring all the yukawa couplings except for the top yukawa coupling , the process @xmath173 proceeds only through a top - quark loop . in the mssm
, we have additional contributions from the s - tops ( see fig .
[ fig : feyn - digluon ] ) , as well as all the other s - quarks through @xmath73-term interactions of the form @xmath174 with coupling of the order @xmath2 . to simplify our analysis
, we will ignore the @xmath73-term interactions except those in the s - top sector , but we note that these contributions can be important when there are light s - quarks and must be taken into account in a full analysis . in the tessm .
, width=408 ] in the mssm with small s - top mixing , the s - top contributions interfere constructively with the top - quark contribution for the gluon - gluon fusion production cross section @xcite@xcite .
however , with small s - top mixing , the s - tops need to be heavy to satisfy the lep bounds on the higgs mass , and the s - top contributions decouple .
( with large - stop mixing , it is possible to have s - top contributions that interfere destructively with the top - quark contribution , leading to a gluo - phobic higgs boson . ) in the tessm and nmssm , we can have light s - tops at the expense of perturbativity at the gut scale , and a large enhancement in the production rate . assuming large @xmath94
so that @xmath175 and there is no s - top mixing , and approximating the lightest @xmath1-even higgs boson @xmath176 as being dominantly composed of @xmath177 ( the @xmath1-even component of @xmath178 ) , we have the interactions @xmath179 where @xmath180 is the top - quark , and @xmath5 ( @xmath6 ) is the superpartner to the left-(right-)handed component of the top - quark . from eqs .
[ eq : di - photon - formula ] , [ eq : diphoton - to - digluon ] , and [ eq : sigma - gamma - conv ] , the ratio of the amplitudes @xmath181 due to the s - top s - quarks and top - quark is then @xmath182 where @xmath183 and @xmath184 are susy - breaking soft masses of the corresponding s - tops , and we have used the relationships @xmath185 . in fig .
[ fig : r - h - to - gg ] , we plot @xmath186 as a function of a common soft s - top mass @xmath187 , assuming @xmath188 gev .
since the s - top mass eigenvalues in this simplified analysis are given by @xmath189 and the current searches limit the s - tops masses to be greater than 120 gev @xcite , we can have @xmath190 ( so @xmath191 ) and @xmath186 can be as large as 0.48 .
this gives a gluo - philic higgs boson whose production cross section via gluon - gluon fusion may be enhanced relative to the sm prediction by a factor of @xmath192 . through s - top loops
normalized with respect to the amplitude through top - quark loop , as a function of a common s - top soft mass @xmath193 , assuming no mixing in the s - top sector .
, width=408 ] imposing perturbativity at the gut scale , we can have a milder gluo - philic higgs boson when one of the s - top is light ( the other is required to be heavy to have a higgs mass satisfying the lep2 bounds ) .
however , when only one s - top is light , the enhancement in the gluon - gluon fusion production cross section is only a factor of @xmath194 larger than that of the sm . in the sm
, the diphoton decay of the higgs boson proceeds through @xmath147-boson loop in addition to top - quark loop , and the contribution from the top - quark destructively interferes with the dominant @xmath147-boson contribution . in the mssm
, we have additional contributions from the s - tops and charginos ( the corresponding superpartners of the top - quark and @xmath147-boson ) , and , as in the case of @xmath195 , contributions from all the electrically charged s - quarks and s - leptons through @xmath73-term interactions . in the tessm .
the top three diagrams are present in the mssm , and the bottom three diagrams involve the coupling @xmath14 .
, width=408 ] in the tessm , we have the additional contributions from the states composed dominantly of the charged triplets , and also new contributions from the mssm matter content induced by @xmath14 ( see fig . [
fig : feyn - diphoton ] ) .
these contributions may be important when @xmath14 is as large as the top yukawa coupling , and so in this subsection we use @xmath97 for our numerical studies . in this work
, we will simplify our analysis by ignoring contributions from the @xmath73-term interactions except those involving the s - tops , and , using the same approximations as in the previous subsection of large @xmath94 and @xmath196 , we have the interactions and fermion masses @xmath197 where @xmath198 and @xmath199 are dirac spinors formed from the higgsinos and the fermionic components of the charged triplet states @xmath200 and @xmath201 are the projection operators @xmath202 although none of the charged states in eq .
[ eq : l - diphoton ] is a mass eigenstate , we approximate the scalar states as mass eigenstates with masses @xmath203 so that the contributions of these states to the amplitude @xmath204 have the same form . in the fermionic sector ,
the contribution to the amplitude @xmath204 comes exclusively from the mixing between @xmath198 and @xmath199 .
we can diagonalize the fermionic mass matrix with two unitary transformations @xmath205 where @xmath206 and @xmath207 are respectively parameterized by @xmath208 and @xmath209 , @xmath210 where @xmath211 and @xmath212 , and @xmath213 and @xmath214 are similarly defined .
these mixing angles are given by @xmath215 in terms of the mass eigenstates and mixing angles , the chargino interactions in eq .
[ eq : l - diphoton ] take the form @xmath216 in figs .
[ fig : diphoton1 ] and [ fig : diphoton2 ] , we illustrate the contributions of light s - tops and the additional charged states to the diphoton partial decay width , normalized with respect to the dominant @xmath147-boson contribution , assuming @xmath188 gev . in fig . [
fig : diphoton1 ] , we show the contributions from the top - quark ( constant line ) , s - tops ( solid line ) , and the charged scalar states ( dotted line ) . for the s - tops ( charged scalar states @xmath217 and @xmath218 ) , the horizontal axis should be interpreted as a common soft susy - breaking mass @xmath219 ( mass of these charged scalar states ) . in fig . [ fig : diphoton2 ] , we show the sum of the fermion contributions as a function of @xmath18 for different values of @xmath8 , and see that even for small values of @xmath8 and @xmath18 ( @xmath220 gev ) , these contributions tend to be small .
we can partially attribute the smallness to a cancellation between the contributions from the two states @xmath221 and @xmath222 , as evident in the relative sign difference between the coefficients of the @xmath223 and @xmath224 interactions .
though these fermionic contributions are small , it is interesting to note that , while the top - quark contribution interferes destructively with the @xmath147-boson contribution , the sum of these fermionic contributions interferes constructively . in any case ,
the additional @xmath14-induced contributions ( both bosonic and fermionic ) to the partial decay width @xmath225 are small compared to the s - top contributions . through scalar s - tops , @xmath217 , and @xmath218 loops , compared to the dominant @xmath147-boson loop contribution , as a function , respectively , of a common soft mass for the s - top ( solid curve ) , and of the mass of the states @xmath217 and @xmath218 ( dashed line ) .
we use @xmath97 .
the constant , solid line denotes the top - quark contribution .
, width=408 ] given by @xmath221 and @xmath222 loops normalized to the dominant @xmath147-boson loop amplitude , as a function of @xmath18 for @xmath226 gev ( solid ) , 200 gev ( dashed ) , and 300 gev ( dot - dashed ) .
note that these fermionic contributions are small compared to the s - top contributions shown in fig .
[ fig : diphoton1 ] . , width=408 ] combining all these contributions to @xmath227 , the diphoton partial width can be significantly reduced ( mostly from the s - top contributions ) .
for example , with @xmath228 , @xmath226 gev , @xmath229 gev , and @xmath230 gev , the amplitude @xmath204 is decreased by ( relative to the sm ) a factor of @xmath231 and the diphoton decay partial width is decreased , relative to the sm partial decay width , by a factor of @xmath232 .
we therefore can have a photo - phobic higgs boson in the tessm from the contribution of light s - tops .
before discussing the fine - tuning in the electroweak sector of the tessm , we briefly review the little hierarchy problem in the mssm . in the mssm with large @xmath94 ,
the higgs doublet @xmath178 is responsible for most of the ewsb since @xmath233 , and it has the potential @xmath234 minimizing the potential then gives @xmath235 so that @xmath236 under radiative corrections , @xmath135 receives large logarithmic corrections from the s - top sector , and we can use the renormalization group equations to infer the value of @xmath135 at a fundamental scale @xmath237 , @xmath238 where @xmath183 and @xmath184 are the susy - breaking s - top masses , @xmath239 is the coupling of the trilinear interaction @xmath7 , and @xmath237 can be taken as the scale of susy - breaking .
the large radiative correction leads to fine - tuning @xmath240 because the electroweak scale @xmath241 depends sensitively on the value of @xmath156 at the fundamental scale of susy - breaking @xmath237 .
we can quantify this fine - tuning as @xcite @xmath242 as a reference of comparison , for @xmath243 tev , and @xmath244 tev , we have @xmath245 so that the higgs sector needs to be fine - tuned to one part in 80 . thus , even though the electroweak scale is no longer quadratically sensitive to the fundamental scale @xmath237 with softly - broken susy , it is quadratic sensitive to the s - top masses and trilinear coupling @xmath246 , which are required to be large to have a higgs mass that satisfies the lep bounds , and this leads to a fine - tuning in the higgs sector of about one part in 100 .
this is the little - hierarchy problem in the mssm .
we can also define other measures of fine - tuning when given a more fundamental theory ( for example , an organizing principle of the soft susy - breaking parameters ) @xcite@xcite@xcite .
however , in this work we are mainly interested in the low - energy phenomenology of the tessm without appealing to a particular fundamental theory , and we will simply define fine - tuning as in eq .
[ eq : higgsfinetune0 ] . in the tessm with @xmath14 comparable to the top yukawa coupling
, we do not need heavy s - top masses nor significant mixing in the s - top sector for the higgs mass to satisfy the lep bound , and as such there is little or no fine - tuning from the s - top sector . on the other hand
, @xmath135 now receives radiative corrections from the triplet sector as well as the s - top sector @xmath247 and we can follow the same steps and reasoning as before to have an estimate of the fine - tuning due to the triplet sector @xmath161 @xmath248 so that @xmath249 , for example , would mean a tuning in @xmath250 to one part in 40 .
the value of @xmath161 indicates the percent change in @xmath164 per a one - percent change in @xmath156 at a fundamental scale of susy - breaking , generally , with large @xmath14 , for a given mass of the lightest @xmath1-even higgs boson , the fine - tuning in @xmath135 is less in the tessm than the mssm . in fig .
[ fig : ntuneplot ] , we plot @xmath161 for the data points shown in fig .
[ fig : nloopplot ] , where we see a rough general trend of increasing fine - tuning with increasing higgs mass . on the other hand ,
it is possible to have points with relatively small @xmath161 @xmath251 that satisfy the lep2 bound of @xmath252 gev , as demonstrated in point 1 of table [ tb : samplepoints ] .
this is a great improvement over the mssm , and it is a consequence of the large tree - level mass we can obtain in tessm , so we do not have to rely on large radiative corrections from @xmath159 and @xmath87 . ) as a function of the mass lightest @xmath1-even higgs boson .
this is typically less than the fine - tuning of the mssm ( as defined in eq .
[ eq : higgsfinetune0 ] ) and the nmssm .
the plot on the left has @xmath166 , and the plot on the right has @xmath167 .
, title="fig : " ] ) as a function of the mass lightest @xmath1-even higgs boson .
this is typically less than the fine - tuning of the mssm ( as defined in eq .
[ eq : higgsfinetune0 ] ) and the nmssm .
the plot on the left has @xmath166 , and the plot on the right has @xmath167 .
, title="fig : " ] the vev of @xmath29 is induced by the vev s of the higgs doublets because the vev s of the higgs doublets @xmath253 induce a tadpole from the trilinear interactions of the form @xmath254 in the second line of eq .
[ eq : vmix ] .
we noted earlier that some cancellation between _ a priori _ unrelated parameters ( @xmath8 and @xmath255 , for example ) is required to keep @xmath82 ( and thus @xmath88 ) small and this leads to fine - tuning in the triplet sector .
however , it is worth pointing out that @xmath82 here does not receive a large radiative correction that requires a fine - tuning as severe as the fine - tuning in the hierarchy problem in the triplet - extended standard model potential analyzed in chivukula et al .
it is easiest to see this in the limit @xmath256 ( susy - limit in the triplet sector ) where the triplet vev @xmath82 in eq .
[ eq : vtcond ] takes a particularly simple form @xmath257 and the 1-loop corrections to @xmath82 then involve 1-loop corrections to the parameters @xmath14 , @xmath253 , and @xmath18 .
the parameters @xmath14 , @xmath8 , and @xmath18 come from the superpotential , and the nonrenormalization theorem dictates that the radiative corrections to these parameters run only in a logarithmical manner due to wavefunction renormalizations only . though the loop corrections to @xmath253 may require a fine - tuning of one part in a few hundreds ( this is the little hierarchy problem in the mssm ) , this is much more benign than the fine - tuning in the triplet - extended sm studied in chivukula et al .
@xcite .
on the other hand , there is a source of fine - tuning in @xmath82 because we often require some degree of cancellation to make @xmath88 small .
we can define a quantitative measure of fine - tuning in @xmath88 by @xmath258 so that @xmath162 is large when there is a large cancellation in the combination @xmath259 that makes @xmath88 unnaturally small . the definition in eq .
[ eq : tripletfinetune ] , however , may not be satisfactory because it does not take into account the range of allowed @xmath88 .
for example , for the parameters listed in eq .
[ eq : texample ] @xmath260 we have viable @xmath88 in the regions @xmath261 and it may be reasonable to expect that any value of @xmath18 in the small range between 250 gev and 375 gev is equally fine - tuned . however , eq . [ eq : tripletfinetune ] would give different values of @xmath162 for different values of @xmath18 , and may even diverge if @xmath18 is such that we have @xmath262 .
it is true that when @xmath262 we have unnatural , complete cancellation , but in our work we only use @xmath82 in a binary way : to distinguish cases with viable @xmath88 from those with unacceptably large @xmath88 .
once @xmath82 is small enough to have viable @xmath88 , we do not care whether @xmath263 gev or @xmath264 gev , for example . as in section
[ sec : tessm ] , we can also estimate the fine - tuning in @xmath88 due to @xmath18 as shown in athron et al .
@xcite when there is a cancellation in the numerator of eq .
[ eq : obliquetformula ] that makes @xmath88 small . with all parameters other than @xmath18 fixed
, we first compute @xmath265 such that for @xmath266 , @xmath88 is always viable ( @xmath267 ) , and define fine - tuning as @xmath268 this definition of fine - tuning is harder to implement because , given a set of parameters except @xmath18 , we first have to find out if regions of @xmath18 allowed by @xmath88 comes about because of cancellations , before we can apply eq .
[ eq : tripletfinetune2 ] .
for example , it is possible that @xmath88 is always viable for any value of @xmath18 ( as are the cases for points 3 through 6 of table [ tb : samplepoints ] ) , so that we can not apply eq .
[ eq : tripletfinetune2 ] as there is no fine - tuning in @xmath88 .
despite its limited applicability compared to @xmath162 , @xmath165 may be a more reasonable measure of fine - tuning when there is a cancellation that leads to a small value for @xmath88 . for point 1(2 ) in table [ tb : samplepoints ] , we have @xmath269 and @xmath270 , corresponding to a 33(11)% change in @xmath88 per a 1% change in @xmath18 , and also cancellation of one part in 4.7(6.4 ) . for the other four points in table [ tb : samplepoints ] where @xmath165 in eq . [ eq : tripletfinetune2 ] is not well - defined , the values of @xmath162 are small , indicating small fine - tuning for these sets of parameters . since a complete analysis of fine - tuning in the triplet sector in the tessm is outside the scope of this work , we will conclude this section noting that in an extreme case ( eq . [ eq : texample ] ) , @xmath271 , so we suspect that the typical fine - tuning in the triplet sector be less than one part in 24 .
in this work we have revisited a very simple extension to the mssm by adding a hypercharge - neutral , @xmath10 triplet chiral superfield .
we considered this model as a reasonably economical extension of the mssm and an alternative to the nmssm , and extended the phenomenological studies in several directions .
in addition to discussing the decoupling behavior of the triplets and comparing it to the decoupling behavior of the singlet of the nmssm , we have computed the mass of the lightest @xmath1-even higgs boson to one - loop in the large quartic coupling @xmath14 . with @xmath14 , the higgs - triplet - higgs coupling in the superpotential ,
being comparable with the top yukawa coupling , we find that the model is able to satisfy lep2 bounds on the higgs mass without contributions from the s - top sector . at the expense of perturbativity at the gut scale
, we have checked that the model can give much smaller fine - tuning in the electroweak sector than the mssm . in the triplet sector
, there may be fine - tuning in having small oblique corrections , but we estimate this fine - tuning to be no worse than about one part in 30 . with large @xmath14 ,
the tessm opens up previously forbidden regions of parameters in the mssm .
in particular , both s - tops can be light in the tessm .
the light s - tops can then lead to phenomenology that is very different from the mssm with important implications for the lhc , such as a higgs boson that is both gluo - philic and photo - phobic .
our simple analysis here can be extended in many directions , and these further studies must be done if the model is going to make precise predictions at the lhc . with large @xmath14
, there can be important higher - loop effects to the mass of the lightest , @xmath1-even higgs boson .
furthermore , important higher - loop qcd effects must also be included to properly study the gluon - gluon fusion production and the diphoton decay of the higgs boson .
we leave these open projects for the future and hope they may add to the already - rich possibilities of phenomenology that will be seen at the lhc .
in this appendix , we list the field - dependent matrices that enter into the coleman - weinberg potential in eq .
[ eq : cwex01 ] .
we have five mass matrices , one for each set of particles : the @xmath1-even higgs bosons ( @xmath272 ) , the @xmath1-odd higgs bosons ( @xmath273 ) , the charged higgs bosons ( @xmath274 ) , the neutralinos ( @xmath275 ) , and the charginos ( @xmath276 ) .
we first list the elements of the higgs bosons .
@xmath277 @xmath278 @xmath279 where @xmath280 satisfy the minimization conditions eqs .
[ eq : min - mhu ] and [ eq : min - mhd ] . for the neutralino and charginos ,
since we do not take into account mixing with the gauginos , we have reduced matrices compared to those in eqs .
[ eq : nmatrix ] and [ eq : cmatrix ] , and here we can simply replace the vevs by the corresponding particle @xmath281 @xmath282
in this appendix , we review the formula for the decay width of a real scalar @xmath283 ( with mass @xmath284 ) decaying into two photons @xmath285 @xcite . generally , given the interactions @xmath286 where @xmath287 ( @xmath288 ) @xmath289 is a charged scalar ( fermion ) @xmath290gauge boson@xmath291 with mass @xmath292 ( @xmath293 ) \{@xmath294 } and electric charge @xmath295 @xmath296 @xmath297 , the diphoton partial decay width is given by @xmath298 where @xmath299 are factors to account for additional degrees of freedom ( such as color ) and @xmath300,\\ f_{\psi}&=-2\tau_{\psi}\left[1+(1-\tau_{\psi})f(\tau_{\psi})\right],\\ f_w&=2 + 3\tau_w + 3\tau_w(2-\tau_w)f(\tau_w),\end{aligned}\ ] ] where @xmath301 this work is supported by the us national science foundation under grants no .
phy-0354226 , no .
phy-0555545 , and no .
we are indebted to r. sekhar chivukula and elizabeth s. simmons for inspiring this project and many useful discussions .
we would also like to thank neil d. christensen and puneet batra for helpful comments and suggestions .
the feynman diagrams in this work are drawn using ` jaxodraw ` @xcite , and we check some of our results using ` hdecay ` @xcite .
99 m. drees , arxiv : hep - ph/9611409 . s. p. martin , arxiv : hep - ph/9709356 .
m. dine , arxiv : hep - ph/9612389 . m. e. peskin , arxiv:0801.1928 [ hep - ph ] . m. acciarri _ et al . _
[ l3 collaboration ] , phys .
b * 519 * , 33 ( 2001 ) [ arxiv : hep - ex/0102025 ] .
a. heister _ et al .
_ [ aleph collaboration ] , phys .
b * 526 * , 191 ( 2002 ) [ arxiv : hep - ex/0201014 ] .
g. abbiendi _ et al . _
[ opal collaboration ] , eur .
j. c * 26 * , 479 ( 2003 ) [ arxiv : hep - ex/0209078 ] .
j. abdallah _ et al .
_ [ delphi collaboration ] , eur .
j. c * 32 * , 145 ( 2004 ) [ arxiv : hep - ex/0303013 ] .
r. barate _
[ lep working group for higgs boson searches and aleph collaboration ] , phys .
b * 565 * , 61 ( 2003 ) [ arxiv : hep - ex/0306033 ] .
y. okada , m. yamaguchi and t. yanagida , prog .
phys . * 85 * , 1 ( 1991 ) .
h. e. haber and r. hempfling , phys .
lett . * 66 * , 1815 ( 1991 ) .
h. e. haber and r. hempfling , phys .
d * 48 * , 4280 ( 1993 ) [ arxiv : hep - ph/9307201 ]
. h. e. haber , r. hempfling and a. h. hoang , z. phys .
c * 75 * , 539 ( 1997 ) [ arxiv : hep - ph/9609331 ] .
s. heinemeyer , w. hollik and g. weiglein , phys .
b * 440 * , 296 ( 1998 ) [ arxiv : hep - ph/9807423 ] .
s. heinemeyer , w. hollik and g. weiglein , phys . lett .
b * 455 * , 179 ( 1999 ) [ arxiv : hep - ph/9903404 ] .
s. heinemeyer , w. hollik and g. weiglein , jhep * 0006 * , 009 ( 2000 ) [ arxiv : hep - ph/9909540 ] . j. r. espinosa and r. j. zhang , jhep * 0003 * , 026 ( 2000 ) [ arxiv : hep - ph/9912236 ] .
m. s. carena , h. e. haber , s. heinemeyer , w. hollik , c. e. m. wagner and g. weiglein , nucl .
b * 580 * , 29 ( 2000 ) [ arxiv : hep - ph/0001002 ] .
j. r. espinosa and r. j. zhang , nucl .
b * 586 * , 3 ( 2000 ) [ arxiv : hep - ph/0003246 ] .
m. bastero - gil , c. hugonie , s. f. king , d. p. roy and s. vempati , phys .
b * 489 * , 359 ( 2000 ) [ arxiv : hep - ph/0006198 ] . c. balazs ,
m. s. carena , a. freitas and c. e. m. wagner , jhep * 0706 * , 066 ( 2007 ) [ arxiv:0705.0431 [ hep - ph ] ] .
n. pandita , z. phys .
c * 59 * , 575 ( 1993 ) .
n. pandita , phys .
b * 318 * , 338 ( 1993 ) .
t. elliott , s. f. king and p. l. white , phys .
d * 49 * , 2435 ( 1994 ) [ arxiv : hep - ph/9308309 ] .
s. codoban and m. jurcisin , acta phys .
slov . * 52 * , 253 ( 2002 ) .
d. j. miller , r. nevzorov and p. m. zerwas , nucl .
b * 681 * , 3 ( 2004 ) [ arxiv : hep - ph/0304049 ] .
u. ellwanger and c. hugonie , phys .
b * 623 * , 93 ( 2005 ) [ arxiv : hep - ph/0504269 ] .
u. ellwanger and c. hugonie , mod .
a * 22 * , 1581 ( 2007 ) [ arxiv : hep - ph/0612133 ] . j. r. espinosa and m. quiros , nucl .
b * 384 * , 113 ( 1992 ) .
j. r. espinosa and m. quiros , phys .
b * 279 * , 92 ( 1992 ) .
o. felix - beltran , int .
j. mod .
a * 17 * , 465 ( 2002 ) .
n. setzer and s. spinner , phys .
d * 75 * , 117701 ( 2007 ) [ arxiv : hep - ph/0612318 ] .
j. l. diaz - cruz , j. hernandez - sanchez , s. moretti and a. rosado , phys .
d * 77 * , 035007 ( 2008 ) [ arxiv:0710.4169 [ hep - ph ] ] .
r. sekhar chivukula , n. d. christensen and e. h. simmons , phys .
d * 77 * , 035001 ( 2008 ) [ arxiv:0712.0546 [ hep - ph ] ] .
r. n. mohapatra and g. senjanovic , phys .
d * 23 * , 165 ( 1981 ) .
r. n. mohapatra , n. okada and h. b. yu , phys . rev .
d * 77 * , 115017 ( 2008 ) [ arxiv:0711.0956 [ hep - ph ] ] .
r. n. mohapatra , n. setzer and s. spinner , jhep * 0804 * , 091 ( 2008 ) [ arxiv:0802.1208 [ hep - ph ] ] . t. blank and w. hollik , nucl .
b * 514 * , 113 ( 1998 ) [ arxiv : hep - ph/9703392 ] .
m. c. chen , s. dawson and t. krupovnickas , int .
j. mod .
a * 21 * , 4045 ( 2006 ) [ arxiv : hep - ph/0504286 ] .
m. c. chen , s. dawson and t. krupovnickas , phys .
d * 74 * , 035001 ( 2006 ) [ arxiv : hep - ph/0604102 ] .
h. chankowski , s. pokorski and j. wagner , eur .
j. c * 50 * , 919 ( 2007 ) [ arxiv : hep - ph/0605302 ] .
p. h. chankowski and j. wagner , phys .
d * 77 * , 025033 ( 2008 ) [ arxiv:0707.2323 [ hep - ph ] ] .
k. s. babu , x. g. he and s. pakvasa , phys .
d * 33 * , 763 ( 1986 ) .
s. willenbrock , phys .
b * 561 * , 130 ( 2003 ) [ arxiv : hep - ph/0302168 ] .
j. sayre , s. wiesenfeldt and s. willenbrock , phys .
d * 73 * , 035013 ( 2006 ) [ arxiv : hep - ph/0601040 ] .
p. batra , a. delgado , d. e. kaplan and t. m. p. tait , jhep * 0402 * , 043 ( 2004 ) [ arxiv : hep - ph/0309149 ] .
r. harnik , g. d. kribs , d. t. larson and h. murayama , phys .
d * 70 * , 015002 ( 2004 ) [ arxiv : hep - ph/0311349 ] .
s. chang , c. kilic and r. mahbubani , phys .
d * 71 * , 015003 ( 2005 ) [ arxiv : hep - ph/0405267 ] .
a. delgado and t. m. p. tait , jhep * 0507 * , 023 ( 2005 ) [ arxiv : hep - ph/0504224 ] .
p. athron and d. j. miller , phys .
d * 76 * , 075010 ( 2007 ) [ arxiv:0705.2241 [ hep - ph ] ] .
s. r. coleman and e. weinberg , phys .
d * 7 * , 1888 ( 1973 ) .
j. f. gunion and h. e. haber , nucl .
b * 272 * , 1 ( 1986 ) [ erratum - ibid .
b * 402 * , 567 ( 1993 ) ] .
s. dawson , a. djouadi and m. spira , phys .
lett . * 77 * , 16 ( 1996 ) [ arxiv : hep - ph/9603423 ] .
r. v. harlander and m. steinhauser , jhep * 0409 * , 066 ( 2004 ) [ arxiv : hep - ph/0409010 ] .
w. m. yao _ et al .
_ [ particle data group ] , j. phys .
g * 33 * , 1 ( 2006 ) .
r. barbieri and g. f. giudice , nucl .
b * 306 * , 63 ( 1988 ) .
g. w. anderson and d. j. castano , phys .
b * 347 * , 300 ( 1995 ) [ arxiv : hep - ph/9409419 ] .
g. w. anderson and d. j. castano , phys .
d * 52 * , 1693 ( 1995 ) [ arxiv : hep - ph/9412322 ] .
d. binosi and l. theussl , comput .
commun . * 161 * , 76 ( 2004 ) [ arxiv : hep - ph/0309015 ] .
a. djouadi , j. kalinowski and m. spira , comput .
commun . * 108 * , 56 ( 1998 ) [ arxiv : hep - ph/9704448 ] . | we revisit an extension of the mssm by adding a hypercharge - neutral , @xmath0(2)-triplet chiral superfield .
similar to the nmssm , the triplet gives an additional contribution to the quartic coupling in the higgs potential , and the mass of the lightest @xmath1-even higgs boson can be greater than @xmath2 at tree - level .
in addition to discussing the perturbativity , fine - tuning , and decoupling issues of this model , we compute the dominant 1-loop corrections to the mass of the lightest @xmath1-even higgs boson from the triplet sector .
when the higgs - higgs - triplet coupling in the superpotential is comparable to the top yukawa coupling , we find that the higgs mass can be as heavy as 140 gev even without the traditional contributions from the top s - top sector , and at the same time consistent with the precision electroweak constraints . at the expense of having landau poles before the gut scale ,
this opens up a previously forbidden region in the mssm parameter space where both s - tops are light .
in addition to having relatively small fine - tuning ( about one part in 30 ) , this leads to a gluo - philic higgs boson whose production via gluon - gluon fusion at the lhc can be twice as large as the sm prediction . |
under the conditions of low temperature and strong perpendicular magnetic field , a two - dimensional electron gas exhibits a striking macroscopic manifestation of a quantum phenomenon , namely the quantum hall effect @xcite : the hall conductivity exhibits quantized plateaus at well defined multiples of @xmath0 ( a fundamental constant ) . in not too clean samples , where the rle of random impurities ( disorder ) is more important than electron - electron interactions ,
the plateaus occur at integer multiples of @xmath0 giving rise to the so - called integer quantum hall effect ( iqhe ) .
it is widely believed that in the absence of a magnetic field , all wavefunctions for non - interacting , disordered electrons in two dimensions are localized . in the presence of a magnetic field however
, a delocalized state occurs at the centre of the ( disorder broadened ) landau level , with energy @xmath1 .
as one tunes the electron energy @xmath2 ( by varying the magnetic field ) , through the centre of a landau level , the localization length , @xmath3 , diverges as @xmath4 , where numerical simulations indicate that @xmath5 . , the number of states with wave functions that reach the boundaries is @xmath6 .
since the density of states , @xmath7 , remains a smooth function of energy , this number is always macroscopic .
however , the density of delocalized states @xmath8 and goes to zero in the limit of infinite sample size . ]
the plateaus with differing @xmath9 are separated by these critical points .
a theoretical description of these points remains one of the most challenging unresolved problems in the theory of disordered systems .
the supersymmetric sigma model describing disordered landau levels has a high symmetry .
so we are lead to believe that the model describing the critical point has at least that symmetry or even higher .
this constitutes a difficulty because on one hand all known critical theories with internal symmetry have at least one free parameter ( for instance , for wess - zumino - novikov - witten ( wznw ) models it would be the level ) and , on the other hand , there is no such free parameter for iqhe critical point . in our previous paper we concentrated on properties of ldos since this is a much simplier problem than a calculation of conductances . with ldos one can work with a closed system and not to worry about boundary conditions .
the first assumption of our work is that the operators representing ldos constitutes a decoupled sector of the theory .
therefore though the system in its entirety is described by some ( yet unknown ) supersymmetric theory , the ldos sector is much simplier .
this assumption was based on the research conducted for one - dimensional limit of the relevant supersymmetric sigma model , where the decoupling of ldos sector was demonstrated explicitely @xcite , @xcite .
these results led us to conclusion that the most likely candidate for the theory describing critical ldos was the h@xmath10 wznw model .
@xmath11 wznw model is defined on a non - compact space sl(2,c)/su(2 ) . since this space differs from the group space sl(2,r ) only by the signature of its metric tensor ( it is ( + , + , + ) in the former case and ( + , + , - ) in the latter one ) , it is legitimate to consider it as a wznw model and not a coset one . in the gaussian parametrization
the action can be written as @xmath12.\ ] ] in the semiclassical limit @xmath13 the primary fields satisfy the diffusion equation on the h@xmath10 space : @xmath14 where @xmath15 and @xmath16 .
this equation provides a natural link between the 2d wznw model and the work done on one - dimensional sigma model .
it was found that the eigenfunctions of the corresponding transfer matrix satisfy a similar equation : @xmath17 where the @xmath18-th power of ldos @xmath19 was identified as @xmath20 .
this identification naturally follows from the form of the correlation function of ldos derived in @xcite .
the relationship between @xmath21 and @xmath22 is obvious : w(y ) ~^2 ( , ^ , y ) , kk^ * = 1[w ] to identify the operator @xmath19 one has to recall some facts from the representation theory . the primary fields @xmath23 of the h@xmath10 model belong to representations of the sl(2,r ) group with angular momentum @xmath24 .
the auxilary coordinate @xmath25 parametrizes the action of generators of the algebra sl(2 ) represented as differential operators : ^+_j = -x^2_x + 2jx , d^0_j = - x_x + j , d^-_j = - _ x the coordinates @xmath26 parametrize a direction of the quantization axis ; this is reflected in the field parametrization used in the majority of papers on the h@xmath10 model : ^(j ) ( , ^ , y = ^-2 ) = ( |- x|^2^ + ^-)^2j [ standard ] since these eigenfunctions depend on @xmath27 , one can replace in eq.([w ] ) the integration over @xmath28 by the integration over @xmath25 and identify : w_j(y ) = ( -2j)^2 x ^j(x|z ) , kk^ * = 1 [ w1 ] where semiclassically w_j(y=^-2 ) = 2 ^ -k_(1 + 2j)(2 ^ - ) [ w2 ] to establish a relationship between field @xmath29 and ( [ standard ] ) , let us recall that the scaling dimensions of h@xmath10 primaries are d(j ) = - where @xmath30 is the level of the corresponding kac - moody algebra . as is known from the numerical simulations , the scaling dimension the @xmath18-th power of ldos fits to this formula with @xmath31 and @xmath32 @xcite .
this publication has also explained that the deviations from the parabolic dependence reported in the earlier work @xcite had their origin in the finite size effects .
the fact that the corresponding value of @xmath33 is not an integer does not constitute a problem for the @xmath11 theory , since in non - compact models @xmath34 is not quantized .
however , this probably means that the parent supersymmetric theory which includes in itself a compact sector , can not possess kac - moody symmetry .
this supports the point of view of zirnbauer who rejected kac - moody symmetry for iqhe critical point @xcite .
so we suggest that @xmath19 is related to the primary field with @xmath31 . to extract it from @xmath35
is the same as to extract @xmath29 from expression ( [ standard ] ) , which can be done by integration around a contour encircling infinity : ( 2q ) = _ cx_c|x
x^(2q -1)|x^(2q -1)^(j = -q)(x,|x , z , |z )
in our previous paper we argued that in order to reproduce existing results on correlation functions of ldos , one has to consider a certain modification of the @xmath36 wznw model .
namely , @xmath37-point correlation functions of ldos should be understood as @xmath38-point function with two additional operators @xmath39 present at zero and infinity . in cylindrical geometry
these points become minus and plus infinity respectively .
the operator @xmath39 was identified as a non - trivial operator with zero conformal dimension .
primary fields of @xmath36 wznw model have been classified and their fusion rules are known .
therefore we have to find @xmath39 among these operators .
it turns out that such non - trivial ( that is non unity ) operator can be identified as _ 0(x|z ) = _ j 0j^-1^j(x|z ) this limit is well defined semiclassically as follows from relation ( [ w2 ] ) . to make sure it holds in quantum field theory
, one can study the fusion rules .
let us consider a three - point correlation function of primary fields in the @xmath36 model @xcite,@xcite : & & < ^j_1(x_1|z_1)^j_2(x_2|z_2)^j_3(x_3|z_3 ) > = + & & [ three ] where the structure constant @xmath40 is the clebsh - gordon coefficient for the sl(2,r ) group .
c_123 = |x_12|^2(j_1 + j_2 - j_3)|x_13|^2(j_1 + j_3 - j_2)|x_23|^2(j_2 + j_3 - j_1 ) and @xmath41 is the quantum correction given by & & d_123 = [ quant ] + & & where @xmath42 the function @xmath43 is a meromorphic function introduced in @xcite .
it has zeroes at @xmath44 and @xmath45 at @xmath46 this function admits the integral representation & & y_w(- x ) = \{_0^ } , + & & 0 < e x < ( 1 + k ) outside this interval it can be defined using the following properties : _
w(j ) = y_w(j -1 ) , y_w(j ) = y_w(j - k ) k^-(2j + 1 ) since at @xmath47 the function @xmath48 , formula ( [ three ] ) has a simple semiclassical limit corresponding to the quantum mechanics of a free particle on the @xmath11 space .
the property crucial for our argument is the fact that structure constant @xmath41 contains the product of @xmath49-functions which vanishes when one of the angular momenta goes to zero . since @xmath50 at @xmath51
, we get from ( [ three ] ) : & & ^2 x_2 |x_2|^(4q - 2 ) < ^j_1(x_1|z_1)^-q(x_2|z_2)_0(0|0 ) > = + & & |x_1|^2(j_1 + q ) [ three1 ] d(j_1,-q ) = [ quant1 ] when two operators have @xmath51 we get : & & < _ 0(x_1|0)[]^q(z)_0(x_3| ) > = & & a(q ) = as it was mentioned above , we suggested that the two - point disorder average of @xmath18-th powers of ldos should be understood as the four - point function of the @xmath11 theory . in the cylindrical geometry where @xmath52 i get the following expression : & & = ^2 x ^*k x * + & & = + & & |2r(w/2r)|^- 2d_q^2 x ( x,|x;z,|z ) , z = ( w_12/r ) to derive the latter formula i used the fact that the confomal blocks depend only on the anharmonic ratios of @xmath53 and @xmath54 .
function @xmath55 satisfies the following differential equation : & & _ x^2f + + & & _ xf + k_zf = 0 at the moment i am still unable to present a complete solution of this equation . | i elaborate on the earlier suggestion that the model describing the plateaux transition in integer quantum hall effect scales to a particular point on the line of critical points of a theory with a higher symmetry .
localization , quantum hall effect , multifractality , conformal symmetry . |
in this section , we first calculate majorana zero - energy states , @xmath36 and @xmath37 , for @xmath57 by solving the bdg equation for the corbino geometry . from eqs .
( 1 ) and ( 2 ) in the main text , the bdg equation for @xmath185 is given by @xmath186 this decouples into two sets of equations , one for spin up and one for spin down .
we consider only the equations with spin down as those for spin up give diverging solutions which are not physical .
due to the absence of rotation symmetry of the system , the wave function in each region should be given in the form of a superposition of different angular momentum eigenstates .
hereafter we will measure energy and length in units of gap magnitude @xmath187 and superconducting coherence length @xmath188 , respectively . with these units ,
the wave function for the region of @xmath189 is expressed as @xmath190 here , @xmath191 are coefficients to be determined below , and @xmath192 is defined as @xmath193 where @xmath194 is the bessel function of the first kind that is regular at the origin @xmath195 . for @xmath196 , the wave function
is found as @xmath197 with coefficients @xmath198 , and @xmath199 with @xmath200 the hankel function of the first kind that goes to zero as @xmath201 . in order to obtain solutions , we determine the coefficients @xmath191 and @xmath198 by matching the wave functions @xmath202 and @xmath203 at @xmath204 , leading to @xmath205 and the following recurrence relations can be derived , @xmath206 where @xmath3 and @xmath207 range over all integers .
eqs . and
allow us to express the coefficients @xmath208 and @xmath209 ( @xmath210 and @xmath211 ) in terms of @xmath212 ( @xmath213 ) .
this fact indicates that we can construct two independent solutions , denoted by @xmath214 and @xmath215 , @xmath216 with one unknown coefficient @xmath212 ( @xmath213 ) for @xmath214 ( @xmath215 ) , where @xmath217 is the unit step function .
according to the particle - hole symmetry , if @xmath214 is a solution , then @xmath218 is also a solution , where @xmath40 is the particle - hole operator with @xmath41 the operator of complex conjugation .
this imposes a constraint @xmath219 , and hence @xmath220 .
we choose @xmath221 and @xmath222 where @xmath223 is the normalization constant .
the following combinations of @xmath214 and @xmath215 give two mbs , @xmath224 which are plotted in fig . 2 in the main text .
we show that @xmath36 and @xmath37 , centered at @xmath225 , respectively , are exponentially localized in the azimuthal direction to justify the assumption in eq .
( 12 ) in the main text that the coupling between the tip and mbs is exponentially small except for @xmath98 where @xmath89 with an integer @xmath90 .
we first solve an effective low - energy hamiltonian which is the @xmath226 hamiltonian obtained by fu and kane in ref .
@xcite incorporating the spatial variation of a superconducting phase difference due to two flux quanta in the josephson junction , @xmath227 where @xmath228 \delta^2_0/(\mu^2 + \delta^2_0)$ ] , @xmath29 is the chemical potential , @xmath229 is the distance between two superconductors , and @xmath230 is the phase difference of the josephson junction .
the effective hamiltonian , whose bases are two branches of counter - propagating helical majorana states after integrating out the transverse ( radial ) degree of freedom , is a good approximation in the limit of @xmath231 where @xmath232 . to calculate approximate zero energy solutions localized at @xmath233 with @xmath234 , we linearize @xmath235 around @xmath236 , @xmath237 for small @xmath238 where @xmath239
. then we obtain the exactly solvable hamiltonian @xmath240 thus allowing us to calculate the zero energy solutions or approximate majorana states which are exponentially localized in the azimuthal direction , given by @xmath241 \left ( \begin{array}{cc } 1 \\
-p \end{array } \right),\end{aligned}\ ] ] and probability densities @xmath242 .
\label{approxmbs}\end{aligned}\ ] ] to compare these approximate solutions with the exact mbs given in eq . , we integrate out the radial degrees of freedom of @xmath243 and @xmath244 , @xmath245 for the values of @xmath246 , @xmath247 and @xmath248 . probability densities of @xmath249 in eq
. are drawn by the solid line , and those of approximate solutions @xmath250 in eq .
are drawn by the dashed line .
, scaledwidth=60.0% ] fig .
[ figs1:mbsdensity ] shows the comparison between @xmath249 and @xmath250 , exhibiting a quantitatively good agreement between them .
this leads us to the conclusion that mbs found in eq .
are exponentially localized in the azimuthal direction , and hence the form of the tunneling hamiltonian of eq .
( 12 ) in the main text is a good approximation .
we prove that majorana bound states remain at zero energy for finite @xmath29 by using a taylor expansion around @xmath57 .
we start the proof by rewriting eq .
( 1 ) in the main text as @xmath251 where @xmath252 are the two lowest eigenvalues which are assumed to be differentiable functions with respect to @xmath29 , and @xmath253 are corresponding eigenstates . in the limit of @xmath254 , @xmath255 and @xmath256 become two majorana bound states , @xmath257 and @xmath258 . throughout this section
, we use the notation @xmath259 as the particle - hole transformed counterpart of @xmath260 where @xmath261 is the particle - hole operator .
we note that @xmath262 by the anti - unitary property of @xmath263 . and
@xmath264 if @xmath265 is a majorana state .
below , we will show that @xmath255 also for finite @xmath29 .
the taylor expansion of the bdg equation in terms of @xmath29 leads to @xmath266 where @xmath267 and @xmath223 is the normalization constant .
we note that @xmath268 and @xmath269 . by substituting these series into eq . and equating the terms in each power of @xmath29
, we obtain equations for zeroth - order , first - order , and so on , each of which is discussed below . the equation for the zeroth order of @xmath29 is given by @xmath270 which is trivial .
next , we consider the first - order equation given by @xmath271 if we multiply by @xmath272 , it yields @xmath273 the first term on the left side is zero by virtue of eq . .
then the second term is real , and we have @xmath274 resulting in @xmath275 .
here we used anti - unitarity of @xmath263 and anticommutation relation @xmath276 .
. with @xmath275 leads to @xmath277 if we apply @xmath263 to eq . @xmath278 allowing us to write @xmath279 which will be useful to prove @xmath280 below , where @xmath281 and @xmath282 are constants .
we can do the similar procedure as above to get @xmath280 .
the @xmath283-order equation with @xmath284 is given by @xmath285 multiplying @xmath272 to this equation gives @xmath286 using eqs . and , @xmath287 and we get @xmath288 . from eq .
, @xmath289 where @xmath290 and @xmath291 are constants .
we now verify that @xmath292 for all positive integer @xmath293 by induction on @xmath293 .
it is assumed that the following conditions from the @xmath294-order equation hold : @xmath295 where @xmath296 are constants . since we have shown above
that these conditions are fulfilled for @xmath297 , it is enough to show that if they hold for @xmath298 , then also @xmath299 holds . using @xmath300 for @xmath301 , the @xmath302-order equation
is given by @xmath303 if we multiply @xmath272 to this equation , we have @xmath304 using eqs . and with @xmath298 , @xmath305 in the left side can be calculated as @xmath306 and @xmath307 . from eq .
we get @xmath308 therefore , if eqs
. - hold for @xmath298 , then they also hold for @xmath309 .
in this section , we provide the details of the calculation of the time - averaged tunneling current @xmath175 in the main text . it is obtained from the average of @xmath135 over a half rotation period , @xmath310 in the limit of very large integer @xmath311 so that @xmath312 . for a time @xmath313 $ ] , @xmath135 is roughly proportional to the tip - majorana tunneling magnitude @xmath314 .
thus @xmath135 is maximum when @xmath315 , and is exponentially small of the order of @xmath316 for @xmath317 .
we first calculate the tunneling current @xmath135 in the time interval @xmath318 $ ] . from eq .
( 14 ) in the main text , we obtain @xmath319,\ ] ] where @xmath320 + \frac{\lambda + i ( \varepsilon_k+ev)/\hbar}{2 \lambda } e^{i q \alpha } e^{-i ( \varepsilon_k+ev)(t - t_{0})/\hbar } + c.c . \right\}\ ] ] for @xmath321 $ ] , and @xmath322 + \frac{\lambda + i ( \varepsilon_k+ev)/\hbar}{2 \lambda } e^{i q \alpha } e^{-i ( \varepsilon_k+ev)(t - t_{0})/\hbar } + c.c . \right\}\ ] ] for @xmath323 $ ] .
@xmath324 is obtained by replacing @xmath138 by @xmath325 from @xmath326 . in this calculation
, we neglected the terms of the order of @xmath316 that give an exponentially small contribution to @xmath326 in our limit of the short duration of the tunneling @xmath101 .
then @xmath175 can be expressed as @xmath327 + n_f(\varepsilon_k ) \left [ \frac{1}{t_j } \int^{t_q + \frac{t_j}{2}}_{t_q -\frac{t_j}{2 } } dt a(t,- \alpha ) \right ] \right\},\ ] ] to calculate @xmath175 we have to calculate the time averages of @xmath326 and @xmath324 .
it is given by @xmath328 \nonumber\\ & + \left [ \frac{-\lambda + i ( \varepsilon_k + ev)/\hbar}{\lambda^2 + ( \varepsilon_k + ev)^2/\hbar^2 } e^{i \alpha q } e^{-i ( \varepsilon_k+ev)q t_j/\hbar } + c.c .
\right ] \bigg\}.\end{aligned}\ ] ] in the limit that @xmath311 goes to infinity , the summation term over @xmath293 on the right side can be written as @xmath329 \nonumber\\ = & \frac{2 \pi \hbar}{t_j } \sum^{\infty}_{l=-\infty } \delta \left[- \varepsilon_k - ev + \frac{(\alpha- 2\pi l ) \hbar}{t_j } \right],\end{aligned}\ ] ] where @xmath143 , while the remaining terms that possess @xmath330 factors can be neglected because of their rapid oscillation as a function of @xmath331 for large @xmath311 @xmath332 thus the time averages of @xmath326 and @xmath324 ( obtained by replacing @xmath138 by @xmath325 in @xmath326 ) have the forms @xmath333 , \nonumber\\ \nonumber\\ \frac{1}{t_j } \int^{t_q + \frac{t_j}{2}}_{t_q -\frac{t_j}{2 } } dt a(t,-\alpha ) & = \frac{2 \lambda}{\lambda^2 + ( \varepsilon_k + ev)^2/\hbar^2 } \frac{2 \pi \hbar}{t^2_j } \sum^{\infty}_{l'=-\infty } \delta \left[- \varepsilon_k - ev - \frac{(\alpha + 2\pi l ' ) \hbar}{t_j } \right ] , \nonumber\\ & = \frac{2 \lambda}{\lambda^2 + ( \varepsilon_k + ev)^2/\hbar^2 } \frac{2 \pi \hbar}{t^2_j } \sum^{\infty}_{l'=-\infty } \delta \left[- \varepsilon_k - ev - \frac{(\alpha - 2\pi l ' ) \hbar}{t_j } \right].\end{aligned}\ ] ] in the last line of the second equation , we changed @xmath207 by @xmath334 for convenience . if we put these into eq
, it follows that @xmath335 ^ 2 \bigg\ { & -(1-n_f(\varepsilon_k ) ) \sum^{\infty}_{l=-\infty } \delta \left[- \varepsilon_k - ev + \frac{(\alpha- 2\pi l ) \hbar}{t_j } \right ] \nonumber\\ & + n_f(\varepsilon_k ) \sum^{\infty}_{l'=-\infty } \delta \left[- \varepsilon_k - ev - \frac{(\alpha - 2\pi l ' ) \hbar}{t_j } \right ] \bigg\}.\end{aligned}\ ] ] by replacing the discrete sum over @xmath331 by an integral with the density of states of the metallic tip @xmath336 @xmath337 assuming a wide - band limit with an energy independent tunneling rate @xmath338 , we obtain @xmath339 ^ 2 \left[-n_f(ev -(\alpha - 2 \pi l ) \hbar
/t_j ) + n_f(-ev -(\alpha - 2 \pi l ) \hbar /t_j ) \right ] \nonumber\\ = & \frac{e}{h } \sum_{l } z_l(\alpha ) \left[-n_f(v^{+}_l ( \alpha ) ) + n_f(v^{-}_l ( \alpha ) ) \right],\end{aligned}\ ] ] where @xmath340 and @xmath341 are defined in the main text .
it is now straightforward to evaluate the time - averaged differential conductance @xmath136 .
since only the fermi - dirac distributions have the voltage dependence , it is enough to compute @xmath342 } \nonumber\\ = & \mp \frac{e}{4 k_b t } \text{sech}^2 \left [ v^{\pm}_l ( \alpha)/(2 k_b t ) \right].\end{aligned}\ ] ] therefore , we get @xmath343 + \text{sech}^2 \left [ \frac{v^{-}_l(\alpha)}{2 k_b t } \right ] \right\}.\ ] ]
in the previous section , we showed that the time - averaged differential conductance has peaks at @xmath149 . here , we provide an alternative derivation of the result by considering an effective floquet hamiltonian for the two rotating majorana bound states in our corbino - geometry josephson junction .
the floquet description is valid as the time - dependent bdg hamiltonian in eq.(7 ) in the main text is periodic in time with period @xmath2 : @xmath344 = h_{\text{bdg}}[\delta\phi(t ' + t_j)]$ ] where @xmath345 is the phase difference across the junction .
the time evolution operator of the hamiltonian from @xmath155 to @xmath156 in our low - energy subspace is @xmath346 in eq .
( 17 ) in the main text , and the effective floquet hamiltonian @xmath347 @xcite is obtained by using @xmath348 which yields @xmath349 where @xmath350 and integer @xmath3 .
the first term is reminiscent to the hamiltonian for two coupled static mbs with hybridisation @xmath351 ( which shows conductance peaks at @xmath352 ) , and the second term shifts energy levels . as a result , tunneling into an effective system described by @xmath347 gives conductance peaks at @xmath353 which is producing our features above .
in this section , we show that two majorana bound states in a corbino - geometry josephson junction are robust against weak disorder on the surface of a topological insulator if we stay in the subspace spanned by the two majorana bound states , i.e. , they remain at zero energy . it can be shown by using a taylor expansion in the same way as in section b. the surface of a topological insulator with a disorder potential is described by @xmath354 where @xmath355 is the disorder potential . then the bdg equation is given by @xmath356\psi'_j(r , \theta ) = e'_j \psi'_j(r , \theta ) .
\end{aligned}\ ] ] here a perturbation parameter @xmath357 is introduced to keep track of the order of the disorder potential in a taylor expansion , and @xmath358 and @xmath359 are eigenfunctions and eigenvalues , respectively , which are assumed to be differentiable functions with respect to @xmath357 .
like in the case of the calculation in section b , we do a taylor expansion for @xmath358 and @xmath359 in @xmath357 , @xmath360 now we substitute these expansions into the bdg equation and equate terms of the same order in @xmath357 .
all calculation details are the same as in section b , except that @xmath29 and @xmath361 in section b are replaced by @xmath357 and @xmath362 where @xmath363 is the pauli matrix in particle - hole space , and we get @xmath364 for all @xmath365 and nonnegative integer @xmath331 . therefore we conclude that two majorana bound states are robust against a weak disorder potential , and hence the results shown in fig . 3 in the main text can be detected even in the presence of weak disorder . | a phase from an adiabatic exchange of majorana bound states ( mbs ) reveals their exotic anyonic nature . for detecting this exchange phase
, we propose an experimental setup consisting of a corbino geometry josephson junction on the surface of a topological insulator , in which two mbs at zero energy can be created and rotated .
we find that if a metallic tip is weakly coupled to a point on the junction , the time - averaged differential conductance of the tip - majorana coupling shows peaks at the tip voltages @xmath0 , where @xmath1 is the exchange phase of the two circulating mbs , @xmath2 is the half rotation time of mbs , and @xmath3 an integer .
this result constitutes a clear experimental signature of majorana fermion exchange .
majorana fermions are charge - neutral quasiparticles obeying non - abelian exchange statistics that may be useful for quantum computation @xcite . over the past years , they have been predicted to appear in certain condensed matter systems , like topological insulators @xcite , semiconductors with spin - orbit interations @xcite , and magnetic atom chains @xcite , in proximity to _ s_-wave superconductors , and the search for experimental signatures of their unusual features has been intensified .
for example , the tunneling conductance between a metallic lead and a majorana fermion shows a zero - bias peak @xcite , and signatures for this prediction have been identified in experiments @xcite .
other schemes , such as interferometry in a dirac - majorana converter @xcite and a majorana - mediated josephson effect @xcite , have also been explored .
so far , however , most of the signatures that are predicted or observed are attributed to majorana features associated with charge neutrality and fermion parity anomaly , and less attention has been paid to find signatures from the exchange statistics , although several schemes to realize exchange or braiding operations have been suggested @xcite .
exchange statistics of majorana fermions , which differs from that of fermions or bosons , can provide alternative routes of detection .
an adiabatic exchange of two majorana fermions @xmath4 and @xmath5 leads to the transformation @xmath6 and @xmath7 , resulting in exotic phase factors @xcite . in ref .
@xcite , a long circular topological josephson junction is considered where the energy spectrum of the junction is found to depend on the exchange phase and fermion parity .
finding further experimental signatures of the exchange phase would be an important task for majorana fermion detection . in this work ,
we propose a transport experiment where the exchange phase of mobile majorana bound states ( mbs ) can be identified .
the setup consists of a corbino geometry josephson junction on the surface of a topological insulator . by solving the bogoliubov - de gennes equation
, we show that two spatially separated mbs at zero excitation energy appear in the junction lacking the hybridization between them when two flux quanta are introduced in the junction .
their positions can be moved along the circle by changing the superconducting phase difference across the junction , allowing for their exchange .
if a metallic tip is weakly coupled locally to the junction , and the two mbs are rotating adiabatically , the electron tunneling between the tip and mbs occurs every half rotation period . in this weak coupling limit , we find that the time - averaged differential conductance shows peaks at the tip voltage @xmath8 with @xmath9 the exchange phase @xcite .
this provides direct experimental signatures of the nature of majorana fermion exchange statistics via a dc charge measurement . and
@xmath10 are deposited on the surface of a topological insulator , forming a corbino - geometry topological josephson junction . in the presence of two flux quanta [ @xmath11 with @xmath12 , two zero - energy mbs ( red balls denoted by @xmath4 and @xmath5 )
appear at opposite sides of the junction .
their positions can be moved slowly while maintaining their relative distance by applying a small voltage across the junction , enabling us to perform an adiabatic exchange .
the resulting exchange phase can be identified in the time - averaged differential conductance of a metallic tip weakly tunnel coupled to the junction .
, scaledwidth=35.0% ] _ rotating mbs. _ we first find mbs in a corbino geometry josephson junction deposited on the surface ( @xmath13-@xmath14 plane ) of a topological insulator .
the setup is illustrated in fig .
[ fig1:setup ] .
the josephson junction is formed by thin films of inner ( @xmath15 ) and outer ( @xmath10 ) _ s_-wave superconductors and contains two magnetic flux quanta . for simplicity , we assume that the distance between @xmath15 and @xmath10 is zero .
we consider the case where the radius of @xmath15 denoted by @xmath16 satisfies @xmath17 so that the vector potential of the magnetic field can be neglected , where @xmath18 is the superconducting coherence length , and @xmath19 is the pearl length @xcite ; neglecting the vector potential is justified provided that the flux through the area where the bound states are localized is smaller than the flux quantum @xcite .
we also assume a short josephson junction , i.e. , @xmath20 where @xmath21 is the josephson penetration depth @xcite .
then the bogoliubov - de gennes ( bdg ) equation with nambu spinors @xmath22 of excitation energy @xmath23 has the form @xcite @xmath24 here @xmath25 is the single - particle hamiltonian describing the surface of the topological insulator , @xmath26 are pauli spin matrices , @xmath27 is the @xmath28 unit matrix , and @xmath29 is the chemical potential .
the induced gap @xmath30 in the presence of the two flux quanta can be described by the position - dependent form as @xmath31 where @xmath32 and @xmath33 are spatially uniform phases in each region , and the polar - angle - dependent term @xmath34 is due to the flux quanta @xcite .
we solve eq .
for the case of @xmath35 to obtain majorana zero - energy states , @xmath36 and @xmath37 , which are spatially separated and orthogonal .
they satisfy @xmath38 , @xmath39 where @xmath40 is the particle - hole operator and @xmath41 the operator for complex conjugation .
the analytical calculation of the mbs we present in the supplemental material @xcite .
note that the mbs remain at the zero - energy for nonzero @xmath29 ( see @xcite for the proof ) .
the probability densities of @xmath42 are shown in fig .
[ fig2:mbsdensity ] for different values of @xmath43 .
the result shows that their positions , which are defined by the locations where the probability density takes its maximum , are on opposite sides of the junction s circumference , and vary as a function of @xmath43 .
we denote the positions by @xmath44 with @xmath45 for @xmath46 .
they are determined by the condition @xmath47 and , thus , can be written as @xmath48 where @xmath49 is the local superconducting phase difference across the junction .
it is consistent with the previous studies that majorana fermions can be found in linear @xcite or circular @xcite josephson junction when the ( local ) difference of the superconducting phase is @xmath50 . and
@xmath51 , for different values of @xmath52 [ see eq . ] .
the interface of @xmath15 and @xmath10 is depicted by the white dashed circle .
the mbs are exchanged ( or rotated by @xmath53 ) when @xmath43 varies from @xmath54 to @xmath55 .
the parameters are @xmath56 and @xmath57 , where @xmath58 is the superconducting coherence length.,scaledwidth=45.0% ] the fact that the positions of the mbs are moved by changing @xmath43 allows us to perform an adiabatic exchange .
when we change adiabatically either @xmath32 from @xmath54 to @xmath59 or @xmath33 from @xmath54 to @xmath60 , the two mbs rotate by @xmath53 in a clockwise direction and are exchanged , as shown in fig .
[ fig2:mbsdensity ] . during this operation
, they continuously evolve as follows @xcite : @xmath61 where @xmath62 .
@xmath63 corresponds to the change of @xmath64 mentioned above .
we note that this transformation for the exchange is exact as long as the adiabaticity is fulfilled because the bound states are exactly at zero energy .
the adiabatic change of @xmath43 can be achieved experimentally by applying a constant voltage @xmath65 across the junction . for the adiabaticity
, @xmath66 should be much smaller than the excitation gap to the excited states .
then @xmath43 varies with time @xmath67 as @xmath68 , where @xmath69 is constant in time . in this case ,
@xmath36 and @xmath37 become instantaneous zero - energy eigenstates of the time - dependent @xmath70 $ ] , and their evolution during a half rotation time @xmath71 is described as in eq . @xcite .
equivalently , in terms of majorana operators , the exchange is expressed as @xmath72 here , @xmath73 and @xmath74 is the four - component field operator . _ exchange phase and transport. _ we now consider a normal metallic tip ( stm ) tunnel coupled to a position of the corbino josephson junction containing two circulating mbs .
in this situation , the coupling between the tip and one majorana state becomes significant or suppressed as the majorana state approaches to or leaves from the tip , respectively , while the other majorana state is on the opposite side of the junction with an exponentially small coupling to the tip , leading to occurrence of time - dependent tunneling periodically with period @xmath2 . in each period
, mbs are exchanged .
we will see below how the resulting exchange phase influences the tunneling differential conductance in this setup .
we perform our calculation in the weak tunneling limit .
also , in order to focus on the low - energy regime where the tip - majorana coupling dominates , we assume that the voltage @xmath75 across the junction and tip voltage @xmath76 are much smaller than the excitation gap .
then the low - energy hamiltonian of our setup is @xmath77 @xmath78 describes the normal metallic tip where @xmath79 represent momentum and spin , respectively , and @xmath80 .
the tunneling hamiltonian between the tip and the junction is @xmath81.\ ] ] as we are interested in low energies , we project @xmath82 onto this low - energy subspace by using @xmath83 .
then we obtain the time - dependent tunneling hamiltonian , @xmath84,\label{tipjunctionh}\ ] ] where @xmath85 is the coupling coefficient between the tip and @xmath86 at time @xmath67 , given by @xmath87 for further development , we consider a specific model for @xmath88 .
we first define the time sequence @xmath89 of the coupling with an integer @xmath90 such that if @xmath90 is even ( odd ) , @xmath91 $ ] is maximal , corresponding to a situation where @xmath92 @xmath93 $ ] is located closest to the tip , while @xmath94 @xmath95 $ ] is on the opposite side of the junction with an exponentially small coupling to the tip .
moreover , since mbs are spin polarized , we can restrict the consideration to electrons of one spin species of the tip and , thus , effectively deal with spinless electrons . by taking nearest - neighbor couplings and writing them in terms of @xmath96 and @xmath97 using eqs . and
, we obtain the tunneling hamiltonian at @xmath98 as @xmath99,\ ] ] where @xmath100 .
next , around @xmath98 , we assume that the magnitude of the coupling coefficient varies with time in an exponential manner , while the phase of the coupling does not change significantly ; the former assumption is justified by the fact that the mbs show in fig .
[ fig2:mbsdensity ] are exponentially localized in the azimuthal direction @xcite , and the latter one is a good approximation provided that @xmath101 , where @xmath102 is the tunneling duration .
then we have @xmath103.\ ] ] the current through the tip is @xmath104 .
@xmath105 is the tip number operator in the heisenberg picture with respect to @xmath106 .
the expectation value @xmath107 is taken over a thermal ensemble of initial states at time @xmath108 being in the far past with @xmath109 .
we switch on @xmath82 at this time @xcite .
since tunneling is assumed to be weak , we approximate the time evolution operator of @xmath106 up to first order in @xmath88 in the interaction picture .
we then obtain the tunneling current @xmath110 \rangle \nonumber\\ = & \frac{2e}{\hbar^2 } \text{re } \bigg\ { \int^{t}_{t_0 } dt ' \sum_{k , q , q ' } \gamma_{k , qq'}(t , t ' ) \nonumber\\ & \times [ g_{k}(t , t')-\bar{g}_{k}(t , t ' ) ] m(t , t ' ) \bigg\ } , \label{linearresponsei}\end{aligned}\ ] ] in the limit of @xmath101 , where the operator notation @xmath111 means an interaction picture operator . the current operator is given as @xmath112.\ ] ] we note that @xmath113 and @xmath114 evolve in time under @xmath115 and @xmath116 , respectively .
the tunneling magnitude is defined as @xmath117 @xmath118 and @xmath119 are the tip electron green s functions @xcite , and @xmath120 is the majorana green s function @xmath121 this function has the information of the majorana exchange that is crucial in our proposal . since the tunneling current is exponentially small except for @xmath98 and @xmath122 due to @xmath123 , we can approximate @xmath124 . in the adiabatic limit ,
the time evolution operator satisfying @xmath125 is @xcite @xmath126,\ ] ] where @xmath127 is the exchange phase where @xmath62 was defined in eq . .
from the transformations @xmath128^{q^{(\prime ) } } \gamma_1(t_0 ) [ u_m]^{q^{(\prime)}}$ ] , @xmath129 can be obtained as @xmath130 \nonumber\\ & + \mathcal{p}(t_0)[e^{i \alpha ( q - q')}-e^{-i \alpha ( q - q ' ) } ] \big\ }
\nonumber\\ \equiv & - i e^{i \eta \alpha ( q - q')}. \label{mbsgreenftn_alpha}\end{aligned}\ ] ] here , @xmath131 is the fermion parity of the junction at @xmath108 and is equal to @xmath132 when the two mbs share no ( one ) fermion , and @xmath133 for @xmath134 . by substituting eq . into eq . , we can obtain the tunneling current @xmath135 as a function of time @xmath67 and the tip voltage @xmath76 .
we discuss the voltage dependence of the time - averaged differential conductance @xmath136 @xcite .
it is enough to consider the current for @xmath1 and @xmath137 as it is the same for @xmath138 and @xmath139 with opposite signs . in the limit of @xmath101
, we obtain @xmath140 + \text{sech}^2 \left [ \frac{v^{+}_l(\alpha)}{2 k_b t } \right ] \bigg\ } , \label{didv}\end{aligned}\ ] ] where @xmath141 ^ 2,\ ] ] and @xmath142 with @xmath143 .
here , @xmath144 is the tunneling energy , and @xmath145 is the tip density of states at the fermi level .
we assume the wide - band limit for the tip so that @xmath146 becomes energy independent .
the calculated @xmath147 is valid for @xmath148 .
it shows peaks at @xmath149 with height @xmath150 as shown in fig .
[ fig3:didv ] . .
the parameters are @xmath151 .
the exchange phase @xmath152 is used . around zero voltage ,
the peaks emerge at @xmath153.,scaledwidth=45.0% ] _ discussion and experimental feasibility. _ we have shown that the time - averaged differential conductance between the metallic tip and the corbino - josephson junction hosting two rotating mbs exhibits peaks at @xmath154 , regardless of the fermion parity of the junction .
this feature results from the coherent interference in the time - dependent tip - majorana tunneling where the exchange phase @xmath138 from the half rotation of mbs introduces a relative phase between the tunneling at time @xmath155 and @xmath156 , as shown in eq . .
we emphasize that , different from the case of tunneling between the tip and a static mbs which yields a zero voltage conductance peak @xcite , we have conductance peaks at _ nonzero voltage _ for _ zero - energy _
mbs due to the exchange operation .
the height of the conductance peaks decreases with increasing tip voltage .
this is because when the voltage increases , the correlation time of electrons in the metallic tip @xmath157 becomes much smaller than @xmath2 , leading to the suppression of the interference between tunneling at @xmath155 and @xmath156 and , hence , to a decreasing peak height .
we note that the position of the conductance peaks can also be understood by considering an effective floquet hamiltonian as the bdg hamiltonian is periodic in time with period @xmath2 @xcite .
we now discuss the experimental feasibility . in our proposal , we require thin - film superconductors whose pearl length @xmath19 is much larger than the values of @xmath16 and @xmath18 .
thin - film nbn superconductors have @xmath158 @xmath29 m , @xmath159 nm , and superconducting gap @xmath160 mev @xcite . assuming the proximity - induced gap @xmath161 mev , the excitation gap of our corbino - josephson junction can be estimated by @xmath162 mev for @xmath163 and @xmath35 , where @xmath164 @xcite .
next , in order to have adiabatic rotation of the mbs and a short tunneling duration , @xmath2 should be long so that @xmath165 , but sufficiently short so that @xmath166 where @xmath167 is the quasiparticle poisoning time that is in the range of 0.1@xmath1681 @xmath29s @xcite .
estimating @xmath169 mev and @xmath170 nev , gives @xmath171 @xmath29ev for adiabatic interference without quasiparticle poisoning . in order to observe clearly the conductance peaks shown in fig .
[ fig3:didv ] , the temperature should be smaller than @xmath172 which provides an experimentally feasible temperature range of 10@xmath16810@xmath173 mk .
we finally note that the mbs are robust against weak disorder ; i.e. , they remain at zero energy @xcite , and , hence , the disorder has no significant effect on our results in fig .
[ fig3:didv ] . in conclusion
, we have proposed an experimental setup where the exchange phase of mobile mbs can be probed as a result of their interference effect .
different from the preceding detection schemes relying on charge neutrality of zero - energy mbs , our setup allows us to identify the nontrivial mbs statistics , providing alternative routes of mbs detection .
we thank c. w. j. beenakker , f. pientka , h .- s .
sim , b. trauzettel , b. van heck , and l. weithofer for valuable discussions .
p.r . acknowledges financial support from the eu - fp7 project se2nd no .
271554 and the dfg grant no .
re 2978/1 - 1 . 50 a. y. kitaev , phys .
usp . * 44 * , 131 ( 2001 ) .
m. z. hasan and c. l. kane , rev .
mod .
phys . * 82 * , 3045 ( 2010 ) .
j. alicea , rep .
. phys . * 75 * , 076501 ( 2012 ) . c. w. j. beenakker , annu .
rev . condens .
matter phys .
* 4 * , 113 ( 2013 ) . c. nayak , s. h. simon , a. stern , m. freedman , and s. das sarma , rev .
mod .
phys . * 80 * , 1083 ( 2008 ) .
g. moore and n. read , nucl .
phys . *
b360 * , 362 ( 1991 ) .
d. a. ivanov , phys .
rev .
lett * 86 * , 268 ( 2001 ) .
l. fu and c. l. kane , phys .
rev .
lett * 100 * , 096407 ( 2008 ) .
j. d. sau , r. m. lutchyn , s. tewari , and s. das sarma , phys .
rev .
lett * 104 * , 040502 ( 2010 ) .
y. oreg , g. refael , and f. von oppen , phys .
rev .
lett . * 105 * , 177002 ( 2010 ) .
t .-
p .
choy , j. m. edge , a. r. akhmerov , and c. w. j. beenakker , phys .
rev .
b * 84 * , 195442 ( 2011 ) .
s. nadj - perge , i. k. drozdov , b. a. bernevig , and a. yazdani , phys .
rev .
b * 88 * , 020407(r ) ( 2013 ) .
s. nadj - perge , i. k. drozdov , j. li , h. chen , s. jeon , j. seo , a. h. macdonald , b. a. bernevig , and a. yazdani , science * 346 * , 602 ( 2014 ) .
k. t. law , p. a. lee , and t. k. ng , phys .
rev .
lett . * 103 * , 237001 ( 2009 ) .
k. flensberg , phys .
rev .
b * 82 * , 180516(r ) ( 2010 ) .
v. mourik , k. zuo , s. m. frolov , s. r. plissard , e. p. a. m. bakkers , and l. p. kouwenhoven , science * 336 * , 1003 ( 2012 ) .
a. das , y. ronen , y. most , y. oreg , m. heiblum , and h. shtrikman , nat .
phys .
* 8 * , 887 ( 2012 ) .
m.t .
deng , c.l .
yu , g.y .
huang , m. larsson , p. caroff , and h.q .
xu , nano lett . * 12 * , 6414 ( 2012 ) .
e. j. h. lee , x. jiang , m. houzet , r. aguado , c. m. lieber , and s. de franceschi , nat . nanotechnol . * 9 * , 79 ( 2014 ) .
l. fu and c. l. kane , phys .
rev .
lett . * 102 * , 216403 ( 2009 ) a. r. akhmerov , j. nilsson , and c. w. j. beenakker , phys .
rev .
lett . * 102 * , 216404 ( 2009 ) .
l. fu and c. l. kane , phys .
rev .
b * 79 * , 161408(r ) ( 2009 ) . p. a. ioselevich and m. v. feigelman , phys .
rev .
lett . * 106 * , 077003 ( 2011 ) .
l. jiang , d. pekker , j. alicea , g. refael , y. oreg , and f. von oppen , phys .
rev .
lett . * 107 * , 236401 ( 2011 ) .
l. p. rokhinson , x. liu , and j. k. furdyna , nat . phys .
* 8 * , 795 ( 2012 ) .
b. sacp , j. b. oostinga , j. li , a. ubaldini , n. j. g. couto , e. giannini , and a. f. morpurgo , nat . commun . * 2 * , 575 ( 2011 ) .
j. r. williams , a. j. bestwick , p. gallagher , s. s. hong , y. cui , a. s. bleich , j. g. analytis , i. r. fisher , and d. goldhaber - gordon , phys .
rev .
lett . * 109 * , 056803 ( 2012 ) .
m. veldhorst , m. snelder , m. hoek , t. gang , v. k. guduru , x. l. wang , u. zeitler , w. g. van der wiel , a. a. golubov , h. hilgenkamp , and a. brinkman , nat . mater . * 11 * , 417 ( 2012 ) .
j. alicea , y. oreg , g. refael , f. von oppen , and m. p. a. fisher , nat . phys .
* 7 * , 412 ( 2011 ) . b. van heck , a. r. akhmerov , f. hassler , m. burrello , and c. w. j. beeankker , new j. phys .
* 14 * , 035019 ( 2012 ) .
j. li , t. neupert , b. a. bernevig , and a. yazdani , arxiv:1404.4058 .
t. karzig , f. pientka , g. refael , and f. von oppen , phys .
rev .
b * 91 * , 201102(r ) ( 2015 ) . e. grosfeld and a. stern , proc .
natl .
acad .
sci . u.s.a . * 108 * , 11810 ( 2011 ) . although majorana exchange in a system of two mbs @xmath4 and @xmath5 can not change the occupation of the non - local fermion @xmath174 in ref .
@xcite , our tunneling setup shown in fig .
[ fig1:setup ] detects the exchange phase of two mbs via an interference effect between different orders of exchange ( or braiding ) cycles .
j. r. clem , phys .
rev .
b * 82 * , 174515 ( 2010 ) .
r. s. akzyanov , a. v. rozhkov , a. l. rakhmanov , and f. nori , phys .
rev .
b * 89 * , 085409 ( 2014 ) .
a. l. rakhmanov , a. v. rozhkov , and f. nori , phys .
rev .
b * 84 * , 075141 ( 2011 ) .
see the supplemental material , which includes refs .
@xcite , for further details on the derivation of mbs , the calculation of @xmath175 , and the effect of weak disorder .
j. cayssol , b. dra , f. simon , and r. moessner , phys .
status solidi rrl * 7 * , 101 ( 2013 ) .
a. c. potter and l. fu , phys .
rev .
b * 88 * , 121109(r ) ( 2013 ) . note that @xmath176 is periodic in time with periodicity @xmath2 if either @xmath32 or @xmath33 is changed in time .
different switching - on times give the same current as long as they are far away from @xmath67 ( by many exchange periods @xmath2 ) .
the tip electron green s functions are given by @xmath177 $ ] , and @xmath178 where @xmath179 $ ] is the fermi - dirac distribution at @xmath108 , and tip voltage @xmath180 s .- z .
lin , o. ayala - valenzuela , r. d. mcdonald , l. n. bulaevskii , t. g. holesinger , f. ronning , n. r. weisse - bernstein , t. l. williamson , a. h. mueller , m. a. hoffbauer , m. w. rabin , and m. j. graf , phys .
rev .
b * 87 * , 184507 ( 2013 ) . d. rainis and d. loss , phys . rev .
b * 85 * , 174533 ( 2012 ) .
* supplemental material for `` detecting the exchange phase of majorana bound states in a corbino geometry topological josephson junction '' * sunghun park@xmath181 and patrik recher@xmath182 @xmath183__institute for mathematical physics , tu braunschweig , 38106 braunschweig , germany _ _
+ @xmath184__laboratory for emerging nanometrology braunschweig , d-38106 braunschweig , germany _ _ in this supplementary material , we analytically derive majorana zero - energy states for @xmath35 and prove that they are still at zero energy for finite @xmath29 .
next , we provide the details of the calculation of the time - averaged tunneling current @xmath175 in the main text , and an alternative explanation for the result shown in fig .
3 using floquet theory .
finally , we discuss disorder effects . |
@xcite identified the phenomenon of shear dispersion in which a passive scalar , e.g. a chemical pollutant , released in a pipe poiseuille flow spreads along the pipe according to a diffusion law .
the corresponding diffusivity , often termed effective diffusivity to distinguish it from molecular diffusivity , is inversely proportional to molecular diffusivity when the latter is small ( see also * ? ? ?
* ; * ? ? ?
this effective diffusivity is associated with a random walk along the pipe that results from the random sampling of the poiseuille flow by molecular brownian motion across the pipe .
the diffusive description of this random walk , and the corresponding gaussian profile of the scalar concentration , of course only apply on time scales that are much longer than the lagrangian correlation time scale .
shear dispersion is a striking example of a broad class of phenomena in which the interaction between fluid motion and brownian motion leads to a strong enhancement of dispersion and to effective diffusivities that are orders of magnitude larger than molecular diffusivity .
the importance of these phenomena in applications , in particular industrial , biological and environmental applications , is obvious .
this has motivated studies of effective diffusivity in many different flows ( see * ? ? ? * for a review ) .
these include spatially periodic flows which can be analysed using the method of homogenisation .
this method , which exploits the separation between the ( small ) scale of the flow and the ( large ) scale of the scalar field that emerges in the long - time limit , has proved highly valuable : it applies to more complicated flows , including time - dependent and random flows , and provides a unifying framework for methods used earlier .
shear dispersion , in particular , can be regarded as a special case of homogenisation applied to periodic flows , where cells repeat in the along pipe direction and the flow in each cell is simple poiseuille flow . in the large literature on shear dispersion
, efforts have been made to overcome the restriction to large times that underlies the diffusive approximation , and improved asymptotic estimates that capture some of the early - time behaviour have been obtained ( see @xcite for a review and @xcite for more recent results ) . for periodic flows , because the effective diffusivity is more difficult to compute , the focus has mainly remained on the derivation of asymptotic estimates and bounds , in particular in the limit of small molecular diffusivity ( e.g. * ? ? ?
* ; * ? ? ?
here we consider a different aspect .
the characterisation of dispersion in the long - time limit @xmath2 by an effective diffusivity and hence by a gaussian scalar distribution holds only close to the centre of mass of the distribution : the results of homogenisation are in essence a manifestation of the central - limit theorem and apply only to particles displaced from the mean by @xmath1 distances .
our aim is to go beyond this and describe the concentration far from the mean . to achieve this
, we derive large - deviation estimates for the concentration , that is , we derive the rate function @xmath4 in an approximation of the form @xmath5 for the scalar concentration at position @xmath6 and time @xmath7 .
( top panel ) and its logarithm @xmath8 ( bottom panel ) in a couette flow as a function of @xmath9 for @xmath10 and @xmath11 ( from left to right , curves have been offset for clarity ) .
monte carlo results ( symbols ) are compared with the large - deviation and diffusive predictions ( solid and dashed lines).,title="fig:",height=226 ] + ( top panel ) and its logarithm @xmath8 ( bottom panel ) in a couette flow as a function of @xmath9 for @xmath10 and @xmath11 ( from left to right , curves have been offset for clarity ) .
monte carlo results ( symbols ) are compared with the large - deviation and diffusive predictions ( solid and dashed lines).,title="fig:",height=226 ] large - deviation theory extends the central - limit theorem and applies to numerous probabilistic problems ( e.g. * ? ? ?
* ; * ? ? ?
when applied to the stochastic differential equations governing the motion of fluid particles advected and diffused in a fluid flow , it naturally yields an improved approximation to the scalar concentration ( interpreted as a particle - position probability function , cf .
this approximation is valid for distances from the mean that are @xmath0 rather than @xmath1 and therefore captures the tails of the distribution .
these are typically non - gaussian and not adequately represented by the diffusive approximation .
this is illustrated in figure [ fig : pdfcouette ] by the example of dispersion in a plane couette flow , one of the shear flows considered in detail in this paper .
the top panel shows the profile along the flow of the cross - stream averaged concentration @xmath12 at four successive times in the case of small molecular diffusivity .
the figure compares the averaged concentration obtained numerically using a monte carlo simulation ( symbols ) with the gaussian , diffusive approximation ( dashed lines ) and the large - deviation approximation derived in
23 ( solid lines ) .
the units of @xmath9 and @xmath7 have been chosen so that the maximum flow velocity and ( taylor ) effective diffusivity are both @xmath13 .
the inadequacy of the diffusive approximation in describing the tails of the concentration and the superiority of the large - deviation approximation are apparent in the top panel for the earliest profile @xmath14 .
they are obvious for all the profiles in the bottom panel which displays the results using logarithmic scale for @xmath12 .
this emphasises the tails of @xmath12 to reveal how the diffusive prediction overestimates dispersion and to demonstrate the effectiveness of the large - deviation approximation .
we note that while large deviation formally applies for @xmath2 , it appears here remarkably accurate for moderate @xmath7 .
( the discrepancies between large - deviation and monte carlo results for @xmath15 are mainly attributable to the limitations of the straightforward monte carlo method used here and are much reduced with the more sophisticated methods discussed in 3 . )
as the couette - flow example illustrates , large - deviation theory provides estimates of the low scalar concentrations in the tails , where the diffusive approximation fails .
this makes it relevant to a range of applications in which low concentrations matter .
examples include the prediction of the first time at which the concentration of a pollutant released in the environment exceeds a low safety threshold , and the quantification of the impact of stirring on chemical reactions in a fluid . in such examples
, there is a strong sensitivity of the response ( physiological or chemical ) to low scalar concentrations that makes the logarithm of the concentration , and hence the rate function @xmath4 , highly relevant quantities .
this broad observation can be made precise for the certain classes of chemical reactions . for f - kpp reactions ( e.g.
* ) , the combination of diffusion and reaction leads to the formation of concentration fronts that propagate at a speed that turns out to be controlled by the large - deviation statistics of the dispersion and given explicitly in terms of the rate function @xmath4 ( @xcite ; see also @xcite , ch .
7 , @xcite , ch . 2 , and @xcite ) .
the present paper starts in [ sec : formulation ] with a relatively general treatment of the large - deviation theory of dispersion which applies to time - independent periodic flows and to shear flows .
the key result is a family of eigenvalue problems parameterised by a variable @xmath16 .
the principal eigenvalue , @xmath17 , is the legendre transform of the rate function @xmath4 .
these eigenvalue problems can be thought of as generalised cell problems in that they resemble and extend the cell problem that appears when homegenization is used to compute effective diffusivities . in
[ sec : ldev][sec : prob ] we present two alternative derivations of the the eigenvalue problems : the first is a direct asymptotic method that treats the large - deviation form of the concentration as an ansatz ( see * ? ? ?
* ) ; the second follows the standard probabilistic approach based on the ellis
grtner theorem and considers the cumulant generating function of the particle position ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we then discuss the relation between large deviation and homogenisation ( [ sec : hom ] ) .
homogenisation , and the corresponding diffusive approximation , are shown to be recovered when the eigenvalue problems yielding @xmath17 are solved perturbatively for small @xmath18 up to @xmath19 errors .
carrying out the perturbation expansion to higher orders provides a systematic way of improving on the diffusive approximation ; in the case of shear dispersion , this recovers earlier results @xcite .
the rest of the paper is devoted to dispersion in specific shear and periodic flows .
we compute the functions @xmath20 and @xmath4 for the classical couette and poiseuille flows in [ sec : shear ] by solving the relevant one - dimensional eigenvalue problem numerically .
we also obtain asymptotic results for the concentration at small and large distances from the centre of mass . while the first limit recovers the well - known expression for the effective diffusivity of shear flows , the second captures the finite propagation speed that exists when diffusion along the pipe is neglected .
this provides a transparent example of the limitations of the diffusive approximation .
section [ sec : cell ] is devoted to a standard example of periodic flow , the two - dimensional cellular flow with streamfunction @xmath21 .
the numerical solution of the corresponding eigenvalue problems for specific values of the pclet number @xmath3 ( measuring the relative strength of advection and diffusion ) reveals interesting features of the dispersion , such as anisotropy , that are not captured in the diffusive approximation . using a regular perturbation expansion
, we derive explicit results in the limit of small @xmath3 .
we examine the opposite , large - pclet - number limit in a companion paper ( * ? ? ?
* hereafter part ii ) .
we conclude the paper with a discussion in
[ sec : disc ] . throughout the present paper and part ii
, we verify the predictions of large - deviation theory against direct monte carlo simulations of particle dispersion .
this is not without challenges since this requires estimating the tails of distributions which are associated with rare events and are , by definition , difficult to sample .
we have therefore used importance sampling and implemented two methods that are applicable broadly .
these are described in appendix [ sec : motecarlo ] .
two other appendices are devoted to technical details of certain asymptotic limits .
we start with the advection diffusion equation for the concentration @xmath22 of a passive scalar . using a characteristic spatial scale @xmath23 as reference length and the corresponding diffusive time scale @xmath24 , where @xmath25 is the molecular diffusivity , as a reference time
, this equation can be written in the non - dimensional form [ eqn : ad - dif ] _ t c + c = ^2 c , where @xmath26 is the pclet number . here
@xmath27 is the typical magnitude of the velocity field , which is assumed to be time independent , @xmath28 , and divergence free , @xmath29 . equation ( [ eqn : ad - dif ] ) can be considered as the fokker
planck equation associated with the stochastic differential equation ( sde ) which governs the position of fluid particles , [ eqn : sde ] = ( ) t + , where @xmath30 denotes a brownian motion . in this interpretation and with @xmath31 , the initial condition for the concentration is @xmath32 and the concentration at later times can then be thought of as the transition probability for a particle to move from @xmath33 at @xmath34 to @xmath6 at @xmath7 .
we focus on this initial condition and use the notation @xmath35 when the dependence on @xmath33 needs to be made explicit . in this paper
we consider two somewhat different flow configurations .
the first , relevant to taylor dispersion , corresponds to parallel shear flows , with @xmath36 unidirectional and varying in the cross - flow direction only , and a domain that is bounded in this direction .
the concentration @xmath35 then satisfies a no - flux condition at the boundary .
the second configuration corresponds to a periodic @xmath36 in an unbounded domain . in both cases ,
our interest is in the dispersion in the unbounded directions of the domain .
the shear - flow configuration can essentially be regarded as a particular case of the more general periodic - flow configuration , with the domain extending over only one period in the streamwise direction and no - flux boundary conditions replacing periodicity conditions .
because of this , we consider the two configurations together when developing the general large - deviation approach in the rest of this section .
any ambiguity that may arise as a result will be clarified in [ sec : shear ] and [ sec : cell ] when we apply the approach separately to shear flows and to two - dimensional periodic flows and obtain explicit results .
mixed configurations , in which the flow is periodic in certain directions and bounded in others , could also be treated with no essential changes .
we are interested in the form of @xmath35 for @xmath2 . under the assumption that @xmath37 , the solution to ( [ eqn : ad - dif ] ) can be sought as the expansion [ eqn : expc ] c(,t|_0 ) = t^-d/2 ^-t g ( ) ( _ 0 ( , ) + t^-1 _ 1 ( , ) + ) , = ( -_0)/t , where @xmath38 is the number of spatial dimensions .
this can be considered to be a wkb expansion with @xmath7 as large parameter .
the leading - order approximation [ eqn : largedevi ] c(,t|_0 ) ~t^-d/2 ( , ) ^-t g ( ) , has the characteristic large - deviation form in which @xmath39 is the cramr or rate function ( e.g. * ? ? ?
* ; * ? ? ?
* and references therein ) .
the conservation of total mass the spatial integral of @xmath35 imposes that [ eqn : g0 ] _ g ( ) = 0 and explains the presence of the prefactor @xmath40 in ( [ eqn : largedevi ] ) , as an application of laplace s method shows .
note that we concentrate on this leading - order approximation throughout and hence omit the subscript @xmath41 from @xmath42 .
introducing the expansion ( [ eqn : expc ] ) into ( [ eqn : ad - dif ] ) and retaining only the leading order terms gives ( _ g - g ) = ^2 - ( + 2 _ g ) + ( _ g + reduces to [ eqn : eig1 ] ^2 - ( + 2 ) + ( + ||^2 ) = f ( ) , where @xmath16 can be regarded as a parameter .
this can be rewritten compactly as [ eqn : expq ] ^ ( ^2 - ) ( ^- ) = f ( ) , in which the form of the operator on the left - hand side makes transparent the connection to the advection
diffusion operator @xmath43 .
the function @xmath42 satisfies no - flux boundary conditions when impermeable boundaries are present or periodic boundary conditions in the case of unbounded domains with periodic @xmath36 .
equation ( [ eqn : eig1 ] ) is central to this paper . together with its associated boundary conditions
, it gives a family of eigenvalue problems for @xmath42 parameterised by @xmath16 , with @xmath17 as the eigenvalue .
solving these eigenvalue problems ( numerically in general ) provides @xmath17 as the principal eigenvalue , that is , the eigenvalue with largest real part . the rate function @xmath39
is then recovered by noting from ( [ eqn : legendre ] ) that @xmath39 and @xmath17 are related by a legendre transform [ eqn : legendre1 ] f ( ) = _ ( - g ( ) ) g()= _ ( - f ( ) ) .
the fact that the critical points of @xmath20 are suprema and the convexity of @xmath20 can be deduced from the probabilistic interpretation of @xmath17 discussed below .
is differentiable ( e.g. * ? ? ?
] it follows that [ eqn : qxi ] = _ f , which gives a one - to - one map between the parameter @xmath16 and the physical variable @xmath44 .
the eigenfunction @xmath42 of ( [ eqn : eig1 ] ) associated with @xmath17 can therefore be equivalently thought of as a function of @xmath45 , as in ( [ eqn : largedevi ] ) , or of @xmath16 , as in ( [ eqn : eig1 ] ) .
note that the maximum principle can be used to show that @xmath17 is real and that @xmath42 is sign definite ( e.g. * ? ? ?
this is consistent with the asymptotics ( [ eqn : largedevi ] ) and the observation that the concentration @xmath35 is positive for all time if it is initially positive . to summarise , solving the eigenvalue problem ( [ eqn : eig1 ] ) for arbitrary @xmath16 and performing a legendre transform of the principal eigenvalue yields the large-@xmath7 approximation ( [ eqn : largedevi ] ) of the concentration .
this approximation is valid for @xmath46 and thus , as discussed below , extends the standard diffusive approximation which requires @xmath47 .
the eigenvalue problem ( [ eqn : eig1 ] ) can be thought of as a generalised cell problem since , as discussed in [ sec : hom ] , it generalises the cell problem of homogenisation theory .
@xcite derive this eigenvalue problem as part of a floquet bloch theory for linear equations with periodic coefficients and term it ` shifted cell problem ' ( see also @xcite , and 4 below ) .
an alternative view of the problem considers the moment generating function [ eqn : genfun ] w(,,t)= ^ , ( 0)=for the position of the fluid particles satisfying ( [ eqn : sde ] ) . here
@xmath48 denotes the expectation over the brownian process in ( [ eqn : sde ] ) .
the generating function obeys the backward kolmogorov equation [ eqn : backkol ] _ t w = w + ^2 w , w(,,0)=^ ( e.g. * ? ? ?
* ; * ? ? ?
a solution can be sought in the form [ eqn : w ] w(,,t)=^+ f ( ) t ^ ( , ) , where the function @xmath17 remains to be determined but will shortly be identified with that in ( [ eqn : legendre ] ) .
introducing ( [ eqn : w ] ) into ( [ eqn : backkol ] ) leads to [ eqn : eig2 ] ^2 ^+ ( + 2 ) ^+ ( + | |^2 ) ^= f ( ) ^ , with no - flux or periodic boundary conditions .
this corresponds to a family of eigenvalue problems , again parameterised by @xmath16 , which are the adjoints of those in ( [ eqn : eig1 ] ) , and hence have the same eigenvalues and in particular the same principal eigenvalue @xmath17 , justifying the notation in ( [ eqn : w ] ) .
this eigenvalue controls @xmath49 for @xmath2 . as a result
, it can alternatively be defined by [ eqn : f ] f ( ) = _ t ^(t ) and interpreted as the limit as @xmath50 of the cumulant generating function scaled by @xmath51 .
this function is convex by definition .
the relationship between the large-@xmath7 asymptotics of @xmath35 encoded in @xmath39 and that of @xmath49 can be made obvious .
noting from the definition ( [ eqn : genfun ] ) that @xmath49 is the legendre transform with respect to @xmath52 of @xmath53 with @xmath54 the variable dual to @xmath52 , we apply laplace s method to obtain @xmath55 where @xmath56 denotes the asymptotic equivalence of the logarithms as @xmath50 and we use ( [ eqn : largedevi ] ) to write @xmath57 . from ( [ eqn : w ] )
we obtain the first part of ( [ eqn : legendre1 ] ) . under the assumption of differentiability of @xmath17 , which ensures that @xmath39 is convex , the second part follows , allowing the computation of the rate function . the argument used in this subsection , which relies on laplace s method to establish a connection between rate function @xmath39 and scaled cumulant generating function @xmath17 ,
is an instance of the grtner
ellis theorem , a fundamental result of large - deviation theory which extends cramr s treatment of the sum of independent random numbers ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
rigorous results for a problem very similar to that defined above can be found in ( * ? ? ?
* ch . 7 ) .
it may be worth contrasting the large - time ( @xmath2 ) large deviations discussed in this paper , with the small - noise ( @xmath58 ) large deviations developed by freidlin & wentzell ( see * ? ? ?
* ) : while for small noise a single ( maximum - likelihood or instanton ) trajectory controls the rate function @xmath4 , this is not generally the case for large time .
as we discuss in the case of shear flows in 3 , it is only for @xmath58 and @xmath18 sufficiently large that @xmath4 can be expressed in terms of single trajectory and that the two forms of large deviations intersect .
some properties of @xmath17 and @xmath39 are useful to infer properties of the dispersion directly from @xmath17 without the need to carry out the legendre transform explicity .
as noted , @xmath17 and @xmath39 are convex .
therefore , from ( [ eqn : qxi ] ) , increasing @xmath16 correspond to increasing @xmath45 , and @xmath16 can be thought of as a proxy for the more physical variable @xmath45 .
it is clear from ( [ eqn : f ] ) that @xmath59 ; correspondingly , _
f(0)= _ * , defines @xmath60 which , by ( [ eqn : legendre1 ] ) , minimizes @xmath4 .
( [ eqn : largedevi ] ) then indicates that the maximum of @xmath22 and its centre of mass are located at @xmath61 .
qualitatively the legendre transform implies that a slow growth of @xmath17 away from its minimum corresponds to a rapid growth of @xmath39 and vice versa .
in particular , linear asymptotes for @xmath17 , say @xmath62 as @xmath63 in the one - dimensional case , correspond to vertical asymptotes for @xmath64 , @xmath65 as @xmath66 .
this implies that @xmath22 vanishes for @xmath67 , reflecting a finite maximum transport speed for the scalar .
exactly linear asymptotes do not arise for @xmath17 because the eigenvalue problem ( [ eqn : eig1 ] ) for @xmath68 has the simple solution @xmath69 which corresponds to a purely diffusive behaviour .
however , for large @xmath3 , there can be a range of values of @xmath16 for which @xmath17 is approximately linear and a finite transport speed controls scalar dispersion .
much of the literature on scalar dispersion focuses on the computation of an effective diffusivity governing the dispersion for @xmath2 and @xmath70 . in this approximation , ( [ eqn : ad - dif ] ) reduces to the diffusion equation [ eqn : effdiff ] _ t c + c = ( c ) , where @xmath71 is the spatial average of @xmath36 , and @xmath72 is an effective diffusivity tensor .
alternatively , @xmath71 and @xmath72 can be obtained from the particle statistics using [ eqn : effdiff2 ] _ t = _ t ( - t ) ( - t ) = .
the form of @xmath72 has been derived for a variety of flows using several essentially equivalent methods , starting with taylor s @xcite work on shear flows . in the last 20 years ,
homogenisation , as reviewed in @xcite and @xcite , has become the systematic method of choice .
the diffusive approximation ( [ eqn : effdiff ] ) can be recovered from the more general large deviation results : since the assumption @xmath73 implies that @xmath74 and hence that @xmath75 , we can expand @xmath17 according to f()= _ * + _ f + o(||^3 ) , where @xmath76 is the hessian of @xmath20 evaluated at @xmath77 . taking the legendre
transform gives g ( ) ~ ( - _ * ) _ f^-1 ( - _ * ) . in this approximation
the concentration is c(,t|_0 ) ^-(- _ * t ) _
f^-1 ( - _ * t)/(2 t ) corresponding to the solution of ( [ eqn : effdiff ] ) with [ eqn : effdiff3 ] = _ * = _ f/2 .
this result also follows from ( [ eqn : effdiff2 ] ) noting that the mean and covariances that appear on the left - hand sides are given by the first and second derivatives with respect to @xmath16 of the cumulant generating function @xmath78 evaluated @xmath77 .
since the diffusive approximation is recovered from the large - deviation results by an expansion for small @xmath16 , it can be expected that the method of homogenisation is equivalent to the perturbative solution of the eigenvalue problem ( [ eqn : eig1 ] ) or ( [ eqn : eig2 ] ) .
this is plainly the case .
consider the periodic - flow configuration and assume that @xmath79 for simplicity .
expanding = 1 + || _ 1 + ||^2 _ 2 + f = || _ 1 + ||^2 _ 2 + , and introducing this into ( [ eqn : eig1 ] ) yields at @xmath80 , @xmath81 where @xmath82 is a unit vector . averaging this equation
gives that @xmath83 .
the solution @xmath84 is then written as @xmath85 in terms of the periodic , zero - average solution @xmath86 of the so - called cell problem [ eqn : cellprob ] ^2 - = . ( see * ? ? ? * ) . at order @xmath87
, the eigenvalue problem reduces to @xmath88 averaging gives @xmath89 where the second equalities follows after some manipulations using ( [ eqn : cellprob ] ) ( see * ? ? ? * for details ) .
this corresponds to an effective diffusivity with components @xmath90 which is the standard homogenisation result .
an analogous computation detailed in appendix [ sec : exp ] shows how the homogenisation results for shear flows are recovered from the large - deviation calculation .
the perturbative solution of the eigenvalue problem ( [ eqn : eig1 ] ) offers a route for the systematic improvement of the diffusive approximation .
such improvements , which have been derived for shear flows by @xcite , @xcite and others ( see * ? ? ?
* for a review ) , extend the diffusion equation ( [ eqn : effdiff ] ) to include higher - order spatial derivatives and increase the accuracy of the approximation for @xmath2 .
they lead to effective equations of the form [ eqn : gendiff ] _ t c + c = _ ij _ ij c + _ ijk^(3)_ijk
c + _ ijkl^(4 ) _ ijlk c + , where summation over repeated indices is understood and we have introduced higher - order effective tensors @xmath91 , etc .
the behaviour of the large - deviation function @xmath17 as @xmath92 encodes all these tensors .
this can be deduced from the large - deviation form ( [ eqn : largedevi ] ) of the concentration which implies that @xmath93 and @xmath94 . combining these
formally leads to the effective equation [ eqn : efff ] _ t c = f(- ) c. comparison with ( [ eqn : gendiff ] ) shows that the various effective tensors that appear are given as derivatives of @xmath17 at @xmath77 .
hence they can be computed by continuing the perturbative solution of the eigenvalue problem ( [ eqn : eig1 ] ) to higher orders in @xmath95 .
this is demonstrated to @xmath96 for shear flows in appendix [ sec : exp ] .
another kind of improvement captures finite - time effects , specifically the fact that the mean and variance of the particle position have @xmath97 corrections to their linear growth which depend on initial conditions .
these corrections have been computed for some shear flows @xcite and termed ` initial displacement ' and ` variance deficit ' .
although we do not consider them further in what follows , it can noted that eq .
( [ eqn : backkol ] ) for the moment generating function is exact .
its solution for finite time can be expressed as a series of the form @xmath98 , where @xmath99 and @xmath100 denote the complete set of eigenvalues and eigenfunctions of ( [ eqn : eig2 ] ) .
the constants @xmath101 can be determined from the initial condition of the concentration .
it is clear , then , that the first 2 terms in the taylor expansion of @xmath102 , where the @xmath103 mode corresponds to the eigenvalue @xmath104 , determine the initial displacement and variance deficit ; the other eigenvalues @xmath105 contribute to exponentially small corrections . in the rest of the paper , we apply the results of this section to several specific shear and periodic flows .
we start with the case of shear flows for which the eigenvalue problems ( [ eqn : eig1 ] ) and ( [ eqn : eig2 ] ) simplify considerably .
consider the advection by a parallel shear flow @xmath106 in two dimensions , in a channel of width @xmath107 corresponding to @xmath108 for the dimensionless coordinate @xmath109 . without loss of generality ( exploiting a suitable galilean transformation as necessary ) the velocity can be assumed to satisfy [ eqn : zeroav ] u = _
-1 ^ 1 u(y ) y = 0 . because it is the longitudinal dispersion that is of interest , we modify ( [ eqn : largedevi ] ) and
take the large - deviation form of the concentration to be [ eqn : cdisp ] c(,t ) ~t^-1/2 ( y , ) ^-t g ( ) , = ^-1 x / t , assuming @xmath110 .
similarly , we write the moment generating function as [ eqn : w - disp ] w(q,,t ) = ^^-1 q x ^^-1 q x + f(q ) t ^(y ) . note that @xmath4 and @xmath20 depend only on the longitudinal variables @xmath111 and @xmath95 and that @xmath42 can be taken @xmath9-independent because of the @xmath9-independence of the flow .
the factors @xmath112 are introduced in ( [ eqn : cdisp])([eqn : w - disp ] ) for convenience : they lead to a legendre pair of functions @xmath113 and @xmath64 that are independent of @xmath3 in the limit @xmath114 , at least for @xmath115 . the eigenvalue problem ( [ eqn : eig1 ] ) then reduces to the schrdinger form [ eqn : eig - disp ] + ( q u(y ) + ^-2 q^2 ) = f(q ) .
this one - dimensional eigenvalue problem is completed by the no - flux boundary conditions [ eqn : bc - disp ] ( -1)=(1)=0 .
note that the operator in ( [ eqn : eig - disp ] ) is self adjoint and hence the same equation arises for the eigenvalue problem ( [ eqn : eig2 ] ) for @xmath116 associated with the moment generating function .
note also that ( [ eqn : eig - disp ] ) can be derived more directly using the feynman kac formula . to see this , write ( [ eqn : sde ] ) explicitly as [ eqn : sdeshear ] x = u(y ) t + w_1 ,
y = w_2 , and note that @xmath117 . the generating function ( [ eqn : w - disp ] )
then becomes @xmath118 using the feynman
kac formula ( e.g. * ? ? ?
* ) , @xmath119 is seen to satisfy @xmath120 and hence , for @xmath2 , to depend on @xmath7 as @xmath121 with @xmath113 the principal eigenvalue in ( [ eqn : eig - disp ] ) .
alternatively , ( [ eqn : eig - disp ] ) is obtained when seeking normal - mode solutions of the form @xmath122 to the advection diffusion equation ( [ eqn : ad - dif ] ) provided that the identification @xmath123 and @xmath124 is made .
the large - deviation form of @xmath125 is then recovered by applying the steepest - descent method to the normal - mode expansion of @xmath126 .
the large - deviation approach makes it clear that the saddle point in the @xmath127 plane is on the imaginary axis with a purely imaginary associated frequency @xmath128 .
below we solve ( [ eqn : w - disp])([eqn : bc - disp ] ) numerically for some classical shear flows . several general remarks can already be made .
first , the term proportional to @xmath129 in ( [ eqn : eig - disp ] ) is associated with longitudinal ( molecular ) diffusion . for @xmath130
, it can be neglected for @xmath58 , leading to the simpler eigenvalue problem [ eqn : eig - disp2 ] + q u(y ) = f(q ) which makes clear that @xmath113 and hence @xmath64 are independent of @xmath3 in the limit @xmath114 with @xmath131 .
the large - deviation form of @xmath22 can be written in terms of dimensional variables @xmath132 and @xmath133 as [ eqn : shearscale ] c(_*,t _ * ) ^-a^-2 t _ * g(x_*/(ut _ * ) ) , and its range of validity as @xmath134 and @xmath135 . in
what follows , we mostly concentrate on the limit @xmath114 and solve ( [ eqn : eig - disp2 ] ) rather than ( [ eqn : eig - disp ] ) : the effect of the neglected longitudinal diffusion on @xmath113 is straightforward , since it simply adds @xmath136 , but the corresponding change in @xmath64 is somewhat more complicated .
it is nonetheless a simple matter to estimate the size of @xmath95 for which the neglect of longitudinal diffusivity ceases to be a good approximation .
second , the perturbative solution of eigenvalue problem ( [ eqn : eig - disp ] ) for @xmath137 , provides an effective diffusivity as sketched in
[ sec : hom ] . in terms of @xmath113
, the dimensional effective diffusivity is expressed from ( [ eqn : shearscale ] ) as [ eqn : effdiff4 ] _ * = f(0 ) , and is inversely proportional to the molecular diffusivity in the limit @xmath114 .
the perturbative calculation carried out in appendix [ sec : exp ] gives [ eqn : fshear ] f(0 ) = ( _ -1^y u(y ) y ) ^2 . and recovers the explicit form of @xmath138 as obtained using homogenisation ( e.g. * ? ? ?
* ; * ? ? ?
the first of the corrections to the diffusive approximation of @xcite and @xcite is also computed in appendix [ sec : exp ] .
third , the asymptotics of ( [ eqn : eig - disp2 ] ) indicates that @xmath113 tends to @xmath139 as @xmath140 , where @xmath141 denote the maximum and minimum velocities in the channel .
this can be seen by noting that @xmath113 is the lowest eigenvalue of a schrdinger operator which , in the semiclassical limit @xmath142 , is given by the minimum of the potential @xmath143 ( e.g. * ? ? ?
the implication , as discussed in
[ sec : prob ] , is that @xmath65 as @xmath144 . physically
, this corresponds to the fact that fluid particles have longitudinal velocities in the range @xmath145 $ ] ; changes in the concentration therefore propagate at finite speeds and the concentration @xmath125 is compactly supported for @xmath146 $ ] .
this is only an approximation of course : when longitudinal molecular diffusion is taken into account , there is no limit on the propagation speed .
it is readily seen that the term @xmath136 becomes comparable to @xmath139 in @xmath113 for @xmath147 and that the rate function is approximately the diffusive @xmath148 for @xmath111 near @xmath149 ( @xmath150 ) or larger ( smaller ) .
this form of @xmath4 can also be shown to arise from an application of the @xcite small - noise large - deviation theory and is controlled by a single maximum - likelihood trajectory .
( this applies only when @xmath95 is sufficiently large : the dimensional expression ( [ eqn : shearscale ] ) makes this clear , with an argument of the exponential that scales like @xmath25 whereas the small - noise large deviation necessarily leads to a @xmath151 scaling , corresponding to a @xmath152 factor with our non - dimensionalisation . )
finally , we note that the eigenfunctions @xmath153 , where the @xmath111 dependence is inferred from the @xmath95-dependence using @xmath154 , have a simple interpretation . for @xmath155
the amount of scalar at @xmath109 for @xmath156 can be approximated as [ eqn : cond ] _
t^c(x , y , t ) x ( , y ) ^-t g
( ) , since , by the convexity of @xmath4 , the integral is dominated by the contribution of the endpoint @xmath157 . therefore @xmath153 gives the scalar distribution across the shear flow of particles with average speed greater than @xmath158 .
similarly , for @xmath159 , @xmath153 gives the distribution of particles with speed less than @xmath111 .
we now examine classical shear flows , starting with the plane couette flow [ eqn : couette ] u(y)=y .
the dispersion in this flow is illustrated in figure [ fig : pdfcouette ] .
the figure shows how the diffusive and large - deviation approximations provide a good approximation in the core of the scalar distribution and how only large deviation captures the tails .
figure [ fig : pdfcouette ] does not resolve the tails of @xmath12 with sufficient detail to assess the validity of the large - deviation approximation fully , however . in
what follows , we test systematically the large - deviation prediction for @xmath113 , defined as [ eqn : f - disp ] f(q ) = _ t ^^-1 q x(t ) with our shear - flow scaling , by comparing the value obtained by solving the eigenvalue problem ( [ eqn : eig - disp ] ) for a range of @xmath95 with careful monte carlo estimates .
the eigenvalue problem is solved using a finite - difference scheme .
( an exact solution can be written in terms of airy functions , but it is not particularly illuminating ) .
the monte carlo estimates approximate the right - hand side of ( [ eqn : f - disp ] ) as an average over a large number of solutions of ( [ eqn : sdeshear ] ) .
however , a straightforward implementation does not provide a reliable estimate for @xmath113 except for small values of @xmath95 .
this is because @xmath113 for moderate to large @xmath95 is controlled by rare realisations which are not sampled satisfactorily . to remedy this , it is essential to use an importance - sampling technique which concentrates the computational effort on these realisations .
for the results reported in this paper , we have implemented a version of grassberger s
@xcite pruning - and - cloning technique which we describe in appendix [ sec : resamp ] .
results for the plane couette flow are displayed in the leftmost panels of figure [ fig : shearall ] .
the top panel shows the eigenvalue and monte carlo approximations of @xmath113 along with asymptotic approximations valid for small and large @xmath95 .
the small-@xmath95 approximation for @xmath113 is found from ( [ eqn : fshear ] ) as [ eqn : smallqcouette ] f(q ) ~ q^2 q 0 .
the large-@xmath160 approximation is obtained by noting that for @xmath140 , the solution to ( [ eqn : eig - disp2 ] ) is localised in boundary layers near @xmath161 .
concentrating on @xmath63 , we introduce @xmath162 and @xmath163 into ( [ eqn : eig - disp2 ] ) . to leading order , this gives - y = , with solution @xmath164 decaying as @xmath165 .
imposing the boundary condition at @xmath166 gives the equation @xmath167 for @xmath168 .
hence we have [ eqn : largeqcouette ] f(q ) ~|q| - 1.019 |q|^2/3 |q| , using symmetry to deal with @xmath169 .
the top left panel of figure [ fig : shearall ] confirms the validity of the eigenvalue calculation and of the asymptotic estimates . in the case of the @xmath170 estimates , a constant
is added to ( [ eqn : largeqcouette ] ) to ensure a good match ; with this @xmath171 correction , the asymptotic formula appears to be accurate for @xmath160 as small as @xmath172 , say . the dispersive approximation corresponding to the parabola ( [ eqn : smallqcouette ] ) overestimates @xmath113 for all @xmath95 , indicating that this approximation overestimates the speed of dispersion or equivalently the magnitude of the tails of the distribution .
the rate function @xmath64 is shown in the second row of figure [ fig : shearall ] .
the solid curve is obtained by legendre transforming the function @xmath113 computed by numerical solution of the eigenvalue problem .
this is compared with direct monte carlo estimates .
again , it is crucial to use importance sampling to obtain a reliable estimate of @xmath64 for @xmath111 not small .
we have chosen to integrate a modified dynamics in which particles , instead of simply diffusing in the @xmath109-direction , also experience of drift towards the wall at @xmath173 ( or @xmath174 ) .
a better sampling is obtained because the wall regions control @xmath64 for large @xmath160 ; the method is described in appendix [ sec : girs ] .
the figure also shows the asymptotic approximations for @xmath64 deduced from ( [ eqn : smallqcouette ] ) and ( [ eqn : largeqcouette ] ) by legendre transform and given by [ eqn : asxicouette ] g ( ) ~ ^2 0 g ( ) ~ 1 .
the match between the values of @xmath64 derived from the eigenvalue problem and those obtained by monte carlo sampling provides a direct check on the validity of the large - deviation theory .
the discrepancy between the exact @xmath64 and its diffusive approximation confirms that diffusion overestimates the dispersion speed , as inferred already from the plot of @xmath113 . the finite support of the concentration distribution for @xmath175 $ ] , arising from the neglect of longitudinal molecular diffusion , is also hinted at by the large slopes of @xmath4 for @xmath176 .
the large-@xmath177 approximation to @xmath64 ( with @xmath171 term fixed by inspection ) is seen to be accurate for @xmath178 and could be combined with the small @xmath111 approximation to provide a satisfactory uniform approximation . the third panel on the left of figure [ fig : shearall ] shows the map between @xmath154 that arises as part of the legendre transform .
this identifies the location @xmath157 which control the corresponding exponential moment @xmath179 for large @xmath7 .
finally , the fourth panel shows profiles of the eigenfunctions @xmath180 of ( [ eqn : eig - disp ] ) for several values of @xmath95 . according to ( [ eqn : cond ] ) ,
these give the structure of the concentration profile for @xmath181 larger than @xmath154 .
thus , for instance , the eigenfunction for @xmath182 approximately corresponds to @xmath183 ( see third panel ) . as @xmath95 and hence @xmath111 increase ( or decrease )
the profile becomes more and more localised in the region of maximum ( or minimum ) velocity , that is , near @xmath173 ( @xmath174 ) .
the eigenfunctions for finite @xmath95 are to be contrasted with the standard ( homogenisation ) results on taylor dispersion which correspond to eigenfunctions that are small , @xmath80 perturbations to the uniform eigenfunction @xmath184 . [ cols="^,^,^,^ " , ] the eigenfunctions of ( [ eqn : eig1 ] ) shown in figure [ fig : efunctionpe250 ] for three different values of @xmath185 illustrate different regimes of dispersion that arise at increasingly larger distances from the scalar - release point . for small @xmath18 and hence for small @xmath186 , @xmath42 is almost uniform : a gentle @xmath187 gradient in the cell interiors is compensated by a rapid change in boundary layers that appear around the separatrices in agreement with the homogenisation treatment . for larger @xmath16 and @xmath186 , @xmath42 inside the cell is no longer close to uniform ; instead , it is approximately constant along streamlines but varies across streamlines , from small values at the centre to large values near the separatrices .
again , boundary layers around the separatrices ensure periodicity .
finally , for large @xmath18 and @xmath186 , @xmath42 is close to zero in the cell interiors and the scalar is confined within boundary layers .
this qualitative description of the eigenfunctions is consistent with the evolution of the scalar field shown in figure [ fig : simu250 ] ; it is supported by the asymptotics results reported in part ii .
this paper discusses the statistics of passive scalars or particles dispersing in fluids under the combined action of advection and molecular diffusion .
it shows how large - deviation theory provides an approximation to the scalar concentration or particle - position pdf in the large - time limit .
this approximation , expressed in terms of the rate function @xmath39 , is valid in the tail of the distribution as well as in the core ; it considerably generalises the more usual diffusive approximation which characterises the dispersion by a single effective - diffusivity tensor .
the rate function is deduced from the solution of the generalised cell problem ( [ eqn : eig1 ] ) , a one or two - parameter family of eigenvalue problems that generalise the cell problem solved when computing the effective diffusivity using the method of homogenisation .
the application to shear flows reveals features of the dispersion that are not captured by the standard theory of shear dispersion initiated by @xcite .
in particular , it shows that the diffusive approximation dramatically overestimates scalar concentrations far away from the centre of mass .
the reason for this is that the mechanism underlying shear dispersion
the interaction between shear and cross - stream molecular diffusion leads to along - flow dispersion with a finite speed , namely the maximum flow speed .
the non - zero concentrations beyond the limits imposed by this finite speed are entirely attributable to molecular diffusion and thus controlled by molecular rather than effective diffusivity . at intermediate distances from the centre of mass , however ,
the non - diffusive effects can in some cases increase and in some cases decrease dispersion .
this can be detected in some of the results for standard shear flows displayed in figure [ fig : shearall ] or be deduced from the order - by - order corrections to the diffusive approximation discussed in
[ sec : hom ] .
our analysis of spatially periodic flows and , in particular , of the classical cellular flow further demonstrates the benefits of large - deviation theory over homogenisation and the resulting diffusive approximation .
the anisotropy of the dispersion in this flow , for instance , although a clear consequence of the streamline arrangement , is overlooked by the diffusive approximation but quantified by large deviation . as for shear flows , there is also a finite speed effect for the dispersion in cellular flow ; this is more subtle and is elucidated in part ii which devoted to a detailed analysis to the large-@xmath3 limit .
the differences between the diffusive and large - deviation approximations for the scalar concentration are significant at large enough distances away from the centre of mass of the scalar .
since the concentration at such distances is small , large deviation applied to problems involving purely passive scalars is of practical importance in situations where low concentrations matter , as would be the case , for instance , for very toxic chemicals . in such applications
the logarithm of the concentration is often a relevant measure of the chemical s impact ; it is read off from the rate function since @xmath188 .
as mentioned in 1 , for scalars that are reacting , the properties of dispersion at large distances embodied in @xmath4 can be critical in determining the main features of the scalar distribution .
this was made explicit in the work of @xcite and @xcite which relates the speed of propagation of fronts for scalars experiencing f - kpp - type reactions to the rate function @xmath39 characterising passive dispersion .
following from this relationship , the results of the present paper and of part ii can be used to predict front speeds in a range of shear and periodic flows .
we will report elsewhere the novel predictions that can be obtained in this manner @xcite .
we conclude by remarking that the large - deviation treatment of scalar dispersion can be extended to a class of flows much broader than that considered in the present paper .
dispersion in time - periodic flows , random flows and turbulent flows can also be characterised by a rate function to improve on the approximation provided by effective diffusivity . in the time - periodic case
an extension of the theory discussed in 2 is straightforward : the eigenfunction @xmath42 in ( [ eqn : largedevi ] ) should depend on @xmath7 as well as on @xmath6 and @xmath45 , leading to an additional term @xmath189 in the eigenvalue problem ( [ eqn : eig1 ] ) and to the requirement that @xmath42 be time periodic which determines the eigenvalue @xmath20 . in the random case , under the assumption of homogeneous and stationary statistics for @xmath190 ,
@xmath20 is determined by the analogous requirement that @xmath42 , a random function , be homogeneous and stationary .
implementing this requirement is not straightforward , however , and monte carlo methods with importance sampling of the types described in appendix b may be best suited for the computation of the rate function .
* acknowledgments .
* jv acknowledges support from grant ep / i028072/1 from the uk engineering and physical sciences research council .
it follows from the scaled large - deviation form of @xmath125 for shear flows ( [ eqn : cdisp ] ) that @xmath191 in these expressions , @xmath95 is related to @xmath192 by @xmath193 and factors @xmath194 describing the error in the wkb - like expansion ( [ eqn : cdisp ] ) are omitted .
thus if we write [ eqn : exp ] f(q ) ~_n=1^n _ n q^n , an equation for @xmath125 follows in the form @xmath195 the solution to this equation gives for @xmath125 a form similar to ( [ eqn : cdisp ] ) with @xmath4 approximated by the legendre transform of the @xmath196-term taylor expansion of @xmath113 at @xmath197 .
in particular , truncating at @xmath198 gives the dispersive approximation with effective diffusivity ( [ eqn : effdiff ] ) .
the perturbative solution of ( [ eqn : eig - disp ] ) is straightforward : introducing ( [ eqn : exp ] ) and @xmath199 into ( [ eqn : eig - disp ] ) and omitting the term in @xmath129 gives at the first three orders , [ eqn:3eqns ] = _ 1 - u , = _ 2 + _ 1 _ 1 - u _ 1 =
_ 3 + _ 2 _ 1 + _ 1 _ 2 - u _ 2 . integrating the first equation and using ( [ eqn : zeroav ] ) gives @xmath200 and = - _ -1^y u(y ) y. an explicit expression for @xmath84 follows , which can be chosen such that @xmath201 .
integrating the second equation in ( [ eqn:3eqns ] ) and using the above gives [ eqn : alpha2 ] _ 2 = u _ 1 = ( _ -1^y u(y ) y ) ^2 . up to the factor @xmath152 , this is the effective diffusivity of taylor and homogenisation theory .
the function @xmath202 can then computed explicitly and the condition @xmath203 imposed .
finally , integrating the third equation in ( [ eqn:3eqns ] ) gives [ eqn : alpha3 ] _
3 = u _ 2 = u _ 1 ^ 2 , in agreement with @xcite .
note that the analogue of ( [ eqn : alpha2 ] ) for pipe flows is [ eqn : alpha2ax ] _ 2 = 2 _ 0 ^ 1 ( _ 0^r r u(r ) r ) ^2 .
we test the theoretical results by estimating the cumulant generating function from monte carlo simulations .
this relies on solving ( [ eqn : sde ] ) for an ensemble of trajectories @xmath204 , then computing [ eqn : samp ] w_k(t ) = _
k=1^k w^(k)(t ) , w^(k)(t ) = ^^(k)(t ) , for fixed @xmath16 . since @xmath205 as @xmath206 , @xmath207 for @xmath7 and
@xmath208 large . when @xmath16 is small , this method provides a good estimate of @xmath17 with @xmath7 moderately large , say @xmath209 or @xmath210 . for @xmath16 of order one or large ,
obtaining even a crude estimate of @xmath17 requires an exceedingly large number of realisations @xmath208 .
this is because the cumulant generating function is determined by exponentially rare , hence difficult to sample , realisations whose weights @xmath211 are exponentially larger than those of typical realisations . to estimate @xmath17 accurately with a reasonable number of realisations ,
it is necessary to use an importance - sampling method which concentrates the computational efforts on realisations that dominate ( [ eqn : samp ] ) .
we have adopted a simple method based on grassberger s @xcite pruning - and - cloning technique ( see also * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) which we now describe .
every few time steps in the numerical integration of ( [ eqn : sde ] ) , the current weight @xmath211 of each realisation is compared to the average @xmath212 . if @xmath213 , where @xmath214 is a parameter of the method ( typically chosen as @xmath215 or @xmath172 ) , the realisation is cloned : an additional realisation @xmath216 is created and integrated forward from the initial condition @xmath217 .
the two clones subsequently follow different trajectories , @xmath218 for @xmath219 because they experience different brownian motions .
the statistics of @xmath212 are left unchanged provided that the weight of the cloned realisations is divided by @xmath220 , that is , the weights @xmath211 in ( [ eqn : samp ] ) are multiplied by additional factors of @xmath221 for each cloning experienced by realisation @xmath127 . if @xmath222 , on the other hand , the realisation is pruned : it is killed with probability @xmath221 and , if surviving , its weight @xmath211 is multiplied by @xmath220 . to keep the number of realisations @xmath208 constant ,
random realisations are either cloned or killed .
we have implemented a slight extension of the method described in which the number of clones for realisations with @xmath213 , is @xmath223 .
the resampling steps make the method very efficient , and the results reported in the paper typically required a few minutes of computation on a modest desktop computer .
crucial to this efficiency is the fact that the cloning - pruning process tailors the ensemble of realisations to a particular value of @xmath16 by selecting those which dominate @xmath224 .
the rate function @xmath4 can be estimated directly by monte carlo simulations , using a binning procedure to approximate @xmath125 .
this is of course highly inefficient for the parts of @xmath4 away from its minimum @xmath60 since these are controlled by exponentially rare realisations which are poorly sampled .
one way of remedying this is to integrate a modified dynamics following the importance - sampling technique discussed in @xcite . for shear flows ,
we have adopted the following approach .
the modified dynamics , denoted by tilde , is given by [ eqn : sdemod ] x = u(y ) t + w_1 , y = r(y ) t + w_2 , instead of ( [ eqn : sdeshear ] ) . here
@xmath225 is a function chosen so that the distribution of @xmath226 better samples the regions where @xmath227 is large ( or small ) which control @xmath64 for @xmath111 away from @xmath228 .
girsanov s formula relates averages under the original dynamics ( [ eqn : sde ] ) to averages under this modified dynamics according to @xmath229 @xcite .
thus @xmath12 can be approximated by integrating numerically ( [ eqn : sdemod ] ) for an ensemble of trajectories and using a discretised version of the relation @xmath230 this result is used for to estimate the tails of @xmath125 and hence the form of @xmath4 for large @xmath177 with a much better sampling than achieved with the original dynamics . for the numerical results reported in [ sec : couette][sec : poiseuille ] , we have used @xmath231 to efficiently sample the portion of @xmath12 controlled by trajectories that remain localised near the wall at @xmath173 ( leading to anomalously large @xmath9 for couette flow and anomalously small @xmath9 for poiseuille flow ) , and @xmath232 to sample trajectories localised near the maximum of the plane poiseuille flow . the value of the parameter @xmath233 was chosen by trial - and - error to obtain the best representation of a portion of the curve @xmath64 . a similar modified dynamics for both @xmath234 and @xmath235
was used in the case of the pipe poiseuille flow in [ sec : pipe ] .
in the limit @xmath236 , the eigenvalue problem ( [ eqn : eig1 ] ) can be solved perturbatively by introducing the expansions @xmath237 of the eigenfunctions and eigenvalue into ( [ eqn : eig1 ] ) .
the leading - order , @xmath97 , equation is solved for @xmath238 and @xmath239 which reduces the @xmath240 equation to @xmath241 on integrating over a period , the left - hand side vanishes , leading to @xmath242 .
the solution is then found in the form [ eqn : phi1smallpe ] _ 1 = a x y + b x y + c x y + d x y , where the constants @xmath243 and @xmath38 are readily computed . integrating the @xmath244 equation @xmath245 over a period leads to the eigenvalue correction @xmath246 substituting ( [ eqn : phi1smallpe ] ) and taking the explicit form of the constants into account yields [ eqn : f2smallpe ] f_2 = . | the dispersion of a passive scalar in a fluid through the combined action of advection and molecular diffusion is often described as a diffusive process , with an effective diffusivity that is enhanced compared to the molecular value .
however , this description fails to capture the tails of the scalar concentration distribution in initial - value problems . to remedy this
, we develop a large - deviation theory of scalar dispersion that provides an approximation to the scalar concentration valid at much larger distances away from the centre of mass , specifically distances that are @xmath0 rather than @xmath1 , where @xmath2 is the time from the scalar release .
the theory centres on the calculation of a rate function characterising the large - time form of the scalar concentration .
this function is deduced from the solution of a one - parameter family of eigenvalue problems which we derive using two alternative approaches , one asymptotic , the other probabilistic .
we emphasise the connection between the large - deviation theory and the homogenisation theory that is often used to compute effective diffusivities : a perturbative solution of the eigenvalue problems in the appropriate limit reduces at leading order to the cell problem of homogenisation theory .
we consider two classes of flows in some detail : shear flows and periodic flows with closed streamlines ( cellular flows ) . in both cases , large deviation generalises classical results on effective diffusivity and
captures new phenomena relevant to the tails of the scalar distribution .
these include approximately finite dispersion speeds arising at large pclet number @xmath3 ( corresponding to small molecular diffusivity ) and , for two - dimensional cellular flows , anisotropic dispersion .
explicit asymptotic results are obtained for shear flows in the limit of large @xmath3 .
( a companion paper , part ii , is devoted to the large-@xmath3 asymptotic treatment of cellular flows . )
the predictions of large - deviation theory are compared with monte carlo simulations that estimate the tails of concentration accurately using importance sampling . |
in the physics of relativistic heavy ion collisions aiming at studying the qcd phase transition of hot and dense matter , two - particle correlation is widely used to investigate the characteristics of the space - time evolution in the collision processes @xcite .
it is well known as hanbury - brown twiss effect that the two - particle intensity correlation function contains the information on the `` size '' of the source of emitted particles due to a quantum symmetry . in recent experiments ,
the sizes are estimated from the correlation function through gaussian fitting and the physical meaning of the sizes depends on the fitting function . for an azimuthally symmetric source , the yano - koonin - podgoretski(ykp )
parametrization @xcite is used as one of the fitting functions . in this
parametrization , three of fitting parameters can be regarded as source sizes directly .
this advantage gives us a chance to observe a signature of quark - gluon matter because large emission duration is expected in the case of a first order phase transition @xcite . in the case of the relativistic heavy ion collisions ,
the reaction is highly dynamic ; the particle source can not be considered to be static one .
so we analyze the size parameters for pions quantitatively based on a hydrodynamic model which takes into account the longitudinal and transverse expansion and a first order phase transition @xcite .
from a hydrodynamic point of view , we consider the pion source as a thermalized one and a completely chaotic source .
hence , the two - particle correlation function is given as @xmath0 where @xmath1 and @xmath2 are momenta of a emitted pion pair and @xmath3 is the interference term .
as usual , we introduce the relative momentum @xmath4 and the average momentum @xmath5 which can be put as @xmath6 by virtue of the cylindrical symmetry . due to the on - shell condition , @xmath7 , only three components of the relative momenta are independent . in the ykp parametrization , @xmath8 , @xmath9 , and @xmath10
are used as the independent components .
then , the fitting function is given by @xmath11 \right.\\ \qquad \qquad \qquad \qquad \left .
-\left [ r_0 ^ 2(k^\mu)+r_\|^2(k^\mu ) \right ] \left [ q_\mu u^\mu ( k^\mu ) \right]^2 \right\ } , \label{ykp}\end{gathered}\ ] ] where @xmath12 is a four - velocity which has only longitudinal component .
the size parameters to be determined by a fit are @xmath13s @xmath14 .
we may choose a special reference frame where the velocity parameter @xmath15 vanishes ( ykp frame ) , because the three size parameters are invariant under longitudinal boosts . introducing the source function , @xmath16 and the weighted average , @xmath17 the size parameters in this frame ( ykp frame )
are expressed as @xmath18,\label{r0}\\ r_\|^2(k^\mu)&=&(\delta z)^2 -\frac{2\beta_\|}{\beta_\bot}\langle \tilde{x}\tilde{z } \rangle + \frac{\beta_\|^2}{\beta_\bot^2}[(\delta x)^2-(\delta y)^2],\label{rl}\end{aligned}\ ] ] where @xmath19 , @xmath20 , @xmath21 and @xmath22 . in order to regard the above
@xmath13s as naive source sizes directly , the second term and the third term in the right hand side of eqs.([r0 ] ) and ( [ rl ] ) must be small compared to the first term of each equation . though it has been shown that this condition holds within a class of thermal model @xcite , it is not trivial in other cases .
the case of a hydrodynamic model as a more realistic one is the point of the present paper . in the following ,
we calculate three quantities : hbt radii extracted from the fit to eq.([c2 ] ) , the space - time extensions , eqs.([rt])-([rl ] ) , and the source sizes , the first terms in eqs.([r0 ] ) and ( [ rl ] ) .
+ fig . 1 @xmath23 dependence of ykp radii .
open and closed circles show the experimental data and hbt radii , respectively .
solid lines stand for space - time extensions ( [ rt])-([rl ] ) .
dotted lines stand for source sizes .
( @xmath24 for @xmath25 and @xmath26 for @xmath27 . ) + figure 1 shows @xmath23 dependence of the pion hbt radii .
close circles stand for the hbt radii .
solid lines are space - time extensions which are expected to agree with the hbt radii when only thermal pions are considered .
dotted lines stand for the source sizes ( @xmath24 for @xmath25 and @xmath26 for @xmath27 ) .
as far as @xmath28 and @xmath25 are concerned , those quantities seem to be consistent with one another and well reproduce the na49 experimental results @xcite ( open circles ) . on the other hand , @xmath27 is smaller than the experimental one in spite of the fact our calculation includes a first order phase transition . the time durations @xmath26 agrees with @xmath27 at small @xmath23 .
however , some deviations from @xmath27 can be seen at large @xmath23 region , that means a failure of interpretation of @xmath27 as the time duration .
we find that this difference occurs from the third term in eq.([r0 ] ) . in eq.([r0 ] ) , the two additional terms means correlation between @xmath29 and @xmath30 ( the second term ) and the thickness of the source ( the third term ) .
though contribution from the second term is small in our model , the third term becomes negative in same order with @xmath31 @xcite .
such a kind of the source is called an `` opaque source '' . the `` opaque '' source
@xcite emits particles dominantly from the surface and the `` transparent '' source emits from whole region .
figure 2 shows the source function ( [ s ] ) projected onto the @xmath32 plane , @xmath33 being normalized as @xmath34 .
note that the average transverse momentum of the emitted pions is parallel to @xmath29 direction in the figure .
the left figure shows _ clearly _ surface emission ; pions are emitted from the crescent region and the source is thinner in the @xmath29 direction than in the @xmath35 direction .
this is the typical property of the opaque source which appears in the factor @xmath36 in eq.([r0 ] ) .
note that this factor also appears in the longitudinal components of the size , eq.([rl ] ) , but the prefactor @xmath37 makes the contribution to @xmath38 small . if we neglected the transverse flow by artificially putting @xmath39 , the source function becomes as shown in the right figure in fig . 2 .
in this case , the source function is proportional to the space - time volume of freeze - out hypersurface . as a consequence of restoration of the azimuthal symmetry which is shown clearly in the figure , the above measure of source opacity vanishes . although the particle emission takes place almost from the thin surface , the source is not opaque in this sense .
when the transverse flow exists , the source function is deformed by thermal boltzmann factor @xmath40 with @xmath41 almost proportional to @xmath42 .
consequently , the number of emitted particles increases in the region where @xmath43 ( i.e. , @xmath44 ) and decreases in the region where @xmath45 ( @xmath46 ) .
this flow effect deforms the surface dominant volume distribution ( right figure ) to the crescent shape ( left figure ) .
2 the source functions as functions of transverse coordinates @xmath29 and @xmath35 for @xmath47 mev and @xmath48 ( @xmath49 ) . the left figure denotes the source function with the transverse flow .
the right figure denotes the source function _ without _ transverse flow .
+ in summary , we analyze @xmath23 dependence of source parameters of the ykp parametrization based on the relativistic hydrodynamics for the cern - sps 158 gev / a pb+pb collisions .
we obtain the results which are almost consistent with the experiment .
however , the temporal source parameter , @xmath27 , shows smaller value than the experiment .
the source opacity makes the interpretation of @xmath27 as the time duration doubtful at large @xmath23 region .
we find that the source opacity is caused by the transverse flow and the characteristics of the freeze - out hypersurface , the surface dominant freeze - out .
the deviations of our results from the experiment would be improved by including the resonance decay and other effects .
the authors are indebted to professor i. ohba and professor h. nakazato for their helpful comments .
this work was partially supported by a grant - in - aid for science research , ministry of education , science and culture , japan ( grant no .
09740221 ) and waseda university media network center .
99 for the recent reviews , u.a .
wiedemann and u. heinz , phys . rep .
* 319 * ( 1999 ) 145 ; r.m .
ibid_. * 327 * ( 2000 ) 249 .
wu , u. heinz , b. tomik and u.a .
wiedemann , eur .
j. * c1 * ( 1998 ) 599 .
s. pratt , phys .
* d33 * ( 1986 ) 1314 . for more details , k. morita , s. muroya , h. nakamura and c. nonaka , phys .
* c61 * ( 2000 ) 034904 .
see also other references therein .
h. appelshuser _ et al .
_ , na49 collaboration , eur .
j. * c2 * ( 1998 ) 661 .
h. heiselberg and a. p. vischer , eur .
j. * c1 * ( 1998 ) 593 ; phys .
* b421 * ( 1998 ) 18 . | we numerically calculate two - particle correlation functions of cern - sps 158 a gev pb+pb central collisions based on a ( 3 + 1)-dimensional relativistic hydrodynamics with a first order phase transition .
we analyze the pair momentum dependence of pion source sizes extracted from the yano - koonin - podgoretski ( ykp ) parametrization which is expected to give the source sizes directly .
we find that , even in the case of the first order phase transition , the collective expansion and surface dominant freeze - out of the fluid naturally lead to the opaque source for which the interpretation of the temporal source parameter as the emission duration breaks down .
( 0,0)(0,0 ) ( 13.93,8.7)wu - hep-00 - 6 ( 14.63,8.2)twc-00 - 5 |
in equs . ( 11 ) and ( 12 ) we show that the decay rate and the change in momentum of an excited atom with initial momentum @xmath72 are given by @xmath104 we shall assume that the atom is heavy and expand only to first order in @xmath105 such that , for @xmath106 , @xmath107 the two polarisation directions @xmath15 and the unit wave vector @xmath16 are mutually orthonormal .
hence summing over the polarisations @xmath6 we get @xmath108 and @xmath109 for any vector @xmath110 , but here specifically for @xmath111 .
generally we have @xmath118 for smooth functions @xmath119 and @xmath120 where @xmath121 are all zeros of @xmath120 within the interval @xmath122 and @xmath123 . in our case
@xmath124 has only one positive root at @xmath125 = \omega_a + \frac{\omega_a}{mc}\left({\boldsymbol{\kappa}}\cdot{\mathbf{p}}_0 - \frac{\hbar \omega_a}{2c}\right ) + \mathcal{o}\big ( \big(\tfrac{\hbar \omega_a}{mc^2}\big)^2 \big)\,,\ ] ] with @xmath126 .
we thus expand , again to first order in @xmath105 , @xmath127 to obtain @xmath128 terms odd in @xmath129 vanish in the following integration @xmath130 where we get @xmath131 which can be checked , for instance , by choosing @xmath132 and @xmath133 . | we show how a simple calculation leads to the surprising result that an excited two - level atom moving through vacuum sees a tiny friction force of first order in @xmath0 . at first sight
this seems to be in obvious contradiction to other calculations showing that the interaction with the vacuum does not change the velocity of an atom .
it is yet more surprising that this change in the atom s momentum turns out to be a necessary result of energy and momentum conservation in special relativity .
atoms moving through and interacting with electromagnetic fields experience velocity dependent forces . for laser fields
this results in famous ( sub- ) doppler cooling schemes which form the foundation of modern atom - optical experiments @xcite while much earlier works by einstein and hopf already showed how such friction forces are necessary to discuss thermal radiation fields self - consistently @xcite .
but these considerations also showed that no such forces will arise from the interaction with the vacuum . here ,
however , we report how the simplest quantum - optical model , an initially excited atom moving through the vacuum , appears to be in contradiction with these results .
our starting point is the usual , non - relativistic atom - field hamiltonian in the dipole approximation including the rntgen term from which we first re - derive the decay rate from the excited state .
based on this well established framework we then go on to calculate the canonical and kinetic momenta of the atom and see that these are not constant in time , indicating the presence of a net force .
such a force emanating from the vacuum is , of course , highly suspicious .
even more suspicious is the fact that this force has the form of a friction , that is , the change in momentum is proportional to the initial momentum .
this suggests that there is no such force in the atom s rest frame and a co - moving observer will stay next to the particle while any other observer will see that the atom decelerates .
such a result would contradict the principle of relativity .
we find that only a proper consideration of energy and momentum conservation can solve this puzzle and that the momentum of the atom does change while its velocity remains constant . for the sake of clarity and brevity
we describe the atom as a two - level system , its dynamics will be calculated using time - dependent perturbation theory .
this model is described by a hamiltonian @xmath1 where @xmath2 describes the kinetic energy and dynamics of the atom with internal states @xmath3 and @xmath4 as well as the quantised electromagnetic field .
the last term includes sums over all modes @xmath5 and over the two polarisation states @xmath6 while @xmath7 is the usual annihilation operator for a photon in mode @xmath5 and polarisation @xmath8 .
the interaction between the atom and the field is given in the dipole approximation where we include the rntgen term @xcite @xmath9 \,,\ ] ] with the electric dipole operator @xmath10 where @xmath11 and electric and magnetic fields given by [ eq:_eandb_vacuum ] @xmath12 here @xmath13 is the electric field strength per photon in the quantisation volume @xmath14 and @xmath15 is a polarisation vector perpendicular to the mode propagating in a direction @xmath16 .
using the fields from equ . and @xmath17 we can rewrite the interaction hamiltonian from equ . in the rotating wave approximation , @xmath18 with @xmath19 and a momentum - dependent coupling term
@xmath20 the notation with a hat indicates that @xmath21 depends on the momentum operator @xmath22 and is thus an operator itself .
more familiar results from a theory ignoring the rntgen interaction term @xmath23 are recovered by setting @xmath24 . the rntgen term included in equ .
describes the coupling between a moving electric dipole and a magnetic field @xcite .
its presence leads to corrections ensuring the correct decay pattern of a moving atom @xcite and to a subtle difference between the canonical momentum @xmath22 and the kinetic momentum @xmath25 where @xcite @xmath26 a simple perturbative treatment suffices to demonstrate the problem . in this work
we consider the evolution of an initially excited atom in an eigenstate of the canonical momentum operator in vacuum , i.e. @xmath27 .
the coupling @xmath28 initiates transitions to the ground state with reduced momentum and a photon of momentum @xmath29 , i.e. @xmath30 with @xmath31 .
these are , of course , eigenstates of @xmath32 with energies @xmath33 and @xmath34 , separated by @xmath35 , @xmath36 the state at time @xmath37 can thus be described by @xmath38 with @xmath39 and @xmath40 .
these amplitudes evolve as [ eq:_cegdot ] @xmath41 where @xmath42 is the eigenvalue of @xmath21 given in equ . .
we can integrate these to find that @xmath43 is @xmath44 here we truncated @xmath45 and used @xmath46 for @xmath47 . using these approximations
we quickly obtain the decay rate @xmath48 , @xmath49 which is what we would expect from fermi s golden rule @xcite .
note that the delta function includes both the doppler shift @xmath50 as well as the recoil shift @xmath51 @xcite .
a calculation of the total decay rate to first order @xmath52 is given in the appendix .
it has been shown @xcite that this expression for the decay rate ensures that the directional decay rate obtained after performing the sum over polarisations @xmath6 and the frequency integral is consistent with a first - order approximation of the emission pattern of a moving dipole .
after calculating the decay rate it is a straightforward task to calculate the expectation value for the canonical momentum @xmath53 . as @xmath22 commutes with @xmath32
we see that @xmath54 such that @xmath55 or @xmath56 comparing this to @xmath57 as given in equ .
we see that the momentum change of the atom is simply given by the probability to emit a photon into a mode @xmath5 multiplied by the momentum of such a photon , @xmath29 .
let us stress again that the directional decay rate given in equ
. is consistent with a first - order approximation of the emission pattern of a moving dipole @xcite .
this result for the momentum change is thus required by momentum conservation . as mentioned in equ .
kinetic and canonical momenta are different in the presence of the rntgen term .
this difference is given by a term @xmath58 with @xmath59 .
in the notation of equ . this is @xmath60 . to see the difference in the change of these momenta
we calculate the time derivative of @xmath61 this derivative will consist of three terms ( cf .
equs . ):
the first is proportional to @xmath62 , it thus goes beyond our perturbation theory ; the second is @xmath63 , its imaginary part is thus zero ; hence the only surviving term is @xmath64 we recognise that this expression is quite similar to what we had in equ . , but here we get @xmath65 .
this is an integral involving a delta function multiplied by its argument , @xmath66 thus _ vanishes _ and the change of our kinetic momentum is equal to the change in canonical momentum given by equ . .
calculating a value for @xmath67 from equ .
involves rather straightforward , albeit lengthy , integration .
in contrast to previous works @xcite we do not assume an infinitely heavy atom here , but keep recoil terms to first order of @xmath68 as these appear in @xmath69 and the delta function .
more details on this calculation are given in the appendix .
in the end we find @xmath70 where @xmath71 is the total decay rate from equ . .
we see that the change in momentum is proportional to the atom s decay rate , the recoil energy and the initial momentum of the atom .
the momentum change given in equ .
is the main result and topic of this work .
it shows that an atom spontaneously emitting a photon changes its average momentum at a rate proportional to its initial momentum @xmath72 . as mentioned in the introduction , such friction terms in the interaction between light and matter
are not unusual . but
here this friction comes from the interaction with vacuum and we see that there is no force in the atom s rest frame where @xmath73 .
. seems to indicate that an observer co - moving with the initially excited atom will ( on average ) observe no change in the atom s motional state , as the photons are emitted in a random direction and the recoil adds up to zero .
but any other observer will see that the atom changes its initial momentum @xmath74 .
thus we note that observers in two different reference frames record different physical behaviour , a clear contradiction to the principle of relativity .
note that this is different from the unruh - davis effect which applies to particles initially _ accelerating _ through the vacuum @xcite .
of course we know that the physics described by the hamiltonian @xmath75 will not include all relativistic effects , but yet we expect its results to be correct to first order in @xmath0 , especially after we included the rntgen term which ensures that the emission pattern on the right hand side of equ .
is consistent with a moving dipole .
it is also of concern that the physics involved in this derivation was not at all complicated .
of course we made some assumptions , especially the dipole and rotating wave approximations and the truncation of the time - dependent perturbation theory , but these simplifications are standard textbook quantum optics @xcite and should not lead to obviously wrong physical results .
another simplification was to put the atom into a momentum eigenstate , but our arguments are repeatable for more physical initial states which are superpositions of these .
the resolution to this puzzle comes when we recognise that , in general , a change in ( kinetic ) momentum @xmath76 as given in equ .
is not only related to a change in velocity , but also to a change in mass as @xmath77 ( this relationship holds also in special relativity , at least to first order @xmath0 ) .
further , we know that the total mass of an atom ( here described as a dipole ) is not just the sum of its constituents ( electron and nucleus ) , but it is reduced by the internal binding energy @xmath78 .
this mass defect is , of course , well known , but it is usually discussed in connection with particle and nuclear physics where the binding energies are considerably larger than in atomic physics . in our case the decaying atom increases its binding energy by an amount @xmath79 , thus changing its mass at a rate @xmath57 as @xmath80 such a change in mass is consistent with a change in momentum , @xmath81 , if the velocity remains unchanged , @xmath82 .
the change in momentum of a decaying atom can therefore be explained by a change in its mass - energy while its _ velocity remains constant_. this also resolves the issue of two observers in different reference frames who now both see the same physics : an atom with constant velocity emitting a photon and thus losing energy at a rate @xmath83 .
this loss of energy is then also connected to a change in inertia and momentum .
but this resolution brings up another question : how did the changing mass enter our description of an atom interacting with an electromagnetic field ?
it is certainly _ not included _ in the hamiltonian which can be derived by explicitly setting @xmath84 @xcite .
we can also directly calculate the change in the particle s velocity as @xmath85/\hbar^2 $ ] to find a nonzero acceleration because @xmath86 , cf .
equ . .
a proper distinction between a change in momentum and an acceleration thus requires the derivation of a new atomic hamiltonian which also includes a dynamical mass - energy term coupled to @xmath22 . by virtue of the rntgen term
the hamiltonian used here properly describes the physics involved in the emission of the photon and this is also what enters the calculation of @xmath87 in equ . .
in the following short calculation we use a simple example to show that the proper description of the emission process is sufficient to derive the friction term . a movable two - way emitter sending photons of equal energy to the left and to the right .
an observer measuring the frequencies can use the doppler shift to measure the relative velocity and will infer that the momentum of the device changed during the emission process . ]
figure [ fig:_towwaytorch ] shows a battery - powered device emitting a pair of photons with the same frequency in two opposite directions .
if @xmath88 [ @xmath89 denotes the total mass - energy of the apparatus before [ after ] the emission process we see that its internal energy changes as @xmath90 .
the motion of the emitter remains constant by design as both photons carry equal momentum @xmath91 in opposite directions .
let this apparatus now move at a speed @xmath92 relative to two perfect detectors each measuring the frequency of one photon . due to the doppler shift
these frequencies are different as @xmath93 and @xmath94 with @xmath95 . using the available data the observer will construct the following relations of energy - momentum conservation [ eq:_twotorchexample_enmom2 ] @xmath96 such that @xmath97 and @xmath98 .
we thus see that the observer would calculate a change in the momentum of the apparatus even though there is no change in its velocity .
this simple example illustrates how the doppler effect directly links the change in internal energy during an emission process to a friction - like change in the momentum .
note again that although this momentum change is real it is not connected to an acceleration , the motion of the object remains unchanged . in our case of the decaying atom the presence of the rntgen term in equ
. led to a correct emission pattern in equ . which in turn was all we needed to calculate the momentum change of the atom in equ . .
our calculations were thus similar to the energy - momentum balance in equs . and also gave the same result @xmath99 . in conclusion
we have shown how a simple calculation of an excited two - level atom in vacuum results in a momentum change proportional to the initial velocity of the atom , cf . equ . .
if this momentum change were linked to a true deceleration of the atom , it would be in contradiction to the principle of relativity .
this riddle is solved , however , by recognising the change in the internal energy ( mass ) of the atom as the source of the momentum change .
thus , the momentum of the atom changes , but its velocity remains constant .
let us stress that this effect is not , intrinsically , a quantum optical phenomenon and is also not limited to the special case of a decaying atom in vacuum .
rather it is a general phenomenon occurring whenever a particle changes its internal energy , for instance , due to stimulated absorption and emission or , more subtle , due to stark or zeeman shifts . in these cases with external fields and resulting ( friction ) forces this effect is concealed within the stronger interactions leading to a true acceleration .
in contrast to nuclear or particle physics , the energies involved in atom optics are far below the atom s rest mass .
it is therefore not surprising that this subtle difference between momentum change and acceleration has not been measured yet .
ion traps have been used to measure nuclear masses to a precision of @xmath100 @xcite which is roughly three orders of magnitude more than the mass defect due to atomic binding energies , @xmath101 .
finally we would like to point out again that the current hamiltonian describing a dipole in an external field does not include a dynamic mass .
although the calculation of the momentum change @xmath102 bypasses this insufficiency , the result obtained for the actual acceleration @xmath103 is flawed .
this calls for a small correction of the atomic hamiltonian to include the coupling between the momentum and a change in internal energy which will be the topic of a future work .
answering the question in the title : we have shown that , yes , a decaying atom sees a force resembling friction .
but this force is a change in momentum due to a change in internal mass - energy and is not connected to decelerated motion .
we thank m. babiker , j. cresser and h. ritsch for stimulating discussions and gratefully acknowledge funding by the austrian science fund fwf ( j 3703-n27 ) , the dfg as part of the crc 1119 crossing and the royal society .
21ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop _ _ ( , ) @noop _ _ , ed .
( , ) @noop _ _ , ed .
( ) @noop * * , ( ) @noop * * , ( ) |
the usual sampling methods , e.g. , langevin dynamics and monte carlo method , fail for the complex action , since the boltzmann factor can not be regarded as the probability density .
this problem , called the sign problem , is still the main obstacle of the computation in lattice quantum field .
three reasons will always lead to the complex action : ( 1 ) grand partition function with finite density ; ( 2 ) fermion systems ; ( 3 ) topological terms in the action . to overcome the sign problem ,
the complex langevin dynamics , which is obtained from the complexification of the langevin dynamics , was used , which is rather successful in xy model @xcite , bose gas @xcite , thirring model @xcite , abelian and non - abelian lattice gauge model @xcite , qcd model @xcite , and its simplified model including one link u(1 ) model , one link su(3 ) model , qcd model in the heavy mass limit @xcite , one link su(n ) model @xcite , su(3 ) spin model @xcite , polykov chain model @xcite . for
some range of chemical potential @xmath0 and large fluctuation , the complex langevin may fails , e.g. , the xy model at finite chemical potential for small @xmath1 ( large fluctuation ) @xcite and in the thirring model in 0 + 1 dimension @xcite .
unfortunately from early studies of complex langevin evolutions @xcite@xcite@xcite until this day , the convergence properties of complex langevin equations are not well understood .
recently aarts etc . provided a criterion for checking the correctness of the complex langevin dynamics @xcite .
it is possible to find suitable variables to represent the partition function such that the action is real .
this is called the dual variable method .
it is successfully applied to many models , including bose gas @xcite , su(3 ) spin model @xcite , u(1 ) and z(3 ) gauge higgs lattice theory @xcite , massive lattice schwinger model @xcite , o(3 ) , o(n ) and cp(n-1 ) model @xcite@xcite@xcite@xcite , fermion bag approach @xcite , 4-fermion lattice theory , including massless thirring model @xcite , gross - neveu model @xcite , yukawa model @xcite , non - abelian yang - mills model @xcite@xcite , and its coupling with fermion field @xcite , lattice chiral model , and sigma model @xcite . although the routine to find the dual variable is case by case , it s technique behind is based on the high temperature expansion . for the fermion system , the dual method is called fermion bag approach @xcite .
this numerical method not only overcome the sign problem for model with small chemical potential , but also a high computational efficiency is achieved for the small or large coupling between the fermions . in this paper
we compare the langvin dynamics , the complex langevin dynamics and the fermion bag approach for the massive thirring model at finite density in 0 + 1 dimension . the chiral condensate and fermion density obtained by these two numerical methods ,
are compared with the exact result .
the arrangement of the paper is as follows . in section [ model ] , the fermion bag approach for thirring model is presented . in section [ langevin dynamics ] , the complex langevin dynamics and ( real ) langevin dynamics are given for this model by introducing a bosonic variable . in section
[ results ] , the implementation of these numerical methods are presented , and are compared with the exact result .
conclusions are given in section [ conclusion ] .
the lattice partition function for the massive thirring model at the finite density in 0 + 1 dimension reads @xmath2 where @xmath3 is the measure of the grassmann fields @xmath4 and @xmath5 , with even number of sites @xmath6 .
we adopt anti - periodic condition for @xmath7 and @xmath8 @xmath9 the staggered fermion action @xmath10 is @xmath11 here @xmath12 is always understood to be @xmath13 , which becomes 0 if @xmath14 since the period is @xmath6 .
the coupling constant @xmath15 between fermions is nonnegative .
the fermion matrix @xmath16 is @xmath17 with the chemical potential @xmath0 and fermion mass @xmath18 .
the antiperiodic condition for the @xmath7 and @xmath19 are accounted for by the sign @xmath20 and @xmath21 @xmath22 a periodic extension for @xmath20 and @xmath21 is used to define them for any @xmath23 .
thus we have @xmath24 for any @xmath23 . by using this formula for @xmath16
, the action @xmath10 in ( [ 2016_5_13_1 ] ) can be written as a sum over edges @xmath25 @xmath26 it is easy to check the symmetry of the fermion matrix @xmath27 with respect to @xmath0 and @xmath18 @xmath28 @xmath29 where @xmath30 , @xmath31 .
thus it is enough to study @xmath32 and @xmath33 in the massive thirring model .
the idea of the fermion bag approach is to expand the boltzmann factor @xmath34 by the high temperature expansion , @xmath35 and the partition function @xmath36 is written as the sum over the configuration @xmath37 @xmath38 where @xmath39 for all two neighboring sites @xmath25 , which satisfies @xmath40 for all site @xmath23 .
if @xmath41 , we say there is a bond connecting @xmath23 and @xmath42 ; otherwise , there are no bonds connecting them . for given configuration @xmath43 , for example
, there are @xmath44 bonds @xmath45 connecting @xmath46 different sites @xmath47 . for such kind of configuration @xmath43 , @xmath48 in ( [ 2016_4_27_2 ] )
is given by @xmath49 where @xmath50 is a @xmath51 matrix of propagators between @xmath46 sites @xmath52 , @xmath53 , whose matrix element are @xmath54 , @xmath55 .
the matrix @xmath50 depends on the order of @xmath56 , but it s determinant does not .
@xmath57 is the @xmath58 matrix which is obtained by deleting rows and columns corresponding to sites @xmath59 .
thus if @xmath44 is small , we use @xmath50 to calculate @xmath60 ; otherwise , we adopted @xmath57 to calculate @xmath60 . because of the symmetry ( [ 2016_4_27_4 ] ) and ( [ 2016_4_27_5 ] ) of @xmath16 , it is easy to show that for any @xmath0 , @xmath18 and any number of different sites @xmath61 @xmath62 where @xmath63 denote the function @xmath48 in ( [ 2016_4_27_6 ] ) since it depends on the fermion matrix @xmath27 .
we can rigorously prove that for any @xmath32 and @xmath64 , @xmath65 for any number of different sites @xmath66 ( [ appendix_1 ] ) .
thus the sign problem is avoided by the presentation ( [ 2016_4_27_2 ] ) of the partition function for the massive thirring model with finite density in 0 + 1 dimensions .
the chiral condensate is @xmath67 where the average is taken with respect to the weight of the partition function ( [ 2016_4_27_2 ] ) . from the first line in ( [ 2016_10_1_1 ] ) , @xmath60 is even in @xmath18 and the chiral condensate @xmath68 vanishes if @xmath69 .
similar to the calculation of @xmath48 in ( [ 2016_4_27_6 ] ) , we also have two methods to calculate its partial derivative @xmath70 @xmath71 the fermion density is @xmath72 which vanishes if @xmath73 since @xmath60 is even in @xmath0 by ( [ 2016_10_1_1 ] ) .
the partial derivative of @xmath48 with respect to @xmath0 is @xmath74 + \\ & & \sum_{t , t-1\neq t_1,\cdots , t_{2j}}\frac{1}{2 } e^{-\mu}s^2_{t}\det d \det g[(t_1,\cdots , t_{2j},t-1),(t_1,\cdots , t_{2j},t)]\end{aligned}\ ] ] where the @xmath75 propagator matrix @xmath76 $ ] has @xmath77 matrix element @xmath78 , @xmath79 .
here we use notations @xmath80 , @xmath81 , @xmath82 , @xmath83 .
the ratio in ( [ 2016_4_27_14 ] ) is @xmath84}{\det g[(t_1,\cdots , t_{2j}),(t_1,\cdots , t_{2j } ) ] } + \\ & & \frac{1}{2 } e^{-\mu}\sum_{t , t-1\neq t_1,\cdots , t_{2j } } s^2_{t } \frac { \det g[(t_1,\cdots , t_{2j},t-1),(t_1,\cdots , t_{2j},t)]}{\det g[(t_1,\cdots , t_{2j}),(t_1,\cdots , t_{2j})]}\end{aligned}\ ] ] note that @xmath85=g(\{t_1,\cdots , t_{2j}\})$ ] .
the ratio between the determinant can be obtained by @xmath86}{\det g[(t_1,\cdots , t_{2j}),(t_1,\cdots , t_{2j } ) ] } = g[(t\pm 1),(t ) ] \\ & & - g[(t\pm 1),\textrm{occu\_sites}]g[\textrm{occu\_sites},\textrm{occu\_sites}]^{-1 } g[\textrm{occu\_sites},(t)]\end{aligned}\ ] ] where @xmath87 denotes @xmath46 occupied sites .
@xmath88 $ ] is the @xmath89 propagator matrix with @xmath77 matrix element @xmath90 , @xmath55 .
another form of @xmath91 is @xmath92 + \\ & & \frac{1}{2}e^{-\mu } \sum_{t , t-1\neq t_1,\cdots , t_{2j } } s^2_{t } ( -1)^{(t)-(t-1 ) } \det d[\backslash(t_1,\cdots , t_{2j},t),\backslash(t_1,\cdots , t_{2j},t-1 ) ] \end{aligned}\ ] ] where @xmath93 and @xmath94 denote the position of @xmath23 and @xmath42 in the @xmath95 @xmath96 , respectively .
@xmath97 $ ] denotes the @xmath98 matrix obtained from @xmath16 by deleting rows @xmath99 and columns @xmath100 . using this formula , one has @xmath101 } { \det d[\backslash(t_1,\cdots , t_{2j}),\backslash(t_1,\cdots , t_{2j } ) ] } + \\ & & \frac{1}{2}e^{-\mu } \sum_{t , t-1\neq t_1,\cdots , t_{2j } } s^2_{t } ( -1)^{(t)-(t-1 ) } \frac { \det d [ \backslash(t_1,\cdots , t_{2j},t),\backslash(t_1,\cdots , t_{2j},t-1 ) ] } { \det d[\backslash(t_1,\cdots , t_{2j}),\backslash(t_1,\cdots , t_{2j } ) ] } \end{aligned}\ ] ] note that @xmath102= d(\backslash \{t_1,\cdots , t_{2j}\})$ ] .
the ratio between the determinant is @xmath103 } { \det d[\backslash(t_1,\cdots , t_{2j}),\backslash(t_1,\cdots , t_{2j } ) ] } = \textrm{d\_inv}[(t\pm 1),(t)]\end{aligned}\ ] ] where @xmath104 $ ] is the @xmath105 matrix element of @xmath58 matrix @xmath106 , which is the inverse of @xmath58 matrix @xmath107 $ ] .
the monte carlo algorithm based on the partition function in ( [ 2016_4_27_2 ] ) can be found in ref.@xcite .
we adopt the following steps to update the current configuration .
assume that the current configuration @xmath43 has @xmath108 bonds @xmath109,\cdots , [ t_{2n_b-1},t_{2n_b}])\ ] ] try to delete a bond , e.g. @xmath110 $ ] from the current configuration @xmath48 to be @xmath111,\cdots , [ t_{2n_b-3},t_{2n_b-2}])\ ] ] according to the detailed balance @xmath112 where @xmath113 and @xmath114 are the weight in the partition function ( [ 2016_4_27_2 ] ) for the configuration @xmath48 and @xmath115 , respectively .
the try probability from @xmath116 to @xmath117 are @xmath118 respectively . here
@xmath119 is the number of bonds which can be created from the configuration @xmath115 .
thus the accept probability from @xmath48 to @xmath115 is @xmath120 try to add a bond , e.g. @xmath121 $ ] from the current configuration @xmath48 to be @xmath122,\cdots , [ t_{2n_b-1},t_{2n_b}],[t_{2n_b+1},t_{2n_b+2}])\end{aligned}\ ] ] the detailed balance is eq .
( [ 2016_4_27_8 ] ) where @xmath123 here @xmath119 is the number of bonds which can be created from the configuration @xmath48 .
thus the accept probability from @xmath48 to @xmath115 is @xmath124 try to delete a bond , e.g. @xmath125 $ ] from the current configuration @xmath48 and then add a bond , e.g. , @xmath126 $ ] to be @xmath111,\cdots , [ t_{2n_b-3},t_{2n_b-2}],[y_{2n_b-1},y_{2n_b}])\ ] ] in the detailed balance ( [ 2016_4_27_8 ] ) , @xmath127 here @xmath119 is the number of bonds which can be created from the configuration @xmath48 where @xmath125 $ ] is deleted .
thus the accept probability to move a bond is @xmath128
the grassmann fields @xmath19 and @xmath7 can be eliminated if an bosonic variable @xmath129 are introduced @xmath130 where we omit a term depending on @xmath15 . inserting this formula to the partition function @xmath36 in ( [ 2016_4_27_0 ] ) and integrating over the grassmann fields
, one has @xmath131 where the @xmath132 fermion matrix @xmath133 under the bosonic variable @xmath134 is @xmath135 for @xmath136 .
the fermion matrix @xmath133 becomes the fermion matrix @xmath16 if @xmath137 .
this form of partition function in ( [ 2016_4_27_10 ] ) was studied in ref.@xcite .
the effective action in ( [ 2016_4_27_10 ] ) is @xmath138 is complex .
the discrete complex langevin dynamics for ( [ 2016_4_27_10 ] ) @xmath139 where @xmath140 denotes the discrete complex langevin time , @xmath141 is the time step .
the real white noise @xmath142 satisfies @xmath143 since @xmath144 is complex , @xmath145 is also complex . using ( [ 2016_5_13_4 ] ) and ( [ 2016_5_4_16 ] ) , the drift force in ( [ 2016_5_2_1 ] ) is @xmath146 the chiral condensate in ( [ 2016_4_27_12 ] ) is written as @xmath147 and the fermion density in ( [ 2016_4_27_14 ] ) reads @xmath148 here the average is taken with respect to weight @xmath149 .
the equation ( [ 2016_5_4_1 ] ) is also valid if @xmath129 and @xmath150 are replaced by @xmath151 , @xmath152 , respectively .
the partition function @xmath36 is @xmath153 where the @xmath132 fermion matrix @xmath154 under the bosonic variable @xmath134 is @xmath155 for @xmath136 .
fortunately the determinant of @xmath156 can be calculated by @xcite @xmath157 where the @xmath158 transfer matrix @xmath159 is @xmath160 inserting ( [ 2016_5_4_3 ] ) into the partition function @xmath36 in ( [ 2016_5_4_2 ] ) , one has @xmath161 where @xmath162 the chiral condensate is @xmath163 where the average is taken with respect to the weight @xmath164 .
the fermion density reads @xmath165 since the observable @xmath166 and @xmath167 for any real fields @xmath168 , the sign problem in this representation is avoided .
these averages @xmath169 can be calculated by the usual monte carlo method or langevin dynamics since these averages is taken over the gaussian weight @xmath170 .
we use ( real ) langevin dynamics @xmath171 to calculate these averages . here
@xmath140 denotes the discrete langevin time and @xmath172 is the time step .
the real white noise @xmath173 satisfies @xmath174 .
the averages @xmath175 and @xmath176 , which does not depend on the chemical potential @xmath0 , are obtained by samples @xmath168 after the equilibrium of the real langevin dynamics . moreover
, the chiral condensate in ( [ 2016_5_4_10 ] ) and fermion density in ( [ 2016_5_4_11 ] ) can be obtained from @xmath177 and @xmath178 for all chemical potential @xmath0 .
the computational complexity for the calculation of @xmath179 and @xmath180 are @xmath181 , which will be explained in the next section .
the exact formula for @xmath182 is known @xcite , @xmath183 @xmath184 thus these averages can be calculated exactly , which is called the exact result in the following sections .
the most advantage of the fermion bag approach is that the weight in ( [ 2016_4_27_2 ] ) is nonnegative and thus the sign problem is avoided in the interesting range of parameters , @xmath185 , @xmath186 , @xmath187 .
table [ tab1 ] shows the values of @xmath188 for randomly chosen @xmath44 sites @xmath189 .
the monte carlo implementation for the fermion bag approach , including three steps of configuration updating and sampling of chiral condensate and fermion density , depends on the ratio of the value of @xmath48 . for small coupling constant @xmath15
, there are few bonds connecting neighboring sites , we use the propagation matrix on the occupied sites to calculate this ratio ; for large coupling constant @xmath15 , many bonds connecting to neighboring sites appear , we use the fermion matrix on the free sites to calculate this ratio .
thus high computational efficiency is achieved for the fermion bag approach with weak or strong coupling between fermions . in our code , we use @xmath190 and @xmath191 to denote the fermion matrix @xmath16 and its inverse matrix , respectively , which does not change in the whole run . @xmath192 and
@xmath193 denote the free sites and occupied sites for the current configuration @xmath43 .
denote by @xmath194}$ ] the inverse of @xmath195}$ ] and by @xmath196}$ ] the inverse of @xmath197}$ ] , respectively .
here @xmath195}$ ] ( @xmath197}$ ] ) denotes the submatrix of @xmath190 ( @xmath191 ) with the rows and columns corresponding to @xmath198 ( @xmath193 ) . thus depending on the number of occupied sites , @xmath106 or @xmath199 changes during monte carlo updating .
the initial configuration for @xmath43 is the one where there are no bonds .
the sampling is taken after @xmath200 monte carlo steps , where in each step at most two bonds change . to reduce the autocorrelation effects ,
two subsequent samples are separated by @xmath201 monte carlo steps .
the sampling is finished after @xmath202 monte carlo steps , thus about @xmath203 samples are taken for each monte carlo simulation .
the implementation of the complex langevin dynamics ( [ 2016_5_2_1 ] ) is rather simple .
the initial condition for @xmath134 is zero everywhere .
we choose the time step @xmath204 .
the sampling is taken after @xmath205 ( i.e. , @xmath206 complex langevin steps ) . to compare with the fermion bag approach ,
two subsequent samples are separated by 10 complex langevin steps .
the end time in the complex langevin dynamics is @xmath207 , thus about @xmath208 samples are taken for each complex langevin simulation . to calculate the sampling ( [ 2016_5_4_10 ] ) with the computational complexity @xmath181
, we introduce @xmath209 where @xmath210 , @xmath211 . using the iteration
@xmath212 we know that the computational complexity for calculating @xmath213 and @xmath214 is @xmath181 .
the trace of @xmath159 is @xmath215 and @xmath216 we use the @xmath217 method to estimate the error for the samples in each monte carlo simulation or complex langevin dynamics @xcite .
a typical result is shown in figure [ fig1 ] where the statistical error for the complex langvin dynamics are larger than those for the fermion bag approach if the chemical potential is close to 1 .
[ tab1 ] . the value of @xmath48 for randomly chosen occupied sites @xmath218 , @xmath219 , @xmath220 , @xmath221 . [ cols="^,^",options="header " , ] , @xmath222 , @xmath223 , @xmath224 and @xmath225 .
bottom : @xmath226 , @xmath227 , @xmath228 , @xmath229 and @xmath230 .
, title="fig:",width=302,height=226 ] , @xmath222 , @xmath223 , @xmath224 and @xmath225 . bottom : @xmath226 , @xmath227 , @xmath228 , @xmath229 and @xmath230 . ,
title="fig:",width=302,height=226 ] the exact result is known for the massive thirring model with one flavor in 0 + 1 dimension @xcite .
we first compare the chiral condensate obtained by the langevin dynamics and exact result . since the sign problem is avoided and the observable is positive , the langvin dynamics ( see eq .
( [ 2016_5_4_10])([2016_5_4_11])([2016_5_4_11_1 ] ) ) should reproduce the exact result .
figure [ fig0 ] shows the agreement between the chiral condensate obtained by the langvin dynamics and exact results .
in fact , the agreement between them are also achieved for the parameters : @xmath231 , @xmath232 , @xmath233 .
figure [ fig1 ] shows the comparison between fermion bag approach , complex langevin dynamics with the exact results .
the chiral condensate and fermion density for fermion bag approach agree very well with the exact result in the range @xmath185 .
compared with the fermion bag approach , the result obtain by complex langevin dynamics agree with exact result for small or large @xmath0 .
moreover the statistical error becomes large for the complex langevin dynamics in the intermediate value @xmath0 .
figure [ fig2 ] shows that the fermion bag approach and complex langevin dynamics agree with the exact results very well when @xmath15 is decreased to be @xmath234 .
since the interaction between fermions is decreased , the statistical error is also small compared with those in figure [ fig1 ] .
when @xmath15 increased , the simulation results obtained by complex langevin dynamics is not reliable . according to ref .
@xcite@xcite , the quantity @xmath235 should vanish for any holomorphic function @xmath236 if the complex langevin dynamics works .
we choose the observable ( the chiral condensate ) @xmath237 . for @xmath238 , @xmath239 and @xmath221 ,
it is @xmath240 which is small if @xmath241 and becomes @xmath242 if @xmath243 .
moreover , the statistical error is also large which means that it is difficult to measure it reliably when @xmath15 is increased .
figure [ fig10 ] shows the dependence of the phase @xmath244 on the coupling strength @xmath15 for fixed chemical potential @xmath238 .
here @xmath245 is the phase quenched approximation of @xmath36 in ( [ 2016_4_27_10 ] ) where @xmath246 is replaced by its module @xmath247 .
when @xmath15 is increased the phase @xmath248 decays to zero , which shows it suffers from a severe sign problem .
moreover , the statistic error is also increased if @xmath15 becomes larger . when @xmath249 , the results obtained by the fermion bag approach agree with exact result rather well , but the complex langevin dynamics is totally wrong , which is shown in figure [ fig3 ] .
the failure of the complex langevin dynamics for strong interaction was also found in ref .
for @xmath220 in figure [ fig3 ] , the average number of occupied bonds in the fermion bag approach is 3.41 , which means almost all sites are occupied by bonds @xmath250 .
this leads to a slight difference of the chiral condensate obtained by the fermion bag approach and exact results .
figure [ fig5 ] show that the average number of occupied bonds drops rapidly near @xmath251 where the simulated results obtained by fermion bag approach agree with the exact result . here
we do nt show the error bar in figure [ fig5 ] since it is very small which can be shown for the error bar of the chiral condensate by fermion bag approach in figure [ fig3 ] . for
fixed @xmath239 and @xmath238 , [ fig6 ] shows the dependence of the average number of occupied bonds ( left ) and relative error for chiral condensate ( right ) on the coupling strength @xmath15 . the average number of occupied bonds increase with the coupling strength @xmath15 .
since the exact solution is known , the relative error for the chiral condensate ( obtained by the fermion bag approach ) can be calculated as shown in the right figure of figure [ fig6 ] where the relative error is increased with the coupling strength when @xmath252 or @xmath253 .
we also notice there is an oscillation of the relative error for @xmath254 .
thus if the coupling strength @xmath15 is not too large the fermion bag approach can always reproduce the exact results .
when @xmath15 is very large , almost all lattice sites are occupied , which will lead a slight difference between results obtained by fermion bag approach and the exact results .
it should be mentioned that the fluctuation of the chiral condensate and fermion density in the fermion bag approach is rather small even when @xmath15 is rather large , e.g. , @xmath255 .
for the parameter : @xmath249 , @xmath239 and @xmath221 , the chiral condensate obtained by langevin dynamics in figure [ fig0 ] ( bottom ) and those by fermion bag approach in figure [ fig3 ] ( upper ) can reproduce the exact result .
compared with the fermion bag approach , the equilibrium time in the langevin dynamics is very long .
however , the advantage of the langevin dynamics over the other two approaches is that the chiral condensate and fermion density for any chemical potential can be obtained when the averages @xmath256 and @xmath257 is calculated from the langevin dynamics , since they does not depend on the chemical potential .
the comparison are also made between the fermion bag approach , complex langevin dynamics with exact results in the range of parameters : @xmath231 , @xmath232 , @xmath233 .
the chiral condensate and fermion density obtained by the fermion bag approach agrees very well with the exact results for all these parameters .
the reason is that the weight in the fermion bag approach is nonnegative and thus the sign problem is avoided .
the results obtained by complex langevin dynamics agree with the exact results only if @xmath15 is not too large , @xmath258 .
when @xmath15 is increased , compared with the drift term , the white noise fluctuation in complex langevin equation becomes large and thus the complex langevin does not work , especially , when the fermion mass @xmath18 , e.g. , @xmath219 , is rather small .
on the fermion coupling strength @xmath259 , where @xmath245 is the phase quenched approximation of @xmath36 in ( [ 2016_4_27_10 ] ) , @xmath260 , @xmath261 , @xmath262 . ,
width=302,height=226 ] by fermion bag approach . ,
width=302,height=226 ] by fermion bag approach .
, width=226,height=151 ]
the fermion bag approach , langevin dynamics and complex langevin dynamics are used to solve the massive thirring model at finite density in 0 + 1 dimension .
the chiral condensate and fermion density by these methods are compared with the exact results .
the complex langevin dynamics only works for not too large interaction @xmath258 .
moreover , this method will also meet the computational difficulties when the fermion mass is too small , or the chemical potential is in the intermediate range @xmath238 . since the sign problem is avoided in the langevin dynamics and the fermion bag approach , these two methods can reproduces the exact results for a large range of parameters .
another advantage of the fermion bag approach over the langevin dynamics is that a high computational efficiency are made for the strong interaction between fermions for the fermion bag approach . in the future paper
we will check these advantages of the fermion bag approach over the other numerical methods for the massive thirring model at finite density in 2 + 1 dimensions .
i would like to thank prof .
shailesh chandrasekharan for discussion .
daming li was supported by the national science foundation of china ( no .
11271258 , 11571234 ) . 00
g. aarts , f.a .
james , e. seiler and i .- o .
stamatescu , adaptive stepsize and instabilities in complex langevin dynamics , phys .
b 687 ( 2010 ) 154 , http://arxiv.org/abs/0912.0617[arxiv:0912.0617[hep-lat ] ] g. aarts , can stochastic quantization evade the sign problem ? the relativistic bose gas at finite chemical potential , phys .
102 ( 2009 ) 131601 , http://arxiv.org/abs/0810.2089[arxiv:0810.2089[hep-lat ] ] j. m. pawlowski and c. zielinski , thirring model at finite density in 2 + 1 dimensions with stochastic quantization , phys .
d 87 ( 2013 ) 094509 , http://arxiv.org/abs/1302.2249[arxiv:1302.2249[hep-lat ] ] j. flower , s. w. otto , and s. callahan , complex langevin equations and lattice gauge theory , phys .
d 34 , 598 ( 1986 ) .
d. sexty , simulating full qcd at nonzero density using the complex langevin equation , http://arxiv.org/abs/1307.7748[arxiv:1307.7748[hep-lat ] ] g. aarts and i .- o . stamatescu , stochastic quantization at finite chemical potential , jhep 09 ( 2008 ) 018 , http://arxiv.org/abs/0807.1597[arxiv:0807.1597[hep-lat ] ] g. aarts , f. a. james , j. m. pawlowski , e. seiler , d. sexty and i. -o .
stamatescu , stability of complex langevin dynamics in effective models , jhep 1303 ( 2013 ) 073 , http://arxiv.org/abs/1212.5231[arxiv:1212.5231[hep-lat ] ] g. aarts and f. a. james , complex langevin dynamics in the su(3 ) spin model at nonzero chemical potential revisited , jhep 01 ( 2012 ) 118 , http://arxiv.org/abs/1112.4655[arxiv:1112.4655[hep-lat ] ] g. aarts , l. bongiovanni , e. seiler , d. sexty and i .- o .
stamatescu , controlling complex langevin dynamics at finite density , eur .
j. a 49 ( 2013 ) 89 , http://arxiv.org/abs/1303.6425[arxiv:1303.6425[hep-lat ] ] g. aarts and f. a. james , on the convergence of complex langevin dynamics : the three - dimensional xy model at finite chemical potential , jhep 1008 , 020 ( 2010 ) , http://arxiv.org/abs/1005.3468[arxiv:1005.3468[hep-lat ] ] j. m. pawlowski and c. zielinski , thirring model at finite density in 0 + 1 dimensions with stochastic quantization : crosscheck with an exact solution , phys.rev .
d87 , 094503 ( 2013 ) , http://arxiv.org/abs/1302.1622[arxiv:1302.1622[hep-lat ] ] h. w. hamber and h. ren , complex probabilities and the langevin equation , phys .
b 159 , ( 1985 ) 330 j. flower , s. w. otto , and s. callahan , complex langevin equations and lattice gauge theory , phys . rev .
d 34 , ( 1986 ) 598 e. -m .
ilgenfritz , complex langevin simulation of chiral symmetry restoration at finite baryonic density , phys .
b 181 , ( 1986 ) 327 g. aarts , f.a . james , e. seiler and i .- o .
stamatescu , eur .
j. c 71 ( 2011 ) 1756 [ arxiv:1101.3270 ] http://arxiv.org/abs/1101.3270[arxiv:1101.3270[hep-lat ] ] c. gattringer and t. kloiber , lattice study of the silver blaze phenomenon for a charged scalar @xmath263 field , nucl .
b 869 , 56 ( 2013 ) , http://arxiv.org/abs/1206.2954[arxiv:1206.2954[hep-lat ] ] y. d. mercado and c. gattringer , monte carlo simulation of the su(3 ) spin model with chemical potential in a flux representation , nucl . phys .
b 862 , 737 ( 2012 ) , http://arxiv.org/abs/1204.6074[arxiv:1204.6074[hep-lat ] ] y. d. mercado , c. gattringer , a. schmidt , surface worm algorithm for abelian gauge - higgs systems on the lattice , comput .
commun . 184
( 2013 ) 1535 , http://arxiv.org/abs/1211.3436[arxiv:1211.3436[hep-lat ] ] c. gattringer , t. kloiber and v. sazonov , solving the sign problems of the massless lattice schwinger model with a dual formulation , nucl .
b 897 ( 2015 ) 732 , http://arxiv.org/abs/1502.05479[arxiv:1502.05479[hep-lat ] ] f. bruckmann , c. gattringer , t. kloiber and t. sulejmanpasic , finite density o(3 ) non - linear sigma model and low energy physics , http://arxiv.org/abs/1512.05482[arxiv:1512.05482[hep-lat ] ] f. bruckmann , c. gattringer , t. kloiber and t. sulejmanpasic , dual lattice representations for o(n ) and cp(n-1 ) models with a chemical potential , physics letters b 749 , 495 ( 2015),http://arxiv.org / abs/1507.04253[arxiv:1507.04253[hep - lat ] ] u. wolff , simulating the all - order strong coupling expansion iii : o(n ) sigma / loop models , nucl . phys .
b 824 ( 2010 ) 254 , http://arxiv.org/abs/0908.0284[arxiv:0908.0284[hep-lat ] ] .
u. wolff , nucl .
b 834 ( 2010 ) 395 ( erratum ) .
u. wolff , simulating the all - order strong coupling expansion iv : cp(n-1 ) as a loop model , nucl .
b 832 ( 2010 ) 520 , http://arxiv.org/abs/1001.2231[arxiv:1001.2231[hep-lat ] ] s. chandrasekharan , fermion bag approach to fermion sign problems , new opportunities in lattice fermion field theories , eur . phys .
j a49 , 1 ( 2013 ) , http://arxiv.org/abs/1304.4900v1 [ arxiv:1304.4900[hep - lat ] ] s. chandrasekharan and a. li , the generalized fermion - bag approach , pos , lattice2011 , 058 ( 2011 ) , http://arxiv.org/abs/1111.5276[arxiv:1111.5276[hep-lat ] ] s. chandrasekharan and a. li , fermion bag solutions to some sign problems in four - fermion field theories , pos , lattice2012 , 225 ( 2012 ) , http://arxiv.org/abs/1202.6572[arxiv:1202.6572[hep-lat ] ] s. chandrasekharan , solutions to sign problems in lattice yukawa models , phys .
d. , 86 , 021701 , http://arxiv.org/abs/1205.0084[arxiv:1205.0084[hep-lat ] ] r. oeckl and h. pfeiffer , the dual of pure non - abelian lattice gauge theory as a spin foam model , nucl .
phys.b 598:400 - 426 , 2001 , http://arxiv.org/abs/hep-th/0008095[arxiv:hep-th/0008095 ] j. w. cherrington , d. christensen , i. khavkine , dual computations of non - abelian yang - mills on the lattice , phys .
d 76 ( 2007 ) 094503 , http://arxiv.org/abs/0705.2629[arxiv:0705.2629[hep-lat ] ] j. w. cherrington , a dual algorithm for non - abelian yang - mills coupled to dynamical fermions , nucl .
b794 , 195 ( 2008 ) , http://arxiv.org/abs/0710.0323[arxiv:0710.0323[hep-lat ] ] h. pfeiffer , exact duality transformations for sigma models and gauge theories , j. math . phys .
53 , 033501 ( 2012 ) ; u. wolff , monte carlo errors with less errors , comput .
commun . 156
( 2004 ) 143 - 153 , http://arxiv.org/abs/hep-lat/0306017[arxiv:hep-lat/0306017 ] l. g. molinari , determinants of block tridiagonal matrices , linear algebra and its applications 429 ( 2008 ) 2221 , http://arxiv.org/abs/0712.0681[arxiv:0712.0681[math-ph ] ]
the @xmath132 ( even @xmath6 ) fermion matrix is @xmath264 according to a formula of the determinant @xcite , the determinant of @xmath16 is @xmath265 where the @xmath158 transfer matrix @xmath159 is @xmath266 .
obviously , @xmath267 for any @xmath268 and @xmath269 .
choose @xmath270 different indices , @xmath271 and delete @xmath270 rows and columns corresponding to these @xmath270 indices from @xmath16 to obtain @xmath272 .
we want to prove that @xmath273 matrix @xmath274 satisfies @xmath275 .
this holds because the structure of @xmath272 is the same with @xmath16 and thus the determinant of @xmath274 can be calculated @xcite , which must be positive .
for example , @xmath276 , @xmath277 , @xmath278 , @xmath279 , @xmath280 since @xmath60 can be presented by the determinant of the submatrix of @xmath16 , which is nonnegative , the sign problem is avoided for the presentation of @xmath36 by fermion bag approach in ( [ 2016_4_27_2 ] ) . | we consider the massive thirring model at finite density in 0 + 1 dimension .
the fermion bag approach , langevin dynamics and complex langevin dynamics are adopted to attack the sign problem for this model .
compared with the complex langevin dynamics , both fermion bag approach and langvin dynamics avoid the sign problem . the fermion density and chiral condensate , which are obtained by these numerical methods , are compared with the exact results . the advantages of the fermion bag approach over the other numerical methods are also discussed .
thirring model , finite density , complex langevin dynamics , fermion bag approach 05.50.+q , 71.10.fd , 02.70.ss |
a key technological hurdle in the study of single molecules in the condensed phase is to achieve a tight focus in the cryostat . here
we show that this problem can be addressed by exploiting the solid - immersion lens ( sil ) technology @xcite . as shown in fig .
[ setup]b , a molded glass aspheric lens with @xmath13 and a working distance of 1.76 mm focused the laser beam onto the sample , and a second identical lens was used to re - collimate it in the forward direction .
a cubic zirconia half - ball lens with a diameter of 3 mm and a refractive index of about 2.18 was used as a sil .
piezoelectrically driven slip - stick sliders were used to adjust the lateral and axial positions of the aspheric lenses with respect to the sil .
[ setup]d shows a confocal fluorescence excitation scan of a single molecule .
we achieved a focal spot with a full width at half - maximum of about 370 nm at @xmath14 .
although this still does not meet the diffraction limit , it constitutes a great improvement over previous cryogenic far - field resolutions and allowed us to perform several new experiments , as described in the following .
figure [ setup]c sketches the experimental arrangement . a tunable dye laser with a linewidth of 1 mhz was used to excite the molecules .
the laser beam could be scanned across the sample with the aid of galvanometric mirrors .
a berek variable waveplate was used to pre - compensate for a slight elliptical polarization that was accumulated along the optical path and to ensure linear polarization at the position of the molecule .
the samples consisted of a shpolskii matrix doped with dibenzanthanthrene ( dbatt ) molecules @xcite . to prepare a sample ,
a small amount of dbatt was dissolved in _
n_-tetradecane , applied to the flat side of the solid immersion lens and then squeezed and held in place by a glass cover slip . the 0 - 0 vibronic transition of dbatt in _
n_-tetradecane is at a wavelength of @xmath15 nm .
we routinely measured near to natural lifetime limited linewidths of 17 - 20mhz consistent with our independent measurements of 9.7 ns for the excited state fluorescence lifetime .
another favorable feature of dbatt is a very short triplet state lifetime @xcite .
figure [ setup]c shows three detection channels that were used for various measurements discussed in this article .
the avalanche photodiode pd1 recorded the intensity of the transmitted light close to the excitation wavelength @xmath2 by using a short - pass filter .
the second detector ( pd2 ) was situated after a long - pass filter and registered the stokes shifted fluorescence on transitions @xmath16 . a third detector ( pd3 )
was used in combination with other optical elements to record the resonance fluorescence intensity in reflection . in each experimental run , single molecules were first identified and characterised on pd2 using conventional fluorescence excitation spectroscopy .
after optimization of the lateral focus position ( see fig .
[ setup]d ) , the polarization of the excitation light was aligned with the dipole moment of the molecule .
the molecules that were selected for the experiments presented here showed no significant axial component in the orientation of their dipole moments and fluoresced with nearly perfect linear polarization .
an example of a raw transmission spectrum , revealing a @xmath17 dip .
the integration time per pixel was 160 ms and the noise amounts to @xmath18.,width=8 ] figure [ dip ] shows an example of the raw transmission spectra recorded on pd1 with the above - mentioned approach .
a dip of @xmath17 is observed directly in transmission when a single molecule is resonant with the laser beam .
the noise level amounts to 0.7% , corresponding to the shot - noise limit of the incident intensity over an integration time of 160 ms per pixel .
we emphasize that no noise suppression technique such as lock - in detection was employed in this work . although one might be inclined to attribute the observed dip to absorption " , the excitation spectrum @xmath19 registered on the detector has to take into account the interference between the excitation field @xmath20 and the field @xmath21 scattered by the molecule @xcite according to @xmath22 in our recent work using snom @xcite we expressed the detector signal in detail , accounting for the modal evolution of the incident and scattered fields and for the spectroscopic properties of the molecule . here
we simplify the full expression from ref .
@xcite in the form @xmath23 .
\label{f - g}\end{aligned}\ ] ] where @xmath24 is a lorentzian line profile , and @xmath25 denotes the laser frequency detuning .
the quantities @xmath26 and @xmath27 stand for the full width at half - maximum values of the homogeneous and natural linewidths of the zero - phonon component of the 0 - 0 transition , respectively .
the parameter @xmath28 is the rabi frequency , where @xmath29 signifies the dipole transition operator .
@xmath30 and @xmath31 are positive coefficients whereas @xmath32 signifies the phase difference between the excitation and the coherently scattered light components .
equation ( [ f - g ] ) describes the spectrum that one registers on a detector as one scans the laser frequency .
the first term is simply the part of the incident intensity that arrives at the detector .
the second term proportional to @xmath30 represents the molecular emission and is always positive . in total , it consists of the narrow - band resonance fluorescence at @xmath2 @xcite , the phonon wing of this transition caused by the vibrational coupling of the molecule with the matrix , and the stokes - shifted fluorescence emitted at @xmath8 . in the signals detected on pd1 and pd3 ,
however , the latter is filtered out . the third term in this equation proportional to @xmath31
denotes the interference between @xmath20 and the coherent part of @xmath21 . in refs .
@xcite , we showed various examples where @xmath32 could be varied , resulting in dispersive transmission spectra . in the weak excitation limit , the transmission signal on detector pd1 is dominated by the interference component , also known as the _ extinction _ term .
we emphasize that the observed interference effect depends on the mode overlap between the molecular emission and the incident beam and can be influenced by the choice of the detection solid angle . in our case
, total internal reflection at the asymmetric geometry of the sil - sample - glass - air layers can lead to a fine structure in the reflected beam @xcite and take away energy from the transmitted beam .
this arrangement also results in a strongly asymmetric molecular emission pattern which favors emission into the sil side by about ten times @xcite .
furthermore , for a given geometry the spectroscopic features of the emitter play an imporant role .
the emission associated with the narrow - band zero - phonon line of the 0 - 0 vibronic transition is lowered by the debye - waller and franck - condon factors @xmath33 and @xmath34 , which amount to 0.25 and 0.3 in our system @xcite , respectively .
thus , as compared to a perfect two - level system in the same geometric configuration , our molecule has had @xmath35 times weaker coherent interaction with the laser beam .
in addition to providing a strong detection signature , direct extinction spectroscopy offers the great advantage that it gives access to the narrow - band emission on the zero - phonon line .
the radiation of a two - level system at the wavelength of the excited optical transition is known as resonance fluorescence . in case of weak resonant excitation and in the absence of dephasing , this emission is purely elastic and coherent @xcite . as the incident intensity
is raised , this component first grows linearly , but it reaches a maximum at saturation parameter @xmath36 . under strong illumination ,
the population of the excited state undergoes rabi oscillations at frequency @xmath37 and the emitted field is contaminated by an inelastic incoherent component @xcite .
the coherent part of the resonance fluorescence is predicted to decrease slowly according to the text - book formula @xmath38 , shown by the solid black curve in fig .
[ coh - incoh ] . .
the solid curves are the results of the theoretical calculations according to text - book formulae discussed in the text.[coh - incoh],width=8 ] the experimental assignment of the coherent and incoherent parts of resonance fluorescence is a nontrivial task because for a given excitation intensity , the detected spectrum contains both .
the lorentzian peak of the @xmath30-term in eq .
( [ f - g ] ) , receives contributions from both components .
however , only the coherent emission can interfere with the incident light to produce the @xmath31-term , which is in turn composed of a lorentzian part and a dispersive signal .
thus , one can not decipher the variation of the coherent fluorescence with the excitation intensity in an unambiguous manner by simply monitoring the dip in fig .
below we show that it is possible to do this by influencing the individual components in a controlled manner . )
is taken to be along the y - axis whereas the molecular dipole moment ( @xmath39 ) and the qwp are set at @xmath40 and @xmath41 from this axis , respectively .
b , c ) two examples of the recorded spectra are shown by the symbols .
the red curves are the fits to the experimental data while the blue and the green lines plot the contributions of the molecular fluorescence intensity and extinction components respectively , as extracted from the series of measurements described in the text .
[ polarization],width=8 ] as illustrated by fig .
[ polarization]a , for this experiment we have introduced a quarter waveplate and a polarizer in the detection path of pd1 at the position marked by the dotted red box in fig . [ setup]c .
we set @xmath20 at @xmath42 with respect to the transition dipole moment of the molecule and at @xmath43 with respect to the axis of the polarizer . after traversing the quarter waveplate ,
the amplitudes and phases of @xmath20 and @xmath21 have evolved differently , thus changing the relative phase @xmath32 and the coefficients @xmath30 and @xmath31 . when rotating the quarter waveplate , both the strength and the shape of the @xmath31-term in eq .
( 2 ) are altered , but the lorentzian peak of the @xmath30-term only changes its magnitude . by recording and analyzing a series of spectra
, we were able to identify each component in a unique manner .
the symbols in figs .
[ polarization]b and c display two examples of such an analysis .
the red curve is the fit to the experimentally recorded spectrum .
the blue and the green spectra show the molecular fluorescence intensity and the extinction components that were extracted from the total signal , respectively . in case of fig .
[ polarization]b , the addition of the @xmath30-term has resulted in a final asymmetric dispersive shape whereas in fig .
[ polarization]c , the peak of @xmath44 associated with the @xmath30-term has reduced a @xmath45 dip of the @xmath31-term to yield a total observable dip of @xmath46 . from the knowledge of the extinction @xmath31-term and @xmath4 we can calculate the absolute number of the coherently scattered photons for a given excitation intensity . to measure the intensity dependence of the coherent resonance fluorescence , we have varied the excitation intensity between 5 pw and 10 nw .
the black symbols in fig .
[ coh - incoh ] show the power of the coherent emission of a single molecule , extracted according to the procedure described above .
we obtain an excellent agreement between the measured and the theoretical data ( black solid curve ) without using any free parameter .
the red dots in fig .
[ coh - incoh ] show the conventional saturation curve obtained from the fluorescence excitation signal recorded simultaneously on pd2 as a direct measure of the excited state population , also in excellent agreement with the function @xmath47 displayed by the solid red curve .
we note that in our work @xmath48 was determined independently by analyzing the measured linewidths of fluorescence excitation spectra , whereby @xmath49 corresponds to a power broadened fwhm linewidth of @xmath50 .
the coherent part of resonance fluorescence has been previously put to evidence using a heterodyne measurement on the emission of a single ion @xcite . however , to the best of our knowledge , a study of its intensity dependence presented in fig .
[ coh - incoh ] has not been reported before .
these measurements are particularly challenging in the solid state because of the omnipresence of background scattering at the excitation wavelength .
the key to success in our experiment has been the efficient coupling of the laser light to the molecule , allowing us to optimize the ratio between the molecular signal and the background scattering .
this is also nicely reflected in the fact that the onset of saturation at @xmath49 could be achieved at the very low incident laser power of 350 pw .
the data in fig . [ coh - incoh ] clearly show that the coherent resonance fluorescence diminishes under strong excitation .
this means that the interference between the laser beam and the molecular emission is reduced , and the dip in the transmission spectra recorded on pd1 gradually disappears . in parallel to this process , the inelastic component of the resonance fluorescence is expected to increase and give rise to a triplet spectrum as first described by b. mollow in 1969 @xcite .
the so - called mollow triplet has been indeed examined in various gaseous systems @xcite , but its direct observation has turned out to be elusive in the solid state . to record and study the incoherent part of the spectrum , we used the detection path in reflection that leads to pd3 in fig .
[ setup]c .
the fluorescence was sent through a short - pass filter to cut out the stokes shifted fluorescence corresponding to the @xmath16 transitions , while transmitting light at @xmath2 and part of the broad phonon wing of the zero phonon line . here
the angle between @xmath20 and that of the molecular dipole was set to @xmath42 and the analyzing polarizer was aligned at @xmath51 , filtering out the background scattering of the laser light with a ratio of 300:1 .
the remaining fluorescence was coupled into a multimode optical fiber with a core diameter of @xmath52 for convenience of handling .
the output of the fiber was sent to a home - built fabry - perot cavity ( fpc ) with a free spectral range of 356 mhz and transmission of approximately @xmath53 . as shown by the lowest spectrum in fig .
[ mollow]a , an instrumental linewidth of about 14 mhz was measured when the reflected laser light was coupled to the resonator . despite polarization filtering of the fluorescence ,
the background laser scattering was of the same order of magnitude as the molecular fluorescence and was exploited for calibrating the frequency axis of each fpc scan .
extracted from the fit ( red curves ) to the @xmath54 data is indicated in each figure together with the corresponding saturation parameter @xmath48 .
[ mollow],width=8 ] figures [ mollow]b - h show a series of emission spectra recorded by exciting a molecule on resonance , collecting its fluorescence close to the same wavelength in the fashion described above and then scanning the fpc several times , equivalent to an integration time of 3 s per scan pixel . as the excitation intensity was raised ( moving upward in the figure ) , side bands appeared in the molecular emission spectrum with increasing frequency separations .
the appearance of the mollow triplet can be understood as the result of the periodic modulation of the density matrix elements and thus the molecular polarization at the rabi frequency . to obtain an independent measure of the rabi frequency , we split the stokes - shifted fluorescence beam on pd2 via a beam splitter onto two photodetectors as sketched in the blue dotted box in fig .
[ setup]c . by using the common start - stop scheme @xcite , we recorded the intensity correlation function @xmath54 @xcite for each measurement presented in figs .
[ mollow]b - h .
the results are plotted in figs .
[ mollow]j - p .
figure [ mollow]i confirms that as expected , we observed pure antibunching at low intensities .
however , as the excitation was made stronger , @xmath54 underwent oscillations @xcite . by fitting the curves in figs .
[ mollow]i - p , we determined the rabi frequencies @xmath37 in each case .
these values together with the corresponding saturation parameters @xmath48 are indicated in figs .
[ mollow]j - p . using the independently measured values of @xmath37 , the laser background and the fpc line profile ( see fig .
[ mollow]a ) , we then calculated the theoretical spectra for the mollow triplet shown by the red curves in figs .
[ mollow]b - h .
the agreement between theory and experiment is excellent with only one scaling parameter for the signal intensity . to our knowledge , this constitutes the first observation of the mollow triplet from a single emitter in the solid state , further confirming the potential of these systems for well - understood and controlled quantum optical experiments .
, width=8 ] commonly a high flux of photons is required in spectroscopy to achieve a single excitation because either the observed transitions are strongly broadened or the light beam is focused very weakly . moreover ,
a certain minimum excitation rate is required to obtain fluorescence count rates larger than the square root of the detector noise level .
a remarkable advantage of coherent detection of single molecules , however , is that the detected intensity can easily be larger than the detector noise even if the molecular emission is much weaker .
this is illustrated in fig .
[ few - photon ] where a single molecule is detected under an incident power of merely 550 photons per second , equivalent to 150 attowatt .
the transmission spectrum shows the raw data integrated over 4 seconds at each pixel . when analyzing this result in the context of an interference phenomenon
, we find that the dip of 50 counts per second originates from a detected molecular emission of only 1.1 photons per second , which is about 10 times lower than the noise associated with the 150 dark counts of our detector .
thus , the high signal contrast of coherent extinction measurements is ideally suited for detecting weak emitters such as single rare earth ions , especially if one would exploit additional improvements in the snr via lock - in detection .
the ability to detect a single emitter with an ultra - faint illumination also shows the feasibility of exciting a single molecule by the emission of a second molecule acting as a source of fourier - limited single photons @xcite . considering that the fluorescence of a single molecule typically amounts to about @xmath55 detected photons per second
, we can easily expect a few thousands of photons per second at the position of the second molecule after various diagnostics and handling optics .
indeed , calculations show that in the case where an emitter is illuminated by a single photon gaussian wavepacket of width @xmath27 , there is a higher than @xmath56 chance for it to act as a @xmath57-pulse to coherently transfer the population of the ground state to the excited state @xcite or to experience a substantial phase shift via its interaction with the molecule .
furthermore , it has been suggested that the molecule can act as a nonlinear medium to couple two ultra - faint photon fields impinging on it at the same time @xcite .
by confining a laser beam close to the diffraction limit at t=1.4 k , we have examined for the first time , the _ coherent _ interaction of a single organic molecule with light over 9 orders of magnitude of excitation power . in the weak excitation regime , we have achieved a large extinction dip of @xmath17 and have demonstrated the detection of a single molecule with an ultra - faint laser source of 150 attowatt . in the strong excitation regime ,
our measurements have allowed us to distinguish the coherent and incoherent parts of resonance fluorescence and to study their intensity dependencies .
furthermore , we could directly record the mollow triplet as a landmark of the nonlinear interactions between an emitter and laser light . from a fundamental point of view ,
an intriguing question that arises is whether one could achieve a perfect coupling between a _
freely _ propagating optical mode and a dipolar emitter in a single - pass encounter , i.e without high - finesse cavities .
when considering the interference nature of the extinction signal , it is clear that the annihilation of a photon in the forward direction requires perfect overlap between its mode and that of a dipolar radiator . clearly , this condition is not satisfied for a strongly focused @xmath58 laser beam @xcite .
however , it has been proposed that the combination of an emitter with an axially oriented transition dipole moment @xcite and light with a radially polarized doughnut mode would be favorable @xcite .
another exciting possibility might be to use nano - antennae for engineering the radiation pattern of the molecule and matching it to the excitation mode @xcite . given the achievements presented in this article and the simplicity of the sil technology , we anticipate that our approach opens doors to the realization of these ideas and a wide range of other investigations on the coherent and nonlinear interaction of light with single molecules and ions in the condensed phase .
we are grateful to a. renn for experimental support .
this work was financed by the schweizerische nationalfond and the eth zurich initiative for quantum systems for information technology ( qsit ) .
+ the authors declare that they have no competing financial interests .
+ correspondence and request for materials should be addressed to v.s .
( email : [email protected] ) . | single dye molecules at cryogenic temperatures display many spectroscopic phenomena known from free atoms and are thus promising candidates for fundamental quantum optical studies .
however , the existing techniques for the detection of single molecules have either sacrificed the information on the coherence of the excited state or have been inefficient . here
we show that these problems can be addressed by focusing the excitation light near to the absorption cross section of a molecule .
our detection scheme allows us to explore resonance fluorescence over 9 orders of magnitude of excitation intensity and to separate its coherent and incoherent parts . in the strong excitation regime
, we demonstrate the first observation of the mollow triplet from a single solid - state emitter . under weak excitation we report the detection of a single molecule with an incident power as faint as 150 attowatt , paving the way for studying nonlinear effects with only a few photons .
single dye molecules embedded in solids have been shown to behave as quantum mechanical two - level systems at cryogenic temperatures , exhibiting phenomena such as antibunching @xcite , ac stark shift @xcite and various nonlinear effects @xcite known from atoms .
the pioneering detection of single molecules was achieved via absorption spectroscopy assisted by a double lock - in method @xcite . in the weak excitation regime ,
the effect of a plane - wave illumination on a molecule with a transition between levels @xmath0 and @xmath1 at wavelength @xmath2 can be formulated as @xmath3 . here
@xmath4 is the detected intensity of the excitation beam after having encountered the emitter , @xmath5 is the absorption cross section of the transition , and @xmath6 and @xmath7 are the area and intensity of the excitation beam , respectively .
the experimental realization of this simple scenario is challenging because the effect of a single molecule has to be registered on top of the noise associated with a considerable laser power and scattering background . to improve the signal - to - noise ratio ( snr ) in single molecule detection , fluorescence excitation spectroscopy
was introduced @xcite . in this technique
the broad stokes shifted fluorescence at @xmath8 is detected on a very low background after filtering out light at @xmath2 ( see fig .
[ setup]a ) .
unfortunately however , the impressive snr of this method comes at the cost of losing the information on the coherences of the transition at @xmath2 .
two recent reports have shown that it is also possible to detect single molecules coherently @xcite via interference between the excitation laser beam and the light scattered by the molecule . in the first scheme , the coherent scattering and a fraction of the reflected excitation beam
were detected through a @xmath9 pinhole @xcite , resulting in a ratio of the order of @xmath10 between the excitation and detected intensities . in the second experiment , cryogenic scanning near - field optical microscopy ( snom )
was employed to confine the excitation light to a subwavelength aperture with a typical transmission of @xmath11 @xcite .
the great shortcoming of these methods is that more than 100,000 photons are required to detect one excitation . in view of quantum optical operations between photons and emitters
, it would be highly desirable to approach a point where as many of the incident photons as possible interact with a single emitter .
such a regime would open the door to a wealth of nonlinear interactions between a single emitter and a single or few photons , which have been so far only achieved using sophisticated high finesse microcavities @xcite . in this work ,
we show that it is possible to reach an efficient coupling between a single emitter and a _ freely _ propagating laser beam in a single - pass encounter if one reduces @xmath6 close to the diffraction limit @xcite .
a ) the energy level scheme of a molecule with ground state @xmath0 and excited state @xmath1 .
manifold @xmath12 shows the vibrational levels of the electronic ground state .
b ) the arrangement of the lenses in the illumination and collection paths in the cryostat .
c ) the schematics of the optical setup .
the dotted boxes are put in place for specific measurements discussed in the text .
abbreviations not explained in the text are bs : beam splitter , sp : short - pass filter , lp : long - pass filter , and pol : polarizer .
d ) a laser scan image of a single molecule , showing a full width at half - maximum focus spot of about 370 nm.,width=8 ] |
experimental information on the partonic structure of the photon can be obtained from the data taken at the hera ep collider experiments .
leading order ( lo ) qcd predicts that photon interactions have a two - component nature . in direct
photon processes the entire momentum of the photon takes part in the hard subprocess with a parton from the proton whereas in resolved photon processes the photon acts as a source of partons and one of these enters the hard subprocess . by measuring inclusive dijet events ( two or more jets ) information on the structure of the real photon can be extracted . by also measuring the scattered electron ,
information can be obtained on the evolution of this structure as a function of the virtuality of the photon , q@xmath0 .
the fraction of the photon s four momentum which enters the hard subprocess at leading order , denoted by @xmath1 , is equal to unity for direct processes , and less than unity for resolved processes .
experimentally it is not possible to measure @xmath1 directly . instead , an observable quantity @xmath2 is defined which is calculable and well - defined to all orders of perturbation theory .
@xmath2 is the fraction of the photon momentum manifest in the two highest transverse energy jets and is defined by the equation @xmath3 where @xmath4 is the transverse energy of jet @xmath5 , @xmath6 is the pseudorapidity of the jet measured in the lab frame , and @xmath7 is the inelasticity of the event .
`` direct enriched '' events are defined as being those with @xmath8 and `` resolved enriched '' events as those with @xmath9 .
this value gives the optimal separation of the leading order direct and resolved event classes .
the kinematic selection cuts made in order to examine the structure of the real photon are * two or more jets ( @xmath10 clustering algorithm ) * q@xmath11 gev@xmath0 * @xmath12 gev , @xmath13 gev * @xmath14 * @xmath15 the advantages of using high @xmath16 dijets are that they provide a hard scale where perturbative qcd ( pqcd ) is expected to work , the hadronisation corrections are small , and the effect of underlying events is small . given these assumptions , the data can be compared directly to nlo pqcd calculations without the need to simulate some hadronisation model .
this analysis concentrates on the high @xmath17 region where the quark densities in the photon are not strongly constrained by @xmath18 experiments .
the measured zeus data is compared to nlo pqcd calculations for three photon pdfs , grv - ho @xcite , afg - ho @xcite and gs96-ho @xcite .
the nlo calculations have been performed by three groups of theorists , harris et al .
@xcite , klasen et al .
@xcite and frixione et al .
since the agreement between the calculations is excellent only one of the calculations is plotted .
figure [ et_fully_both_4b ] shows the differential dijet cross section as a function of @xmath19 for 0.20 @xmath20 0.85 .
there is an excess in the data above theory for central jets ( 0 @xmath21 1 ) below an @xmath22 of 25 gev .
the assumption that the hadronisation corrections are negligible is not true for backward jets ( -1 @xmath21 0 ) , so no conclusions are drawn about the backward region , rather the discrepancy between data and theory is ascribed to a theoretical uncertainty .
figure [ eta_highy_both_32 ] shows the cross section for a narrower range in @xmath7 .
this provides a better sensitivity to the photon structure since the cross section is no longer averaged over the broader @xmath7 range .
there is an excess in the data seen for central ( 0 @xmath21 1 ) and forward ( 1 @xmath21 2 ) jets , both for the whole @xmath2 range and the high @xmath2 range .
the excess in the data above theory for jets with @xmath23 gev and for central and forward rapidities suggests that the available photon pdf parametrisations , based on @xmath18 @xmath24 data , are too small in this kinematic regime .
the kinematic selection cuts made for the analysis of virtual photons are three different q@xmath0 regions are available for measurement .
q@xmath30 gev@xmath0 corresponds to quasi - real photons , where the electron is not measured , and the large bulk of such events lead to a median q@xmath0 of 0.001 gev@xmath0 .
zeus has a small angle electron detector for tagging events in the transition region between photoproduction and dis ( @xmath25 gev@xmath0 ) . for @xmath31 gev@xmath0 the electron
is detected in the zeus main calorimeter .
figure [ plotb ] shows the ratio of the resolved to the direct dijet cross sections as a function of q@xmath0 . since some of the systematic uncertainties cancel in the ratio , this is a more precise measurement than an absolute differential cross section .
the data fall with increasing q@xmath0 .
this is compared to the dijet ratio obtained with herwig 5.9 @xcite using a photon pdf which does not evolve with q@xmath0 ( grv lo @xcite ) and one which does ( sas1d @xcite ) .
the herwig ratio with grv lo is flat , while that obtained with sas 1d falls with increasing q@xmath0
. therefore the fall of the ratio with q@xmath0 observed in the data indicates that the data require a virtual photon pdf which evolves with photon virtuality .
the lepto 6.5.1 @xcite curve shows the prediction for leading order direct processes in dis only .
note that this ratio is non - zero as may be understood given that lepto contains higher - order processes in the approximation of parton - showering and hadronization .
the data approach this direct only limit , however resolved processes are still required in order to describe the data up to at least q@xmath0 = 4.5 gev@xmath0 . | measurements sensitive to the structure of both real and virtual photons are presented and compared to theoretical models with various photon parton distribution functions ( pdfs ) .
measurements for real photons show a tendency for the available photon pdfs to be too small to describe the data . for virtual photons , the photon pdf is seen to decrease with increasing photon virtuality . in order to describe the data , resolved photon processes are required up to a photon virtuality of at least 4.5 gev@xmath0 .
glas - ppe/199906may 1999 |
let @xmath9 , @xmath10 , be an isometric immersion of a riemannian manifold with nonnegative sectional curvature .
a. weinstein observed that , in this case , the curvature operator @xmath114 is nonnegative .
indeed , if @xmath115 is the second fundamental form , the gauss equations imply that @xmath116 at every @xmath117 .
hence , in the connected subset @xmath118 intersected with the unit circle all points have distance at most @xmath119 .
this implies that this set lies in the first quadrant with respect to some orthonormal basis @xmath103 , while it lies in its interior if it has positive sectional curvature .
we thus have the shape operators @xmath120 since @xmath121 . by the gauss equations we have @xmath122 , and hence @xmath123 ( and @xmath124 if the sectional curvature is positive ) .
furthermore , the cone of shape operators which are positive semidefinite contains the first quadrant , while the cone of shape operators which are negative semidefinite contains the third .
we will see that the interesting immersions are quite rigid as a consequence of the following purely algebraic key result which will be applied to the shape operators in .
[ l : eq ] let @xmath104 and @xmath105 be nonnegative semidefinite self
adjoint operators on an euclidean space @xmath125 .
then , @xmath126 moreover , equality holds for all @xmath127 if and only if either @xmath128 , or @xmath129 .
first , observe that the lemma follows easily if one of the operators is nonsingular . indeed ,
if , say , @xmath105 is invertible , @xmath130 , where @xmath131 is the characteristic polynomial of the positive semidefinite operator @xmath132 .
since all the coefficients of @xmath131 are nonnegative the inequality holds , with equality if and only if @xmath133 .
therefore , we can assume that both @xmath104 and @xmath105 are singular .
let @xmath134 , and decompose @xmath104 as its block endomorphisms @xmath135 , according to the orthogonal decomposition @xmath136 , @xmath137 since @xmath138 , we have @xmath139 and @xmath140 . for @xmath141
we get @xmath142 , for all @xmath143 .
hence , @xmath144 and @xmath145 .
therefore either @xmath146 or the kernels intersect and equality holds .
so let us then assume further that @xmath146 and @xmath147 .
if we call @xmath148 the symmetric operator @xmath105 restricted and projected to @xmath149 , then @xmath150 and , for all @xmath151 , @xmath152 thus , @xmath153 where @xmath154 is the symmetric operator on @xmath149 given by @xmath155 .
now , observe that @xmath154 is nonnegative since @xmath156 this gives the inequality in the lemma as in the nonsingular case .
moreover , this computation also shows that @xmath157 if and only if @xmath158 .
thus , @xmath159 is again , up to a positive constant , the characteristic polynomial of the positive semidefinite operator @xmath160 , which is an odd or even function if and only if @xmath161 . in this case ,
@xmath162 , as claimed .
the converse of the equality case is straightforward .
since @xmath123 we can use the following rigidity result , where @xmath163 denotes the lie algebra of the holonomy group of @xmath164 ( see theorem 1.13 in @xcite ) : [ rigidity ] if @xmath164 is a compact and simply connected riemannian manifold with @xmath123 , then one of the following holds : * @xmath165 and @xmath164 is diffeomorphic to @xmath12 , with the ricci flow converging to a metric with constant positive curvature ; * @xmath166 and @xmath164 is diffeomorphic to @xmath167 , with the ricci flow converging to the fubini study metric ; * @xmath164 is isometric to an irreducible symmetric space @xmath168 with @xmath169 ; * arbitrary isometric products of the above cases .
notice in particular that if @xmath165 in the above , then @xmath164 is diffeomorphic to a sphere since , if the lie algebra @xmath170 of @xmath102 for a symmetric space @xmath168 satisfies @xmath171 , then @xmath168 is in fact isometric to a round @xmath172-sphere .
we will also need the next result due to bishop ( @xcite ) about the holonomy group of a general compact euclidean submanifold in codimension two .
[ bishop ] if @xmath11 is a compact submanifold immersed in @xmath28 , then either @xmath173 and @xmath174 , or @xmath175 for some @xmath176 $ ] .
here @xmath177 corresponds to @xmath165 and @xmath178 to @xmath179 , i.e. , the manifold splits a flat factor locally .
in particular , there can be at most a one dimensional flat factor .
in this section we provide the proof of theorem [ a ] . in the process , we reprove , simplify and strengthen known results about compact euclidean submanifolds in codimension 2 with nonnegative sectional curvature by using proposition [ rigidity ] . along the way
, we will see how the proof implies further rigidity of the immersion .
let @xmath15 and @xmath11 be as in theorem [ a ] . since @xmath11 is compact with nonnegative sectional curvature , a well - known consequence of the cheeger
gromoll splitting theorem , see @xcite theorem c , implies that its universal cover @xmath180 splits isometrically as @xmath181 where @xmath182 is compact and simply connected .
since the lie algebra of the holonomy group of @xmath164 and @xmath183 coincide , proposition [ bishop ] implies that @xmath184 .
now observe that in our context @xmath173 and @xmath185 is not possible in proposition [ bishop ] since then the universal cover would be compact , and by proposition [ rigidity ] diffeomorphic to @xmath186 . but @xmath186 does not admit an immersion in @xmath187 as follows , e.g. , from @xmath188 , the product formula for pontrjagin classes , and @xmath189 for the normal 2 plane bundle @xmath190 of the immersion . therefore , the existence of an isometric immersion implies that @xmath191 for some @xmath176 $ ] .
we can now combine , and proposition [ rigidity ] .
first , it follows that @xmath184 and @xmath192 if and only if @xmath178 . if @xmath193 , then @xmath180 is compact and hence diffeomorphic to @xmath12 , while if @xmath194 and @xmath195 , then @xmath180 is diffeomorphic to @xmath13 . in the latter case , by @xcite the immersion must be a product immersion of convex hypersurfaces and , in particular , @xmath11 itself is simply connected and diffeomorphic to @xmath13 .
if @xmath196 , and hence @xmath179 , implies that @xmath180 splits isometrically @xmath197 , with @xmath43 compact and simply connected .
but then @xmath198 , and again by proposition [ rigidity ] we get that @xmath43 is diffeomorphic to @xmath199 .
thus @xmath200 isometrically , with @xmath201 a discrete group acting by isometries .
we will see in [ z ] below that @xmath202 and that @xmath11 is diffeomorphic to either @xmath19 if @xmath11 is orientable , or to the nonorientable quotient @xmath20 otherwise . to conclude the proof of theorem [ a ] , it remains to study the possibilities for the fundamental group when @xmath193 and @xmath203 , or when @xmath196 and @xmath204 for some metrics of nonnegative curvature on the spheres . in order to do this
, we first recall the relationship , due to kuiper , between height functions and shape operators of compact euclidean submanifolds . following @xcite , for each @xmath205 consider the height function @xmath92 and define @xmath206 and @xmath207 since on a set of full measure the height functions are nondegenerate
, we can apply the usual morse inequalities for morse functions . after integrating them
we get @xmath208 and @xmath209 where @xmath210 for some field @xmath211 .
clearly , @xmath212 is a critical point of @xmath94 if and only if @xmath213 for some @xmath214 , where @xmath215 is the normal bundle of the immersion .
moreover , @xmath212 is a nondegenerate critical point of @xmath216 if and only if @xmath217 is nonsingular , and the index of the critical point is equal to @xmath218 , the number of negative eigenvalues of @xmath217 .
we now use the gauss map @xmath219 with @xmath220 , which for compact submanifolds is surjective .
since @xmath221 , we have that @xmath212 is a nondegenerate critical point of @xmath216 if and only if @xmath222 is a regular point of @xmath223 .
thus if @xmath224 is the set of regular values of @xmath223 ( an open and dense set of full measure ) , then @xmath94 with @xmath225 is a morse function and hence @xmath226 is constant on every connected component of @xmath227 .
furthermore , for @xmath225 , the set @xmath228 contains @xmath93 many points of index @xmath95 .
thus by change of variables , @xmath229 since @xmath230 , we also have @xmath231 for all @xmath95 .
we proceed by applying lemma [ l : eq ] to the shape operators @xmath232 in in order to estimate @xmath233 for @xmath214 , by first integrating over a circle in the normal space @xmath234 , and then over @xmath117 . for a fixed normal space @xmath234 , lemma [ l : eq ] implies that @xmath235 furthermore , if a regular value @xmath236 lies in the first quadrant , then , and if it lies in the third quadrant then @xmath237 , whereas all saddle points ( and possibly some local maxima and minima as well ) lie in the second and fourth quadrants .
thus , together with @xmath231 , implies that @xmath238 as we will see , in the case of strict inequality @xmath164 is diffeomorphic to a sphere .
we are thus interested from now on in the equality case , where one expects a certain amount of rigidity .
this motivates our next key definition .
_ definition .
_ given a compact riemannian manifold @xmath11 with nonnegative sectional curvature , an isometric immersion @xmath9 is said to be _ wide _ if @xmath239 .
we can now use lemma [ l : eq ] to express the rigidity in terms of the shape operators @xmath104 and @xmath105 in .
[ wide ] if @xmath240 is wide , then at each point @xmath241 we have that either @xmath242 with @xmath243 and @xmath244 , or @xmath245 .
equality in holds if and only if we have equality in at every point and for every angle @xmath246 .
thus lemma [ l : eq ] implies that either @xmath242 or @xmath245 .
furthermore , if @xmath133 at some point , then @xmath105 is nonsingular , and hence in a neighborhood of the point as well .
but then this neighborhood will not have any saddle points , and we would have a strict inequality in . notice that , along the ( open ) set where the first case holds , the orthonormal basis @xmath103 in , and hence @xmath104 and @xmath105 , are unique and thus smooth . on the other hand , no such information can be obtained in the second case .
altogether , we have the following improvement of a result in @xcite , where we obtain additional information on the structure of the immersion . [ main ] if @xmath11 is compact , nonnegatively curved and immersed in euclidean space in codimension @xmath247 , then we have for any coefficient field @xmath211 over which @xmath11 is orientable that @xmath248 with equality if and only if @xmath249 in particular , equality implies that the immersion is wide . by the morse inequalities
@xmath250 for @xmath251 and @xmath252 , i.e. @xmath253 . if @xmath11 is compact and orientable mod @xmath211 , then @xmath254 and , in general , @xmath255 .
hence , @xmath256 which together with yields @xmath257 with equality as claimed .
so far we have shown that there are three situations : either @xmath258 diffeomorphically , or @xmath259 splits isometrically as either @xmath260 or @xmath261 , for certain metrics of nonnegative curvature on the spheres . as a corollary of proposition [ main ] , we conclude that in the first case @xmath164 itself is a sphere when @xmath262 : [ sn ] if @xmath263 and @xmath264 is finite , then @xmath11 is simply connected . in particular
, either @xmath11 is diffeomorphic to @xmath12 , or isometric to a riemannian product , @xmath265 , for certain metrics of nonnegative sectional curvature on the spheres . in the latter case , the immersion is a product of two convex hypersurfaces , and thus it is wide .
assume @xmath11 is not simply connected , and let @xmath266 be an element of prime order @xmath267 , so @xmath268 .
then , the lens space @xmath269 is a finite cover of @xmath11 and with the covering metric it has an isometric immersion into @xmath28 .
but @xmath270 for all @xmath95 which contradicts proposition [ main ] if @xmath262 .
notice that if @xmath271 the lens space is orientable , and if @xmath272 it is orientable mod @xmath247 . in the case where @xmath11 is isometric to a riemannian product ,
@xcite implies that the immersion is a product of two convex hypersurfaces .
one easily sees that in this case the immersion is wide . for the three dimensional case
we have the following . [ lens ] if @xmath273 and @xmath274 is finite , then @xmath4 is diffeomorphic to a lens space @xmath23 , and the immersion is wide with @xmath275 . furthermore , the height functions satisfy @xmath276 as well as @xmath277 .
@xmath4 is again covered by a lens space with @xmath278 and we have equality in proposition [ main ] , which implies that @xmath275 ( they do not have to be @xmath279 , though ) .
the morse inequality tells us that @xmath280 for a.e .
@xmath281 , and since @xmath282 , it follows that @xmath283 . from @xmath284
we get @xmath285 for a.e .
@xmath286 . therefore for almost all @xmath286 the morse function @xmath94 has the same number of critical points of index 0 and 1 ( first for the lens space , and then for @xmath4 as well ) .
but if this is the case , the cancellation result in @xcite says that there exists another ( abstract ) morse function on @xmath4 with only one critical point of index 0 and 1 .
in particular , this means that @xmath4 is a cw complex whose 1 skeleton is a circle , which implies by transversality that the map @xmath287 induced by the inclusion is onto and hence @xmath288 is finite cyclic . for infinite fundamental group
, we have : [ z ] if @xmath264 is infinite , then @xmath289 is also cyclic .
moreover , @xmath290 splits isometrically for some metric of nonnegative curvature in @xmath199 , and @xmath11 is diffeomorphic to @xmath19 if @xmath11 is orientable , or to the nonorientable quotient @xmath20 otherwise .
in addition , @xmath291 , and @xmath292 , and hence the immersion is wide .
furthermore , the height functions satisfy @xmath276 as well as @xmath293 , and @xmath294 .
we already saw that under this assumption , @xmath11 is diffeomorphic to @xmath16 for some group @xmath201 acting properly discontinuously .
since the quotient is compact , there exists a subgroup @xmath295 which acts via translations on the @xmath100 factor .
choose a covering @xmath296 such that the image of the fundamental group under @xmath297 is equal to @xmath298 .
thus @xmath299 , and we can apply proposition [ main ] to the induced immersion of @xmath300 . as in the proof of [ lens ] , it follows that for the height functions on @xmath300 we have @xmath283 and @xmath301 for a.e .
the same thus holds for the height functions of the immersion of @xmath11 , and again as in the proof of [ lens ] , we see that @xmath302 is cyclic , hence @xmath201 is isomorphic to @xmath42
. finally , since the functions @xmath303 are nonnegative , and @xmath304 , it follows that @xmath305 as well .
now , projection onto the first factor gives rise to a fiber bundle @xmath306 with fiber @xmath199 which is hence isomorphic to @xmath307/(p,0)\sim ( \sigma(p),1)$ ] for some diffeomorphism @xmath308 .
if @xmath308 is orientation preserving , @xmath308 is homotopic to the identity and hence the bundle is trivial , in which case @xmath11 is diffeomorphic to @xmath19 .
if on the other hand @xmath308 is orientation reversing , @xmath11 is nonorientable and the orientable double cover is diffeomorphic to @xmath19 and hence @xmath11 is diffeomorphic to @xmath20 .
notice that , in both cases , the product structure is not necessarily isometric .
the presence of points with positive curvatures imposes further restrictions .
[ posric ] if there exists a point with positive ricci curvature , then either @xmath11 is diffeomorphic to @xmath12 , or @xmath11 is isometric to a product @xmath13 and @xmath15 is a product of two convex hypersurfaces , for some @xmath309 . in particular ,
if there is a point with positive sectional curvature , then @xmath11 is diffeomorphic to @xmath12 .
the fundamental group must be finite since otherwise [ z ] implies that @xmath259 splits off a real line .
if @xmath273 , then proposition [ wide ] implies that at a point with @xmath310 , one of @xmath232 , say @xmath104 , has rank 2 , and @xmath105 has rank one .
but then @xmath311 and hence all 2-planes containing @xmath312 have curvature 0 , thus contradicting @xmath310 . in the remaining case [ sn ]
proves our claim . in particular , corollaries [ sn ] , [ lens ] and [ z ]
imply the following .
[ mustbewide ] if @xmath11 is not diffeomorphic to a sphere , then the immersion is wide
. 0.1 cm
in this section we obtain further information about the way the manifold @xmath11 is immersed , that is , about the isometric immersion @xmath15 . as we saw in [ mustbewide ] , unless @xmath11 is diffeomorphic to the sphere , for which the space of nonnegatively curved immersions is quite rich , the isometric immersion @xmath15 must be wide .
hence the purpose of this section is to understand these immersions .
we have chosen this approach since there are interesting wide immersions of spheres , as shown in example 2 .
we assume from now on that @xmath11 is a riemannian manifold with nonnegative sectional curvature , @xmath10 , and @xmath9 is an isometric immersion . to provide a deeper understanding of the local phenomena
, we do not require @xmath11 to be complete or compact unless otherwise stated .
instead , we only assume that @xmath15 satisfies the local consequence in proposition [ wide ] of being wide .
in other words , again following the notations in , we say that @xmath15 is _ locally wide _ if , at every point @xmath117 , either @xmath313 from now on we will also assume in the second case that @xmath314 . recall that the _ index of nullity _ @xmath315 and the _ nullity _ @xmath201 of @xmath164 are defined as @xmath316 furthermore , we have the _ index of relative nullity _ @xmath317 of @xmath15 at @xmath212 defined as the dimension of the _ relative nullity _ @xmath318 of @xmath15 at @xmath212 , @xmath319 notice that @xmath315 is an intrinsic invariant , while @xmath320 is extrinsic . by the gauss equation , @xmath321 and hence @xmath322 .
we will also use the well know fact ( see e.g. @xcite ) that the nullity distribution , as well as the relative nullity distribution , is integrable on any open set where it has constant dimension , and its leaves are totally geodesic in @xmath11 . in the case of the relative nullity ,
the images of the leaves under @xmath15 are open subsets of affine subspaces of the euclidean space as well .
in addition , if @xmath11 is complete , the leaves of both distributions are complete on the open set of minimal nullity or minimal relative nullity .
choosing an orthonormal normal frame @xmath103 as in , by @xmath11 can be written as the disjoint union @xmath323},\ ] ] where @xmath102 is the subset of positive index of relative nullity , @xmath324 and , for @xmath325 $ ] , @xmath326 is given by @xmath327 from the continuity of the eigenvalues and @xmath122 it follows that : * @xmath102 is closed , and the sets @xmath326 are open with @xmath328 ; * @xmath329}$ ] is the set of all points with positive ricci curvature , @xmath310 ; * on @xmath330 we have @xmath331 and hence @xmath332 and @xmath333 , @xmath334 .
[ ku ] the above discussion and [ posric ] imply that , if @xmath11 is compact and @xmath15 is wide , then @xmath15 is isometric to a product of convex hypersurfaces unless @xmath335 for all @xmath336 . therefore , for the immersions of types @xmath24 and @xmath25 in theorem [ a ] we have that @xmath337 , and , in particular , @xmath338 everywhere . in what follows , on @xmath339 where we have seen that the frame @xmath103 is unique and smooth , @xmath340 will denote the normal connection form of @xmath15 given by @xmath341 for @xmath342 , and thus @xmath343 and @xmath344 .
hence the codazzi equations @xmath345 become @xmath346 @xmath347 or , equivalently , @xmath348)+b(w(x)y - w(y)x)\ ] ] @xmath349)-a(w(x)y - w(y)x),\ ] ] while the ricci equation is @xmath350x , y{\rangle}.\ ] ] for the sake of completeness , let us first analyze the local behaviour of @xmath15 on @xmath326 for @xmath336 .
we will see that it is already a product of strictly convex hypersurfaces , hence showing that case @xmath27 in theorem [ a ] has its roots in a local phenomenon . [
prod ] assume that @xmath15 is locally wide and @xmath108 $ ] .
then the immersion @xmath109 is locally a product immersion of two strictly convex euclidean hypersurfaces whose respective normal vectors are @xmath351 and @xmath352 . for @xmath355 ,
the codazzi equations imply that @xmath356 = b(w(x)y - w(y)x)$ ] .
since @xmath357 , both sides have to be @xmath358 and hence @xmath312 is integrable .
furthermore , @xmath359 since @xmath360 and @xmath361 . indeed , if @xmath362 , choose a linearly independent @xmath363 , which implies @xmath364 and hence @xmath365 .
analogously , @xmath366 is integrable and @xmath367 .
therefore , @xmath368 which implies that @xmath109 has flat normal bundle . from the ricci equation
it now follows that @xmath104 and @xmath105 commute , and hence @xmath369 .
for @xmath370 and @xmath371 the codazzi equations imply that @xmath372 = \nabla_x au$ ] . since @xmath373 is arbitrary , @xmath374 as well .
hence @xmath375 which implies that the distribution @xmath312 is totally geodesic , and similarly so is @xmath366 .
we conclude that both @xmath312 and @xmath366 are mutually orthogonal transversal totally geodesic distributions , and hence both are parallel , and @xmath11 is locally a product .
furthermore , @xmath376 for all @xmath377 and @xmath378 .
the proposition then follows from the main lemma in @xcite .
observe that each factor is strictly convex since @xmath104 and @xmath105 are positive definite on the corresponding factors .
consider @xmath380 the universal cover of @xmath11 with the covering metric , and @xmath381 . by hypothesis ,
@xmath102 and @xmath330 are empty .
since @xmath382 for all @xmath95 , there is @xmath383 such that @xmath384 . but then proposition [ prod ] and the derham decomposition theorem imply that @xmath259 is globally a product . from @xcite
it follows that @xmath385 is a product immersion of two strictly convex euclidean hypersurfaces . by @xcite ,
a complete strictly convex euclidean hypersurface is the boundary of a strictly convex body and hence embedded .
thus @xmath385 is injective and so @xmath386 and @xmath387 .
the description of @xmath15 on the set @xmath330 is considerably more delicate .
this case is of main interest to us since , as pointed out in remark [ ku ] , for the immersion in case @xmath24 and @xmath25 of theorem [ a ] , we have @xmath388 .
furthermore , @xmath330 is nonempty since otherwise the open subset of minimal relative nullity in @xmath102 would be foliated by complete straight lines in euclidean space , contradicting compactness .
[ comp ] given an isometric immersion @xmath389 , we say that another isometric immersion @xmath390 is a _ composition ( of @xmath391 ) _ when there is an isometric embedding @xmath392 into a flat manifold @xmath393 , an isometric immersion @xmath394 ( that is , a local isometry ) satisfying @xmath395 , and an isometric immersion @xmath396 such that @xmath397 . since @xmath402 , and hence @xmath331 , the shape operator @xmath104 alone satisfies the gauss equation along @xmath403 .
we claim that @xmath104 also satisfies the codazzi equation for hypersurfaces , that is , the skew symmetric tensor @xmath404 $ ] vanishes . for any @xmath405 , the codazzi equation for @xmath104
tells us that @xmath406 . for @xmath407 , the codazzi equation for @xmath105
says that @xmath408=-a(w(x)y - w(y)x)$ ] .
since @xmath409 , it follows as in the proof of proposition [ prod ] that @xmath367 .
thus either @xmath410 or @xmath411 . in the first case , since @xmath412 is skew symmetric , we can assume that @xmath413 and @xmath414 are linearly independent , and hence can assume that @xmath415 .
but then @xmath416 .
altogether , @xmath412 vanishes . by the fundamental theorem of submanifolds , locally on @xmath330 ( or globally on @xmath330 if it is simply connected ) , there exists a euclidean hypersurface @xmath391 whose second fundamental form is @xmath104 . since @xmath417 , @xmath418
now , since @xmath367 , by proposition 8 in @xcite we have that @xmath15 is a composition of @xmath391 ( see also proposition 9 in @xcite ) . since @xmath419 , the nullity geodesics in @xmath330 are ( locally ) mapped by @xmath391 into straight lines of @xmath400 .
if @xmath420 is complete and simply connected , we will see in the proof of theorem [ b1 ] that these straight lines are parallel and @xmath391 is globally a cylinder .
we will also discuss properties of the metric and the immersion on the set of index of nullity @xmath279 , i.e. , on @xmath421 , and the open set @xmath422 where @xmath105 does not vanish ( and hence @xmath15 is 1-regular , i.e. , @xmath423 ) , @xmath424 recall that by the gauss equations on the set @xmath330 we have @xmath333 and , by definition , @xmath334 as well .
thus , @xmath425 on the complement of @xmath330 in @xmath125 we have @xmath426 . in any case
, @xmath125 is the set of minimal index of nullity of @xmath11 , so its leaves of nullity are complete if @xmath11 is complete .
[ b1 ] assume that @xmath11 is compact with nonnegative sectional curvature , and @xmath15 is locally wide .
furthermore , assume that @xmath11 has no points with positive ricci curvature , i.e. , @xmath427 .
let @xmath428 the universal cover of @xmath125 , @xmath429 the lift of @xmath430 , and set @xmath431 .
then we have : * @xmath432 splits globally and isometrically as a product @xmath433 , where @xmath434 is a strictly locally convex hypersurface .
in particular , @xmath432 is itself an euclidean hypersurface via the cylinder over @xmath391 , @xmath435 ; * the restriction @xmath436 is a composition of the cylinder over @xmath437 ; * along each connected component @xmath438 of the interior of @xmath439 , @xmath440 is a composition of the cylinder over @xmath441 with a linear inclusion @xmath442 . here
we continue with the notations in the proof of proposition [ b0 ] .
define on @xmath125 the _ splitting tensor _ of the ( totally geodesic and complete ) nullity distribution , @xmath443 given by @xmath444 , where @xmath445 is a unit vector field tangent to @xmath201 .
since @xmath201 is totally geodesic , the distribution @xmath446 is totally geodesic if and only if @xmath227 vanishes .
observe that @xmath227 satisfies the riccati type differential equation @xmath447 when restricted to a geodesic with @xmath448 .
indeed , @xmath449}t -r(t , x)t+ \nabla_{\nabla_tx}t = \nabla_{\nabla_xt}t = c^2x$ ] since @xmath450 and @xmath451 .
since @xmath452 is complete , this ode holds true over the entire real line , and then @xmath453 , where @xmath454 .
therefore , along each nullity geodesic , all real eigenvalues of @xmath227 vanish .
we claim that @xmath455 .
observe first that every nullity line in @xmath125 has to intersect @xmath330 since in @xmath456 we have @xmath457 and hence a complete nullity line is also a relative nullity line .
but then its image under @xmath15 is a straight line in @xmath28 , contradicting compactness of @xmath11 . as we saw in the proof of proposition [ b0 ] , on @xmath330 the shape operator @xmath104 satisfies the codazzi equation of a hypersurface and hence @xmath458=a[t , x]$ ] since on @xmath330 we also have @xmath459 . thus if @xmath460 and @xmath461 we have @xmath462-a\nabla_{t}x = -a\nabla_xt = acx.\ ] ] hence , from the symmetry of @xmath463 , we have @xmath464 .
furthermore , since @xmath465 on @xmath446 , we have the inner product @xmath466 , positive definite on @xmath446 .
but then @xmath227 is self adjoint since @xmath467 .
thus all eigenvalues of @xmath227 are real , and hence @xmath455 on @xmath125 which proves the claim . in particular
, @xmath125 is locally a riemannian product of a line and a manifold with positive sectional curvature .
moreover , by , @xmath104 is parallel along @xmath452 in @xmath330 , which implies in particular that @xmath351 and @xmath352 are parallel along @xmath330 as well .
now , in @xmath456 , @xmath445 spans the relative nullity of @xmath15 . for any parallel normal vector field @xmath308 along @xmath452 we obtain as in that @xmath468 .
in particular , by taking the parallel transport of @xmath351 we see that , along the whole @xmath125 , there is a unique smooth unit normal vector field @xmath351 such that @xmath469 has rank equal to @xmath470 , and for a local unit normal vector field @xmath471 , @xmath472 has rank at most 1 .
in addition , @xmath104 is parallel along the complete lines of nullity in @xmath125 , and @xmath473 along @xmath456 .
we now show that @xmath432 is a cylinder over a strictly convex euclidean hypersurface , thus proving part @xmath474 .
first , the ricci equation along @xmath456 tells us that @xmath475 since @xmath476 in @xmath456 .
therefore , @xmath477 is parallel along @xmath452 .
moreover , since @xmath478 , @xmath366 is also parallel along @xmath452 .
but in the proof of proposition [ b0 ] we showed that , on @xmath330 , @xmath479 .
we conclude that @xmath479 on the whole @xmath125 , and , again as in the proof of proposition [ b0 ] , @xmath104 satisfies gauss and codazzi equations for @xmath125 .
so , the tensor @xmath480 satisfies gauss and codazzi equations for @xmath432 , and thus there exists an isometric immersion @xmath481 with gauss map @xmath482 whose second fundamental form is @xmath483 . by construction
, @xmath201 is now the relative nullity of @xmath484 , and hence the complete geodesics of nullity are mapped into complete straight affine lines of @xmath400 . on the other hand ,
since @xmath485 for the standard connection @xmath486 of @xmath400 , the complete leaves of @xmath201 are mapped by @xmath484 into locally parallel lines of @xmath400 .
therefore , @xmath432 is globally a product @xmath487 and @xmath484 is a cylinder , i.e. , @xmath488 , where @xmath489 is a strictly convex hypersurface , as claimed .
finally , since @xmath367 , part @xmath490 follows from proposition 8 in @xcite as in proposition [ b0 ] , while part @xmath491 is immediate from @xmath492 and @xmath411 ( i.e. @xmath352 is locally constant ) on @xmath456 .
[ 2pos ] under the assumptions of theorem [ b1 ] , suppose further that @xmath11 has 2-positive ricci curvature . then , for the universal cover @xmath380 we have @xmath112 , where @xmath113 is a compact strictly convex embedded hypersurface . moreover , if @xmath15 is 1-regular , then @xmath401 is globally a composition of its cylinder : @xmath495 . for @xmath273 ,
if @xmath4 is compact but not diffeomorphic to a sphere , we have the decomposition @xmath499 where @xmath211 is the set of flat points of @xmath4 .
indeed , in dimension 3 , @xmath500 implies that @xmath501 .
[ 3 ] let @xmath502 be an isometric immersion of a compact manifold with nonnegative sectional curvature .
if @xmath4 has no complete geodesic of flat points , then @xmath4 is either diffeomorphic to @xmath51 , or its universal cover @xmath503 is isometric to @xmath3 for some metric of positive gaussian curvature on the sphere @xmath504 . by [ mustbewide ] , @xmath15 is wide , hence locally wide .
as we saw in the proof of theorem [ b1 ] , @xmath125 is then foliated by complete geodesics of nullity , and therefore its boundary as well .
but since @xmath505 , the boundary of @xmath125 is made of flat points .
therefore , @xmath4 has no flat points , and @xmath333 everywhere . by part @xmath474 of theorem
[ b1 ] , @xmath506 splits globally and isometrically . that @xmath81 is a sphere is a consequence of theorem [ a ] .
[ lens1 ] let @xmath77 with @xmath78 be an isometric immersion with nonnegative sectional curvature .
then @xmath4 has flat points , and the complement @xmath125 of its flat points is isometric to a twisted cylinder @xmath5 , where @xmath507 is a surface with positive gaussian curvature .
[ taus ] in the situation of [ lens1 ] and hence theorem [ a ] part @xmath25 , the type numbers are given by @xmath97 for all @xmath508 . here , @xmath102 denotes the gaussian curvature of @xmath81 and @xmath99 the total curvature of the leaf of @xmath201 through @xmath101 , considered as a curve in @xmath52 .
notice first that the computation of @xmath91 in only takes into account the normal bundle over the set with vanishing relative nullity , which in our case is @xmath509 .
writing @xmath236 , a direct computation using @xmath510 and @xmath511 shows that @xmath512 on @xmath125 , where @xmath154 is the restriction of @xmath104 to @xmath513 and @xmath514 for @xmath515 $ ] . along @xmath125 the surface
@xmath81 in [ lens1 ] has @xmath516 .
furthermore , @xmath517 is the geodesic curvature of @xmath518 in @xmath52 since @xmath519 and hence @xmath520 .
thus implies that @xmath521 here , @xmath522 is the total geodesic curvature of @xmath523 until the return time @xmath524 , i.e. , the cyclic group @xmath42 in [ lens1 ] is spanned by @xmath525 . | we provide a classification of compact euclidean submanifolds @xmath0 with nonnegative sectional curvature , for @xmath1 .
the classification is in terms of the induced metric ( including the diffeomorphism classification of the manifold ) , and we study the structure of the immersions as well . in particular , we provide the first known example of a nonorientable quotient @xmath2 with nonnegative curvature . for the 3-dimensional case ,
we show that either the universal cover is isometric to @xmath3 , or @xmath4 is diffeomorphic to a lens space , and the complement of the ( nonempty ) set of flat points is isometric to a twisted cylinder @xmath5 . as a consequence
we conclude that , if the set of flat points is not too big , there exists a unique flat totally geodesic surface in @xmath6 whose complement is the union of one or two twisted cylinders over disks .
it is well known that a compact immersed positively curved hypersurface in euclidean space is diffeomorphic to a sphere ( and is in fact the boundary of a convex body ) , the diffeomorphism simply given by the gauss map . a deeper result , that such a submanifold still has to be homotopy equivalent to a sphere in codimension 2 , is due to a. weinstein @xcite and d. moore @xcite .
however , the situation where the sectional curvature of the submanifold is only nonnegative is more delicate .
although it is still not known whether there exists an isometric immersion of @xmath7 into @xmath8 with nonnegative curvature , the higher dimensional problem was studied in @xcite . using recent results about the ricci flow , we strengthen these by showing the following .
[ a ] let @xmath9 , @xmath10 , be an isometric immersion of a compact riemannian manifold with nonnegative sectional curvature .
then one of the following holds : * @xmath11 is diffeomorphic to @xmath12 ; * @xmath11 is isometric to a product metric on @xmath13 for some @xmath14 , and @xmath15 is the product embedding of two convex euclidean hypersurfaces ; * @xmath11 is isometric to @xmath16 with a product metric on @xmath17 and @xmath18 acting isometrically . as a manifold , @xmath11 is diffeomorphic to @xmath19 if orientable , or diffeomorphic to the nonorientable quotient @xmath20 otherwise , where @xmath21 denotes some diagonal @xmath22 action ; * @xmath11 is diffeomorphic to a 3 dimensional lens space @xmath23 . not only is this statement stronger than previously known results , but thanks to the use of the ricci flow its proof becomes substantially simpler as well .
more importantly , in the process we will also provide new strong restrictions on the structure of these submanifolds .
in fact , the particular nature of the submanifolds in case @xmath24 , and even if a nonorientable immersion exists , was unsettled in the literature .
furthermore , if case @xmath25 is possible at all was never discussed .
the main purpose of our paper is to address both issues . regarding case @xmath26
, there are plenty of immersions of @xmath12 with nonnegative curvature in codimension two .
for example , take any compact convex hypersurface and then a composition with a flat ( not necessarily complete ) hypersurface . on the other hand , the submanifolds in case @xmath27
are isometrically rigid .
it was the study of the structure of the immersions that led us to the following non - orientable example in case @xmath24 . _
example 1[ex1 ] : the nonorientable quotient @xmath20 embedded in @xmath28 .
_ consider a flat strip isometrically immersed in @xmath29 @xmath30 such that @xmath31 the image is a flat moebius band immersed or embedded in @xmath29 .
large families of analytic moebius bands of this type , together with some classification results , have been given in @xcite .
the product immersion of @xmath32 with the identity gives a flat hypersurface @xmath33 using a convex hypersurface @xmath34 invariant under the reflection in the first coordinate of @xmath35 , we define the cylinder @xmath36 .
the composition @xmath37 then satisfies @xmath38 , where @xmath39 is the map @xmath40 .
hence the image of @xmath15 is isometric to the _ twisted cylinder _
@xmath41 , where @xmath42 is generated by the orientation reversing isometry of @xmath43 induced by @xmath39 .
thus @xmath44 descends to the desired immersion @xmath45 observe that we can also choose @xmath32 as a cylinder and obtain immersions of @xmath46 which are not products of immersions as in part @xmath27
. actually , they are not even locally product of immersions , as is the case in example 2 below .
we say that an isometric immersion @xmath15 is a _ composition ( of @xmath47 ) _ when @xmath48 , where @xmath49 is an isometric immersion into a ( not necessarily complete ) flat euclidean hypersurface @xmath50 .
we will see that all submanifolds in theorem [ a ] @xmath24 are almost everywhere compositions of a cylinder over a strictly convex euclidean hypersurface when the ricci curvature is 2-positive . for complete
simply connected nowhere flat nonnegatively curved manifolds which split off a line , this structure of a composition was shown in @xcite under certain regularity assumptions .
it is not clear though what additional restrictions hold , or whether regularity is necessary , if the immersion @xmath15 descends to a compact quotient , as required in .
the following example , where the induced intrinsic metric is well known in the theory of graph manifolds , illustrates the structure necessary for case @xmath25 to occur .
_ example 2 : the switched @xmath51 in @xmath52 .
_ consider a closed strictly convex hemisphere inside a closed halfspace , @xmath53 such that the boundary is a closed geodesic in @xmath54 along which the gauss curvature vanishes to infinite order , and with image contained in @xmath55 .
its product with another @xmath56 gives the nonnegatively curved three manifold @xmath57 whose boundary is the totally geodesic flat torus @xmath58 now , reversing the role of the @xmath59 factors we similarly construct another manifold @xmath60 with the same boundary @xmath61 . thus @xmath62 gives an embedded euclidean submanifold with nonnegative sectional curvature : a switched @xmath51 isometrically embedded in @xmath52 as a smooth manifold @xmath4 is diffeomorphic to the 3-sphere since this is the usual description of @xmath63 as the union of 2 solid tori .
the set of flat points of @xmath4 contains the totally geodesic flat torus @xmath64 which disconnects @xmath4 .
moreover , each connected component of @xmath65 is obtained as a composition of the cylinder @xmath66 with a local isometric immersion from @xmath8 into @xmath52 . by replacing the torus @xmath67 with a small flat cylinder @xmath68\times{\mathbb{s}}^1\subset { { \mathbb{r}}}^5
$ ] , we obtain @xmath69 the resulting embeddings @xmath70 for @xmath71 have thus the following properties : * @xmath72 is compact with nonnegative sectional curvature ; * the set of flat points contains a totally geodesic flat torus that disconnects @xmath72 ; * @xmath72 is locally an isometric product of two manifolds for @xmath73 ( almost everywhere for @xmath74 ) ; * the embedding @xmath75 is locally , yet not globally , a composition of a cylindrical hypersurface for @xmath73 ( almost everywhere for @xmath74 ) ; * @xmath72 has no points with positive ricci curvature ; * @xmath72 has 2-positive ricci curvature outside the set of flat points . the submanifold @xmath72 can be changed further so that the boundary of the set of flat points is not regular , and arbitrarily complicated sets of flat points can be introduced in @xmath76 as well .
observe that , by taking the two disks @xmath76 with strictly convex boundaries as plane curves , all these properties can be achieved even with 1-regular full embeddings , i.e. , with the dimension of the space spanned by the second fundamental form having constant dimension 2 everywhere .
the following result recovers some of this structure for any immersion of a lens space with nonnegative sectional curvature . [ 3a ]
let @xmath77 with @xmath78 be an isometric immersion with nonnegative sectional curvature .
then the set @xmath79 of flat points of @xmath4 is nonempty , and @xmath80 is isometric to a twisted cylinder @xmath5 , where @xmath81 is a ( not necessarily connected ) surface with positive gaussian curvature .
moreover , @xmath15 is a composition almost everywhere on @xmath80 .
the natural intrinsic problem raised by this result is to try to understand how twisted cylinders can be glued together with flat regions in order to build a compact manifold .
one way is to glue the cylinders through compact totally geodesic hypersurfaces , as we did for @xmath82 in example 2 above . in @xcite
we study this intrinsic structure under the additional assumption that the set of nonflat points @xmath83 is dense and locally finite ( without assuming that the curvature is necessarily nonnegative ) .
the main result in @xcite is that under this assumption there exist pairwise disjoint flat totally geodesic compact surfaces in @xmath4 whose complement is a union of twisted cylinders .
we also give examples of metrics on @xmath84 which show that the assumption on the nonflat points is necessary . in the case of nonnegative curvature ,
theorem b above , together with theorem c in @xcite , implies that : [ main2 ] let @xmath77 , @xmath78 , be an isometric immersion with nonnegative sectional curvature , such that @xmath80 is dense and its number of connected components is locally finite .
then , either @xmath79 contains a unique flat totally geodesic torus @xmath67 that disconnects @xmath4 and @xmath85 , or @xmath4 is the lens space @xmath86 and @xmath79 contains a unique flat totally geodesic klein bottle @xmath87 satisfying @xmath88 . in both cases , @xmath89 is a twisted cylinder , where @xmath90 is a disc with nonnegative gaussian curvature and boundary a closed geodesic . yet , despite the rigid structure described above , we still do not know if a nonnegatively curved lens space in @xmath52 actually exists . for the proof of theorem [ a ] we will need a fundamental property of our submanifolds that was proved in @xcite and @xcite .
in order to state it , recall that we have the _ type numbers _
@xmath91 of the immersion : if @xmath92 is a height function , and @xmath93 the number of critical points of @xmath94 with index @xmath95 , then @xmath91 is the average of @xmath93 , integrated over the unit sphere in @xmath28 .
if @xmath11 is compact with nonnegative sectional curvature immersed in @xmath28 , then the type numbers satisfy @xmath96 furthermore , if the inequality is strict , we will see in section [ rg ] that @xmath11 is diffeomorphic to a sphere .
we are thus mainly interested in the case of equality , and call such immersions _
wide_. in the situation of theorem [ 3a ] , we will show that @xmath97 for @xmath98 , where @xmath99 denotes the total curvature of the image of the @xmath100 factor through @xmath101 , and @xmath102 the gaussian curvature of @xmath81 . for codimension two euclidean submanifolds with nonnegative curvature
, a. weinstein @xcite showed that at every point there exists a basis @xmath103 of the normal space such that its shape operators @xmath104 and @xmath105 are nonnegative , and hence the curvature operator is nonnegative .
the crucial new property that we will prove is that for a wide immersion we have in addition that @xmath106 and call the ( not necessarily complete ) immersions which satisfy only this property _ locally wide_. these immersions already have strong rigidity properties .
we examine their behavior on the open subsets @xmath107 , which fill out the complement of the points of positive relative nullity .
we will show in particular that : * if @xmath108 $ ] , then @xmath109 is locally a product immersion of two strictly convex euclidean hypersurfaces ; * if @xmath110 is the universal cover , then @xmath111 is a composition of a convex euclidean hypersurface with constant index of relative nullity one ( see theorem [ b1 ] ) ; * if @xmath11 is complete and has no points of positive ricci curvature , but has 2-positive ricci curvature , then @xmath112 , where @xmath113 is a compact strictly convex hypersurface . |
while the first statistically significant observation of cp violation in the b meson system will most likely take place in one of the @xmath0 @xmath1-factory experiments ( babar , belle ) , this talk will try to prove that the contribution from experiments in hadron colliders will be dominant in the future , starting in 2001 .
figure 1 shows the standard cp violation triangle .
the angle @xmath2 is the most easily accessible , in particular through the asymmetry in the decays @xmath3 .
first determinations by cdf @xcite , opal @xcite and aleph @xcite have recently become available .
when combined , they give a two - standard deviation measurement of @xmath4 : @xmath5 the combination is totally dominated by the cdf result . by summer 2000 , the electro - positron colliders should get a measurement with an uncertainty about @xmath6 @xcite .
the task is substantially more involved for the other two angles , @xmath7 and @xmath8 .
it is not obvious when will babar or belle be able to obtain a measurement .
hadronic uncertainties diminish the usefulness of a determination of the length of the side of the triangle opposite @xmath2 .
the length of the side opposite @xmath8 , can be obtained from the determination of @xmath9 , the frequency of the oscillations in the @xmath10 system .
the current limit from lep is @xcite : @xmath11 standard model expectations for @xmath12 are in the range 2030 @xcite . when the determination of @xmath12 will become available
, it will be possible to combine it with @xmath13 , the similar parameter in the @xmath14 system , now measured to be @xcite @xmath15 in order to obtain the length of the side of the triangle opposite @xmath8 through the relation : @xmath16 since the lepton machines are only scheduled to run at the energy of the @xmath17(4s ) resonance , no @xmath18 mesons will be produced .
therefore , @xmath12 will only be determined in hadron machine experiments . the standard cp triangle in fig .
1 is the only one obtained from the standard wolfenstein parametrization of the ckm matrix @xcite .
however , one should remember that this is just an expansion of the complete matrix in powers of @xmath19 , the sinus of the cabibbo angle , valid up to order @xmath20 . if one proceeds with the expansion up to @xmath21 , one encounters a new imaginary contribution to @xmath22 , which gives raise to a new unitarity triangle .
a new angle , @xmath23 , appears , which in the standard model is proportional to @xmath24 , and , therefore , very small , of order @xmath25 .
it is easily accessible through the decay @xmath26 , the @xmath18 equivalent to the `` golden channel '' @xmath27 .
again , this is not possible in the lepton machines , which do not produce @xmath18 .
the experimental program in the coming years has two clear goals : * first , to observe for the first time cp violation in the b system . as explained above , this
will most likely be achieved by babar and/or belle , followed by hera - b at desy , cdf and d0 .
* then , to test whether cp violation is generated by the standard model . here , all the experiments mentioned before will contribute but , most probably , the second - generation dedicated experiments , lhcb and btev , will be needed . in the remaining of the talk , the capabilities of the experiments foreseen in future hadron colliders will be reviewed .
section 2 will cover the near future : cdf and d0 at the fermilab tevatron , scheduled to start physics data - taking in spring 2001 .
the more distant future , from 2005 on , will be covered in section 3 .
atlas , cms and lhcb are approved to take data at cern s lhc , while btev , if approved , will run at the tevatron .
special emphasis will be put in the two dedicated @xmath1-experiments , lhcb and btev .
finally , section 4 will give a short summary of the talk .
the upgraded tevatron at fermilab is scheduled to start collider physics operation in spring 2001 with two upgraded detectors , cdf and d0 .
the luminosity delivered in the first two years of operation is expected to reach about 2 fb@xmath28 , which is equivalent to a production of about 10@xmath29 @xmath1 pairs per year .
the challenge in a hadron collider experiments is not producing the @xmath1 pairs , but rather triggering on them , selecting them and getting rid of the background .
cdf has proven to be able to do all this with their recent determination @xcite @xmath30 using the @xmath31 channel .
both cdf and d0 are undergoing substantial upgrades @xcite . among other improvements , cdf is getting a new , longer , vertex detector , that will provide 3d coordinates in a larger acceptance region . a new time - of - flight system
will allow kaon identification at low momentum , and can , therefore , be used for kaon tagging : using the kaon charge to decide on the flavor of the decaying @xmath1 hadron .
more importantly , a new pipelined , deadtimeless trigger system will allow purely hadronic events ( no leptons ) to be triggered efficiently .
this will vastly increase the physics capabilities of cdf , as it will be shown later .
the new all - hadron trigger requires two high transverse momentum tracks at level 1 and , using vertex detector information , finds their vertex at level 2 and requires the decay length to be positive .
d0 upgrades are not less important .
the detector will have for the first time a solenoidal coil , with a 2 tesla field that , together , with the new scintillating - fiber central tracking system , will provide precise momentum measurements for all charged tracks .
furthermore , a new four - layer vertex detector will also be added .
these changes should allow d0 to reconstruct and tag b decays using standard displaced vertes techniques .
a comparison between the @xmath1-physics capabilities of cdf and d0 shows a clear advantadge for the former , due , in particular , to its unique all - hadron trigger . in order to trigger on hadronic
b decays , d0 has to rely on semileptonic decays of the opposite - side b hadron , therefore paying the price of the 10% semileptonic branching ratio . as mentioned in the introduction , the measurement of the mass difference of the two @xmath18 sates , @xmath9 , is needed in order to determine the length of the side of the standard ckm triangle opposite the angle @xmath8 .
the mass difference is obtained from the difference between mixed and unmixed @xmath18 decays as a function of proper time : @xmath32 where @xmath33 is the so - called dilution factor , explained later . in order to be able to perform the measurement one needs to determine : * the proper decay time .
hence good decay length resolution is needed . * the @xmath18 flavor at production time ( tagging ) .
the dilution factor @xmath33 in the previous equation is @xmath34 , where @xmath35 is the probability of getting the flavor wrong .
* the @xmath18 flavor at decay time , using , for instance , flavor - specific decays .
cdf has studied the flavor - specific decay channel @xmath36 or @xmath37 with @xmath38 .
as one can see , there are no leptons and , therefore , the purely hadronic trigger is mandatory .
cdf expects about 20000 selected events in the first two years of operation with a signal - to - background ratio between 1/2 and 2 @xcite .
the significance of the measurement of @xmath9 ( in number of standard deviations ) or of the related , dimensionless parameter @xmath12 introduced above , is given by @xmath39 where @xmath40 is the number of selected events , @xmath41 is the tagging efficiency , @xmath33 is the dilution factor , @xmath42 is the proper time resolution , @xmath43 is the @xmath18 lifetime and @xmath44 and @xmath45 are the number of signal and background events selected , respectively . looking at the formula , one sees that the `` quality factor '' @xmath46 gives the effective tagging efficiency and that good proper time resolution is , clearly , crucial .
no matter how many events one can collect , the reach in @xmath9 is going to be limited by the time resolution to something of order @xmath47 .
cdf expects a time resolution around 50 fs and a quality factor @xmath48 , which translate into a sensitivity to @xmath49 at the five standard deviation level @xcite .
this goes well beyond the current lep limit of 21.8 and the expected range in the standard model , from 20 to 30 . in summary
, cdf should be able to determine @xmath12 quite precisely unless it is much larger than expected .
the standard prodecure to determine the angle @xmath2 uses the time integrated asymmetry in the @xmath50 decays to @xmath51 : @xmath52 here @xmath53 is a dilution factor due to @xmath54 mixing .
since @xmath12 is much larger than @xmath13 this dilution factor is very small in the @xmath55 asymmetries , so that time - dependent measurements are needed .
the experimentally observed asymmetry is further diluted by mistagging and background : @xmath56 so that the final uncertainty on @xmath4 can be written as @xmath57 cdf believes they can obtain a total @xmath58 around 9.1 % , including the kaon tagging using the proposed time - of - flight system . with the statistics available in the first two years of data - taking , that would imply @xmath59 @xcite . on the other hand , total @xmath58 for d0
is expected to be around 5.0 % , implying @xmath60 , using only the muon decay of the @xmath31 @xcite .
analyses are underway to quantify the precision attainable in the electron channel
. determination of cp angles other than @xmath2 can also be attempted by cdf
. it will be more difficult for d0 because of their lack of a fully hadron trigger .
a quick summary of cdf capabilities follows : * @xmath61 could , in principle , be obtained from the @xmath62 asymmetry in the @xmath63 decay . however , the large penguin contribution , with different phase , my preclude the effective extraction of @xmath7 from this channel .
some other methods will be described in section 3 . in any case
, cdf can have a quality factor around 9.1% for this channel , leading to a precision in the determination of the asymmetry of @xmath64 @xcite .
* there are several ways of studying the angle @xmath8 .
cdf has explored , for instance , the use of the asymmetries in the decays @xmath65 and @xmath66 , where @xmath67 is the @xmath68 eigenstate of the @xmath69 system . for the moment
, however , there are no quantitative estimates of the possible @xmath70 reach .
* finally , the asymmetry in the decays @xmath71 , the `` golden decay '' equivalent in the @xmath18 sector , can be determined to about 10% by cdf , depending on the value of @xmath12 @xcite .
this asymmetry provides a direct determination of @xmath23 .
they expect about 6000 events with @xmath72 .
while only an asymmetry of order a few percent is expected in the standard model , a larger value , if found , would be a sign of new physics .
it should be noted that since in this channel the triggering is based on the leptonic decays of the @xmath31 , d0 should be able to do a similar job .
in about 2005 atlas , cms and lhcb will start taking data at the lhc accelerator at cern . at about the same time , btev , if approved , will start at the tevatron . at that time
a lot of progress will have been made by the earlier experiments , both at hadron machines and at the electron - positron factories .
in particular , one could expect @xmath4 to be measured with a precision around 0.04 by a combination of babar , belle , hera - b , cdf and d0 .
the length of the side opposite @xmath8 will be known thanks to the @xmath18 oscillations measurements at hera - b , cdf and d0 but it is rather unclear the amount of information that will be gathered about @xmath7 and @xmath8 .
the outcome of the combination of all the different measurements should be one of the following : * either a clear inconsistency with the standard model predictions will be seen in the precise measurements ( @xmath2 and @xmath18 mixing ) ; * or a hint of inconsistency with the less precise measurements ( @xmath7 , kaon results ) will appear ; * or all measurements will look consistent with the standard model . in any case , the next generation of experiments will be needed for a full understanding of the cp violation mechanism . that will involve more precise measurements of the same parameters in the same channels , measuring the same parameters in different channels and determining for the first time some parameters , notably the angle @xmath8 . around year 2005
, four new hadron - machine experiments with interesting capabilities in @xmath1 physics will start taking data .
two of them are specifically designed for @xmath1 physics , lhcb @xcite and btev @xcite .
both of them cover only the forward region , since most of the @xmath73 s produced are in this region .
btev actually covers also the backward region , thus doubling the acceptace . the larger cross section available at lhc energies ( 500 @xmath74b vs. 100@xmath74b ) more than compensates for the smaller acceptance of lhcb . covering only the forward / backward regions
has clear advantages : * @xmath73 production peaks forward ; * limited solid angle coverage leads to limited cost ; * the @xmath1 hadrons have higher momentum which leads to easier vertex finding and improved decay time resolution ; * open geometry allows for easy installation and maintenance .
these advantadges compensate for the higher minimum bias background , occupancies and radiation dose that have to be overcome at low angles .
the main advantadges that lhcb and btev have over the all - purpose detectors , atlas and cms , are in particle identification and , especially , in the triggering capabilities .
lhcb and btev have rich systems that allow @xmath75 separation in a large momentun range ( @xmath76 for lhcb , @xmath77 gev for btev ) .
this is mandatory to separate signal from background in the @xmath78 channels and adds the possibility of using kaon tagging , as mentioned in chapter 2 for cdf . both lhcb and btev can trigger efficiently in all - hadron events , thanks to their vertex triggers .
in contrast , atlas and cms need a high-@xmath79 lepton to start the triggering sequence .
this considerably reduces the @xmath1-physics capabilities of the two multipurpose detectors .
the angle @xmath2 can be measured very precisely by the second generation @xmath1 experiments using the same `` golden '' decay used by cdf and d0 : @xmath80 . since the @xmath31 is only reconstructed through its lepton - pair decays ,
the triggering and particle - id advantadges of lhcb and btev are of no importance and the kind of precision that can be achieved by the four experiments is very similar , ranging from @xmath81 for lhcb to @xmath82 for cms @xcite , with atlas and btev somewhere in between .
the @xmath18 mixing analysis is much improved by the availability of a pure hadron trigger and , therefore , the differences between the reach of the experiments are large . whereas atlas and cms could only reach @xmath12 values around 50 @xcite , lower than what can be explored already at cdf with their hadron trigger
, the dedicated experiments can go up to 60 ( btev ) or 75 ( lhcb ) @xcite .
it should be said , however , that standard model expectations for @xmath12 are in the range 2030 , easily accessible by the first generation experiments .
the asymmetry in the @xmath83 channel has already been mentioned in section 2 as a possibility for measuring @xmath7 , although , at present , its usefulness is not clear due to the large penguin contribution with different phase . in any event , lhcb and btev can measure the asymmetry in this all - hadron channel with a @xmath82 precision @xcite .
more promising , if more involved , seems to be the @xmath84 channel , which proceeds through @xmath85 intermediate states .
here , all amplitudes can be determined and the remaining ambiguities can be resolved using the intereference regions in the dalitz plot @xcite .
the analysis seems feasible , but no experiment has yet made public the precision it could achieve in @xmath7 . there are several ways of getting to the angle @xmath8 , all of them rather challenging .
for example , it can be obtained from the simultaneous measurement of six time - integrated decay rates in the @xmath86 channel : @xmath87 , with @xmath69 decaying to @xmath88 , @xmath89 to @xmath90 , and @xmath67 to either @xmath91 or @xmath63 .
the final state , therefore , consists in all cases in four charged kaons or pions resonating in different masses .
it is clear that proper pion / kaon separation is crucial .
lhcb claims a precision in @xmath8 around 10@xmath92 from this channel @xcite .
btev has studied a similar channel : @xmath93 , where @xmath69 and @xmath89 go to the same final state , and the corresponding @xmath94 decays .
using now nine time - integrated decays , the precision in @xmath8 is found to be about 13@xmath92 @xcite .
apart from @xmath7 , @xmath2 and @xmath8 , other cp angles are also accessible to the second generation cp experiments : * the small angle @xmath23 introduced in section 1 can be measured precisely using the @xmath71 asymmetry as already mentioned above .
precisions around @xmath95 seem possible , even for atlas and cms @xcite , because of the leptonic decays of the @xmath31 . *
the combination @xmath96 can be determined with a precision around 10@xmath92 @xcite by both btev and lhcb through the decay @xmath97 and its charge conjugated .
since @xmath23 is expected to be small in the standard model , this can be viewed as yet another way of getting @xmath8 , only in this case through @xmath18 decays , and , hence , a very interesting check .
* lhcb has also studied the extraction of @xmath98 from @xmath99 .
the precision that can be obtained is around 9@xmath92 @xcite .
again , since @xmath2 will be well known , this method could be used to obtain @xmath8 or @xmath7 .
the ability of measuring a single angle with different processes can be very useful in checking whether the standard model by itself is able to explain cp violation .
this talk has tried to convey the message that hadron machines have a very comprehensive program for understanding the origin of cp violation in the period 20012015 or so , at the tevatron and lhc . in a first phase , starting in 2001 , cdf and , to a lesser extend , d0 will be able to measure @xmath4 to about 0.07 and study @xmath18 mixing for values of the mixing parameter @xmath12 up to 63 . in a second phase , from 2005 on , lhcb , btev ,
if approved , plus atlas and cms will improve on @xmath2 and @xmath12 and will be able to determine both @xmath7 and @xmath8 to about 10@xmath92 , thus putting strong constraints on the standard model ability to explain all the phenomenology of cp violation in the @xmath1 sector .
it is a pleasure to thank giorgio capon and the rest of the organizers of the workshop for the kind invitation to give this talk and for their patience while waiting for me to finish writing this manuscript . 99 cdf collaboration , f. abe _ et al .
lett . * 51 * , 5513 ( 1998 ) .
+ cdf collaboration , `` a measurement of @xmath4 from @xmath100 with the cdf detector '' , fermilab - pub-99/225-e ( september 1999 ) .
opal collaboration , k. ackerstaff _ et al .
j. * c5 * , 379 ( 1998 ) .
aleph collaboration , `` study of the cp asymmetry in the @xmath100 decays in aleph '' , aleph 99 - 099 , conf 99 - 054 , contributed to the 3rd int .
conf . on b physics and cp violation , taipei , taiwan , december 1999 .
i. peruzzi , `` b factories : prospects and preliminary experimental results '' , these proceedings .
the lep b oscillations working group , lepbosc 98/3 , and updates in http://www.cern.ch/lepbosc/ , presented at the 1999 lepton - photon symposyum , stanford university ( usa ) , august 1999 .
y. nir , phys .
b327 * , 85 ( 1994 ) .
l. wolfenstein , phys .
lett . * 51 * , 1945 ( 1984 ) .
cdf ii technical design report , fermilab - pub-96/390-e ( 1996 ) .
the d0 upgrade : the detector and its physics , fermilab - pub-96/357-e ( 1996 ) .
j. kroll , `` b physics in run ii from cdf '' , in _ workshop on b physics at the tevatron run
ii and beyond _ , fermilab ( usa ) , september 1999 .
f. stichelbaut,``b physics at d0 in run ii '' , in _ nsf b physics review _
, arlington ( usa ) , january 1998 . lhcb technical proposal , cern / lhcc 98 - 4 ( 1998 ) .
btev : an expression of interest for a heavy quark program at c0 , fermilab - proposal-897 ( 1997 ) . n. harnew , `` b - physics potential of atlas , cms , lhcb and btev '' , contribution to the 8th international symposium on heavy flavour physics , southampton ( uk ) , july 1999
snyder and h.r .
quinn , phys . rev . *
d48 * , 2139 ( 1993 ) .
t. skwarnicki , `` btev experiment for cp violating measurements '' , in _ workshop on b physics at the tevatron run ii and beyond _ , fermilab ( usa ) , september 1999 . | the capabilities of future cp violation experiments in hadron machines are reviewed .
special emphasis is put on the cdf experiment in the run ii of the tevatron , due to start taking data in 2001 , and on lhcb and btev , on a longer time scale .
= 11.6pt |
matching the large scale structure of the universe remains one of the most important constraints for models of structure formation .
models have been primarily compared to observations using statistics such as the correlation function , power spectrum and counts - in - cells .
these two - point statistics have shown that galaxies are indeed clustered and have allowed us to reject models , such as the standard cold dark matter model ( scdm ) with @xmath5 , as it does not have enough power on large scales .
similar constraints from the cosmic microwave background measurements , supernova ia experiments , large scale structure and clusters ( see bahcall et al .
1999 for a recent review ) have led to the development of a consensus flat cosmological constant - dominated cdm model .
all currently examined cdm models begin with gaussian random phase initial conditions .
inflation predicts that the seeds for structure formation should derive from a gaussian random phase distribution ( bardeen , steinhardt & turner 1983 ) .
this can be tested using topological statistics such as the genus statistic ( gott et al .
the topology of the large scale structure is invariant during the linear growth of structure , thus , after appropriate smoothing on large scales , the topology of the present galaxy distribution can be related to that of the initial density field .
this allows a test of the random phase hypothesis as any deviation of the measured topology might be evidence for non - gaussian initial conditions . on
relatively smaller scales , the topology of the smoothed galaxy density quantifies the degree of non - linear evolution and/or biasing of galaxy formation with respect to the mass density at the present epoch .
the genus statistic has been measured from many different galaxy redshift surveys .
gott et al .
( 1989 ) applied the 3d genus statistic to small samples of galaxies .
it has since been applied to subsequently larger surveys ; the 3d genus of the ssrs was measured by park , gott & da costa ( 1992 ) , moore et al .
( 1992 ) applied it to the qdot survey , rhoads et al .
( 1994 ) analyzed abell clusters , vogeley et al .
( 1994 ) analyzed the cfa survey and canavezes et al .
( 1998 ) analyzed the pscz survey . in 2d , the method has been applied to the cfa survey ( park et al .
1993 ) and the lcrs survey ( colley 1997 ) .
the 2d genus of the microwave background has also been measured ( colley , gott & park 1996 ) .
these papers have all concluded that the genus curve obtained was consistent with the universe having gaussian initial conditions but in some cases , results were limited by the size of the surveys .
some evidence for departure from gaussianity was found as a result of non - linear gravitational evolution and/or biasing of galaxies as compared to the mass . in this paper
, we estimate the genus of the largest galaxy redshift survey that is publicly available , the 2 degree field galaxy redshift survey ( 2dfgrs ) . in section [ sec : survey ] we describe the survey in more detail . in section [ sec : genus ]
we describe the genus statistic and in section [ sec : res ] we present our results .
we draw conclusions in section [ sec : conc ] .
the 2dfgrs is an optical spectroscopic survey of objects brighter than b@xmath0=19.45 selected from the apm galaxy survey ( maddox et al .
1990a , b ) .
the 2dfgrs survey is divided into two main regions , with additional random fields observed to improve the angular coverage for statistics such as the power spectrum . for this analysis
we do not use the random fields .
the two areas of interest are the south galactic pole ( sgp ) region , which covers the region @xmath6 , @xmath7 and a region which is close to the north galactic pole ( ngp ) , @xmath8 , @xmath9 . the data that we analyze are from the public release that was distributed to the community in june 2001 ( colless et al . 2001 and references therein ) .
102,426 galaxy redshifts were contained in the data release , making this the largest redshift survey that is publicly available .
as the survey is not finished , the angular selection function is complicated . to match this
when we construct random catalogs , required when estimating the genus , we use the software developed by the 2dfgrs team and distributed as part of the data release . for any given coordinates ( @xmath10 ) , the expected probability of a galaxy being contained in the 2dfgrs survey region is returned .
we construct a volume - limited sample from the 2dfgrs in order to obtain a radial selection function that is uniform , thus the only variation in the space density of galaxies with radial distance is due to clustering .
this means no weighting scheme is required when constructing the density field but it does restricts the number of objects in a volume - limited sample extracted from this data .
we select a volume limit of @xmath11=0.138 .
this value maintains the maximum number of galaxies in the sample .
we adopt the global @xmath12-correction + evolution correction found for b@xmath0 selected galaxies in the eso slice project ( zucca et al .
1997 ) and adopted by the 2dfgrs team ( norberg et al . 2001 ) and use @xmath13 following the 2dfgrs team and to maintain consistency with the simulation we will describe in section [ sec : res ] , we adopt a @xmath14 cosmology when converting redshift into comoving distances .
for this cosmology , the redshift limit of @xmath11=0.138 corresponds to a comoving distance of 398@xmath15mpc . [ cols="^,^ " , ] as a final comparison , we compare the genus of the 2dfgrs with the genus of a gaussian random field ( grf ) that has the same power spectrum as the 2dfgrs .
the power spectrum of the 2dfgrs has been measured by percival et al .
( 2001 ) but is only presented in a form where it is divided by a linear theory power spectrum with values of @xmath16 with @xmath17=1 and @xmath18=1 .
we use cmbfast to construct such a power spectrum , assuming that @xmath19=0.7 and that @xmath20=1 .
we multiply the points presented in percival et al . to obtain an estimate of the 2dfgrs power spectrum .
we also measure the power spectrum from the two ngp and sgp volume - limited samples we have constructed here to check consistency and find that the three power spectra have approximately the same shape . on scales larger than 10@xmath15mpc and up to 400@xmath15mpc , the 2dfgrs p(k ) is fairly similar in shape to the cmbfast power spectrum . to test
how sensitive the genus curve is to the exact features in the power spectrum curve , we compare the genus curves we obtain using a power law with p@xmath21 , the cmbfast power spectrum and the estimate of the 2dfgrs power spectrum we obtain here and find little difference between the cmbfast and 2dfgrs genus curves and only small differences between these genus estimates and ones created with the power law input .
we construct the grf on a 256@xmath22 array by assigning values to the real and imaginary parts of each fourier mode which are gaussian deviates .
we then imprint the shape of the ngp and sgp regions onto the grid before we smooth the density field and calculate the genus , exactly as described above .
we compare the genus of the 2dfgrs with that of a grf with the same power spectrum to see if the non - linear clustering that appears in the data and in the simulations affects the amplitude of the 2d genus . in previous papers that examine the genus in 3d , significant
damping of the amplitude of the data genus curve has been found as compared to the genus curve of a grf .
vogeley et al .
( 1994 ) found that for smoothing scales of 6@xmath15mpc that a grf had a 2@xmath4 higher amplitude than the genus curve of the cfa survey . on larger smoothing scales
this difference was reduced and at 16@xmath15mpc , the two curves had consistent amplitude .
we find in 2d that we do not see the same damping that has previously been seen in the 3d genus .
for values of @xmath23 , the amplitude of the grf genus curve is fairly similar to that of the 2dfgrs but on positive @xmath24 scales , the amplitude is lower than the 2dfgrs . on these large scales
, the data is definitely showing an excess of overdensities , as shown earlier in figure [ fig : simcomp ] .
the best fitting gaussian curve has a comparable amplitude to the underdense part of the genus curve but a lower amplitude than the overdense part of the genus curve .
several factors could account for the apparent discrepancy between the 2 and 3d genus amplitude ratios
. it may be that our smoothing length of 10@xmath15 is close enough to the linear regime that no amplitude difference due to non - linear clustering may be seen .
note that our 10@xmath15mpc smoothing length would be @xmath25mpc if we used the smoothing length definition adopted in in older genus papers .
vogeley et al .
( 1994 ) found no effect in 3d for @xmath26mpc .
it may also be that the non - linear clustering has a smaller effect on the 2d genus , where data are projected and thereby smoothed over the width of the survey , thus making the effective smoothing length larger .
however , we repeated this analysis using a 2d smoothing length of only 5@xmath15mpc and found similar results . because an excess of clusters may reflect biasing of the galaxies with respect to the mass distribution , it will be interesting to repeat this analysis with the final data set to examine whether the completed sample exhibits this amplitude excess in the genus curve at high density thresholds .
if these partial data in the 2dfgrs strips were slightly biased toward high density regions , perhaps because denser fields were given targeting priority over lower density regions during early observations , then the complete sample might lack such a feature .
we measure the genus from a 100,000 galaxy redshift sample of the 2dfgrs , the largest redshift sample publicly available to date . we find that the genus curves of the ngp and sgp are slightly different , with the ngp region displaying an excess of isolated clusters and the sgp region displays a slight excess of isolated voids .
these shifts are only significant at the 1@xmath4 level when we compare the genus curves to mock catalogs drawn from a @xmath2cdm simulation .
the results are qualitatively the same for two choices of the smoothing length .
we average the ngp and sgp data and find that the average genus curve is consistent with the @xmath2cdm simulation .
we find that the average 2dfgrs genus curve reveals a smaller number of isolated voids than clusters in the survey , although the points in the genus curve that show this slight excess of clusters are highly correlated .
we demonstrate that in spite of the current incompleteness in the geometry , the genus can be accurately recovered .
no bias is introduced into the shape of the genus curve by the current angular coverage .
we compare the 2dfgrs genus curve with that for a grf with the same power spectrum .
we do not find any significant amplitude depression in the amplitude of the 2dfgrs genus curve due to non - linear clustering .
this may be due to the large smoothing length , may be a feature of the 2d genus , or may indicate a real excess in the number of isolated clusters in the the 2dfgrs that is not predicted directly from the power spectrum alone .
colley ( 1997 ) found that the 2d genus of the las campanas redshift survey agrees well with a gaussian random field . a larger sample of galaxies will reveal whether there really is an excess of over dense regions , or whether this is a feature of the observing strategy applied so far .
msv acknowledges support from nsf grant ast-0071201 and the john templeton foundation .
jrg acknowledges the support from nsf grant ast-9900772 .
we thank david weinberg for providing the contour2d code .
we also thank the 2dfgrs team for their tremendous efforts in obtaining and releasing this data set and for providing the related analysis tools . | we study the topology of the publicly available data released by the 2 degree field galaxy redshift survey team ( 2dfgrs ) .
the 2dfgrs data contains over 100,000 galaxy redshifts with a magnitude limit of b@xmath0=19.45 and is the largest such survey to date .
the data lie over a wide range of right ascension ( 75@xmath1 strips ) but only within a narrow range of declination ( 10@xmath1 and 15@xmath1 strips ) .
this allows measurements of the two - dimensional genus to be made .
we find the genus curves of the ngp and sgp are slightly different .
the ngp displays a slight meatball shift topology , whereas the sgp displays a bubble like topology .
the current sgp data also have a slightly higher genus amplitude . in both cases , a slight excess of overdense regions
are found over underdense regions .
we assess the significance of these features using mock catalogs drawn from the virgo consortium s hubble volume @xmath2cdm @xmath3=0 simulation .
we find that differences between the ngp and sgp genus curves are only significant at the 1@xmath4 level .
the average genus curve of the 2dfgrs agrees well with that extracted from the @xmath2cdm mock catalogs .
we also use the simulations to assess how the current incompleteness of the survey ( the strips are not completely filled in ) affects the measurement of the genus and find that we are not sensitive to the geometry ; there are enough data in the current sample to trace the isolated high and low density regions .
we compare the amplitude of the 2dfgrs genus curve to the amplitude of a gaussian random field with the same power spectrum as the 2dfgrs and find , contradictory to results for the 3d genus of other samples , that the amplitude of the grf genus curve is slightly lower than that of the 2dfgrs .
this could be due to a a feature in the current data set or the 2d genus may not be as sensitive as the 3d genus to non - linear clustering due to the averaging over the thickness of the slice in 2d . |
in this supplemental material , we present the derivation details of the anisotropic chern - simons term through opposite coupling . as mentioned in the main text , it comes from the summation of the three diagrams with one and two s .
let @xmath122 denote the summation . applying feynman rules we have @xmath123,\end{aligned}\ ] ] noting that the three diagrams are equal .
then the above express can be rewritten as @xmath124 where @xmath125.\ ] ] introducing feynman s parameters @xmath126 and @xmath127 and replacing @xmath107 by @xmath128 , the above expression can be simplified as @xmath129.\end{aligned}\ ] ] as the @xmath11 gauge symmetry is fundamental , we adopt the dimensional regularization scheme for @xmath107 .
accordingly , @xmath107 is decomposed as @xmath130 in the above notation , @xmath109 is still in the four - dimensional spacetime , but @xmath110 is in the extended spacetime , anti - commuting with @xmath111 and commuting with @xmath112 .
it is noted that @xmath131 and @xmath132 are still in the four - dimensional spacetime without extension .
the only non - vanishing terms within the trace are @xmath133\gamma^{\lambda}\slashed l_{\perp}\gamma^{\nu}\slashed l_{\perp}\right\ } \\ + & \mathrm{tr}\left\ { \gamma^{5}\gamma^{\mu}\slashed l_{\perp}\gamma^{\lambda}\slashed l_{\perp}\gamma^{\nu}[x\slashed k+(1-y)\slashed p]\right\ } \\
+ & \mathrm{tr}\left\ { \gamma^{5}\gamma^{\mu}\slashed l_{\perp}\gamma^{\lambda}(x\slashed k - y\slashed p)\gamma^{\nu}\slashed l_{\perp}\right\ } .
\end{array}\ ] ] their summation can be simplified as @xmath134l_{\perp}^{2},\ ] ] using the identity @xmath135 thus @xmath136\\ & & \int\frac{d^{d}l}{(2\pi)^{d}}\frac{l_{\perp}^{2}}{(l^{2}-\delta)^{3}}\\ & = & -4\epsilon^{\mu\lambda\nu\alpha}(k_{\alpha}-p_{\alpha})\int\frac{d^{d}l}{(2\pi)^{d}}\frac{l_{\perp}^{2}}{(l^{2}-\delta)^{3}}.\end{aligned}\ ] ] using @xmath137 , the integration above is evaluated as @xmath138 so we have @xmath139 now we are ready to evaluate @xmath122 as @xmath140 thus the effective theory also has the anisotropic chern - simons term @xmath141 | we develop a topological theory for disordered weyl semimetals in the framework of gauge invariance of replica formalism and boundary - bulk correspondence of chern insulators .
an anisotropic topological @xmath0-term is analytically derived for the effective non - linear sigma model .
it is this nontrivial topological term that ensures the bulk transverse transport of weyl semimetals to be robust against disorders .
moreover , we establish a general diagram that reveals the intrinsic relations among topological terms in the non - linear sigma models and gauge response theories respectively for @xmath1-dimensional topological insulators , @xmath2-dimensional chiral fermions , @xmath2-dimensional chiral semimetals , and @xmath3-dimensional topological insulators with @xmath4 being a positive integer .
_ introduction _ recently , weyl semimetal ( wsm ) has been attracting more and more attention due to its interdisciplinary interest from anomalous transport in condensed matter physics @xcite and quantum field theory anomalies as well as a topological character of its gapless modes @xcite .
weyl fermion as the quasi - particle of wsm has a definite chirality , left - handed or right handed , depending on its sign of spin polarization along the momentum direction@xcite .
according to the no - go theorem@xcite , gapless weyl points in a wsm appear as left and right - handed pairs in the momentum space .
since all the energy bands of a system with both time - reversal symmetry(trs ) and inversion symmetry(is ) are doubly degenerate for both chiral modes , a wsm can only be realized in a system breaking trs and/or is symmetry .
here we focus essentially on the simplest model of wsm with two weyl points , whose low - energy effective hamiltonian reads @xmath5 where @xmath6 s are pauli matrices and @xmath7 is the displacement of weyl points from the origin of the momentum space . under semi - classical approximation ,
the spatial dependence of @xmath8 is assumed to be adiabatic .
the model does not have trs @xcite .
each weyl point has a nontrivial topological charge @xcite , namely the chern number on the gapped two - dimensional sphere enclosing a single weyl point in @xmath9 space is @xmath10 , which leads to topological terms in the @xmath11 response of this model @xcite . for real materials ,
disorders are normally unavoidable , and can be studied by the replica method , where a non - linear sigma model ( nl@xmath6 m ) encodes fluctuations of nambu - goldstone modes that are on a curved manifold@xcite . particularly for topological materials ,
the global topology of the target manifold may have essential effects , such that topological terms in the nl@xmath6 m are significantly important for describing physics of the system under disorders @xcite , which highly motivates us to develop a general theory for disordered wsms .
-th unit of a whole topological family of class a @xcite . for each node ,
the upper , below - left , and below - right boxes denote the model , the topological term in nl@xmath6 m , and gauge theory , respectively .
the dashed arrow is effectively composed by the three successive solid arrows . nodes : ti(topological insulator ) , @xmath12(@xmath0-term ) , cs(chern - simons term ) , cf(chiral fermion ) , wzw(wzw - term ) , cc(chern character ) , csm(chiral semimetal ) , a-@xmath12(anisotropic @xmath12 ) and a - cs(anisotropic cs ) .
arrows : bbc(boundary - bulk correspondence ) , oc(opposite coupling ) and dr(dimension reduction ) .
notably , the thick blue - arrow represents a key step that deduces the central result of this work .
[ diagram ] ] in this letter , for the model of eq .
( [ model ] ) under disorders and preserving no discrete symmetry , we derive for the first time an anisotropic topological term in the action of the nl@xmath6 m as @xmath13 where @xmath14 is the completely anti - symmetric tensor , @xmath15 denotes the @xmath16-th component of @xmath7 , and @xmath17 stands for a sigma field to be specified later .
we refer to this term as an anisotropic @xmath0-term , as it is just a usual @xmath0-term on the plane perpendicular to @xmath7 for a constant @xmath7 @xcite .
actually this term can be generalized to any chiral semimetal ( csm ) with two chiral points separated in the @xmath9 space of odd dimensions @xcite .
we also reveal that this new anisotropic @xmath0-term is associated with the so - called opposite coupling in csm , which plays a key role in a relationship diagram of topological terms for a family of topological matter , as illustrated in fig.([diagram ] ) , where each topological term in the nl@xmath6 m has a counterpart in the @xmath11 gauge response theory .
_ disorder model _ we now consider that the wsm , eq.([model ] ) , is subjected to a white - noise and random scalar potential @xmath18 , namely @xmath19 assuming that the probability density of @xmath20 is gaussian , we have @xmath21 and @xmath22 with @xmath23 indicating the strength of the random potential .
after applying the replica scenario @xcite and averaging over the random potential , the disordered system may be described by a lagrangian @xmath24 where the spinor is decomposed into retarded and advanced space , namely @xmath25 , and @xmath26 .
the pauli matrix @xmath27 acts in this space , and @xmath28 is an infinitesimal positive number .
the subscript as the replica index ranges from @xmath29 to @xmath30 with repeated one being summed over .
the four - operator term comes from the averaging over the random potential .
since our system is in the topological class a , we introduce the sigma field @xmath31 to decouple the four - operator term @xcite , @xmath32 where the greek subscripts indicate both replica and retarded - advanced indexes .
accordingly the partition function can be transformed to be in the form , @xmath33 where @xmath34 @xmath35 is determined by the consistence equation @xmath36 $ ] with @xmath37 being the short - distance cutoff .
assuming that @xmath38 is small and @xmath39 , the dynamic terms of the effective nl@xmath6 m are given by @xmath40 it is observed that the operator in the determinant is diagonal in the chirality space , recalling that @xmath41 , where @xmath42 and @xmath43 are weyl spinors with opposite chirality .
this means the effective nl@xmath6 m @xmath44 for @xmath45 is a summation of the nl@xmath6ms for the left and right - handed chiral components , namely @xmath46=s_{-,eff}[q,-\mathbf{b}]+s_{+,eff}[q,\mathbf{b } ] , \label{sumrule}\ ] ] where the signs in front of @xmath7 indicate that it couples oppositely to the left and right - handed ones .
we below derive @xmath47 separately . _ boundary - bulk correspondence _ before deriving @xmath48 $ ] , let us introduce the wess - zumino - witten term ( wzw - term ) for a single weyl point at the center of @xmath9 space with disorders but without any anti - unitary symmetry , from a viewpoint of boundary - bulk correspondence(bbc ) @xcite . for a @xmath1-dimensional
( d ) ti in the class a with a nontrivial chern number @xmath49 in its bulk @xcite , there are @xmath49 flavors of chiral fermions with the same chirality on each of its @xmath2d boundaries @xcite .
the nl@xmath6 m of the @xmath1d ti under disorders has a @xmath0-term @xcite . in the infrared limit the behavior of the ti
is entirely determined by its gapless boundary , and the coupling constant of the @xmath0-term becomes the chern number @xmath49 under the renormalization flow @xcite . to be concrete , for a 4d ti , the @xmath0-term of the nl@xmath6
m is given by @xmath50 viewing from the boundary of the ti , this implies that there exists a wzw - term at level @xmath49 in the nl@xmath6 m of the chiral fermions on the boundary @xcite . for the case of eq.([4dtheta ] ) , the corresponding wzw - term is found to be @xmath51 where @xmath52 $ ] is the extending parameter , and @xmath53 on the @xmath54 is extended continuously to @xmath55 with @xmath56 and @xmath57 being constant@xcite . in this case , @xmath58 can be regarded as the parameter of the radial direction if the geometry of the 4d ti is a disc @xmath59 . here
@xmath60 is the total @xmath61-type topological charge of the boundary gapless modes , which is defined as the chern number on the gapped sphere enclosing these gapless points in momentum space , and it is equal to the bulk chern number @xmath49 according to an established index theorem @xcite .
inherently , the @xmath61-type topological charge @xmath62 also implies the wzw - term to be at level @xmath60 in the nl@xmath6 m .
since this wzw - term has a topological character with a non - perturbative discrete coupling constant , gapless chiral modes have a topologically protected finite conductance , free from the anderson localization @xcite . as the counterpart of the above results for nl@xmath6 m
, the bbc can also be used to deduce the chern character ( cc ) term in the gauge response of @xmath2d chiral fermions from the chern - simons(cs ) term of the gauge response of @xmath1d tis , since the cs term is not gauge invariant on a manifold with boundary , leading to the boundary cc terms . _ the coupling of disordered weyl fermions with a gauge field _
if @xmath63 , @xmath64 just corresponds to chiral fermions under disorders . according to the boundary - bulk correspondence , the nl@xmath6 m for a model @xmath65 with only one gapless point contains a wzw - term , @xmath66=\frac{im}{128\pi}\int d\tau d^{3}x\;\epsilon^{\mu\nu\rho\lambda}\mathrm{tr}\tilde{q}\partial_{\mu}\tilde{q}\partial_{\nu}\tilde{q}\partial_{\rho}\tilde{q}\partial_{\lambda}\tilde{q},\ ] ] where @xmath67 is the topological charge of the gapless modes in the model , raising from @xmath68=\int d^{3}x\;\chi^{\dagger}\left(\omega-\mathcal{h}+i\eta\tau_{3}^{ra}-v(\mathbf{r})\right)\chi .\ ] ] in the present case , @xmath69 correspond to the left and right - handed fermions , respectively .
so it is clear that the wzw - terms are cancelled if ever @xmath63 , leading to vanishing topological terms and corresponding to a dirac hamiltonian under disorders .
when @xmath7 is nonzero , we can regard it as a chiral gauge field coupling oppositely to the left- and right - handed fermions .
let us consider @xmath70 coupled with a @xmath11 gauge field @xmath71 under disorders , noting that @xmath72 can be treated similarly .
after application of the replica method , the lagrangian reads @xmath73 it is found that the @xmath11 gauge transformation can be made for each component of @xmath43 independently , namely the action is invariant under the transformation , @xmath74 and @xmath75 with the superscript ` @xmath76 ' being the index of the retarded - advanced space . accordingly in the non - linear sigma version , @xmath77 transforms under the gauge transformation as @xmath78 in the absence of gauge field , the nl@xmath6
m is @xmath79=\frac{1}{\lambda_+}\int d^{3}x\:\mathrm{tr}(\partial_{j}q\partial_{j}q)+\gamma_{+}[q],\ ] ] where @xmath80 determined by the microscopic details of the weyl point is proportional to the longitudinal conductivity .
our strategy is to find the minimal coupling of the nl@xmath6 m with the @xmath11 gauge field , which is invariant under the above gauge transformation . to simplify our calculation but without loss of generality
, we consider a specific gauge transformation as @xmath81 and @xmath82 which are the opposite constants in advanced and retarded spaces , respectively . following a seminar work on current algebra of witten @xcite
, we highlight our derivations below .
the infinitesimal variation of @xmath17 is @xmath83.\ ] ] @xmath84 operates on retarded and advanced spaces , and hereafter we drop the superscript for simplicity . since the minimal coupling for the ordinary term is readily obtained by the substitution , @xmath85 $ ] , we thus focus only on the wzw - term .
the variation of @xmath86 under the infinitesimal local gauge transformation is @xmath87 where the exterior derivative and wedge product have been used for brevity .
thus we have @xmath88 we may expect that the coupling takes the form @xmath89 . however , since the current @xmath90 is not gauge invariant , additional terms are needed to cancel its variation under gauge transformation , which turns out to be @xmath91 . as a result ,
the total gauge invariant action for the @xmath92 case is given by @xmath93 _ opposite coupling _ we are now ready to obtain the nl@xmath6 m of the wsm .
in the wsm @xmath94(@xmath95 ) is coupled with @xmath96(@xmath97 ) .
thus from eqs.([sumrule ] ) and ( [ em ] ) , it is found that @xmath46=\frac{1}{\lambda}\int d^{3}x\:\mathrm{tr}(\partial_{j}q\partial_{j}q)+s_{a\theta}[q ] , \label{seff}\ ] ] where @xmath98 , recalling that @xmath99 , as our most main result , is given by eq.([atheta ] ) .
also note that the first term in eq.([seff ] ) is actually a normal action that accounts for usual non - topological properties .
a physical meaning of the anisotropic @xmath0-term becomes clear if @xmath7 is constant .
intuitively , a 2d slice at any @xmath100 along @xmath7 direction may be viewed as a 2d ti with unit chern number , which is accompanied with the bulk transverse conductivity and the edge chiral gapless modes ( see the latter discussion around eq.([2dtheta ] ) ) .
it is interesting to note that the transverse conductivity in the plane perpendicular to @xmath7 is proportional to the magnitude of @xmath7 , being completely independent of the disorder strength @xmath101 .
the anisotropic form in the bulk is also consistent with the edge currents traveling perpendicular to @xmath7 , enabling them to have topological protection similar to that of the wzw - term . on the other hand ,
it is seen from the derivation that the wzw - terms of two weyl points are cancelled , which relies only on the fact that the two weyl points have opposite topological charges , independent of their positions in the @xmath9-space .
when the wzw - term is absent after the opposite coupling , its corresponding topological protection is lost , consistent with the fact that disorders may mix the two weyl points leading to localization@xcite . however , remarkabally , the new nontrivial anisotropic @xmath0-term emerges , such that it embodies the remaining anisotropic topological protection originated from topological charges after the global cancellation , analogous to the net electric field generated from an electric dipole .
at this stage , we elaborate how to derive the a - cs term in fig.([diagram ] ) ( see supplemental material for details @xcite ) , from a trick of opposite coupling . treating @xmath7 as a gauge field coupling oppositely to the two weyl points
, the corresponding lagrangian reads @xmath102 , where @xmath7 is promoted to be a space - time vector @xmath103 with @xmath104 corresponding to deviation in energy , and notations about dirac matrices are consistent with those in @xcite .
there are two types of vertexes , @xmath105 and @xmath106 .
the a - cs term is given by the three - vertex diagrams with two s and one , shown in fig.([feynman ] ) .
as the @xmath11 gauge symmetry is fundamental , we adopt the dimensional regularization scheme for the internal momentum @xmath107 in the loop .
accordingly @xmath107 is decomposed as @xmath108 . in the above notations
, @xmath109 is still in the four - dimensional spacetime , but @xmath110 is in the extended infinitesimal dimensions , anti - commuting with @xmath111 and commuting with @xmath112 @xcite .
the three diagrams contribute equally , and after a dramatic cancellation only a finite term is left similar to the derivation of the axial anomaly @xcite , leading to the anisotropic cs term , @xmath113 ] _ dimension reduction _ the last arrow process to be established is the dimension reduction from @xmath2d csm to @xmath3d ti in class a. first it is observed that if @xmath7 is a constant vector in the model ( [ model ] ) , along the direction of @xmath7 , we may regard the 3d system as a collection of 2d systems perpendicular to @xmath7 in @xmath9 space , and thus the 2d systems gapped over @xmath100 are 2d tis with unit chern number and the others outside this range are trivial , except at the two gapless points . setting @xmath114 constant and all fields independent of @xmath115 in eqs.([atheta ] ) and ( [ acs ] ) , the dimension reduction gives @xmath116 and @xmath117 since @xmath118 correspond to a collection of unit topological insulators ( chern insulators ) , divided by the dimension constant @xmath119 , the above equations give the well - known @xmath0-term and cs term for 2d tis , @xmath120 since the above treatment is applicable for any integer @xmath121 , the dimension reduction(dr ) from @xmath2d csm to @xmath3d ti in fig.([diagram ] ) has been completed .
the dimension reduction actually illustrates the correspondence between the anisotropic @xmath0-term of eq.([atheta ] ) , and the anisotropic cs term of eq.([acs ] ) , recalling that the cs term for 2d ti indicates the transverse conductivity and the @xmath0-term implies the stability of this transverse transportation under disorders .
_ remark _ we now make comments about the essence of emergent topological terms . in quantum field theory , this emergence is usually related to some quantum anomaly , where two regularization schemes with different symmetries are contradictory @xcite . from a viewpoint of band theory , these contradictions are usually originated from nontrivial topological configurations of the berry fiber bundle . the coupling constant of a topological term can formally be expressed as a topological invariant of the berry fiber bundle , which affects directly the transport properties @xcite , and thus its nontrivial topological configurations may account for anomalous transport in quantum anomalies .
this work was supported by the grf ( grant nos .
hku7045/13p and hku173051/14p ) and the crf ( hku8/11 g ) of hong kong .
references s. murakami , new j. phys .
* 9 * , 356 ( 2007 ) .
x. wan , a. m. turner , a. vishwanath , s. y. savrasov , phys .
rev .
b * 83 * , 205101 ( 2011 ) . w. w. krempa and y .- b .
kim , phys .
rev .
b * 85 * , 045124 ( 2012 ) . a.a .
burkov and l. balents , phys .
rev .
lett . * 107 * , 127205 ( 2011 ) .
a.a .
burkov , m. d. hook , and l. balents , phys .
rev .
b * 84 * , 235126 ( 2011 ) .
a.a .
zyuzin , s. wu , a.a .
burkov , phys .
rev .
b * 85 * , 165110 ( 2012 ) . g. b. halasz and l. balents , phys . rev .
b 85 , 035103 ( 2012 ) .
f. haldane , arxiv:1401.0529 ( 2014 ) .
s. y. xu , et al . , arxiv:1502.03807 . c. zhang , et al .
, arxiv:1503.02630 .
s. adler , phys . rev . * 177 * , 2426 ( 1969 ) . j.s .
bell and r. jackiw , nuovo cimento * 60a * , 4 ( 1969 ) .
g. e. volovik , _ the universe in a helium droplet _
( clarendon , oxford , 2003 ) .
s. m. carroll , g. b. field , and r. jackiw , physical review d * 41 * , 1231 ( 1990 ) .
r. jackiw and s .- y .
pi , phys .
rev .
d * 68 * , 104012 ( 2003 ) .
c .- x .
liu , p. ye , x .-
l .
qi , phys .
rev .
b * 87 * , 235306 ( 2013 ) .
a. a. zyuzin and a.a .
burkov , phys .
rev .
b * 86 * , 115133 ( 2012 ) . d.t .
son and n. yamamoto , phys . rev . lett . * 109 * , 181602 ( 2012 ) .
a. g. grushin , phys .
rev .
d * 86 * , 045001 ( 2012 ) .
m. m. vazifeh and m. franz , phys .
rev .
lett . * 111 * , 027201 ( 2013 ) .
m. e. peskin and d. v. schroeder , _ an introduction to quantum field theory _ ( westview press , 1995 ) .
h. b. nielsen and m. ninomiya , phys .
lett .
b * 105 * , 219 ( 1981 ) .
although the displacements of two weyl points are assumed to be placed oppositely in the momentum space due to is for simplicity , our main results are valid for a general case without is .
y. x. zhao and z. d. wang , phys .
rev . lett.*110 * , 240404 ( 2013 ) .
a. m. m. pruisken and l. schafer , phys .
rev .
lett * 46 * , 490 ( 1981 ) .
a. m. m. pruisken and l. schafer , nucl .
phys .
b * 200 * , 20 ( 1982 ) .
f. wegner , z. phys b * 35 * 207 ( 1979 ) ; z. phys b * 35 * , 209 ( 1980 ) and references therein .
a. m. m. pruisken , nucl .
phys .
b * 235 * , 277 ( 1984 ) .
a. p. schnyder , s. ryu , a. furusaki , and a. w. w. ludwig , phys .
rev .
b * 78 * , 195125 ( 2008 ) ; aip conf . proc . * 1134 * , 10 ( 2009 ) .
s. ryu , a. p. schnyder , a. furusaki and a. w. w ludwig , new j. phys .
* 12 * , 065010 ( 2010 ) .
s. ryu , c. mudry , h. obuse , and a. furusaki , phys .
rev .
lett . * 99 * , 116601 ( 2007 ) .
p. m. ostrovsky , i.v .
gornyi , and a. d. mirlin , phys .
rev .
lett . * 98 * , 256801 ( 2007 ) .
here the term ` chiral point ' refers to a gapless point in any odd dimensional momentum space such that the topological charge is nontrivial .
m. r. zirnbauer , j. math . phys . * 37 * , 4986 ( 1996 ) . a. altland and m. r. zirnbauer , phys .
rev .
b * 55 * , 1142 ( 1997 ) .
e. witten , nucl .
phys .
b. * 223 * , 422 ( 1983 ) .
m.z.hasan and c. l.kane , rev .
mod .
phys . * 82 * , 3045 ( 2010 ) .
x. l. qi and s. c. zhang , rev .
mod .
phys * 83 * , 1057 ( 2011 ) . for a direct mathematical derivation of eq.([wzw ] )
, please see ref.@xcite .
y. x. zhao and z. d. wang , phys .
rev .
b * 89 * , 075111 ( 2014 ) .
e. witten , commun .
math .
phys .
* 92 * , 455 ( 1984 ) .
z. huang , d. p. arovas , and a. v. balatsky , new j. phys .
* 15 * , 123019 ( 2013 ) . supplemental material . g. t hooft and m. j. g. veltman , nucl . phys .
b * 44 * , 198 ( 1972 ) .
d. xiao , m .- c .
chang , and q. niu , rev .
mod .
phys . * 82 * , 1959 ( 2010 ) .
y. x. zhao and z. d. wang , arxiv:1504.02373 . |
for an affine real smooth algebraic hypersurface @xmath2 with real part @xmath3 there is an universal inequality between the total curvatures of the real part @xmath4 and the one of @xmath5 ( @xcite ) , similar to the `` smith s inequality '' between the sums of mod.2 betti numbers . in this paper
we prove an analogous result in the tropical setting ; it turns out that in the non - singular tropical case , this inequality becomes an equality ; this fact is true in the algebraic world only up to any positive @xmath6 and for special maximal varieties . for plane
algebraic curves the only cases when the equality holds are the line , the ellipse and the parabola .
+ let us describe briefly the main results of the paper .
+ if @xmath0 is a tropical hypersurface defined by a polynomial with coefficients in the field of real puiseux series , it has a real part @xmath1 ( see subsection [ subsec : real - tropical - hypersurfaces ] ) .
using the fact that @xmath0 ( resp .
@xmath1 ) is limit of amoebas ( resp .
real amoebas ) , we define the total curvature of @xmath0 ( resp .
@xmath1 ) by using the total curvature of amoebas and passing to the limit .
+ for the real part @xmath1 , which is a polyhedral manifold , we also define its total curvature geometrically ( as a polyhedral hypersurface ) , and call it the `` polyhedral total curvature '' .
+ non - singular tropical hypersurfaces are those whose dual subdivision is primitive ; they are the tropical counterpart of primitive t - hypersurfaces of viro s patchworking ( see @xcite and @xcite ) .
the main results we prove about these notions are the following : 1 . the fact that the notions of total curvature and polyhedral total curvature coincide for real non - singular tropical hypersurfaces ( section [ sec : polyhedral - curvature ] ) .
2 . a universal inequality between the total curvatures of @xmath0 and @xmath1 ( subsection [ subsec : real - tropical - curvature ] ) based on a similar inequality between the logarithmic gaussian curvatures of the complex and real parts of a real algebraic hypersurface ( section [ sec : amoebas - curvature ] ) .
3 . in the non - singular case ,
this inequality turns out to be an equality ( subsection [ subsec : real - tropical - curvature ] ) .
4 . a `` gauss - bonnet s style '' formula for the total curvature of a non - singular ( complex ) tropical hypersurface ( subsection [ subsec : gauss - bonnet ] ) . we would like to thank the referee for their useful comments .
the structure of the paper is the following : section 2 is a preliminary one : it introduces notation and basic properties of tropical and real tropical hypersurfaces .
section 3 treats the case of amoebas , using the `` logarithmic curvature '' to define their curvature .
section 4 describes how we can define the tropical total curvature passing to the limit from the one of amoebas , both in the real and complex cases , and contains the main results of the paper .
section 5 is devoted to defining directly the `` polyhedral curvature '' for a real non - singular tropical hypersurface and to proving that this technique gives the same notion of total curvature than the previous one in the tropical non - singular case .
section 6 gives some complements and applications , in particular a gauss - bonnet s style formula for non - singular tropical hypersurfaces , comparing the euler characteristic of a generic complex hypersurface in @xmath7 and the total curvature of its tropicalisation .
in all the paper , @xmath8 is considered with its canonical orientation , @xmath9 will be the volume of the unit sphere @xmath10 , and we will set @xmath11 .
+ we have then : @xmath12 let @xmath13 be a smooth oriented hypersurface , @xmath14 the gauss map @xmath15 , where @xmath16 is a non - zero normal vector to @xmath17 at @xmath18 .
we define the curvature function @xmath19 on @xmath17 as the jacobian of @xmath20 .
the curvature of a measurable set @xmath21 is the integral @xmath22 of the curvature function on @xmath23 .
the * total curvature * of @xmath24 is then by definition : @xmath25 where @xmath26 is the canonical euclidean volume form on @xmath13 .
it clearly satisfies the following equality : @xmath27 where @xmath28 is the canonical volume form on @xmath29 and assuming @xmath30 is almost everywhere finite .
one can think of the total curvature as the volume of im@xmath31 taken `` with multiplicities '' , the multiplicity of a point @xmath18 being the cardinality of the fiber @xmath32 .
+ if now @xmath33 is a smooth analytic complex hypersurface , one may define its curvature as the `` lipschitz- killing '' curvature @xmath34 , where @xmath35 is the canonical volume form on @xmath33 and @xmath36 the curvature function . + another way , due to milnor , to define the function @xmath37 is the following .
let @xmath38 be the `` complex gauss map '' @xmath39 $ ] , where @xmath40 is the complex normal vector to @xmath41 at @xmath18 .
one then has : @xmath42 where @xmath43 is the volume form on @xmath44 ( see @xcite ) ; note that @xmath45 is a positive function on @xmath41 . + the following inequality is proved in @xcite in the algebraic case : @xmath46 this paper is devoted to defining similar notions and proving similar results in the tropical case .
in particular we prove the same type of inequality for the logarithmic gaussian curvatures which are the natural curvatures in the tropical setting and study its sharpness . in order to fix notations and definitions we use in this text , we recall briefly basic notions in tropical geometry . we use the following notation : the scalar product is written @xmath47 ; for @xmath48 and @xmath49 we write @xmath50 ; the set of vertices of a polytope @xmath51 is denoted @xmath52 .
we tend to identify a point and its coordinate vector when it makes the notation less cumbersome .
we consider the tropical semi - field @xmath53 , where the tropical operations are defined by @xmath54 and @xmath55 .
a tropical polynomial @xmath56 $ ] is of the form @xmath57 where @xmath58 denotes the set of exponents of @xmath59 .
the newton polytope of @xmath59 will be denoted by @xmath60 .
given a tropical polynomial @xmath59 and a point @xmath61 , the @xmath62-initial part of @xmath59 is @xmath63 a tropical polynomial @xmath59 determines naturally a polyhedral subdivision of @xmath8 whose cells are formed of the sets of points defining the same initial part of @xmath59 .
dually , the set of cells @xmath64 realises a regular subdivision of @xmath60 .
it is dual to the subdivision of @xmath8 determined by @xmath59 in the following usual sense .
if @xmath65 is a cell of the subdivision of @xmath8 determined by @xmath59 there exists a unique dual cell @xmath66 of @xmath67 .
it satisfies @xmath68 and @xmath65 and @xmath66 generate orthogonal affine subspaces .
we say that a tropical polynomial @xmath59 is * generic * if @xmath67 is simplicial , * non - singular * if all the maximal dimensional cells of @xmath67 are primitives simplices , and * primitive * if @xmath60 is a primitive simplex of dimension @xmath69 .
the * corner locus * of a tropical polynomial @xmath70 $ ] is the set of points @xmath61 where the value of @xmath59 is attained by at least two of its monomials . or equivalently , the set of points contained in cells of dimensions at most @xmath71 . in this text ,
such a set will be called a * tropical hypersurface*. [ def : tropical - hypersurface]let @xmath70 $ ] . the set @xmath72 is by definition the tropical hypersurface defined by @xmath59 .
we denote by @xmath73 the @xmath74-dimensional cells of @xmath75 and more generally @xmath76 will always denote the @xmath74-dimensional cells of the natural subdivision of a piecewise linear variety @xmath77 .
a formal series @xmath78 is a * locally convergent generalised puiseux series * if @xmath79 is a well - ordered set , @xmath80 , and the series is convergent for @xmath81 small enough .
denote by @xmath82 the set of all locally convergent generalised puiseux series .
it is an algebraically closed field of characteristic @xmath74 .
a series @xmath83 is said to be * real * if all its coefficients are real numbers .
we denote by @xmath84 the subfield of @xmath82 composed by the real series . since the coefficients of a polynomial @xmath85 $ ] are locally convergent near @xmath74 , any polynomial over @xmath82 ( resp .
@xmath86 ) can be thought as a one parametric family of complex ( resp .
real ) polynomials . for any @xmath87 , the complex ( resp .
real ) polynomial @xmath88 is the polynomial resulting of the evaluation of the coefficients at @xmath89 . by hypothesis
, the set of exponents of an element of @xmath90 has a first element .
the map that sends each element of @xmath90 to the first element of its set of exponents and @xmath74 to @xmath91 is a non - archimedean valuation .
we denote by @xmath92 the opposite of such a map . in other words , @xmath93 maps @xmath94 to @xmath95 and @xmath74 to @xmath96 .
the map @xmath92 extends naturally to the map @xmath97 by applying @xmath92 coordinate - wise .
the image of a variety @xmath98 under the map @xmath99 is called the * non - archimedean amoeba * of @xmath0 . given a polynomial @xmath85
$ ] one can associate a tropical polynomial .
this map is called tropicalisation and acts as follows : if @xmath100 , the * tropicalisation * of @xmath101 is the tropical polynomial @xmath102 kapranov s theorem establishes that the non - archimedean amoeba of @xmath101 and the tropical hypersurface @xmath103 coincide . given a tropical hypersurface @xmath104 and a polynomial @xmath85 $ ] , we will say that @xmath101 _ realises _ @xmath0 if @xmath105 .
real tropical hypersurfaces are very closely related to viro patchworking ( see @xcite and @xcite ) .
a description can be found in @xcite and one can look at @xcite pp .
25 and 37 , @xcite , and the appendix of @xcite in the case of amoebas for further details .
we recall some definitions here for the convenience of the reader .
let @xmath106 $ ] be a * real * polynomial defined over the field of real puiseux series .
let @xmath107 be the map sending a real puiseux series @xmath108 to @xmath109 and @xmath110 be the map defined by @xmath111 coordinate - wise .
the map @xmath111 restricts to @xmath112 which sends @xmath108 to @xmath113 and we denote @xmath114 the corresponding restriction of @xmath115 . for any @xmath116 we will call * orthant * of the torus over the field of puiseux series and denote by @xmath117 the preimage of @xmath118 under @xmath119 .
as in the case of @xmath120 an orthant is thus a choice of sign for each coordinate . the map @xmath115 allows to consider the collection of images under @xmath121 of each orthant @xmath117 .
[ def : real - tropical - hypersurface-1 ] a * real tropical hypersurface * @xmath122 is the data of @xmath123 for a polynomial @xmath106 $ ] .
the * real part * of the real tropical hypersurface is @xmath124 .
let us now compare this definition to the following patchworking procedure .
let @xmath59 be a generic tropical polynomial , and @xmath125 be a distribution of signs .
let @xmath126 be the canonical basis of @xmath127 . for @xmath128
, let @xmath129 be the symmetry mapping @xmath130 to @xmath131 .
the map @xmath132 only depends on the reduction modulo @xmath133 of the coordinates of @xmath134 , so we will indifferently use the notation @xmath132 for @xmath135 or @xmath136 . define the symmetrised distribution of sign @xmath137 by @xmath138 .
the maps @xmath139 are involutions on the set @xmath140 of sign distributions on @xmath58 .
they define an action of @xmath141 on sign distributions .
we will consider this action via the maps @xmath139 below in subsection [ sec : polyhedral - curvature ] ( for example in proposition [ prop : transitivity ] ) .
[ def : real - tropical - hypersurface ] let @xmath59 be a generic tropical polynomial , and @xmath125 a distribution of signs .
the * patchworked real tropical hypersurface * @xmath142 is the data of @xmath75 and @xmath143 .
the * real part * @xmath144 of @xmath142 is a subset of @xmath145 consisting of relevant symmetric copies of cells of @xmath75 .
namely for each given @xmath136 and @xmath65 a cell of @xmath75 , @xmath146 if and only if @xmath147 here and below we identify the set of elements of @xmath148 with @xmath149 or @xmath150 depending on which is more convenient . with this identification ,
the above equality @xmath151 just means that not all vertices of @xmath66 carry the same sign .
the set @xmath152 is either empty or a polyhedral hypersurface .
one can identify the real tropical hypersurfaces @xmath142 and @xmath153 one being obtained from the other by simultaneously reversing all signs ( _ i.e. _ , multiplying @xmath154 by @xmath155 for all @xmath156 ) .
the real parts @xmath144 and @xmath157 are the same . if @xmath158 $ ] is a polynomial with real series coefficients , we associate to each monomial the sign of the first term of its coefficient . in particular
this defines a natural sign distribution @xmath159 at the vertices of the subdivision @xmath160 .
the proposition below is a direct consequence of viro s patchworking .
[ prop : patchworking ] let @xmath161 $ ] be a polynomial such that @xmath162 is generic .
for @xmath163 let @xmath164 be in @xmath165 . if for every such pair @xmath166 the identity @xmath167 holds only if @xmath164 is a vertex of @xmath160 then , @xmath168 , the real part of the real tropical hypersurface coincides with the real part of the patchworked tropical hypersurface .
in order to recover a hypersurface in @xmath169 from @xmath144 one uses the map @xmath170 defined by @xmath171 where the exponential is applied component - wise .
[ exa : real - conic ] let @xmath172 be a second degree tropical polynomial @xmath173 such that @xmath174 is the tropical conic represented on figure [ fig : newton ] a ) and @xmath143 be the signed distribution shown on figure [ fig : newton ] b ) . on figure [ fig : real - conic ] , we have drawn the real part @xmath175 of the real tropical conic @xmath176 in each of the four quadrants @xmath177 corresponding to the four elements of @xmath178 . the axes are the dashed blue lines .
the dotted - dashed black segments are the parts of the symmetric copies @xmath179 of @xmath174 which do not belong to @xmath175 .
the real part @xmath175 is depicted in plain thick red . in each region of @xmath180 we indicated the sign of each vertex in the sign distribution @xmath181 . on figure
[ fig : exp ] we represented the image of @xmath175 under the map @xmath182 ( still in plain thick red ) .
[ cols="^,^ " , ] after applying the map @xmath182 . ]
when no confusion is possible we will abuse notation and terminology and write @xmath142 instead of @xmath144 and `` real tropical hypersurface '' instead of `` real part of the real tropical hypersurface '' . for a vertex @xmath183 of @xmath142 ,
when @xmath184 is a vertex of @xmath144 , it will be implicitly denoted by @xmath185 .
notice that if all vertices of @xmath186 have the same sign , @xmath184 is not in @xmath144 .
let @xmath187 $ ] be a polynomial defining a non - singular hypersurface in @xmath188 .
for @xmath189 , @xmath190 , @xmath191 $ ] defines a non - singular hypersurface @xmath192 .
we call it a real algebraic variety because @xmath88 is defined over @xmath193 and denote by @xmath194 the real part of @xmath195 . + we will set @xmath196 for the newton polygon of @xmath101 ( or @xmath88 for @xmath197 ) .
we denote by @xmath198 the map from @xmath7 to @xmath8 that sends @xmath199 to @xmath200 where @xmath201 is the base @xmath89 logarithm .
[ def : amoebas ] set @xmath202 .
\1 ) for @xmath203 , 0<t \ll 1 $ ] , @xmath204 $ ] , the amoeba of @xmath195 is @xmath205 \2 ) for @xmath206 , 0<t \ll 1 $ ] , @xmath207 $ ] , the real amoeba of @xmath208 is @xmath209 in each orthant @xmath210 , the map @xmath211 is a diffeomorphism onto @xmath8 ; then we may define the gauss map `` @xmath212 '' by taking the gauss map for the image of each orthant ( then for some points of the amoeba , the `` map '' @xmath213 may be multivalued ) .
we then have the following diagram : @xmath214^{\gamma_t^{{\mathbb r } } } \ar[d]^{\operatorname{{\textup{log}}}_t } \\
\mathcal{a}^{{\mathbb r}}(f_t ) \ar[r ] _ { g_t } & { { \mathbb r}}{{\mathbb p}}^n}\end{aligned}\ ] ] where @xmath213 is the gauss map and @xmath215 the logarithmic gauss map , defined as : @xmath216.\ ] ] [ def : amoebas - real - curvature ] according to diagram ( [ diag : log - gauss ] ) , the total curvature of the amoeba @xmath217 is @xmath218 in the complex case , contrary to the real case , the amoeba @xmath219 is not in general an immersed manifold , therefore there is no natural definition of a gauss map @xmath220 .
however , the logarithmic gauss map @xmath221 : @xmath222 is meaningfull , and we have the following diagram : @xmath223^{\gamma_t } \ar[d]^{\operatorname{{\textup{log}}}_t } \\
\mathcal{a}(f_t ) & { { \mathbb c}}{{\mathbb p}}^n}\ ] ] it is then natural to give the following definition ( see ( [ compcurv ] ) ) : [ def : amoebas - complex - curvature]the total curvature of the amoeba @xmath219 is defined by @xmath224 [ rem : multiplicative - translation - invariance ] we consider subvarieties of the torus @xmath225 therefore for any @xmath226 , @xmath88 and @xmath227 define the same variety .
moreover , the logarithmic gauss map @xmath228{rccc } \gamma~ : & v(f_t ) & \rightarrow & { { \mathbb c}}{{\mathbb p}}^n \\ & x & \mapsto & [ x_i\frac{\partial f_t}{\partial x_i } ] \end{array}}\ ] ] is also invariant when multiplying @xmath88 by @xmath229 ( since the tangent space to @xmath195 at a given point is unaffected by this action ) .
these definitions lead us to consider the systems @xmath230 = [ \beta_1:\dotso : \beta_{n+1 } ] \end{cases}\ ] ] where @xmath231=[\beta_1:\beta_2:\dotso : \beta_{n+1}]$ ] is an element of @xmath232 . since we only consider solutions in the torus , the system @xmath233 is equivalent to @xmath234 where we introduce a new variable @xmath235 .
we say that a polynomial @xmath88 is b - generic if , for a generic @xmath236 , the system @xmath237 satisfies the genericity conditions of @xcite _ i.e. _ , the restriction of the system to any proper face of the convex hull of its set of exponent has no solution in the torus .
[ rem : multiplicative - translation - invariance-2 ] let @xmath238 be the system @xmath239 = [ \beta_1:\beta_2:\dotso : \beta_{n+1 } ] \end{cases}.\ ] ] if the newton polytopes of @xmath88 and @xmath227 are included in @xmath240 we find algebraically the statement of remark [ rem : multiplicative - translation - invariance ] .
indeed , @xmath238 is then clearly equivalent to : @xmath241 = [ \frac{\beta_1}{x^\omega}:\frac{\beta_2}{x^\omega}:\dotso : \frac{\beta_{n+1}}{x^\omega } ] \end{cases}\ ] ] since we consider solutions in @xmath7 .
thus @xmath238 is equivalent to @xmath242 and for every @xmath231\in{{\mathbb c}}{{\mathbb p}}^{n+1}$ ]
, both corresponding logarithmic gauss maps have the same fiber . let @xmath243 denote the truncation of @xmath88 to a face @xmath244 of the newton polytope @xmath245 of @xmath88 .
let @xmath246 $ ] be a polynomial , @xmath247 its newton polytope and @xmath248 the toric variety associated to @xmath247 .
we say that @xmath101 is * completely non - degenerate * if for each ( not necessarily proper ) face @xmath244 of @xmath247 the restriction @xmath249 defines a non - singular variety in @xmath7 .
[ rem : tprimitive - nondegenerate ] let @xmath250 $ ] be such that @xmath251 is non - singular . then @xmath88 is completely non - degenerate for @xmath252 .
[ prop : nondegenerate ] a polynomial @xmath253 $ ] is b - generic if and only if it is completely non - degenerate . the support polytope @xmath254 of the system @xmath255 is the cone with apex @xmath256 over @xmath257 .
in fact , @xmath254 is the newton polytope of all polynomial of @xmath255 but @xmath88
. the restrictions of @xmath255 to the faces of @xmath254 are either of the form @xmath258 or of the form , @xmath259 for @xmath244 a face of @xmath245 .
the system @xmath260 is the resctriction of @xmath255 to the cone with apex @xmath256 over a face @xmath244 of @xmath254 . by definition ,
@xmath88 is b - generic if and only if for all proper faces @xmath244 of @xmath245 the systems @xmath260 have no solution for a generic @xmath236 and for all ( not necessarily proper ) faces @xmath244 of @xmath245 the systems @xmath261 have no solution .
note that @xmath261 has no solution if and only if @xmath262 is non - singular in @xmath7 .
assume first that @xmath88 is b - generic .
then , in particular for all ( not necessarily proper ) faces @xmath244 of @xmath245 the systems @xmath261 have no solution .
so , for any face @xmath244 , @xmath262 is non - singular which is equivalent to @xmath88 being completely non - degenerate .
assume now that @xmath88 is completely non - degenerate .
we saw above that this implies that for all ( not necessarily proper ) faces @xmath244 of @xmath245 the systems @xmath261 have no solution .
we only need to check that for a generic @xmath236 , the systems @xmath260 have no solution for all proper faces @xmath244 of @xmath245 .
let us now fix a proper face @xmath244 of @xmath245 and consider the system @xmath260 .
denote by @xmath263 the map
@xmath264 @xmath265 note that @xmath266 if and only if @xmath267 is singular .
let us write @xmath268 . since @xmath269 , @xmath244 is contained in a hyperplane and there exist @xmath270 and @xmath271 such that for any exponent @xmath164 of @xmath243 , @xmath272 for any @xmath273 , @xmath274 @xmath275 in particular , if @xmath276 , @xmath277 .
then , @xmath278 is contained in the linear hyperplane @xmath279 orthogonal to @xmath280 .
then , for any @xmath244 the system @xmath260 has no solution in the torus if the line @xmath281 is not contained in @xmath279 .
since there exists a finite number of hyperplanes containing all the proper faces of @xmath245 , @xmath282 is contained in a finite union of linear hyperplanes .
then , for a generic @xmath231 $ ] none of the systems in @xmath283 has a solution .
thus @xmath88 is b - generic . from now on we consider completely non - degenerate polynomials .
a polynomial @xmath88 is completely non - degenerate if and only if the degree of the map @xmath221 is @xmath284 it follows directly from proposition [ prop : nondegenerate ] applying bernstein theorem to the systems @xmath255 . as a direct consequence of proposition [ prop : nondegenerate ]
, we have the following corollary .
[ rem : complexamoeba ] @xmath88 is completely non - degenerate if and only if @xmath285 which does not depend on @xmath89 .
we have then the following inequality , similar to ( [ ineqcurv ] ) for the logarithmic curvatures of the real and complex parts of a real algebraic hypersurface : [ prop : inequality ] for any completely non - degenerate @xmath286 $ ] , @xmath287 we have seen that @xmath288 . for @xmath289 ,
the cardinality of the real fiber @xmath290 is smaller than the cardinality of the complex one .
so we similarly have that @xmath291 , with @xmath292 and @xmath293 from which the inequality follows .
there is equality in the above theorem if and only if the map @xmath221 is totally real , _
i.e. _ , @xmath294 for @xmath289 . one can prove ( see @xcite ) that for a non - singular real curve , the maximality for the above inequality characterises the harnack curves in the sense of @xcite .
let @xmath59 be a tropical polynomial , @xmath203 $ ] realising @xmath59 and such that @xmath88 is completely non - degenerate for @xmath197 .
since the total curvature of the amoeba @xmath219 does not depend on @xmath89 for @xmath197 , we define the total curvature of the tropical variety @xmath75 by passing to the limit in the trivial way : [ def : complex - tropical - curvature]let @xmath59 be a tropical polynomial .
we define the complex total curvature of @xmath75 as @xmath295 notice that @xmath296 is a positive function , then @xmath297 , therefore @xmath298 [ cor : hyperplane - complex - tropical - curvature ] for any primitive tropical hypersurface @xmath299 , @xmath300 [ cor : complex - tropical - curvature ] let @xmath59 be a tropical polynomial and let @xmath301 be the set of vertices of @xmath75
. then @xmath302 [ def : real - tropical - curvature ] let @xmath303 $ ] be a real polynomial , @xmath304 its tropicalization .
+ the real total curvature of @xmath305 is defined as @xmath306 [ rem : limit ] the proof of proposition [ totcurvr ] below implies that if @xmath75 is non - singular , @xmath307 has a limit when @xmath308 .
recall the following diagram : @xmath309^{\gamma_t^{{\mathbb r } } } \ar[d]^{\operatorname{{\textup{log}}}_t } \\
\mathcal{a}^{{\mathbb r}}(f_t ) \ar[r ] _ { g_t } & { { \mathbb r}}{{\mathbb p}}^n}\ ] ] where @xmath213 is the gauss map and @xmath215 the logarithmic gauss map , defined as : @xmath216.\ ] ] it follows immediately from theorem [ prop : inequality ] and definitions [ def : complex - tropical - curvature ] and [ def : real - tropical - curvature ] that real and complex tropical curvatures satisfy an inequality similar to inequality ( [ ineqcurv ] ) .
[ thm : tropical - inequality]let @xmath310 $ ] be a real polynomial and @xmath311 be its tropicalisation .
we have @xmath312 we now establish one the main results of this article ; namely that the inequality ( [ eq : tropical - inequality ] ) of theorem [ thm : tropical - inequality ] is an equality for real non - singular tropical hypersurfaces ( theorem [ thm : tropical - equality ] ) .
let us first look at the case of a primitive hypersurface .
[ prop : realcurv ] let @xmath59 be a primitive tropical polynomial , tropicalisation of @xmath303 $ ] .
then , @xmath313 we have @xmath314vol@xmath315 by definition .
but the map @xmath221 is generically of degree one by berstein s theorem , and for a generic @xmath316 ( namely in the complementary of the set of normal directions to non - compact cells of the tropical variety @xmath75 ) , we have @xmath317 and @xmath318 ; therefore @xmath319 and we have that vol(@xmath320vol@xmath321 vol@xmath322 , and ( [ eq : realcurv ] ) passing to the limit .
let now @xmath323 $ ] be a polynomial such that @xmath324 is non - singular , @xmath325 the vertices of @xmath75 , @xmath326 the subdivision of @xmath196 in simplices dual to @xmath75 .
then for each vertex @xmath327 , we set @xmath328 , with @xmath329 .
notice that @xmath330 .
the main step in the proof of theorem [ thm : tropical - equality ] is the following : + [ totcurvr ] let @xmath310 $ ] be such that @xmath331trop@xmath332 is non - singular . then @xmath333 before proving the proposition , we need a lemma of `` localisation '' at a vertex .
+ let @xmath334 be a vertex , @xmath335 .
let @xmath336 be a generic element ( see the proof of proposition [ prop : realcurv ] ) and @xmath337 be given .
then there exist @xmath338 and @xmath339 such that for any @xmath236 such that @xmath340 and @xmath89 such that @xmath341 , we have @xmath342 . here
@xmath343 is induced on @xmath29 by the distance on the unit sphere in @xmath8 . up to multiplying each @xmath344 by @xmath345 ( which has the effect of translating the vertex @xmath156 to @xmath74 ) and multiplying @xmath88 by the relevant power of @xmath89 , one may write : @xmath346 where @xmath347
, @xmath348 is a positive real number and @xmath349 $ ] is a polynomial whose coefficients have non - negative valuation .
we set @xmath350 so that @xmath351 . for @xmath316 ,
the points of @xmath352 are the images under @xmath198 of the solutions of the system : @xmath353 = [ \beta ] \end{cases}.\ ] ] for a generic @xmath354 , we know by the proof of proposition [ prop : realcurv ] that the system : @xmath355 = [ \beta_0 ] \end{cases}\ ] ] has a non - degenerated solution @xmath356 . then the system : @xmath357 = [ \beta ] \end{cases}\ ] ] has a solution @xmath358 such that @xmath359 is in the ball @xmath360 for @xmath361 sufficiently small .
the system @xmath233 is a one parameter deformation of @xmath362 therefore it has a solution @xmath363 such that @xmath364 for @xmath89 small enough .
then @xmath365 is in the ball @xmath366 which is included , for @xmath89 small enough , in @xmath367 .
let us now prove proposition [ totcurvr ] .
let @xmath368 be a compact set , @xmath339 and @xmath369 such that : \1 ) vol@xmath370 \2 ) for any direction @xmath371 and any vertex @xmath156 of @xmath75 , the system @xmath233 has a non - degenerated solution .
+ then , if @xmath372 is the number of vertices of @xmath75 , we have that for @xmath341 : @xmath373 passing to the limit when @xmath374 and @xmath375 gives the result .
we now deduce easily the main result of the paper .
[ thm : tropical - equality ] let @xmath310 $ ] be such that @xmath331trop@xmath332 is non - singular .
then @xmath376 by proposition [ totcurvr ] and corollary [ cor : complex - tropical - curvature ] , it is enough to prove the theorem in the primitive case .
we have then to prove that @xmath377 with @xmath378 , that is immediate .
our definition of the curvature in the polyhedral case is similar to banchoff s in @xcite ( see also @xcite and @xcite ) but , exactly as in the complex case , we only consider absolute value of the curvature here .
it amounts to the following .
the * solid angle * of a cone is the portion of the unit sphere centred at the vertex of the cone that it intersects , its * measure * is the volume of this spherical portion .
we might abuse terminology and write `` solid angle '' when we mean its measure .
let @xmath183 be a vertex of a polyhedral hypersurface @xmath379 ( here our real tropical hypersurface @xmath144 ) .
for sufficiently small neighbourhoods @xmath23 of @xmath183 , @xmath380 has two connected components .
label one by @xmath381 and the other by @xmath382 ( for @xmath144 these will be the signs of the corresponding vertices of the dual subdivision ) .
for each maximal dimensional cell of @xmath379 containing @xmath183 , choose a normal vector oriented from @xmath382 to @xmath381 .
the * curvature cone * @xmath383 at @xmath183 is the cone generated by these vectors .
[ def : local - polyhedral - curvature ] the * curvature * @xmath384 at @xmath183 is the measure of the solid angle of the curvature cone @xmath383 .
[ rem : curvature - cone ] for @xmath142 the normal vectors above can be chosen to correspond to the vectors @xmath385 supported by the edges of the simplex @xmath186 dual to @xmath183 and oriented from a vertex with @xmath382 sign to a vertex with @xmath381 sign .
thus the curvature cone @xmath383 is naturally identified to the cone generated by the @xmath385 s and depends only on the simplex @xmath186 dual to @xmath183 and the sign distribution at the vertices of @xmath186 .
[ rem : sign - inversion ] changing all the vectors to their opposites clearly leaves @xmath384 invariant .
[ def : polyhedral - curvature ] let @xmath59 be a generic tropical polynomial and let @xmath143 be a distribution of signs in @xmath58 .
the polyhedral total curvature of @xmath142 is @xmath386 _ i.e. _ , it is the sum of the curvatures at all vertices of the real part of @xmath142 .
it follows from the definition that we have the equality below .
[ prop : by - def]let @xmath59 be a generic tropical polynomial let @xmath143 a distribution of signs in @xmath58 . if @xmath387 are the vertices of @xmath75 , then @xmath388 where @xmath389 we denote by @xmath390 the curvature cone at the vertex @xmath185 of the real tropical hypersurface @xmath144 .
it is the cone generated by vector edges of @xmath391 oriented form vertices with minus sign to vertices with plus sign in the sign distribution @xmath181 .
[ rem : alternative - sum ] one can also define the curvature @xmath392 at a vertex @xmath183 of @xmath75 to be the sum over all symmetric copies @xmath185 of @xmath183 appearing in @xmath144 of the solid angles of the corresponding curvature cones @xmath390 _ i.e. _ , @xmath393
. then @xmath394 .
[ def : elementary ] let @xmath395 be a simplex with integer vertices and @xmath396 be the collection of the vectors defined by edges issuing from one of its vertices .
the simplex @xmath397 is * elementary * if @xmath398 is a basis of @xmath141 where @xmath399 is the reduction modulo @xmath133 of @xmath400 .
[ prop : hyperplane - polyhedral - curvature ] let @xmath59 be a tropical polynomial such that @xmath60 is an elementary simplex and that @xmath401 . then , for any distribution of signs @xmath143 , @xmath402 in particular , proposition [ prop : hyperplane - polyhedral - curvature ] holds for primitive real tropical hypersurfaces ( those whose newton polytope is primitive ) since a primitive simplex is elementary .
the two corollaries below follow from proposition [ prop : hyperplane - polyhedral - curvature ] and lemma [ prop : by - def ] .
[ cor : generic - tropical - curvature]let @xmath59 be a generic tropical polynomial and let @xmath403 be the vertices of @xmath75 . if @xmath404 is an elementary simplex for all @xmath405 then , for any distributions of signs @xmath143 , @xmath406 [ cor : primitive - tropical - curvature]let @xmath59 be a non - singular tropical polynomial .
then , for any distribution of signs @xmath143 , @xmath407 from proposition [ prop : hyperplane - polyhedral - curvature ] and proposition [ totcurvr ] we deduce the equality between the polyhedral curvature and real total curvature of a non - singular real tropical hypersurface : [ thm : polhedral = limit]let @xmath59 be a non - singular tropical polynomial .
then , for any distribution of signs @xmath143 , @xmath408 both can be expressed as @xmath409 times the volume of @xmath410 ( see corollary [ cor : primitive - tropical - curvature ] in the polyhedral case ) .
we will use the following proposition which essentially follows from itenberg s prop 3.1 in @xcite .
recall that @xmath411 so for our study we might as well consider sign distributions up to total inversion of signs .
we will denote @xmath412 the set of sign distributions on @xmath413 up to simultaneous change of all signs .
[ prop : transitivity ] let @xmath397 be an elementary simplex .
the group @xmath141 acts transitively on @xmath414 via the maps @xmath139 ( see subsection [ subsec : real - tropical - hypersurfaces ] ) . in notation of definition [ def :
elementary ] the @xmath399 s form a basis of @xmath141 .
let @xmath415 be the dual basis .
then for each vertex @xmath416 of @xmath397 , * either @xmath417 and @xmath418 for @xmath419 * or @xmath420 and @xmath421 for @xmath419
. we will prove that the curvature cones defined by a real tropical hypersurface dual to an elementary simplex @xmath397 give rise to a partition of a half - space , which yields the result .
since we are considering the sign distributions up to total inversion of signs , we can assume that for one vertex @xmath422 of @xmath397 , @xmath423 for all @xmath424 .
by proposition [ prop : transitivity ] and remark [ rem : curvature - cone ] , the cones we need to consider are exactly those corresponding to all distributions of signs @xmath425 on @xmath426 such that @xmath422 carries a minus sign .
let us denote by @xmath427 the curvature cone corresponding to such a distribution @xmath425 .
let us prove that the curvature cones @xmath427 naturally define a fan which covers a half - space . to each vertex @xmath156 of @xmath397 one
associates its opposite facet @xmath428 , the vectorial hyperplane @xmath429 parallel to @xmath428 and a vector @xmath430 normal to @xmath428 pointing from @xmath428 to the interior of @xmath397 . for a vertex @xmath156 of @xmath397 let @xmath431 and @xmath432 be respectively the half - space defined by @xmath433 and @xmath434 .
the key point is the following fact .
[ lem : conesandhyperplanes0 ] for a sign distribution @xmath425 ( with @xmath435 ) , the curvature cone @xmath427 is the intersection of the half - spaces @xmath436 for all @xmath437 .
moreover it is enough to intersect only those @xmath436 such that @xmath438 .
we need to prove that the cone @xmath427 is defined by @xmath439 .
let us denote by @xmath440 the set of edges of @xmath397 whose vertices have different signs and by @xmath441 the set of vectors such that @xmath442 is supported by @xmath443 and oriented from `` - '' to `` + '' .
for any @xmath444 , @xmath445 .
if @xmath443 is not in the face @xmath428 of @xmath397 opposite to @xmath156 , it points towards the ( affine ) half - space containing @xmath397 if @xmath446 is `` + '' and towards the other half - space determined by @xmath428 if @xmath446 is `` - '' .
thus @xmath447 .
let us prove that each facet of @xmath427 is parallel to a facet of @xmath397 and thus included in some @xmath429 which leads to the equality in the above inclusion .
each facet of @xmath427 is a cone generated by a subset @xmath448 of @xmath41 whose linear span @xmath449 is of dimension @xmath71 and such that all vectors of @xmath450 are on the same side of @xmath449 .
let @xmath448 be a subset of @xmath41 , @xmath451 be the corresponding subset of @xmath452 and assume that @xmath453 and @xmath449 is included in no @xmath429 .
let us prove that the cone generated by @xmath448 is not a face of @xmath427 .
each vertex of @xmath397 belongs to an edge of @xmath451 ( otherwise @xmath449 would be parallel to a facet of @xmath397 ) .
let @xmath454 ( resp .
@xmath455 ) be the set of vertices of @xmath397 with `` + '' respectively `` - '' signs .
if either @xmath456 or @xmath457 is @xmath458 then @xmath427 is just the cone of apex a vertex @xmath156 over its opposite face @xmath428 and its facets are cones on facets of @xmath428 .
let us then assume that @xmath459 and @xmath460 .
if every pair of vertices in @xmath426 were connected by a chain of edges in @xmath451 then every vertex would be connected by a chain of edge in @xmath451 thus @xmath461 which would contradict the hypothesis . thus the edges in @xmath451 split in several connected components @xmath462 .
each one contains at least one element of @xmath463 and one element of @xmath454 .
the affine span @xmath464 of @xmath462 is just the affine span of the vertices it contains , thus it is the affine span of the corresponding face of @xmath397 .
each @xmath464 is parallel to @xmath449 which is of codimension one . since @xmath451 covers all vertices of @xmath397 ,
all @xmath464 are not in the same affine hyperplane parallel to @xmath449 .
then an affine hyperplane parallel to @xmath449 separates the affine spans @xmath465 and @xmath466 of two connected components @xmath467 and @xmath468 .
let us pick two vertices with minus sign @xmath469 and @xmath470 respectively in @xmath467 and @xmath468 .
let @xmath471 ( resp .
@xmath472 ) be vectors in @xmath450 having origin @xmath473 ( resp @xmath474 ) and extremity a vertex in @xmath468 ( resp . in @xmath467 ) .
the connected components @xmath467 and @xmath468 being separated by an hyperplane parallel to @xmath449 , the vectors @xmath471 and @xmath472 are not on the same side of @xmath449 and @xmath448 does not generate a facet of @xmath427 .
thus each facet of @xmath427 is contained in one of the hyperplanes @xmath429 and , since we already have that @xmath475 , then @xmath476 .
it is enough to intersect only those @xmath436 such that @xmath438 . indeed when the signs of all the vertices of a facet are the same , the intersection of @xmath427 with @xmath477 is the origin . in the proof of lemma [ lem : conesandhyperplanes0 ]
one can easily see that , if @xmath453 and @xmath449 is included in no @xmath429 , @xmath451 has exactly two connected components .
indeed , the affine spans @xmath464 are just the spans of pairwise disjoint faces of @xmath397 .
let @xmath478 be the number of connected components of @xmath451 .
each @xmath462 contains @xmath479 vertices and @xmath480 .
but the number of vertices in @xmath397 is @xmath481 , thus @xmath482 and @xmath483 .
[ lem : conesandhyperplanes ] consider the union @xmath484 of all linear hyperplanes @xmath429 and the collection of curvature cones @xmath485 .
the cones in @xmath486 are precisely the maximal dimensional closed cones in @xmath487 defined by @xmath488 . by lemma [ lem : conesandhyperplanes0 ] for a sign distribution @xmath425 ( with @xmath435 )
, the curvature cone @xmath427 is the intersection of the half - spaces @xmath436 for all @xmath437 .
any cone @xmath489 which is the closure of a connected component of the complement of @xmath490 in @xmath491 is of the form @xmath427 .
indeed it is defined by a choice of a side for each hyperplane @xmath429 _ i.e. _ , by the choice of the sign of the scalar product of vectors in the interior of @xmath489 with @xmath430 for each @xmath437 .
the sign distribution @xmath425 is then given by @xmath492 for any @xmath18 in the interior of @xmath489 .
indeed , setting a sign on a vertex of @xmath397 amounts to choosing on which side of @xmath429 are all vectors not in @xmath429 generating @xmath427 .
the sign `` - '' corresponds to pointing from @xmath428 to the exterior of @xmath397 and `` + '' from @xmath428 to the interior of @xmath397 .
( of course cones defined by a choice of side for each @xmath481 hyperplanes @xmath429 can sometimes be reduced to the origin ; this corresponds exactly to a sign distribution @xmath425 on @xmath426 which does not surjects on @xmath148 , _
i.e. _ , to an empty orthant on the real tropical variety side . )
thus the closed cones @xmath427 clearly cover @xmath487 .
( a vector @xmath18 in @xmath487 not belonging to one of the @xmath429 s is in @xmath427 if and only if for all @xmath437 , @xmath492 . )
moreover by lemma [ lem : conesandhyperplanes ] they realise a subdivision of @xmath487 .
so , since by proposition [ prop : transitivity ] we get all possible sign distributions such that @xmath435 , we proved that @xmath493 .
[ ex : total - curvature - real - line ] on figure [ fig : example ] and figure [ fig : line - total - curvature ] we illustrate proposition [ prop : hyperplane - polyhedral - curvature ] on the trivial case of a real tropical line .
we depict the angles formed by the vectors generating the curvature cones which in this case are the angles of the newton triangle .
[ htbp ] [ htbp ]
in a forthcoming paper , the second author studies , using tropical geometry , the limit when @xmath89 goes to zero of the real total curvature of a family @xmath208 of real algebraic hypersurfaces . this study allows to give a lower bound to this limit , for any polynomial @xmath494 $ ] realising a generic tropical hypersurface .
this bound depends only on the tropicalisation of @xmath101 and will be called the _ tropical bound_. + then , if @xmath495 is the classical curvature function , @xmath496 will be bounded from above by the risler s complex bound ( see inequation ( [ ineqcurv ] ) ) and from below by the tropical bound . a polynomial @xmath494 $ ] is called _ maximal _ with respect to the real total curvature if the risler s upper bound is sharp . in other words ,
a polynomial @xmath494 $ ] is maximal with respect to the real total curvature if @xmath497 the idea behind the construction of the tropical bound is to look for the valuation of points in @xmath498 that _ concentrate _ the real total curvature : a point @xmath499 concentrates the real total curvature if for any family of neighbourhoods @xmath500 of the family of points @xmath501 , @xmath502 via tropical methods , the second author gives a lower bound for the number of points in @xmath503 that concentrates the real total curvature and the tropical bound arises as a direct consequence . using this study , an infinite family of polynomials in @xmath504 $ ] whose tropical bound is equal to their risler s complex bound
is constructed .
this is a tropical proof of the following theorem : + one deduces from this theorem a tropical proof of orevkov s observation ( see @xcite ) about the sharpness ( up to any @xmath508 ) of risler s complex bound for affine real algebraic hypersurfaces . in the viro s patchworking language ,
this result has been also proved in @xcite .
let @xmath75 be a non - singular tropical hypersurface with newton polytope @xmath60 , and @xmath510 be a generic complex hypersurface with newton polyhedra @xmath60 .
then : @xmath511 where @xmath512 stands for the _ euler characteristic _ of @xmath0 .
( of ( [ g - b ] ) ) . + if @xmath510 is a generic hypersurface with newton polytope @xmath60,one has @xmath520 ( hovansky s formula , cf .
the tropical variety @xmath75 is by hypothesis dual to a primitive triangulation @xmath509 with @xmath521 .
+ let @xmath372 be the number of @xmath522 s ( _ i.e. _ , the number of vertices of @xmath75 ) ; then one has : @xmath523 by definition [ def : complex - tropical - curvature ] .
+ this proves ( [ g - b ] ) , because @xmath524 , and then @xmath525 .
oleg viro , _ gluing of algebraic hypersurfaces , smoothing of singularities and construction of curves.(in russian ) _ , proc .
leningrad int .
topological conf . ,
leningrad , 1982 , nauka , leningrad ( 1983 ) , 149197 . | this paper studies the curvatures of amoebas and real amoebas ( i.e. essentially logarithmic curvatures of the complex and real parts of a real algebraic hypersurface ) and of tropical and real tropical hypersurfaces .
if @xmath0 is a tropical hypersurface defined over the field of real puiseux series , it has a real part @xmath1 which is a polyhedral complex .
we define the total curvature of @xmath0 ( resp .
@xmath1 ) by using the total curvature of amoebas and passing to the limit .
we also define the `` polyhedral total curvature '' of the real part @xmath1 of a generic tropical hypersurface .
the main results we prove about these notions are the following : 1 . the fact that the total curvature and the polyhedral total curvature coincide for real non - singular tropical hypersurfaces .
2 . a universal inequality between the total curvatures of @xmath0 and @xmath1 and another between the logarithmic curvatures of the real and complex parts of a real algebraic hypersurface .
3 . the fact that this inequality is sharp in the non - singular case . |
turbulence in superfluids , often named quantum turbulence , is a very active area of current research @xcite .
quantum mechanical constraints require that only vortices of quantized circulation can exist in an otherwise irrotational flow @xcite . in the limit of very low temperature when almost all thermal excitations are frozen out , turbulence in a pure superfluid
is considered to be the most simple paradigm of a difficult subject .
this applies to superfluid @xmath0he at millikelvin temperatures , to superfluid @xmath7he in the microkelvin regime , and to bose - einstein condensates ( bec ) at nanokelvin temperatures .
the method to generate vorticity that leads to turbulence in the form of a vortex tangle is to move an object through the superfluid , or to rotate it . in superfluid helium
using oscillating objects like spheres , wires , grids or tuning forks is the easiest way to study the transition from potential flow to turbulence at some critical velocity amplitude . for recent reviews
see , e.g. , @xcite .
new techniques are now available to obtain more information of the turbulent state , like improved electronics @xcite and visualization by means of particle image velocimetry ( piv ) @xcite . in case of a bec
the moving object is typically a laser beam that presents an obstacle to the condensate , and thus may lead to vortex shedding above a critical velocity at which the beam is moved repeatedly or back and forth through the condensate .
this phenomenon , that was theoretically predicted and observed by several groups , is related to our experiments in the dense bose liquid of superfluid @xmath0he , as will be shown below . as mentioned above , quantum turbulence
can also be investigated by rotating the superfluids .
although this technique will not be discussed in the following , interesting review articles can be found in @xcite for bec , and @xcite for the helium liquids . in our present article
we review and summarize our earlier experimental and theoretical results from experiments with an oscillating microsphere immersed in superfluid @xmath0he at temperatures down to 25 mk . the experimental technique ( for details , see appendix a ) makes use of superconducting levitation of a ferromagnetic sphere ( radius @xmath8 = 124 @xmath9 m , mass @xmath10 = 27@xmath9 g ) between superconducting niobium electrodes of a horizontal parallel plate capacitor ( spacing @xmath11 = 1 mm ) . before cooling the capacitor into the superconducting state ( for nb at 9.2
k ) , we apply several hundred volts to the bottom electrode charging the sphere to about @xmath12 1 pc .
vertical oscillations around the equilibrium position of the levitating sphere can be excited by applying an ac voltage @xmath13 at resonance ( @xmath14 120 hz ) in the range from 0.1 mv to several volts , exerting on the sphere a driving force @xmath15 .
the oscillations induce an ac current @xmath16 that is detected by an electrometer .
because no mechanical support is needed , the sphere moves at a well - defined velocity in contrast to wires , grids or tuning forks , where the velocity changes from zero at the mechanical support to some maximum value at the center of the wire or at the tip of the fork .
moreover , the simple spherical geometry makes the data transparent and more directly accessible in a quantitative way , in particular the laminar and the turbulent drag forces on the sphere can be identified quantitatively , in contrast to the complicated geometry of the other oscillating objects . + potential flow and the transition to turbulent flow at a sharp critical velocity @xmath1 can be easily identified due to the very different drag forces .
most interestingly , in a small interval of velocities @xmath17 above @xmath1 where @xmath5 , we observe an instability of the flow pattern switching intermittently between phases of potential flow and turbulent flow .
an investigation of this phenomenon leads to detailed information on the stabibility of the turbulent state and allows an interesting comparison with vortex shedding in a bec . +
this article is organized as follows : in section 2 the results for stable potential flow and stable turbulent flow will be presented ; in section 3 we review our results for the critical velocity .
the central part of the article is the intermittent switching phenomenon discussed in section 4 : in subsection 4.1 the stability of the phases of potential flow is discussed ; in subsection 4.2 the lifetimes of the turbulent phases are analyzed ; in subsection 4.3 vortex shedding from the sphere is compared with experiments of other authors obtained by moving a laser beam through a bec ; in subsection 4.4 we find from a recent definition of a superfluid reynolds number that the transition to turbulence in our experiments is a new example of `` supertransient '' chaos .
after the final summary , we present experimental details in appendix a. some peculiar results obtained with dilute @xmath7he -@xmath0he mixtures are described in appendix b.
at a fixed temperature @xmath18 we measure the velocity amplitude as a function of the amplitude of the driving force @xmath19 at resonance frequency .
( details of the determination of the resonance frequency are described in appendix a. ) as an example , the data at @xmath18 = 300 mk are shown in fig.[fig:1 ] .
we identify three different regimes : a linear rise of @xmath19 at small driving forces up to a critical velocity ; a nonlinear increase at larger drives ; and the shaded region in between where the flow pattern is unstable , switching between both patterns @xcite . in the following
we will discuss these regimes in detail .
the linear regime at small velocities indicates potential flow with a small linear drag force @xmath20 that , because of the @xmath21 temperature dependence of @xmath22 , we attribute to ballistic phonon scattering , whereby @xmath23 where the phonon density @xmath24 rapidly varies as @xmath21 , and @xmath25 is the velocity of sound .
there might be a numerical factor of order one in eq .
, depending on the details of the phonon scattering ( specular or diffuse ) .
because the wavelength of a thermal phonon in helium at 0.1 k is about 0.1 @xmath9 m which is much smaller than the size of the sphere and likely also of the surface roughness , we expect both a geometric cross section and a diffuse scattering , but we are not aware of a calculation of the prefactor .
we are assuming that it will be of order 1 . using the value of the radius @xmath26 m as determined
optically , we obtain a perfect quantitative agreement with the data if we set this factor equal to one , see fig.[fig:2 ] . at temperatures below ca .
100 mk , the phonon drag falls below the background damping measured in vacuum . at temperatures above 1 k ,
the drag force is described by stokes solution extended to the two - fluid model @xcite . of the laminar phase .
the solid line is calculated for ballistic quasiparticle scattering . at the lower temperatures it follows a @xmath21 law due to phonon scattering ,
see eq .. also note that the drag is larger near 0.7 k than at 0.9 k where the transition to the hydrodynamic regime begins .
a background damping of @xmath27kg / s measured in vacuum has been subtracted.,scaledwidth=80.0% ] we use this linear regime to determine the charge @xmath28 of the sphere .
when a stable amplitude is reached at resonance , the drag force @xmath29 and driving force @xmath30 cancel . from the slope @xmath31 of the induced current @xmath32 as a function of @xmath13 , and from the time constant of the freely decaying amplitude @xmath33 measured separately at the same temperature ,
we obtain the charge ( the mass @xmath10 and the spacing @xmath11 are known , see above ) .
we apply this procedure at @xmath34 where @xmath35 has a convenient value of @xmath36 . during the course of the experiment
, we find the charge to remain constant , and occasionally a small loss of few percent can be observed after several weeks .
the nonlinear dependence of @xmath19 in fig.[fig:1 ] and fig.[fig:3 ] can be properly described by a quadratic drag force .
in contrast to a classical liquid , the apex of the parabolic shape of @xmath19 is shifted to the left of the origin by 0.4 nn , and the resulting finite intercept at @xmath30 = 0 indicates a velocity range of frictionless flow which is the paradigm of superfluidity .
moreover , we observe a sharp onset of the turbulent regime , whereas in a classical liquid there are about three orders of magnitude in flow velocity between stokes regime of laminar flow and fully developed turbulence , where the classical drag on a sphere is given by @xmath37 with @xmath38 ( @xmath39 is the density of the liquid and the drag coefficient @xmath40 of a sphere is approximately 0.4 ) . , see eq ..
,scaledwidth=80.0% ] for a quantitative analysis of the data , we have to take into account the following considerations .
for oscillatory flow subject to a quadratic drag force , the concept of energy balance requires that per half - period the power injected by the drive and dissipation due to the drag must cancel for a stable oscillation amplitude . in the case of a linear drag
, this leads to the simple result that the equilibrium amplitude is reached when driving force and drag force cancel , see above .
but in the case of a quadratic drag of the form @xmath41 that is obviously appropriate here , a short calculation gives for the driving force @xcite @xmath42 there are 2 corrections , firstly a factor 8/3@xmath43 0.85 , and secondly , an intercept for @xmath44 at @xmath45 = 3/2 @xmath46 .
thus , the intercept velocity is larger than @xmath47 by a factor @xmath48 . before fitting eq . to the data , the linear drag ( phonons and background ) ,
see eq . , must be subtracted .
if we parametrize @xmath49 like in the classical case , i.e. , @xmath50 , the only fitting parameter is the drag coefficient @xmath40 in @xmath51 .
we find @xmath52 0.40 to within 10 % .
hence , surprisingly , the expressions of @xmath51 for oscillatory superflow and for classical uniform flow are identical . we
define _ the critical velocity @xmath1 as the largest velocity at @xmath30 = 0 , i.e. , @xmath53 , which can be obtained either from the intercept in fig.[fig:4 ] , or directly from the kink in the @xmath19 curve , see fig.[fig:5 ] .
+ = 0.36 .
the intercept at @xmath44 occurs at @xmath54 5.0 ( cm / s)@xmath55 .
the background damping of the empty cell has been subtracted , phonon drag is negligible at 33 mk .
( color figure online),scaledwidth=80.0% ]
when the velocity amplitude is increased beyond a certain value , the linear regime abruptly changes to a nonlinear regime at lower amplitude .
this signals the transition to a large turbulent drag force that is followed by a hysteresis with the down sweep , see fig.[fig:5 ] .
= 46 mm / s can only be determined with the down sweep .
the oscillation frequency was 236hz .
( color figure online),scaledwidth=80.0% ] we interpret the hysteretic behavior as follows . at the beginning of the first up - sweeps there are few remanent vortices left from the first cool - down .
potential flow will break down when the oscillating sphere collides with such a vortex and then sheds more vortices .
in fact , detailed experiments with vibrating wires , e.g. , by the osaka group , demonstrate that remanent vorticity can be reduced , or even avoided , if the measuring cell is filled very slowly at low temperatures @xcite . in that case , no transition to turbulence was observable up to very high velocity amplitudes of 1.5 m / s .
obviously , the wire does not shed vortices by itself .
we expect the same scenario to occur in our case . from fig.[fig:5 ]
the velocity at initial breakdown is @xmath56 = 82 mm / s and the critical velocity is @xmath1 = 46 mm / s .
we may infer the intervortex spacing @xmath57 where @xmath58 is the remanent vortex density ( defined as the length of all vortices per unit volume , having a dimension 1/m@xmath55 ) by postulating that the critical oscillation amplitude is determined by the spacing of the remanent vortices , i.e. , @xmath59 we find a ratio of 0.56 for the spacings and hence a ratio @xmath60 = 0.31 . if for some reason @xmath58 were too large , a too small @xmath1 would be measured .
this and the hysteresis must be taken into account when determining @xmath1 .
we now turn to dependence of @xmath1 on the oscillation frequency @xmath61 .
empirically we find a scaling of @xmath1 as @xmath2 , see fig.[fig:6 ] , in agreement with results obtained with a vibrating wire at osaka @xcite and with a tuning fork at lancaster @xcite .
on dimensional grounds this leads to @xmath62 . because there is still no rigorous theory available the numerical prefactor
must be determined from the experiment . in our case
, we find from fig.[fig:6 ] @xmath63 + the other authors , working with vibrating wires or tuning forks , find slightly different numerical prefactors , both are a little smaller but also of order 1 .
the shape of the oscillating object might have an effect .
for example , sharp corners strongly increase the velocity locally , making the prefactor smaller .
m ; blue dots : 100@xmath64 m ; black line : calculated from eq .. there is a large scatter of the data near 200 hz that might be caused by some unusual remanent vorticity .
( color figure online),scaledwidth=75.0% ] although we have no theory for oscillating superfluid flow , there are several arguments that make the @xmath2 dependence of @xmath1 plausible @xcite .
firstly , starting from the superfluid reynolds number @xmath65 1 for the onset of turbulence , and choosing the oscillation amplitude as the characteristic length scale @xmath66 we obtain @xmath67 .
this is a qualitative , but very general argument .
note that the size @xmath8 of the oscillating object is irrelevant here in contrast to the feynman critical velocity for uniform motion @xmath68 , where the size is the relevant length scale @xmath69 .
secondly , making use of results obtained by kopnin @xcite based on the well known vinen equation for uniform motion , a relaxation time @xmath70 of the vortex tangle after a change of the superfluid flow velocity @xmath71 has been calculated : @xmath72 here the numerical parameter @xmath73 = a(1 - @xmath74 ) - b@xmath75 is determined by the coefficients of mutual friction @xmath74 and @xmath75 and the constants a , b @xmath76 1 . applying this to oscillatory flow ( which remains to be proven valid ) ,
we take the time the velocity changes from zero to the maximum , i.e. , one quarter of a period , in which the vortex tangle can follow and thus arrive at the condition @xmath77 and therefore we have @xmath78 + to compare eq . with the experimental data we evaluate @xmath73 for @xmath0he by referring to tabulated values of @xmath79 and @xmath80 @xcite . setting the constants a , b = 1 , we find @xmath73 = 0.95 ( at 1.3k ) , 0.89 ( at 1.6k ) , and 0.79 ( at 1.9k ) .
below 1k mutual friction is very small , hence we set @xmath81 .
this implies a slow increase of @xmath1 by about 10% in qualitative agreement with experimental results as displayed in fig.[fig:7 ] .
closer to the @xmath82-point , the critical velocity could not be identified accurately any more because there was no sharp onset of turbulence detectable in the @xmath19 curves .
, e.g. , in a double - logarithmic plot , indicates a velocity above which the turbulent drag dominates the linear drag , but not the _ onset _ of turbulent drag , hence not @xmath1 that actually could be much smaller @xcite . ]
is indicated by the dashed line .
( color figure online),scaledwidth=80.0% ] at present , the most accurate theoretical support of the @xmath83 dependence of @xmath1 comes from a dynamical scaling of the critical velocity @xcite . instead of presenting here all details of this approach
, we only outline the following concept .
the equation of motion of a vortex filament was shown by schwarz @xcite to be invariant when scaling the length @xmath84 , the velocity @xmath85 , and the time @xmath86 .
this result was extended to oscillatory motion by kotsubo and swift @xcite by recognizing that the frequency @xmath87 is a scaling variable .
the final result , that is exact to within a small logarithmic correction of about 4% in our frequency range @xmath88 7 , is @xmath89 + it is quite clear that the @xmath83 dependence of @xmath1 can not remain valid when the frequency approaches zero for uniform motion , because a finite critical velocity is known to exist in that case too .
we are suggesting that the relevant length scale in this case is the size @xmath8 of the sphere .
thus , as shown above , we recover the feynman critical velocity @xmath90 = 0.5 mm / s .
the transition to steady flow will occur when @xmath91 , that is at @xmath92 = 6.5 s@xmath93 , which is much smaller than our oscillation frequencies ( @xmath94 700 s@xmath93 ) , see fig.[fig:8 ] . for much smaller objects this will matter . from oscillatory flow to steady flow @xmath95 at @xmath92 .
( color figure online),scaledwidth=70.0% ] finally , we should mention that with our experimental method , we have no quantitative control of the resonance frequency of the sphere .
it depends on the state of levitation , in particular on how much flux is trapped in the superconducting electrodes .
a more rapid cool down in our @xmath7he cryostat in general produced higher frequencies than in our dilution cryostat .
below @xmath96 the hysteretic regime discussed in the previous section is replaced by an instability where neither potential flow nor turbulence are stable , but instead the flow switches intermittently between both patterns ( shaded area in fig.[fig:1 ] ) .
the unstable regime extends from the critical velocity @xmath1 up to @xmath97 where @xmath98 3% .
it is plausible that stable turbulence can not exist exactly at @xmath1 , because in the limit @xmath99 no power is available at vanishing drive to create vorticity
. corresponds to turbulent flow while the increase occurs during a laminar phase . with
increasing drive the lifetimes of the laminar phases become shorter , whereas the lifetimes of turbulent phases grow rapidly .
the time interval shown here extends over 1000 s @xmath100 17 min.,scaledwidth=80.0% ] we have recorded time series ( lasting up to 36 hours ) of this switching phenomenon at various constant driving forces and temperatures , see fig.[fig:9 ] , for three different driving forces at @xmath101
. the amplitude switches between a low level @xmath102 corresponding to turbulent drag and an exponential recovery of the laminar level corresponding to phonon scattering . because the velocity amplitudes are above @xmath1 , these phases of potential flow break down at some velocity @xmath103 .
if the laminar phase lasts long enough , the stationary value for laminar flow @xmath104 is reached ( see upper time series ) .
when the laminar flow breaks down , a rapid drop to the turbulent level occurs due to the onset of the large and nonlinear drag .
it is obvious from fig.[fig:9 ] , that with increasing driving force the lifetimes of the laminar phases become shorter and those of the turbulent phases grow very rapidly . in the following
we analyze these time series statistically .
the statistical properties of the laminar phases @xcite can be analyzed either from the distribution of the signal height @xmath105 ( velocity at breakdown minus turbulent velocity @xmath102 ) , or by considering the distribution of the lifetimes .
both variables are related by the exponential recovery of the laminar amplitude ( starting at @xmath106 ) @xmath107 where @xmath108 is the difference between the stationary laminar velocity amplitude @xmath109 and @xmath102 , and @xmath110 is the time constant . of the velocity increase @xmath111 above the turbulent level .
b ) the fitting parameter @xmath112 is shown to be independent of temperature and driving force @xmath113 , i.e. , linear drag force subtracted.,scaledwidth=80.0% ] in fig.[fig:10]a the distribution of the signal height @xmath111 is shown to be of the gaussian form @xmath114 where @xmath112 is a fitting parameter .
the distribution function describes the probability @xmath115 that a given @xmath111 is exceeded ( reliability function " ) . in our case , it is the so - called rayleigh distribution , and the fitting parameter @xmath116 ( resonance frequency 120 hz ) is found to be completely independent of both temperature and driving force , see fig.[fig:10]b .
it is a property of the rayleigh distribution that @xmath112 is the rms value of the distribution . from the rayleigh distribution we obtain the failure rate @xmath117 which is the conditional probability that the potential flow decays in a small interval just above @xmath111 ,
provided it has not decayed until @xmath111 : @xmath118 we note that the failure rate is proportional to @xmath111 , and hence , to the increase of the oscillation amplitude @xmath119 .
this fits into our picture that a collision of the sphere with remanent vortices will cause a breakdown of the potential flow , see fig.[fig:5 ] , where the breakdown at @xmath103 is related to the intervortex spacing @xmath120 of the remanent vortex density .
another property of the failure rate can be found when changing the variable from @xmath111 to the lifetime @xmath121 of the potential flow : @xmath122 where @xmath123 is given by eq .. initially we have a linear increase of @xmath123 and therefore @xmath124 . at a time @xmath121 =
@xmath125 the failure rate has a maximum . and , if the maximum level @xmath108 is approached for large @xmath121 , we find @xmath126 .
thus , the flow becomes stable , although the velocity @xmath104 is clearly above @xmath1 .
these metastable states of potential flow above @xmath1 will also break down after a mean lifetime of 25 min , because natural radioactivity produces vorticity in the fluid .
this is proven by placing a weak radioactive source @xmath127 ( activity of @xmath128 = 2 @xmath9ci ) near the cryostat , resulting in a dramatic effect : the mean lifetime of the laminar phase is reduced to 3.0 minutes , see fig.[fig:11 ] . .
for times @xmath129 ( at 300 mk ) a mean lifetime of 25 minutes is measured .
the additional radioactive source reduces the mean lifetime to 3.0 minutes .
( color figure online),scaledwidth=75.0% ] we have measured the dose rate of the source at the position of the measuring cell inside the cryostat ( taking into account a measured 20% loss in the dewar walls ) to be 440ngy / h ( @xmath1305% ) . comparing this value with the measured dose rate due to natural background radiation in our laboratory of 57.6 ( @xmath130 1.5)ngy / h , we obtain an increase of the dose rate due to the source by a factor of @xmath131 . within the error bars this compares well with the measured reduction of the mean lifetime of the laminar phases by a factor of @xmath132 .
the effect of the radioactivity is attributed to the creation of he ions in the superfluid , which produce local vorticity either during the creation and recombination processes , or when the ions are accelerated by the electric field which exists in our measuring cell .
this vorticity then triggers the breakdown of the potential flow when the velocity amplitude of the sphere reaches the corresponding critical velocity @xmath56 .
we therefore conclude that the lifetime of the metastable laminar phases which we observe above @xmath1 is limited only by natural background radioactivity . + below 100
mk the phonon drag is negligibly small , only the residual damping of the empty cell @xmath133 = 4.4 @xmath134 kg / s determines the oscillation amplitude at a given drive .
equivalent is a time constant @xmath35 = 2m/@xmath133 = 1200 s. therefore , the exponential function in eq .
can be expanded : @xmath135 during potential flow is described by eq .
, see text .
the lifetimes of the turbulent phases can not be resolved ( not even on an expanded time scale ) .
we take the minima near 19 mm / s as @xmath136 . ,
scaledwidth=80.0% ] in fig.[fig:12 ] the time series recorded at 27 mk with a driving force of 4.6 pn shows a stochastic saw tooth pattern of the laminar phases .
the lifetimes of the turbulent phases can not be resolved any more .
the slope of the linear increase of the velocity amplitude during potential flow amounts to 6.9 @xmath137 m / s@xmath55 . inserting in eq .
the relevant numbers for @xmath30 = 4.6 pn , @xmath138 19 mm / s , @xmath10 = 27 @xmath9 g , and @xmath133 = 4.4 @xmath134 kg / s , we find a slope of 7.0 @xmath137 m / s@xmath55 , in agreement with the experimental value . from eq
. follows that the lifetimes of the phases of potential flow now also have a rayleigh distribution . from the measured rms value @xmath112 = 6.5 mm / s ( resonance frequency 107 hz ) of the @xmath111 distribution ,
we obtain the rms lifetime @xmath139 = 93 s. the second term in the brackets of the rhs of eq . which is due to the residual damping , is here 18% of the first one , and decreases further at larger drives .
finally , only the first term will dominate , i.e. , we may neglect damping and approach the case of an undamped oscillator : @xmath140 the rms lifetime @xmath141 of the potential flow is then proportional to @xmath142 : @xmath143 and the failure rate , eq .
, would simply be given by @xmath144 we observe a weak frequency dependence of @xmath112 , see fig.[fig:13 ] .
for six different frequencies .
each data point is obtained from the average of several individual time series recorded at different temperatures and driving forces .
the error bars represent the standard deviations .
a power law fit that takes into account the error bars indicates a slope of @xmath145.,scaledwidth=70.0% ] a power law fit to the data indicates a slope of @xmath146 .
this may be interpreted as an @xmath147 law .
however , our frequency range is rather limited ( @xmath14 factor 2 ) and there is some scatter .
therefore , this dependence is not firmly established .
however , the data point at 212 hz is the mean of 14 individual @xmath112 data measured at different temperatures ( 28 mk and 300 mk ) and 11 different driving forces , and therefore can not simply be neglected .
a theoretical interpretation is not possible as long as a theory of the transition to turbulence in oscillating superflows is not available .
however , a strong support of this frequency dependence of @xmath112 comes from the failure rate eq .. in that case , it follows that @xmath148 , i.e. , the failure rate is proportional to the number of the completed cycles .
this appears to be plausible .
the analysis of our data allows us to obtain informations concerning the properties of the unstable phases of potential flow above the critical velocity .
the failure rate eq . is proportional to the increase of the oscillation amplitude beyond @xmath1 , and is independent of temperature and driving force .
the final breakdown of potential flow at @xmath103 is known to be affected by the remanent vorticity and occurs statistically during the switching process at mk temperatures . moreover
, the stability of the phases of potential flow when @xmath104 is reached ( see upper trace in fig.[fig:9 ] ) leads us to assume that in this case the intervortex spacing @xmath57 of the remanent vorticity @xmath58 can not be reached by the sphere : @xmath149 40 @xmath9 m ( less than half of the radius of the sphere ) .
defining @xmath150 we obtain from eq .
the relation @xmath151 the rms value @xmath112 of the @xmath111 distribution is now related to the rms value @xmath152 of the @xmath153 distribution : @xmath154 inserting @xmath112 = 6.5 mm / s and @xmath1 = 19 mm / s ( data from fig.[fig:12 ] ) , we get @xmath155 = 0.34 and thus an average @xmath120 = 1.34 @xmath156 or an average @xmath157 = 0.56 .
this means that on the average 56 % of the vortex density at @xmath1 have survived the breakdown of the turbulent phase . but
as long as the value @xmath158 is unknown for oscillatory flow , we can not determine absolute values of the vortex densities .
also , since our sphere radius @xmath8 may be larger than the intervortex distance @xmath120 , it is rather peculiar that the transition to turbulence is not triggered by the vortices likely to be attached to the sphere .
however , it is possible that such vortices are absent , and in the laminar state the surface of the sphere becomes free from vorticity , or is only covered by very tiny loops which require a large velocity in order to expand .
the transition to turbulence can then occur at oscillation amplitudes that are smaller than @xmath8 .
this would happen when the sphere comes again in contact with the tangle that still remains in the vicinity of the sphere .
the correct physical picture requires experiments that could visualize the vortices around the sphere or realistic computer simulations .
we understand now why intermittent switching is not observed above 0.5 k : because of the large phonon drag , the velocity increase @xmath108 is rather small and so is the failure rate . only by increasing the drive , the laminar phase will finally break down and the turbulent phase will be stable . only a considerable reduction of the drive will cause a transition back to potential flow . hence , instead of intermittent switching , a hysteresis is observed at higher temperatures .
hysteresis and switching of the flow have been observed also with vibrating wires @xcite and tuning forks @xcite , and the significance of remanent vorticity for the critical velocity @xmath56 at breakdown was demonstrated . in our work
we are relating @xmath56 directly to the intervortex spacing of the remanent vortex density .
+ the turbulent phases are found to be exponentially distributed as @xmath159 , see fig.[fig:14 ] .
= 24.8 s. very short lifetimes ( below ca . 1 s ) could not be resolved .
( color figure online ) ] the mean lifetime @xmath160 increases very rapidly with increasing driving force @xmath30 , namely as @xmath161 \
, , \label{eq:21}\ ] ] see fig.[fig:15 ] .
these results were confirmed by the osaka group in experiments with a vibrating wire @xcite . + from a fit of eq . to the data , we obtain for the 119 hz oscillator @xmath162 = 0.5 s and @xmath163 = 18 pn , while for the 160 hz oscillator we have @xmath162 = 0.25 s and @xmath163 = 20 pn .
although we have data only for two frequencies , it is possible that @xmath163 scales as @xmath164 because @xmath165 = 1.16 and 20/18 = 1.11 . in the following ,
we will first concentrate on the properties of @xmath163 by using dimensional arguments .
he ( red squares ) ; at 30 mk with 0.5% of @xmath7he ( blue diamonds ) . note that the slopes @xmath166 and the intercepts @xmath162 are independent of temperature and @xmath7he concentration , but both depend on the oscillation frequency .
( color figure online ) ] the force @xmath163 must also depend on the density of liquid helium @xmath167 kg / m@xmath7 , a length scale which we take as the radius @xmath168 mm of the sphere , and the circulation quantum @xmath169 m@xmath55/s .
the result is @xmath170 .
the latter expression is more appropriate because @xmath171 determines the critical velocity , see above . of course , we must allow for a numerical factor @xmath25 which can be obtained only from the experiment : @xmath172 from our experimental results for @xmath163 we find @xmath173 .
we propose the following model for @xmath163 .
we assume that a vortex ring of the size @xmath8 is shed by the sphere during each half - period @xmath174 .
the power lost by the sphere amounts to the energy @xmath175 of the ring per half - period .
for @xmath175 we take the usual formula , namely @xmath176 \ , .\label{eq:23}\ ] ] the term in the brackets is roughly 14 .
the dissipated power is then given by @xmath177 i.e. , @xmath178 this result agrees with eq . and with our experiment :
@xmath163 scales as @xmath179 and is independent of temperature , the numerical factor of 1.6 is close to our experimental value of 1.3 .
we interpret the number @xmath180 as the average number of vortex rings emitted per half - period .
our data in fig.[fig:15 ] lie in the range @xmath181 . in the regime where we observe the intermittent switching @xmath182
we may approximate @xmath30 , see eq .
, by @xmath183 we proceed by changing variables from forces to velocities , and write @xmath184 where the characteristic " velocity is @xmath185 = 0.40 mm / s .
we note that @xmath68 is of the same order of magnitude as the self - induced velocity of a vortex ring of size @xmath8 . using eq .
we can plot the normalized lifetimes @xmath186 \ , , \label{eq:27}\ ] ] see fig.[fig:16 ] . from eq .
it also follows that the force can now be written as a function of @xmath187 : @xmath188 . as a function of @xmath17 for the 119 hz oscillator at 301 mk ( blue squares ) and the 160 hz oscillator at 30 mk with 0.05% @xmath7he ( black dots ) .
note the rapid increase of @xmath189 by 3 orders of magnitude over the small velocity interval of ca .
0.7 mm / s . at the top axis ,
the corresponding values of the superfluid reynolds number are given , see text .
the dashed line is calculated from eq ..
( color figure online),scaledwidth=80.0% ] the exponential dependence of @xmath160 on @xmath190 may be described by rice s formula for gaussian fluctuations crossing a given level @xcite .
it is applied here to fluctuations of @xmath111 crossing @xmath1 , where turbulence breaks down .
the average number @xmath191 per unit time of crossings of a given level @xmath192 below a mean at zero with negative slope is proportional to @xmath193 $ ] , with @xmath194 being the variance of the fluctuations .
we assume , that the breakdown occurs at the first crossing .
thus , we have @xmath195 . comparing this with eq . , we set @xmath196 and @xmath197 = 0.34 @xmath68 = 0.27 mm / s .
therefore , @xmath198 determines the width of the gaussian fluctuations of the velocity .
these fluctuations can not be attributed to the sphere , because at @xmath76 0.1 k the relaxation time of the oscillator is larger than 1000 s. we thus must conclude , that the fluctuations are due to the flow velocity of the superfluid around the sphere . + the prefactor @xmath199 is interpreted to be twice the average time between two crossings of the level @xmath1 , i.e. , for @xmath162 = 0.5 s or 0.25 s there are 4 or 8 crossings per second which means that the velocity of superfluid around the oscillating sphere fluctuates at an average frequency of 2 hz or 4 hz , respectively .
this could not be detected in our experiments , nor can we firmly establish @xmath200 from those 2 values .
+ in this subsection we compare our own experiments in superfluid helium @xcite with those in a bec as observed by other authors , where a moving laser beam sheds vortices above a critical velocity .
in particular , the frequency @xmath201 , with which vortices are shed , is found to be similar .
beginning with our experiments , we obtain from the average number @xmath187 of vortex rings shed per half - period , see eq . , @xmath202 where the coefficient @xmath203 = 2@xmath204 .
at @xmath205 = 119 hz we obtain @xmath203 = 0.60 @xmath9m@xmath93 and at 160 hz @xmath203 = 0.80 @xmath9m@xmath93 . from @xmath206
we have a characteristic length scale , which is given here by @xmath207 = 0.48 @xmath208 .
this can be interpreted as the distance a vortex ring travels during one - half period .
the linear increase of @xmath203 with @xmath205 likely breaks down at very large frequencies , i.e. , when the period becomes shorter than the time it takes a vortex ring to be shed .
then our picture of shedding of individual rings may no longer be applicable .
reconnections and annihilations near the surface of the sphere will become important . in the opposite limit ,
when @xmath205 goes to zero for uniform motion , it is clear that @xmath201 must remain finite because vortices can clearly be created also in the case of steady motion .
therefore , there must be a crossover from the linear frequency dependence to a constant value .
in that case , the relevant length scale is assumed to be the radius @xmath8 of the sphere .
we set @xmath209 and obtain @xmath210 0.008@xmath9m@xmath93 .
this is a small but finite value instead of zero . for a sketch of the frequency dependence of @xmath203 , see fig.[fig:17 ] .
of @xmath211 as a function of the oscillation frequency @xmath212 . at small frequencies
the radius @xmath8 of the sphere is taken as the characteristic length scale , hence @xmath209 , whereas at large frequencies @xmath203 scales as @xmath213 , see eq .. the characteristic frequency that marks the transition between both regimes is given by @xmath214 .
numerical factors of order 1 are neglected .
( color figure online),scaledwidth=80.0% ] above we used analogous arguments to estimate the change of the critical velocity for oscillatory flow @xmath215 when @xmath87 approaches zero for steady motion , see section [ s.vc ] .
the characteristic length scale is assumed to change from the oscillation amplitude to the size of the sphere , reproducing the feynman critical velocity @xmath216 .
the transition occurs at a characteristic oscillation frequency @xmath217 .
it is remarkable , that @xmath218 is the same ( except for some numerical factor ) as the frequency where the shedding @xmath201 has its transition , see above .
thus , our experiments with the sphere in @xmath0he are clearly in the regime of oscillatory flow , concerning both @xmath1 and @xmath201 .
shedding of vortex dipoles in a stable and periodic manner has been observed recently at seoul national university by moving a repulsive gaussian laser beam steadily through a bec of @xmath219na atoms @xcite .
the shedding frequency @xmath201 is also given by @xmath220 , where now @xmath1 = 0.99 mm / s and @xmath203 = 0.25 @xmath9m@xmath93 .
because the beam was moved steadily , we assume that the relevant length scale is given by the size of the beam 2@xmath8 = 9.1 @xmath9 m , in accordance with the arguments presented above .
therefore , we estimate @xmath209 = 0.22 @xmath9m@xmath93 , in fair agreement with the experimental result , and @xmath221 = 3.7 mm / s ( where @xmath222 = 1.7 @xmath223 m@xmath55/s for the @xmath219na bec ) . if the beam would have been oscillating at a frequency substantially larger than @xmath224 = 803 s@xmath93 ( or 128 hz ) , we would expect a shedding frequency @xmath201 proportional to @xmath212 and a critical velocity scaling as @xmath225 , in accordance with our results in @xmath0he .
earlier experiments with a moving beam were performed by ketterle s group @xcite . in that work ,
the beam was moved back and forth at frequencies below 200 hz and the diameter of the beam was 10 @xmath9 m .
although the motion was not a sinusoidal one , we can estimate @xmath226 680 s@xmath93 ( or 108 hz ) , which is of the same order of magnitude as the applied frequencies , and therefore the frequencies are too close to the steady regime to show the frequency dependencies of both @xmath201 and @xmath1 .
more details of the work can be found in @xcite where care was taken to distinguish between the critical velocity for heating the bec due to phonon emission by the moving beam and that of vortex production . in our experiments in helium ,
the situation is simpler because of the very different ratio of the critical velocity to the speed of sound that is of the order of @xmath227 in @xmath0he , while in the becs it is typically 0.1 to 0.5 .
therefore , in our case the critical velocity is not affected by phonon emission .
there are several numerical investigations @xcite confirming the linear increase of the energy dissipation above @xmath1 which is equivalent to a linear increase of @xmath201 .
that result agrees also with earlier work on the drag force at steady motion @xcite from which also a linear behavior of @xmath201 can be inferred .
both the dissipation and the drag were found to be caused by vortex shedding above a critical velocity .
this supports our general picture presented above .
recently , an oscillating object in a bec has been investigated theoretically @xcite , but shedding frequencies have not been calculated so far . the frequency range in that work
was limited to 0.2 @xmath228 1.0 , and therefore is also in the regime of steady flow , which is also evident from the frequency independence of the critical velocity for multiple vortex pair production @xmath229 . a very recent work by v.p .
singh et al .
@xcite presents exciting new details of the effect of stirring a bec with a laser beam .
if the stirrer is repulsive , the critical velocity is governed by vortex production , while for an attractive one , phonon emission is the relevant mechanism for heating . both circular and linear motion of the beam were used .
vortex shedding frequencies were not an issue in that work but an estimate of the characteristic frequency @xmath218 gives a value of about 100 s@xmath93 which is of the same order of the rotation frequency , decomposed into 2 linear oscillations .
therefore , the flow regime was probably near the steady limit .
more investigations on vortex shedding in becs and in superfluid helium could demonstrate the similarity of both types of superfluids .
+ in a numerical investigation of the 2-dimensional gross - pitaevskii equation @xcite a new superfluid reynolds number has been introduced to describe the onset of vortex shedding : @xmath230 solving for @xmath111 and setting the characteristic length scale @xmath231 , we obtain from eq .
@xmath232 where c = 1.04 .
+ in fig.[fig:16 ] the dependence of @xmath233 on @xmath234 is shown at the top axis ( note that @xmath234 = 0 corresponds to @xmath111 = 0 and @xmath233 = 1 ) .
all values of @xmath234 are below 3 .
an extension towards larger values of @xmath234 and @xmath233 would require time series lasting much longer than our longest one of 36 hours .
therefore , the validity of eq . can not safely be extrapolated .
nevertheless , we can state that the fast increase of @xmath233 with @xmath234 is a new example of supertransient chaos , where the lifetime of the turbulent state grows faster than exponentially with the reynolds number @xcite , here for the first time in a superfluid .
summarizing our experiments with the oscillating sphere in superfluid @xmath0he we conclude , that we have a fairly good understanding of the flow in the various regimes .
however , several of our results require more work , both experimental and theoretical . in particular , a theory of the transition to turbulence for oscillatory flow is needed in order to understand the origin of the critical velocity @xmath235 and to calculate the vortex density @xmath236 , especially at @xmath1 .
furthermore , the frequency dependence of the rms of the rayleigh distribution of the laminar phases @xmath237 , and the lifetimes of the turbulent phases @xmath238 need to be established and explained . still surprising and unexplained , even after 15 years of publication , is the linear temperature dependence of the laminar drag in the dilute @xmath7he - @xmath0he mixtures . on the experimental side , it would be interesting to use other spheres , having a different size and a smoother surface .
an extension of our method to superfluid @xmath7he will probably yield more surprises .
finally , more experiments on vortex shedding in becs should test our predictions concerning the diameter of the beam and its oscillation frequency .
\1 . the experimental setup : the measuring cell is shown in fig.[fig:18 ] .
the central part is the parallel plate capacitor made of niobium with the sphere in between the electrodes .
the sphere of diameter 240 @xmath64 m ( @xmath239 8 @xmath64 m ) as seen through a microscope is depicted in fig.[fig:19 ] .
a schematic of the electronics is shown in fig.[fig:20 ] .
embedded in a polymer matrix .
while the overall structure is rather spherical , surface protrusions are visible .
( color figure online),scaledwidth=40.0% ] \2 .
the experimental procedure : the evacuated measuring cell is cooled down . just before the niobium capacitor becomes superconducting we apply several 100 volt to the capacitor .
the surface charge of the sphere puts it into an up - and - down motion that is detected by the electrometer .
the motion stops when the capacitor plates are superconducting .
the high - voltage supply is disconnected . + after some waiting time at 4 k ( usually overnight ) we detect the resonance frequency simply by knocking gently at the cryostat that is isolated from vibrations of the laboratory .
the signal from the vibrating sphere is inspected on a scope to make sure that there are no beats , and that the signal is sinusoidal within our range of amplitudes .
the oscillator has a nonlinear return force ( duffing oscillator ) with an amplitude dependent resonance frequency @xcite .
we continue to cool the cell further down .
+ next , we determine the background damping of the sphere in the empty cell from the exponential free decay .
finally , we fill the cell with superfluid helium and begin the data acquisition by applying small driving forces at constant temperature to measure the linear damping .
a free exponential decay of the signal gives the time constant , as described in the introduction .
every morning we check for stability of the charge by reproducing some data of the day before .
usually the charge is quite stable . +
a change of the resonance frequency requires a new levitation status that is achieved first by evacuating the cell while heating it above @xmath240 of nb and then following the same procedure as above .
for various @xmath7he concentrations x@xmath241 , for `` natural '' and for purified @xmath0he . note that from 5@xmath24210@xmath243 to 5@xmath24210@xmath244 @xmath82 scales as @xmath245 .
the solid line is calculated for ballistic quasiparticle scattering in pure @xmath0he .
( background damping subtracted).,scaledwidth=60.0% ] [ [ the - linear - drag - force - due - to-3he - impurities - in - dilute - mixtures ] ] * the linear drag force due to @xmath7he impurities in dilute mixtures * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ we have extended our experiments to dilute @xmath7he - @xmath0he mixtures in order to measure the drag caused by the @xmath7he impurities . especially in the limit of ballistic scattering at low concentrations and low temperatures , we had expected the drag due to @xmath7he - scattering to be given simply by @xmath246 where @xmath25 is a numerical factor depending on details of the scattering mechanism , @xmath39 is the density of the @xmath7he atoms having an effective mass @xmath247 , and a thermal velocity @xmath248 .
this would lead to @xmath249 , where @xmath250 is the @xmath7he concentration . from fig.[fig:21 ] we see that instead @xmath82 scales as @xmath245 .
so far , there is no explanation for this peculiar result that we know of @xcite . for more details of our work
we refer to the original publication @xcite .
we acknowledge early contributions by k. gloos , j. simola , and j. tuoriniemi ( then all at helsinki university of technology , finland ) , who invented the technique to oscillate a levitating particle , and shared their expertise with us
. experiments with a levitating sphere immersed in superfluid helium were conducted at regensburg university from 1993 on by h. barowski and j. jger .
kerscher was a co - worker on the experiment in 2000 .
m. bleher ( federal office for radiation protection ( bfs ) , braunschweig ) performed the measurements of the dose rate of the natural background radiation in our laboratory . a helpful discussion with the late shaun fisher is gratefully acknowledged .
support and encouragement came from m. krusius ( aalto university , helsinki ) and from k.f .
renk ( regensburg university ) .
r.h . was supported by the academy of finland , and m.n . by the deutsche forschungsgemeinschaft .
w.s . is grateful to the o.v .
lounasmaa laboratory ( aalto university ) for the warm hospitality experienced during many visits .
w. schoepe , http://dx.doi.org/10.1134/s0021364015140106[jetp lett .
* 102 * , 105107 ( 2015 ) ] ; http://arxiv.org/abs/1506.05003v3[arxiv:1506.05003v3 ] [ cond-mat.other ] m.r
. leadbetter , g. lindgren , h. rootzn , _ extremes and related properties of random sequences and processes _ ( springer - verlag , new york , 1983 ) v.i .
tikhonov , http://dx.doi.org/10.1070/pu1963v005n04abeh003439[sov .
* 5 * , 594611 ( 1963 ) ] w. schoepe , http://dx.doi.org/10.1007/s10909-016-1653-2[j .
( 2016 ) ; doi:10.1007/s10909 - 016 - 1653 - 2 ] | the flow of superfluid @xmath0he around a translationally oscillating sphere , levitating without mechanical support , can either be laminar or turbulent , depending on the velocity amplitude . below a critical velocity @xmath1 that scales as @xmath2 , and is temperature independent below 1k , the flow is laminar ( potential flow ) . below 0.5k
the linear drag force is caused by ballistic phonon scattering that vanishes as t@xmath0 until background damping , measured in the empty cell , becomes dominant for t @xmath3 0.1k .
increasing the velocity amplitude above @xmath1 leads to a transition from potential flow to turbulence , where the large turbulent drag force varies as @xmath4 . in a small velocity interval @xmath5 above @xmath1 ,
the flow is unstable below 0.5k , switching intermittently between both patterns . from time
series recorded at constant temperature and driving force , the lifetimes of both phases are analyzed statistically .
we observe metastable states of potential flow which , after a mean lifetime of 25 minutes , ultimately break down due to vorticity created by natural background radioactivity .
the lifetimes of the turbulent phases have an exponential distribution , and the mean increases exponentially with @xmath6 .
we investigate the frequency at which the vortex rings are shed from the sphere .
our results are compared with recent data of other authors on vortex shedding by moving a laser beam through a bose - einstein condensate .
finally , we show that our observed transition to turbulence belongs to the class of `` supertransient chaos '' where lifetimes of the turbulent states increase faster than exponentially .
peculiar results obtained in dilute @xmath7he - @xmath0he mixtures are presented in the appendix . |
depinning phenomena and dynamic phases of collective transport through quenched disorder @xcite arise in a wide range of condensed matter systems including flux lines in superconductors @xcite , sliding charge density waves @xcite , moving electron crystals @xcite , magnetic skyrmions @xcite , and driven pattern forming systems such as electronic states with competing interactions @xcite . in materials science systems
such dynamics is relevant to sliding friction @xcite , dislocation motion @xcite , yielding transitions @xcite , and models of fault lines and earthquakes @xcite . in soft matter ,
similar dynamics occurs in the depinning of contact lines @xcite , driven colloidal motion on disordered substrates @xcite , or sliding of incommensurate colloidal structures on ordered substrates @xcite .
typically , these systems are in the pinned state at small external drives , and as the drive increases , a transition to a moving state occurs at a specific critical value of the external force @xcite . at higher drives , distinct types of sliding motions
can appear along with transitions between different dynamic states that can be deduced from features in the velocity - force curves .
the flow can be plastic or disordered @xcite , with the particles moving through riverlike channels@xcite , or elastic , where the particles maintain the same neighbors while moving @xcite .
transitions from disordered to more ordered or coherent flow can occur @xcite , such as from plastic flow to a moving smectic state @xcite .
such transitions are associated with changes in the moving structure morphologies , the density of topological defects @xcite , and the noise fluctuations @xcite . in all of the systems mentioned above , an external driving force produces a nonequilibrium condition . other types of nonequilibrium systems that involve no external driving force include self - propelled or active matter systems . these may be biological systems , such as swimming bacteria , or artificial swimmers , such as self - propelled colloidal systems , each of which can exhibit distinct nonequilibrium phases as a function of the activity or particle density @xcite . many types of active matter systems can be effectively modeled as a collection of sterically interacting disks undergoing driven brownian diffusion @xcite or run and tumble dynamics @xcite .
these disks can form uniform liquid states as well as phase separated states in which dense disk clusters are separated by a low density disk gas .
an open question is whether such systems would exhibit depinning transitions and different types of sliding states if an external drive were applied when the disks are coupled to a random substrate , as the driven dynamics of particles interacting with obstacles generally are distinct from those of particles on random pinning substrates .
previous studies of active matter systems driven through obstacle arrays showed that the average drift mobility varies non - monotonically with increasing activity , dropping when the system forms a phase - separated clump state @xcite .
other studies of active matter involving swarming or vicsek flocking models moving over disordered substrates in the absence of an external drive showed that there can be an optimal noise at which flocking occurs @xcite as well as transitions from flocking to non - flocking behavior @xcite . here
we use large scale gpu - based computer simulations to characterize the different states of run and tumble disks driven over a random pinning substrate , and show that a rich variety of distinct dynamical phases are possible that can be identified by the morphologies of the moving structures as well as by features in bulk transport properties . despite the additional complexity introduced by the self - propulsion
, we find several generic features in the dynamic phase diagrams that are similar to those observed for non - active driven systems such as superconducting vortex lattices , including a disorder to order transition at higher drives . in the limit of no quenched disorder
, we find a cluster or phase - separated state , while in the presence of strong quenched disorder , the system forms a uniform pinned liquid state which undergoes plastic depinning as the drive is increased , followed at higher drives by a transition to a stripe state coexisting with the pinned liquid state . at even higher drives
, there is a transition to a more fully phase separated state .
we also find strong differences in the dynamic phases for pinning arrays compared to those observed in obstacle arrays , where only a limited number of dynamic phases occur .
we perform a simulation of @xmath0 to @xmath1 run and tumble disks in a system with periodic boundary conditions of size @xmath2 with @xmath3 .
the disks have a short range repulsive interaction modeled as a harmonic spring force @xmath4 , where @xmath5 is the distance between the centers of the disks , @xmath6 is the displacement vector between the disk centers , @xmath7 is the spring constant , and @xmath8 is the heaviside function .
we use a disk radius of @xmath9 . the system density @xmath10 is defined by the area coverage of disks , @xmath11 , giving us a range of @xmath12 to @xmath13 .
we randomly place @xmath14 non - overlapping pinning sites which are modeled as parabolic traps that exert a force on the disks of the form @xmath15 , where @xmath16 is the displacement vector from the pinning site center to the disk center , @xmath17 , @xmath18 is the maximum pinning force , and @xmath19 is the pinning site radius .
we set @xmath20 to ensure that at most one disk can be pinned by a given pinning site .
we apply a uniform external drive @xmath21 on each disk in the @xmath22-direction and measure the resulting average drift velocity per disk in the direction of the drive @xmath23 to produce velocity - force curves .
the disk dynamics is obtained by integrating the following overdamped equation of motion : @xmath24 , where @xmath25 is the drag coefficient .
the disk self - propulsion is produced by the motor force @xmath26 which acts on a given disk in a fixed randomly chosen direction for a run time @xmath27 that is chosen from a uniform random distribution over the range @xmath28 $ ] .
after @xmath27 simulation time steps have elapsed , the motor force instantly reorients to a new randomly chosen direction , which it maintains for the next @xmath27 simulation time steps before reorienting again . in the absence of other disks ,
pinning sites , or obstacles , a disk would move a distance @xmath29 during one running time , where @xmath29 is uniformly randomly distributed over the range @xmath30 $ ] with @xmath31 .
we use @xmath32 and run times ranging from @xmath33 to @xmath34 simulation steps , with a simulation time step of @xmath35 . to initialize the system , we place the disks in random configurations and increase @xmath36 by increments of @xmath37 , spending @xmath38 simulation time steps at each increment . , @xmath39 , @xmath40 , and @xmath41 .
disks are colored according to the cluster to which they belong , with the largest cluster shown in blue , the second largest in red , the third largest in green , the fourth largest in purple , and the fifth largest in orange . (
a ) at @xmath42 , the system forms a uniform disordered state .
( b ) disk trajectories ( lines ) in a portion of the sample at @xmath43 , where the system is in a plastic flow state .
( c ) at @xmath44 , a dense stripe coexists with a pinned liquid .
( d ) at @xmath45 , a fully phase separated state appears .
, width=336 ]
in fig . [ fig:1 ]
we illustrate the disk locations for a system with @xmath46 , @xmath47 , @xmath40 , and @xmath41 . for these values of @xmath10 and @xmath48 , in the absence of pinning the system forms a phase separated state in which the disks condense into a single large dense cluster . when pinning is present , fig .
[ fig:1](a ) shows that at @xmath49 , the system forms a homogeneous disordered phase .
we identify disks that are in contact with one another using the luding and herrmann cluster algorithm @xcite , and plot the largest cluster in blue , the second largest cluster in red , and the next largest clusters in green , purple , and orange . in fig .
[ fig:1](a ) for @xmath50 , the largest cluster percolates across the entire sample . at @xmath43 in fig .
[ fig:1](b ) , we observe a plastic flow state , as indicated by the riverlike disk trajectories , in which the overall disk density remains uniform . as @xmath36 increases , a transition occurs into a moving stripe state as shown in fig .
[ fig:1](c ) for @xmath44 . here
a portion of the disks form a dense moving phase that is aligned with the driving direction .
the disks that are not in the dense phase are directly trapped by the pinning sites . as @xmath51 approaches @xmath52 ,
more of the pinned disks become mobile and the system enters the fully phase separated state illustrated in fig .
[ fig:1](d ) for @xmath45 , where a single large clump appears that has no particular orientation with respect to the driving direction .
vs @xmath51 for the system in fig .
[ fig:1](a ) with @xmath41 , @xmath53 , @xmath54 , and varied @xmath10 ranging from @xmath55 to @xmath56 , corresponding to @xmath57 to @xmath58 in the figure legend .
the letters * a * to * d * indicate the values of @xmath51 at which the images in fig .
[ fig:1 ] were obtained for the @xmath46 sample .
( b ) the corresponding @xmath59 vs @xmath36 curves .
( c ) the size of the largest cluster @xmath60 vs @xmath51 .
the dashed line indicates the transition into the moving stripe state , and the solid line denotes the transition to the moving phase separated state .
( d ) the number of six - fold coordinated disks @xmath61 vs @xmath51 . , width=336 ] in fig
. [ fig:2](a ) we plot the velocity - force curves for the system in fig .
[ fig:1 ] at disk densities ranging from @xmath55 to @xmath56 in increments of @xmath62 , while in fig .
[ fig:2](b ) we show the corresponding @xmath59 curves .
the letters * a * to * d * indicate the values of @xmath51 at which the images in fig .
[ fig:1 ] were obtained for the @xmath46 sample , demonstrating that the different phases correlate with distinct features in the transport curves .
previous studies of superconducting vortex systems showed that a peak in @xmath63 , the derivative of the velocity - force curve , is associated with a dynamical phase transition from plastic flow to a moving lattice state @xcite . in more complex systems , such as driven pattern forming systems , multiple peaks in the @xmath63 curves
coincide with multiple transitions in the structure of the moving state @xcite . in fig .
[ fig:2](a ) , a depinning transition occurs at a critical driving force called @xmath64 which shifts to lower values as @xmath10 increases .
the @xmath63 curves in fig .
[ fig:2](b ) have a double peak feature , with the largest peak near near @xmath65 and a smaller peak at lower values of @xmath51 .
both peak features become less distinct as @xmath10 increases .
the transition from a pinned state to plastic flow occurs at the point at which @xmath63 first rises above zero , while the moving stripe phase appears just above the first peak in @xmath63 .
the second peak in @xmath63 is associated with a transition to a moving fully phase separated state or to a uniformly dense state in which all the disks are in motion . in fig .
[ fig:2](c , d ) we plot the size of the largest cluster @xmath60 and the number of six - fold coordinated disks @xmath61 , obtained from a voronoi tessellation , versus @xmath51 for the system in fig .
[ fig:2](a , b ) . for clarity , both @xmath60 and @xmath61
are plotted in terms of the total number of disks and are not normalized to range between 0 and 1 . for @xmath66 ,
the system is in the moving disordered state for all values of @xmath51 . for @xmath67
, there is a pinned labyrinth phase similar to that shown in fig .
[ fig:1](a ) for @xmath68 , while for @xmath69 there is a pinned liquid state similar to that illustrated in fig .
[ fig:3](c ) at @xmath70 . at low drives ,
a drop in @xmath71 occurs when the labyrinth phase breaks apart into a moving disordered phase , as illustrated in fig .
[ fig:1](b ) .
the dashed line highlights an increase in @xmath60 that occurs when the system enters a moving stripe phase . in fig .
[ fig:2](c ) the thick solid line highlights an increase in @xmath60 that occurs near @xmath65 , which is also the location of the largest peak in @xmath63 . in fig .
[ fig:2](d ) , there is an upward jump in @xmath72 near @xmath65 , marked with a solid line , at the transition into the moving fully phase separated state . at lower @xmath36
there is a smaller jump in @xmath72 marked with a dashed line connected with the transition into a moving stripe state .
there is an increase in @xmath72 whenever dense regions of disks form since the densely packed regions have triangular ordering with disks that are mostly sixfold coordinated .
we note that in the labyrinth phase , @xmath60 can be large since a cluster can percolate across the entire system as illustrated in fig .
[ fig:1](a ) , but @xmath61 remains low since the disks within the labyrinth are disordered . ,
@xmath54 , and @xmath41 .
disks are colored according to the cluster to which they belong , with the largest cluster shown in blue . ( a ) the uniform disordered state for @xmath13 and @xmath67 .
( b ) the moving stripe state for @xmath13 and @xmath73 .
( c ) the pinned liquid state for @xmath74 and @xmath67 .
( d ) the moving phase separated state for @xmath74 and @xmath75 . , width=336 ] in fig .
[ fig:3](a ) we illustrate the pinned cluster state at @xmath67 and @xmath76 , while fig . [ fig:3](b ) shows the moving stripe state at the same disk density for @xmath73 .
figure [ fig:3](c ) shows the pinned liquid phase at @xmath67 and @xmath70 , where the clusters are absent , and fig .
[ fig:3](d ) illustrates the moving phase separated state at @xmath70 and @xmath77 .
vs @xmath10 constructed from the features in fig .
[ fig:2 ] for a sample with @xmath41 , @xmath54 , and @xmath78 .
i : pinned phase ; ii : plastic flow phase ; iii : moving stripe phase ; iv : moving fully phase separated state ; v : moving liquid phase . the dashed line in phase
i indicates the separation between a pinned liquid and a pinned labyrinth state .
( b ) the dynamic phase diagram as a function of @xmath51 vs @xmath18 for a sample with @xmath46 , @xmath40 , and @xmath39 .
the phases i through v are marked as above .
phase vi is the moving phase separated state in which some disks can be temporarily pinned .
, width=336 ] vs @xmath79 constructed from the features in fig .
[ fig:2 ] for a sample with @xmath46 , @xmath54 , and @xmath41 .
i : pinned phase ; ii : plastic flow phase ; iii : moving stripe phase ; iv : moving fully phase separated state ; v : moving liquid phase .
( b ) the dynamic phase diagram as a function of @xmath51 vs the number of pinning sites @xmath14 in samples with @xmath46 , @xmath41 , and @xmath39 .
, width=336 ] from the features in @xmath60 , @xmath61 , and the transport curves in fig .
[ fig:2 ] , we identify various dynamic phases as plotted in figs .
[ fig:4ab ] and [ fig:4cd ] . as a function of @xmath36 versus @xmath10 , shown in fig .
[ fig:4ab](a ) , we find five dynamic phases , marked i through v. there is a transition out of phase i , the pinned phase , at the depinning threshold @xmath80 , which drops to lower @xmath36 with increasing @xmath10 .
a dashed line marks the transition between @xmath69 , where the pinned clusters do not percolate and a pinned liquid of the type shown in fig .
[ fig:3](c ) forms , and @xmath81 , where a pinned percolating cluster appears as illustrated in fig .
[ fig:1](a ) and fig .
[ fig:3](a ) . in phase ii ,
plastic flow of the type illustrated in fig .
[ fig:1](b ) occurs , and the disk density is mostly homogeneous .
phase iii , the moving stripe phase , is shown in fig .
[ fig:1](c ) and fig .
[ fig:3](b ) .
figures [ fig:1](d ) and 3(d ) illustrate phase iv , the moving fully phase separated state , while in phase v , the moving liquid state , no clustering occurs but all the disks are moving .
vs @xmath36 in samples with @xmath46 , @xmath40 , and @xmath39 from the system in fig .
[ fig:4ab](b ) for different values of @xmath52 ranging from @xmath82 to @xmath83 .
as @xmath52 increases , a second peak appears and shifts to higher @xmath36 .
( b ) the corresponding @xmath84 vs @xmath51 .
, width=336 ] vs @xmath51 for samples with @xmath46 , @xmath41 , and @xmath40 for varied @xmath79 ranging from @xmath85 to @xmath86 .
the magnitude of the second peak in @xmath59 decreases with decreasing @xmath48 .
( b ) the corresponding @xmath87 vs @xmath36 curves .
( c ) the corresponding @xmath84 vs @xmath51 curves , indicating that phase iv disappears for @xmath88 .
, width=336 ] we have also examined the dynamic phases as a function of the pinning strength @xmath52 . in fig .
[ fig:5ab](a ) we plot @xmath59 versus @xmath51 for a system with @xmath46 and @xmath40 over the range @xmath89 to @xmath83 . a double peak in @xmath63 occurs only when @xmath90 , and both peaks shift to higher values of @xmath51 with increasing @xmath18 .
figure [ fig:5ab](b ) shows the corresponding normalized fraction of sixfold coordinated disks @xmath91 versus @xmath51 . for @xmath92 ,
the disks are disordered and @xmath93 .
when the system enters phase iii , a feature appears near @xmath94 , and there is a jump up in @xmath84 at the onset of phase iv . for low pinning strengths of @xmath95 , a moving phase separated state called phase vi appears in which some disks can be temporarily pinned , while for large enough @xmath36 all the disks are moving and the system enters phase iv , the flowing state .
the dynamic phase diagram as a function of @xmath36 versus @xmath52 in figure [ fig:4ab](b ) indicates that phase iii flow only occurs for @xmath96 , while phases i and iii grow in extent with increasing @xmath52 . in fig .
[ fig:5cde ] we plot the velocity - force and @xmath59 curves for a system with @xmath41 , @xmath46 , and @xmath97 for varied run lengths from @xmath98 to @xmath86 .
the curves follow each other closely except for the second peak in @xmath59 , which increases in magnitude with increasing @xmath79 .
the second peak is absent at lower @xmath48 when a transition from phase ii to phase v occurs , but materializes at larger @xmath48 once a transition from phase iii to phase iv begins to occur . in fig .
[ fig:5cde](c ) we plot the corresponding @xmath84 versus @xmath51 curves which show the onset of the transition into region iv in the form of an increase in @xmath99 . in fig . [ fig:4cd](a ) , the dynamic phase diagram as a function of @xmath36 versus @xmath48 for the system in fig .
[ fig:5cde ] shows that phases iv and iii only occur for @xmath100 . for @xmath88 , phase iv , the moving phase separated state , disappears , and the system passes directly from phase ii plastic flow to the phase v moving liquid . here
, the line marking the phase ii to phase v transition remains flat at @xmath101 .
we have also examined the case of fixed @xmath46 and varied @xmath48 to obtain the dynamic phase diagram as a function of @xmath36 versus @xmath102 in fig .
[ fig:4cd](b ) .
phase iii diminishes in width until @xmath103 . for larger @xmath102
, there is a direct transition from phase ii to phase iv , and phase iii flow disappears .
our results indicate that the dynamic phases we observe persist over a wide range of system parameters and represent generic features of this class of system .
vs @xmath51 for systems with @xmath46 and @xmath39 for varied numbers of obstacles @xmath104 to @xmath105 .
( b ) @xmath59 vs @xmath51 corresponding to the system in panel ( a ) , showing a lack of peak features .
( c ) the corresponding normalized @xmath84 vs @xmath51 has a smooth monotonic behavior .
( d ) the normalized cluster size @xmath106 vs number of obstacles @xmath107 for @xmath46 and @xmath39 at @xmath42 , showing a transition from a phase separated state at low @xmath36 to a uniform state at high @xmath36 . , width=336 ] and @xmath39 . ( a ) at @xmath67 and @xmath108 obstacles , a disordered state appears . ( b ) at @xmath45 and @xmath108 ,
the system is still disordered .
( c ) @xmath67 and @xmath104 .
( d ) at @xmath109 and @xmath104 , a moving stripe state forms.,width=336 ] active disks moving through obstacle arrays have a very different behavior than that described above for active disks in pinning arrays .
to explore this , we set up an obstacle landscape in which the obstacles are modeled as immobile disks with a radius of @xmath110 .
we consider a system with @xmath53 , a mobile disk density of @xmath46 , and @xmath107 obstacles .
an obstacle - free system at this disk density and running length @xmath48 forms a phase separated state for @xmath67 .
as we increase the number of obstacles @xmath107 , we find a transition from a phase separated state to a uniform liquid state , as indicated in the plot of the normalized value of @xmath106 versus @xmath107 in the inset of fig .
[ fig:6](a ) .
we show snapshots of the disk and obstacle positions in fig .
[ fig:7 ] . at @xmath50 , fig .
[ fig:7](a ) indicates that a sample containing @xmath108 obstacles is disordered , while fig .
[ fig:7](c ) shows that when @xmath104 obstacles are present , the system is density phase separated . in fig .
[ fig:6](a , b , c ) we plot the velocity @xmath87 , @xmath59 , and @xmath84 versus @xmath51 for varied numbers of obstacles @xmath107 .
as @xmath107 increases , the average value of @xmath87 monotonically deceases , and the @xmath63 curve contains no peaks , unlike the behavior for a pinning substrate .
there is a gradual decrease in @xmath84 with increasing @xmath51 , and there are no peaks or dips in @xmath99 of the type that appear for the pinning substrate . in general , we find that the obstacles produce no clear changes in the structure of the moving phase .
when the number of obstacles @xmath111 , the system undergoes plastic depinning into river - like channel flow , and remains in a uniform density , disordered flow state up to the highest drives @xmath36 that we considered , as shown in fig .
[ fig:7](b ) for a sample with @xmath108 at @xmath45 .
conversely , for @xmath112 , the system remains in a phase separated state , and forms a moving stripe structure resembling that illustrated in fig .
[ fig:7](d ) for @xmath104 at @xmath113 .
this result indicates that there are no dynamical transitions in the moving state for self - propelled disks moving through obstacle arrays ; however , there is a transition in the drive - free limit of @xmath67 as a function of obstacle density .
we note that in course of completing this work we became aware of an experimental paper by morin _
@xcite , who examined a colloidal flocking active matter system drifting through an obstacle array and found that as the obstacle number increases , the flow develops riverlike properties similar to that observed in other depinning systems and to what we find in fig .
[ fig:1](b ) .
in addition , morin _
_ report that there is a critical obstacle density above which the flocking behavior disappears , not unlike the cluster disappearance at high obstacle densities that we observe in fig .
[ fig:7](a ) .
differences between our system and that of morin _
_ include the fact that the underlying physics behind flock formation and cluster formation in the two systems is not the same ; additionally , morin _
_ considered only obstacles but not pinning sites .
the results of ref .
@xcite indicate that experiments on active matter interacting with a substrate are feasible .
in addition to obstacle arrays , it is also possible to have active matter move over landscapes of pinning sites created via optical means , as has been demonstrated in other recent experiments @xcite .
we have shown that active matter assemblies driven over random pinning arrays represent a new class of system that exhibit pinning , depinning and nonequilibrium phase transitions similar to those found for driven vortices and colloids moving over random disorder . in a regime where the system forms a phase separated state in the absence of pinning , we find that the addition of pinning causes the formation of a disordered uniform state that depins plastically into a flowing uniform state with river - like features . at higher drives ,
this is followed by a transition to a moving dense stripe phase coexisting with a pinned liquid , until at the highest drives , the system transitions into a moving fully phase separated state .
the different transitions are associated with features in the velocity - force and @xmath63 curves that are similar to the features observed in driven non - active systems with quenched disorder .
in contrast , for substrates composed of obstacle arrays , there are no dynamical transitions in the moving state and correspondingly there is a lack of features in the transport curves . at zero drive
, there is a transition from a phase separated state to a disordered state as the obstacle density increases .
this work was carried out under the auspices of the nnsa of the u.s .
doe at lanl under contract no .
de - ac52 - 06na25396 .
sndor and a. libl thank the nvidia corporation for their graphical card donation that was used in carrying out these simulations .
williams , p.a .
wright , r.g .
clark , e.y .
andrei , g. deville , d.c .
glattli , o. probst , b. etienne , c. dorin , c.t .
foxon , and j.j .
harris , conduction threshold and pinning frequency of magnetically induced wigner solid , phys .
* 66 * , 3285 ( 1991 ) .
i. buttinoni , j. bialk ' e , f. k " ummel , h. l " owen , and c. bechinger , dynamical clustering and phase separation in suspensions of self - propelled colloidal particles , phys .
lett . * 110 * , 238301 ( 2013 ) . | depinning and nonequilibrium transitions within sliding states in systems driven over quenched disorder arise across a wide spectrum of size scales ranging from atomic friction at the nanoscale , flux motion in type - ii superconductors at the mesoscale , colloidal motion in disordered media at the microscale , and plate tectonics at geological length scales . here
we show that active matter or self - propelled particles interacting with quenched disorder under an external drive represents a new class of system that can also exhibit pinning - depinning phenomena , plastic flow phases , and nonequilibrium sliding transitions that are correlated with distinct morphologies and velocity - force curve signatures . when interactions with the substrate are strong , a homogeneous pinned liquid phase forms that depins plastically into a uniform disordered phase and then dynamically transitions first into a moving stripe coexisting with a pinned liquid and then into a moving phase separated state at higher drives .
we numerically map the resulting dynamical phase diagrams as a function of external drive , substrate interaction strength , and self - propulsion correlation length .
these phases can be observed for active matter moving through random disorder .
our results indicate that intrinsically nonequilibrium systems can exhibit additional nonequilibrium transitions when subjected to an external drive . |
let @xmath19 be a simple undirected graph with @xmath20 vertices and @xmath21 edges .
we say that a vertex @xmath7 and an edge @xmath22 are incident if @xmath23 .
assign a weight @xmath24 . in this paper
we view @xmath25 as a flow in @xmath26 .
the value of @xmath25 at @xmath7 , denoted as @xmath9 , is given by @xmath27 . by abuse of notation
we view @xmath28 as column vectors in @xmath29 , respectively . for
a given set @xmath30 , @xmath25 is called an @xmath3-flow if @xmath31 . thus an @xmath32-flow is just a flow defined above .
given a vector @xmath33 we say that @xmath25 is a @xmath34-@xmath3-flow if the value of an @xmath3-flow @xmath25 on each vertex @xmath8 is @xmath35 .
let @xmath36 .
then @xmath34-@xmath3-flow is called a @xmath11-sum @xmath3-flow if @xmath37 for all @xmath7 and @xmath31 . in this paper
we study the existence problem of @xmath34-@xmath3-flow on undirected graphs .
the problem of finding @xmath11-sum @xmath38-flows was studied in the papers @xcite . for simplicity of exposition
we will assume that @xmath26 is a connected graph .
the existence of @xmath6-@xmath32-flow is a linear algebra problem .
let @xmath30 be an interval .
( it may be open , closed , half open , finite or infinite . )
then the existence of @xmath6-l - flow is a problem linear programming related to graphs .
see for example @xcite .
given a value @xmath6 and an interval @xmath3 the most basic is whether a graph @xmath26 has a @xmath34-@xmath3-flow or not .
if @xmath3 is the entire real line this a purely linear algebraic question , and when @xmath3 is a proper subinterval of reals we can apply methods from linear programming to find conditions for its solvability . in this section
we will first strengthen an existence result from @xcite for @xmath34-@xmath32-flows and then look at the case when @xmath3 is a proper subinterval .
let @xmath19 be a simple undirected graph .
let @xmath39\in{\mathord{\mathbb r}}^{v\times e}$ ] be the vertex edge incidence matrix of @xmath26 .
that is @xmath40 if @xmath41 and @xmath42 otherwise .
it is well known that @xmath43 is unimodular , i.e. all its minors have values in the set @xmath44 , if and only if @xmath26 is bipartite @xcite .
( see @xcite for a textbook reference . )
assume that @xmath26 is connected .
then @xmath45 if @xmath26 contains an odd cycle and @xmath46 if @xmath26 is bipartite @xcite .
the following result is a more detailed version of the result proved in @xcite .
[ existgamrflow ] let @xmath19 be a connected graph and @xmath47 is given .
then 1 .
if @xmath26 is not bipartite then there exists a @xmath34-@xmath32-flow .
furthermore , if @xmath48 then there exists a solution @xmath25 such that @xmath49 .
2 . assume that @xmath26 is bipartite and @xmath50 is the bipartite decomposition of vertices of @xmath26 .
then there exists a @xmath34-@xmath32-flow if and only if @xmath51 equivalently , let @xmath52 be a vector such that @xmath53 if @xmath54 and @xmath55 if @xmath56 .
that is , @xmath57 .
then @xmath58 is a basis of the null space of @xmath59 .
furthermore , if @xmath60 and the condition holds then there exists a solution @xmath61 .
recall that the existence of @xmath34-@xmath32-flow is equivalent to the solvability of the system : @xmath62 1 .
@xmath26 is not bipartite if and only if it contains an odd cycle .
so @xmath63 . hence is solvable .
we now show that if @xmath60 then there exists a solution @xmath25 such that @xmath64 . since @xmath26 is not bipartite it contains an odd cycle @xmath65 .
+ first , assume that @xmath65 is a hamiltonian cycle .
we can assume that @xmath66 .
we claim that @xmath67 .
indeed , @xmath68 , where @xmath69 is the permutation matrix corresponding to an odd cycle @xmath70 .
so @xmath71 since @xmath69 is a cyclic matrix of an odd order @xmath72 .
note that @xmath73 corresponds also to a cyclic matrix of order @xmath74 .
that is , @xmath73 is similar to @xmath69 . hence @xmath75 .
recall that the eigenvalues of @xmath69 are all the @xmath74-th roots of @xmath15 .
that is , @xmath76 .
let @xmath77 be @xmath78 roots of @xmath15 .
so @xmath79 here , @xmath80 is the @xmath81 elementary polynomial of @xmath77 for @xmath82 .
as @xmath77 are the roots of @xmath83 , it follows that @xmath84 for @xmath85 . as @xmath74 is odd , @xmath86 . hence @xmath67 .
+ next , assume that @xmath87 if @xmath88 .
let @xmath89 be the unique solution of @xmath90 .
so the coordinates of @xmath25 coincide with coordinates of @xmath91 on @xmath92 .
clearly , @xmath93 .
recall that @xmath94 . here , @xmath95 is the adjoint matrix of @xmath96 whose entries are minors of @xmath43 of order @xmath97 .
since the entries of @xmath96 are integers it follows that the entries of @xmath95 are integers .
hence @xmath98 and @xmath99 .
so @xmath100 and @xmath101 .
+ we now assume that @xmath65 is not hamiltonian .
let @xmath102 where @xmath103 is odd and @xmath104 .
delete the edge @xmath105 from @xmath65 to obtain a path @xmath106 .
extend @xmath106 to a spanning tree @xmath107 of @xmath26 .
let @xmath108 .
so @xmath109 has exactly one odd cycle @xmath65 .
we claim that @xmath110 .
observe if we delete the edges of @xmath92 in @xmath109 we obtain a forest .
hence , @xmath109 contains at least one vertex @xmath111 of degree @xmath15 .
expand @xmath112 by the row corresponding to @xmath111 .
then @xmath113 , where @xmath114 is obtained from @xmath109 by deleting the vertex @xmath111 . continue this process to deduce that @xmath115 .
+ let @xmath25 be the unique solution of where @xmath87 if @xmath116 .
the above arguments show that @xmath101 .
2 . assume that @xmath26 is bipartite and @xmath50 is the bipartite decomposition of @xmath117 .
clearly , @xmath118 . since @xmath119 then @xmath58 spans the null space of @xmath59 .
hence the system is solvable if and only if the condition holds .
+ let @xmath60 and assume that the condition holds .
we now construct a solution @xmath61 .
let @xmath107 be a spanning tree of @xmath26 .
let @xmath25 be the unique solution of such that @xmath87 if @xmath120 .
recall that @xmath121 .
since @xmath58 spans the null space of @xmath122 it follows any @xmath97 rows of @xmath123 are linearly independent .
let @xmath124 be a square submatrix of @xmath43 obtained by deleting a row in @xmath123 corresponding to a vertex @xmath125 .
denote by @xmath126 the vector obtained from @xmath6 be deleting coordinate @xmath9 . as @xmath123 unimodular it follows that @xmath127 .
hence the solution of the system @xmath128 is given by @xmath129 .
as @xmath130 for each @xmath131 we deduce that @xmath132 . in this section
we apply linear programming methods to study the conditions for existence of @xmath34-@xmath3-flow , where @xmath3 is an interval of @xmath32 . for simplicity of exposition
we assume that @xmath3 is a closed bounded interval @xmath133 $ ] .
our methods and arguments are closed to those given in @xcite .
we denote @xmath134=\{1,\ldots , n\}$ ] .
we will identify @xmath135 \quad e\equiv [ m ] , \quad { \mathord{\mathbb r}}^v\equiv{\mathord{\mathbb r}}^n,\quad { \mathord{\mathbb r}}^e\equiv { \mathord{\mathbb r}}^m,\ ] ] and no ambiguity will arise .
let @xmath136 be a column vector with @xmath21 coordinates equal to 1 . for two vectors
@xmath137 we denote @xmath138 if @xmath139 for @xmath140 .
we are looking for a solution of such that @xmath141 denote by @xmath142 the identity matrix of order @xmath143 and by @xmath144 the set of nonnegative real numbers .
let @xmath145 be the degree sequence of @xmath26 .
note that @xmath146 .
[ farkaslem ] consider the system satisfying the conditions .
then this system of equations is solvable if and only if the following conditions are satisfied : for each @xmath147 one has the inequality @xmath148 clearly , the system satisfying the conditions is equivalent to the following conditions .
@xmath149,\ ; { \mathbf{f}}= \left[\begin{array}{r}{\boldsymbol{\gamma}}\\-{\boldsymbol{\gamma}}\\b{\mathbf{1}}_m\\-a{\mathbf{1}}_m\end{array}\right].\ ] ] farkas lemma claims @xcite that the above system is solvable if and only if the following implication holds : @xmath150 where @xmath151 .
the equation @xmath152 is equivalent to @xmath153 the condition @xmath154 is equivalent to the inequalities @xmath155 ( note that if the above conditions hold , one can always choose @xmath156 such that @xmath157 . ) clearly , these conditions are satisfiable for @xmath158 and @xmath159 . finally , the condition @xmath160 is equivalent to the following the inequality @xmath161 set @xmath162 and recall that @xmath163 to deduce the lemma .
the condition can be stated as the following nonlinear inequality in @xmath164 .
let @xmath165 be an arbitrary vector in @xmath166 .
define @xmath167 as follows : @xmath168 then the condition is equivalent to @xmath169 we state an equivalent necessary and sufficient condition for solvability of the system satisfying the conditions which can be stated in terms of nonnegative solutions of a corresponding variant of .
[ equivcondsolv ] the following are equivalent : 1 .
the system satisfying the conditions is solvable .
the system @xmath170 is solvable .
the proof is straightforward by noting that @xmath5 is a solution satisfying - if and only if @xmath171 satisfies .
we now give the condition for the existence of nonnegative solutions of .
[ nonnegsol ] consider the system with @xmath172 .
then this system has a nonnegative solution if and only if @xmath173 as in the proof of lemma [ farkaslem ] the existence of nonnegative solutions of the system is equivalent to the system @xmath174,\ ; { \mathbf{f}}= \left[\begin{array}{r}{\boldsymbol{\gamma}}\\-{\boldsymbol{\gamma}}\\{\mathbf{0}}\end{array}\right].\ ] ] the above system is solvable if and only if each nonnegative solution of @xmath152 satisfies the inequality @xmath175 .
let @xmath176 , where @xmath177 and @xmath178 .
then the condition @xmath154 and @xmath179 are equivalent to the condition that @xmath180 , where @xmath181 .
the condition @xmath175 is equivalent to @xmath182 . note that
if we choose @xmath183 then @xmath184 and @xmath185 .
this implies .
we now restate our results for @xmath11-@xmath133$]-flows .
that is , we let @xmath186 .
[ char1pm1flow ] let @xmath19 be a simple undirected graph with no isolated vertices .
the following are equivalent : 1 .
@xmath26 has @xmath11-@xmath133$]-flow .
2 . if @xmath26 has a nonnegative @xmath187-flow such that the value of this flow on each edge is at most @xmath188 .
3 . for each @xmath147 one has the inequality @xmath189
once a graph has been shown to have a @xmath34-@xmath32-flow it is natural to ask which values the edge weights in such a flow can take . in this section
we will look at questions of this type for the specific case of 1-sum flows .
given a 1-sum flow on a graph @xmath26 we call the smallest interval which contains all the edge weights of the flow is called the range of the flow .
a natural question now is : given a graph @xmath26 , which is the shortest interval @xmath3 such that @xmath3 is the range of a @xmath15-sum flow on @xmath26 ?
starting from the other end we can also ask for a characterization of the graphs which have a 1-sum flow with range in some given interval @xmath3 .
we will prove some results of both these forms .
first we will look at 1-sum flows on trees , which have a unique 1-sum flow or none at all , and find the optimal range for this class of graphs .
after that we do the same for graphs with a single cycle , and then give some bounds for the range of 1-sum flows on general graphs .
after this we instead look at conditions guaranteeing that a graph has a 1-sum @xmath190$]-flow , or a non - negative flow . for a given graph @xmath19 and
the weight function @xmath191 , for each subset @xmath106 of @xmath192 we denote by @xmath193 .
we agree that @xmath194 . in this section
we analyze the the range of values of @xmath15-flow on a tree @xmath195 with @xmath74 vertices , i.e. @xmath20 and we let @xmath196=\{1,\ldots , n\}$ ] .
recall that @xmath197 and @xmath198 is bipartite .
let @xmath199 .
then the system is solvable if and only if the condition holds .
assume that holds .
we now estimate the coordinates of the solution of .
we perform the following _ pruning _ procedure of a tree @xmath198 .
let @xmath200 and @xmath201 be leaves . if @xmath202 or the star @xmath203 then we are done .
otherwise , let @xmath204 be the subtree of @xmath205 obtained by deleting the leaves @xmath206 and the corresponding @xmath207 edges . denote by @xmath208 the subset of edges attached to @xmath206 .
we now continue this process on @xmath204 .
we obtain a sequence of subtrees @xmath209 , where @xmath210 .
the leaves of @xmath211 are @xmath212 .
then @xmath213 for @xmath214 .
( @xmath215 . )
note @xmath216 indeed , if we delete all leaves of @xmath205 which are neighbors of @xmath111 , then it is possible that @xmath111 is not a leaf of @xmath204 . on the other hand
if @xmath111 is a leaf in @xmath204 then @xmath111 is not a leaf in @xmath205 and @xmath111 has at least leaf neighbor in @xmath205 .
we consider the system .
let @xmath217 and @xmath218 for @xmath219 and @xmath41 .
the values of @xmath220 is the value of @xmath221 , where @xmath222 is the unique edge in @xmath205 that contains the vertex @xmath223 .
let @xmath224 and @xmath225 be defined recursively as follows for @xmath226 : @xmath227 it is easy to see that each @xmath228 can appear at most in one of the coordinates of @xmath229 with coefficient @xmath230 .
( this is also follows from the condition . )
now consider @xmath231 .
assume first that @xmath232 .
so @xmath233 .
in order to be able to solve the original system one needs that condition @xmath234 .
assume @xmath210 , where @xmath235 .
let @xmath111 be the center of the star .
then the solvability condition is : @xmath236 in both cases , since each @xmath237 appears exactly once in some degree of @xmath231 with coefficient @xmath230 , we deduce that this is equivalent to the fact that a basis to the null space of @xmath238 is @xmath57 .
[ 1admisflowtree ] assume that a tree @xmath198 has @xmath15-flow , i.e. @xmath198 is a balanced bipartite graph .
let @xmath239 be the subtrees defined as above .
then the following conditions hold : 1 .
the unique flow is integer valued .
2 . @xmath240 for @xmath219 .
hence @xmath241 .
3 . if @xmath242 is even then @xmath243 for @xmath244 . if @xmath242 is odd then @xmath245 for @xmath244 .
@xmath246 5 .
let @xmath247 be the bipartite decomposition of the balanced tree @xmath198 .
then both @xmath248 and @xmath249 contain a leaf .
if @xmath250 then @xmath198 is a path and @xmath5 is @xmath251-flow . 7 .
if @xmath252 the @xmath198 has the shape `` t '' and @xmath5 is @xmath251-flow .
assume that @xmath253 .
then @xmath254 and the flow is 1-sum @xmath255 .
9 . in particular , the flow is in @xmath256 $ ] .
the lower bound achieved only for the unique tree @xmath257 , where @xmath257 is @xmath258 with appended @xmath259 vertices to each vertex of @xmath258 .
for @xmath257 we obtain that the flow is 1-sum @xmath260 flow .
the upper bound is obtained on for the unique tree @xmath261 , @xmath261 the path on @xmath262 vertices , @xmath263 , with @xmath264 vertices appended to each leaf of @xmath263 .
for @xmath261 we obtain the flow is 1-sum @xmath265 . 10 .
the other optimal tree @xmath266 , different from @xmath261 and @xmath257 , on @xmath267 vertices is obtained as follows .
take the path @xmath268 , add @xmath264 leaves at @xmath269 , @xmath270 leaf at @xmath271 and one leaf at @xmath272 .
then this flows is 1-sum @xmath273 . 1 .
this follows from part _ 2 .
_ of lemma [ existgamrflow ] .
2 . self evident .
use , the fact that @xmath274 and each @xmath240 for @xmath219 to deduce that that each @xmath275 for @xmath276 .
continuing in this manner , using we deduce the claim .
let @xmath277 be the subset of all edges in @xmath278 which are connected to @xmath279 .
( note that some leaves in @xmath280 may be connected to nonleaf vertices in @xmath281 . ) then summing the @xmath15-flow on all vertices in @xmath212 , for @xmath282 we get @xmath283 let @xmath284 .
as @xmath240 for each @xmath219 we deduce that @xmath285 .
this establishes for @xmath284 .
+ assume now that @xmath286 .
as @xmath275 for each @xmath276 the equality and the inequality yields the inequality for @xmath284 yields : @xmath287 this establishes for @xmath286 .
continuing in this manner we deduce for @xmath288 .
assume to the contrary that @xmath248 does not have a leaf .
so @xmath289 as @xmath198 is a balanced bipartite .
this is impossible .
hence @xmath248 contains a leaf .
similarly , @xmath249 contains a leaf .
@xmath290 it is enough to show that @xmath5 is 1-sum @xmath291 $ ] flow .
we will prove the claim on induction on @xmath74 .
in view of _ 6.-7 .
_ the claim holds for @xmath292 .
assume that the claim holds for all even @xmath74 , where @xmath293 .
assume that @xmath294 .
in view of _ 6.-7 .
_ we assume that @xmath295 .
let @xmath198 be a balanced tree on @xmath296 vertices with @xmath295 leaves .
let @xmath297 be the unique @xmath15-flow on @xmath198 .
let @xmath298 be two leaves of @xmath198 .
assume that @xmath299 .
take the path @xmath106 in @xmath198 connecting @xmath111 and @xmath8 given by @xmath300 .
note that there is the following flow on @xmath106 : @xmath301 let @xmath107 be the tree obtained from @xmath198 by deleting the vertices @xmath302 .
denote @xmath303 and assume that @xmath107 has @xmath304 leaves .
let @xmath305 be the unique @xmath15-flow on @xmath107 . then @xmath306 if @xmath307 and @xmath308 for @xmath309 .
so @xmath310 for each @xmath131 .
+ suppose first that @xmath311
. then @xmath312 .
by induction hypothesis @xmath313 this proves _ 8 . _ in this case .
+ suppose now that @xmath314
. then @xmath315 and @xmath316 .
delete vertices @xmath317 in @xmath198 to obtain a balanced tree @xmath318 with @xmath319 vertices and @xmath320 pendant vertices .
clearly @xmath321 .
also the @xmath15-flow on @xmath318 coincides .
use the induction hypothesis to deduce _
clearly the maximal number of leaves in a balanced tree is @xmath322 .
this equality is achieved only for the tree @xmath257 .
apply _ 8 . _ to deduce that the value of each @xmath15-flow on a balanced tree on @xmath74 vertices is not less than @xmath323 .
for @xmath257 the @xmath15-flow is @xmath260-flow .
other balanced tree on @xmath74 vertices have at most @xmath324 leaves .
_ to deduce that the value of each @xmath15-flow on a balanced tree on @xmath74 vertices is not more than @xmath264
. there are four nonisomorphic balanced trees with @xmath324 leaves . @xmath325 and @xmath326 .
@xmath327 is obtained from @xmath257 by deleting one leaf in @xmath328 and adjoining one vertex of a leaf in @xmath329 .
@xmath330 is obtained from @xmath261 by removing one leaf from @xmath331 and from @xmath332 and adjoining these two leaves to @xmath272 and @xmath333 , respectively .
for @xmath261 the @xmath15-flow is @xmath334 .
for @xmath266 the @xmath15-flow is @xmath273-flow .
for @xmath327 the @xmath15-flow is @xmath335-flow .
for @xmath330 the @xmath15-flow is @xmath336-flow . +
if @xmath198 has at most @xmath337 leaves then _ 8 _ implies that the range of @xmath15-flow is in @xmath338 $ ] .
we can also find a bound for the range of a 1-sum flow on a connected unicyclic graph , i.e. a graph which is obtained from a tree by adding a single edge .
as for trees we call a vertex of degree one a leaf , and just as for trees the number of leaves turns out to control the range of the 1-sum flows . the bound is also strongly dependent on whether the graph is bipartite or not , with bipartite graphs giving us a narrower range , and in each case we find graphs for which the stated bound is optimal .
[ 1flowconntr ] let @xmath19 be a connected unicyclic graph , with @xmath339 , which has a 1-sum flow .
assume that @xmath26 has @xmath340 leaves .
then one of the following conditions holds : 1 .
@xmath341 . in this case
@xmath26 is a cycle and has a 1-sum @xmath342-flow 2 .
if @xmath26 has a @xmath15-sum flow then it has a 1-@xmath344$]-flow 3 .
@xmath345 and @xmath26 is not bipartite .
then @xmath26 has a 1-sum @xmath346$]-flow . + this
bound is optimal for the graph obtained by taking the disjoint union of a triangle and @xmath347 and joining one vertex on the triangle to one of the leaves of the @xmath347 .
4 . @xmath345 and @xmath26 is a balanced bipartite graph .
then @xmath26 admits a 1-sum @xmath348$]-flow . +
this bound is optimal for the graph obtained by taking two copies of @xmath349 and joining the two high degree vertices by a six vertex path , giving a total of @xmath350 vertices in the graph , and then adding an edge so that the middle 4 vertices of the path form a 4-cycle . 1 .
set each the value on each edge to @xmath351 .
2 . if @xmath343 then @xmath26 consists of a cycle @xmath65 joined to a path @xmath69 by a single edge @xmath352 , where @xmath353 .
the flow on the path is uniquely determined , and is locally a flow with only values 0 and 1 .
if the flow on @xmath222 is 0 we can set the weight on every edge in @xmath65 to @xmath351 and we are done . if the flow on @xmath222 is 1 then @xmath354 is a path with an even number of vertices , since a 1-sum flow exists , and we can set the weight on a perfect matching in that path to 1 and 0 on the remaining edges , and so we have a flow on @xmath26 with only weights 0 and 1 .
we inductively assume that the theorem is true for smaller @xmath74 and @xmath355 .
let @xmath111 and @xmath8 be two leaves of @xmath26 . if @xmath111 is adjacent to a vertex @xmath356 of degree 2 then @xmath357 has a 1-sum @xmath346$]-flow , by induction on @xmath74 , and by setting the weight on the edge @xmath358 to 1 we can extend this to a 1-sum @xmath346$]-flow on @xmath26 , and we can follow the same procedure if @xmath8 is adjacent to a vertex of degree 2 . hence we can assume that @xmath111 and @xmath8 are not adjacent to vertices of degree 2 .
+ since @xmath26 is not bipartite there exists a walk @xmath359 of odd length in @xmath26 from @xmath111 to @xmath8 . by induction on @xmath355 the graph
@xmath360 has a 1-sum flow of the desired range .
we can now build a 1-sum flow on @xmath26 by setting the flow on the edges incident to @xmath111 and @xmath8 to 1 , and then alternatingly subtract and add 1 to the weight of the edges along @xmath356 . in this way we get a 1-sum flow on @xmath26 , and since @xmath109 had two less leaves than @xmath26 and our modification changed each weight by at most 2 , which happens if the edge was traversed twice by the walk @xmath359 , we get a flow of the desired range .
4 . in this case
@xmath26 is a balanced bipartite graph containing a single even cycle @xmath65 .
let @xmath111 and @xmath8 be two leaves of @xmath26 belonging to different parts of the bipartition .
if either one of them , say @xmath111 , is adjacent to a vertex @xmath356 of degree 2 then @xmath357 is also a balanced bipartite graph and , by induction on @xmath74 , it has a 1-sum flow with the desired range . by setting the weight on the edge @xmath361 to 1 and the weight on the other edge incident to @xmath356 to 0 ,
extend this to 1-sum flow of the desired range on @xmath26 .
hence we can assume that @xmath111 and @xmath8 are not adjacent to vertices of degree 2 .
+ since @xmath111 and @xmath8 are in different parts there exists a path from @xmath111 to @xmath8 in @xmath26 of odd length . by induction on @xmath355
the graph @xmath360 has a 1-sum flow of the desired range .
we can now build a 1-sum flow on @xmath26 by setting the flow on the edges incident to @xmath111 and @xmath8 to 1 , and then alternatingly subtract and add 1 to the weight of the edges along @xmath356 . in this way we get a 1-sum flow on @xmath26 , and since @xmath109 had two less leaves than @xmath26 and our modification changed each weight by at most 1 we get a flow of the desired range .
our last two results give optimal bounds for the range of a 1-sum flow on very sparse graphs . for denser graphs
there is a wider variety of behaviors and we do not have , or expect , an optimal bound in terms of any simple graph parameter .
when one considers denser graphs @xmath26 the expectation in general is that the optimal range should become narrower , especially since a 1-sum @xmath133$]-flow on a spanning subgraph can be extended to a 1-sum flow on @xmath26 with range @xmath362 $ ] in the obvious way . for a graph with a @xmath12-factor
this immediately leads to a 1-sum flow with a narrow range .
if @xmath26 has a @xmath12-regular spanning subgraph then @xmath26 has a @xmath15-sum @xmath363$]-flow .
we know that very dense graphs have 1-factors , and that random graphs with positive density have @xmath12-factors for quite large values of @xmath12 , but there are of course quite dense graphs which do not even have a 1-factor .
however all connected graphs have a spanning tree and using this fact our earlier results for trees implies a bound on the range for general graphs as well . [ 1flowgengraph ]
let @xmath364 be a connected graph .
then there exists a @xmath15-flow if and only @xmath26 is not a bipartite nonbalanced graph .
if a @xmath15-flow exists , then there exists a flow of the following type : 1 .
@xmath26 is a balanced bipartite graph with @xmath267 .
then there exists an integer valued flow with values @xmath365 .
@xmath26 is nonbipartite graph .
then there exists a flow @xmath366 such that @xmath367 .
moreover for @xmath254 its values are in the interval @xmath368 $ ] .
this results can be sharpened a bit by including information about the independence number @xmath369 of @xmath26 . in @xcite
it was proven that unless @xmath26 is a cycle , a complete graph , or a balanced complete bipartite graph it has a spanning tree the end vertices of which form an independent set in @xmath26 . using this we get the following .
[ 1flowbdinds ] let @xmath19 be a connected graph with independence number @xmath369 . 1 .
if @xmath26 is @xmath12-regular then @xmath26 has a @xmath15-sum @xmath370-flow .
2 . if @xmath26 is not regular and not bipartite then @xmath26 has a @xmath15-sum @xmath371$]-flow .
if @xmath26 is not regular , but is bipartite and balanced , then @xmath26 has a @xmath15-@xmath372$]-flow .
these bounds are quite far from the actual range for most graphs , since we know that for any fixed @xmath373 a random graph with minimum degree at least @xmath373 almost surely has an @xmath373-factor , and hence a 1-sum @xmath374$]-flow .
if a graph @xmath26 has @xmath12 disjoint spanning trees and we have a 1-sum flow @xmath375 on each tree @xmath198 then the average of these flows , seen as flows on @xmath26 , will also be a 1-sum flow on @xmath26 .
so in this case we can reduce the bounds in corollary [ 1flowgengraph ] by a factor of @xmath376 . here
we recall that nash - williams @xcite and tutte @xcite have characterized the graphs which have @xmath12 disjoint spanning trees , and so their characterization together with corollary [ 1flowgengraph ] give us a collection of graph classes with smaller ranges for their 1-sum flows . in principle
one could use the averaging procedure from the last paragraph for a collection of non - disjoint trees too . on one hand
we might now be averaging several positive weights on a single edge , in which case we are no longer guaranteed to gain a factor of @xmath376 over the single tree bound but on the other hand we might have both positive and negative weights on the same edge , and the cancellation could lead to even greater gains .
it would be interesting to see what can be said about a flow obtained in this way by taking a random collection of spanning trees in @xmath26 .
let @xmath377 be the set of doubly stochastic matrices .
that is @xmath378_{i , j=1}^n$ ] is a nonnegative matrix such that each row and column has sum @xmath15 .
denote by @xmath379 the group of @xmath380 permutation matrices .
recall the classical result of g. birkhoff @xcite , which is also called birkhoff - von neumann theorem @xcite .
namely , the extreme points of doubly stochastic matrices are the permutation matrices .
let @xmath381 be the subset of symmetric doubly stochastic matrices .
the following result due to m. katz @xcite : [ katzthm ] let @xmath382 be the set of symmetric doubly stochastic matrices .
then @xmath383 is an extreme point of @xmath382 if and only if @xmath384 for some permutation matrix @xmath385
. equivalently , there exists a permutation matrix @xmath386 such that @xmath387 , where @xmath388 and each @xmath389 is a doubly stochastic symmetric matrix of the following form : 1 . the @xmath390 matrix @xmath391 $ ] .
2 . @xmath392 .
@xmath393 , where @xmath65 is a cycle .
[ katzthmcor ] let @xmath394 be the set of symmetric doubly stochastic matrices with zero diagonal .
then @xmath395 is an extreme point of @xmath394 if and only if @xmath384 for some permutation matrix @xmath385 which does not fix any @xmath396 $ ] .
equivalently , there exists a permutation matrix @xmath386 such that @xmath387 , where @xmath388 and each @xmath389 is a doubly stochastic symmetric matrix of the forms 2 or 3 given in theorem [ katzthm ] .
let @xmath397 be a simple graph .
@xmath397 is called @xmath15-factor , or perfect matching , if each connected component is @xmath258 .
@xmath397 is called @xmath398-factor if each connected component of @xmath397 is either @xmath258 or a cycle .
@xmath26 has @xmath15-factor , or perfect matching , if @xmath26 has a spanning subgraph which is @xmath15-factor .
@xmath26 has @xmath398-factor if @xmath26 has a spanning subgraph which is @xmath398-factor .
[ exist1nonfl ] let @xmath19 be a simple graph
. then @xmath26 has @xmath15-@xmath344$]-flow if and only if one of the following conditions hold : 1 .
assume that @xmath26 is not bipartite .
then @xmath26 has @xmath398-factor .
2 . assume that @xmath26 is bipartite .
then @xmath26 has @xmath15-factor .
furthermore , @xmath26 has a @xmath15-@xmath399$]-flow if and only if for each @xmath22 one of the following conditions holds : 1 .
assume that @xmath26 is not bipartite .
then there exists @xmath398-factor of @xmath26 that contains @xmath222 .
2 . assume that @xmath26 is bipartite .
then there exists @xmath15-factor of @xmath26 that contains @xmath222 .
suppose first that @xmath26 is not bipartite .
assume that @xmath20 .
view @xmath196=\{1,\ldots , n\}$ ] and @xmath192 as a subset of all pairs @xmath400 , where @xmath401 $ ] . clearly , @xmath26 has @xmath15-@xmath344 $ ] if and only if there exists @xmath402_{i , j=1}^n \in \omega_{n , s,0}$ ] , such that @xmath403 if @xmath404 . corollary [ katzthmcor ] yields that @xmath65 is a convex combination of @xmath378 $ ] such that @xmath405 if @xmath406 .
take such an extreme point .
corollary [ katzthmcor ] implies that @xmath43 corresponds to a @xmath398-factor of @xmath26 .
vice versa , assume that @xmath397 is @xmath398-factor of @xmath26 .
let @xmath407 be the following flow on @xmath397 . on each edge of the connected component @xmath258 of @xmath397
the value of @xmath408 is @xmath15 . on each edge of the cycle in @xmath397
the value of the edge is @xmath351 .
extend this flow to @xmath409 by letting @xmath410 for @xmath411 . note that @xmath397 induces a unique extremal point @xmath412 .
assume that @xmath26 has @xmath15-@xmath344$]-flow .
denote by @xmath413 all the symmetric doubly stochastic matrices corresponding to the @xmath15-@xmath344$]-flow on @xmath26 .
let @xmath414 be all the extremal points of of @xmath413 .
so @xmath415 where @xmath416 is @xmath398-factor of @xmath26 .
so any @xmath15-@xmath344$]-flow is a convex combination of @xmath417 .
suppose there exists @xmath15-@xmath399$]-flow @xmath5 on @xmath26 .
let @xmath22 .
since @xmath418 it follows that @xmath222 is contained in some @xmath416 .
vice versa , suppose @xmath419 are all @xmath420 @xmath398-factors of @xmath26 .
assume that each @xmath22 is contained in some @xmath416 .
consider the @xmath15-flow @xmath421 .
then @xmath5 is @xmath15-@xmath399$]-flow .
assume now that @xmath26 is a bipartite graph .
so @xmath15-@xmath32-flow exists if and only if @xmath26 is balanced bipartite graph @xmath422 .
let @xmath423 .
so each edge @xmath22 is of the form @xmath424 .
then @xmath15-@xmath344$]-flow @xmath5 on @xmath26 corresponds to @xmath378_{i , j=1}^n\in\omega_n$ ] where @xmath405 if @xmath425 .
recall birkhoff s theorem which shows that @xmath426 is the set of extreme points on @xmath427 .
assume first that @xmath26 has @xmath15-@xmath344$]-flow @xmath5 .
then @xmath428 represents @xmath5 .
so @xmath429 where each @xmath430 , @xmath431 and @xmath432 .
hence each @xmath433 represents @xmath15-factor of @xmath26 .
vice versa , assume that @xmath26 has @xmath15-factor @xmath397 . the arguments above imply that @xmath434 is @xmath15-@xmath344$]-flow on @xmath26 . as in the non - bipartite graph
we deduce that @xmath26 has @xmath15-@xmath399$]-flow if and only if each edge is covered by some @xmath15-factor of @xmath26 .
the fundamental works of tutte give necessary and sufficient conditions for existence @xmath15 and @xmath398 factors @xcite .
[ exist01fdelge2 ] let @xmath26 be a graph and @xmath435 .
if @xmath26 has no even cycle , then @xmath26 admits a @xmath15-@xmath436$]-flow .
we claim that @xmath19 has a @xmath398-factor .
we prove this claim by induction on @xmath437 . for @xmath438
the claim is trivial .
consider the block decomposition of @xmath26 .
it is well - known that every block of @xmath26 is @xmath258 or an odd cycle , see @xcite . now , choose a leaf block of @xmath26 .
obviously , it is an odd cycle @xmath65 on 2l+1 vertices .
suppose first that @xmath65 has a common vertex @xmath8 , with another odd cycle @xmath439 .
remove all vertices of @xmath65 except @xmath8 .
the remaining graph @xmath109 satisfies the assumption of the corollary . by the induction hypothesis @xmath109 has a @xmath398-factor .
the subgraph of @xmath65 on @xmath440 vertices has a @xmath15-factor .
hence @xmath26 has a @xmath398-factor .
it is left to discuss the case where the leaf cycle has one vertex @xmath8 of degree @xmath441 which is common with a @xmath258-block .
consider the the shortest path , @xmath69 , between @xmath8 and another vertex of degree at least 3 , say @xmath442 .
remove all the vertices on @xmath65 and the path @xmath69 except the vertex @xmath356 .
the remaining graph @xmath443 has a @xmath398-factor by induction .
consider now the subgraph @xmath114 of @xmath26 on the vertices @xmath444 .
if the length of path is odd then @xmath114 has a @xmath398-factor consisting of @xmath65 and a matching , where the matching may be empty .
if @xmath69 is even then @xmath114 has a @xmath15-factor .
hence @xmath26 has a @xmath398-factor . in this section
we assume that @xmath26 is connected graph with the minimal degree @xmath447 at least @xmath448 .
we believe that it would be interesting to characterize the graphs which admit a 1-sum @xmath445$]-flow .
this class clearly extends the class of graphs which have a 1-sum @xmath344$]-flow , but the inclusion of negative edge weights adds more flexibility .
lemma [ farkaslem ] gives a necessary and sufficient conditions on the existence of these flows but at the moment we do not have a good interpretation of this result in terms of structural properties of the graphs .
however , we can show that not all graphs have a 1-sum @xmath445$]-flow , and in fact that given an integer @xmath449 there are graphs of arbitrarily high edge - connectivity which does not have a 1-sum @xmath450-flow .
we start out with a simple example and then proceed with the generalization to higher connectivity .
$ ] -flow . ]
[ examp1 ] the graph @xmath26 on @xmath451 vertices with @xmath452 given in figure [ fig1 ] , does not have @xmath15-sum @xmath445 $ ] flow .
a direct computation shows that the center edge of @xmath26 has weight 2 in all 1-sum flows and that the narrowest range is given by a @xmath15-sum @xmath453$]-flow .
[ examp2 ] for two positive integers @xmath449 and @xmath454 , there is an @xmath454-edge connected bipartite graph @xmath26 which admits a @xmath15-sum @xmath455-flow but admits no 1-sum @xmath450-flow .
consider two disjoint copies of @xmath456 .
call the vertex parts of the first one by @xmath457 and the second one by @xmath458 , where @xmath459 and @xmath460 .
choose an arbitrary vertex @xmath461 and join @xmath8 to all vertices in @xmath462 with the edges @xmath463 and call the resulting graph by @xmath26 .
we claim that @xmath26 is the desired graph .
clearly , @xmath26 is an @xmath454-edge connected bipartite graph .
note that @xmath26 is a balanced bipartite graph and so it admits a @xmath15-sum @xmath455-flow . by contradiction assume that @xmath464 is a @xmath15-sum @xmath450-flow of @xmath26
. then we have @xmath465 where @xmath466 is the sum values of all edges incident with @xmath467 .
this implies that @xmath468 .
we know that for each @xmath242 , @xmath469 , a contradiction .
characterize the graphs which admit a 1-sum @xmath445$]-flow .
as mentioned in the introduction the problem of finding a @xmath15-sum @xmath3-flow when @xmath3 is an interval is a linear programming problem .
as soon as @xmath3 is not an interval we are no longer working with a convex problem and many of the tools we have used so far do not apply .
nonetheless we shall prove some results for two cases of this type .
first we will consider the real line with the single point 0 removed , a second we will look at the case when @xmath3 consist of just a finite list of real numbers . it is not hard to see that if a graph @xmath26 admits a @xmath15-sum @xmath471-flow , then the order of @xmath26 should be even . in
[ 4 ] it has been proved that a connected bipartite graph admits a @xmath15-sum @xmath455-flow if and only if it is balanced .
let @xmath26 be a connected balanced bipartite graph .
then @xmath26 admits a @xmath15-sum @xmath16-flow if and only if the removing of every cut edge of @xmath26 does not make a balanced bipartite connected component .
first assume that @xmath26 admits a @xmath15-sum @xmath16-flow , say @xmath5 , and @xmath222 is a cut edge its removing makes a balanced bipartite connected component .
call this component by @xmath397 .
assume that @xmath457 be two vertex parts of @xmath397 and @xmath472 .
we have @xmath473 where @xmath474 denotes the sum of the values of all incident edges to @xmath8 .
this implies that @xmath87 , a contradiction .
now , assume that the removing of every cut edge does not make a balanced bipartite connected component .
let @xmath475 and for @xmath476 , @xmath477 is the set of all @xmath13-sum flows of @xmath26 in which the value of @xmath478 are zero .
let @xmath479 be the set of all @xmath13-sum flows of @xmath26 .
clearly , @xmath117 and @xmath480 are vector spaces over @xmath455 . by theorem 3 of [ 4 ] , there is a @xmath15-sum @xmath455-flow for @xmath26 . if @xmath481 , then there exists a @xmath13-sum @xmath16-flow @xmath482 of @xmath26 .
it is obvious that there exists a suitable real number @xmath483 such that @xmath484 is a @xmath15-sum @xmath16-flow and we are done .
now , assume that there exists @xmath485 such that for every @xmath486 , @xmath487 and for any @xmath488 , @xmath489 .
since @xmath455 is infinite , it is well - known that @xmath490 .
so , there exists a vector @xmath491 , such that the @xmath492th component of @xmath493 is non - zero for every @xmath486 .
now , let @xmath488 .
if @xmath494 is not a cut edge , then it is contained in an even cycle .
if we assign @xmath15 and @xmath495 to all edges of this cycle , alternatively and assign @xmath13 to all other edges of @xmath26 , we obtain a vector in @xmath496 , a contradiction .
hence @xmath494 is a cut edge of @xmath26 . by assumption
@xmath497 has a non - balanced bipartite component @xmath397 .
note that since @xmath26 is balanced and @xmath397 is not balanced , the other component , @xmath498 , is not balanced too .
let @xmath499 be two vertex parts of @xmath397 and @xmath500 . without loss of generality
assume that @xmath501 and @xmath494 is incident with @xmath8 .
assign 1 to every vertex in @xmath502 and assign @xmath503 to @xmath8 . then by theorem 3 of @xcite , @xmath397 admits a flow such that @xmath504 and @xmath505 , for each @xmath506 .
similarly , if @xmath507 and @xmath508 and @xmath494 is incident with @xmath509 , then there exists a flow for @xmath498 such that @xmath510 and @xmath505 , for each @xmath511 .
since @xmath26 is balanced , we have @xmath512 .
this yields that @xmath513 .
now , assign @xmath514 to @xmath494 to obtain a @xmath15-sum flow for @xmath494 .
note that since @xmath515 , in any @xmath15-sum @xmath455-flow of @xmath26 , the value of @xmath494 should be @xmath514 , which is non - zero .
it is not hard to see that there exists a suitable @xmath483 such that @xmath516 is a @xmath15-sum @xmath16-flow of @xmath26 , as desired .
let @xmath26 be a @xmath448-edge connected bipartite graph and @xmath517 and @xmath518 be two real numbers and @xmath519 .
then @xmath26 admits a @xmath15-sum @xmath3-flow if and only if @xmath26 admits a @xmath15-sum @xmath520-flow . to see this , by a theorem
, @xmath26 has a @xmath13-sum @xmath16-flow , say @xmath482 .
let @xmath5 be a @xmath15-sum @xmath3-flow of @xmath26 . then if @xmath521 is small enough , @xmath522 is a @xmath15-sum @xmath16-flow of @xmath26 .
consider @xmath34-flow on a connected @xmath26 .
assume that @xmath524 and consider @xmath34-@xmath525-flow on @xmath26 .
lemma [ existgamrflow ] applies also to @xmath525-flows .
let @xmath526 be a subset of @xmath525 of cardinality @xmath12 .
we now interested in the problem when there exists @xmath34-@xmath3-flow .
this is a problem in algebraic geometry .
define @xmath527 so , what we are asking for is a solution of the linear system and the polynomial conditions @xmath528 since @xmath529 we have an overdetermined system of equations with @xmath530 parameters which will have to satisfy @xmath143 polynomial conditions .
for relatively small graphs the existence of a solution can be determined by standard computer algebraic software like mathematica . 1 .
first we assign a bipartite graph @xmath397 to @xmath26 .
suppose that @xmath532 and let @xmath397 be a bipartite graph with two parts @xmath533 and @xmath534 .
join @xmath467 and @xmath535 if and only if two vertices @xmath242 and @xmath492 are adjacent in @xmath26 .
+ since @xmath397 is a @xmath12-regular bipartite graph , the edges of @xmath397 can be decomposed into @xmath12 , @xmath15-factors , @xmath536 .
assign @xmath537 to all edges of @xmath538 , @xmath539 .
so , for each vertex @xmath540 , we have @xmath541 . for two adjacent vertices
@xmath542 and @xmath543 in @xmath26 , assume @xmath544 is the edge between @xmath542 and @xmath543 .
let @xmath545 be the value of the edge @xmath546 , @xmath547 .
assign the value @xmath548 to @xmath544 . by our assumption , @xmath549 .
we have @xmath550 it is not hard to see that using this assignment we find a @xmath15-sum @xmath44-flow for @xmath26 .
since @xmath12 is even , @xmath26 is an eulerian graph and so it is @xmath448-edge connected .
now , by theorem 3.10 , part ( ii ) of @xcite , the edges of @xmath26 can be decomposed into two spanning @xmath551-regular graphs @xmath114 and @xmath552 . by part ( i )
, @xmath114 has a @xmath15-sum @xmath44-flow .
now , assign @xmath13 to all edges of @xmath552 .
so , @xmath26 admits a @xmath15-sum @xmath44-flow , as desired .
let @xmath556 be an odd positive integer .
then every @xmath373-regular graph admits a @xmath15-sum @xmath441-flow .
moreover , every @xmath448-edge connected @xmath373-regular graph admits a @xmath15-sum @xmath448-flow .
first let @xmath557 . by theorem [ kano ] , @xmath26 has a @xmath558$]-factor @xmath397 whose each component is regular .
now , assign @xmath495 to any edge in @xmath559 and @xmath15 to all edges of any @xmath441-regular component and @xmath448 to all edges of any @xmath448-regular component to obtain a @xmath15-sum @xmath441-flow .
now , let @xmath560 .
we have @xmath561 . by theorem [ kano ] , @xmath26 has a @xmath562$]-factor whose each component is regular .
let @xmath397 be the union of all @xmath449-regular components and @xmath563 be the union of all @xmath564-regular components of @xmath26 .
first assume that @xmath449 is odd .
assign @xmath15 to all edges in @xmath565 . also , assign @xmath495 to all edges of @xmath397 . since @xmath566 is even , by petersen theorem
, @xmath563 has a @xmath262-regular factor say @xmath3 .
since @xmath3 is a union of two @xmath448-factors , it admits a @xmath567-sum @xmath441-flow .
now , assign @xmath495 to all edges of @xmath568 to obtain a @xmath15-sum @xmath441-flow for @xmath26 .
now , let @xmath449 be even .
assign @xmath495 to all edges in @xmath565 .
assign @xmath15 to all edges of @xmath563 .
since @xmath449 is even , @xmath397 has a @xmath448-factor , say @xmath3 .
assign @xmath15 to all edges in @xmath569 and @xmath448 to all edges in @xmath570 to obtain a @xmath15-sum @xmath441-flow for @xmath26 .
1 . @xmath572 . by theorem 3.10 ,
part ( v ) of @xcite , @xmath26 has a @xmath449-factor .
thus @xmath573 can be decomposed into one @xmath449-factor and one @xmath574-factor .
assign @xmath448 and @xmath495 to each edge of @xmath449-factor and @xmath574-factor , respectively to obtain a @xmath13-sum @xmath441-flow for @xmath26 .
2 . @xmath575 .
since @xmath373 is odd , @xmath566 is odd . by theorem 3.10 ,
part ( v ) of @xcite , @xmath26 has a @xmath564-factor . by petersen
s theorem @xmath573 can be decomposed into one @xmath564-factor , one @xmath576-factor and two @xmath448-factors @xmath577 and @xmath578 .
now , assign @xmath448 , @xmath495 , @xmath567 and @xmath495 to each edge of @xmath564-factor , @xmath576-factor , @xmath577 and @xmath578 , respectively , to obtain a @xmath13-sum @xmath441-flow for @xmath26 .
3 . @xmath579 .
since @xmath373 is odd , @xmath580 is odd . by theorem 3.10 ,
part ( v ) of @xcite , @xmath26 has a @xmath581-factor .
so by petersen s theorem @xmath573 can be decomposed into one @xmath581-factor , one @xmath576-factor and one @xmath262-factor .
now , assign @xmath582 and @xmath567 to each edge of @xmath581-factor , @xmath576-factor and @xmath262-factor , respectively to obtain a @xmath13-sum @xmath441-flow for @xmath26 .
it is not hard to see if @xmath222 is a cut edge of a graph @xmath26 , then in any @xmath13-sum @xmath12-flow of @xmath26 , the value of @xmath222 should even .
now , let @xmath373 be an odd positive integer and @xmath26 be an @xmath373-regular graph containing a vertex @xmath8 such that all edges incident with @xmath8 is a cut edge . thus @xmath26 does not admit a @xmath13-sum @xmath262-flow . in @xcite and @xcite ,
it was proved that every @xmath373-regular graph ( @xmath584 ) admits a @xmath13-sum @xmath583-flow . | let @xmath0 be a simple undirected graph .
for a given set @xmath1 , a function @xmath2 is called an @xmath3-flow . given a vector @xmath4 we say that @xmath5 is a @xmath6-@xmath3-flow if for each @xmath7 , the sum of the values on the edges incident to @xmath8 is @xmath9 .
if @xmath10 , for all @xmath7 , then the @xmath6-@xmath3-flow is called a @xmath11-sum @xmath3-flow . in this paper
we study the existence of @xmath6-@xmath3-flows for various choices of sets @xmath3 of real numbers , with an emphasis on 1-sum flows . given a natural @xmath12 number , a _
@xmath11-sum @xmath12-flow _ is a @xmath11-sum flow with values from the set .
let @xmath3 be a subset of real numbers containing @xmath13 and denote @xmath14 .
answering a question from @xcite we characterize which bipartite graphs admit a @xmath15-sum @xmath16-flow or a @xmath15-sum @xmath17-flow . we also show that that every @xmath12-regular graph , with @xmath12 either odd or congruent to 2 modulo 4 , admits a @xmath15-sum @xmath18-flow .
_ keywords _ : @xmath3-flow , @xmath6-@xmath3-flow , c - sum flow , bipartite graph . *
2010 mathematics subject classification * : 0521 , 90c05 . |
the discovery of superconductivity in feas - based materials@xcite has re - opened new interest in superconductivity research . unfortunately , despite a large theoretical@xcite and experimental@xcite effort the nature of the superconductivity in these materials including the pairing mechanism and the symmetry of the order parameter remains unclear . one way to address the nature of the superconductivity is to study the low - temperature heat capacity .
the measurements of the temperature dependence of the specific heat and its magnetic filed response in superconducting state give an important information about the symmetry of the order parameter@xcite .
for optimally co doped material@xcite such a analysis suggest the presence of an anisotropic gap structure which may or may not have nodes ( see ref.@xcite ) .
moreover , the specific heat displays significant residual specific heat possibly coming from the non - superconducting fraction that is present in the system@xcite .
thus , the heat capacity at very low temperatures is very sensitive on the quality of the crystals studied and it is very often dominated by a schottky - like contribution@xcite .
this could make the the gap structure analysis quite complex . in this paper , we present results of our specific heat studies of optimally doped ba(fe@xmath0co@xmath1)@xmath2as@xmath2 samples .
we have measured two samples extracted from the same batch : sample one has been measured just after growing and without any additional heat treatment .
the sample shows @xmath3=20 k , residual specific heat @xmath4=3.6 mj / mol k@xmath5 and a schottky - like contribution at low temperatures .
sample two has been annealed at 800 @xmath6 for two weeks with @xmath3=25 k and @xmath4=1.4 mj / mol k@xmath5 . for this material
no schottky term is observed ( see fig.1 inset ) . then using similar approach as previously reported in ref.@xcite we extract and study the full electronic @xmath10-dependence for these two samples .
single crystals were grown out of feas flux with a typical size of about [email protected]@xmath110.2 mm@xmath12@xcite .
they crystalize in well - formed plates with the [ 001 ] direction perpendicular to the plane of the crystals .
the doping level was determined by microprobe analysis . after synthesis
a second piece of the sample was annealed for two weeks at 800 @xmath6 while the first one has not been heat treated .
the heat capacity was measured down to 400 mk using a thermal relaxation method implemented in a quantum design ppms-9 device .
co@xmath1)@xmath2as@xmath2 samples . circles : annealed sample with @xmath3 = 25 k. squares : non - annealed sample with @xmath3 = 20 k. the solid line describes the normal state specific heat ( see text ) .
inset : low - temperature specific heat of these two materials.,scaledwidth=58.0% ] fig.1 shows the temperature dependence of the specific heat of the annealed and non - annealed ba(fe@xmath0co@xmath1)@xmath2as@xmath2 samples . as may be seen the annealing process has increased @xmath3 quite remarkably from 20 k for the non - annealed sample up to 25 k for the annealed one . it is worth noting the transition temperature of about 25 k is one of the highest reported within the co doped ba122 family .
above @xmath3 the @xmath13 curves indicate very similar temperature dependence of the phonon part in these two materials as expected .
as may be seen from the inset of fig.1 the annealing strongly suppress the residual specific heat @xmath4 from 3.6 to 1.4 mj / mol k@xmath5 respectively for the non - annealed and annealed sample .
moreover , no schottky - like anomaly is observed for the annealed sample .
the considerably increase of @xmath3 , decrease of @xmath4 and lack of schottky - type contribution at low temperature indicate significant improvement of the quality of the single crystals after the heat treatment . in order to obtain the electronic part of the specific heat
, we have used the same approach used previously for the optimal and doped ba(fe@xmath14co@xmath15)@xmath2as@xmath2 samples ( see ref.@xcite ) .
we have assumed that the phonon contribution of the specific heat is independent of doping and we use the lattice specific heat obtained from the parent compound .
we separate the lattice contribution to the specific heat of the pure ba122 compound as @xmath16 = @xmath17 - @xmath18 where @xmath19 is the t@xmath200 intercept of @xmath21 of bafe@xmath2as@xmath2 as shown by a solid line in fig.1 .
so , the electronic specific heat is simply determined by @xmath22 .
the obtained normal state @xmath23 from above @xmath3 provide perfect entropy balance for the electronic specific heat in the superconducting state below @xmath3 for both , non - annealed@xcite and annealed samples . co@xmath1)@xmath2as@xmath2 samples .
solid and dashed lines are fits of the bcs @xmath24-model ( see text ) .
inset : normalized temperature dependence of the superconducting part of the specific heat.,scaledwidth=58.0% ] fig.2 shows the electronic part of the specific heat of ba(fe@xmath14co@xmath15)@xmath2as@xmath2 samples . for the non - annealed sample a small schottky contribution below 1 k has also been subtracted .
for both samples pronounced peaks are observed at the transition temperature . for the annealed sample
the specific heat exhibits a jump @xmath25 = 33.6 mj / mol k@xmath5 at @xmath3 .
taking @xmath26 = 22 mj / mol k@xmath5 the ratio @xmath27 = 1.53 is slightly larger then the weak - coupling bcs value 1.43 .
it is also larger then @xmath27 = 1.34 obtained for the non - annealed sample@xcite . below @xmath3
the @xmath28 data in fig.2 were analyzed by the bcs @xmath24-model : @xmath29\ ] ] where @xmath30 , @xmath31 is the fermi function @xmath32 , @xmath33 and @xmath34 ( see ref.@xcite ) plus residual normal state component .
as has been shown before a single @xmath8-wave gap scenario can not describe the specific heat data in co - doped ba122 material@xcite .
it clearly indicates the presence of an anisotropic gap structure in this material .
moreover , the specific heat data of optimally doped ba122 can not be accounted by using a clean d - wave gap aproach either@xcite . in order to sufficiently describe
the specific heat data of annealed ba(fe@xmath0co@xmath1)@xmath2as@xmath2 two separate @xmath8-wave gaps ( solid line ) have been used as shown in fig.2 .
this parametrization is not unique but provides a reasonable estimate of the dominant energy scales in the gap structure .
it gives parameters : @xmath35 = 1.83 mev , @xmath36 = 4.76 mev and @xmath37 = 0.72 .
the values of the gap are larger then the ones observed in non - annealed materials@xcite @xmath35 = 1.65 mev , @xmath36 = 3.75 mev and @xmath37 = 0.62 ( see dashed line in fig.2 ) .
it is consistent with the lower transition temperature observed in the non - annealed materials .
inset of fig.2 presents the electronic part of the specific heat of the superconducting portion of the ba(fe@xmath0co@xmath1)@xmath2as@xmath2 samples .
it has been obtained by subtracting the residual linear term @xmath4 , together with a small schottky contribution below 1 k for the non - annealed sample , and normalized by ( @xmath26-@xmath4 ) .
interestingly , these two curves match , indicating no significant change of the gap structure in these two samples .
in summary , using the low - temperature specific heat we explore details of the superconducting state of ba(fe@xmath0co@xmath1)@xmath2as@xmath2 with @xmath3 = 25 and 20 k. by subtracting the phonon specific heat the temperature dependence of the electronic specific heat has been studied .
the temperature variation of @xmath38 as well as its field dependence ( not shown ) can not be described by a single isotropic @xmath8-wave gap , indicating the presence of an anisotropic gap structure in the system .
the temperature variation of the electronic specific heat below @xmath3 may be well described by the presence of two superconducting gaps , pointing to a complex gap structure in the system .
it has been shown recently that for optimally co - doped bafe@xmath2as@xmath2 samples a minimum of two superconducting gaps are necessary to describe the temperature dependence of the electronic specific heat@xcite . moreover , for these two optimally doped samples with different quality the details of the superconducting state do not change significantly .
it is in agreement with the lack of doping dependence in ba(fe@xmath14co@xmath15)@xmath2as@xmath2@xcite , indicating that the gap structure does not change significantly as a function of doping@xcite .
this does not address the possibility that accidental nodes may or may not to be present and could be sample , doping , and/or material dependent .
+ work at los alamos national laboratory was performed under the auspices of the u.s .
department of energy , office of science and supported in part by the los alamos ldrd program .
research at oak ridge national laboratory is sponsored by the division of material sciences and engineering office of basic energy sciences . | we report the temperature dependence of the low - temperature specific heat down to 400 mk of the electron - doped ba(fe@xmath0co@xmath1)@xmath2as@xmath2 superconductors . we have measured two samples extracted from the same batch :
first sample has been measured just after preparation with no additional heat treatment .
the sample shows @xmath3=20 k , residual specific heat @xmath4=3.6 mj / mol k@xmath5 and a schottky - like contribution at low temperatures .
a second sample has been annealed at 800 @xmath6 for two weeks and shows @xmath3=25 k and @xmath4=1.4 mj / mol k@xmath5 . by subtracting the lattice specific heat , from pure bafe@xmath2as@xmath2 ,
the temperature dependence of the electronic specific heat has been obtained and studied .
for both samples the temperature dependence of @xmath7 clearly indicate the presence of low - energy excitations in the system .
their specific heat data can not be described by single clean @xmath8- or @xmath9-wave models and the data requires an anisotropic gap scenario which may or may not have nodes . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.