article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the advent of the new class of 10-m ground based telescopes is having a strong impact on the study of galaxy evolution .
for instance , instruments as lris at the keck allow observers to regularly secure redshifts for dozens of @xmath3 galaxies in several hours of exposure .
technical advances in the instrumentation , combined with the proliferation of similar telescopes in the next years guarantees a vast increase in the number of galaxies , bright and faint , for which spectroscopical redshifts will be obtained in the near future . notwithstanding
this progress in the sheer numbers of available spectra , the @xmath3 ` barrier ' ( for reasonably complete samples ) is likely to stand for a time , as there are not foreseeable dramatic improvements in the telescope area or detection techniques . despite the recent spectacular findings of very high redshift galaxies , ( @xcite , @xcite , @xcite ) , it is extremely difficult to secure redshifts for such objects . on the other hand , even moderately deep ground based imaging routinely contain many high redshift galaxies ( although hidden amongst myriads of foreground ones ) , not to mention the hubble deep field or the images that will be available with the upcoming advanced camera .
to push further in redshift the study of galaxy evolution is therefore very important to develop techniques able to extract galaxy redshifts from multicolor photometry data .
this paper applies the methods of bayesian probability theory to photometric redshift estimation . despite the efforts of thomas loredo , who has written stimulating reviews on the subject ( loredo 1990 , 1992 ) , bayesian methods are still far from being one of the staple statistical techniques in astrophysics .
most courses and monographs on statistics only include a small section on bayes theorem , and perhaps as a consequence of that , bayesian techniques are frequently used _
ad hoc _ , as another tool from the available panoply of statistical methods . however , as any reader of the fundamental treatise by e.t .
jaynes ( 1998 ) can learn , bayesian probability theory represents an unified look to probability and statistics , which does not intend to complement , but to fully substitute the traditional , ` frequentist ' statistical techniques ( see also bretthorst 1988 , 1990 ) one of the fundamental differences between ` orthodox ' statistics and bayesian theory , is that the probability is not defined as a frequency of occurrence , but as a reasonable degree of belief .
bayesian probability theory is developed as a rigorous full flegded alternative to traditional probability and statistics based on this definition and three _ desiderata _ : a)degrees of belief should be represented by real numbers , b)one should reason consistently , and c)the theory should reduce to aristotelian logic when the truth values of hypothesis are known .
one of the most attractive features of bayesian inference lies on its simplicity .
there are two basic rules to manipulate probability , the product rule @xmath4 and the sum rule @xmath5 where `` @xmath6 '' means `` @xmath7 and @xmath8 are true '' , and `` @xmath9 '' means `` either @xmath7 or @xmath8 or both are true '' . from the product rule , and taking into account that the propositions `` @xmath6 '' and `` @xmath10 '' are identical , it is straightforward to derive bayes theorem @xmath11 if the set of proposals @xmath12 are mutually exclusive and exhaustive , using the sum rule one can write @xmath13 which is known as bayesian marginalization .
these are the basic tools of bayesian inference .
properly used and combined with the rules to assign prior probabilities , they are in principle enough to solve most statistical problems .
there are several differences between the methodology presented in this paper and that of @xcite , the most significant being the treatment of priors ( see sec .
[ bpz ] ) .
the procedures developed here offer a major improvement in the redshift estimation and based on them it is possible to generate new statistical methods for applications which make use of photometric redshifts ( sec . [ appli ] ) .
the outlay of the paper is the following : sec .
2 reviews the current methods of photometric redshifts estimation making emphasis on their main sources of error .
3 introduces an expression for the redshift likelihood slightly different from the one used by other groups when applying the sed
fitting technique . in sec .
4 it is described in detail how to apply bayesian probability to photometric redshift estimation ; the resulting method is called bpz . sec 5
compares the performance of traditional statistical techniques , as maximum likelihood , with bpz by applying both methods to the hdf spectroscopic sample and to a simulated catalog .
6 briefly describes how bpz may be developed to deal with problems in galaxy evolution and cosmology which make use of photometric redshifts .
sec 7 briefly summarizes the main conclusions of the paper .
there are two basic approaches to photometric redshift estimation . using the terminology of @xcite ,
they may be termed ` sed fitting ' and ` empirical training set ' methods .
the first technique ( @xcite , @xcite , @xcite , @xcite , @xcite , etc . ) involves compiling a library of template spectra , empirical or generated with population synthesis techniques .
these templates , after being redshifted and corrected for intergalactic extinction , are compared with the galaxy colors to determine the redshift @xmath14 which best fits the observations . the training set technique
( @xcite , @xcite , @xcite ) starts with a multicolor galaxy sample with apparent magnitudes @xmath15 and colors @xmath16 which has been spectroscopically identified . using this sample , a relationship of the kind @xmath17
is determined using a multiparametric fit .
it should be said that these two methods are more similar than what it is usually considered . to understand this ,
let s analyze how the empirical training set method works . for simplicity ,
let s forget about the magnitude dependence and let s suppose that only two colors @xmath18 are enough to estimate the photometric redshifts , that is , given a set of spectroscopic redshifts @xmath19 and colors @xmath20 , the training set method tries to fit a surface @xmath21 to the data .
it must be realized that this method makes a very strong assumption , namely that the surface @xmath21 is a _ function _ defined on the color space : each value of @xmath16 is assigned one and only one redshift .
visually this means that the surface @xmath21 does not ` bend ' over itself in the redshift direction .
although this functionality of the redshift / color relationship can not be taken for granted in the general case ( at faint magnitudes there are numerous examples of galaxies with very similar colors but totally different redshifts ) , it seems to be a good approximation to the real picture at @xmath22 redshifts and bright magnitudes ( @xcite ) .
a certain scatter around this surface is allowed : galaxies with the same value of @xmath23 may have slightly different redshifts and it seems to be assumed implicitly that this scatter is what limits the accuracy of the method .
the sed fitting method is based on the color / redshift relationships generated by each of the library templates @xmath24 , @xmath25 . a galaxy at the position @xmath16 is assigned the redshift corresponding to the closest point of any of the @xmath26 curves in the color space .
if these @xmath26 functions are inverted , one ends up with the curves @xmath27 , which , in general , are not functions ; they may present self crossings ( and of course they may also cross each other ) .
if we limit ourselves to the region in the color / redshift space in which the training set method defines the surface @xmath21 , for a realistic template set the curves @xmath27 would be embedded in the surface @xmath21 , conforming its ` skeleton ' and defining its main features .
the fact that the surface @xmath21 is continuous , whereas the template - defined curves are sparsely distributed , does not have a great practical difference .
the gaps may be filled by finely interpolating between the templates ( @xcite ) , but this is not strictly necessary : usually the statistical procedure employed to search for the best redshift performs its own interpolation between templates .
when the colors of a galaxy do not exactly coincide with one of the templates , @xmath28 or the maximum likelihood method will assign the redshift corresponding to the nearest template in the color space .
this is equivalent to the curves @xmath27 having extended ` influence areas ' around them , which conform a sort of step like surface which interpolates across the gaps , and also extends beyond the region limited by them in the color space .
therefore , the sed - fitting method comes with a built - in interpolation ( and extrapolation ) procedure . for this reason ,
the accuracy of the photometric redshifts does not change dramatically when using a sparse template set as the one of @xcite ( @xcite ) or a fine grid of template spectra ( @xcite ) .
the most crucial factor is that the template library , even if it contains few spectra , adequately reflects the main features of real galaxy spectra and therefore the main ` geographical accidents ' of the surface @xmath21 the intrinsic similarity between both photometric redshift methods explains their comparable performance , especially at @xmath29 redshift ( @xcite ) . when the topology of the color
redshift relationship is simple , as apparently happens at low redshift , the training set method will probably work slightly better than the template fitting procedure , if only because it avoids the possible systematics due to mismatches between the predicted template colors and the real ones , and also partially because it includes not only the colors of the galaxies , but also their magnitudes , what helps to break the color / redshift degeneracies ( see below ) .
however , it must be kept in mind that although the fits to the spectroscopic redshifts give only a dispersion @xmath30 ( @xcite ) , there is not a strong guarantee that the predictive capabilities of the training set method will keep such an accuracy , even within the same magnitude and redshift ranges . as a matter of fact , they do not seem to work spectacularly better than the sed fitting techniques ( @xcite ) , even at low and intermediate redshifts .
however , the main drawback of the training set method is that , due to its empirical and _ ad hoc _ basis , in principle it can only be reliably extended as far as the spectroscopic redshift limit . because of this , it may represent a cheaper method of obtaining redshifts than the spectrograph , but which can not really go much fainter than it . besides it is difficult to transfer the information obtained with a given set of filters , to another survey which uses a different set .
such an extrapolation has to be done with the help of templates , what makes the method lose its empirical purity . and last but not least
, it is obvious that as one goes to higher redshifts / fainter magnitudes the topology of the color - redshift distribution @xmath31 displays several nasty degeneracies , even if the near - ir information is included , and it is impossible to fit a single functional form to the color - redshift relationship .
although the sed fitting method is not affected by some of these limitations , it also comes with its own set of problems .
several authors have analyzed in detail the main sources of errors affecting this method ( @xcite,@xcite ) .
these errors may be divided into two broad classes : fig .
. shows @xmath32 vs @xmath33 for the morphological types employed in sec [ test ] and @xmath34 .
the color / redshift degeneracies happen when the line corresponding to a single template intersects itself or when two lines cross each other at points corresponding to different redshifts for each of them ( these cases correspond to `` bendings '' in the redshift / color relationship @xmath21 ) .
it is obvious that the likelihood of such crossings increases with the extension of the considered redshift range and the number of templates included .
it may seem that even considering a very extended redshift range , such confusions could in principle be easily avoided by using enough filters .
however , the presence of color / redshift degeneracies is highly increased by random photometric errors , which can be visualized as a blurring or thickening of the @xmath35 relationship ( fig .
[ colors]b ) : each point of the curves in fig .
[ colors]a is expanded into a square of size @xmath36 , the error in the measured color .
the first consequence of this is a ` continuous ' ( @xmath37 ) increase in the rms of the ` small - scale ' errors in the redshift estimation , and , what it is worse , the overlaps in the color - color space become more frequent , with the corresponding rise in the number of ` catastrophic ' redshift errors . in addition , multicolor information may often be degenerate , so increasing the number of filters does not break the degeneracies ; for instance , by applying a simple pca analysis to the photometric data of the hdf spectroscopic sample it can be shown that the information contained in the seven @xmath38 filters for the hdf galaxies can be condensed using only three parameters , the coefficients of the principal components of the flux vectors ( see also @xcite ) .
therefore , if the photometric errors are large , it is not always possible to get totally rid of the degeneracies even increasing the number of filters .
this means that the presence of color / redshift degeneracies is unavoidable for faint galaxy samples .
the training set method somehow alleviates this problem by introducing an additional parameter in the estimation , the magnitude , which in some cases may break the degeneracy .
however , it is obvious that color / redshift degeneracies also affect galaxies with the same magnitude , and the training set method does not even contemplate the possibility of their existence ! the sed
fitting method at least allows for the existence of this problem , although it is not extremely efficient in dealing with it , especially with noisy data .
its choice of redshift is exclusively based on the goodness of fit between the observed colors and the templates . in cases as the one described above , where two or more redshift / morphological type combinations have practically the same colors , the value of the likelihood @xmath39 would have two or more approximately equally high maxima at different redshifts ( see fig . [ peaks ] ) .
depending on the random photometric error , one maximum would prevail over the others , and a small change in the flux could involve a catastrophic change in the estimated redshift ( see fig .
[ peaks ] ) .
however , in many cases there is additional information , discarded by ml , which could potentially help to solve such conundrums .
for instance , it may be known from previous experience that one of the possible redshift / type combinations is much more likely than any other given the galaxy magnitude , angular size , shape , etc . in that case , and since the likelihoods are not informative enough , it seems clear that the more reasonable decision would be to choose the option which is more likely _ a priori _ as the best estimate .
plain common sense dictates that one should compare all the possible hypotheses with the data , as ml does , but simultaneously keeping in mind the degrees of plausibility assigned to them by previous experience .
there is not a simple way of doing this within ml , at best one may remove or change the redshift of the problematic objects by hand or devise _ ad hoc _
solutions for each case .
in contrast , bayesian probability theory allows to include this additional information in a rigorous and consistent way , effectively dealing with this kind of errors ( see sec [ bpz ] ) in some cases , the spectra of observed galaxies have no close equivalents in the template library .
such galaxies will be assigned the redshift corresponding to the nearest template in the color / redshift space , no matter how distant from the observed color it is in absolute terms .
the solution is obvious , one has to include enough templates in the library so that all the possible galaxy types are considered . as was explained above , the sed fitting techniques perform their own ` automatic ' interpolation and extrapolation , so once the main spectral types are included in the template library , the results are not greatly affected if one finely interpolates among the main spectra .
the effects of using a correct but incomplete set of spectra are shown in sec [ test ] .
both sources of errors described above are exacerbated at high redshifts .
high redshift galaxies are usually faint , therefore with large photometric errors , and as the color / redshift space has a very extended range in @xmath14 , the degeneracies are more likely ; in addition the template incompleteness is worsened as there are few or no empirical spectra with which compare the template library .
the accuracy of any photometric redshift technique is usually established by contrasting its output with a sample of galaxies with spectroscopic redshifts .
it should be kept in mind , though , that the results of this comparison may be misleading , as the available spectroscopic samples are almost ` by definition ' especially well suited for photometric
redshift estimation : relatively bright ( and thus with small photometric errors ) and often filling a privileged niche in the color - redshift space , far from degeneracies ( e.g. lyman - break galaxies ) .
thus , it is risky to extrapolate the accuracy reached by current methods as estimated from spectroscopic samples ( and this also applies to bpz ) to fainter magnitudes .
this is especially true for the training set methods , which deliberately minimize the difference between the spectroscopic and photometric redshifts .
photometric redshift techniques based on template fitting look for the best estimate of a galaxy redshift from the comparison of its measured fluxes in @xmath40 filters @xmath41 , @xmath42 , with a set of @xmath43 template spectra which try to represent the different morphological types , and which have fluxes @xmath44 .
these methods find their estimate @xmath45 by maximizing the likelihood @xmath39 ( or equivalently minimizing @xmath28 ) over all the possible values of the redshift @xmath14 , the templates @xmath24 and the normalization constant @xmath46 .
@xmath47 since the normalization constant @xmath46 is considered a free parameter , the only information relevant to the redshift determination is contained in the ratios among the fluxes @xmath41 , that is , in the galaxy colors . the definition of the likelihood in eq .
( [ li ] ) is not convenient for applying bayesian methods as it depends on a normalization parameter @xmath46 , which is not convenient to define useful priors either theoretically or from previous observations .
here we prefer to normalize the total fluxes in each band by the flux in a ` base ' filter , e.g. the one corresponding to the band in which the galaxy sample was selected and is considered to be complete .
then the ` colors ' @xmath48 , are defined as @xmath49 @xmath50 , where @xmath51 is the base flux .
the exact way in which the colors are defined is not relevant , other combinations of filters are equally valid .
hereinafter the magnitude @xmath15 ( corresponding to the flux @xmath51 ) will be used instead of @xmath51 in the expressions for the priors . and
so , assuming that the magnitude errors @xmath52 are gaussianly distributed , the likelihood can be defined as @xmath53 where @xmath54[c_j - c_{tj}(z)]\ ] ] and the matrix of moments @xmath55 can be expressed as @xmath56 by normalizing by @xmath51 instead of @xmath46 , one reduces the computational burden as it is not necessary to maximize over @xmath51 , which is already the ` maximum likelihood ' estimate for the value of the galaxy flux in that filter .
it is obvious that this assumes that the errors in the colors are gaussian , which in general is not the case , even if the flux errors are .
fortunately , the practical test performed below ( sec .
[ test ] ) shows that there is little change between the results using both likelihood definitions ( see fig .
[ comparison]a ) .
within the framework of bayesian probability , the problem of photometric redshift estimation can be posed as finding the probability @xmath57 , i.e. , the probability of a galaxy having redshift @xmath14 given the data @xmath58 , _ and _ the prior information @xmath59 . as it was mentioned in the introduction ,
bayesian theory states that _ all _ the probabilities are conditional ; they do not represent frequencies , but states of knowledge about hypothesis , and therefore always depend on other data or information ( for a detailed discussion of this and many other interesting issues see jaynes , 1998 ) .
the prior information @xmath59 is an ample term which in general should include any knowledge that may be relevant to the hypothesis under consideration and is not already included in the data @xmath60 .
note that in bayesian probability the relationship between the prior and posterior information is _ logical _ ; it does not have to be temporal or even causal .
for instance , data from a new observation may be included as prior information to estimate the photometric redshifts of an old data set .
although some authors recommend that the @xmath61 should not be dropped from the expressions of probability ( as a remainder of the fact that all probabilities are conditional and especially to avoid confusions when two probabilities based on different prior informations are considered as equal ) , here the rule of simplifying the mathematical notation whenever there is no danger of confusion will be followed , and from now @xmath62 will stand for @xmath63 , @xmath64 for @xmath65 etc . as a trivial example of the application of bayes s theorem ,
let s consider the case if which there is only one template and the likelihood @xmath39 only depends on the redshift @xmath14 . then ,
applying bayes theorem @xmath66 the expression @xmath67 is simply the likelihood : the probability of observing the colors @xmath16 if the galaxy has redshift @xmath14 ( it is assumed for simplicity that @xmath39 only depends on the redshift and morphological type , and not on @xmath15 ) the probability @xmath68 is a normalization constant , and usually there is no need to calculate it . the first factor , the _ prior _ probability @xmath69 , is the redshift distribution for galaxies with magnitude @xmath15 .
this function allows to include information as the existence of upper or lower limits on the galaxy redshifts , the presence of a cluster in the field , etc .
the effect of the prior @xmath69 on the estimation depends on how informative it is .
it is obvious that for a constant prior ( all redshifts equally likely _ a priori _ ) the estimate obtained from eq .
( [ 1 ] ) will exactly coincide with the ml result .
this is also roughly true if the prior is ` smooth ' enough and does not present significant structure .
however , in other cases , values of the redshifts which are considered very improbable from the prior information would be `` discriminated '' ; they must fit the data much better than any other redshift in order to be selected .
note that in rigor , one should write the prior in eq .
( [ 1 ] ) as @xmath70 where @xmath71 is the ` true ' value of the observed magnitude @xmath15 , @xmath72 is proportional to the number counts as a function of the magnitude @xmath15 and @xmath73 $ ] , i.e , the probability of observing @xmath15 if the true magnitude is @xmath71 . the above convolution accounts for the uncertainty in the value of the magnitude @xmath15 , which has the effect of slightly ` blurring ' and biasing the redshift distribution @xmath69 .
to simplify our exposition this effect would not be consider hereinafter , and just @xmath69 and its equivalents will be used .
it may seem from eq . [ 1 ] ( and unfortunately it is quite a widespread misconception ) that the only difference between bayesian and ml estimates is the introduction of a prior , in this case , @xmath69 . however , there is more to bayesian probability than just priors .
the galaxy under consideration may belong to different morphological types represented by a set of @xmath43 templates .
this set is considered to be _ exhaustive _ , i.e including all possible types , and _ exclusive _ : the galaxy can not belong to two types at the same time . in that case , using bayesian marginalization ( eq . [ mar ] ) the probability @xmath74 can be ` expanded ' into a ` basis ' formed by the hypothesis @xmath75 ( the probability of the galaxy redshift being @xmath14 _ and _ the galaxy type being @xmath24 ) .
the sum over all these ` atomic ' hypothesis will give the total probability @xmath74 .
that is , @xmath76 @xmath77 is the likelihood of the data @xmath16 given @xmath14 and @xmath24 .
the prior @xmath78 may be developed using the product rule .
for instance @xmath79 where @xmath80 is the galaxy type fraction as a function of magnitude and @xmath81 is the redshift distribution for galaxies of a given spectral type and magnitude .
( [ bas ] ) and fig .
[ peaks ] clearly illustrate the main differences between the bayesian and ml methods .
ml would just pick the highest maximum over all the @xmath77 as the best redshift estimate , without looking at the plausibility of the corresponding values of @xmath14 or @xmath24 . on the contrary ,
bayesian probability averages all these likelihood functions after weighting them by their prior probabilities @xmath78 . in this way
the estimation is not affected by spurious likelihood peaks caused by noise as it is shown in fig .
[ peaks ] ( see also the results of sec .
[ test ] ) .
of course that in an ideal situation with perfect , noiseless observations ( and a nondegenerate template space , i.e , only one @xmath16 for each @xmath82 pair ) the results obtained with ml and bayesian inference would be the same . instead of a discrete set of templates ,
the comparison library may contain spectra which are a function of continuous parameters .
for instance , synthetic spectral templates depend on the metallicity @xmath83 , the dust content , the star formation history , etc . even starting from a set of a few templates
, they may be expanded using the principal component analysis ( pca ) technique ( @xcite ) . in general ,
if the spectra are characterized by @xmath84 possible parameters @xmath85 ( which may be physical characteristics of the models or just pca coefficients ) , the probability of @xmath14 given @xmath86 can be expressed as @xmath87 sometimes , instead of finding a ` point ' estimate for a galaxy redshift , one needs to establish if that redshift belongs within a certain interval .
for instance , the problem may be to determine whether the galaxy has @xmath88 , where @xmath89 is a given threshold , or whether its redshift falls within a given @xmath90 , e.g. in the selection of cluster members or background galaxies for lensing studies .
as an example , let s consider the classification of galaxies into the background - foreground classes with respect to a certain redshift threshold @xmath91 .
one must choose between the hypothesis @xmath92 and its opposite , @xmath93 .
the corresponding probabilities may be written as @xmath94 and @xmath95 the ( ` bookmaker ' ) odds of hypothesis @xmath96 are defined as the probability of @xmath96 being true over the probability of @xmath96 being false ( jaynes 1998 ) @xmath97 when @xmath98 , there is not enough information to choose between both hypothesis .
a galaxy is considered to have @xmath99 if @xmath100 , where @xmath101 is a certain decision threshold .
there are no fixed rules to choose the value of @xmath101 , and the most appropriate value depends on the task at hand ; for instance , to be really sure that no foreground galaxy has sneaked into the background sample , @xmath101 would have to be high , but if the main goal is selecting all the background galaxies and one does not mind including some foreground ones , then @xmath101 would be lower , etc .
basically this is a problem concerning decision theory . in the same way
, the cluster galaxies can be selected by choosing a redshift threshold @xmath102 which defines whether a galaxy belongs to the cluster .
the corresponding hypothesis would be @xmath103 .
@xmath104 and @xmath105 similarly , the odds of @xmath106 are defined as @xmath107 in those cases where the prior information is vague and does not allow to choose a definite expression prior probability , bayesian inference offers the possibility of `` calibrating '' the prior , if needed using the very data sample under consideration . let s suppose that the distribution @xmath108 is parametrized using @xmath109 continuous parameters @xmath110 .
they may be the coefficients of a polynomial fit , a wavelet expansion , etc . in that case , including @xmath110 in eq .
( [ bas ] ) , the probability can be written as @xmath111 where @xmath112 is the prior probability of @xmath110 , and @xmath113 is the prior probability of @xmath114 and @xmath15 as a function of the parameters @xmath110 .
the latter have not been included in the likelihood expression since @xmath16 is totally determined once the values of @xmath14 and @xmath24 are known .
now let s suppose that the galaxy belongs to a sample containing @xmath115 galaxies .
each @xmath116th galaxy has a ` base ' magnitude @xmath117 and colors @xmath118 .
the sets @xmath119 and @xmath120 contain respectively the colors and magnitudes of all the galaxies in the sample .
then , the probability of the @xmath121th galaxy having redshift @xmath122 given the full sample data @xmath123 and @xmath124 can be written as @xmath125 the sets @xmath126 and @xmath127 , @xmath128 are identical to @xmath123 and @xmath124 except for the exclusion of the data @xmath129 and @xmath130 . applying bayes theorem , the product rule and simplifying @xmath131 where as before it has been considered that the likelihood of @xmath129 only depends on @xmath132 and that the probability of @xmath122 and @xmath24 only depend on @xmath133 and @xmath134 through @xmath110 . the expression to which we arrived is very similar to eq .
( [ 20 ] ) only that now the shape of the prior is estimated from the data @xmath135 .
this means that even if one starts with a very sketchy idea about the shape of the prior , the very galaxy sample under study can be used to determine the value of the parameters @xmath110 , and thus to provide a more accurate estimate of the individual galaxy characteristics . assuming that the data @xmath133 ( as well as @xmath124 ) are independent among themselves @xmath136 where @xmath137 if the number of galaxies in our sample is large enough , it can be reasonably assumed that the prior probability @xmath138 will not change appreciably with the inclusion of the data @xmath139 belonging to a single galaxy . in that case , a time - saving approximation is to use as a prior the probability @xmath140 , calculated using the whole data set , instead of finding @xmath138 for each galaxy . in addition
, it should be noted that @xmath140 represents the bayesian estimate of the parameters which define the shape of the redshift distribution ( see fig .
[ nz ] ) . in some cases spectroscopical redshifts
@xmath141 are available for a fraction of the galaxy sample .
it is straightforward to include them in the prior calibration procedure described above , using a delta function likelihood weighted by the probability of the galaxy belonging to a morphological type , as it is done to determine the priors in sec [ test ] .
this gives the spectroscopical subsample a ( deserved ) larger weight in the determination of the redshift and morphological priors in comparison with the rest of the galaxies , at least within a certain color and magnitude region , but , unlike what happens with the training set method , the information contained in the rest of the sample is not thrown away .
if nevertheless one wants to follow the training set approach and use only the spectroscopic sample , it is easy to develop a bayesian variant of this method .
as before , the goal is to find an expression of the sort @xmath142 , which would give us the redshift probability for a galaxy given its colors and magnitude .
if the color / magnitude / redshift multidimensional surface @xmath31 were infinitely thin , the probability would just be @xmath143 , where @xmath144 is a delta - function .
but in the real world there is always some scatter around the surface defined by @xmath145 ( even without taking into account the color / redshift degeneracies ) , and it is therefore more appropriate to describe @xmath142 as e.g. a gaussian of width @xmath146 centered on each point of the surface @xmath145 .
let s assume that all the parameters which define the shape of this relationship , together with @xmath146 are included in the set @xmath147 . using the prior calibration method introduced above
, the probability distribution for these parameters @xmath148 can be determined from the training set @xmath149 .
@xmath150 the expression for the redshift probability of a galaxy with colors @xmath16 and @xmath15 would then be @xmath151 the redshift probability obtained from eq .
( [ 25 ] ) is compatible with the one obtained in eq .
( [ bas ] ) using the sed fitting procedure .
therefore it is possible to combine them in a same expression . as an approximation ,
let s suppose that both of them are given equal weights , then @xmath152 in fact , due to the above described redundancy between the sed
fitting method and the training set method ( sec .
[ sed ] ) , it would be more appropriate to combine both probabilities using weights which would take these redundancies into account in a consistent way , roughly using eq.([25 ] ) at brighter magnitudes , where the galaxies are well studied spectroscopically and leaving eq.([bas ] ) for fainter magnitudes .
the exploration of this combined , training set / sed - fitting approach will be left for a future paper , and in the practical tests performed below the followed procedure uses the sed
fitting likelihood .
the hubble deep field ( hdf ; @xcite ) has become _ the _ benchmark in the development of photometric redshift techniques . in this section bpz
will be applied to the hdf and its performance contrasted with the results obtained with the standard ` frequentist ' ( in the bayesian terminology ) method , the procedure usually applied to the hdf ( @xcite,@xcite , @xcite , etc . ) .
the photometry used for the hdf is that of @xcite , which , in addition to magnitudes in the four hdf filters includes jhk magnitudes from the observations of @xcite .
@xmath153 is chosen as the base magnitude .
the colors are defined as described in sec .
[ ml ] .
the template library was selected after several tests with the hdf subsample which has spectroscopic redshifts ( 108 galaxies ) , spanning the range @xmath2 .
the set of spectra which worked best is similar to that used by @xcite .
it contains four @xcite templates ( e / s0 , sbc , scd , irr ) , that is the same spectral types used by @xcite , plus the spectra of 2 starbursting galaxies from @xcite(@xcite used two very blue seds from gissel ) .
all the spectra were extended to the uv using a linear extrapolation and a cutoff at @xmath154 , and to the near ir using gissel synthetic templates .
the spectra are corrected for intergalactic absorption following @xcite .
it could seem in principle that a synthetic template set which takes ( at least tentatively ) into account galaxy evolution is more appropriate than ` frozen ' template library obtained at low redshift and then extrapolated to very high redshifts .
however , as @xcite has convincingly shown , the extended cww set offers much better results than the gissel synthetic models @xcite .
i have also tried to use the rvf set of spectra and the agreement with the spectroscopic redshifts is considerably worse than using the empirical template set . and if the synthetic models do not work well within the magnitude range corresponding to the hdf spectroscopic sample is relatively bright , there is little reason to suppose that their performance will improve at fainter magnitudes .
however , even working with empirical templates , it is important to be sure that the template library is complete enough .
[ comparison ] illustrates the effects of template incompleteness in the redshift estimation .
the left plot displays the results obtained using ml ( sec [ ml ] ) redshift estimation using only the four cww templates ( this plot is very similar to the @xmath155 diagram shown in @xcite , which confirms the validity of the expression for the likelihood introduced in sec [ ml ] ) . on the right , the results obtained also using ml ( no bpz yet ) but including two more templates , sb2 and sb3 from @xcite .
it can be seen that the new templates almost do not affect the low redshift range , but the changes at @xmath156 are quite dramatic , the ` sagging ' of the cww only diagram disappears and the general scatter of the diagram decreases by @xmath157 .
this shows how important it is to include enough galaxy types in the template library .
no matter how sophisticated the statistical treatment is , it will do little to improve the results obtained with a deficient template set .
the first step in the application of bpz is choosing the shape of the priors . due to the depth of the hdf
there is little previous information about the redshift priors , so this is a good example in which the prior calibration procedure described in sec[bpz ] has to be applied .
it will be assumed that the early types ( e / s0 ) and spirals ( sbc , scd ) have a spectral type prior ( eq . [ pri ] ) of the form @xmath158 with @xmath159 for early types and @xmath160 for spirals .
the irregulars ( the remaining three templates ; @xmath161 ) complete the galaxy mix .
the fraction of early types at @xmath162 is assumed to be @xmath163 and that of spirals @xmath164 .
the parameters @xmath165 and @xmath166 are left as free .
based on the result from redshift surveys the following shape for the redshift prior has been chosen : @xmath167^{\alpha_t } \ } \label{par1}\ ] ] where @xmath168 and @xmath169 , @xmath170 and @xmath171 are considered free parameters . in total , 11 parameters have to be determined using the calibration procedure . for those objects with spectroscopic redshifts ,
a ` delta - function ' located at the spectroscopic redshift of the galaxy has been used instead of the likelihood @xmath77 .
table 1 shows the values of the ` best ' values of the parameters in eq .
( [ par1],[par2 ] ) found by maximizing the probability in eq .
( [ 25 ] ) using the subroutine _ amoeba _ ( @xcite ) .
the errors roughly indicate the parameter range which encloses @xmath172 of the probability .
the values of the parameters in eq .
( [ par ] ) are @xmath173 and @xmath174 .
the prior in redshift @xmath69 can obviously be found by summing over the ` nuisance ' parameter ( jaynes 1998 ) , in this case @xmath24 : @xmath175 fig .
[ priors ] plots this prior for different magnitudes .
with the priors thus found , one can proceed with the redshift estimation using eq .
( [ 20 ] ) . here the multiplication by the probability distribution @xmath112 and the integration over @xmath176 will be skipped . as it can be seen from table 1 ,
the uncertainties in the parameters are rather small and it is obvious that the results would not change appreciably , so the additional computational effort of performing a 11-dimensional integral is not justified .
there are several options to convert the continuous probability @xmath142 to a point estimate of the ` best ' redshift @xmath177 . here
the ` mode ' of the final probability is chosen , although taking the ` median ' value of @xmath14 , corresponding to @xmath178 of the cumulative probability , or even the ` average ' @xmath179 is also valid .
it was mentioned in sec [ bpz ] that bayesian probability offers a way to characterize the accuracy of the redshift estimation using the odds or a similar indicator , for instance by analogy with the gaussian distribution a ` @xmath180 ' error may be defined using a interval with contains @xmath172 of the integral of @xmath142 around @xmath177 , etc .
here it has been chosen as an indicator of the redshift reliability the quantity @xmath181 , the probability of @xmath182 , where @xmath14 is the galaxy redshift . in this way ,
when the value of @xmath181 is low , we are warned that the redshift prediction is unreliable . as it will be shown below
, @xmath181 is extremely efficient in picking out galaxies with ` catastrophic errors ' in their redshifts .
the photometric redshifts resulting from applying bpz to the spectroscopic sample are plotted in fig .
galaxies with a probability @xmath183 ( there are three of them ) have been discarded , where @xmath102 is chosen to be @xmath184 , to take into account that the uncertainty grows with the redshift of the galaxies .
it is evident from fig .
[ hdf ] that the agreement is very good at all redshifts .
the residuals @xmath185 have @xmath186 .
if @xmath187 is divided by a factor @xmath188 , as suggested in @xcite , the rms of the quantity @xmath189 is only @xmath190 .
there are no appreciable systematic effects in the residuals .
one of the three objects discarded because of their having @xmath183 is the only clear outlier in our ml estimation , with @xmath191 and @xmath192 ( see fig .
[ comparison]b ) , evidence of the usefulness of @xmath181 to generate a reliable sample . from the comparison of fig .
[ comparison]b with fig .
[ hdf ] , it may seem that , apart from the exclusion of the outlier , there is not much profit in applying bpz with respect to ml .
this is not surprising in the particular case of the hdf spectroscopic sample , which is formed mostly by galaxies either very bright or occupying privileged regions in the color space .
the corresponding likelihood peaks are thus rather sharp , and little affected by smooth prior probabilities .
to illustrate the effectiveness of bpz under worse than ideal conditions , the photometric redshifts for the spectroscopic sample are estimated again using ml and bpz but restricting the color information to the ubvi hst filters .
the results are plotted in fig .
the ml redshift diagram displays 5 ` catastrophic errors ' ( @xmath193 ) .
note that these are the same kind of errors pointed out by @xcite in the first hdf photometric redshifts estimations .
bpz with a @xmath194 threshold ( which eliminates a total of 7 galaxies ) totally eliminates those outliers .
this is a clear example of the capabilities of bpz ( combined with an adequate template set ) to obtain reliable photometric redshift estimates .
note that even using near
ir colors , the ml estimates shown in fig .
[ comparison ] presented outliers .
this shows that applying bpz to uv only data may yield results more reliable than those obtained with ml including near - ir information ! although of course no more accurate ; the scatter of fig .
[ comparison]b , once the outliers are removed is @xmath195 , whereas fig .
[ 4f]b has a scatter of @xmath196 , which incidentally is approximately the scatter of fig .
[ comparison]a . another obvious way of testing the efficiency of bpz is with a simulated sample .
the latter can be generated using the procedure described in @xcite .
each galaxy in the hdf is assigned a redshift and type using ml ( this is done deliberately to avoid biasing the test towards bpz ) and then a mock catalog is created containing the colors corresponding to the best fitting redshifts and templates . to represent the photometric errors present in observations , a random photometric noise of the same amplitude as the photometric error
is added to each object .
[ 90]b shows the ml estimated redshifts for the mock catalog ( @xmath197 ) against the ` true ' redshifts ; although in general the agreement is not bad ( as could be expected ) there are a large number of outliers ( @xmath198 ) , whose positions illustrate the main source of color / redshift degeneracies : high @xmath14 galaxies which are erroneously assigned @xmath29 redshifts and vice versa .
this shortcoming of the ml method is analyzed in detail in @xcite .
in contrast , fig .
[ 90]a shows the results of applying bpz with a threshold of @xmath199 .
this eliminates @xmath157 of the initial sample ( almost half of which have catastrophically wrong redshifts ) , but the number of outliers is reduced to a remarkable @xmath200 .
is it possible to define some ` reliability estimator ' , similar to @xmath181 within the ml framework ?
the obvious choice seems to be @xmath28 .
[ odds]b plots the value of @xmath28 vs. the ml redshift error for the mock catalog .
it is clear that @xmath28 is almost useless to pick out the outliers .
the dashed line marks the upper @xmath201 quartile in @xmath28 ; most of the outliers are below it , at smaller @xmath28 values . in stark contrast ,
[ odds]a plots the errors in the bpz redshifts _ vs. _ the values of @xmath181 .
the lower @xmath201 quartile , under the dashed line , contains practically all the outliers . by setting an appropriate threshold one
can virtually eliminate the ` catastrophic errors ' .
[ oddsmz ] shows the numbers of galaxies above a given @xmath181 threshold in the hdf as a function of magnitude and redshifts .
it shows how risky it is to estimate photometric redshifts using ml for faint , @xmath202 objects ; the fraction of objects with possible catastrophic errors grows steadily with magnitude .
there is one caveat regarding the use of @xmath181 or similar quantities as a reliability estimator .
they provide a safety check against the color / redshift degeneracies , since basically they tell us if there are other probability peaks comparable to the highest one , but they can not protect us from template incompleteness .
if the template library does not contain any spectra similar to the one corresponding to the galaxy , there is no indicator able to warn us about the unreliability of the prediction .
because of this , no matter how sophisticated the statistical methods become , it is fundamental to have a good template set , which contains even if only approximately all the possible galaxy types present in the sample .
finally , fig .
[ nz ] shows the redshift distributions for the hdf galaxies with @xmath203 .
no objects have been removed on the basis of @xmath181 , so the values of the histogram bins should be taken with care .
the overplotted continous curves are the distributions used as priors and which simultaneously are the bayesian fits to the final redshift distributions .
the results obtained from the hdf will be analyzed in more detail , using a revised photometry , in a forthcoming paper .
as we argue above , the use of bpz for photometric redshift estimation offers obvious advantages over standard ml techniques .
however , quite often obtaining photometric redshifts is not an end in itself , but an intermediate step towards measuring other quantities , like the evolution of the star formation rate ( @xcite ) , the galaxy galaxy correlation function ( @xcite,@xcite ) , galaxy or cluster mass distributions ( @xcite ) , etc .
the usual procedure consists in obtaining photometric redshifts for all the galaxies in the sample , using ml or the training set method , and then work with them as if these estimates were accurate , reliable spectroscopic redshifts .
the results of the previous sections alert us to the dangers inherent in that approach , as it hardly takes into account the uncertainties involved in photometric redshift estimation .
in contrast , within the bayesian framework there is no need to work with the discrete , point
like ` best ' redshift estimates .
the whole redshift probability distribution can be taken into account , so that the uncertainties in the redshift estimation are accounted for in the final result . to illustrate this point ,
let s outline how bpz can be applied to several problems which use photometric redshift estimation .
if , instead of working with a discrete set of templates , one uses a spectral library whose templates depend of parameters as the metallicity , the star - formation history , initial mass function , etc . , represented by @xmath7 in sec [ bpz ] , it is obvious from equation ( [ cont ] ) that the same technique used to estimate the redshift can be applied to estimate any of the parameters @xmath7 which characterize the galaxy spectrum .
for instance , let s suppose that one want to estimate the parameter @xmath204 . then defining @xmath205
, we have @xmath206 that is , the likelihoods @xmath207 and the weights @xmath208 are the same ones used for the redshift estimation ( eq . [ cont ] ) , only that now the integration is performed over the variables @xmath14 and @xmath209 instead of @xmath7 . in this way ,
depending on the template library which is being used , one can estimate galaxy characteristics as the metallicity , dust content , etc .
an important advantage of this method over ml is that the estimates of the parameter @xmath204 automatically include the uncertainty of the redshift estimation , which is reflected in the final value of @xmath210 . besides , by integrating the probability over all the parameters @xmath209 , one precisely includes the uncertainties caused by possible parameter degeneracies in the final result for @xmath204 .
it should also be noted that as many of the results obtained in this paper , this method can be almost straightforwardly applied to spectroscopical observations ; one has only to modify the likelihood expression which compares the observed fluxes with the spectral template .
the rest of the formalism remains practically identical .
one frequent application of photometric redshift techniques is the study of galaxy cluster fields .
the goals may be the selection of cluster galaxies to characterize their properties , especially at high redshifts , or the identification of distant , background galaxies to be used in gravitational lensing analysis ( @xcite ) .
bpz offers an effective way of dealing with such problems . to simplify the problem , the effects of gravitational lensing on the background galaxies ( magnification , number counts depletion , etc
. ) will be neglected ( see however the next subsection ) .
let s suppose that we already have an estimate of the projected surface density of cluster galaxies ( which can roughly be obtained without any photometric redshifts , just from the number counts surface distribution ) @xmath211 , where @xmath212 is the position with respect to the cluster center .
the surface density of ` field ' , non cluster galaxies is represented by @xmath213 . for each galaxy in the sample
we know its magnitude and colors @xmath214 and also its position @xmath212 , which is now a relevant parameter in the redshift estimation .
following eq .
( [ bas ] ) we can write @xmath215 a dependence on the magnitude ( e.g. for the early types cluster sequence ) could easily be included in the likelihood @xmath77 if needed .
the prior can be divided into the sum of two different terms : @xmath216 where @xmath217 represents the prior probability of the galaxy belonging to the cluster , whereas @xmath218 corresponds to the prior probability of the galaxy belonging to the general field population .
the expression for @xmath217 can be written as @xmath219 the probability @xmath220 corresponds to the expected galaxy mix fraction in the cluster , which in general will depend on the magnitude and will be different from that of field galaxies .
the function @xmath221 is the redshift profile of the cluster ; a good approximation could be a gaussian with a width corresponding to the cluster velocity dispersion .
the second prior takes the form @xmath222 which uses the priors for the general field galaxy population ( sec [ test ] ) . finally , the hypothesis that the galaxy belongs to the cluster or not can be decided about with the help of a properly defined @xmath181 , or with the odds introduced in sec [ bpz ] .
we have assumed above that the cluster redshift and its galaxy surface density distribution are known . however , in some cases , there is a reasonable suspicion about the presence of a cluster at a certain redshift , but not total certainty , and our goal is to confirm its existence .
an example using ml photometric redshift estimation is shown in @xcite .
an extreme case with minimal prior information occur in optical cluster surveys , galaxy catalogs covering large areas of the sky are searched for clusters . in those cases there are no previous guesses about the position or redshift of the cluster , and a ` blind ' , automatized search algorithm has to be used ( @xcite ) .
the prior expression used in the previous subsection offers a way to build such a searching method . instead of assuming that the cluster redshift and its surface distribution are known , the redshift can be left as a free parameter @xmath223 and the expression characterizing the cluster galaxy surface density distribution @xmath211 can be parametrized using the quantities @xmath224 .
for simplicity , let s suppose that @xmath225 where @xmath226 is the cluster ` amplitude ' , @xmath227 is the number counts distribution expected for the cluster ( which in general will depend on the redshift @xmath223 ) and @xmath228 represents the cluster profile , centered on @xmath229 and with a scale width of @xmath230 .
this expression , except for the dependence on the redshift is very similar to that used by @xcite to define their likelihood .
then for a multicolor galaxy sample with data @xmath124 , @xmath231 and @xmath232 , the probability @xmath233 can be developed analogously to how it was done in sec .
[ bpz ] . the probability assigned to the existence of a cluster at a certain redshift and position may be simply defined as @xmath234 .
it seems that the most obvious application of bpz to cluster lensing analysis is the selection of background galaxies with the technique described in the previous subsection in order to apply the standard cluster mass reconstruction techniques ( @xcite , @xcite , @xcite , @xcite ) .
however , using bayesian probability it is possible to develop an unified approach which simultaneously considers the lensing and photometric information in an optimal way . in a simplified fashion
, the problem of determining the mass distribution of a galaxy cluster from observables can be stated as finding the probability @xmath235 where @xmath236 represent the parameters which describe the cluster mass distribution ; their number may range from a few , if the cluster is described with a simplified analytical model or as many as wanted if the mass distribution is characterized by e.g. fourier coefficients ( @xcite ) .
@xmath237 represents the cosmological parameters , which sensitively affect the lensing strength .
the parameter set @xmath238 represents the properties of the background galaxy population which affect the lensing , as its redshift distribution , number counts slope , etc .
and it is assumed to be known previously .
the data @xmath239 correspond to the galaxy ellipticities , @xmath240 to their angular positions . as above , @xmath241 correspond to their colors and magnitudes . for simplicity , it will be assumed that the cluster and foreground galaxies have been already removed and we are dealing only with the background galaxy population .
analogous to eq .
( [ 23 ] ) , we can develop eq .
( [ 38 ] ) as @xmath242 where the last factor may be written as @xmath243 the meaning of the three factors on the right side of the equation is the following : @xmath244 represents the likelihood of measuring a certain ellipticity @xmath245 in a galaxy given its redshift , position , etc .
the second factor @xmath246 corresponds to the so called `` broadhurst effect '' , the number counts depletion of background galaxies caused by the cluster magnification @xmath247 ( broadhurst 1995 , bentez & broadhurst 1998 ) .
the last factor , @xmath248 is the redshift probability , but including a correction which takes into account that the observed magnitude of a galaxy @xmath15 is affected by the magnification @xmath249 .
it is clear that the simplified method outlined here is not the only way of applying bayesian probability to cluster mass reconstruction .
my purpose here is to show that this can be done considering the photometric redshifts in a integrated way with the rest of the information .
as it has been shown in section ( [ bpz ] ) , bpz can be used to estimate the parameters characterizing the joint magnitude redshift
morphological type galaxy distribution . for small fields
, this distribution may be dominated by local perturbations , and the redshift distribution may be ` spiky ' , as it is observed in redshift surveys of small fields .
however , if one were to average over a large number of fields , the resulting distribution would contain important information about galaxy evolution and the fundamental cosmological parameters .
@xcite included galaxy counts as one of the four fundamental tests of observational cosmology , although noting that the number - redshift distribution is in fact more sensitive to the value of @xmath250 .
as @xcite also notes , the color distribution of the galaxies in a survey hold also much more information about the process of galaxy evolution that the raw number counts . however , quite often the only method of analyzing multicolor observations is just comparing them with the number counts model predictions , or at most , with color distributions .
there are several attempts at using photometric redshifts to study global galaxy evolution parameters ( e.g. @xcite , @xcite ) , but so far there is not an integrated statistical method which would simultaneously considers all the information , magnitudes and colors , contained in the data , and set it against the model predictions .
it is then clear that eq .
( [ 23 ] ) can be used to estimate these parameters from large enough samples of multicolor data . if it is assumed that all the galaxies belong to a few morphological types , the joint redshift - magnitude-`type ' distribution can be written as @xmath251 where @xmath252 is the comoving volume as a function of redshift , which depends on the cosmological parameters @xmath253 and @xmath254 , and @xmath255 is the schecter luminosity function for each morphological type @xmath24 , where the absolute magnitude @xmath256 has been substituted by the apparent magnitude @xmath15 ( a transformation which depends on the redshifts , cosmological parameters and morphological type ) .
schecter s function also depend on @xmath257 , @xmath258 and @xmath259 , and on the evolutionary parameters @xmath260 , such as the merging rate , the luminosity evolution , etc .
therefore , the prior probability of @xmath14,@xmath7 and @xmath261 depends on the parameters @xmath262 , @xmath263 and @xmath260 . as an example , let s suppose that one wants to estimate @xmath260 , independently of the rest of the parameters , given the data @xmath264 .
then @xmath265 @xmath266 the prior @xmath267 can be derived from @xmath268 in eq .
( [ 5 ] ) .
the prior @xmath269 allows to include the uncertainties derived from previous observations or theory in the values of these parameters , even when they are strongly correlated among themselves , as in the case of the schecter function parameters @xmath270 .
the narrower the prior @xmath269 is , the less ` diluted ' the probability of @xmath260 and the more accurate the estimation .
despite the remarkable progress of faint galaxy spectroscopical surveys , photometric redshift techniques will become increasingly important in the future .
the most frequent approaches , the template fitting and empirical training set methods , present several problems related which hinder their practical application .
here it is shown that by consistently applying bayesian probability to photometric redshift estimation , most of those problems are efficiently solved . the use of prior probabilities and bayesian marginalization allows the inclusion of valuable information as the shape of the redshift distributions or the galaxy type fractions , which is usually ignored by other methods .
it is possible to characterize the accuracy of the redshift estimation in a way with no equivalents in other statistical approaches ; this property allows to select galaxy samples for which the redshift estimation is extremely reliable . in those cases
when the _ a priori _ information is insufficient , it is shown how to ` calibrate ' the prior distributions , using even the data under consideration . in this way it is possible to determine the properties of individual galaxies more accurately and simultaneously
estimate their statistical properties in an optimal fashion .
the photometric redshifts obtained for the hubble deep field using optical and near - ir photometry show an excellent agreement with the @xmath0 spectroscopic redshifts published up to date in the interval @xmath271 , yielding a rms error @xmath272 and no outliers .
note that these results , obtained with an empirical set of templates , have not been reached by minimizing the difference between spectroscopic and photometric redshifts ( as for empirical training set techniques , which may lead to an overestimation of their precision ) and thus offer a reasonable estimate of the predictive capabilities of bpz .
the reliability of the method is also tested by estimating redshifts in the hdf but restricting the color information to the ubvi filters ; the results are shown to be more reliable than those obtained with the existing techniques even including the near - ir information .
the bayesian formalism developed here can be generalized to deal with a wide range of problems which make use of photometric redshifts .
several applications are outlined , e.g. the estimation of individual galaxy characteristics as the metallicity , dust content , etc . , or the study of galaxy evolution and the cosmological parameters from large multicolor surveys .
finally , using bayesian probability it is possible to develop an integrated statistical method for cluster mass reconstruction which simultaneously considers the information provided by gravitational lensing and photometric redshift estimation .
i would like to thank tom broadhurst and rychard bouwens for careful reading the manuscript and making valuable comments .
thanks also to alberto fernndez - soto and collaborators for kindly providing me with the hdf photometry and filter transmissions , and to brenda frye for help with the intergalactic absorption correction .
the author acknowledges a basque government postdoctoral fellowship .
jaynes , `` probability theory : the logic of science '' , to be published in cambridge university press . a preliminary version can be obtained from thomas loredo s web page at _ http://astrosun.tn.cornell.edu / staff / loredo / bayes/_. | photometric redshift estimation is becoming an increasingly important technique , although the currently existing methods present several shortcomings which hinder their application . here it is shown that most of those drawbacks are efficiently eliminated when bayesian probability is consistently applied to this problem . the use of prior probabilities and bayesian marginalization allows the inclusion of valuable information , e.g. the redshift distributions or the galaxy type mix , which is often ignored by other methods .
it is possible to quantify the accuracy of the redshift estimation in a way with no equivalents in other statistical approaches ; this property permits the selection of galaxy samples for which the redshift estimation is extremely reliable . in those cases
when the _ a priori _ information is insufficient , it is shown how to ` calibrate ' the prior distributions , using even the data under consideration .
there is an excellent agreement between the @xmath0 hdf spectroscopic redshifts and the predictions of the method , with a rms error @xmath1 up to @xmath2 and no systematic biases nor outliers .
note that these results have not been reached by minimizing the difference between spectroscopic and photometric redshifts ( as is the case with empirical training set techniques ) , which may lead to an overestimation of the accuracy .
the reliability of the method is further tested by restricting the color information to the ubvi filters .
the results thus obtained are shown to be more reliable than those of standard techniques even when the latter include near - ir colors .
the bayesian formalism developed here can be generalized to deal with a wide range of problems which make use of photometric redshifts .
several applications are outlined , e.g. the estimation of individual galaxy characteristics as the metallicity , dust content , etc . , or the study of galaxy evolution and the cosmological parameters from large multicolor surveys .
finally , using bayesian probability it is possible to develop an integrated statistical method for cluster mass reconstruction which simultaneously considers the information provided by gravitational lensing and photometric redshift estimation . |
there are various phenomena in nature including complex , chaotic , turbulent , critical , fractal and anomalous transport phenomena having erratic trajectories with often non - differentiable characteristics . such ` anomalous ' phenomena as a rule can not be described by standard approaches involving integer order partial differential equations .
however , it has been shown that they often can be described by non - integer order , i.e. fractional differential equations @xcite .
there are many definitions for fractional derivatives and integrals ( riemann , liouville , caputo , grnwald - letnikow , marchaud , weyl , riesz , feller , and others ) , see e.g. @xcite and the references therein .
this diversity of definitions is due to the fact that fractional operators take different kernel representations in different function spaces which is a consequence of the nonlocal character of fractional kernels .
whereas fractional operators are well known in the continuous space and obtained as power law convolutional kernels , the fractional calculus on discrete networks and lattices is more involved and much less developed .
an approach to define fractional differential operators on lattices was suggested in the papers @xcite . in @xcite
was suggested the fractional calculus which includes fractional difference operators and fractional integro - differential operators on lattices where exact results have been presented in these works ( see also @xcite ) .
in contrast to that approach , the goal of the present work is to introduce _ fractional centered difference operators _ on the lattice appearing as a natural fractional generalization of born von karman s centered symmetric second order difference operator . in the same time all good properties of the classical born von karman lattice approach such as translational symmetry and elastic stability is conserved by the fractional lattice model to be developed in the present paper . in this way
the present approach opens the door towards a generalization of the classical lattice approaches @xcite . in the context of markovian processes on networks ,
the concept of ` fractional diffusion on undirected networks ' generalizing the ` normal random walk ' was recently introduced by riascos and mateos @xcite .
such random motions on lattices are defined by diffusion equations where instead of discrete laplacian matrices defined by second order difference operators their fractional generalizations come into play . in these works
it has been demonstrated that fractional generalizations of lattice models have a huge interdisciplinary potential as they are able to describe phenomena which account for nonlocal interactions including the emergence of lvy flights on lattices @xcite .
fractional generalizations second order difference operators introduce nonlocal long - range features .
in contrast to fractional orders , the standard finite differences of integer orders describe local nearest neighbor interactions .
the nonlocality increases with the integer order of the power .
the difference operators of integer orders at the lattice level correspond to spatial forms of non - local lattice sums and the relations of differences and differential operators defined by infinite series are analyzed in recent articles @xcite , and further a fractional generalization of lattice derivative approach was introduced in @xcite . beside the applications on diffusion problems on the lattice , the importance of fractional lattice models appears also for a description of fractional lattice vibrational phenomena , a generalization of crystal lattice dynamics .
some initial steps towards such a fractional generalization generalization of nonlinear classical lattice dynamics has been introduced by laskin and zaslavsky @xcite and tarasov and zaslavsky @xcite . in a lattice dynamics model which defines by hamilton s variational principle the ` laplacian matrix ' which contains all constitutive information of the harmonic interparticle interactions , it is therefore desirable to develop a ` fractional generalization ' of the existing lattice dynamics approach . to this end in the present paper
we utilize the methodology of characteristic matrix function which was introduced recently @xcite .
the interdisciplinary fractional lattice approach developed in the present paper allows to describe all types of physical fields in the harmonic approximation defined on lattices having a continuum limit governed by a non - classical , ` anomalous ' power - law nonlocality in the form of riesz fractional derivative kernels .
physical applications include especially problems of fractionally generalized lattice dynamics : lattice vibrations , elastic wave propagation , nonlocal elasticity , nonlocal electrodynamics , fractional quantum mechanics on lattices and networks , and last but not least above mentioned anomalous transport and diffusion phenomena with ` fractional random walk models ' and emergence of lvy flights . the entire set of problems of classical lattice dynamics in the harmonic approximation @xcite can be fractionally generalized by the present approach .
this generalization contains only one characteristic parameter , namely the power - law index @xmath3 . newly emerging phenomena of this ` fractionalization '
are to be expected .
the classical ( born von karman ) theories are contained in the approach at the value @xmath2 . since the present fractional lattice approach is well defined by hamilton s variational principle
, it provides a physically admissible nonlocal generalization of any classical harmonic lattice model .
the present paper is organized as follows . in the first part of the paper we deduce from fractional harmonic lattice potentials " on the cyclically closed linear chain a discrete fractional laplacian matrix .
we do so by applying our recent approach to generate nonlocal lattice models by matrix functions where the generator operator is the discrete centered born von karman laplacian @xcite .
first we obtain the discrete fractional laplacian in explicit form for the infinite chain for particle numbers @xmath4 , being in accordance with the fractional centered difference models of ortiguiera @xcite and zoia et al .
@xcite . utilizing the discrete infinite chain fractional laplacian matrix we construct an explicit representation for the _ fractional laplacian matrix on the @xmath5-periodic finite 1d lattice _ where the particle number @xmath5 can be arbitrary not necessarily large .
then we analyse continuum limits of the discrete fractional model : the infinite space continuum limit of the fractional laplacian matrix yields the well known infinite space kernel of the standard fractional laplacian .
the periodic string continuum limit yields an explicit representation for the kernel of the fractional laplacian ( riesz fractional derivative ) which fulfills periodic boundary conditions and is defined on the finite @xmath6-periodic string .
in the second part of the paper we suggest an extension of the fractional approach on nd periodic and infinite lattices .
we deduce an integral representation for fractional laplacian on the infinite nd lattice and proof that as asymptotic representation the well known riesz fractional derivative of the nd infinite space is emerging .
more detailed derivations of some of the results of the present paper can be found in recent articles @xcite .
all these results are fully equivalent and can also be deduced by employing the more general approach of riascos and mateos for fractional diffusion problems on networks @xcite , and see also the references therein .
we consider first a periodic , cyclically closed linear chain ( 1d periodic lattice or ring ) with equidistant lattice points
@xmath7 consisting of @xmath5 identical particles with particle mass @xmath8 .
each mass point @xmath9 has equilibrium position at @xmath10 where @xmath6 denotes the length of the chain and @xmath11 the interparticle distance ( lattice constant ) .
further we impose periodicity ( cyclic closure of the chain ) . for convenience of our demonstration
we introduce the unitary shift operator @xmath12 defined by @xmath13 and its adjoint @xmath14 with @xmath15 .
we employ periodic boundary conditions ( cyclic closure of the chain ) @xmath16 ( @xmath17 ) and equivalently , cyclic index convention @xmath18 .
any elastic potential in the _ harmonic approximation _ defined on the 1d periodic lattice can be written in the representation @xcite @xmath19 where @xmath20 indicates the ( negative - semidefinite ) laplacian @xmath21-matrix , @xmath22 the identity matrix , and @xmath23 we refer to as the characteristic function : physically admissible , elastically stable and translational invariant positive elastic potentials require for the 1d periodic lattice ( cyclic ring ) that the characteristic function @xmath23 which is defined as a scalar function to have the following properties @xmath24 for @xmath25 ( elastic stability ) and @xmath26 ( translational invariance , zero elastic energy for uniform translations of the lattice ) . for the approach to be developed we propose the characteristic function to assume power law form @xmath27 which fulfills for @xmath28 and @xmath29 the above required good properties for the characteristic function .
@xmath30 denotes a dimensional constant of physical dimension @xmath31 .
note that @xmath32 is the central symmetric second difference operator which is defined by @xmath33 .
the matrix function @xmath34 is in general a self - adjoint ( symmetric ) positive semidefinite @xmath21-matrix function of the simple @xmath21 generator matrix @xmath35_{pq}=2\delta_{pq}-\delta_{p+1,q}-\delta_{p-1,q}$ ] .
it can be easily seen that @xmath34 has tplitz structure , i.e. its additional symmetry consists in the form @xmath36 , @xmath37 giving the fractional generalization of the born von karman centered difference operator .
the fractional elastic potential has then with ( [ compfo ] ) the representation @xmath38 with the matrix elements @xmath39_{|p - q|}$ ] of the _ fractional characteristic matrix function_. in full analogy to the negative semidefinite continuous laplacian ( second derivative operator ) we define here the fractional laplacian matrix as the negative semidefinite matrix defined through hamilton s variational principle . ]
@xmath40_{|p - q| } = -\mu f^{(\alpha)}_{|p - q| } .\ ] ] we emphasize again that the fractional laplacian matrix @xmath41 is self - adjoint having a symmetric tplitz structure matrix representation @xmath42 .
this property appears naturally as the shift operators @xmath12 and their adjoint @xmath43 are symmetrically contained .
this is true for all admissible @xmath28 and includes @xmath2 for the conventional central symmetric second difference operator . in this relation
we have introduced the positive - semidefinite fractional characteristic matrix @xmath44 which is in our convention up to the prefactor @xmath45 identical with the fractional laplacian matrix ( which we define as negative - semidefinite being the fractional analogue of @xmath46 . to determine the fractional laplacian matrix it is useful to consider the spectral representation @xmath47 where we take advantage of the @xmath5-periodicity of the chain , i.e. the eigenvectors of the fractional laplacian matrix are ortho - normal bloch - vectors having the components @xmath48 . for the infinite chain in the limit @xmath4 the matrix elements of the fractional laplacian matrix ( [ fractlap ] )
can be evaluated explicitly .
@xmath49 this expression can be obtained in explicit form @xcite @xmath50 where we introduced the generalized factorial function @xmath51 . in view of ( [ matrixelei ] )
we observe that for noninteger @xmath52 any matrix element @xmath53 is non - vanishing indicating the nonlocality of the harmonic fractional interparticle interaction ( [ fractlap ] ) .
for @xmath54 the matrix elements ( [ matrixelei ] ) take the values of the standard binomial coefficients .
( [ fractlattice])@xmath55 can be read as the fourier coefficients of the infinite fourier series @xmath56 representing the fractional dispersion relation .
putting @xmath57 in the dispersion relation , we can verify directly that the fractional laplacian matrix conserves translational symmetry which is expressed by @xmath58 this equation can also be read as @xmath59 , i.e. the fractional centered difference operator applied to a constant is vanishing .
this property appears as the fractional generalization of the same property of second order centered differences when @xmath2 .
we further observe in view of ( [ fouser ] ) the positive semi - definiteness of the fractional characteristic matrix @xmath44 where positiveness of ( [ fouser ] ) for @xmath60 indicates elastic stability of the chain .
the fractional dispersion relation ( [ fouser ] ) leads to the remarkable relation which holds _ only _ for complex numbers on the unit circle @xmath61 , namely @xmath62 this laurent series converges nowhere except on the unit circle @xmath63 .
for instance the zero eigenvalue @xmath64 which corresponds to translational invariance ( zero elastic energy for uniform translations ) is obtained by putting @xmath65 in ( [ fouserb ] ) . for integer
@xmath66 ( [ matrixelei ] ) takes the form of the standard binomial coefficients and the series ( [ fouser ] ) , ( [ fouserb ] ) then take the representations of standard binomial series of @xmath67 breaking at @xmath68 corresponding to zero values for the matrix elements for ( [ matrixelei ] ) for @xmath69 .
we further observe for noninteger @xmath70 the _ power law asymptotics _ for @xmath71 which can be obtained by utilizing stirling s asymptotic formula for the @xmath72-function @xcite @xmath73 the asymptotic power law ( scale free ) characteristics of the fractional laplacian matrix @xmath74 is the essential property which gives rise to many ` anomalous phenomena ' such as in ` fractional diffusion ' problems on networks such as the emergence of lvy flights @xcite ( and references therein ) .
the fractional continuum limit kernels are discussed in the subsequent section .
the expressions ( [ fractlattice])-([asymp ] ) hold for the infinite 1d lattice corresponding to @xmath75 . as everything in nature
is limited we shall consider now the fractional laplacian matrix for a finite periodic lattice where the particle number @xmath5 is arbitrary and not necessarily large .
it is only a small step to construct the finite lattice laplacian matrix in terms of infinite lattice laplacian matrix .
we can perform this step by the following consideration : let @xmath76 the laplacian matrix of the infinite lattice , and @xmath77 the continuous dispersion relation of the infinite lattice matrix @xmath78 obeying the eigenvalue relation @xmath79 this relation holds identically in the entire principal interval @xmath80 and is @xmath81-periodic in the @xmath82-space .
let us now choose @xmath83 with @xmath84 being the bloch wave number of the _ finite _ periodic lattice of @xmath5 lattice points where @xmath5 is not necessarily large .
since the bloch wave numbers of the chain are discrete points within the interval @xmath85 , then relation ( [ eigval ] ) holds as well for these @xmath5 @xmath82-points , namely @xcite in ( [ eigval ] ) has been put to zero .
] @xmath86 in the second relation the @xmath5-periodicity of the finite lattice bloch eigenvector @xmath87 has been used .
the last relation can be read as the eigenvalue relation for the @xmath5-periodic lattice matrix of tplitz structure @xmath88 it follows that in the limiting case @xmath75 the finite lattice matrix ( [ identiperiodic ] ) recovers the infinite lattice matrix @xmath89 . from ( [ identiperiodic ] ) we read of for the fractional lattice laplacian of the finite periodic 1d lattice @xmath90 with @xmath91 we observe @xmath5-periodicity of ( [ finitecharmat ] ) and furthermore the necessary property that in the limit of infinite chain @xmath4 , ( [ finitecharmat ] ) recovers the infinite lattice expression of eq .
( [ matrixelei ] ) .
in this section we investigate the interlink between the lattice fractional approach introduced above and continuum fractional derivatives . to this end
we introduce the following hypotheses which are to be observed when performing continuum limits .
following @xcite we require in the continuum limit that extensive physical quantities , i.e. quantities which scale with the length of the 1d system , such as the total mass @xmath92 and the total elastic energy of the chain remain finite when its length @xmath6 is kept finite we require the mass per unit length and elastic energy per unit length to remain finite . ] , i.e. neither vanish nor diverge .
let @xmath93 be the length of the chain and @xmath11 the lattice constant ( distance between two neighbor atoms or lattice points ) .
we can define two kinds of continuum limits : ( i ) the _ periodic string continuum limit _ where the length of the chain @xmath93 is kept finite and @xmath94 ( i.e. @xmath95 ) .
( ii ) the _ infinite space continuum limit _ where @xmath94 , however , the length of the chain tends to infinity @xmath96 where @xmath97 . ] .
the kernels of the infinite space limit can be recovered from those of the periodic string limit by letting @xmath98 . from the finiteness of the total mass of the chain
, it follows that the particle mass @xmath99 scales as @xmath100 . then by employing expression ( [ valpha ] ) for the fractional elastic potential , the total continuum limit elastic energy @xmath101 can be defined by @xmath102 putting @xmath103 ( @xmath104 ) and accounting for @xmath105
we get @xmath106 the formal relation ( [ fraclap ] ) shows that the continuum limit kernels to be deduced in explicit forms have the interpretation of the _ fractional laplacian _ or also in the literature referred to as _ riesz fractional derivative_. to maintain finiteness of the elastic energy in the continuum limit @xmath94 the following scaling relations for the characteristic model constants , the mass @xmath8 and the frequency @xmath30 are required @xcite @xmath107 where @xmath108 denotes the mass density with dimension @xmath109 and @xmath110 denotes a positive dimensional constant of dimension @xmath111 , where the new constants @xmath112 are independent of @xmath11 .
note that the dimensional constant @xmath110 is only defined up to a non - dimensional positive scaling factor as its absolute value does not matter due to the scale - freeness of the power law .
we obtain then as continuum limit of the elastic energy by taking into account @xmath113 and @xmath114 , @xmath115 , @xmath116 the continuum limit laplacian kernel @xmath117 can then formally be represented by the distributional kernel representation in the spirit of generalized functions @xcite
@xmath118 the last relation contains the distributional representation of the fractional laplacian and is obtained for the infinite space limit ( ii ) in explicit form as ( where @xmath119 indicates the real part of @xmath120 ) @xcite @xmath121 being defined ` under the integral ' in the distributional sense and yields for noninteger @xmath122 for @xmath123 the well known riesz fractional derivative kernel of the infinite space @xmath124 with a characteristic @xmath125 power law nonlocality reflecting the asymptotic power law behavior ( [ asymp ] ) of ( [ matrixelei ] ) for sufficiently large @xmath71 .
the continuum procedure of @xmath6-periodic string where @xmath6 is kept finite is then obtained as @xcite denotes the real part of a quantity @xmath120 ] @xmath127 this kernel can be conceived as the explicit representation of the fractional laplacian ( riesz fractional derivative ) on the @xmath6-periodic string .
the last relation is the distributional representation and is expressed by standard hurwitz @xmath128-functions denoted by @xmath129 .
the two variants of @xmath128- functions which occur in above relation are defined by @xmath130 we see for @xmath131 and @xmath123 that the series in ( [ perfrac ] ) are absolutely convergent as good as the power function integral @xmath132 . for integer powers
@xmath133 the distributional representations ( [ perfrac])@xmath134 take the ( distributional ) forms of the ( negative semi - definite ) 1d integer power laplacian operators , namely @xmath135 where @xmath136 and @xmath137 indicate the dirac s @xmath138-functions of the infinite and the @xmath6-periodic string , respectively .
we further observe in full correspondence to the discrete fractional laplacian matrix , the necessary property that in the limit of an infinite string @xmath139 ( [ perfrac ] ) recovers the expression of the standard 1d infinite space fractional laplacian kernel ( [ inffyla ] ) known from the literature ( see for a further discussion @xcite and references therein ) .
in this section we deduce the @xmath140 counterpart of the fractional laplacian matrix introduced above . with that approach the fundamentals of ` _ fractional lattice dynamics _ '
can be deduced as a generalization of conventional lattice dynamics . in this section
our goal is to generalize the above 1d lattice approach to cubic periodic lattices in @xmath1 dimensions of the physical space where the 1d lattice case is contained .
we assume the lattice contains @xmath141 lattice points , each covered by identical atoms with mass @xmath8 .
each mass point is characterized by @xmath142 ( @xmath143 ) and @xmath1 denotes the dimension of the physical space embedding the lattice . in order to define the lattice fractional laplacian matrix , it is sufficient to consider a _ scalar _ generalized displacement field @xmath144 ( one field degree of freedom ) associated to each mass point @xmath145 only .
the physical nature of this scalar field can be any scalar field , such as for instance a one degree of freedom displacement field , an electric potential or , in a stochastic context a probability density function ( pdf ) or in a fractional quantum mechanics context a schrdinger wave function .
this demonstrates the interdisciplinary character of the present fractional lattice approach .
the fractional laplacian matrix for general networks was only recently and to our knowledge for the first time introduced by riascos and mateos @xcite in the framework of fractional diffusion analysis on networks which include nd periodic lattices ( nd tori ) as special cases being subject of the present analysis . for cubic nd lattices
the fractional laplacian matrix can be written as @xcite @xmath146 where @xmath22 denotes the identity matrix , @xmath0 indicates the dimension of the physical space and @xmath147 indicates the connectivity , i.e. the number of next neighbors of a lattice point in the nd cubic lattice . in ( [ fraalphn ] )
we introduced the adjacency matrix @xmath148 which has for the cubic lattice with next neighbor connections the form @xmath149 where then @xmath150 and @xmath151 denote the next neighbor shift operators in the @xmath152-directions defined by @xmath153 and @xmath154 , i.e. @xmath150 shifts the field associated to lattice point @xmath155 to the field associated with the adjacent lattice point in the positive @xmath156-direction @xmath157 , and the inverse ( adjoint ) shift operator @xmath151 to the adjacent lattice point in the negative @xmath156-direction @xmath158 .
all matrices introduced in ( [ fraalphn ] ) and ( [ cubicadj ] ) are defined on the nd lattice being @xmath159 matrices ( @xmath160 ) . as in the case of 1d lattice
the so defined fractional laplacian matrix ( [ fraalphn ] ) describes for non - integer powers @xmath161 , @xmath162 nonlocal elastic interactions , whereas they are generated by the ` local ' next neighbor born von karman laplacian which is in our definition up to a negative dimension factor @xmath163 equal to @xmath164 .
we therefore refer to @xmath164 as ` generator matrix ' .
we emphasize that the sign convention of what we call ` ( fractional ) laplacian matrix ' varies in the literature ( e.g. by denoting the positive semidefinite matrix @xmath165 as ` fractional laplacian matrix ' , this convention is chosen , e.g. in @xcite ) .
we have chosen to refer to as ` fractional laplacian matrix ' the negative - semidefinite matrix @xmath166 to be in accordance with the negative definiteness of continuum limit fractional laplacian ( [ integerperiodic ] ) containing as a special case @xmath167 the negative semidefinite conventional laplacian @xmath168 ) . for a discussion of some general properties of the fractional laplacian ( [ fraalphn ] ) well defined on general networks including @xmath140 lattices ,
we refer to @xcite . in the periodic and infinite lattice
the shift operators are unitary .
assuming @xmath169-periodicity in each direction @xmath156 , the fractional laplacian matrix is defined by the spectral properties of the @xmath164-matrix , namely by @xmath170_{(\vec{p}-\vec{q})}=\frac{1}{n } \sum_{\vec{\ell } } e^{i\vec{\kappa}_{\vec{\ell}}\cdot(\vec{p}-\vec{q})}\lambda_{\vec{\ell}}^{\frac{\alpha}{2 } } , \hspace{0.5 cm } \lambda_{\vec{\ell } } = \left(2n-2\sum_{j=1}^n\cos{(\kappa_{\ell_j})}\right ) , \,\alpha > 0,\ ] ] where we denoted @xmath171 and @xmath172 denotes the bloch wave vectors of the brillouin zone where their components can the values @xmath173 ( @xmath174 ) .
it can be seen that ( [ spectralfrac ] ) has tplitz structure depending only on @xmath175 ) .
for the infinite lattice when all @xmath176 in ( [ spectralfrac ] ) , the summation over the reciprocal lattice points assumes asymptotically the form of an integral @xmath177 , where the integration intervals @xmath178 $ ] can be chosen instead of @xmath179 $ ] for @xmath81-periodic functions @xmath180 .
0.5 cm -2 cm _ @xmath181 ( a - b ) show the dispersion surfaces @xmath182 for the 2d cubic lattice ( @xmath183 ) of ( [ spectralfrac ] ) for four values of @xmath3 , while ( c - d ) illustrate cross - sections of these dispersion sheets with the planes ( 0 1 0 ) and ( 1 1 0 ) , respectively .
_ for @xmath3 fixed , the circular frequency is given by @xmath184 .
the linear frequency spectra @xmath185 , for @xmath183 , are normalized by the maximum frequency @xmath186 obtained for a wave vector located in @xmath187 plane .
it will be noted that the sheets cut at dimensionless frequency @xmath188 and the dispersion relations of the classical next neighbor born von karman lattice are recovered ( indicated by @xmath189 for @xmath2 and @xmath190 , @xmath191 for @xmath2 and @xmath192 ) .
when the value of @xmath3 decreases , one observes in agreement with another work @xcite , namely a decrease of the maximum dimensionless frequency in end of the first brillouin zone .
the goal is now to deduce a more convenient integral representation of ( [ spectralfrac ] ) . to this end
we utilize the following observation : let in the following @xmath193 be a positive semidefinite of this matrix are non - negative . ] matrix and @xmath28 like ( [ fraalphn ] ) with the spectral representation @xmath194 where we have to put for the periodic nd lattice of ( [ spectralfrac ] ) the bloch - eigenvectors @xmath195
. then it will be useful to define the matrix dirac @xmath138-function by @xmath196 where @xmath197 is a scalar parameter and @xmath22 the identity matrix and @xmath198 the conventional scalar dirac @xmath138-function .
then with the matrix @xmath138-function defined in ( [ matrixdelta ] ) we can write @xmath199 and by utilizing @xmath200 together with the kernel @xmath201 of the 1d fractional laplacian ( riesz fractional derivative ) of order @xmath52 in its distributional form @xcite @xmath202 then we can write for the matrix power function ( [ matrixpower ] ) the representation @xmath203 where the exponential @xmath204 of the matrix @xmath205 can be determined more easily for the generator @xmath206 ( @xmath207 ) being the sum of the 1d generator matrices of the @xmath169-periodic 1d lattices and having therefore the eigenvalues @xmath208 and as a consequence having a cartesian product space spanned by the periodic bloch eigenvectors @xmath209 .
the matrix elements of the spectral representation of the exponential of @xmath164 can hence be written as @xmath210_{\vec{p}-\vec{q } } = \sum_{\vec{\ell } } \frac{e^{i{\vec \kappa}_{\vec{\ell}}\cdot ( \vec{p}-\vec{q})}}{n}e^{i\xi \lambda_{\vec{\ell } } } = \prod_{j=1}^n \sum_{\ell_j=1}^{n_j-1 } \frac{e^{i(p_j - q_j)\kappa_{\ell_j}}}{n_j}e^{i2k(1-\cos{\kappa_{\ell_j } } ) } .\ ] ] * infinite nd lattice * + in the limiting case of an infinite nd lattice when all @xmath176 we can write by using @xmath211 to arrive at @xmath212_{\vec{p}-\vec{q } } = [ e^{i\xi l_n}]_{|p_1-q_1|, .. ,|p_n - q_n| } = \prod_{j=1}^n \frac{1}{(2\pi)}\int_{-\pi}^{\pi } e^{i(p_j - q_j)\kappa}e^{i2\xi ( 1-\cos{\kappa})}{\rm d}\kappa .\ ] ] taking into account the definition of the modified bessel functions of the first kind @xmath213 where @xmath214 denotes non - negative integers @xcite , we can write the exponential matrix ( [ expolaplinfini ] ) in the form @xmath215_{|p_1-q_1|, .. ,|p_n - q_n| } = e^{i 2n\xi } \prod_{j=1}^ni_{|p_j - q_j|}(-2i\xi ) .\ ] ] applying now the matrix relation ( [ matrixpower ] ) and plugging in the exponential ( [ expobessel ] ) yields an integral representation of the ( negative semidefinite ) fractional laplacian matrix ( [ fraalphn ] ) in terms of a product of modified bessel functions of the first kind , namely @xmath216_{|p_1-q_1|, ..
,|p_n - q_n| } = -\mu \omega_{\alpha , n}^2[l_n^{\frac{\alpha}{2}}]_{|p_1-q_1|, ..
,|p_n - q_n| } \\ \\
\hspace{1 cm } \displaystyle = -\mu \omega_{\alpha , n}^2 \int_{-\infty}^{\infty } { \rm d}\xi \ , e^{i 2n\xi } { \cal d}_{\frac{\alpha}{2}}(\xi ) \prod_{j=1}^ni_{|p_j - q_j|}(-2i\xi ) , \end{array}\ ] ] with @xmath217 indicating the riesz fractional derivative kernel of ( [ distrifoufracder ] ) . * asymptotic behavior * + introducing the new vector valued integration variable @xmath218 ( @xmath219 ) we can write for the infinite lattice integral of ( [ spectralfrac ] ) by utilizing spherical polar coordinates @xmath220 ( @xmath221 , @xmath222 ) @xmath223 the dominating term for @xmath224 becomes @xmath225 having the form @xmath226 where the positive normalization constant is obtained explicitly as @xmath227 , e.g. @xcite , and holds only for @xmath228 .
relation ( [ recrit3 ] ) does not hold for @xmath2 where the fractional laplacian takes asymptotically for @xmath229 the localized singular distributional representation @xmath230 of the standard laplacian .
the details of this limiting case calculation can be found in @xcite .
we can identify the asymptotic representation ( [ recrit2 ] ) , ( [ recrit3 ] ) with the kernel of riesz fractional derivative ( fractional laplacian ) of the nd infinite space . for a more detailed discussion of properties
we refer to @xcite .
in the present paper we have developed a fractional lattice approach on nd periodic and infinite lattices .
the fractional laplacian matrices conserve the ` good ' properties of the laplacian matrices ( translational symmetry and in our sign convention negative semi - definiteness ) .
the fractional lattice approach generalizes the concept of second order centered difference operator appearing in the context of classical lattice models @xcite to the concept of centered fractional order difference operator . for @xmath2
the fractional lattice approach contains the classical lattice approach , and for integer orders @xmath231 finite centered differences of integer orders of the second difference operator are generated . for a discussion of properties of fractional laplacian matrix on the cyclic chain ,
we refer to @xcite . in the infinite space and periodic lattice continuum
limits these fractional laplacian matrices take the representations of the well known respective riesz fractional derivative kernels , i.e. the convolutional kernels of the ( continuous ) fractional laplacians .
the approach allows to model ` anomalous diffusion ' phenomena on lattices with fractional transport phenomena including asymptotic emergence of lvy flights .
in such a fractional lattice diffusion model , the conventional laplacian matrix is generalized by its fractional power law matrix function counterpart .
the formulation of our approach is consistent with recent works on the fractional approach developed on undirected networks by riascos and mateos @xcite .
the present fractional lattice approach represents a point of departure to investigate anomalous diffusion and fractional random walk phenomena on lattices .
such problems are defined by master equations involving fractional laplacian matrices such as deduced in the present work as generator matrices for the random dynamics .
fractional random walks on lattices and undirected networks open currently a huge interdisciplinary research field @xcite .
_ this work has been performed in the framework of the ercoftac sig/42 cooperation during the sabbatical stay of f.c.g.a nicolleau at institut jean le rond dalembert .
we thank the two anonymous reviewers for their useful remarks . _ abramowitz and stegun , handbook of mathematical function , national bureau of standards : applied mathematics series - 55 , ( 1972 ) , p. 376 .
m. born , k. huang ( 1954 ) dynamical theory of crystal lattices , london : oxford university press .
gelfand , g.e .
shilov , generalized functions , vol .
i : properties and operations ( new york : academic ) , ( 1964 ) .
r. hilfer , threefold introduction to fractional derivatives , in : anomalous transport : foundations and applications , r. klages et al .
( eds . ) , wiley - vch , weinheim , 2008 , pp 17 , isbn : 978 - 3 - 527 - 40722 - 4
. n. laskin , fractional schrdinger equation , phys .
e 66 , 056108 ( 2002 ) .
n. laskin , a. zaslavsky , nonlinear fractional dynamics on a lattice with long - range interactions , physica a 368 ( 2006 ) , 38 - 45 .
r. metzler , j. klafter , the random walk s guide to anomalous diffusion : a fractional dynamics approach .
physics reports 339 , pp
. 1 - 77 ( 2000 ) .
r. metzler , j. klafter , the restaurant at the end of the random walk : recent developments in the description of anomalous transport by fractional dynamics , j. phys . a : math .
37 r161r208 ( 2004 ) .
michelitsch , b. collet , a.f nowakowski , f.c.g.a .
nicolleau , fractional laplacian matrix on the finite periodic linear chain and its periodic riesz fractional derivative continuum limit , j. phys .
a : math . theor .
48 295202 ( 2015 ) .
( arxiv:1412.5904 ) .
michelitsch , b. collet , a.f .
nowakowski , f.c.g.a .
nicolleau , lattice fractional laplacian and its continuum limit kernel on the finite cyclic chain , chaos , solitons & fractals 82 , pp .
38 - 47 ( 2016 ) .
( arxiv:1511.01251 ) .
t. michelitsch , b. collet , x. wang , nonlocal constitutive laws generated by matrix functions : lattice dynamics models and their continuum limits , international journal of engineering science * 80 * , 106123 ( 2014 ) .
michelitsch , g.a .
maugin , a.f .
nowakowski , f.c.g.a .
nicolleau , m. rahman , the fractional laplacian as a limiting case of a self - similar spring model and applications to n - dimensional anomalous diffusion .
fractional calclulus and applied analysis vol .
16 , no.4 , 827 - 859 ( 2013 ) .
michelitsch ; g.a .
maugin , s. derogar , m. rahman , a regularized representation of the fractional laplacian in n dimensions and its relation to weierstrass - mandelbrot - type fractal functions , i m a journal of applied mathematics 79 , 753777 ( 2014 ) .
michelitsch , b.a .
collet , a.p .
riascos , a.f .
nowakowski , f.c.g.a nicolleau , fractional random walk lattice dynamics , to be published .
maradudin , a. a. , montroll , e. a. , & weiss , g. n. ( 1963 ) .
theory of lattice dynamics in the harmonic approximation .
solid state of physics .
new york : academic press .
i. podlubny , fractional differential equations , mathematics in science and engineering , vol 198 , academic press california 1999 .
a. p. riascos , jos l. mateos , long - range navigation on complex networks using lvy random walks , phys .
e 86 , 056110 ( 2012 ) a.p .
riascos , j.l .
mateos , fractional dynamics on networks : emergence of anomalous diffusion and l evy flights , phys .
e 90 , 032809 ( 2014 ) .
( arxiv:1506.06167 ) .
riascos , j.l .
mateos , fractional diffusion on circulant networks : emergence of a dynamical small world , j. stat .
( 2015 ) p07015 .
a. p. riascos and jos l. mateos , fractional quantum mechanics on networks : long - range dynamics and quantum transport , phys .
e 92 , 052814 ( 2015 ) .
ortiguera , riesz potential operators and inverses via fractional centered derivatives , international journal of mathematics and mathematical sciences , no .
48391 , 112 ( 2006 ) .
s. samko , a. kilbas and o. marichev , fractional integrals and derivatives : theory and applications , gordon and breach , london ( 1993 ) .
s. samko , fractional weyl - riesz integrodifferentiation of periodic functions of two variables via the periodization of the riesz kernel , applicable analysis , vol .
82 , no 3 , 269 - 299 ( 2003 ) .
tarasov , map of discrete system into continuous , journal of mathematical physics . vol.47 .
no.9 . ( 2006 ) 092901 .
tarasov , continuous limit of discrete systems with long - range interaction , journal of physics a. vol.39 .
( 2006 ) 14895 - 14910 .
tarasov , lattice fractional calculus , applied mathematics and computation 257 ( 2015 ) 1233 .
tarasov , exact discrete analogs of derivatives of integer orders : differences as infinite series , v.e .
tarasov , exact discretization by fourier transforms , communications in nonlinear science and numerical simulation .
( 2016 ) 31 - 61 .
journal of mathematics .
vol . 2015 .
( 2015 ) article i d 134842 .
tarasov , united lattice fractional integro - differentiation " , fractional calculus and applied analysis .
( 2016 ) 625 - 664 .
tarasov , g.m .
zaslavsky , fractional dynamics of coupled oscillators with long - range interaction , chaos . vol.16 .
( 2006 ) 023110 .
a. zoia a , a. rosso , m. kardar , fractional laplacian in bounded domains , phys . rev .
e 76 , 021116 ( 2007 ) . | we develop physically admissible lattice models in the harmonic approximation which define by hamilton s variational principle fractional laplacian matrices of the forms of power law matrix functions on the @xmath0-dimensional periodic and infinite lattice in @xmath1 dimensions .
the present model which is based on hamilton s variational principle is confined to conservative non - dissipative isolated systems .
the present approach yields the discrete analogue of the continuous space fractional laplacian kernel . as continuous fractional calculus generalizes differential operators such as the laplacian to non - integer powers of laplacian operators , the fractional lattice approach developed in this paper generalized difference operators such as second difference operators to their fractional ( non - integer ) powers .
whereas differential operators and difference operators constitute local operations , their fractional generalizations introduce nonlocal long - range features .
this is true for discrete and continuous fractional operators .
the nonlocality property of the lattice fractional laplacian matrix allows to describe numerous anomalous transport phenomena such as anomalous fractional diffusion and random walks on lattices .
we deduce explicit results for the fractional laplacian matrix in 1d for finite periodic and infinite linear chains and their riesz fractional derivative continuum limit kernels .
the fractional lattice laplacian matrix contains for @xmath2 the classical local lattice approach with well known continuum limit of classic local standard elasticity , and for other integer powers to gradient elasticity .
we also present a generalization of the fractional laplacian matrix to n - dimensional cubic periodic ( nd tori ) and infinite lattices .
we show that in the continuum limit the fractional laplacian matrix yields the well - known kernel of the riesz fractional laplacian derivative being the kernel of the fractional power of laplacian operator . in this way
we demonstrate the interlink of the fractional lattice approach with existing continuous fractional calculus .
the developed approach appears to be useful to analyze fractional random walks on lattices as well as fractional wave propagation phenomena in lattices . |
in this paper we comment on the nature of the superconducting gap in the cuprates . recently ,
deutscher has identified two distinct energy scales at low temperatures ( in the superconducting state ) , which he calls the coherence energy @xmath0 and the single - particle excitation energy @xmath3 @xcite .
the former corresponds to the superconducting gap seen in phase - sensitive experiments ( andreev reflection , penetration depth and raman scattering ) , while the later is the `` pseudogap '' seen in experiments that probe the one - particle spectrum ( photoemission and tunnelling ) .
in fact some experiments belonging to the later class seem to be able to detect both `` gaps '' simultaneously : specifically , @xmath0 and @xmath3 appear to have been resolved , as two distinct features in the single - particle excitation spectrum , in angle - integrated photoemission experiments on lsco by sato et al .
@xcite and in intrinsic c - axis tunneling spectrosopy experiments on bscco by krasnov et al .
@xcite here we address the following question : `` which , if any , of these two energies is the ` superconducting gap ' , in the bcs sense ? ''
since the issue that we wish to explore is fairly general , we use a minimal model which might be expected to contain the relevant physics .
the basic idea is to exploit the fact that certain properties of cuprate superconductors seem to be approximately material - independent .
in particular , the critical temperature @xmath4 @xcite as well as the two energy scales @xmath5 and @xmath3 @xcite , when normalised to the critical temperature at optimal doping , @xmath6 , appear to be fairly universal functions of the number of holes per @xmath7 unit , @xmath8 . thus one could expect that a very simple model which captures the essential physics in these systems can describe this behaviour .
more realistic models would only be required to calculate material - specific properties , such as @xmath2 ( as opposed to @xmath9 ) accurately . in a sense
the rationale behind our approach is the same as in the bcs model which , with a very simplified description of the electronic structure and the electron - electron interaction , can nevertheless predict the isotope effect and the constant value of the `` bcs ratio '' @xmath10 very accurately , for most of the conventional superconductors .
unfortunately our model of the electron - electron interaction in the cuprates is only pehnomenological , and hence we shall have little to say about the nature of the physical mechanism giving rise to pairing . nevertheless , as we shall show presently the model has a number of physical features which if confirmed , by comparison with experiments , will shed some light on the central problem of the high-@xmath2 phenomena .
in particular , the model that we shall use has been showed to describe quantitatively the rise and fall of @xmath11 @xcite .
moreover , it facilitates the investigations of superconductivity in both the bcs and bose - einstein ( be ) limits @xcite .
the central new result , presented here , is that the prediction of the model for the superconducting gap in the single - particle excitation spectrum , @xmath12 , coincides , quite accurately , with the _ lower _ of the two energy scales identified by deutscher , namely the `` coherence energy '' @xmath13 .
on the basis of this we argue that , although the model is too simple to describe the superconducting state of the cuprates in detail , some of its properties may generically characterise these materials . in short , it provides evidence that @xmath0 can be understood as a bcs - like @xmath14-wave gap , while the other energy scale , @xmath3 , seems to be quite unrelated to superconductivity .
interestingly similar conclusions were reached in ref .
@xcite on the basis of experimental investigations .
moreover , our approach , based on the universal properties of cuprate superconductors , complements the one recently adopted by szotek , blg and temmerman @xcite , who used a detailed 8-band model of the electronic structure of a specific material ( ybco ) to study the same question , reaching similar conclusions .
our model consists of free electrons ( holes ) of effective mass @xmath15 interacting _ via _ the attractive `` delta - shell '' potential ( between opposite spins ) @xmath16 this is , arguably , the simplest generalisation of the bcs model that can describe exotic pairing @xcite .
the four parameters of the delta - shell model are the `` coupling constant '' @xmath17 ( which has dimensions of @xmath18 ) , the distance at which the attraction takes place @xmath19 , the effective mass @xmath20 , and the density of charge carriers , per unit volume , @xmath21 .
although formally similar to the `` contact potential '' of bcs theory , the physics described by the delta - shell potential are very different : it corresponds to non - retarded attraction at a finite distance @xmath19 . in that sense , it is more similar to the hubbard - like models with nearest - neighbour attraction @xcite . but note that in the delta - shell model the distance @xmath19 at which the electrons attract each other is a free parameter that can be varied continuously , and the non - interacting dispersion relation is that of free electrons with an effective mass @xmath15 . a preliminary analysis of this model has been carried out using a standard functional integral technique .
the details of these calculations are to be found in ref .
@xcite , however some of the results will be summarised in what follows as the need arises and , for conveninence , the relevant formulae have been reproduced in the appendix .
a remarkable feature of the experimental data on the superconducting cuprates is the dependence of @xmath11 on the number of holes , @xmath8 , per cuo@xmath23 @xcite .
namely , @xmath11 rises and falls as @xmath8 increases from below optimal doping , @xmath24 , to above , and the functional form of this relation is the same for the different materials @xcite .
explicitely , it is given by @xmath25\ ] ] remarkably , in the delta - shell model , the ratio @xmath11 does not depend on @xmath17 , @xmath19 , @xmath15 and @xmath21 independently , but only on the two dimensionless quantities @xmath26 and @xmath27.[def - g - n ] the dimensionless density @xmath28 can be related to @xmath8 through the value of the ratio @xmath29 , where the effective cu
cu distance , @xmath30 , is defined in terms of the proportionality between the number of holes per unit and their density per unit volume : @xmath31 .
thus the values of the dimensionless coupling constant @xmath32 and the rescaled range of the attraction , @xmath33 , completely determine , for our model , the dependence of @xmath11 on @xmath8 .
we find that , as in the cuprates , @xmath11 rises and falls as a function of @xmath8 , for fixed @xmath32 and @xmath34 .
moreover , as we have pointed out earlier @xcite , we can choose these two parameters so that this rise and fall reproduces the shape of the experimental curve , an inverted parabola defined by the position of its maximum and its width .
the condition that @xmath32 and @xmath34 are fixed to the values that produce the best fit , in the @xmath35-channel ( cooper pairs with angular momentum quantum number @xmath36 ) , is equivalent to the following two relations between the four parameters @xmath17 , @xmath19 , @xmath15 and @xmath37 : @xmath38 and @xmath39 note that @xmath15 and @xmath30 are normal state properties independent of the attractive interaction in eq .
( [ delta - shell interaction ] ) . for clarity , the corresponding fit is recorded in figure [ quigyof1 ] . evidently , the above relations determine @xmath17 and @xmath19 for each system : ybco , lsco , etc . , characterised by the frankly phenomenological , but neverteless clearly material - specific , parameters @xmath15 and @xmath37 .
thus although the four parameters are independent of doping , their fixed values are different for different systems .
the universal character of the empirical law of eq .
( [ eq5 ] ) , on the other hand , is embodied in the fact that the relations ( [ parameter - values-1],[parameter - values-2 ] ) between the four quantities are the same for all systems . by fitting relative quantities like @xmath40 and universal trends with doping
we can hope to have obtained results dependent more on the overall properties of our model than on its material - specific , no doubt inadequate , details .
we would also like to stress that other , @xmath41 , singlet pairing channels could not fit the oberved dependence of @xmath11 on the doping level . in particular , for pairs with angular momentum quantum number @xmath42 ( @xmath43-wave ) , the critical temperature is finite for arbitrarily small values of the density , in contradiction with the observed behaviour , while other pairing channels are negligible in the range of densities of interest @xcite .
thus the interpretation of the observed rise - and - fall of @xmath2 with doping in terms of the delta - shell model implies @xmath35-wave symmetry in agreement with the current consensus @xcite .
as a further validation of our model we will consider , briefly , the dependence of @xmath2 on pressure . experimentally , there are two contributions to the variation of @xmath2 with applied hydrostatic pressure @xcite .
the extrinsic contribution is caused by a pressure - induced change of the doping level , combined with the universal @xmath2 vs doping law ( [ eq5 ] ) .
on the other hand , there is also an _ intrinsic _ enhancement of @xmath4 , i.e. an increase of the value of @xmath44 . for the values of the parameters given in eqs .
( [ parameter - values-1],[parameter - values-2 ] ) , @xmath44 scales with @xmath45 and is given by @xmath46 thus although @xmath11 is universal , in the sense that it only depends on @xmath8 , the absolute value of @xmath2 is material - specific , as one would expect in view of the experimental data @xcite , because it depends also on @xmath15 and @xmath47 .
although , along the lines set in the introduction , we do not expect this simple model to reproduce the absolute values of @xmath44 accurately , if it contains any of the relevant physics must be compatible with the values of @xmath44 observed experimentally .
notably , in the cuprates , @xmath44 can go from @xmath48 to @xmath49 , which is consistent with our prediction if @xmath50 and the effective masses are of the order of a few times the bare electron mass .
moreover eq . ( [ tcmax ] ) provides a simple interpretation for the origin of the _ intrinsic _ enhancement of the critical temperature : it is due to the increase of the `` localisation energy '' @xmath45 associated with the reduction in size of @xmath47 . by taking the derivative of @xmath44 with respect to the volume of the unit cell @xmath51 ( which is proportional to @xmath52 )
we can obtain an estimate of such enhancement : @xmath53 k_{v}\ ] ] here @xmath54 is the compressibility of the material .
note that in writing ( [ eq9 ] ) we have assumed that the relations ( [ parameter - values-1],[parameter - values-2 ] ) do not depend on applied pressure .
this would be consistent with the universality of ( [ parameter - values-1],[parameter - values-2 ] ) if that universality were exact , but note that such universality is probably only approximate .
even so , we can easily use ( [ eq9 ] ) to obtain a specific value as follows .
we can employ the usual way of estimating the volume dependance of the electronic effective mass , which is of interest in many fields of solid state physics , by an appeal to the relationship between @xmath55 and an appropriate hopping integral , @xmath56 , in the tight - binding description of the relevant band . in the simplest case of a cubic lattice with constant @xmath57
this is @xmath58 .
then we use the dependance of @xmath56 on the nearest - neighbour distance to estimate the changes in @xmath15 with changes in the lattice parameters .
thus we may use @xmath59 and , assuming that we are dealing with @xmath35-orbitals , take @xmath60 @xcite . this yields @xmath61 unfortunately , we are not aware of many measurements of the variation of the _ maximum _ value of the critical temperature under applied hydrostatic pressure
. however wijngaarden _ _ _ _ _ et al .
_ @xcite have studied the single layer compound @xmath62 and from their data we deduce @xmath63 . using the value of the compressibility calculated by cornelius _
@xcite , @xmath64 , and our formula in equation ( [ eq10 ] ) , we predict @xmath65 , in reasonable agreement with the experiment . similarly ,
_ @xcite have obtained @xmath66 for @xmath67 which , using @xmath68 , yields @xmath69 . the theoretical prediction in this case , using the value @xmath70 quoted in ref .
@xcite , is @xmath71 .
thus the _ intrinsic _ enhancement of the critical temperature with applied hydrostatic pressure can be understood essentially , at the qualitative level , as the result of the change of the length @xmath47 , and therefore of the corresponding localisation energy @xmath45 , under pressure .
, which is more general than ( [ tcmax ] ) and , in fact , would hold for any model for which @xmath72 is the relevant energy setting the scale of @xmath44 . ]
we end by stressing that the results presented in this section ( unlike those in the previous and following ones ) refer to material - specific properties of the cuprates and therefore , in accordance with the philosophy outlined in the introduction , one must not expect our simple model to provide a detailed description of them .
for example , it can not describe the observed effect of the number of adjacent cuo@xmath23 planes per unit cell , nor of applied uniaxial stress @xcite , on @xmath2 , since obviously the parameters @xmath73 and @xmath15 do not describe the varying degrees of anisotropy of the normal states .
we shall now return to the main topic of this paper . on the basis of the discussion in the previous sections
, we can expect that the delta - shell model , together with the relations ( [ parameter - values-1],[parameter - values-2 ] ) , can describe quantitatively other material - independent properties of the superconducting state of the cuprates .
in fact it _ predicts _ , with no adjustable parameters , the value of any such quantities .
the most interesting one is the ratio between the energy gap to one - particle excitations , at zero temperature , and the critical temperature at optimal doping : @xmath74 . in the model ,
this quantity is universal , in the same sense as @xmath11 , because @xmath75 also scales with @xmath45 . in a very general context , there are two distinct scenarios , which we shall now briefly recall ( see @xcite , and references within ) . in the bcs regime ( @xmath76 )
the fermionic quasiparticles with minimum excitation energy lie on a surface in * @xmath77 * -space that is defined by @xmath78 and the corresponding gap is @xmath79 , where @xmath80 is the usual bcs `` gap function '' .
the extreme case of this is the bcs _ limit _ , defined by @xmath81 for all * @xmath82 * , in which the chemical potential is approximately equal to the fermi energy , @xmath83 , and so the fermionic quasiparticles with minimum energy have wave vector @xmath84 , i.e. they lie in the fermi surface . in the opposite ,
be regime ( @xmath85 ) , quasiparticles with minimum excitation energy have @xmath86 and the energy gap to single - particle ( fermionic ) excitations is @xmath87 , which , in the be _ limit _ ( @xmath88 ) , becomes the binding energy @xmath89 of a pair : @xmath90 .
however in the be regime the lowest - lying excitations are not fermionic , but bosonic , and therefore there is a second , smaller energy scale , which is associated with the break - up of coherence in the condensate without the dissociation of any pairs , whose value could be related to @xmath91 .
the delta - shell model can display both bcs - like and be - like behaviour , depending on the values of the dimensionless parameters @xmath32 and @xmath28 introduced above . specifically , for the values that correspond to eqs .
( [ parameter - values-1],[parameter - values-2 ] ) , the model is close to the bcs limit @xcite .
in fact the rise - and - fall of @xmath2 with the doping level that we have found in the delta - shell model is only present at low values of @xmath92 . at higher values of the dimensionless coupling constant @xmath32 ,
the critical temperature becomes a monotonic function of the density , which furthermore has a finite value for any finite value of @xmath21 , as it tends towards the bose - einstein condensation temperature @xmath93 .
moreover the rotational symmetry of the model is restored in this limit due to the fragmentation of the condensate . thus on the basis of the discussion presented in the previous section
we conclude the first of the above scenarios to be operating in the cuprates .
namely , we expect there to be a single energy scale @xmath94 coming from the superconductivity , which would manifest itself both as the characteristic energy seen in phase sensitive experiments and as a gap in the single - particle excitation spectrum . in this picture ,
the origin of the larger `` pseudogap '' energy @xmath95 would be quite independent from superconductivity .
figure [ gd ] shows the experimental data from ref .
@xcite for the two energy scales @xmath0 and @xmath3 , together with our theoretical prediciton for the maximum value of the superconducting energy gap in the single - particle spectrum across the fermi surface , @xmath75 , for the purported @xmath14 symmetry of the order parameter @xcite .
all quantities are normalised to @xmath96 , which is the way in which , both in the model and , approximately , in the experiments , their values become material - independent .
remarkably , the agreement between @xmath75 and @xmath0 is rather good .
in particular the ratio of @xmath75 to @xmath96 is @xmath97 at optimal doping , both in the theory and the experiments . obviously , because of the simplicity of our model it is not easy to decide what to make of the above very good agreement .
nevertheless , at the minimum we conclude that it provides evidence in favour of the bcs - like scenario .
moreover the model seems again to confirm the current consensus on the symmetry of the order parameter : for pairs with @xmath98 symmetry,[d3z2-r2 ] @xmath99 at optimal doping , which is too high to fit the experiments ( note that , unlike at @xmath2 , at @xmath100 the different @xmath35-wave channels are non - degenerate ) . presumably , it is the presence of the crystal field that makes one of the @xmath35-wave symmetries preferred over the others .
since this is absent from our model all we can do is examine each of the possible symmetries independently and compare them to the experiments . of course , our interpretation of the data presented in ref .
@xcite differs from that of the author . in effect
, it was assumed at the time that there was only one energy gap in the single - particle excitation spectrum , corresponding to the `` pseudogap '' energy @xmath3 .
this was legitimate back then because only experiments that probed the phase coherence of the condensate ( andreev , raman , penetration depth ) had been able to measure @xmath0 .
however , as we mentioned in the introduction , it has recently been shown experimentally @xcite that there are actually two gaps in the spectrum of single - particle excitations : one of them corresponds to @xmath3 , while the other , smaller one closes in at @xmath101 , very much like a bcs gap .
it is this second gap that we have obtained in our calculations , and in our picture it should coincide with the coherence energy @xmath0 .
as regards the other energy scale , @xmath3 , we have no sign of it in our theory and hence we are inclined to conclude that its origin is not superconductivity
. in particular , it can not be due to `` preformed pair '' fluctuations of the superconducting order parameter which , as we have discussed , are ruled out on the basis of the empirically determined values of the parameters , eqs .
( [ parameter - values-1],[parameter - values-2 ] ) , and their interepretation in terms of the calculations presented in ref .
we have addressed the question of whether the coherence energy range @xmath0 , identified by deutscher , at low temperatures , in cuprate superconductors , as distinct from the `` pseudo - gap '' energy @xmath3 , can be understood as a superconducting gap in the sense of bcs theory .
capitalising on the fact that certain properties of cuprate superconductors are approximately material - independent , we have studied a simple model , obtained as the simplest generalisation of the model compatible with exotic pairing .
it features free electrons with a non - retarded , effective interaction potential which is attractive at a well - defined finite distance @xmath19 and is zero for all other separations .
this `` delta - shell '' model provides a mechanism for the rise and fall of @xmath2 with doping , as a consequence of the interplay between the carrier - carrier distance or , in other words , the fermi wavelength @xmath102 and the new microscopic length scale @xmath19 .
morover , it can correlate this behaviour with the @xmath35-wave symmetry of the order parameter and it provides an interpretation of its approximate universality in terms of certain relations between the parameters of the model .
on this basis , the model can be used to predict , with no fitting parameters , the corresponding universal behaviour of the gap in the quasiparticle spectrum due to the superconductivity . for the empirically determined values of the parameters
the model is in the bcs regime , as opposed to that of bose - einstein condensation of `` preformed pairs '' , which leads to the prediction that the gap to single - particle excitations arising from the superconductivity is equal to @xmath0 .
in fact the prediction for that gap agrees quite satisfactorily with the behaviour of @xmath0 , suggesting a different origin for the pseudogap @xmath3 .
the above conclusion is compatible with general arguments based on the available experimental data which indicate that the descripition of the charge carriers in the bose - einstein limit is inappropriate @xcite .
although a model that accomodates this fact while still describing the pseudogap in terms of pairing fluctuations has been devised @xcite , it has also been suggested @xcite that the pseudogap may follow from the spectral properties of stripe phases @xcite .
more generally , a pseudogap may be a generic feature of systems that are intrinsically inhomogeneous due to the competition between ordered states separated by a first - order phase transition @xcite .
such picture becomes especially appealing in the light of recent scanning tunnelling microscope studies of bscco @xcite .
it could be investigated in our model by evaluating the density fluctuations induced by the attractive interaction , in addition to the pairing fluctuations considered in the present work .
our analysis complements the one carried out by szotek , blg and temmerman @xcite , who also modeled @xmath0 as a bcs superconducting gap , reaching similar conclusions .
the main difference between our approach and theirs is that , instead of focusing on a specific material using a detailed description of the electronic structure , we have used a very simple model to capture those features that are approximately material - independent .
specifically , our calculations are not based on a fit to the value of @xmath44 ( which can not be predicted accurately by such a simple model ) , but rather to the width of the @xmath11 vs @xmath8 curve , and the position of its maximum , which are determined by the values of the range of the interaction @xmath19 and its strength @xmath103 , rescaled to the effective cu
cu distance @xmath73 and the corresponding `` localisation energy '' @xmath104 , respectively .
our result that the model is in the bcs limit for the empirically determined values of these parameters speaks in support of the methodology employed in ref .
@xcite , where this was assumed _ a priori _ by using a bcs - like mean - field theory to calculate @xmath2 .
finally we stress that the rise and fall of the critical temperature @xmath2 and the superconducting gap @xmath75 , in the bcs limit , is probably a generic behaviour for attractive potentials with a minimum at some finite separation .
essentially , the gap function depends , in this case , on the _ magnitude _ of @xmath82 as well as on its direction : @xmath105 , and thus its value on the fermi surface @xmath106 depends strongly on the size of the fermi surface , given by @xmath107 .
this situation is quite different from the bcs model , and its usual @xmath108-wave generalisations ( * ? ? ?
* for example ) , for which the gap function is independent of @xmath109 .
note that the present scenario for the rise and fall of @xmath2 is quite different from previous proposals .
in particular , in our picture this rise and fall is not due to the competition of superconductivity with other kinds of order or the suppression of @xmath2 by fluctuations away from optimal doping .
moreover , the value of @xmath2 does not necessarily correlate with the dos on the fermi surface .
can have a non - monotonic dependence on doping .
interestingly , a rise and fall of the critical temperature with doping , which does not seem to correlate to the dos , has been observed in experiments on gate voltage - doped c@xmath110 films @xcite . ]
it could be tested if an experimental method to measure directly the variation of the gap function @xmath111 with @xmath82 in the radial direction of increasing @xmath112 , rather than along the fermi surface , e.g. one that would give the value of @xmath113 , were available .
we wish to thank james annett , jon wallington and klaus capelle for useful discussions , and phillip howell for detailed , constructive criticism of an earlier version of the manuscript .
blg wishes to thank the institute of theoretical physics at the university of california at santa barbara for hospitality during the preparation of an earlier version of the manuscript .
jq acknowledges financial support from the tmr programme ( eu ) , _ via _ a marie curie fellowship ( contract no .
erbfmbict983194 ) and , during the final stages of the preparation of the manuscript , from fapesp ( brazil ) , _ via _ a post - doctoral fellowship ( process no . 01/10461 - 8 ) .
@xmath114present address : instituto de fsica de so carlos ( grupo de fsica terica ) , universidade de so paulo , caixa postal 369 , 13560 - 970 so carlos , sp , brazil .
@xmath115electronic address : [email protected] .
in ref . @xcite the ground state of the delta - shell model
was studied in the mean - field approximation in which the gap function @xmath116 and the chemical potential @xmath117 are determined by two coupled equations @xcite .
the theory is particularly transparent in terms of their dimensionless counterparts @xmath118 and @xmath119 ( where @xmath120 is the dimensionless wave vector ) .
if @xmath121 is expanded in spherical harmonics @xmath122 , it is found to have the form @xmath123 i.e. the radial part is given by the spherical bessel functions @xmath124 , and the two equations are , for singlet pairing ( @xmath125 for @xmath126 odd ) , @xmath127 where @xmath128 and @xmath129 are the dimensionless coupling constant and density defined in the text , and @xmath130 all the ground - state properties that we discuss in the text were obtained by solving the above two equations self - consistently , for `` trial gound states '' in which specific constraints were imposed on the @xmath131 .
note that these equations are valid approximations not only in the bcs limit , encountered for large @xmath129 or when @xmath128 is below the threshold to bind an isolated pair , but also in the opposite , bose - einstein condensation limit @xcite . to find the condensation temperature @xmath132 we went a small step beyond the mean - field theory , including quadratic fluctuations about it . in a manner similar to that reviewed in ref .
@xcite we found that @xmath133 is determined by @xmath134 } { 2\left(|\tilde{\bf k}|^2-\tilde{\mu}_{c}\right ) } \label{rescaled - tc - eq } \\
\tilde{n}&= & \frac{4}{3\pi } \int _ { 0}^{\infty } d|\tilde{\mathbf{k}}| \left\lbrace f\left [ \tilde{\beta } _ { c}\left(|\tilde{\bf k}|^2-\tilde{\mu}_{c}\right)\right ] + g\left ( \tilde{\beta } _ { c}\frac{|\tilde{\mathbf{k}}|^{2}}{2}\right ) \tilde{w}^{\tilde{\beta } _ { c},\tilde{\mu } _ { c}}_{l,| \tilde{\mathbf{k}}| } \right\rbrace \label{density - eq - rescaled}\end{aligned}\ ] ] where @xmath135 is the value of @xmath136 at the transition and @xmath137 and
@xmath138 denote the usual fermi and bose distribution functions .
the term containing the bose function , times the `` weight '' @xmath139 , is the contribution from the above mentioned fluctuations , described here to the lowest non - trivial order in their frequency and momentum , and it signals the presence of preformed pairs , with the same internal angular momentum quantum number @xmath126 as the condensate , above @xmath132 ( in these expressions , the contribution from other pairs has been neglected ) .
this contribution dominates in the bose - einstein limit ( where its `` weight '' takes the form @xmath140 , and thus it simply accounts for the degeneracy of the internal state of the pairs ) while being negligible in the bcs limit .
note that the critical temperature is degenerate in the magnetic quantum number @xmath141 , although it still depends on @xmath126 . from the point of view of the present paper
it is particularly relevant that both @xmath142 and @xmath143 depend only on the dimensionless parameters @xmath128 and @xmath129 , as direct consequences of eqs .
( [ rescaled - gap - eq],[rescaled - n - eq ] ) and ( [ rescaled - tc - eq],[density - eq - rescaled ] ) , respectively . | recent experiments indicate that the excitation spectrum of the cuprates is characterised , in the superconducting state , by two energy scales : the `` coherence energy '' @xmath0 and the `` pseudogap '' @xmath1 . here
we consider a simple generalisation of the bcs model that yields exotic pairing and can describe , phenomenologically , the generic trends in the critical temperature @xmath2 of cuprate superconductors .
we use the model to predict the gap in the single - particle spectrum arising from the superconductivity and we find evidence that it corresponds to the lower of the two energy scales , @xmath0 , seen in the experiments .
this provides further support to the view that the origin of the pseudogap is not superconducting fluctuations .
+ pacs numbers : 74.72.-h , 74.25.dw , 74.62.dh , 74.20.rp , 74.62.fj |
in the context of the canonical quantization of diffeomorphism invariant theories of connections for a compact gauge group , in particular , quantum gravity in four spacetime dimensions , the quantum configuration space is coordinatized by certain elementary operators which could be called generalized wilson functions .
these functions are labelled in general by a set of paths in the manifold in question .
it turns out that the precise nature of these paths has a deep impact on the structure on the resulting quantum theory .
all the considerations in quantum gravity to date concern the case that the paths defining a given wilson function can intersect only finitely many times . to insure this , piecewise analyticity was often assumed .
therefore in the sequel we loosely denote that case by the analytic category " . in the smooth category " , however , the known results are sparse . in this letter
we extend the pioneering work by baez and sawin on this issue . specifically , this letter is organized as follows : + in section 2 we state the problem with the extension of results from the analytic to the smooth category and recall the results due to baez and sawin .
then we prove a master theorem which extends the results due to these authors and which is key to sections 3 , 4 and 5 .
in section 3 we apply the master theorem to show how the results concerning the ashtekar - isham algebra can be immediately extended . in section 4 and 5
we define the notion of a spin web in analogy to the spin - nets of the analytic category in order to apply the master theorem to the extension of results by ashtekar , marolf , mouro and the authors concerning diffeomorphism invariant states of connections and associated diffeomorphism invariant operators .
finally , in section 6 we apply the machinery of the previous sections to the operators that have proved to be elementary building blocks for the quantization of gravitational field .
this section is divided into three parts . in the fist part we recall the basic notions of cylindrical functions on the space of distributional connections and inductive limit structures and
review the results that one obtains if one labels cylindrical functions by piecewise analytic , finitely generated graphs . in the second part we review
the construction , due to baez and sawin , of a new label set , called webs " , which are appropriate for the case that the loops defining the wilson functions are only piecewise smooth and may intersect each other an infinite number of times .
finally , in the third part we prove the master theorem on which the rest of this letter is based . in the loop quantization of gravity as initiated by ashtekar @xcite and jacobson , rovelli and smolin @xcite and gambini et .
( see e.g. @xcite and references therein ) through embedding general relativity into a yang - mills theory phase space and making use of wilson loop variables as well as in a general framework for quantization of a diffeomorphism invariant theory of connections with local degrees of freedom @xcite one uses the algebra of cylindrical functions defined by the parallel transport maps as the main device [ 4 - 30 ] .
the algebra , let us call it @xmath4 , is a sub - algebra of the algebra of functions defined on a space of @xmath5 connections @xmath6 .
the connections are defined on a bundle over a manifold @xmath7 . for the simplicity
we assume here that the bundle is trivial and we fix a global gauge ( section ) .
we are assuming that @xmath5 is compact and semi - simple .
a complex valued function @xmath8 is called cylindrical , if there are piecewise smooth paths @xmath9\ \rightarrow\ \sigma$ ] continues on the whole interval of @xmath10 $ ] such that its ( one - sided ) derivative never vanishes except for the constant path ; here we are actually interested in equivalence classes of paths where two paths are equivalent iff they differ by an orientation preserving piecewise smooth diffeomorphism @xmath10\mapsto [ s_0,s_1]$ ] . ] in @xmath11 and a function @xmath12 such that [ cyl ] ( a ) = ( u_p_1(a), ... ,u_p_n(a ) ) where @xmath13 represents the parallel transport along the path @xmath14 with respect to the connection @xmath15 . to control the algebra it is useful to decompose @xmath4 into a union [ union ] = _ w_w of sub - algebras that are easier to handle ( @xmath16 is a labeling set and will be specified below ) .
this decomposition is particularly convenient if it defines a so called ` inductive limit ' , that is , if for every @xmath17 , there exists @xmath18 such that _ w_1,_w_2_w. one might take for @xmath16 the set of all the finite families of piecewise smooth paths in @xmath7 and associate to each finite family of paths @xmath19 the sub - algebra @xmath20 given by all the paths obtained from the elements of @xmath19 and their inverses by using the path product .
that does define an inductive limit decomposition , however , given a general family of paths @xmath19 , we do not know much about the corresponding @xmath21 .
an example of a sub - algebra @xmath21 we do know much about is the sub - algebra corresponding to a family of paths @xmath19 which is an embedded graph
. such that every two can share only one or two points of the boundary . given a graph ,
its elements are called edges . by an edge , we shall also mean an oriented 1-sub - manifold with boundary . ] then , @xmath21 can be identified with the algebra of functions @xmath22 where @xmath23 equals the number of edges in @xmath19 , via the map [ proj ] a ( u_e_1(a), ... ,u_e_n(a))_wg^n . where @xmath24 are the edges of @xmath19 and @xmath25 is the image of that map .
such a map is defined for any family of paths , but in the case of a graph the map ( [ proj ] ) is _ onto _ , therefore _ w = g^n .
if one fixes an analytic structure on @xmath7 and defines @xmath26 to be the sub - algebra given by the piecewise analytic paths , then the analytic graphs are enough to decompose @xmath27 into the following inductive limit sum [ unan ] _ _ = _ ( ) , where the sum ranges over the set of piecewise analytic graphs in @xmath11 .
that is true due to the following fact . given two analytic graphs @xmath28 there is an analytic graph @xmath29 such that @xmath30 , where the inequality relation is defined as follows .
two families @xmath31 of paths , are in the relation [ le ] w w if every path - element of @xmath19 can be obtained from the elements of @xmath32 and their inverses by using the path product and @xmath33 such that the end point of @xmath34 is a starting point of @xmath33 , the path product @xmath35 is a path obtained by connecting the end of @xmath34 with the beginning point of @xmath33 . ] .
therefore , because of the composition rule of the parallel transports u_pp = u_pu_p we have _
_ 1,__2_. the above decomposition was key to the following issues : * the development of integral calculus on @xmath27 @xcite . * the characterization of the gelfand spectrum of the corresponding @xmath36 algebra when @xmath5 is compact @xcite . * measure theory on the spectrum @xcite .
* the introduction of differential calculus on the spectrum useful for introducing and studying the operators of quantum gravity @xcite . * the introduction of spin - networks as the orthonormal basis with respect to the natural integral in @xmath27 @xcite .
our goal now is to see to what extent the above applications can be generalized to the case of smooth paths .
the virtue of piecewise analytical graphs is that their edges can not intersect or overlap in an infinite number of distinct , isolated points or segments .
this is key to the definition of the algebra @xmath27 as an inductive limit since the union of two piecewise analytic graphs ( after a finite number of subdivisions ) again defines a piecewise analytic graph .
this is no longer true when two graphs are just piecewise smooth , an example being given by the graph that consists of the union of two smooth curves which intersect in a cantor set @xcite or the intersection of the graph of the function @xmath37\ni x\to e^{-1/x^2}sin(1/x)$ ] with the @xmath38-axis .
the graphs are not both contained in a same graph .
thus , for the full algebra @xmath4 the graphs are not enough to provide an appropriate decomposition .
recently baez and sawin @xcite introduced a suitable generalization of an embedded graph which they called a _
web_. for the purpose of this paper , it is enough to know the following properties of the webs due to by baez and sawin : @xmath39 a web is a finite family of piecewise smooth paths
@xmath40 which do not self intersect ; @xmath41 for every web @xmath42 the image of the map ( [ proj ] ) is a lie subgroup of @xmath43 ; @xmath44 for every finite family @xmath19 of smooth paths there exists a web @xmath32 such that @xmath45 ; thus , in contrast to the picewise analytic graphs which are labelled by its edges ( i.e. paths which intersect at most at their endpoints ) a baez - sawin web is labelled by the paths @xmath46 which may overlap or intersect each other possibly an infinite number of times . abusing the notation , we shall nevertheless denote the labels of a web by edges again to distinguish them from an arbitrary path . henceforth , we denote by @xmath16 the set of all the webs in @xmath7 .
we should also notice that baez - sawin s @xmath16 is preserved by the diffeomorphisms of @xmath7 .
the highly non - trivial result proved in @xcite is that @xmath16 indeed _
exists_. due to property @xmath44 , = _ w_w , and @xmath47 is a projective family .
the point @xmath41 gives us certain control on @xmath21 .
our first result will tell us more about @xmath21 .
we end this section with the baez - sawin definition of a web . for more details see @xcite .
a _ web _ is the ( set ) union of a finite number of families of edges so - called _
tassels_. first , to give the reader an intuitive picture , we explain how a graph gives rise to a web . given a graph
@xmath48 subdivide each of its edges @xmath40 into two edges oriented in such a way that they end in the subdivision point .
denote the resulting graph by @xmath19 .
a vertex of @xmath19 is either a vertex of @xmath48 or one of the subdivision points .
let @xmath49 be a vertex of @xmath19 which is also a vertex of @xmath48 .
the set of edges of @xmath19 leaving @xmath49 forms a tassel based at @xmath49 and the subdivided graph @xmath19 is a web ( a special case of a web ) . in general , edges of a tassel are allowed to intersect at points different from their end points and can overlap . given a family @xmath50 of edges , a point @xmath51 of a family of curves
@xmath50 is the union of the ranges of each element of @xmath50 . ]
is a _ regular point _ of @xmath50 if there is a neighborhood @xmath52 of @xmath53 such that the intersection @xmath54 is an embedding of an open interval .
a segment of that embedded interval we will call a _ regular segment_. ( in other words , if a curve @xmath55 intersects @xmath53 then it intersects @xmath53 exactly once , and any other @xmath56 either overlaps @xmath40 at @xmath53 or does not intersect @xmath53 at all . ) given a regular point @xmath53 of @xmath50 its _ type _ is the set of the elements of @xmath50 which overlap at @xmath53 .
* general definition of a web . *
a a family of edges is a _ tassel _
if it has the following properties : @xmath0 all the edges of @xmath50 begin at a same point ; we call it the _ base point _ of @xmath50 ; @xmath1 the edges @xmath57 , @xmath58 , of @xmath50 can be parameterized in such a way that e_i(t
) = e_j(s ) t = s ; @xmath2 two edges of @xmath50 that intersect at a point different from the base point @xmath14 intersect in every neighborhood of @xmath14 ; @xmath3 any type which occurs at some point @xmath51 occurs in every neighborhood of the base point @xmath14 .
paths @xmath9 are _ holonomically independent _ if for every @xmath59 there is a smooth connection @xmath60 such that ( u_p_1(a), ...
,u_p_n(a ) ) = ( g_1, ... ,g_n)g^n . as we already indicated in section ( [ s1 ] ) , the edges of any graph @xmath61 are holonomically independent , hence the corresponding @xmath62 can be identified with the whole of @xmath22 .
however , the results of @xcite only show that for an @xmath23-element web @xmath19 the set @xmath21 can be identified with @xmath63 where a priori @xmath64 can be any lie subgroup of @xmath43 that may even vary with @xmath19 .
our master result is that it continues to hold that @xmath65 for the case that @xmath5 is semi - simple .
let @xmath66 be a finite family of paths which satisfies the properties @xmath39 and @xmath41 above .
then the paths are holonomically independent .
* proof :* + for @xmath67 the assertion is obvious . for @xmath68
we need to notice that since @xmath69 , there is a point @xmath70 such that a neighborhood @xmath71 of @xmath38 does not intersect @xmath72 and a similar point @xmath73 and a neighborhood @xmath74 .
the neighborhoods may be chosen not to intersect each other .
therefore , given any @xmath75 one can easily construct a single connection @xmath15 such that @xmath76 and @xmath77 .
suppose now that the theorem is true for every @xmath78 and consider the case when @xmath79 .
denote by @xmath80 the lie sub - algebra of the lie algebra of @xmath43 corresponding to the subgroup @xmath81 obtained as the image of the map [ projl ] ( u_e_1(a), ... ,u_e_n(a))g^n .
it suffices to prove that @xmath82 coincides with the whole lie algebra of @xmath43 .
since , according to our assumption , the theorem holds for @xmath83 for any edge @xmath24 , it is enough to show that for every @xmath84 in the lie algebra of @xmath5 the @xmath23-tuple @xmath85 is in @xmath80 since @xmath80 is a vector space ( owing to baez and sawin ) . since @xmath5 is semi - simple , for any @xmath86
there are @xmath87 such that @xmath88=a$ ] . by induction assumption
, we can freely specify @xmath23 of the entries @xmath89 of an element @xmath90 of the lie algebra @xmath80 , however , the last entry may depend on the other @xmath23 ones .
let us choose freely @xmath91 and @xmath92 . then there exist certain @xmath93 which may depend on the already specified @xmath89 and @xmath94 respectively such that ( b,,0,
,0 ) , ( b,0,,0, ... ,0)_w . because @xmath80 is closed with respect to the commutator ( again owing to baez and sawin ) , we have = ( a,0, ... ,0)_w , which completes the proof
. + @xmath95
in this section we list the immediate consequences of theorem [ th1 ] concerning the issue of ( mandelstam ) group identities which are an important building block in the construction of the ashtekar - isham algebra @xcite , the spectrum of that algebra and a natural uniform measure thereon .
the issue is the following : + suppose that @xmath96 complex valued functions @xmath97 defined on @xmath43 satisfy a group identity [ i d ] p(_1, ...
,_k ) = 0 ( g_1, .. ,g_n)g^n , where @xmath98 is a complex valued function of @xmath96 complex variables which is characteristic for the group @xmath5 . examples of such identities are the famous mandelstam identities which for the group @xmath99 are given by [ su(2 ) ] ( g)(g)=(g g)+(g ( g)^-1 ) .
now take any @xmath23-tuple of paths @xmath9 in @xmath7 and consider the following cylindrical functions defined on @xmath100 _
i(a ) = _ i(u_p_1(a), ... ,u_p_n(a ) ) , a , i=1, ... ,k .
obviously these cylindrical functions satisfy then the identity p(_1, ...
,_k ) = 0 a. thus , we see that every group identity gives rise to _ one _ identity on the algebra @xmath4 .
+ the question that arises is whether _ every _ identity that holds on the algebra @xmath4 ( or one of its sub - algebras ) declines from a group identity .
owing to the master theorem , the answer turns out to be affirmative ( one could not prove this result without knowing theorem [ th1 ] ) . indeed , consider any system of cylindrical functions @xmath101 satisfying some identity p(_1 ,
... , _ k ) = 0 .
any of the cylindrical functions is defined by ( [ cyl ] ) with respect to some finite set of paths in @xmath7 .
there exists a web @xmath102 whose range contains all the paths used to define the functions @xmath103 .
thus @xmath104 for some functions @xmath105 defined on the image @xmath106 of ( [ projl ] ) .
consequently , the functions @xmath105 satisfy the corresponding identity ( [ i d ] ) on all of @xmath64 . but
according to theorem [ th1 ] we have @xmath65 where @xmath23 is the number of the edges of @xmath19 .
thus the identity on cylindrical functions came from a group identity .
notice that , in particular , the path identities on cylindrical functions u_p(a)u_q(a ) = u_pq(a ) , , u_pp^-1=1_g for any two paths @xmath107 and the mandelstam identities , which for @xmath108 take the form u _
+ u_^-1 = u _ u _ , decline from a group identity .
+ a related question is : when are two paths @xmath14 and @xmath53 holonomically equivalent .
that is , for which paths @xmath14 and
@xmath53 does the identity u_p(a ) = u_q(a ) hold on @xmath109 ( i.e. for all @xmath60 ) ?
let paths @xmath14 and @xmath53 be holonomically equivalent .
choose a web @xmath110 such that @xmath111 .
decompose @xmath107 into the path products of the elements of @xmath19 and their inverses : p = e_i(1)^(1) ... e_i(n)^(n ) + q = e_j(1)^(1) ... e_j(m)^(m ) .
suppose , that in this decomposition , all the sequences of the form @xmath112 have been cancelled .
the properties of the parallel transport , the holonomic equivalence of the paths gives an identity g_i(1)^(1 ) ...
g_i(n)^(n ) = g_j(1)^(1 ) ...
g_j(m)^(m ) true for _ every _ @xmath113
. however , precisely because theorem [ th1 ] holds , @xmath65 which enables us , at least when @xmath5 is non - abelian and compact , to to conclude that modulo the cancellation of pairs @xmath114 , n = m , i(r ) = j(r ) , ( r ) = (r ) . that is _ for a non - abelian compact lie group , paths @xmath14 and @xmath53 are holonomically equivalent iff @xmath115 modulo cancellation of the segments of the form @xmath116 .
_ ( we are here relying on a related fact from the theory of compact non - abelian groups @xcite ) . notice that we could not conclude this if @xmath64 was a proper sub lie group of @xmath43 . the main motivation for raising the question about the precise image of the map ( [ proj ] ) was the wish to give a characterization of the ashtekar - isham quantum configuration space which in the analytical category turned out to be rather straightforward .
recall , that when @xmath5 is compact , we define in the algebra of cylindrical functions @xmath4 a norm : = _ a|(a)| .
the completion of @xmath4 with respect to that norm is a @xmath36 algebra .
it s gelfand spectrum @xmath117 has been promoted by ashtekar and isham to the role of the quantum configuration space for a connection theory . according to the results of @xcite an element @xmath118 of the spectrum of @xmath4
can be naturally identified with a family of points @xmath119 which satisfies certain consistency condition .
namely , given two webs @xmath45 , the projections @xmath120 and @xmath121 defined in ( [ proj ] ) determine uniquely a projection p_ww : _ w _ w , s.t .
p_wwp_w = p_w .
the consistency condition satisfied by @xmath119 is that for every pair of webs @xmath45 , [ con ] p_ww |a_w = |a_w .
more precisely , there is a 1-to-1 correspondence between the set of solutions of the consistency condition and the spectrum . to solve the consistency conditions one considers a map @xmath122 which assigns to every path @xmath14 an element of @xmath5 , pu(|a)_pg , such that u(|a)_pq = u(|a)_p u(|a)_q , and u(|a)_p^-1 = ( u(|a)_p)^-1 . indeed , a map @xmath122 defines a family @xmath119 by _
w : = ( u(|a)_p_1, ... ,u(|a)_p_n)g^n=_w where @xmath110 varies , the points @xmath123 fill all of @xmath43 . due to
@xmath124 , every @xmath125 defines a point in @xmath25 . ] .
it is not hard to see , that the result solves the consistency conditions and that every solution can be obtained in that way .
the algebra @xmath4 contains a sub - algebra @xmath126 of gauge invariant cylindrical functions .
also the results of @xcite apply , with the graphs replaced by webs . in particular
, the spectrum @xmath127 is = / , where @xmath128 is the group of _ all _ the maps from @xmath7 into @xmath5 , the gelfand completion of the gauge group acting in @xmath100 .
a measure @xmath129 on the spectrum @xmath100 is defined by any family of measures @xmath130 such that given a cylindrical function @xmath131 in the form ( [ cyl ] ) , the integral d_w = : _ d_0 is independent of the choice of the web @xmath19 . by the same arguments as those used in @xcite , if @xmath5 is compact and we choose for each @xmath132 the probability haar measure , the resulting family of measures does satisfy the consistency condition and defines a natural , diffeomorphism invariant measure on @xmath100 .
the resulting integral restricted to the sub - algebra of gauge invariant elements of @xmath4 coincides with the natural measure defined in @xcite .
this result was derived earlier by baez and sawin and our theorem 2.1 mainly simplifies the argument .
with the natural measure , the space of the cylindrical functions is naturally completed to become the hilbert space : = l^2(,d_0 ) .
the aim of this section is to find a generalization of the spin - network states @xcite constructed from graphs .
the spin - networks gave an orthogonal decomposition of the hilbert completion of the subspace of @xmath133 given by the piecewise analytic paths and were used to define the averaging with respect to all the diffeomorphisms of @xmath11 .
given a web @xmath19 we associate a hilbert subspace @xmath134 , the hilbert completion of @xmath21 . that subspace is isometric with @xmath135 .
that identification provides an orthogonal decomposition given by @xmath136 where @xmath137 are subspaces labeled by the irreducible representations of @xmath5 . in this way , to each _ labelled web _
@xmath138 , that is , a pair consisting of a web @xmath19 and a labelling @xmath139 of the edges of @xmath19 by irreducible representations , we associate a finite dimensional subspace @xmath140 called a _ spin - web space_. of a web @xmath19 and an irreducible representation @xmath141 one associates the linear span @xmath142 of the the functions @xmath143 , where @xmath144 stands for an @xmath145 entry of the matrix corresponding to an element @xmath146 in the @xmath147 representation and with respect to some fixed basis in the fundamental representation space . then , @xmath148 , @xmath14 running through the edges of @xmath19 . ]
the spin - web subspaces @xmath149 of @xmath134 corresponding to all possible labellings @xmath139 span @xmath134 , are finite dimensional and for different labellings they are orthogonal .
if @xmath150 for an edge @xmath14 of a labelled web @xmath138 , then we can consider another labelled spin - web @xmath151 given by removing the edge @xmath14 from @xmath19 and maintaining the labelling of the remaining edges . obviously , the corresponding spin - web spaces are equal . in the case of graphs , that degeneracy can be removed by admitting only nontrivial representations for the labelling of its edges and vertices .
however , in the case of the proper webs , that is not enough and still there are spin - web spaces associated to different webs @xmath152 such that @xmath153 fails to be orthogonal to @xmath154 .
since that is an important difficulty in diffeomorphism averaging of the spin - webs , we illustrate it by the following example .
+ * example .
* the first web is @xmath155 .
this is just one tassel of a base point @xmath14 .
the second web consists of the deformed edges .
it is @xmath156 .
( see fig.1 ) .
for the labellings of @xmath19 and @xmath32 respectively we take @xmath157 .
our claim is that the corresponding spin - web spaces @xmath158 and @xmath159 are not orthogonal . to see this ,
consider a web @xmath160 obtained from @xmath19 and @xmath32 by subdividing their edges , such that the segments @xmath161 and @xmath162 become path products of entire edges of @xmath160 ( see fig.2 ) .
the spin - web space @xmath158 is not any longer a spin - web space with respect to the web @xmath160 .
it admits the non - vanishing projection on the spin - web space given by the labelling @xmath163 of @xmath160 which takes the value @xmath164 on the segments of @xmath161 and @xmath162 , and @xmath165 otherwise . the same is true for the spin - web @xmath159 .
the lesson the above example teaches us is that a smooth deformation of a segment @xmath161 in @xmath138 such that it does not overlap the remaining part of @xmath19 any longer does not necessarily result in a labelled spin - web @xmath151 such that @xmath159 and @xmath158 are orthogonal to each other . this would not be true if we were working in the analytical category and with spin - networks @xcite .
we could also see how two different webs define a refined , bigger one .
our goal now is to show that there exist certain non - degenerate " spaces @xmath158 with the property that they are orthogonal to every other space @xmath159 whenever the ranges of the webs do not coincide , whether @xmath151 is also non - degenerate or not .
[ def1 ] consider a web @xmath19 and a labelling @xmath139 of its edges with irreducible and nontrivial representations .
consider a regular point @xmath38 ( see sec .
s1.2 ) of @xmath19 and let the type of @xmath38 be such that the edges @xmath166 are overlapping at @xmath38 .
we will say that @xmath38 is a * non - degenerate point * whenever the trivial representation does not emerge in the decomposition of the tensor product of the representations @xmath167 .
given a labelled web @xmath138 , if every regular point is nondegenerate , then we will say that @xmath138 is * nondegenerate*. we then have the following result . [ th2 ]
suppose @xmath158 is the spin - web space assigned to a non - degenerate labelled web @xmath138 and the spin - web space @xmath159 assigned to @xmath151 is not orthogonal to @xmath158 ; assume also that @xmath168 for every @xmath169 ; then : @xmath0 the labelled spin - web @xmath151 is also non - degenerate ; @xmath1 the ranges of the spin - webs coincide , r(w ) = r(w ) ; @xmath2 there is a web @xmath170 which can be obtained from each of the webs @xmath31 by subdividing its edges .
proof of theorem [ th2 ] : + the key fact is the existence of a web @xmath171 such that ww , w. one has to remember ( see example of sec .
[ s3.1 ] ) , however , that for proper webs , that is for webs which are not just graphs , given the spin - web space @xmath158 and the bigger web @xmath171 , the space @xmath158 is not in general one of the spaces @xmath172 with some labelling @xmath173 .
all that we know is that in general v_w , j , v_w,j _ jv_w,j , where @xmath173 is some finite set of labellings . since the spin - web spaces @xmath158 and @xmath159 are not orthogonal to each other ,
there is a labelling @xmath174 of the web @xmath171 such that both the projections [ w ] v_w , j v_w,j_0 , and v_w,j v_w,j_0 are not trivial .
an easy observation is the following : [ la1 ] let @xmath138 and @xmath151 be as in the hypothesis above ; there is a web @xmath175 which can be obtained by sub - dividing the edges of @xmath32 .
proof of lemma [ la1 ] : + let @xmath171 be a web such that @xmath176 .
let @xmath177 be webs which consist of the segments of the edges of @xmath19 , and respectively , @xmath32 .
thus we can write , w_1 = \{e_1, ... ,e_k , a_1, ... ,a_n } , + w_2 = \{e_1, ... ,e_k ,
b_1, ... ,b_m } , + w = \{e_1, ... ,e_k , a_1, ... ,b_m , c_1 , ... } , that is the webs @xmath178 and @xmath179 share the edges @xmath180 , the edges @xmath181 denote those of the edges of @xmath178 which are not contained in @xmath179 , whereas @xmath182 denote the edges of @xmath179 not contained in @xmath178 .
the first observation is , that the web obtained from the third web above by removing the extra edges @xmath183 , that is , w_0 : = \{e_1, ...
,e_k , a_1, ... ,b_m } also satisfies @xmath175 .
it is constructed from segments of the edges of @xmath19 and @xmath32 .
consider now the labelling @xmath174 of ( [ w ] ) . since the first projection is not trivial and due to the non - degeneracy of @xmath138 we have [ j ] j_0(e_i ) , j_0(a_j ) = 0 , and j_0(b_i ) = 0 . on the other hand , because the second projection is non - trivial either , we have j_0(a_j ) = 0 .
therefore , the extra elements @xmath181 of the web @xmath178 do not exist , and we finally have [ ww ] w_1 = \{e_1, ... ,e_k } , + w_0 = w_2 = \{e_1, ... ,e_k , b_1, ... ,b_m}.
+ this completes lemma [ la1 ] .
+ @xmath95 + + > from now on , let w=w_0w , w be the web given by lemma [ la1 ] .
eventually , we will show that if spin - web spaces @xmath158 and @xmath159 are not orthogonal to each other , then the non - degeneracy of @xmath138 implies the non - degeneracy of @xmath151 . together with lemma [ la1 ] that will complete the proof .
but before that we need two intermediate lemmas .
consider two webs @xmath184 .
an edge @xmath185 is the path product of the elements of @xmath171 , e = e_k ... e_1 , where @xmath186 is the one that contains the base point of @xmath40 .
we will see below that given a labelling @xmath139 of the web @xmath19 , in the decomposition of the corresponding spin - web space @xmath158 into the spaces @xmath172 the labelling @xmath173 of the beginning segment @xmath187 coincides with the labelling @xmath139 of the corresponding edge @xmath40 , for every edge @xmath185 .
[ la2 ] let @xmath138 be a labelled web and @xmath188 ; consider an edge @xmath185 ; let @xmath186 be the segment of @xmath40 containing its base point ; then v_w , j _ j:j(e_1)=j(e)v_w,j .
proof of lemma [ la2 ] : + the spin - web space @xmath158 is spanned by the cylindrical functions of the form _
ew ^(j(e))_n(e)^m(e)(u_e ) .
each such function is expressed by the parallel transports along the elements of @xmath171 through the decomposition : [ prod ] ^(j(e))_n(e)^m(e)(u_(e ) ) = ( ^(j(e_k))(u_e_k ) ... ^(j(e_1)))(u_e_1))_n(e)^m(e ) where @xmath189 since @xmath190 is a representation and @xmath191 .
notice that any of the factors @xmath192 appears only once in the whole product ( [ prod ] ) . to see this
, we note that different edges @xmath193 and @xmath72 may share the same segment @xmath180 .
however , by the definition of a web @xmath19 ( @xcite , see also section [ s1.2 ] ) , a segment of an edge @xmath193 of @xmath19 connected to the base point of a tassel in @xmath19 can not be contained in a different edge @xmath72 of the same web .
+ @xmath95 [ la3 ]
suppose @xmath138 is a nondegenerate labelled web and @xmath151 is another labelled web such that the associated spin - web spaces @xmath158 and @xmath159 are not orthogonal ; then @xmath151 is also nondegenerate .
proof of lemma [ la3 ] : + consider a regular point @xmath194 of the web @xmath32 .
denote by @xmath195 the edges of @xmath32 which overlap at @xmath194 .
what we have to show is that the tensor product of the corresponding representations @xmath196 , ... , @xmath197 does not contain the trivial representation .
use again the web @xmath171 of lemma [ la1 ] and consider the labelling @xmath174 of ( [ w ] ) . let the sub - webs @xmath198 be defined as in ( refww ) .
one of the edges of @xmath171 is the segment of the edge @xmath199 which contains its base point of the corresponding tassel .
denote it by @xmath200 .
notice , that according to lemma [ la2 ] j_0(e ) = j(e_1 ) = 0 .
comparison with ( [ j ] ) shows , that @xmath200 is one of the edges @xmath180 , say @xmath201 , of the sub - web obtained by subdividing the edges of @xmath32 .
the same argument applies to each of the edges overlapping at @xmath194 .
denote the corresponding segments by @xmath202 .
since they are segments of edges @xmath203 of a single tassel and each of them contains the base point , there is a regular point @xmath204 of @xmath171 such the edges @xmath205 overlap at @xmath204 and @xmath204 is not intersected by any other of the edges of @xmath171 .
but the tensor product of the representation @xmath206 does not give the trivial one , because the labels @xmath207 come from the decomposition of the non - degenerate labelled web @xmath138 into the spin web spaces @xmath172 and the projection onto @xmath208 is non - trivial .
this concludes the proof of lemma [ la3 ] and the proof of theorem [ th2 ] .
+ @xmath95 + + what we learnt from this section is that despite of the difficulties shown in example of sec .
[ s3.1 ] there is a class of spin - webs which have the property , that a space deformation of a given spin - work results , generically , in an orthogonal spin - web space . in @xcite
the spin - networks of the analytic category can be employed to construct solutions to the diffeomorphism constraint , the topological dual of @xmath4 . ] .
the construction consists of averaging the state over the diffeomorphisms of @xmath11 . with the baez - sawin results @xcite the averaging
was easily extended to averaging of a spin - network state with respect to the smooth diffeomorphisms @xcite .
our goal now is to extend that construction to a spin - web state .
we need first some notation . given a labelled web @xmath138 the gauge group acts in @xmath158 at the end points @xmath209 of an edge @xmath14 through its action on the parallel transport along the edges , u_e g(y)^-1 u_p g(x ) where the @xmath5-valued function @xmath210 defines an element of the group of the gauge transformations .
the end points of the edges of a web will be called vertices in the sequel .
one can further decompose each @xmath158 into the irreducible representation @xmath211 , of that action , assigning to each vertex @xmath49 of the web @xmath19 an irreducible representation @xmath212 and an extra label @xmath213 , because several equivalent but mutually orthogonal representations can appear more then once ( see @xcite for the details about the gauge invariant spin - networks and @xcite ) .
the quadruple @xmath214 will be also called a _ labelled _ web . as in @xcite
an important notion will be the group of trivial action diffeomorphisms of a web whose vertices are labelled by irreducible representations .
we define it as follows .
given a web @xmath215 fix a labelling @xmath216 of its vertices by the irreducible representations .
the group @xmath217 is the subgroup of those diffeomorphisms of @xmath11 which act trivially in the space @xmath218 for every labelled web @xmath219 with the fixed web @xmath215 and the labelling @xmath216 , and with arbitrary labellings @xmath220 .
in fact , the group depends quite weakly on @xmath216 .
if the representation @xmath221 is non - trivial for every vertex @xmath49 , then the group @xmath217 consists of those diffeomorphisms which preserve every edge ( including its orientation ) of @xmath215 .
however , there are some more diffeomorphisms acting ` trivially ' from the above point of view . consider a 2-valent vertex @xmath222 of the web @xmath215 which is the intersection point of two edges
@xmath223 such that , modulo change of orientation , one edge is a smooth extension of the other .
suppose @xmath224 is the trivial representation .
then , every element of the associated space @xmath218 , as a function on @xmath6 , depends on the connection @xmath15 through the parallel transport @xmath225 .
therefore in this case , every diffeomorphism preserving the oriented non - parameterized curve @xmath226 is an element of @xmath217 .
the group of smooth diffeomorphisms of @xmath11 acts naturally in the space of the cylindrical functions @xmath4 . given a diffeomorphism @xmath227 and
@xmath228 denote the result of the action by @xmath229 .
consider a labelled web @xmath214 , and @xmath230 .
we attempt to define an averaged state @xmath231 by the following action on a cylindrical function @xmath232 : [ av ] |(f ) : = 1(w , j , l ) _ [ ] ( ) /(w , l ) ( |f ) , where @xmath233 is a constant .
the following two conditions should be satisfied by the averaging : @xmath0 finiteness ; given @xmath234 the result should be finite for a large class of @xmath232 ; @xmath1 consistency ; since the spin - web spaces are not orthogonal to each other , we have to ensure that given two different labelled webs @xmath138 and @xmath151 and a state @xmath235 our definition of the averaging gives the same result regardless of which of the above webs we use .
suppose @xmath236 is contained in the subspace @xmath159 associated with some labelled web @xmath151 and that @xmath168 for every @xmath169 .
now , owing to theorem [ th2 ] , the only nonzero terms in ( [ av ] ) come from the diffeomorphisms such that both , @xmath237 and the webs @xmath19 and @xmath238 satisfy the property @xmath2 of theorem [ th2 ] . define the _ symmetry group _ @xmath239 of a labelled web @xmath214 to be ( w , l ) = \{[]()/(w , l)| ( r(w))=r(w ) + and w:=(w ) satisfies ( iii ) of theorem [ th2]}. + what we need to assume is that @xmath239 is finite .
a general cylindrical function can be always written as an infinite sum , f = _ j a_j f_j , f_jv_w,j , with respect to the labellings @xmath240 of the edges of a single web @xmath32 . to see that also for a general cylindrical function @xmath236 the sum ( [ av ] )
is finite as long as the symmetry group @xmath239 is finite , consider @xmath241 $ ] in the sum ( [ av ] ) such that ( |f ) = 0 .
the projection of @xmath242 onto @xmath159 is not trivial only for finite number of the labellings @xmath240 .
( to see that , take @xmath243 , to represent @xmath236 ) .
turn now to the consistency condition @xmath1 .
it will constrain the freedom of the constants @xmath233 . consider a labelled web @xmath244 and a state [ id2 ] a_0|w_0,j_0,l_0,c_0,m_0 > = a_1 |w_1,j_1,l_1,c_1,m_1 > + ... + a_r|w_r , j_r , l_r , c_r , m_r>. without loss of generality , we assume that the identity can not be obtained as a sum of two non - trivial identities .
that is we may assume that @xmath245 on the left hand side is not orthogonal to any of the terms on the right hand side . therefore , the web @xmath19 is in the relation @xmath2 of theorem [ th2 ] with each of the webs @xmath246 on the right hand side .
moreover , if @xmath247 for any of the vertices @xmath248 of any of the webs above , then the point @xmath248 is a vertex of every of the webs in the identity , and l_1(v_s ) = ...
= l_r(v_s ) .
we may also write the identity in such a way , that the labelled webs @xmath249 are all different . to ensure that the averaging is consistent with the identity we observe the following : [ lb1 ]
suppose a labelled spin - web @xmath214 is non - degenerate . if @xmath250 is another labelled spin - web such that the corresponding spin - web spaces @xmath211 and @xmath251 are not orthogonal to each other , then the corresponding groups of the trivially acting diffeomorphisms coincide , ( w , l ) = ( w,l ) . proof of lemma [ lb1 ] : + we again take advantage of teorem [ th2 ] .
let @xmath171 be the web given by theorem [ th2 ] which satisfies @xmath2 of that theorem .
consider the following labelling @xmath252 of the vertices of @xmath171 with the irreducible representations .
for a vertex @xmath49 of @xmath171 set l(v ) = recall that every vertex @xmath49 of @xmath171 which is not a vertex of @xmath19 is obtained by subdividing an edge of @xmath19 .
it is easy to check , that the corresponding groups of trivially acting diffeomorphisms coincide , ( w , l ) = ( w,l ) .
but it is also true for @xmath253 and @xmath254 , which completes the proof of lemma [ lb1 ] .
+ @xmath95 + + the formula ( [ av ] ) for the averaging depends on the group @xmath255 of a given labelled web @xmath214 and on the unspecified constant @xmath233 . for every term in the identity @xmath256
the group of the trivially acting diffeomorphisms is the same , and equals @xmath217 .
therefore , to ensure the consistency , we have to take the constants @xmath233 depending only on @xmath255 , [ av2 ] ( w , j , l ) = f((w , l ) ) .
a quite natural choice is for example [ av3 ] ( w , j , l ) = |(w , l)| , where @xmath257 is the number of elements of the symmetry group of a labelled web .
let us summarize our results : denote by @xmath258 and @xmath259 the subspaces of @xmath4 spanned by the spin - web subspaces corresponding to , respectively , non - degenerate and degenerate labelled spin - webs .
next , let @xmath260 and @xmath261 be spanned by the spin - web spaces corresponding to nondegenerate labelled webs of , respectively , finite and infinite symmetry groups .
[ th3 ] the hilbert space @xmath133 of all the square integrable functions , can be orthogonally decomposed in the following way , [ decomp ] = = , where the over lining denotes the hilbert closure ; the diffeomorphism averaging given by ( [ av],[av2 ] ) defines a linear map _
nondegfin | , @xmath262 being the topological dual to the space @xmath4 .
the proof follows from the above arguments , theorem [ th2 ] and lemma [ la3 ] .
some remarks are in order about the subspaces @xmath261 and @xmath259 which are excluded from the domain of the averaging map .
the first subspace is given by non - degenerate spin - webs with infinite symmetry groups .
an example of such a spin - web was given in @xcite .
the spin - web considered there is degenerate in terms of the present paper , but it is not hard to modify it to a nondegenerate spin - web of the infinite symmetry group .
the idea of the construction of such an example is that a part of a web can be non trivially mapped into itself by a diffeomorphism which preserves the rest of the web .
if similar parts emerge infinitely many times ( in every neighborhood of a base of a tassel ) then we have infinitely many diffeomorphisms . which are symmetries of the web .
the second group of states is given by degenerate labelled webs @xmath138 .
an example of such a web is shown in fig .
1 . in all the examples of degenerate labelled webs we are aware of
, an associated cylindrical function can be decomposed into a converging infinite sum of cylindrical functions each being associated to a web obtained from the original one by removing a ( degenerate ) segment .
see section [ s4.3 ] .
> from the point of view of a diffeomorphism invariant theory of connections , the averaged states are solutions of the diffeomorphism constraint . the action of the hamiltonian constraint operator of the gravitational field can be extended to those states .
we go back to that issue in sec .
in the analytic category @xcite , the hilbert space of cylindrical functions is orthogonally decomposed into subspaces labeled by graphs in such a way , that every diffeomorphism invariant operator preserves the subspaces . the aim of this subsection is to show , that in the smooth case , certain subspaces are also necessarily preserved by diffeomorphism invariant operators .
however , the result that we derive below is not as strong as in the analytic category . given a web @xmath19 denote by @xmath263 the hilbert spaces obtained by using all the cylindrical functions which are associated to the webs whose ranges are contained in the range of @xmath19 , _
r(w ) : = .
[ th4 ] suppose @xmath264 is a diffeomorphism invariant operator defined in a domain @xmath265 in the hilbert space @xmath266 ; then , @xmath39 for every web @xmath19 , the hilbert subspace @xmath263 associated to its range is preserved by the operator @xmath264 , @xmath41 in particular , when a web is a graph , then the space , @xmath139 running over over all the labellings of the edges of @xmath19 by the irreducible ( possibly trivial ) representations , is preserved by the operator @xmath264 , @xmath44 suppose , that for every labelled web @xmath151 , it is true that , @xmath267 ; let @xmath268 and @xmath269 where @xmath138 are labelled webs ; then , ( | ) = 0 , r(w)=r(w ) , provided the labellings @xmath270 do not assign the trivial representation to any of the edges of the webs @xmath19 and , respectively , @xmath32 ; in particular , if the web @xmath19 is a graph , then the operator preserves the space _
jv_w , j , @xmath139 running through all the labellings of the edges of @xmath19 by irreducible and non - trivial representations .
proof : + consider a web @xmath19 and a state @xmath271 . for the proof of @xmath39 ,
it is enough to show , that for every labelled web @xmath151 such that @xmath243 containing an edge @xmath169 such that the range @xmath272 is not contained in @xmath273 j(e ) = 0 v_w,j. ( notice , that the subspaces @xmath159 span a dense subspace in the whole of @xmath133 , even when we restrict ourselves , to @xmath274 a given @xmath19 because we allow @xmath240 to contain trivial representations . )
let @xmath275 be an edge such that the range @xmath276 is not contained in @xmath273 . in @xmath277 , there is a regular point @xmath278 of the web @xmath32 which is not contained in @xmath273 . according to the definition of a tassel , the edge @xmath277 contains a sequence of the regular points @xmath279 such that each of the points has the same type as @xmath278 , and the sequence converges to the beginning point of @xmath277 , the base point of the corresponding tassel . for each of the points @xmath280 take an open neighborhood @xmath281 diffeomorphic to a ball , such that the intersection @xmath282 is an embedded open interval , say @xmath283 .
consider a diffeomorphism @xmath284 such that @xmath285 @xmath227 preserves each of the @xmath286 and acts trivially outside of @xmath287 , @xmath288 @xmath289 , for every point of the sequence , @xmath290 @xmath227 deforms each segment @xmath283 in the same way , modulo diffeomorphisms ; that is , for every pair of the segments @xmath283 and @xmath291 , the pair @xmath292 is diffeomorphic to @xmath293 .
let @xmath294 be the edges of @xmath32 which overlap the edge @xmath277 at @xmath295 .
hence , they overlap it at each of the points @xmath296 .
it follows from ( @xmath297 ) above , that the family of edges obtained from @xmath32 and its deformation , namely w : = w(w ) = \{(e_0),(e_1 ) , ... , ( e_m)}w , is also a web .
now , given a labelling @xmath240 of @xmath32 , we can trivially extend it to a labelling @xmath173 of @xmath171 by setting j((e_0 ) ) = ... j((e_m))= 0 .
on the other hand , the diffeomorphism @xmath227 carries @xmath240 into another labelling @xmath298 of the web @xmath171 given by @xmath299 for every edge of @xmath32 .
obviously we then have [ nonzero ] j(e_0 ) = 0 ( j ) = j . therefore ,
if @xmath240 satisfies ( [ nonzero ] ) then for every @xmath269 , the functions @xmath300 and @xmath301 are orthogonal to each other .
but , the diffeomorphism acts trivially on @xmath234 .
therefore from the diffeomorphism invariance of @xmath264 , the projection of @xmath302 on @xmath300 is @xmath227-invariant , ( | ) = ( ^-1 | ) = ( ^-1 | ) = ( | ) .
moreover , it is easy to iterate the above construction to obtain an infinite family of diffeomorphisms each of which satisfies the conditions @xmath303 above and such that for every pair of this set of diffeomorphisms ,
= ( _w,p | _ w,p ) = 0 . if @xmath227 varies through all possible diffeomorphisms like this , the set of the resulting states @xmath304 has an infinite number of mutually orthogonal elements . since a densely defined operator can not have an infinite number of non - vanishing matrix elements of the same value it follows that ( | _ w,j ) = 0 .
this is sufficient to complete the proof of part @xmath39 of the statement .
+ if @xmath19 is a graph , then @xmath305 implies that @xmath32 is also a graph .
let @xmath306 be a vertex of @xmath32 which is not a vertex of the graph @xmath19 .
therefore , it is contained in the interior of an edge @xmath185 .
for every labelling @xmath250 of the graph @xmath32 such that @xmath307 , and for every spin - web state @xmath308 , we prove that ( | ) = 0 , in the same way as before .
we construct an infinite class of the states diffeomorphic to @xmath300 and orthogonal to each other .
the diffeomorphisms we use move the vertex @xmath306 along the edge @xmath40 but preserve the edges of the web @xmath19 , hence they also preserve the function @xmath234 .
this completes the proof of the part @xmath41 .
the point @xmath44 follows easily from @xmath309 and their applications to the adjoint operator @xmath310 .
this completes the proof of theorem [ th4 ] .
+ @xmath95 in the previous section we have spited the hilbert space into the orthogonal sum of degenerate and nondegenerate spin - web states and the last sector has again been orthogonally decomposed with respect to possible symmetry groups of the labelled spin - webs .
the diffeomorphism averaging is well defined only in one of those sectors , namely in the subspace @xmath260 spanned by all the spin - web states of a non - degenerate labelled webs who have at most finite symmetry groups .
for this reason it is natural to ask if a diffeomorphism invariant operator is necessarily consistent with that decomposition and whether it preserves the domain of the diffeomorphism averaging .
it is easy to see , that owing to theorem [ th4].@xmath213 , a symmetric diffeomorphism invariant operator preserves the finiteness of the symmetry groups . regarding the non - degeneracy
, we will show below , that the image of a non - degenerate spin - web state under the action of a diffeomorphism invariant operator is orthogonal to what we will define as ` strongly degenerate ' states .
we will also indicate a large class of webs for which the degeneracy is equivalent to the ` strong degeneracy ' , and we will conjecture , that the equivalence holds for all the spin - webs
. given an edge @xmath40 let _ e : be the projection onto the subspace given by the cylindrical functions independent of the holonomy along @xmath40 .
explicitely , the projection can be defined as follows .
given a cylindrical function @xmath311 take a web @xmath312 such that @xmath313 .
let @xmath314 be all those elements of @xmath19 which are the segments of @xmath40 .
we will use the orthogonal projection , _ e , w : _ w _ j : j(e_1)= ... =j(e_k)=0v_w , j , to define _ e = _ e , w .
a labelled web @xmath138 is _ strongly degenerate _ if there is an infinite sequence of disjoint regular segments @xmath315 such that for every @xmath316 [ strdeg ] _
(1-_s_l ) = 0 , in the hilbert space topology . indeed
, the strong degeneracy implies the week degeneracy .
a labelled web @xmath138 is either degenerate or non - degenerate .
if it is non - degenerate , then , for every regular segment @xmath161 _ s(v_w , j ) = 0 , and the left hand side of ( [ strdeg ] ) is @xmath234 itself for every @xmath316 .
[ th5 ] suppose @xmath264 is a diffeomorphism invariant operator in @xmath133 such that the domain @xmath317 contains all the spin - web subspaces @xmath159 .
then , for every non - degenerate spin - web state @xmath234 and every strongly degenerate labelled spin - web state @xmath300 , ( | ) = 0 .
proof : + it follows from the non - degeneracy of @xmath234 and from theorem [ th2 ] , that any diffeomorphism @xmath227 which deforms only the segment @xmath318 and acts trivially in @xmath319 satisfies ( | ^_s_1 ) = ( |^_s_1 ) , ( | ) = 0 .. so , there exists an infinite family of states @xmath320 , such that ( _ k_1| ^_s_1 ) = ( _ k_2| ^_s_1 ) , + ( _ k_1|_k_2 ) = 0
. it follows , that ( |_s_1 ) = 0 . that shows , that ( | ) = ( | ( 1-_s_1 ) ) .
next , using the same argument for the sequences @xmath321 we prove , that for arbitrary @xmath96 ( | ) = ( |(1-_s_k) ... (1-_s_1 ) ) . taking limit @xmath322 and using the strong degeneracy of @xmath300 completes the proof of theorem [ th5 ] .
+ @xmath95 + + our conjecture is : * conjecture . *
_ every degenerate labelled web is strongly degenerate .
_ so if the conjecture is true , then every symmetric , diffeomorphism invariant operator preserves the subspace @xmath323 in which the diffeomorphism averaging has been well defined
. we will advocate the conjecture in the next subsection .
there is a large class of webs , for which we can prove that the degeneracy of a labelling does imply the strong degeneracy .
consider throughout this section the webs for which the following is true : for every tassel @xmath324 of a given web @xmath19 and for a parameterization of the edges of @xmath50 given by the point @xmath1 of the definition of tassel ( say , the parameterization is defined on @xmath37 $ ] ) for every @xmath325 there is @xmath326 such that the segments @xmath327 of all the edges @xmath328 are holonomically indepenent ( the above property is satisfied for the web in example given in sec [ s3.1 ] . in fact , we do not know of any counter example , any web that would fail to satisfy that property . ) now , consider a degenerate labeling @xmath138 of a web @xmath19 and a state @xmath316 .
we will calculate explicitely the action of the projection operator @xmath329 for appropriately selected segments .
let @xmath161 be a regular segment , such that the tensor product @xmath330 contains the trivial representation in the decomposition into the irreducible representations , where @xmath331 are the edges of @xmath19 that overlap @xmath161 .
the degeneracy guarantees the existence of such a segment . then , owing to the properties of a tassel , there is a segment @xmath332 of the same type as @xmath161 which is close enough to the base point of the tassel for the following to be true : if we write each of the edges overlapping at @xmath161 as e_i = e_i+se_i- , i=1, ... ,k , then the segments @xmath333 are holonomically independent ( the beginning segments @xmath334 are holonomically independent for any choice of @xmath161 by the very construction of a tassel . )
we will show , that [ c ] ( 1-_s ) = c(j(e_1), ... ,j(e_k ) ) .
the main point of this result is that the coefficient @xmath213 depends only on the labels assigned by the labelling @xmath139 to the edges .
of course the coefficient is less then one , because the operator is a projection and the trivial representation does appear in the decomposition above .
the new function @xmath335 can be written as ( 1-_s ) = _ n_n , where each @xmath336 is again a spin - web state associated to a labelled web @xmath337 which contains a segment @xmath318 of the same properties as the segment @xmath332 , namely @xmath0 @xmath318 is overlapped by @xmath96 edges , @xmath338 ; @xmath1 the segments @xmath203 overlapping @xmath318 are labelled in the same way as the edges @xmath57 , j_n(e_i ) = j(e_i ) , i=1, ... ,k ; @xmath2 the new segments @xmath339 are holonomically independent ( where @xmath340 ) . therefore we can use the result again to conclude that ( 1-_s_1 ) ( 1-_s_0 ) = c(j(e_1),
... ,j(e_k))^2 . finally , repeating the construction we find an infinite sequence of segments @xmath341 such that ( 1-_s_n)
(1-_s_0 ) = c(j(e_1), ... ,j(e_k))^n+1 .
the sequence converges to @xmath164 as @xmath342 hence indeed @xmath234 is strongly degenerate .
let us show now , that equation ( [ c ] ) is true .
the function @xmath234 can be written as = l_a_1 ...
a_k ^(j(e_1))^a_1_b_1(u_e_1 ) ... ^(j(e_k))^a_k_b_k(u_e_k)r^b_1 ...
b_k , where @xmath343 do not involve any of the holonomies @xmath344 . then , the action of the projection operator @xmath329 on @xmath234 is _
s = l_a_1 ... a_k^(j(e_1))^a_1_c_1(u_e_1
+ ) ... ^(j(e_k))^a_k_c_k(u_e_k+)c^c_1 ...
c_k + c_d_1 ... d_k ^(j(e_1))^d_1_b_1(u_e_1- ) ... ^(j(e_k-))^d_k_b_k(u_e_k)r^b_1 ... b_k , where c^c_1 ...
c_kc_d_1 ... d_k = du ^(j(e_1))_d_1^c_1(u ) ...
^(j(e_k))_d_k^c_k(u ) .
the comparison of the norms gives the following result , _
s^2 = c^c_1 ... c_k _ c_1c_1 ... _c_kc_kc_d_1 ... d_k ^d_1d_1 ... ^d_kd_k^2 . from the orthogonality of @xmath345 and @xmath335 we conclude ( [ c ] ) .
it is easy to see , that as mentioned , the web of example in sec . [ s3.1 ] belongs to the class of webs considered in this section .
indeed , each of the two edges contains a segment which is not overlapped by the other edge .
interestingly , even if we use the range of the web to introduce more edges such that each regular segment is overlapped by more than one edge , in all the examples we considered we found only webs belonging to the same class .
it is likely that for a semi - simple group this class contains all the webs .
what we have learnt from this section is that , as in the case of the analytic category considered in @xcite , the algebra of the observables strongly commuting with the diffeomorphisms is reducible to the sectors labelled by the ranges of the webs . in the case
when a web @xmath19 is a graph , the corresponding sector is @xmath134 .
in this section for the group @xmath5 we take su(2 ) . for the quantization of gravity
there are two kinds of operators that play a key role .
the first kind consists of the so - called geometrical operators @xcite corresponding respectively to lengths , areas and volumes of curves , areas and volumes .
the second kind consists of constraint operators @xcite , most importantly the hamiltonian constraint .
these have been defined in the analytical category so far .
we will now extend them to the smooth category .
the regularizations of the operators representing geometric functionals can be easily modified to the case of the smooth curves .
a problem one encounters then is that when a functional depending on @xmath346 is given by an integral over a surface or a region in @xmath11 the resulting operator often is not well defined in any dense domain .
this means that a given surface or region is just infinite in certain states of the system . to begin with ,
consider a functional associated with an oriented 2-surface @xmath347 ( see @xcite for the details ) e^i_s = _ s 12e^ia_abc dx^adx^b , and the corresponding quantum operator ^i_s = _ s12i a^i_a_abc dx^adx^b .
let @xmath348 where @xmath19 is a web . `
generically ' , every edge @xmath14 of @xmath19 has only finite number of isolated intersections with @xmath347 and only those contribute to its action on @xmath131 .
generally , however , that number is infinite .
every point @xmath38 of @xmath40 which is not contained in @xmath347 defines a segment @xmath161 of @xmath40 whose interior does not intersect @xmath347 whereas its ends lie on @xmath347 or coincide with the ends of @xmath40 .
however , for each edge @xmath40 the number of such segments is at most countable . split each such segment into two pieces , oriented in such a way that they are outgoing from the surface and
denote the resulting segments by @xmath349 . repeating the derivation of the first reference of
@xcite we see that for a cylindrical function @xmath131 compatible with @xmath19 , the only contribution comes from those end points @xmath350 of the segments @xmath351 which are contained in @xmath347 , according to the formula ^i_s
= 12_s_e , i _
s(s_e , i)j^i_x_e , i , s_e , i , where @xmath352 is @xmath353 if the segment intersects @xmath347 and goes up / down ( the above notation excludes segments which overlap @xmath347 and contribute zero ) .
the operators @xmath354 were already defined in @xcite where it is understood that we replace graphs by webs . in the general case , for infinitely many intersections , ^i_s = .
therefore , the domain of the operator is defined by the generic webs which intersect the surface @xmath347 finitely many times . using the same notation , for the area of @xmath347
we obtain = ^22_x_e , i , where @xmath350 runs through all the intersection points of the segments @xmath351 with @xmath347 and @xmath161 and @xmath162 run through all the segments @xmath355 which intersect the same @xmath350 . but
again , the operator may take the infinity .
for instance , on a wilson loop function ( a ) = u_(a ) given by a non - self intersecting loop @xmath356 which defines infinitely many , transversally intersecting the surface @xmath347 segments @xmath351 defined above , we have = . in the case of the volume of a region @xmath357 in @xmath11 , the ` external ' regularization of rovelli , smolin and depietri @xcite
does not appear to be easily applicable to the smooth curves case .
on the other hand , the regularization of the second reference of @xcite proposed for the volume operator can be extended to that case .
then , the contribution is given by all the points of intersection of three edges of @xmath19 such that the directions tangent to the edges are linearly independent and the intersection point is contained in @xmath357 .
the number of such generic intersection points is at most countable .
the generic intersection points are isolated from each other , so their set is countable ; the number of triples is finite , determined by the number of edges of the web . ]
the action of the resulting volume operator on a cylindrical function compatible with the web @xmath19 reads [ volgamma ] _
r & = & _ o _ v + _ v & = & 148_ijk_s , s,s(s , s,s ) j^i_v , sj^j_v , sj^k_v , s . where @xmath49 runs through all the intersection points between the edges of @xmath19 and each of @xmath358 runs through all the segments intersecting at and
are outgoing from @xmath49 ; the number @xmath359 equals @xmath360 depending on the orientation of the vectors tangent to @xmath358 at @xmath49 . + + similar statements
apply to the length operator @xmath361 @xcite for any piecewise smooth curve @xmath213 since it uses the volume operator as a basic building block .
+ + summarizing , for all three operators considered above , the infinity arrises as a sum of a countable number of finite terms . therefore ,
if we exclude most of those points from the curve @xmath213 , the surface @xmath347 or the region @xmath357 respectively then we will obtain a finite length , area or volume .
the process of gluing back the removed points one after the other can be interpreted as extending the curve , surface or region to infinity . in section [ s3.2 ]
we have defined new solutions of the diffeomorphism constraint .
they were , however , so far not shown to be in the domain of the dual of the loop hamiltonian constraint , as it was defined in @xcite .
we observe below , that in fact the action of the dual @xmath362 of the hamiltonian constraint can be naturally defined on them and that it is free of the infinities emerging in the case of the above 3-geometry operators .
one can easily repeat the regularization of @xcite of the hamiltonian constraint , with appropriate modifications , to the case of a web with an infinite number of intersection points among its defining curves .
recall , that in the hilbert space @xmath133 of the cylindrical functions , one first defines a regulator depending hamiltonian operator @xmath363 where @xmath364 denotes the regulator .
its domain is not dense in the smooth category any longer , in contrast to what happens in the analytical category .
the reason for this is the same as in the case of the volume operator .
explicitly , the regulated hamiltonian operator acts on a spin - web state @xmath365 in the following way [ reg ] _
| = _ vn(v)_(v ) all the intersection points of the edges in a web and for each @xmath49 the operator @xmath366 is well defined for all the spin - web states .
the number of the intersection points is in general infinite , hence the total operator ( [ reg ] ) is not well defined on such states .
( the operator @xmath366 involves the volume operator in such a way that as in the volume operator case , only the triples of generically intersecting segments contribute ) .
remarkably however , the dual operator @xmath367 is well defined for any dual spin - web state , that is if @xmath316 and @xmath368 for labelled webs @xmath138 and @xmath369 , then @xmath370 is finite .
finally , the regulator free dual hamiltonian operator @xmath362 is well defined on the diffeomorphism averaged dual spin web states @xmath371 provided that we restrict ourselves to the space @xmath260 on which the diffeomorphism averaging has been defined in sec .
[ s3.2 ] . more specifically , whenever @xmath372 then for any lapse function the functional @xmath373 is a well defined linear functional on the entire space @xmath260 . + acknowledgements .
the authors thank abhay ashtekar , john baez and steve sawin for helpful discussions and correspondence .
j.l . was supported by alexander von humboldt - stiftung and the polish committee on scientific research ( kbn , grant no .
2 p03b 017 12 ) .
j. baez , lett .
* 31 * ( 1994 ) 213 - 223 + j. baez , diffeomorphism invariant generalized measures on the space of connections modulo gauge transformations " , hep - th/9305045 , in the proceedings of the conference on quantum topology , d. yetter ( ed ) ( world scientific , singapore , 1994 ) .
a. ashtekar , j. lewandowski , d. marolf , j. mouro and t. thiemann , `` a manifestly gauge invariant approach to quantum theories of gauge fields '' , in _ geometry of constrained dynamical systems _ , j. charap ( ed ) ( cambridge university press , cambridge , 1994 ) ; `` constructive quantum gauge field theory in two space - time dimensions '' ( cgpg preprint ) . c. rovelli , l. smolin , phys
* d53 * ( 1995 ) 5743 j. baez , spin network states in gauge theory " , adv . math .
* 117 * ( 1996 ) 253 + j. baez , `` spin networks in non - perturbative quantum gravity '' , in the interface of knots and physics " , ed .
louis kauffman , american mathematical society , providence , rhode island , 1996 j. baez , s. sawin , jour .
analysis * 150 * ( 1997 ) , 1 c. rovelli , l. smolin , `` discreteness of volume and area in quantum gravity '' nucl .
b * 442 * ( 1995 ) 593 , erratum : nucl .
b * 456 * ( 1995 ) 734 r. de pietri , c. rovelli , phys.rev .
d * 54 * ( 1996 ) 2664 | in the canonical quantization of gravity in terms of the ashtekar variables one uses paths in the 3-space to construct the quantum states . usually , one restricts oneself to families of paths admitting only finite number of isolated intersections .
this assumption implies a limitation on the diffeomorphisms invariance of the introduced structures . in this work , using the previous results of baez and sawin , we extend the existing results to a theory admitting all the possible piecewise smooth finite paths and loops . in particular , we @xmath0 characterize the spectrum of the ashtekar - isham configuration space , @xmath1 introduce spin - web states , a generalization of the spin - network states , @xmath2 extend the diffeomorphism averaging to the spin - web states and derive a large class of diffeomorphism invariant states and finally @xmath3 extend the 3-geometry operators and the hamiltonian operator .
epsf.tex = -1 cm h o
[ section ] [ section ] [ section ] |
the sharp increase in the mean - square charge radius is observed in the odd mass @xmath0hg isotopes with respect to their even neighbors @xcite . a hint that a very different configuration may constitute the ground state of these nuclei was given by an isomer shift measurement of @xmath1hg @xcite , where they found an excited state of @xmath1hg which has a charge radius consistent with the charge radius of heavier neighbors .
the sharp increase in the charge radius was interpreted as the transition from a nearly spherical to a deformed shape in the ground state of the odd mass @xmath0hg isotopes .
frauendorf and pashkevich @xcite calculated the deformation energy of hg isotopes by means of strutinsky s shell correction method .
they found that the light hg isotope has an oblate and prolate minimum of nearly the same depth at the respective deformations @xmath2 and 0.22 .
they discussed a mechanism which causes the transition from a small oblate to a large prolate deformation in the odd mass @xmath0hg isotopes .
they noticed that the neutron level density is very high in the almost spherical oblate minimum , whereas the neutron level density is low in the well deformed prolate minimum .
they concluded that the loss of pairing energy due to the blocking of the fermi level by the odd neutron is larger in the oblate minimum than in the prolate one , which gives rise to the oblate - prolate transition in the ground state of these isotopes .
as the prolate minimum corresponds to deformation twice as large as the oblate one , the sharp increase in the mean - square radius is observed in the odd - mass light hg isotopes .
there are several mean field calculations of hg isotopes .
however , all of them are limited to even isotopes @xcite . in order to see if the mechanism of frauendorf and pashkevich works well , we perform hartree - fock - bogoliubov ( hfb ) calculations of hg isotopes with effective interactions siii @xcite , ski3 @xcite and sly4 @xcite .
we calculate lipkin - nogami corrections @xcite to both mean field and pairing energies @xcite .
we investigate the influence of the time - odd mean - field terms as well as the blocking of a level by the odd neutron on the oblate - prolate shape transition . in section 2 ,
we present the skyrme hfb method with lipkin - nogami corrections to describe both even and odd hg isotopes . in section 3 , we present calculations with effective interactions siii , ski3 and sly4 .
conclusions are given in section 4 .
the hfb equation is derived by the variation of energy with respect to the quasi - particle vacuum @xcite , @xmath3 @xmath4 here the quasi - particle vacuum @xmath5 is given by @xmath6 @xmath7 for all quasi - particle states @xmath8 . in the present study
, we employ skyrme forces in the particle - hole channel and the zero - range density - dependent pairing force in the particle - particle channel , @xmath9 the hfb energy is then given by @xmath10 where @xmath11 is the skyrme energy density @xcite and @xmath12 is the gap field , @xmath13 the pairing strength @xmath14 is determined from the odd - even mass difference and the density parameter @xmath15 is taken as the saturation density of symmetric nuclear matter which is obtained by using the respective skyrme parameters .
the particle density @xmath16 and the anomalous density @xmath17 are written as @xmath18 @xmath19 with the density matrix and pairing tensor , @xmath20 @xmath21 variation of eq .
( [ ehfb ] ) with respect to @xmath22 and @xmath23 yields the hfb equation , @xmath24 diagonalization is carried out in a deformed oscillator basis with axial symmetry @xcite . a greek index in eq .
( [ hfbeq ] ) denotes a set of quantum numbers \{@xmath25 } of a deformed oscillator wave function . in the present study
we will restrict our basis space as @xmath26 with @xmath27 . the oscillator constant
is held fixed as @xmath28 .
the deformation parameter @xmath29 is varied to obtain the maximum deformation energy in the oblate and prolate minimum , respectively .
the chemical potential @xmath30 is adjusted by the particle number condition , @xmath31 . using the occupation amplitude @xmath32 and quasi - particle energy @xmath33
, we calculate an auxiliary single - particle energy and pairing gap in order to find the optimum value of @xmath30 @xcite , @xmath34 @xmath35 as in ref .
@xcite , the cut - off procedure of quasi - particle states is imposed on the auxiliary single - particle energy .
the value of @xmath36 mev is assumed in the present study .
the above formalism can be straightforwardly extended to the case of one quasi - particle state in odd mass nuclei .
all the equation retain their form . only the density matrix @xmath37 and pairing tensor @xmath38
should be modified due to the blocked quasi - particle @xmath39 in the following way @xcite ; @xmath40 we introduce a reference state @xmath41 and its energy expectation value @xmath42
. the state is constructed as an _ even _ vacuum without quasi - particle creation and without breaking time - reversal invariance but with an _ odd _ average particle number .
the difference of energy @xmath43 is thus composed of the loss of pairing energy due to the blocking of the fermi level and the polarization energy of the mean field by the presence of an odd neutron @xcite .
a proper description of an odd nucleus by mean field theories requires to break the time - reversal symmetry @xcite . when the time - reversal symmetry is broken , the skyrme energy density @xmath11 is written by the sum of kinetic energy and potential energy of isoscalar @xmath44 and isovector @xmath45 terms ; @xmath46 .
\label{functional}\end{aligned}\ ] ] the coupling constants @xmath47 are expressed by the skyrme force parameters @xcite .
the isospin index @xmath48 can have values 0 or 1 .
the isoscalar and isovector parts of the particle densities @xmath49 are defined as @xmath50 and analogous expressions are used to define other densities .
it should be noted that since we consider the skyrme force as the density - dependent two - body interaction , the time - odd terms @xmath51 in eq .
( [ functional ] ) are different from the original expressions @xcite which were derived from the three - body interaction .
the characteristic combinations of @xmath52 , @xmath53 and @xmath54 $ ] in eq .
( [ functional ] ) are the results of the local gauge invariance of the skyrme force @xcite .
the time - odd coupling constants are uniquely determined from the original skyrme - force parameters and therefore the time - odd mean field has no adjustable parameters once the time - even mean field is given . however , as discussed by dobaczewski and dudek @xcite , one may consider the energy density to be a more fundamental construction than the skyrme interaction itself .
in such a case , all 20 coupling constants in eq .
( [ functional ] ) can be treated and adjusted independently .
however , the gauge - invariant conditions restrict the values of six time - odd coupling constants and leave the freedom to modify the values of @xmath51 and @xmath55 .
we are going to vary @xmath51 and @xmath55 to see the influence of these terms on the oblate - prolate transition in the odd - mass @xmath0hg isotopes .
we will omit the terms @xmath53 because @xmath56 are omitted in the parameterization of the present skyrme forces . in light hg isotopes ,
the neutron level density is high in the oblate minimum , while it is low in the prolate minimum .
because of the difference of level densities in the respective minima , it happens that neutrons in the oblate minimum are superconductive , whereas they are normal in the prolate minimum within the same isotope .
the situation is extremely inconvenient because we want discuss the energy difference between the oblate and prolate minimum . besides , a sudden phase - transition from the normal to superconductive state or vice versa at certain neutron numbers may be unphysical and it may be due to the defect of the number non - conserving pairing scheme . in the present study we employ the approximate number - projection method of lipkin and nogami ( ln ) @xcite and
calculate its corrections to both mean field and pairing energies @xcite .
the number projected energy is expressed as @xmath57 where @xmath58 and @xmath59 are constant parameters @xcite .
the second order expansion is considered in the lipkin - nogami method .
variation with respect to the density matrix and pairing tensor as in the case of the hfb equation yields the hfbln equation applicable to both even and odd nuclei ( hereafter simply called hfb ) .
the parameter @xmath58 is given by the particle number condition , whereas the parameter @xmath59 is held fixed during the variation and is determined after the variation from the additional condition , @xmath60 @xmath61 equation ( [ lam2 ] ) is calculated by the following density matrix and pairing tensor @xcite ; @xmath62 for an odd nucleus , the density matrix and pairing tensor of eqs .
( [ density matrix2 ] ) and ( [ pairing tensor2 ] ) are used on the right hand side of the above equations @xcite .
we have analysed the oblate - prolate transition in the odd - mass light hg isotopes using the skyrme hartree - fock - bogoliubov method with effective interactions siii , ski3 and sly4 .
shape transition occurs only in the odd - mass isotopes as the result of a subtle balance between the time - even mean - field energy , the loss of pairing energy by the odd neutron and the time - odd mean - field energy .
we have examined effective interactions siii , sgii @xcite , skm@xmath63 @xcite , skx @xcite , ski1 , ski2 , ski3 , ski4 and ski5 @xcite and sly4 whether they can predict the oblate - prolate shape coexistence in light hg isotopes .
we have found that both siii and ski3 interactions have desirable isospin dependence and deformation energies .
we have assumed the zero - range density - dependent pairing interaction .
the pairing strength @xmath14 was adjusted to reproduce the one - particle separation energy @xcite .
the density parameter @xmath64 is taken as the saturation density of symmetric nuclear matter which is obtained from skyrme parameters .
adjusted values are given in the respective figure captions .
we have investigated the effect of time - odd mean - field terms on the hfb energy , first by using the time - odd coupling constants derived from the skyrme force parameters and second by switching on and off the coupling constants @xmath55 and @xmath51 , respectively .
these coupling constants in @xmath47-representation @xcite are given in table [ param ] .
they vary very much with different skyrme forces , e.g. , the coupling constant @xmath65 is 14 , 84 and 208 for the respective siii , ski3 and sly4 forces .
the polarization energy depends strongly on the blocked level by an odd neutron .
even the sign of energy changes with different choice of the blocked level . in the present study we simply choose as the blocked level the lowest quasi - particle state that is obtained from the hfbe calculation .
mercury isotopes are very interesting because the shape of the ground state varies very much as the neutron number decreases from the closed shell .
figure [ pes ] shows the deformation energy of even hg isotopes calculated by the constraint siii - hf method .
we see the spherical shape in the neutron - closed @xmath66hg .
the oblate minimum with deformation @xmath67 then develops in @xmath68hg .
a rather complicated prolate minimum appears in @xmath69hg . finally the deep prolate minimum with deformation @xmath70 develops in @xmath71hg . the oblate - prolate shape coexistence in light hg isotopes has been predicted by many authors , e.g. , @xcite . figure [ p - o ] shows the siii calculation of the oblate - prolate energy difference .
the hf calculation ( white circle ) predicts prolate deformation in the ground state of all hg isotopes .
the hfbe calculation ( black circle ) , the fully paired state without the blocking effect , predicts oblate deformation in the ground state of all hg isotopes .
the shape changes due to the large gain of pairing energy in the oblate minimum because of the high level density as compared to the level density in the prolate minimum . in the siii - hfb calculation ( white square )
, on the other hand , the shape of odd mass @xmath0hg isotopes returns to the prolate shape .
this is because the loss of pairing energy due to the blocking by the odd neutron is also large in the oblate minimum than in the prolate one .
the effect of the time - odd mean fields on the oblate - prolate transition is weak and a few tenths of the blocking effect .
figure [ is ] shows the siii - hfb calculation of ( a ) the mean - square charge radius and ( b ) one - neutron separation energy of @xmath72hg isotopes .
the sharp increase in the mean - square radius of the odd - mass @xmath0hg isotopes is very well explained by the calculation .
the one - neutron separation energy is also well reproduced in the calculation .
this is so because the depth of the pairing force has been adjusted to reproduce the observed separation energies .
it is noted that the white circle in fig .
[ is](b ) is a derived value from systematic trends @xcite and not an experiment .
as is well known , a proper description of odd nuclei by mean field theory requires to break the time - reversal symmetry and introduces the time - odd components ( toc ) in the mean field .
the effect of an odd neutron on the hfb energy is therefore twofold ; ( a ) the loss of pairing energy due to the blocking of the fermi level by the odd neutron , and ( b ) the time - odd mean - field energy arising from the blocked level .
the time - odd mean fields have been studied both in the analysis of super - deformed rotational bands @xcite and the odd - even nuclear mass difference @xcite . in order to see the blocking energy ( the loss of pairing energy ) and the polarization energy ( the time - odd mean - field energy ) separately , we define the blocking energy @xmath73 as the energy difference between @xmath74 with no toc and @xmath42 .
we also define the polarization energy @xmath75 as the energy difference between @xmath74 and @xmath74 with no toc , i.e. , @xmath76 figure [ p - o_siii ] shows the siii - hfb calculation of ( a ) the blocking energy in the oblate and prolate minimum , ( b ) the polarization energy in the oblate minimum , ( c ) the polarization energy in the prolate minimum and ( d ) the oblate - prolate difference of hfb energy . in fig .
[ p - o_siii](a ) , the blocking energy in the oblate minimum is larger by @xmath77200 kev than the one in the prolate minimum .
this results from higher neutron level - density in the former than in the latter .
the blocking energies are reversed in @xmath78hg because of the small deformation @xmath79 in the prolate minimum of this isotope and the level density in the minimum is high accordingly . as mentioned in section 2.2 ,
the @xmath80 and @xmath81 terms are free from the time - even parameterization .
to investigate their effects on the hfb energy , we have further divided the polarization energy into contributions with ( i ) the full time - odd terms , ( ii ) the time - odd terms without @xmath51 , ( iii ) the time - odd terms without @xmath55 , and ( iv ) the time - odd terms without both @xmath51 and @xmath55 . in the following we assume simultaneous modifications of the isoscalar and isovector coupling constants of a given species .
figures [ p - o_siii ] ( b ) and ( c ) display the contributions to the polarization energy .
the polarization energy with full time - odd terms ( white circle ) is quite small .
it is attractive and less than 50 kev in the oblate minimum , while it is repulsive and less than 50 kev in the prolate minimum .
when the @xmath51 terms are omitted ( black circle ) , the polarization energy becomes attractive and more than 250 kev in the oblate minimum and it is attractive around 150 kev in the prolate minimum .
the @xmath51 terms have repulsive contribution to the hfb energy .
when the @xmath55 terms are omitted ( white square ) , on the other hand , the polarization energy becomes repulsive and @xmath7750 kev in both minima .
the @xmath55 terms have large attractive contributions to the hfb energy .
when both @xmath51 and @xmath55 terms are omitted ( black square ) , the polarization energy becomes small .
the effect of the @xmath51 and @xmath55 terms is opposite and cancels with each other .
figure [ p - o_siii](d ) shows the difference of energy in the oblate and prolate minimum .
the figure shows the calculation with the full time - odd terms ( white circle ) , without the @xmath51 terms ( black circle ) , without the @xmath55 terms ( white square ) , without both @xmath51 and @xmath55 terms ( black square ) and without all time - odd terms ( white triangle ) .
when both @xmath51 and @xmath55 terms are omitted , the oblate shape is predicted in @xmath82hg , while the prolate shape is calculated in @xmath1hg with marginal energy difference between the oblate and prolate minima .
table [ p - o_siiit ] summarizes the oblate - prolate difference of the hfbe energy @xmath83 , the blocking energy @xmath84 , the polarization energy @xmath85 and the sum of these contributions @xmath86 .
the polarization energy @xmath85 is calculated with four different time - odd mean fields .
they are the hfb calculation ( i ) with full time - odd terms , ( ii ) without the @xmath51 terms , ( iii ) without the @xmath55 terms and ( iv ) without both @xmath51 and @xmath55 terms .
the energy difference @xmath83 shows that the minimum is found at the neutron number @xmath87 and the oblate minimum gets deeper than the prolate one as the neutron number is away from @xmath87 in both directions .
for the blocking energy @xmath73 , it is repulsive and larger by 400 kev in the oblate minimum than in the prolate one .
the blocked level is labeled with the quantum number @xmath88 , the component of neutron angular momentum along the symmetry axis . for the odd mass @xmath0hg
, the blocked level has @xmath89 in the oblate minimum , while it has @xmath90 or @xmath91 in the prolate minimum . in the calculation with full time - odd terms ( case ( i ) ) , the polarization energy is attractive in the oblate minimum , while it is repulsive in the prolate minimum , leading to the the attractive energy difference of @xmath85 .
the attractive @xmath85 weakens the repulsive @xmath84 and prevents the oblate - prolate shape transition in these isotopes . in the calculation without the @xmath51 terms ( case ( ii ) ) , the polarization energy becomes attractive because of the absence of the repulsive @xmath51 terms .
in this case , the polarization energy @xmath85 almost cancels the blocking energy @xmath84 . in the calculation without the @xmath92 terms ( case ( iii ) ) , the polarization energy is weakly repulsive because of the absence of strongly attractive @xmath55 terms . finally in the calculation without both @xmath51 and @xmath55 terms , the polarization energy is weakly attractive in both minima .
the effects of the @xmath51 and @xmath92 terms on the polarization energy will be discussed more generally in the last paragraph of this section .
figure [ p - o_ski3 ] shows the results of the ski3-hfb calculation .
we had the difficulty of convergence in this calculation when the full time - odd components are considered .
the difficulty arises mainly due to the strong coupling - constant @xmath93 derived from the ski3 parameters ( see table [ param ] ) and partly due to the large hf basis of the present study ( @xmath27 ) .
the difficulty occurs when the energy splitting between the time - reversed partners becomes large and a few of the quasi - particle states have negative energy .
since we have no way to correct this difficulty , we simply omit the @xmath55 terms and display the results in fig .
[ p - o_ski3 ] .
the blocking energy @xmath73 in fig .
[ p - o_ski3](a ) is similar to the one in the siii - hfb calculation .
figures [ p - o_ski3](b ) and [ p - o_ski3](c ) show the polarization energy ( iii ) without @xmath55 terms ( white square ) and ( iv ) without both @xmath55 and @xmath51 terms ( black square ) .
the @xmath51 terms have repulsive contributions of @xmath77150 kev in the oblate minimum while it is @xmath77200 kev in the prolate minimum . in fig .
[ p - o_ski3](d ) , the oblate - prolate transition is very well explained when the terms @xmath55 are omitted .
it is also well explained by omitting the whole time - odd components .
table [ p - o_ski3 t ] summarizes the results of the ski3-hfb calculation .
the energy difference @xmath83 is nearly the same for even and odd @xmath94hg isotopes .
this fact helps very much in reproducing the oblate - prolate transition in the odd mass @xmath0hg isotopes .
the ski3 force is very well suited to describe the mean fields of light hg isotopes .
figure [ p - o_sly4 ] shows the results of the sly4-hfb calculation .
the blocking energy @xmath73 in the odd - mass @xmath0hg is larger by @xmath77200 kev in the oblate minimum than in the prolate one .
however , their magnitudes are reversed at the neutron number @xmath95 and @xmath96 .
the reason is the same as the siii calculation .
the equilibrium deformation in the prolate minimum is reduced to @xmath97 in these isotopes .
figures [ p - o_sly4](b ) and [ p - o_sly4](c ) show the terms @xmath51 have repulsive contributions to the polarization energy .
in addition , the terms @xmath51 and @xmath55 have opposite signs and nearly cancel with each other . the oblate - prolate energy difference in fig .
[ p - o_sly4](d ) shows that the sly4 force is not very well suited to describe deformation properties of light hg isotopes .
here we also had the difficulty of convergence when the @xmath51 terms are not included in the calculation .
the difficulty may be due to the @xmath55 terms .
the repulsive @xmath51 terms cancel the attractive @xmath55 terms . without the @xmath51 terms ,
the attractive @xmath55 terms are quite strong and cause the difficulty of convergence . table [ p - o_sly4 t ]
summarizes the results of the sly4-hfb calculation .
the energy difference @xmath83 shows that the sly4 force favors the oblate shape in light hg isotopes much more than the siii and ski3 forces .
this fact results in the oblate - prolate transition only in the ground state of @xmath82hg isotope .
the oblate shape is predicted for all heavier isotopes . to see the effects of the time - odd @xmath51 and @xmath55 terms
more closely , we have analysed the odd - mass sn , ba , yb , hg and pb isotopes by the ski3-hfb method with the basis space of @xmath98 .
a small basis space was chosen in order to avoid the difficulty of convergence .
results of the ski3-hf calculation are shown in fig .
the polarization energy is plotted against @xmath99 of the blocked level .
polarization energies are calculated ( a ) with the full time - odd terms , ( b ) without @xmath55 terms and ( c ) without both @xmath55 and @xmath51 terms .
we may conclude that the @xmath55 terms are attractive and have a large effect on the blocked level with low @xmath88 , in particular @xmath100 .
the terms @xmath51 are repulsive for levels with any value of @xmath88 .
the remaining time - odd terms are attractive for levels with high @xmath88 and are negligible for levels with low @xmath88 .
the same features of the time - odd terms are observed also for the siii and sly4 interactions . for the odd - mass @xmath0hg in tables [ p - o_siiit ] , [ p - o_ski3 t ] and [ p - o_sly4 t ]
, we have seen that the time - odd terms have attractive contributions to @xmath86 in the cases ( i ) and ( ii ) , while the terms have repulsive contributions in the case ( iv ) .
the shape and the mean - square radius of hg isotopes have been investigated by using the skyrme hartree - fock - bogoliubov method .
we have first examined effective interactions siii , sgii , skm@xmath63 , skx , ski1 , ski2 , ski3 , ski4 , ski5 and sly4 to see if they can predict the oblate - prolate shape coexistence in light hg isotopes .
we have then examined the loss of pairing energy due to the odd neutron by employing the zero - range density - dependent pairing force .
lipkin - nogami corrections to both even and odd nuclei are calculated within the hfb formalism .
we have also investigated the effect of the time - odd @xmath55 and @xmath51 terms on the hfb energy in relation to the blocked level by the odd neutron .
we have obtained the following conclusions .
( 1 ) the effective interactions siii and ski3 have desirable deformation properties in predicting the oblate - prolate shape coexistence in light hg isotopes . ( 2 ) the blocking energy of the pairing is the most important factor to change the shape of the odd - mass @xmath0hg isotopes .
( 3 ) the effect of the time - odd terms on the prolate - oblate transition is weak and cancels the effect of the blocking energy by a few tenths . we have also investigated the effects of the time - odd terms @xmath51 and @xmath55 by blocking several levels near the fermi surface of sn , ba , yb , hg and pb isotopes . we have found that the terms @xmath55 are attractive for the blocked levels with low @xmath88 , in particular @xmath100 . on the other hand , the terms @xmath51 are repulsive for the blocked levels with all values of @xmath88 .
the remaining time - odd terms are attractive for the blocked levels with high @xmath88 and negligible for the blocked levels with low @xmath88 .
we had the difficulty of convergence in the ski3-hfb calculation when the @xmath55 terms are included .
the difficulty also arises from the sly4-hfb calculation when the @xmath51 terms are omitted .
these difficulties are mainly due to the strong @xmath55 terms and partly due to the large basis space of the calculation ( @xmath27 ) . for large @xmath55 ,
the energy splitting between the time - reversed partners becomes large and a few of the quasi - particle states have negative energy .
we would like to thank k. arita for many illuminating discussions .
numerical computation in this work was partly carried out at the yukawa - institute computer facility , kyoto university .
99 p. aufmuth , k. heilig and a. steudel , at .
data nucl .
data tables 37 ( 1987 ) 455 .
p. dabkiewicz , f. buchinger , h. fischer , h .- j.kluge , h. kremmling , t. khl , a. c. mller and h. a. schuessler , phys .
lett . 82b ( 1979 ) 199 .
s. frauendorf and v. v. pashkevich , phys .
lett . 55b ( 1975 ) 365 .
m. calilliau , j. letessier , h. flocard and p. quentin , phys .
lett . 46b ( 1973 ) 11 .
s. yoshida and n. takigawa , phys .
c55 ( 1997 ) 1255 .
reinhard , c. reiss , m. bender , t. brvench , t. cornelius and j. a. maruhn , hyp .
127 ( 2000 ) 13 .
t. niksi , d. vretenar , p. ring and g. a. lalazissis , phys .
rev c65(2002 ) 054320 .
m. beiner , h. flocard , nguyen van giai and p. quentin , nucl .
a238 ( 1975 ) 29 .
reinhard and h. flocard , nucl .
a584 ( 1995 ) 467 .
e. chabanat , p. bonche , p. haensel , j. meyer and r. schaeffer , nucl .
a635 ( 1998 ) 231 .
h. j. lipkin , ann .
( n.y . ) 9 ( 1960 ) 272 . y. nogami , phys . rev . 134 ( 1964 ) b313
. m. bender , k. rutz , p. -g .
reinhard and j. a. maruhn , eur .
j. a8 ( 2000 ) 59 .
w. satua , r. wyss and p. magierski , nucl .
a578 ( 1994 ) 45 .
p. ring and p. schuck , _ the nuclear many - body problem _ , ( springer , berlin , 1980 ) .
d. vautherin and d. m. brink , phys .
c5 ( 1972 ) 626 . y. m. engel , d. m. brink , k. goeke , s. j. krieger and d. vautherin , nucl . phys .
a249 ( 1975 ) 215 .
d. vautherin , phys .
c7 ( 1973 ) 296 .
j. dobaczewski , h. flocard and j. treiner , nucl .
a422 ( 1984 ) 103 .
m. v. stoitsov , j. dobaczewski , p. ring and s. pittel , phys .
c61 ( 2000 ) 034311 .
j. dobaczewski , w. nazarewicz and m. v. stoitsov , proc .
on _ the nuclear many - body problem 2001 _ , brijuni national park , pula , croatia , june 2 - 5 , 2001 ( nucl - th/0109073 ) .
k. sugawara , prog .
35 ( 1966 ) 44 . p. ring , r. beck and h. j. mang , z. phys . 231 ( 1970 ) 10 .
t. duguet , p. bonche , p. -h .
heenen and j. meyer , phys .
c65 ( 2001 ) 014310 , 014311 .
j. dobaczewski and j. dudek , phys .
c52 ( 1995 ) 1827 , phys rev c55 ( 1997 ) e 3177 .
nguyen van giai and h. sagawa , phys .
lett . 106b ( 1981 ) 379 .
j. bartel , p. quentin , m. brack , c. guet and h. -b .
hkansson , nucl . phys .
a386 ( 1982 ) 79 .
b. alex brown , phys . rev .
c58 ( 1998 ) 220 . g. audi and a. h. wapstra , nucl .
a595 ( 1995 ) 409 .
a. faessler , u. gtz , b. slavov and t. ledergerber , phys .
lett . 39b ( 1972 ) 579 .
r. bengtsson , t. bengtsson , j. dudek , g. leander , w. nazarewicz and j. zhang , phys .
b183 ( 1987 ) 1 .
.[param ] the time - odd coupling constants @xmath51 and @xmath55 for the skyrme siii , ski3 and sly4 forces .
the coupling constants @xmath51 are given in mev@xmath101@xmath102 , while @xmath55 are given in mev@xmath101@xmath103 . [ cols="<,^,^,^,^,^,^",options="header " , ] | anomalous isotope shifts in the chain of light hg isotopes are investigated by using the hartree - fock - bogoliubov method with the skyrme siii , ski3 and sly4 forces .
the sharp increase in the mean - square radius of the odd mass @xmath0hg isotopes is well explained in terms of the transition from an oblate to a prolate shape in the ground state of these isotopes .
we discuss the polarization energy of time - odd mean - field terms in relation to the blocked level by the odd neutron .
# 1|#1 # 1 # 1 _ pacs _ : 21.10.dr ; 21.10.ft ; 21.10.ma ; 21.60.jz + _ keywords _ : isotope shifts ; shape coexistence ; hartree - fock - bogoliubov method ; lipkin - nogami corrections ; time - odd mean fields |
understanding the origin of the hierarchical pattern of quark masses and mixings has long been recognized as a challenge for physics beyond the standard model @xcite .
the discovery of neutrino masses and mixings with totally different flavor pattern than quarks ( i.e. @xmath1 and @xmath2 as against @xmath3 and @xmath4 ) has added more mystery to the flavor problem . in generic bottom - up pictures where quarks and leptons are treated as different species of particles with no particular relation between them ,
this problem is not so serious since one can simply focus on each sector separately , as is often done for neutrinos @xcite .
however , in grand unified theories where the quarks and leptons unify at a very high scale , one would naively expect that their masses and mixings would exhibit a similar pattern .
the fact that they are so different may be hint of some really new exciting underlying physics . in this note
we address this question in the context of supersymmetric so(10 ) models with renormalizable yukawa couplings being responsible for fermion masses .
we show that in so(10 ) models with * 10 * , * 126 * plus possibly another * 10 * or * 120 * higgs fields where fermion masses are generated by renormalizable yukawa couplings @xcite only and where type ii seesaw is responsible for neutrino masses @xcite , there is a natural way to have a unified understanding of both large lepton mixings and small quark ones .
the basic idea is to require that one of the * 10 * yukawa couplings is the dominant one contributing to up , down and charged lepton masses and has rank one with other smaller couplings providing neutrino masses as well as most of the quark lepton flavor hierarchy .
rank one plus small corrections as a way to unravel fermion flavor in d - brane models was discussed in @xcite .
we find that so(10 ) models with type ii seesaw @xcite are ideally suited for such an ansatz
. a specific form of the rank one matrix can lead to tri - bimaximal mixing with corrections dictated by the quark flavor pattern .
this paper is organized as follows : in sec .
ii , we review the mass formulae in so(10 ) models with renormalizable couplings ; in sec .
iii , we summarize our basic strategy for understanding the quark lepton flavor in a unified manner , discuss the rank one ansatz and apply it both the two generation case ( iiia ) and three generation cases ( iiib ) . in sec .
iiic , we present realistic three generation models and outline their predictions . sec .
iv is devoted to some specific conjectures for the rank one matrix which could emerge from discrete symmetry models with specific discussion on the correction to tri - bimaximal mixings .
v is devoted to a possible way to obtain the rank one ansatz in so(10 ) models and in sec .
vi , we present our conclusions .
the basic idea in this class of models is to consider susy so(10 ) theory with higgs fields that give fermion masses to be in * 10 * ( denoted by @xmath5 ) and @xmath6 ( denoted by @xmath7 and @xmath8 ) plus either an extra * 10 * ( @xmath9 ) or * 120 * ( @xmath10 ) . the gut symmetry is broken by * 210*+*54*+*126*+@xmath11 @xcite .
the yukawa superpotential of this model is : @xmath12 where the symbol @xmath13 stands for the * 16 * dimensional representation of so(10 ) that represents the matter fields .
the coupling matrices @xmath14 and @xmath15 are symmetric , and @xmath16 is symmetric or anti - symmetric depending on whether we adopt @xmath17 or @xmath10 .
the representations @xmath5 , @xmath9 and @xmath7 have two standard model ( sm ) doublets in each of them whereas @xmath10 has four such doublets . the general way to understand so
many sm doublets is that at the gut scale @xmath18 , once the gut and the @xmath19 symmetry are broken , one linear combination of the up - type doublets and one of down - type ones remain almost massless whereas the remaining ones acquire gut scale masses just like the color triplet and other non - mssm multiplets .
the electroweak symmetry is broken after the light mssm doublets ( to be called @xmath20 ) acquire vacuum expectation values ( vevs ) and they then generate the fermion masses .
the resulting mass formulae for different fermion masses are given by : @xmath21 where @xmath22 are mass matrices divided by the electro - weak vev @xmath23 and @xmath24 and @xmath25 are the mixing parameters which relate the @xmath20 to the doublets in the various gut multiplets .
more precisely , the matrices @xmath14 , @xmath15 and @xmath16 in @xmath22 are multiplied by the higgs mixings
. the precise definitions of the couplings and the higgs mixings are given in ref.@xcite .
when @xmath17 is adopted for the @xmath16 coupling , @xmath26 and @xmath27 . in generic so(10 ) models of this type , the neutrino mass formula has a type i @xcite and a type ii @xcite contributions : @xmath28 where @xmath29 is the vev of the @xmath30 triplet in the * 126 * higgs field and is given by @xmath31 .
note that in general , the two contributions to neutrino mass depend on two different parameters and it is easy to have symmetry breaking pattern in so(10 ) @xcite where the first contribution ( the type ii term ) dominates over the type i term .
the neutrino mass formula then becomes @xmath32 note that @xmath15 is the same coupling matrix that appears in the charged fermion masses in eq .
( [ eq2 ] ) , up to factors from the higgs mixings and the clebsch - gordan coefficients .
the equations ( [ eq2 ] ) and ( [ eq4 ] ) are the key equations in our unified approach to address the flavor problem .
the main hypothesis of our approach is that the fermion mass formula of eq .
( [ eq2 ] ) are dominated by the matrix @xmath14 with the contributions of @xmath15 and @xmath33 being small perturbations . in the limit of @xmath34 ,
the quark and lepton mixings vanish as do the neutrino masses
. we will show below that this simple hypothesis combined with eq .
( [ eq4 ] ) can simultaneously explain large lepton mixings while keeping the quark mixings being proportional to @xmath35 and hence small .
we will subsequently assume that the matrix @xmath14 has rank one in which case the mass hierarchy can also be explained in a natural manner .
the quark and lepton mixing matrices are given by the product of diagonalizing unitary matrices for quark and lepton mass matrices as follows : denoting the diagonalizing matrices of @xmath36 , @xmath37 by @xmath38 and @xmath39 respectively ( e.g. , @xmath40 and similarly for the down quark mass matrix ) , the ckm ( cabibbo - kobayashi - maskawa ) quark mixing matrix is given as @xmath41 . the pmns ( pontecorvo - maki - nakagawa - sakata ) lepton mixing matrix is given as @xmath42 in the similar notation ( e.g. , @xmath43 ) .
in general , when two matrices with random @xmath44 elements are considered , the mixing angles of the relative diagonalizing unitary matrices are all @xmath44 in radian , while the eigenvalues can have a hierarchy of @xmath45 .
in such an anarchical scenario , the neutrino masses and mixings can be explained ( except for the chooz bound of 13 neutrino mixing ) : the neutrino mixings are generically @xmath44 and there is a little hierarchy for the neutrino mass squared difference ratio @xmath46 @xcite . on the other hand , since the quark mixings are all smaller than @xmath44 and the masses of quarks and charged leptons are very much hierarchical , anarchic mass matrices in general provide no explanation of these observations .
besides , it appears that the mass ratios and ckm mixings have several correlations among them .
it is therefore to be expected that the quark and lepton matrices instead of being independent anarchic matrices must have some relations among them and an underlying theory leading to these relations . in this paper
we find that so(10 ) with type ii seesaw could be such a theory . when the fermion masses are given by the eqs .
( [ eq2 ] ) and ( [ eq4 ] ) , several possible outcomes are obtained by simple assumptions . to understand these possible outcomes from the eqs.([eq2 ] ) and
( [ eq4 ] ) , let us first ignore @xmath16 .
we then have the following possibilities : _ assumption 1 : _ take @xmath14 , @xmath15 are general rank 3 matrices , and @xmath15 is small .
this is the case analyzed to fit observed experimental data and to obtain predictions from the minimality of the number of parameters in various papers @xcite . here
, we list the properties resulting from the smallness of @xmath15 without resorting to any numerical fit . *
the ckm mixings are small , due to the fact that there is an approximate up - down symmetry and @xmath47 @xcite .
* bottom - tau unification up to @xmath48 . *
the 3 neutrino mixings are generically of @xmath44 since @xmath14 and @xmath15 are unrelated matrices .
the type ii seesaw dominance of the neutrino mass is crucial for the generic largeness of the neutrino mixings .
thus it is interesting that without any special assumption , the gross features of fermion mixings can be reproduced .
this does not , however , throw any light on the mass hierarchies among quarks and leptons , though one can fit the experimental results by the choices of parameters ( even in the type i seesaw ) @xcite . since we use the experimental data as an input
, these scenarios do not provide a fundamental understanding of either the mass hierarchy for quarks and charged leptons , or why the 13 neutrino mixing ( @xmath0 ) is less than @xmath44 .
similar situation holds for models where @xmath16 is added @xcite . _ assumption 2 : _
let us next consider the specific case when @xmath14 is a rank 1 matrix @xcite , and @xmath15 is a rank 3 matrix with eigenvalues of @xmath15 being hierarchical ( @xmath49 ) and small compared to the elements of @xmath14 . as we noted in ref.@xcite , this choice helps to suppress proton decay in susy so(10 ) models without invoking huge cancellations among the colored higgsino exchange amplitudes . in this case
we will show in the next two subsections that the following results follow : 1 .
ckm mixings are small .
2 . approximate bottom - tau unification occurs .
3 . @xmath50 .
the quark mixing are related as @xmath51 ( where @xmath52 is a phase ) and @xmath53 .
atmospheric and solar neutrino mixings are generically large , but 13 mixing is @xmath54 .
all these predictions are in qualitative agreement with observations .
the advantage of the rank one assumption is that it naturally explains the mass hierarchies among quarks and leptons in addition to large lepton mixings .
we emphasize that these features are obtained from the rank 1 assumption above without using any numerical inputs , and our claim here is not based on the scenario of the numerical predictions from a fit in which the minimality of parameters plays a key role . before we do a full demonstration of these results in the context of a three generation model ,
let us illustrate the first four points in the context of a two generation model . in this subsection , we apply our rank one hypothesis to the second and the third generation .
we will confirm the results 1 - 4 mentioned above .
the starting point is the mass relation from eqs .
( [ eq2 ] ) and ( [ eq4 ] ) where we ignore the @xmath16 contribution .
using our assumption , we have @xmath55 and @xmath56 ( without loss of generality , we can parameterize @xmath15 to be diagonal ) .
the parameter @xmath57 is of @xmath44 in general .
we now have ten parameters ( @xmath57 , @xmath58 and @xmath59 as real parameters , @xmath60 , @xmath61 , and @xmath62 as complex parameters , and @xmath29 for neutrino mass scale ) describing ten observables of all lepton and quark mixings and masses .
one can easily obtain @xmath63 at the leading order neglecting @xmath64 correction , where @xmath65 is a ratio of vevs of @xmath66 and @xmath67 .
therefore , @xmath59 corresponds to the freedom of @xmath65 , @xmath68 . when @xmath69 , we obtain @xmath70 because @xmath71 , @xmath62 is small , i.e. @xmath72 . to proceed further , we first diagonalize the charged fermion mass matrices to zeroth order in @xmath73 .
the matrix diagonalizing this is given by @xmath74 since @xmath75 .
let us now see how the small quark mixings arise despite large mixings in @xmath76 . because @xmath77 has a small off - diagonal element , @xmath78 , the down - type quark mass matrix is diagonalized by @xmath79 , where @xmath80 is close to unit matrix whose off - diagonal element is @xmath81 .
the up - type quark mass matrix is diagonalized by @xmath82 , where the off - diagonal element of @xmath83 is @xmath84 .
the quark mixing matrix is then given by @xmath85 , in this product @xmath76 cancels out leaving small mixings between the two generations i.e. small @xmath86 , where @xmath52 is a phase of @xmath62 .
coming now to lepton mixings , suppose that the charged lepton mass matrix is diagonalized by @xmath87 , then it can be written as @xmath88 , where @xmath89 is close to a unit matrix similarly to the quark sector , and is roughly equal to @xmath90 .
since the neutrino mass matrix is already diagonal as a parameterization , the pmns matrix is given by the charged lepton mixings so that @xmath91 .
this leads to a large lepton mixing as desired . in the two generation case
, @xmath57 describes approximately ( up to small corrections of order @xmath92 ) the atmospheric mixing angle " .
since this was an input into our rank one ansatz , we can choose to be large to explain the observations .
if @xmath93 and @xmath62 are assumed to be real , there are six real parameters in this model . in this case ,
@xmath94 , @xmath95 and @xmath96 can be written as a function of @xmath97 , @xmath98 and @xmath92 for example .
even if @xmath93 and @xmath62 are all complex , we have the following approximate relation at the grand unified scale : @xmath99 since the relations are satisfied under the assumption of approximate rank 1 property irrespective of the counting of freedom , they are stable even in the case of three generation model . indeed , the predictivity from the minimality of the parameter is related to the @xmath64 corrections , and the minimality does not play a crucial role in the approximate relations from the rank 1 assumption .
it is known that there is a solution that the large atmospheric mixing is obtained even if the smallness of @xmath15 is not assumed a priori . in the scenario , the @xmath100-@xmath101 mass convergence as well as the other experimental inputs predict the neutrino mixing as an output @xcite .
our main goal in this section is not to give numerical predictions but rather to show how one can get qualitatively expected hierarchical pattern for masses and mixings .
later on we of course study the detailed numerical predictions . as it turns out there is a fine - tuned solution to fit the experimental data even if @xmath61 is comparable to @xmath58 .
however such a fine - tuned solution is sensitive to the numerical inputs , and therefore the numerical predictions in this case may be unstable under a possible higher order correction . in our case
where the quark and lepton mass hierarchy is predicted by the rank 1 assumption , they are stable under radiative corrections . while the qualitative predictions are in the expected range , we note that the approximate relation @xmath102 is not very good agreement with the current observation , and small @xmath16 will be invoked to obtain the best fit of the experimental data .
we emphasize that our final solutions do not use any fine tuned cancellations , and thus are stable even if we add small corrections to fit the numerical experimental data .
the fermion mass equations for this case are those in eq .
( [ eq2 ] ) with all coupling matrices being @xmath103 .
the assumption that @xmath14 has rank one means that we can write it as @xmath104 @xmath105 again , we can parameterize @xmath15 to be diagonal and @xmath106 to be real without loss of generality . at first , we ignore @xmath16 . in order to analyze the detailed consequences of this assumption
, we go to the basis where @xmath14 is diagonal .
this is achieved by the matrix : @xmath107 where @xmath108 and @xmath109 with @xmath110 where @xmath111 .
it is interesting to note that in the diagonalization matrix there is an ambiguity resulting from the residual su(2 ) flavor symmetry in @xmath14 ( i.e. one of three mixing angles is not fixed at this stage ) .
we choose the unitary matrix @xmath76 to be an approximate leading order diagonalization matrix of @xmath112 , @xmath113 , and @xmath114 as in the previous subsection ( @xmath82 , @xmath79 , and @xmath115 where @xmath116 are close to a unit matrix ) .
then , once the @xmath15 contribution is included , the afore mentioned su(2 ) flavor symmetry is broken and the ambiguity in mixing angles alluded to above is removed .
we wish to note that in our original parameterization of @xmath76 , we chose the 13 element to be zero since even after including the @xmath15-contribution , the 13 element goes to zero in the limit @xmath117 ( which is the limit where @xmath118 is rank 2 ) .
we have not used prejudices from neutrino experiment . by the same argument as in the case of two generations ,
@xmath76 is cancelled out in the ckm mixing matrix and the quark mixings are small .
the pmns matrix is given by @xmath119 and since the off - diagonal elements of @xmath120 are small ( being related to quark mixings ) , neglecting the @xmath121 and @xmath122 quark mixings , we get for the solar and atmospheric mixing angles @xcite @xmath123 where @xmath124 is defined as the diagonal phase matrix @xmath125 needed to diagonalize the charged lepton mass matrix @xcite . we also get a formula for @xmath0 as follows : @xmath126 to proceed with the rest of the masses and mixings , let us define the matrices in the @xmath76 rotation : @xmath127 , @xmath128 , and so on . in this notation
, @xmath129 is a diagonalization matrix of @xmath130 . because @xmath131 and @xmath132 , one can obtain @xmath133 neglecting @xmath64 and @xmath134 corrections , @xmath129 can be approximately as : @xmath135 where @xmath136 and thus @xmath137 thus we have obtained all the features listed before . due to the generic largeness of the relative mixing angles of the unrelated matrices , solar and atmospheric neutrino mixing angles are of @xmath44 generically . on the other hand , 13
mixing is not in the category of the generic largeness since it is related to the ratio of eigenvalues of @xmath15 .
the eigenvalue ratio is also related to @xmath138 implying that the 13 mixing angle has to be small in our approach .
it is important to note that we do not assume a particular flavor texture such like hierarchical pattern in one matrix to obtain the feature .
the key property to obtain the features for the neutrino mixings is that the correction to the rank one charged lepton mass matrix and the type ii seesaw term are unified ( or more roughly , simultaneously diagonalized ) , as a result of so(10 ) unification .
the discussion above gives the qualitative consequences of the rank one property , and the experimental inputs are not used to obtain the features .
the discussion below will address the issue of the experimental data for the first generation .
actually , we have not listed the first generation masses and @xmath139 before .
in fact , if @xmath140 , one obtains the following relation among the fermion masses : @xmath141 where @xmath142 .
when one fits down quark and electron masses ( e.g. , @xmath143 ) , the up quark mass is clearly too large since @xmath144 . as a result ,
one of the first generation masses can not be fitted . besides , since @xmath145 in eq.([check - v ] ) is common for up- and down - type quarks , @xmath139 becomes too small compared to observations , since @xmath146 .
therefore , one needs non - vanishing contribution from @xmath16 to obtain realistic masses for the first generation and @xmath139 under the rank 1 assumption . as is well - known , the empirical relation @xmath147 is obtained when @xmath148 and @xmath149
. therefore we choose @xmath150 .
when @xmath151 is assumed , the choice of @xmath152 satisfies the georgi - jarskog ( gj ) relation ( @xmath153 ) for the down - type quarks and charged lepton masses . the up quark mass can be fit by using the freedom of @xmath154 . as a result
, we have the following two solutions typically to fit the first generation masses and @xmath139 in a simple manner .
_ case a _ : @xmath155 and @xmath156 .
the smallness of up quark mass is realized by a cancellation in @xmath157 .
in this case , @xmath16 has to be symmetric which we can obtain by employing an extra @xmath158 higgs field .
for example , @xmath159 is the simplest solution , giving @xmath160 where @xmath161 is given in eq.([check - th ] ) and used a relation @xmath162 from @xmath163 .
assuming that the corrections from the other elements of @xmath164 ( e.g. , @xmath165 ) are small , we have an approximate relation @xmath166 which is in good agreement with the experiment . _ case b _ : in this case we have @xmath167
. then , 11 and 12 elements of @xmath168 are zero , where @xmath145 in eq.([check - v ] ) is an approximate diagonalization matrix in the limit @xmath169 .
the 12 element of @xmath170 produces the cabibbo angle .
the gj relation is manifest when @xmath171 and @xmath172 .
the up quark mass is fitted by the smallness of @xmath154 , @xmath173 . in this case
, @xmath16 can be either symmetric or anti - symmetric .
since 11 element vanishes automatically , anti - symmetric coupling from * 120 * higgs field is a better choice . as is noted , @xmath145 contributes to @xmath0 , but it does not contribute to the cabibbo angle . as a result , we obtain from eq.([ue3-v12 ] ) @xmath174 where @xmath175 is a relative phase between @xmath176 and @xmath177 roughly , and we have used a relation @xmath162 . since @xmath139 is generated purely from @xmath16 , it is not directly correlated to @xmath138 contrary to the case a. as we have noted , to fit @xmath92 , @xmath178 and @xmath95 very well , one needs a correction in @xmath179 .
however , the correction does not affect the approximate expressions for @xmath0 very much .
it is interesting that the @xmath0 is related to the mass ratio of neutrino in both cases .
since the gj relation and the empirical relation of @xmath139 are not exact relations , there can be a shift from them in a numerical fit analysis .
here we assumed @xmath180 to satisfy the gj relation and the empirical relation of @xmath139 in a simple manner .
when one introduces other parameters especially for symmetric @xmath16 , there will be an accidental fine - tuned solution for the relations in a general fit for @xmath181 .
actually , when @xmath16 is symmetric ( @xmath182 ) and @xmath183 , it results in a minimal model in which only one @xmath158 and @xmath11 higgs fields couple to fermions with @xmath14 being rank 3 . in the minimal model for the fermion sector
, it is known that there is a fine - tuned solution to fit fermion masses and mixings @xcite unless the minimality of the higgs potential is taken into account . in this case , when the first generation masses and @xmath139 are tuned , there is no freedom to adjust @xmath0 and thus the approximate relation in the previous subsection holds , @xmath184 . when @xmath185 ( but @xmath186 ) , @xmath0 can be tuned to be any value ( including zero ) since
first generation masses and @xmath139 can be fitted even if @xmath187 .
( when @xmath14 is rank 1 and @xmath16 is anti - symmetric , there is no such fine - tuned solution .
when @xmath14 is rank 3 , the fine - tuning fit for @xmath188 is allowed @xcite . ) therefore , the assumption @xmath180 to satisfy gj relation in a simple manner is crucial to keep the @xmath0 prediction
. actually , when @xmath180 is assumed , the fine - tune solutions are removed , and the @xmath0 is predicted as we have noted , irrespective of the number of parameters .
we also note that the assumption @xmath180 is preferable to suppress nucleon decay amplitudes naturally . the case with the assumption that @xmath14 ( rank 1 ) and @xmath15 are real and anti - symmetric @xmath16 is pure imaginary ( in which case , the charged fermion mass matrices are hermitian ) is in fact the model discussed in @xcite .
for this case , cancellation can not happen between @xmath15 and @xmath16 and thus the numerical fit does not shift very much from the above expression . in the numerical fit ,
it predicts @xmath189 in the case where @xmath190 ( and the gj relation is manifest ) . under the hermiticity assumption ,
one obtains @xmath191 and it is consistent with the above expression . in this case , since @xmath192 is real , the pmns phase is roughly same as the phase in the expression in eq.([ue3-case - b ] ) and thus @xmath193 and we obtain @xmath194 or @xmath195 using the experimental inputs . for the case of @xmath196 , one can also fit the experimental data and the prediction is @xmath197 . since the cancellation is not allowed between @xmath198 and @xmath199 , the experimental data of @xmath200
cuts the upper region of experimentally allowed mass squared ratio difference , and then gives an upper bound of @xmath0 .
in the previous section , we incorporated large lepton mixings but their values were inputs into the theory . in this section
, we consider special cases where the dominant part of the lepton mixing is in the tri - bimaximal form @xcite .
this would require special form for the rank one matrix @xmath14 .
we envisage that the rank one form for @xmath14 as well as the matrix forms for @xmath15 and @xmath33 come from some vacuum alignment of flavon fields , e.g. , @xcite . in the triplet flavon models ,
the @xmath103 matrix can be expanded by the tensor products of the flavon fields when there are three independent flavons .
the three flavon fields can be expressed ( without loss of generality , by making unitary transformations ) as @xmath201 in general , there is no reason for the flavon vevs to be hierarchical , and the large neutrino mixings can originate from @xmath202 , @xmath203 @xcite .
the experimental result from the neutrino oscillation seems to imply a special alignment of flavon vevs rather than the generic largeness , namely @xcite , @xmath204 the vacuum alignment can be obtained by imposing a discrete flavor symmetry @xcite , leading to tri - bimaximal neutrino mixings .
it is worth pointing out at the beginning that the aligned flavon fields can be written in several ways by choice of coordinates ( or by making unitary transformations ) and the final results are independent of the coordinate choice .
it is interesting to note that the aligned flavon vevs correspond to a link of a hexahedron , a diagonal line of a lateral surface , a diagonal line of a regular hexahedron , respectively .
the interpretation becomes clear when the flavon fields are expanded in terms of the following orthogonal axes of coordinates ( called hexahedral coordinate ) @xmath205 which correspond to the three lateral links of the regular hexahedron .
the hexahedral coordinate is convenient to describe the @xmath206 rotation around the surface - diagonal axes of the hexahedron .
in fact , the regular hexahedron has different coordinates to describe the symmetry of the shape .
one can consider a coordinate system which proves convenient to describe the @xmath207 rotation around the vertex - diagonal axes of the hexahedron , @xmath208 the axes of the coordinates @xmath209 and @xmath210 are on the regular triangle which is formed by three of the hexahedron s vertices , and @xmath211 is perpendicular to the triangle .
we call this tetrahedral coordinate .
the unitary matrix for the coordinate transformation from hexahedral ( unprimed ) to tetrahedral ( primed ) coordinate is the tri - bimaximal ( tb ) matrix i.e. @xmath212 , where @xmath213 therefore , in general ( irrespective of the rank one assumption ) , if the charged - lepton mass matrix is ( nearly ) diagonal in the hexahedral coordinate and the neutrino mass matrix is ( nearly ) diagonal in the tetrahedral coordinate , then the neutrino mixing matrix is ( nearly ) tri - bimaximal and given by : @xmath214 where @xmath215 is a diagonalizing matrix of @xmath114 in the hexahedral coordinate , and @xmath216 is a diagonalizing matrix of @xmath217 in the tetrahedral coordinate .
suppose that the vacuum alignment in eq.([alignment ] ) is given in the hexahedral coordinate as we have noted .
then , those flavons in the tetrahedral coordinate are given as ( @xmath218 ) @xmath219 therefore , from the discussion in the previous section , one can easily check that the nearly tri - bimaximal neutrino mixings are obtained when @xmath14 is rank one formed by @xmath220 ( irrespective of the choice of the coordinate ) , and @xmath15 is formed by @xmath221 and @xmath222 .
we define @xmath223 which is obtained an outer product of @xmath221 and @xmath222 , i.e. @xmath224 . in the tetrahedral
coordinate , @xmath225 .
as we have mentioned , the yukawa matrices can be expressed in terms of the tensor products of the flavon fields .
the symmetric matrices can be formed by six bases .
since we set 11 element to be zero in the hexahedral coordinate , we define the following matrices as the bases to form the linear space of symmetric matrices : @xmath226 where the elements of the matrices are presented in the hexahedral coordinate . in the following
, we will consider two models : ( i ) @xmath216 is a unit matrix ( @xmath15 is diagonal in the tetrahedral coordinate ) , ( ii ) @xmath216 is close to a unit matrix . the model i can have both case a and case b solutions as described in the previous section .
the case a solutions are , however , more natural in this model . in order to obtain such a solution
, we employ additional @xmath158 higgs to obtain a correction matrix @xmath16 , and @xmath16 is a symmetric matrix .
we arrange the @xmath14 , @xmath15 , @xmath16 couplings as follows : @xmath228 then , since the ratio of eigenvalues of @xmath15 is @xmath229 , we obtain @xmath230 . in the parameterization in the previous section , @xmath76 in eq.([u0 ] ) is the tri - bimaximal matrix because @xmath231 .
the fermion yukawa matrices are @xmath232 where @xmath233 when @xmath234 , and @xmath235 when @xmath236 . for the numerical fits ,
the parameter @xmath237 is given by @xmath238 , and the parameter @xmath239 is given by @xmath240 .
when @xmath241 , then goergi - jarskog relation is satisfied naturally . at that time ,
this is interesting since the empirical relation @xmath243 is satisfied simultaneously when @xmath244 are small ( up - type quark masses are more hierarchical rather than down - type quark masses ) . the parameter @xmath62 is fixed as @xmath245 .
the up quark mass can be made small by a choice @xmath246 .
cabibbo angle , @xmath0 and the ratio of mass squared differences @xmath247 are all correlated by the parameter @xmath239 .
the naive approximate relation is @xmath248 as we have derived in the previous section . in fig . 1
, we plot @xmath0 as a function of the mass squared difference ratio .
in the plot , we fit @xmath249 and @xmath250 using @xmath251 and @xmath237 ( which are assumed to be real in the plot ) .
then , @xmath0 is calculated as a function of @xmath239 .
( the mixing angles do not depend on @xmath58 ) .
we note that a correction from @xmath16 is needed to fit @xmath92 , e.g. , @xmath252 .
as we have mentioned , such correction do not modify the @xmath0 very much .
in fig . 1 , we also show the plot of @xmath0 as a function of @xmath253 . using the experimental constraint on @xmath254 and the other input from quark masses and mixings , we find that @xmath0 is predicted to be 0.07 - 0.08 .
we note that the smaller side of experimental range of @xmath255 is preferred from the numerical fit , which obeys from the naive relation @xmath256 . the solar mixing angle @xmath253 is found to be @xmath257 from the plot which is obeyed by the approximation eq.([th - sol ] ) when the pmns phase is 0 ( or @xmath258 , which is resulting from the assumption where @xmath251 and @xmath237 are real . the atmospheric mixing angle @xmath95 is @xmath259 up to @xmath2602 - 3@xmath261 correction from @xmath92 for model i irrespective of case a or case b solutions .
the current allowed range for the neutrino parameters at 2@xmath52 level are as follows @xcite : @xmath262 , @xmath263 and @xmath264 . the model ii can have both case a and case b solutions as well . in this model ,
the case b solutions are more natural .
it is possible that the @xmath15 coupling is not completely diagonal in the tetrahedral coordinate . using the available freedom , we choose the 12 element of @xmath15 in the hexahedral coordinate to obtain the case b solution .
since the 12 elements of @xmath266 and @xmath267 are zero , one can consider the choice : @xmath268 the matrix @xmath269 can be replaced with @xmath270 .
one can also choose @xmath16 to be antisymmetric , e.g. , @xmath271 . in the tetrahedral coordinate , the @xmath15 coupling is written when @xmath272 ( case b1 ) as @xmath273 and if the notation in the previous section is used , we obtain @xmath274 where @xmath275 .
the mass squared ratio is @xmath276 .
when we use @xmath277 ( case b2 ) , @xmath278 we obtain @xmath279 where @xmath280 .
the mass squared ratio is @xmath281 .
we note that if there is a @xmath282 correction in the 12 element in the tetrahedral coordinate , it modifies the @xmath283 angle largely , and it separates from the nearly tri - bimaximal mixing , and thus we do not use the choice .
as we have obtained in the previous section , @xmath0 prediction is @xmath284 clearly , if the parameters @xmath251 , @xmath239 , @xmath237 are all real , then , @xmath285 or @xmath286 , and the maximal and minimal values of @xmath0 are obtained . at that time
, there is no phase in the pmns mixing matrix .
( the kobayashi - maskawa phase can be obtained from a phase of @xmath62 and/or @xmath154 . ) in fig.2 ( case b1 , eq.([caseb1 ] ) ) and fig.3 ( case b2 , eq.([caseb2 ] ) ) , we plot @xmath0 when @xmath251 @xmath239 , @xmath237 are real to find the lower and upper limits . due to the off - diagonal elements of @xmath15 in
the tetrahedral coordinate , the atmospheric angle shifts from @xmath259 , and the shift is correlated to @xmath0 , unlike the case of model i. in case b1 , using the experimental constraint on @xmath255 , we find that @xmath0 is predicted to be 0.05 - 0.08 .
this solution corresponds to the sign choice @xmath287 in eq.([eq - ue3 ] ) .
we also plot atmospheric mixing angle in fig.2 ( right ) .
it is interesting to note that for the case b1 , @xmath239 should be negative for @xmath288 to fit mass squared difference ratio since @xmath289 . as a result ,
the direction of the shift is determined to fit experimental values , i.e. @xmath290 . in case b2 , the mass squared ratio can be fitted for both signatures of @xmath239 ( we find two branches in both graphs in fig.3 ) . the plot in fig.3
is shown constraining @xmath291 , and one can find that @xmath247 becomes too small for one of the branch to fit solar mixing angle , and @xmath290 is favored in this case as well .
larger @xmath0 values ( @xmath292 ) is preferred once we include the experimental limit on @xmath247 .
the solution corresponds to the sign choice @xmath293 in eq.([eq - ue3 ] ) .
as we discussed above that both case a and case b solutions can be obtained in models i and ii .
the interesting question is how can we distinguish these two models .
for example , the predictions of @xmath0 ( as shown in figs 1,2 and 3 ) distinguishes between the cases a and b. actually , if @xmath0 is just below the current chooz bound , the case b solution with @xmath294 is preferred . since the case a is a more natural solution for model i and case b is a more natural solution for model ii , one may weakly conclude that this prediction distinguishes between models i and ii .
but a stronger way to distinguish these models would be to use the predictions of @xmath95 .
in model i , @xmath95 is fixed to be 45@xmath261 up to @xmath295 corrections from @xmath92 , where as model ii prefers @xmath296 ( figs 2 and 3 ) in the experimentally allowed region .
this difference is directly due to the rigorousness of @xmath297 symmetry in the @xmath15 coupling from the tri - bimaximal ansatz ( @xmath298 ( model i ) ) where the @xmath15 coupling has @xmath299 symmetry . in model
i , the deviation from the maximal angle is related to @xmath92 , while in model ii , the deviation is related to @xmath0 .
the current best fit value of the atmospheric mixing angle ( @xmath300 @xcite ) is nearly the maximal mixing , and it implies the model i. however , the error is still large .
the accurate deviation from the maximal angle will be obtained in future three generation fit of the neutrino oscillations @xcite , and it will give us an important test for the tri - bimaximal ansatz .
the predictions for @xmath253 are similar with a small margin in these models , since the deviations from the tri - bimaximal angle ( @xmath301 ) are related to @xmath0 in all cases . in model
i , @xmath253 is predicted to be @xmath257 , where as , in model ii ( fig .
4 ) , @xmath253 is predicted to be @xmath302 ( case b1 ) , @xmath303 ( case b2 ) , once we include all the experimental bounds .
the pmns phase is assumed to be 0 or @xmath286 in the plot , and the general phase fit will change the predictions of the angle , especially for model ii .
the rank one yukawa coupling with * 10 * higgs field generates the features of flavor hierarchy , and rank 1 matrices can often appear in various ways ( flavor symmetry , discrete symmetry , and string models ) . in this section ,
we give an so(10 ) model , where the rank one ansatz used in our discussion of flavor emerges from a discrete symmetry .
when the direct couplings of chiral fermions with a higgs field are forbidden by a symmetry , and the effective yukawa couplings are generated by propagating vector - like matter fields , the rank of the effective yukawa matrix depends on the number of the vector - like fields . actually , when there are only one pair of vector - like matter fields as a flavor singlet , the effective yukawa matrix is rank 1 .
the model we assume has one extra vector - like pair of matter fields with mass slightly above the gut scale ( denoted by @xmath304 ) and three gauge singlet fields @xmath22 .
we add a @xmath206 discrete symmetry to the model under which the fields @xmath305 , and @xmath306 . the * 10 * -higgs field @xmath5 is invariant under this symmetry .
the gauge invariant yukawa superpotential under this assumption is given by @xmath307 when we give vevs @xmath308 , @xmath309 and @xmath310 are mixed .
the heavy vector - like fields , @xmath311 and a linear combination of @xmath309 and @xmath310 ( i.e. @xmath312 ) , and the effective operator below its scale and at the gut scale is given by : @xmath313 h \left[\sum_b y_b\psi_b\right].\end{aligned}\ ] ] this gives rise to a rank one @xmath14 coupling .
we note that it does not contradict the @xmath44 top yukawa coupling , when @xmath314 ( or @xmath315 ) .
if we let the @xmath11 higgs field transform like @xmath316 under @xmath206 , it can induce the @xmath15 coupling with rank three . another way to get mass matrix patterns of types in sec .
iv is to assume that there are three component flavon fields ( denoted by the dimensionless field @xmath317 ) which are representations of some internal flavor group and constrain their couplings to fermions by other symmetries . as an example , let us choose four flavon fields which transform as follows under a @xmath318 group : @xmath319 ; @xmath320 ; @xmath321 ; @xmath322 ; @xmath323 ; @xmath324 ; @xmath325 ; @xmath326 .
the invariant yukawa coupling under these symmetries is : @xmath327 in general the vevs of the @xmath328 fields align as in eq .
( 25 ) which can then lead to our type of rank one models . with suitable discrete symmetries , e.g. , @xmath329 @xcite ,
the flavon vevs can align as in eq .
( 26 ) , leading to models of the type considered here ( model i ) . when @xmath330 group is chosen instead of @xmath299 group such that @xmath331 ; @xmath332 ; @xmath333 ; @xmath334 ; @xmath335 ; @xmath336 ; @xmath337 ; @xmath338 , @xmath339 term is allowed in the @xmath340 coupling and model ii ( b1 ) can be considered .
more details on the flavon vev alignment with discrete symmetries and implications for rank one models is currently under investigation .
in conclusion , we have shown how a simple ansatz for the dominant yukawa coupling matrix in renormalizable so(10 ) models can lead to a unified understanding of the diverse quark and lepton flavor hierarchies .
we suggest this as a possible way to address the challenge of a unified description of quark - lepton flavor . we have not attempted in this note to derive our ansatz from any specific discrete symmetries , although we show a guideline to obtain a rank one form from a higher than gut scale theory
. this may be the next step towards a complete theory of flavor . within our rank one hypothesis ,
we have considered two classes of models which are nearly tri - bimaximal and point out that both a measurement of @xmath341 and the atmospheric mixing angle @xmath95 can distinguish between these models . in these models ,
the natural minimal value of @xmath341 is around 0.05 , whereas the maximal value can be larger than 0.15 .
this range of @xmath341 can be probed in the upcoming experiments .
the current estimate of @xmath341 using the 1.5@xmath342 excess of events in minos @xmath343 appearance channel @xcite and all other experimental data is @xmath344 ( 1@xmath342 ) @xcite .
the work of r. n. m. and y. m. is supported by the us national science foundation under grant no .
phy-0652363 and that of b. d. is supported in part by the doe grant de - fg02 - 95er40917 . for reviews ,
see r. n. mohapatra and a. y. smirnov , ann .
nucl . part .
sci . * 56 * , 569 ( 2006 ) [ hep - ph/0603118 ] ; g. altarelli and f. feruglio , new j. phys .
* 6 * , 106 ( 2004 ) [ hep - ph/0405048 ] ; g. altarelli , arxiv:0905.2350 [ hep - ph ] .
b. bajc , g. senjanovic and f. vissani , phys .
lett . * 90 * , 051802 ( 2003 ) [ hep - ph/0210207 ] ; h. s. goh , r. n. mohapatra and s. p. ng , phys .
b * 570 * , 215 ( 2003 ) [ hep - ph/0303055 ] ; phys .
d * 68 * , 115008 ( 2003 ) [ hep - ph/0308197 ] .
b. brahmachari and r. n. mohapatra , phys .
d * 58 * , 015001 ( 1998 ) [ hep - ph/9710371 ] ; t. fukuyama and n. okada , jhep * 0211 * , 011 ( 2002 ) [ hep - ph/0205066 ] ; c. s. aulakh , b. bajc , a. melfo , g. senjanovic and f. vissani , phys .
b * 588 * , 196 ( 2004 ) [ hep - ph/0306242 ] ; s. bertolini , m. frigerio and m. malinsky , phys .
d * 70 * , 095002 ( 2004 ) [ hep - ph/0406117 ] ; s. bertolini and m. malinsky , phys . rev .
d * 72 * , 055021 ( 2005 ) [ hep - ph/0504241 ] ; b. bajc , a. melfo , g. senjanovic and f. vissani , phys . lett .
b * 634 * , 272 ( 2006 ) [ hep - ph/0511352 ] .
h. s. goh , r. n. mohapatra and s. nasri , phys .
d * 70 * , 075022 ( 2004 ) [ hep - ph/0408139 ] ; t. fukuyama , a. ilakovac , t. kikuchi , s. meljanac and n. okada , eur .
j. c * 42 * , 191 ( 2005 ) [ hep - ph/0401213 ] ; j. math .
* 46 * , 033505 ( 2005 ) [ hep - ph/0405300 ] ; phys .
d * 72 * , 051701 ( 2005 ) [ hep - ph/0412348 ] ; b. bajc , a. melfo , g. senjanovic and f. vissani , phys .
d * 70 * , 035007 ( 2004 ) [ hep - ph/0402122 ] ; c. s. aulakh and a. girdhar , nucl .
b * 711 * , 275 ( 2005 ) [ hep - ph/0405074 ] ; c. s. aulakh and s. k. garg , nucl .
b * 757 * , 47 ( 2006 ) [ hep - ph/0512224 ] .
p. minkowski , phys .
b * 67 * ( 1977 ) 421 ; t. yanagida in _
workshop on unified theories , kek report 79 - 18 _ , p. 95
, 1979 ; m. gell - mann , p. ramond and r. slansky , _ supergravity _ , p. 315 .
amsterdam : north holland , 1979 ; s. l. glashow , _
1979 cargese summer institute on quarks and leptons _ , p. 687 .
new york : plenum , 1980 ; r. n. mohapatra and g. senjanovic , phys . rev .
lett , * 44 * ( 1980 ) 912 .
g. lazarides , q. shafi and c. wetterich , nucl .
b * 181 * , 287 ( 1981 ) ; j. schechter and j. w. f. valle , phys .
d * 22 * , 2227 ( 1980 ) ; r. n. mohapatra and g. senjanovic , phys . rev .
d * 23 * , 165 ( 1981 ) . l. j. hall , h. murayama and n. weiner , phys . rev
. lett . * 84 * , 2572 ( 2000 ) [ hep - ph/9911341 ] ; n. haba and h. murayama , phys .
d * 63 * , 053010 ( 2001 ) [ hep - ph/0009174 ] .
k. s. babu , b. dutta and r. n. mohapatra , phys .
d * 60 * , 095004 ( 1999 ) [ hep - ph/9812421 ] ; phys .
b * 458 * , 93 ( 1999 ) [ hep - ph/9904366 ] .
k. matsuda , y. koide and t. fukuyama , phys .
d * 64 * , 053015 ( 2001 ) [ hep - ph/0010026 ] .
k. matsuda , y. koide , t. fukuyama and h. nishiura , phys .
d * 65 * , 033008 ( 2002 ) [ erratum - ibid .
d * 65 * , 079904 ( 2002 ) ] [ hep - ph/0108202 ] .
k. s. babu and c. macesanu , phys .
d * 72 * , 115003 ( 2005 ) [ hep - ph/0505200 ] ; s. bertolini , t. schwetz and m. malinsky , phys .
d * 73 * , 115012 ( 2006 ) [ hep - ph/0605006 ] ; b. bajc , i. dorsner and m. nemevsek , jhep * 0811 * , 007 ( 2008 ) [ arxiv:0809.1069 [ hep - ph ] ] .
w. grimus and h. kuhbock , eur .
j. c * 51 * , 721 ( 2007 ) [ hep - ph/0612132 ] ; phys . rev .
d * 77 * , 055008 ( 2008 ) [ arxiv:0710.1585 [ hep - ph ] ] ; a. s. joshipura , b. p. kodrani and k. m. patel , phys .
d * 79 * , 115017 ( 2009 ) [ arxiv:0903.2161 [ hep - ph ] ] .
a @xmath103 unitary matrix @xmath345 can be expressed as @xmath346 where @xmath347 and @xmath348 are diagonal phase matrices and @xmath349 is a standard form of the unitary matrix which includes one phase and three mixings .
the usual expression of the ckm matrix is parameterized in the @xmath349 part since @xmath347 and @xmath348 are unphysical .
the phase matrix @xmath125 corresponds to the @xmath348 part of @xmath89 which is physical in the pmns matrix .
actually , one can find that the phase @xmath124 corresponds to the pmns phase ( which appears in the long baseline neutrino oscillation ) approximately up to a quadrant .
p. f. harrison , d. h. perkins and w. g. scott , phys .
b * 530 * , 167 ( 2002 ) [ hep - ph/0202074 ] ; l. wolfenstein , phys .
d * 18 * , 958 ( 1978 ) ; y. yamanaka , h. sugawara and s. pakvasa , phys .
d * 25 * , 1895 ( 1982 ) [ erratum - ibid .
d * 29 * , 2135 ( 1984 ) ] ; z. z. xing , phys .
b * 533 * , 85 ( 2002 ) [ hep - ph/0204049 ] .
s. f. king and g. g. ross , phys .
b * 520 * , 243 ( 2001 ) [ hep - ph/0108112 ] ; s. f. king , jhep * 0508 * , 105 ( 2005 ) [ hep - ph/0506297 ] ; i. de medeiros varzielas and g. g. ross , nucl .
b * 733 * , 31 ( 2006 ) [ hep - ph/0507176 ] .
i. de medeiros varzielas , s. f. king and g. g. ross , phys . lett .
b * 644 * , 153 ( 2007 ) [ hep - ph/0512313 ] ; i. de medeiros varzielas , s. f. king and g. g. ross , phys .
b * 648 * , 201 ( 2007 ) [ hep - ph/0607045 ] ; c. luhn , s. nasri and p. ramond , phys . lett .
b * 652 * , 27 ( 2007 ) [ arxiv:0706.2341 [ hep - ph ] ] . | diverse mass and mixing patterns between the quarks and leptons makes it challenging to construct a simple grand unified theory of flavor .
we show that so(10 ) susy guts with type ii seesaw mechanism giving neutrino masses provide a natural framework for addressing this issue .
a simple ansatz that the dominant yukawa matrix ( the * 10 * -higgs coupling to matter ) has rank one , appears to simultaneously explain both the large lepton mixings as well as the observed quark flavor hierarchy in these models .
a testable prediction of this ansatz is the neutrino mixing , @xmath0 , which should be observable in planned long baseline experiments . 1.0 in |
the weak neutral current was a critical prediction of the electroweak standard model ( sm ) @xcite . following its discovery in 1973 by the gargamelle and hpw experiments , there were generations of ever more precise wnc experiments , typically at the few % level .
these included pure weak @xmath2 and @xmath3 scattering processes , and weak - electromagnetic interference processes such as polarized @xmath4 or @xmath5 , @xmath6 ( hadron or charged lepton ) cross sections and asymmetries below the @xmath0 pole , and parity - violating effects in heavy atoms ( apv ) .
there were also early direct observations of the @xmath1 and @xmath0 by ua1 and ua2 .
the early 1990 s witnessed the very precise @xmath0-pole experiments at lep and the slc , in which the lineshape , decay modes , and various asymmetries were measured at the 0.1% level .
the subsequent lep 2 program at higher energies measured , searched for the higgs and other new particles , and constrained anomalous gauge self - interactions .
parallel efforts at the tevatron by cdf and dled to the direct discovery of the @xmath7 and measurements of and , while a fourth generation of weak neutral current experiments continued to search for new physics to which the ( more precise ) @xmath0-pole experiments were blind .
the program was supported by theoretical efforts in the calculation of qcd , electroweak , and mixed radiative corrections ; the expectations for observables in the standard model , large classes of extensions , and alternative models ; and global analyses of the data .
the precision program has established that the standard model ( sm ) is correct and unique to first approximation , establishing the gauge principle as well as the sm gauge group and representations ; shown that the sm is correct at loop level , confirming the basic principles of renormalizable gauge theory and allowing the successful prediction or constraint on @xmath8 , @xmath9 , and the higgs mass @xmath10 ; severely constrained new physics at the tev scale , with the ideas of unification favored over tev - scale dynamics or compositeness ; and yielded precise values for the gauge couplings , consistent with ( supersymmetric ) gauge unification .
even before the beginning of the @xmath0-pole experiments at lep and slc in 1989 , the precision program had established @xcite-@xcite : * global analyses of all data carried more information than the analysis of individual experiments , but care has to be taken with systematic and theoretical uncertainties . * the sm is correct to first approximation .
the four - fermion operators for @xmath11 , @xmath3 , and @xmath12 were uniquely determined , in agreement with the standard model , in model ( i.e. , gauge group ) independent analyses .
the @xmath1 and @xmath0 masses agreed with the expectations of the gauge group and canonical higgs mechanism , eliminating more complicated alternative models with the same four - fermi interactions as the standard model .
* qcd evolved structure functions and electroweak radiative corrections were necessary for the agreement of theory and experiment . * the weak mixing angle ( in the on - shell renormalization scheme ) was determined to be = 0.230 @xmath13 ; consistency of the various observations , including radiative corrections , required @xmath14 gev . *
theoretical uncertainties , especially in the @xmath15 threshold in deep inelastic weak charge current ( wcc ) scattering , dominated . *
the combination of wnc and wcc data uniquely determined the @xmath16 representations of all of the known fermions , i.e. , @xmath17 and @xmath18 , as well as the @xmath19 and @xmath20 components of the @xmath21 and @xmath15 @xcite .
in particular , the left - handed @xmath22 and @xmath23 were the lower components of @xmath16 doublets , implying unambiguously that the @xmath7 quark and @xmath24 had to exist .
this was independent of theoretical arguments based on anomaly cancellation ( which could have been evaded in alternative models involving a vector - like third family ) , and of constraints on from electroweak loops . *
the electroweak gauge couplings were well - determined , allowing a detailed comparison with the gauge unification predictions of the simplest grand unified theories ( gut ) .
ordinary was excluded ( consistent with the non - observation of proton decay ) , but the supersymmetric extension was allowed , `` perhaps even the first harbinger of supersymmetry '' @xcite .
* there were stringent limits on new physics at the tev scale , including additional @xmath25 bosons , exotic fermions ( for which both wnc and wcc constraints were crucial ) , exotic higgs representations , leptoquarks , and new four - fermion operators .
the lep / slc era greatly improved the precision of the electroweak program .
it allowed the differentiation between non - decoupling extensions to the sm ( such as most forms of dynamical symmetry breaking and other types of tev - scale compositeness ) , which typically predicted several % deviations , and decoupling extensions ( such as most of the parameter space for supersymmetry ) , for which the deviations are typically 0.1% .
the first phase of the lep / slc program involved running at the @xmath0 pole , @xmath26 and @xmath27 . during the period 1989 - 1995 the four lep experiments aleph , delphi , l3 , and opal at cern observed @xmath28 bosons .
the sld experiment at the slc at slac observed some @xmath29 events . despite the much lower statistics , the slc had the considerable advantage of a highly polarized @xmath30 beam , with @xmath31 75% .
there were quite a few @xmath0 pole observables , including : * the lineshape : @xmath32 and the peak cross section @xmath33 . * the branching ratios for @xmath34 and @xmath35 .
one could also determine the invisible width , @xmath36 , from which one can derive the number @xmath37 of active ( weak doublet ) neutrinos with @xmath38 , i.e. , there are only 3 conventional families with light neutrinos .
@xmath36 also constrains other invisible particles , such as light sneutrinos and the light majorons associated with some models of neutrino mass .
* a number of asymmetries , including forward - backward ( fb ) asymmetries ; the @xmath23 polarization , @xmath39 ; the polarization asymmetry @xmath40 associated with @xmath41 ; and mixed polarization - fb asymmetries . the expressions for the observables are summarized in @xcite , and the experimental values and sm predictions in table [ tab1 ] .
the precision of the @xmath0 mass determination was extraordinary for a high energy experiment .
these combinations of observables could be used to isolate many @xmath0-fermion couplings , verify lepton family universality , determine in numerous ways , and determine or constrain , , and .
lep and slc simultaneously carried out other programs , most notably studies and tests of qcd , and heavy quark physics .
@xmath42 [ gev ] & @xmath43 & @xmath44 & @xmath45 & @xmath46 @xmath47 [ gev ] & @xmath48 & @xmath49 & @xmath50 & @xmath51 @xmath52 [ gev ] & @xmath53 & @xmath54 & & @xmath36 [ mev ] & @xmath55 & @xmath56 & & @xmath57 [ mev ] & @xmath58 & @xmath59 & & @xmath60 [ nb ] & @xmath61 & @xmath62 & @xmath63 & @xmath63 @xmath64 & @xmath65 & @xmath66 & @xmath67 & @xmath68 @xmath69 & @xmath70 & @xmath66 & @xmath71 & @xmath67 @xmath72 & @xmath73 & @xmath74 & @xmath75 & @xmath76 @xmath77 & @xmath78 & @xmath79 & @xmath80 & @xmath80 @xmath81 & @xmath82 & @xmath83 & @xmath84 & @xmath84 @xmath85 & @xmath86 & @xmath87 & @xmath50 & @xmath88 @xmath89 & @xmath90 & & @xmath91 & @xmath80 @xmath92 & @xmath93 & & @xmath94 & @xmath95 @xmath96 & @xmath97 & @xmath98 & @xmath99 & @xmath100 @xmath101 & @xmath102 & @xmath103 & @xmath75 & @xmath50 @xmath104 & @xmath105 & @xmath106 & @xmath51 & @xmath107 @xmath108 & @xmath109 & @xmath110 & @xmath71 & @xmath111 & @xmath112 & & @xmath113 & @xmath114 @xmath115 & @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & & @xmath121 & @xmath122 & @xmath123 & & @xmath91 & @xmath80 @xmath124 & @xmath125 & & @xmath107 & @xmath126 @xmath127 & @xmath128 & & @xmath76 & @xmath50 & @xmath129 & & @xmath76 & @xmath51 @xmath130 & @xmath131 & @xmath132 & @xmath88 & @xmath88 @xmath133 & @xmath134 & @xmath135 & @xmath136 & @xmath136 @xmath137 & @xmath138 & @xmath139 & @xmath107 & @xmath107 lep 2 ran from 1995 - 2000 , with energies gradually increasing from @xmath140 to @xmath141 gev .
the principal electroweak results were precise measurements of the @xmath1 mass , as well as its width and branching ratios ( these were measured independently at the tevatron ) ; a measurement of @xmath142 , @xmath143 , and single @xmath1 , as a function of center of mass ( cm ) energy , which tests the cancellations between diagrams that is characteristic of a renormalizable gauge field theory , or , equivalently , probes the triple gauge vertices ; limits on anomalous quartic gauge vertices ; measurements of various cross sections and asymmetries for @xmath144 for @xmath145 and @xmath15 , in reasonable agreement with sm predictions ; a stringent lower limit of 114.4 gev on the higgs mass , and even hints of an observation at @xmath146 116 gev ; and searches for supersymmetric or other exotic particles . in parallel with the lep / slc program , there were precise ( @xmath147 1% ) measurements of atomic parity violation ( apv ) in cesium at boulder , along with the atomic calculations and related measurements needed for the interpretation ; precise new measurements of deep inelastic scattering by the nutev collaboration at fermilab , with a sign - selected beam which allowed them to minimize the effects of the @xmath15 threshold and reduce uncertainties to around 1% ; and few % measurements of @xmath148 by charm ii at cern .
although the precision of these wnc processes was lower than the @xmath0 pole measurements , they are still of considerable importance : the @xmath0 pole experiments are blind to types of new physics that do not directly affect the @xmath0 , such as a heavy @xmath25 if there is no @xmath149 mixing , while the wnc experiments are often very sensitive . during the same period there were important electroweak results from cdf and d@xmath150 at the tevatron , most notably a precise value for @xmath151 , competitive with and complementary to the lep 2 value ; a direct measure of , and direct searches for @xmath25 , @xmath152 , exotic fermions , and supersymmetric particles .
the tevatron program continues to the present day , and there have been recent precise measurements of the @xmath153 ( mller ) polarization asymmetry at slac ; polarization asymmetries in @xmath30-hadron scattering at mit - bates , mainz , and jefferson lab ; asymmetry measurements at the tevatron ; and measurements of the @xmath1 propagator and of @xmath0 exchange effects at hera .
many of these non-@xmath0 pole results are summarized in table [ tab2 ] .
@xmath8 [ gev ] & @xmath154 & @xmath155 & @xmath156 & @xmath157 @xmath151 ( @xmath158 ) & @xmath159 & @xmath160 & 1.4 & 1.7 @xmath151 ( lep ) & @xmath161 & & 0.0 & 0.5 @xmath162 & @xmath163 & @xmath164&@xmath165 & @xmath166 @xmath167 & @xmath168 & @xmath169 & 0.7 & 0.7 @xmath170 & @xmath171 & @xmath172 & 0.0 & 0.0 @xmath173 & @xmath174 & @xmath175 & 0.0 & 0.0 @xmath176 & @xmath177 & @xmath178 & 1.3 & 1.2 @xmath179 & @xmath180 & @xmath181 & 1.2 & 1.2 @xmath182 & @xmath183 & @xmath184 & 0.1 & 0.1 @xmath185 & @xmath186 & @xmath187 & 0.8 & 0.7 @xmath188 & @xmath189 & @xmath190 & 2.7 & 2.7 @xmath191 [ fs ] & @xmath192 & @xmath193 & @xmath194 & @xmath195 the effort required the calculation of the needed electromagnetic , electroweak , qcd , and mixed radiative corrections to the predictions of the sm .
careful consideration of the competing definitions of the renormalized was needed .
the principal theoretical uncertainty is the hadronic contribution to the running of @xmath196 from its precisely known value at low energies to the @xmath0-pole , where it is needed to compare the @xmath0 mass with the asymmetries and other observables .
the radiative corrections , renormalization schemes , and running of @xmath196 are further discussed in @xcite .
the lep electroweak working group ( lepewwg ) @xcite combined the results of the four lep experiments , and also those of sld and some wnc and tevatron results , taking proper account of common systematic and theoretical uncertainties .
much theoretical effort also went into the development , testing , and comparison of radiative corrections packages , and into the study of how various classes of new physics would modify the observables , and how they could most efficiently be parametrized .
* as can be seen in table [ tab1 ] most of the @xmath0-pole measurements are in excellent agreement with the standard model predictions using the parameters from the global best fit .
one exception is the lep value for @xmath96 , the forward - backward asymmetry into @xmath22 quarks , which is 2.5@xmath197 below the best fit expectation , and 2.0@xmath197 below the fit with @xmath198 gev . if not just a statistical fluctuation or systematic problem , @xmath96 could be a hint of new physics
however , any such effect should not contribute too much to @xmath77 .
the deviation is only around 3.9% , but if the new physics involved a radiative correction to the coefficient @xmath199 of , the change would have to be around 20% .
hence , the new physics would most likely be at tree - level type affecting preferentially the third generation .
examples include the decay of a scalar neutrino resonance @xcite , mixing of the @xmath22 quark with heavy exotics @xcite , and a heavy @xmath25 with family - nonuniversal couplings @xcite .
* there is a strong correlation between @xmath96 and the predicted higgs mass @xmath10 in the global fits , and in fact a fit to @xmath200 and @xmath42 alone yields a prediction @xmath201 gev @xcite .
in contrast , the sld polarization asymmetry @xmath40 combined with @xmath42 yields a lower value @xmath202 gev , with the other measurements closer to the average @xmath203 gev . it has been emphasized @xcite that if one eliminated @xmath96 from the fit ( e.g. , because it is affected by new physics ) then the global fit prediction for @xmath10 would be lowered , with the central value well below the lower limit of @xmath204 gev from the direct searches at lep 2 .
one resolution , assuming @xmath96 is due to new physics or a large fluctuation , is to invoke a supersymmetric extension of the standard model with light sparticles and second higgs doublet @xcite , which modify the radiative corrections and higgs constraints . *
the nutev collaboration at fermilab @xcite has reported the results of its deep inelastic measurements of @xmath205 .
they greatly reduce the uncertainty in the charm quark threshold in the charged current denominator by taking appropriate combinations of @xmath18 and @xmath206 .
they find a value for the on - shell weak angle @xmath207 of 0.2277(16 ) , which is 3.0@xmath197 above the global fit value of 0.2231(3 ) .
most of the difference is in the left handed neutral current coupling @xmath162 .
the discrepancy is reduced to @xmath208 if one incorporates ( as is done here ) the effects of the difference between the strange and antistrange quark momentum distributions , @xmath209 = 0.00196 \pm 0.00135 $ ] , from dimuon events , recently reported by nutev @xcite .
other possible effects that could contribute are large isospin violation in the nucleon sea , next to leading order qcd effects and electroweak corrections , and nuclear shadowing @xcite .
a full reanalysis of all deep inelastic data taking into account these issues and all of the uncertainties would be extremely useful .
possible new physics explanations of the nutev anomaly , such as a @xmath25 with specific couplings and neutrino mixing are reviewed in @xcite . *
the brookhaven @xmath210 experiment has reported a very precise value @xcite , leading to a world average @xmath211 .
the qed contribution has been calculated to four loops , and the predicted sm electroweak contribution is @xmath212 @xcite .
the largest uncertainty in the standard model prediction is from the hadronic vacuum polarization contribution , which has been estimated to two loops .
this can not be calculated perturbatively , but involves a dispersion relation that can be evaluated using experimental data from @xmath213 hadrons or hadronic @xmath23 decays .
a recent analysis @xcite indicates a @xmath214 discrepancy between the standard model prediction and the experimental value of @xmath215 when the @xmath216 data are used , but only a 0.9@xmath197 difference using the hadronic @xmath23 decays .
the issue is still not settled , but most recent authors advocate the @xmath216 value because the @xmath23 decays involve uncertainties from isospin violation .
there is also a small but hard to pin down uncertaintly from the hadronic light by light scattering diagrams .
+ because of the confused situation with the vacuum polarization , it is hard to know how seriously to take the discrepancy
. nevertheless , @xmath217 is more sensitive than the electron moment to most types of new physics , so it is important .
one obvious candidate for a new physics explanation would be supersymmetry @xcite , with relatively low masses for the relevant sparticles and high @xmath218 ( roughly , one requires an effective mass scale of @xmath219 ) .
there is a correlation between the theoretical uncertainty in the vacuum polarization and in the hadronic contribution to the running of @xmath196 to the @xmath0 pole @xcite , leading to a slight reduction in the predicted higgs mass when @xmath220 is included in the global fit assuming the standard model ( as is done here ) . * , the hadronic contribution to the running of @xmath196 up to the @xmath0-pole , introduces the largest theoretical uncertainty into the precision program , in particular to the relation between and the weak angle @xmath221 ( extracted mainly from the asymmetries ) .
the uncertainty is closely related to that in @xmath222 .
* the lep and slc @xmath0-pole experiments are the most precise tests of the standard electroweak theory , but they are insensitive to any new physics that does nt affect the @xmath0 or its couplings .
non-@xmath0-pole experiments are therefore extremely important , especially given the possible nutev anomaly .
the recent measurement of polarized mller scattering from slac @xcite is in agreement with the standard model and observes the running of the weak mixing angle in the scheme at the 6.4@xmath197 level , as can be seen in figure [ s2w ] .
an even more precise result is anticipated from the qweak polarized electron experiment at jefferson lab @xcite .
the running is observed to even lower @xmath223 in atomic parity violation @xcite .
+ -dis refer to atomic parity violation , the mller asymmetry , and @xmath224 deep inelastic scattering , respectively . from @xcite . ] * although the @xmath0-pole program has ended for the time being , there are prospects for future programs using the giga-@xmath0 option at a linear collider , which might yield a factor @xmath225 more events .
this would enormously improve the sensitivity @xcite , but would also require a large theoretical effort to improve the radiative correction calculations .
a global fit to all data contains more information than the individual experiments , but care must be taken with experimental and theoretical systematics and correlations . here
we report our most recent fits for the particle data group @xcite .
they utilize the fully program gapp for the radiative corrections @xcite .
the results are generally in good agreement with those of the lep electroweak working group @xcite ( which uses the on - shell renormalization scheme ) . however , the pdg fits use a more complete set of low energy data , which can be important for constraining certain types of new physics . as of november , 2007 ,
the result of the global fit was m_h & = & 77^+28_-22 , + m_t & = & 171.1 1.9 + _ s & = & 0.1217 0.0017 + ( m_z)^-1 & = & 127.909 0.019 + ^2_z & = & 0.23119 0.00014 + s^2_w & = & 0.22308 0.00030 + _
had^(5)(m_z ) & = & 0.02799 0.00014 , [ results ] with a good overall @xmath226 of @xmath227 .
the three values of the weak angle @xmath228 refer respectively to the , effective @xmath0-lepton vertex , and on - shell values @xcite .
the latter has a larger uncertainty because of a stronger dependence on the top mass .
the precision data alone yield @xmath229 gev from loop corrections , in impressive agreement with the direct tevatron value @xmath230 .
the fit actually uses the mass @xmath231 , which is @xmath232 10 gev lower , and converts to the pole mass at end . the tevatron value and the global fit value that it dominates is lower than the value obtained during run i , which leads to a lower predicted higgs mass .
the result @xmath233 for the strong coupling is somewhat above the previous world average @xmath234 , which includes other determinations , most of which are dominated by theoretical uncertainties @xcite .
this is due in part to the inclusion of the @xmath23 lifetime result @xcite .
( without it , one would obtain = @xmath235 from the @xmath0-pole data . )
the @xmath0-pole value has negligible theoretical uncertainty if one assumes the exact validity of the standard model , and is also insensitive to oblique ( propagator ) new physics .
however , it is very sensitive to non - universal new physics , such as those which affect the @xmath236 vertex . the @xmath23 decay value , on the other hand
, is less sensitive to new physics but is dominated by theoretical uncertainties . the prediction for the higgs mass from indirect data , @xmath237 gev ,
should be compared with the direct lep 2 limit @xmath238 gev @xcite .
there is no direct conflict given the large uncertainty in the prediction , but the central value is in the excluded region , as can be seen in figure [ higgspdf ] .
including the direct lep 2 exclusion results , one finds @xmath239 gev at 95% .
the theoretical range in the standard model is 115 gev @xmath240 750 gev , where the lower ( upper ) bound is from vacuum stability ( triviality ) . in the mssm ,
one has a theoretical upper limit @xmath241 gev , while can be as high as 150 gev in generalizations . in the decoupling limit
in which the second higgs doublet is much heavier the direct search lower limit is similar to the standard model .
however , the direct limit is considerably lower in the non - decoupling region in which the new supersymmetric particles and second higgs are relatively light @xcite .
enters the expressions for the radiative corrections logarithmically .
it is fairly robust to many types of new physics , with some exceptions . in particular , a much larger
would be allowed for negative values for the @xmath242 parameter or positive values for @xmath243 .
the predicted value would decrease if new physics accounted for the value of @xmath244 @xcite .
allowed regions in vs and the 90% cl global fit region from precision data , compared with the direct exclusion limits from lep 2 , from @xcite . ] vs. @xmath8 from direct ( tevatron and lep 2 ) and indirect data , and from the global fit .
also shown are the standard model expectations as a function of the higgs mass @xmath10 . from @xcite . ]
the @xmath245 or @xmath242 , @xmath246 and @xmath247 parameters describe the tree level effects of higgs triplets , or the loop effects on the @xmath1 and @xmath0 propagators due to such new physics as nondegenerate fermions or scalars , or chiral families ( expected , for example , in extended technicolor ) .
the current values are vs. @xmath242 produced by the lep electroweak working group @xcite shows larger values , @xmath248 and @xmath249 , based on the @xmath0-pole data and @xmath151 .
we almost exactly reproduce their values for the same inputs .
the lower values reported here are due to the inclusion of the low - energy data , such as atomic parity violation and neutrino scattering , as well as allowing @xmath9 to float and a different evaluation of @xmath250 . ] : s & = & -0.04 0.09 ( -0.07 ) + t & = & 0.02 0.09 ( + 0.09 ) for @xmath251 gev and @xmath252 , where these represent the effects of new physics only ( the and effects are treated separately ) .
the numbers in parentheses are the changes in the central values when one assumes @xmath253 gev instead .
similarly , @xmath254 and 114.4 gev @xmath255 215 gev ( for @xmath256 ) , implying limits on doublet mass splittings @xmath257 at 95% cl , where @xmath258 for leptons ( quarks ) .
the standard model higgs mass limits are weakened or can be evaded entirely for @xmath259 and @xmath260 , as can be seen in figure [ stcontour ] .
most types of new physics lead to positive contributions to @xmath242 , but negative values can be obtained due to higgs doublet or triplet loops or majorana fermions . and @xmath243 from a global fit to all of the data assuming @xmath252 , for @xmath261 and 1000 gev . also shown
are the @xmath262 constraints from individual inputs for @xmath251 gev . @xmath242 and @xmath243 are defined to include only the contributions of new physics , i.e. , @xmath8 and @xmath10 are treated separately . from @xcite . ]
the @xmath242 constraints strongly restrict the possibility of additional chiral fermions .
for example , a _ degenerate _
heavy family or mirror familiy is excluded at 6@xmath197 .
a nondegenerate additional family ( with a neutrino mass heavier than @xmath263 to evade the lineshape constraint ) can to some extent balance the effects of @xmath264 and @xmath265 @xcite , but is still significantly disfavored compared to the standard model fit @xcite . in the decoupling limit of supersymmetry , in which the sparticles and second higgs doublet are heavier than @xmath266 gev
, there is little effect on the precision observables , other than that there is necessarily a light sm - like higgs , consistent with the data .
there is little improvement on the sm fit , and in fact one can somewhat constrain the supersymmetry breaking parameters @xcite .
however , in the light higgs / sparticle limit the tension between the direct higgs searches and the indirect precision results is relaxed .
light sparticles could also account for a muon magnetic moment anomaly .
heavy @xmath25 bosons are predicted by many grand unified and string theories @xcite .
limits on the @xmath25 mass are model dependent , but are typically around @xmath267 gev from direct searches at the tevatron , with ( usually ) weaker limits from indirect constraints from wnc and lep 2 data .
the @xmath0-pole data severely constrains the @xmath149 mixing , typically @xmath268 .
a heavy @xmath25 would have many other theoretical and experimental implications @xcite .
precision data constrains mixings between ordinary and exotic fermions , large extra dimensions , new four - fermion operators , and leptoquark bosons @xcite .
gauge unification is predicted in guts and some string theories .
the simplest non - supersymmetric unification is excluded by the precision data . for the mssm , and assuming no new thresholds between 1 tev and the unification scale , one can use the precisely known @xmath196 and @xmath221 to predict @xmath269 and a unification scale @xmath270 gev @xcite .
the uncertainties are mainly theoretical , from the tev and gut thresholds , etc .
is high compared to the experimental value , but barely consistent given the uncertainties .
@xmath271 is reasonable for a gut ( and is consistent with simple seesaw models of neutrino mass ) , but is somewhat below the expectations @xmath272 gev of the simplest perturbative heterotic string models .
however , this is only a 10% effect in the appropriate variable @xmath273 .
the new exotic particles often present in such models ( or higher ka - moody levels ) can easily shift the @xmath273 and predictions significantly , so the problem is really why the gauge unification works so well .
it is always possible that the apparent success is accidental ( cf . ,
the discovery of pluto ) .
the precision @xmath0-pole , lep 2 , wnc , and tevatron experiments have successfully tested the sm at the 0.1% level , including electroweak loops , thus confirming the gauge principle , sm group , representations , and the basic structure of renormalizable field theory .
the standard model parameters @xmath274 , @xmath8 , and @xmath9 were precisely determined .
in fact , was successfully predicted from its indirect loop effects prior to the direct discovery at the tevatron , while the indirect value of , mainly from the @xmath0-lineshape , agreed with more direct qcd determinations .
similarly , and @xmath275 were constrained .
the indirect ( loop ) effects implied @xmath276 gev , while direct searches at lep 2 yielded @xmath277 gev , with a hint of a signal at 116 gev .
the combined direct and indirect data imply @xmath278 gev at 95% c.l .
this range is consistent with , but does not prove , the expectations of the supersymmetric extension of the sm ( mssm ) , which predicts a light sm - like higgs for much of its parameter space .
the agreement of the data with the sm imposes a severe constraint on possible new physics at the tev scale , and points towards decoupling theories ( such as most versions of supersymmetry and unification ) , which typically lead to 0.1% effects , rather than new tev - scale dynamics or compositeness ( e.g. , little higgs , dynamical symmetry breaking or composite fermions ) , which usually ( but not always ) imply deviations of several % , and often large flavor changing neutral currents .
finally , the precisely measured gauge couplings were consistent with the simplest form of grand unification if the sm is extended to the mssm .
this work was supported in part by dgapa unam contract papiit in115207 , by the friends of the ias , and by nsf grant pht-0503584 .
pl is also grateful to the conference organizers for support and to the aspen center for physics . for complete references , see j. erler and p. langacker , _ electroweak model and constraints on new physics _ , in w. m. yao _ et al .
_ [ particle data group ] , _
j. phys .
g _ * 33 * , 1 ( 2006 ) , 2008 edition ( in press ) , and _ phys .
_ d * 52 * , 441 ( 1995 ) ; _ precision tests of the standard electroweak model _
p. langacker ( singapore , world , 1995 ) ; p. langacker , m. luo and a. k. mann , _ rev .
_ * 64 * , 87 ( 1992 ) .
j. e. kim , p. langacker , m. levine and h. h. williams , _ rev .
phys . _ * 53 * , 211 ( 1981 ) .
u. amaldi _ et al .
_ , _ phys .
_ * d 36 * , 1385 ( 1987 ) .
g. costa , j. r. ellis , g. l. fogli , d. v. nanopoulos and f. zwirner , _ nucl .
_ * b297 * , 244 ( 1988 ) .
the lep collaborations aleph , delphi , l3 , opal , the lep electroweak working group and the sld heavy flavour group : s. schael , _ phys .
_ * 427 * , 257(2006 ) ; j. alcarez , hep - ex/0612034 and 0712.0929 ; for updates see http://lepewwg.web.cern.ch/lepewwg/ for reviews ,
see , t. kinoshita , _ j. phys .
g _ * 29 * , 9 ( 2003 ) ; m. davier and w.j .
marciano , _ ann .
nucl . part .
sci . _ * 54 * , 115 ( 2004 ) ; j.p .
miller , e. de rafael , and b.l .
roberts , _ rept .
* 70 * , 795 ( 2007 ) ; f. jegerlehner , _ acta phys .
_ * b38 * , 3021 ( 2007 ) ; and @xcite .
r. barate _
[ lep working group for higgs boson searches ] , phys .
b * 565 * , 61 ( 2003 ) [ arxiv : hep - ex/0306033 ] .
see , for example , v. a. novikov , l. b. okun , a. n. rozanov and m. i. vysotsky , jetp lett . *
76 * , 127 ( 2002 ) [ arxiv : hep - ph/0203132 ] ; g. d. kribs , t. plehn , m. spannowsky and t. m. p. tait , phys . rev .
d * 76 * , 075016 ( 2007 ) [ arxiv:0706.3718 [ hep - ph ] ] . | the results of high precision weak neutral current ( wnc ) , @xmath0-pole , and high energy collider electroweak experiments have been the primary prediction and test of electroweak unification .
the electroweak program is briefly reviewed from a historical perspective .
the current status and the implications for the standard model and beyond are discussed . |
the observation of the ridge , a long range pseudorapidity ( @xmath1 ) correlation at near - side @xcite , in dihadron correlations has motivated many theoretical investigations @xcite .
understanding the formation mechanism of the ridge may provide insight into the early dynamics of relativistic heavy ion collision .
we investigate the ridge using two tools : dihadron correlation relative to event plane and 3-particle @xmath1-@xmath1 correlations .
we discuss what have already been learned from previous studies @xcite and present new findings at this conference .
the ridge magnitude was found to decrease with trigger particle azimuth relative to event plane ( @xmath3 ) from in - plane ( @xmath4@xmath5@xmath4@xmath6@xmath7 ) to out - of - plane ( @xmath4@xmath5@xmath4@xmath6@xmath8 ) @xcite .
motivated by our data , chiu and hwa @xcite proposed the correlated emission model ( cem ) where the ridge is formed by correlated particle emission due to aligned jet propagation and medium flow and predicted asymmetric near - side @xmath0 correlations .
we present a study of this asymmetry from our data .
we found the ridge is asymmetric and the asymmetry peaks at @xmath5@xmath6@xmath9 .
the broadness of the ridge was found to be event - by - event from 3-particle @xmath1-@xmath1 correlations @xcite . by further analyzing their charge dependence , we may separate jet - like and ridge components without assuming the shape of the ridge in @xmath1 .
we find like - sign triplets are dominated by ridge and use this observation to separate the jet - like and ridge components .
we find jet - like correlation is narrow , and ridge is broad and approximately uniform in @xmath1 .
dihadron correlations are studied for trigger particles separately in @xmath5@xmath10@xmath7 and @xmath5@xmath11@xmath7 for 20 - 60@xmath12 au+au collisions at @xmath2 = 200 gev .
the @xmath13 ranges of the trigger and associated particles are 3@xmath11@xmath14@xmath114 gev / c and 1@xmath11@xmath15@xmath11 2 gev / c , respectively , and both within @xmath4@xmath16@xmath4@xmath111 .
the event plane angle is constructed from particles outside the @xmath13 range used in the correlation study .
the dihadron correlation signal obtained for each @xmath5 slice has a large background contribution from anisotropic flow ( @xmath17 and @xmath18 ) .
the construction and subtraction of this background is described in detail in ref .
the jet - like yield is obtained from the difference between the raw correlation in @xmath4@[email protected] , containing both jet - like component and the ridge , and that in @xmath4@[email protected] ( scaled by @xmath1 acceptance factor ) containing presumably only the ridge component .
the systematic uncertainties are largely canceled in the jet - like result . to obtain the near - side ridge
, we fit the away - side correlation within @xmath4@[email protected] to two gaussian of the same width , and then subtract the extrapolated fit to the near - side to remove away - side leakage to the ridge @xcite .
the remaining is the near - side ridge @xmath0 correlation .
we analyze the asymmetry parameter of the @xmath0 correlations by @xmath21 where @xmath22 . for 20 - 60@xmath12 au+au collisions .
the shaded bands represent the systematic uncertainties in the flow estimation . also shown by the black curve
is the prediction from cem . ]
figure [ fig1 ] shows @xmath23 for jet - like and ridge @xmath0 correlations as a function of @xmath5 for 20 - 60@xmath12 au+au collisions .
the ridge is found to be asymmetric with the maximum asymmetry appearing at @xmath5@xmath645@xmath24 , while jet is symmetric .
the ridge asymmetry is persistent for different associated @xmath13 .
the ridge asymmetry is qualitatively consistent with the cem prediction @xcite , suggesting that formation of the ridge may be due to jet - flow alignment .
however , we note that cem does not have dynamics depending on particle @xmath13 , nor does it have mechanisms to create correlations between particles separated by large @xmath1 . in general , the ridge @xmath0 asymmetry may be obtained by models which incorporate interplay between jet propagation and radial flow .
we have analyzed the 3-particle @xmath1-@xmath1 correlations data in @xmath25+au , 40 - 80@xmath12 and 0 - 12@xmath12 au+au collisions at @xmath2 = 200 gev @xcite . the trigger and associated particles
are restricted to @xmath4@xmath16@xmath4@xmath111 and @xmath4@xmath0@[email protected] .
the @xmath13 ranges are 1@xmath11@xmath15@xmath113@xmath11@xmath14@xmath1110 gev / c .
we further analyze different charge combinations in an attempt to separate the jet - like and ridge components and their structures .
the 3-particle correlation raw signal is obtained from all triplet of one trigger and two associated particles from the same event .
the signal is binned in @xmath26 and @xmath27 , the pseudorapidity differences between the associated particles and the trigger .
combinatorial background arises when one or neither of the two associated particles are correlated with the trigger particle besides flow correlation .
details about background construction , systematic uncertainties and correction factors can be found in ref .
the 2- and 3-particle correlation yields are corrected for the centrality- , @xmath13- and @xmath28-dependent reconstruction efficiencies for associated particles and the @xmath28-dependent efficiency for trigger particles , and are normalized per corrected trigger particle .
( @xmath4@xmath0@[email protected] ) corrected for @xmath1 acceptance in 0 - 12@xmath12 au+au collisions for 3@xmath11@xmath14@xmath1110 gev / c and 1@xmath11@xmath15@xmath113 gev / c .
the distributions are separated between like - sign ( square ) and unlike - sign ( triangle ) trigger - associated pairs . ]
figure [ fig2 ] shows the dihadron @xmath1 correlation between associated and trigger particles on the near - side ( @xmath4@xmath0@[email protected] ) in 0 - 12@xmath12 au+au collisions .
the correlation is separated for like - sign and unlike - sign pairs .
there are more unlike - sign than like - sign pairs at @xmath4@xmath1@xmath4@xmath60 .
this is due to the charge ordering in jet fragmentation . at larger @xmath4@xmath1@[email protected] ,
the unlike - sign and like - sign correlations are similar indicating the ridge particles have no charge dependence .
correlation in same - sign triplets should be dominated by ridge correlation because the probability of a jet fragmenting into 3 same - sign particles in our @xmath13 range is negligible .
therefore , we separate the charge ordered triplets in 3-particle @xmath1-@xmath1 correlations with like - sign triplets ( @xmath29 ) and like - sign associated pairs with an opposite - sign trigger particle ( @xmath30 ) .
-@xmath1 correlations for 3@xmath11@xmath14@xmath1110 gev / c and 1@xmath11@xmath15@xmath113 gev / c for @xmath29 in 0 - 12@xmath12 au+au collisions .
right : average 3-particle correlations signal for associated particles are of same charge sign as a function of @xmath31 on @xmath1-@xmath1 plane in 0 - 12@xmath12 au+au collisions .
systematic errors are shown in the shaded box due to background normalization ] figure [ fig3](left ) shows the background subtracted 3-particle @xmath1-@xmath1 correlations for @xmath29 in 0 - 12@xmath12 au+au collisions .
the average correlation signal seems to be uniform over @xmath1-@xmath1 region . to quantify this we study the average signal as a function of @xmath31 on @xmath1-@xmath1 plane , shown in fig [ fig3](right ) for @xmath32 ( @xmath29+@xmath30 ) and @xmath29 in 0 - 12@xmath12 au+au collisions .
@xmath32 contains both jet - like and ridge contributions while @xmath29 is dominated by ridge .
since the ridge at large @xmath1 is identical for like - sign and unlike - sign trigger - associated pairs in dihadron correlations , it should also be the same at large @xmath1 in @xmath29 and @xmath30 because the change from the dihadron to trihadron correlation is the addition of third particle of the same charge sign . on @xmath1-@xmath1 plane for 3@xmath11@xmath14@xmath1110 gev / c and 1@xmath11@xmath15@xmath113 gev / c in 0 - 12@xmath12 au+au collisions .
the curves are gaussian fits to the data . ]
therefore , the total jet - like component can be obtained by : jet = 3(@xmath30 - @xmath29 ) , where the factor of 3 takes into account the jet contribution from @xmath30 and @xmath33 charge combinations .
the ridge can then be obtained by subtracting the jet - like component from the total 3-particle correlation where all charge combinations are included . alternatively the ridge can be taken simply as 4 times @xmath29 .
the two methods yield consistent results .
figure [ fig4 ] shows the average signal of the separated ridge and jet - like components from 3-particle @xmath1-@xmath1 correlations as a function of @xmath31 on @xmath1-@xmath1 plane for 0 - 12@xmath12 au+au collisions .
the jet - like component is narrow and symmetric between @xmath28 and @xmath16 having a gaussian width @[email protected] . the separated ridge , while narrow in @xmath0 , is extremely broad with gaussian width @[email protected] .
the ridge is also consistent with a uniform distribution within our acceptance with reasonable @xmath36/ndf .
in summary , we have presented dihadron correlations relative to event plane for au+au 20 - 60@xmath12 collisions . while jet - like correlation is symmetric , ridge is found to be asymmetric when trigger particle azimuth is between in- and out - of - plane .
the asymmetry is qualitatively consistent with the correlated particle emission model prediction , suggesting the formation mechanism of the ridge may be due to alignment of jet propagation and medium flow .
we have used the charge ordering properties to separate the jet - like and ridge components via 3-particle @xmath1-@xmath1 correlations in 0 - 12@xmath12 au+au collisions .
this stems out of the observation that like - sign triplets are dominated by the ridge .
we found that the jet - like correlation is narrowly confined , and the ridge is broadly distributed being approximately uniform in @xmath1 in our measured acceptance .
except small angle correlation in azimuth , the ridge particles appear to be uncorrelated not only with the trigger particle , but also between themselves in @xmath1 .
n. armesto _ et al .
_ , _ phys .
lett . _ * 93 * , 242301 ( 2004 ) ; s. a. voloshin , _ phys .
b _ * 632 * , 490 ( 2006 ) ; m. strickland _ et al . _ , _ eur .
j. _ * a 29 * , 59 ( 2006 ) ; a. majumder _ et al .
_ , _ phys .
lett . _ * 99 * , 042301 ( 2007 ) ; e. shuryak , _ phys .
c _ * 76 * , 047901 ( 2007 ) ; c. y. wong , arxive:0712.3282 ( 2007 ) .
a. dumitru _ et al .
_ , _ nucl .
a _ * 91 * , 810 ( 2008 ) . c. b. chiu and r. c. hwa , _ phys .
c _ * 79 * , 034901 ( 2009 ) ; a. feng ( for star ) , _ j. phys .
g _ * 35 * , 104082 ( 2008 ) .
p. k. netrakanti ( for star ) , _ j. phys .
g _ * 35 * , 104010 ( 2008 ) .
j. konzer and f. wang , nucl . inst . and methods * a 606 * 713 ( 2009 )
. j. konzer ( for star ) , qm2009 , these proccedings . | we present new results on 2-particle azimuthal ( @xmath0 ) correlation relative to event plane and 3-particle pseudorapidity ( @xmath1 ) correlation at mid - rapidity in au+au collisions at @xmath2 = 200 gev , measured by the star experiment . while jet - like correlation is symmetric , ridge is found to be asymmetric when trigger particle azimuth is between in- and out - of - plane .
the charge ordering properties between associated and trigger particles are exploited to separate jet - like and ridge contributions in 3-particle @xmath1-@xmath1 correlations .
we found that like - sign triplets are dominated by ridge . the separated ridge , while narrow in @xmath0 , is extremely broad in @xmath1 .
the ridge particles are not only uncorrelated to the trigger particle in @xmath1 , but also uncorrelated between themselves . |
the investigation of nano - scale electronic devices , such as low - capacitance tunnel junctions or quantum dot systems , has always been motivated by the perspective of future applications . by now
several have been demonstrated , e.g. the use of single - electron transistors ( set ) as ultra - sensitive electro - meters and single - electron pumps . from the beginning it also appeared attractive to use these systems for digital operations needed in classical computation ( see ref . for a review ) .
obviously _ single - electron _ devices would constitute the ultimate electronic memory .
unfortunately , their extreme sensitivity makes them also very susceptible to fluctuations and random background charges . due to these problems
and the continuing progress of conventional techniques the future of set devices in _ classical _ digital applications remains uncertain .
the situation is different when we turn to elements for quantum computers .
they could perform certain calculations which no classical computer could do in acceptable times by exploiting the quantum mechanical coherent evolution of superpositions of states @xcite . here
conventional systems provide no alternative . in this context , ions in a trap , manipulated by laser irradiation are the best studied system @xcite .
however , alternatives need to be explored , in particular those which are more easily embedded in an electronic circuit . from this point of view nano - electronic devices appear particularly promising . the simplest choice ,
normal - metal single - electron devices are ruled out , since due to the large number of electron states involved different , sequential tunneling processes are incoherent
. ultra - small quantum dots with discrete levels , and in particular , spin degrees of freedom embedded in nano - scale structured materials are candidates .
they can be manipulated by tuning potentials and barriers @xcite . however , these systems are difficult to fabricate in a controlled way . more attractive
appear systems of josephson contacts , where the coherence of the superconducting state can be exploited and the technology is quite advanced .
macroscopic quantum effects associated with the flux in a squid have been demonstrated @xcite .
quantum extension of elements based on single flux logic have been suggested @xcite , and efforts are made to observe one of the elementary processes , the coherent oscillation of the flux between degenerate states @xcite .
we have suggested @xcite to use low - capacitance josephson junctions , where cooper pairs tunnel coherently while coulomb blockade effects allow the control of individual charges .
they provide physical realizations of quantum bits ( qubits ) with logical states differing by the number of cooper pair charges on an island .
these junctions can be fabricated by present - day technology .
the coherent tunneling of cooper pairs and the related properties of quantum mechanical superpositions of charge states have been discussed and demonstrated in experiments @xcite . in particular , in a recent experiment @xcite nakamura et al . have observed time - resolved coherent oscillations of charge in such josephson junction setup .
we concentrate here on the simplest design of the qubit , which is sufficient to introduce the ideas , and it is most easy for fabrication . it should be added that other ideas in this direction have been discussed @xcite . in ref .
@xcite an improved design has been proposed that relaxes requirements to the parameters of the circuit and makes the manipulation procedures more flexible . in section [ section_qubits ] we will introduce the system and show how single- and two - bit operations ( gates ) can be performed by the application of sequences of gate voltages . in section [ section_dephasing ] we analyze the influence of dissipation and fluctuations and conclude that for a proper choice of parameters the phase coherence time is sufficiently long to allow a large number of gate operations .
in addition to the controlled manipulations of qubits , quantum computation requires a quantum measurement process to read out the final state .
the requirements for both steps appear to contradict each other . during manipulations
the dephasing should be minimized , whereas a measurement should dephase the state of the qubit as fast as possible .
the option to couple the measuring device to the qubit only when needed is hard to achieve in nano - scale systems .
the alternative described in section [ section_measurement ] , is to couple a normal - state single - electron transistor capacitively to a qubit @xcite . during the manipulations the transport voltage of the set is turned off , and the set only acts as an extra capacitor . to perform the measurement the transport voltage is turned on .
the dissipative current through the transistor dephases the qubit and provides the read - out signal for the quantum state .
we describe this quantum measurement process by considering explicitly the time - evolution of the density matrix of the coupled system .
we find that the process is characterized by three time scales : a short dephasing time , the longer ` measurement time ' when the signal resolves the different quantum states , and finally the mixing time after which the measurement process itself destroys the information about the initial state .
similar nonequilibrium dephasing processes @xcite have recently been demonstrated experimentally @xcite . in section [ section_discussion ]
we discuss system parameters and suggest suitable sets which can be realized by present - day technology .
we further compare with related work .
the simplest josephson junction qubit is shown in fig .
[ bits_ideal]a .
it consists of a small superconducting island connected by a tunnel junction , with capacitance @xmath0 and josephson coupling energy @xmath1 , to a superconducting electrode .
an ideal voltage source , @xmath2 , is connected to the system via a gate capacitor @xmath3 ( fluctuation effects will be discussed later ) .
we choose a material such that the superconducting energy gap @xmath4 is the largest energy in the problem , larger even than the single - electron charging energy to be discussed below . in
this case quasi - particle tunneling is suppressed at low temperatures , and a situation can be achieved where no quasiparticle excitations exist on the island .
this ` parity effect ' has been established in experiments below a crossover temperature @xmath5 , where @xmath6 is the number of electrons in the system near the fermi energy @xcite . for a small enough island , @xmath7 is typically one order of magnitude smaller than the superconducting transition temperature . for the case that e.g. due to the initial preparation an unpaired excitation exists on the island , a channel should be provided for the quasiparticle to escape to normal parts of the system @xcite . ] .
= 14 0.8 cm in the following we will consider the situation where only cooper pairs tunnel in the superconducting junction . this system is described by the hamiltonian @xmath8 here , @xmath9 is the number ( operator ) of extra cooper pair charges on the island ( relative to some neutral reference state ) and the phase variable @xmath10 is its conjugate @xmath11 . the charging energy of the superconducting island is characterized by the scale @xmath12 , while the dimensionless gate charge , @xmath13 , acts as a control field .
we consider systems where @xmath14 is much larger than the josephson coupling energy , @xmath15 . in this regime
a convenient basis is formed by the charge states , parameterized by the number of cooper pairs @xmath9 on the island . in this basis the hamiltonian ( [ 1bit_hamiltonian_eqb ] ) reads @xmath16 for most values of @xmath17 the energy levels are dominated by the charging part of the hamiltonian .
however , when @xmath17 is approximately half - integer and the charging energies of two adjacent states are close to each other , the josephson tunneling mixes them strongly ( see fig .
[ parabolas ] ) .
= 10 0.8 cm we concentrate on the voltage interval near a degeneracy point of two charge states , say @xmath18 and @xmath19 . for the parameters chosen
, further charge states can be ignored , and the system ( [ 1bit_hamiltonian_eqb ] ) reduces to a two - state model , with hamiltonian which can be written in spin-@xmath20 notation in terms of pauli spin matrices @xmath21 as @xmath22 the charge states @xmath18 and @xmath19 correspond to the spin basis states @xmath23 and @xmath24 , respectively .
the charging energy @xmath25 , equivalent to a magnetic field in @xmath26-direction in the spin problem , is controlled by the gate voltage . for convenience
we can further rewrite the hamiltonian as @xmath27 where the mixing angle @xmath28 $ ] determines the direction of the effective magnetic field in the @xmath29-plane , and the energy splitting between the eigenstates is @xmath30 . at the degeneracy point ,
@xmath31 , it reduces to @xmath1 .
the eigenstates are denoted in the following as @xmath32 and @xmath33 .
for some chosen value of @xmath17 they are @xmath34 for later convenience we can rewrite the hamiltonian in the basis of eigenstates . to avoid confusion
we introduce a second set of pauli matrices , @xmath35 , which operate in the basis @xmath32 and @xmath33 , while reserving @xmath36 for the basis of charge states @xmath37 and @xmath38 . by definition
the hamiltonian then becomes @xmath39 . by changing the gate voltage we can perform the required one - bit operations ( gates ) .
if , for example , one chooses the idle state far away from degeneracy , the eigenstates @xmath32 and @xmath33 are close to @xmath40 and @xmath38 , respectively . for definiteness , we choose the eigenstates @xmath32 and @xmath33 at the idle point as logical basis states .
then switching the system suddenly to the degeneracy point for a time @xmath41 and then switching back produces a rotation in spin space , @xmath42 where @xmath43 .
depending on the value of @xmath41 , a spin flip can be produced , or , starting from @xmath32 , a superposition of states with any chosen weights can be reached .
this is exactly the way the experiments of nakamura et al .
@xcite were performed . at the same time , keeping the system at the idle point we achieve a phase shift between two logical states by the angle @xmath44 .
a different phase shift during the same time period @xmath41 can be achieved using a temporary change of @xmath17 by a small amount which changes the energy difference between the ground and excited states .
the example presented above provides an approximate spin flip for the situation where the idle point is far from degeneracy and @xmath15 . but a spin flip in the logical basis can also be performed exactly .
it requires that we switch from the idle point @xmath45 to the point where the effective magnetic field is orthogonal to the idle one , @xmath46 .
this changes the hamiltonian from @xmath47 to @xmath48 . to achieve that , the dimensionless gate charge @xmath17
should be increased by @xmath49 . in the limit
discussed above , @xmath50 , the operating point is close to the degeneracy , @xmath51 .
unitary @xmath52- and @xmath53-rotations described above are sufficient for all manipulations with a single qubit . by using a sequence ( of no more than three ) such elementary rotations we can achieve any unitary transformation of the qubit s state . in our discussion
so far elementary rotations should have been performed one immediately after another .
however , sometimes the quantum state should be kept intact for a certain time interval , for instance , while other qubits are manipulated during the computation . even in the idle state
@xmath54 , the energies of the two eigenstates are different .
hence their phases evolve relative to each other , which leads to the quantum mechanical ` coherent oscillations ' of a system which is in a superposition of eigenstates .
we have to keep track of this time dependence as is demonstrated by the following example ( for simplicity again on the approximate level ) .
imagine we try to perform a rotation by angle @xmath55 in the spin space , which we can do by a suitable choice of @xmath41 in ( [ 1bit_spin_flip ] ) , and after some time delay @xmath56 we perform the reverse operation .
the unitary transformation for this combined process is @xmath57 clearly the result depends on the intermediate time @xmath56 and differs from unity or a simple phase shift , unless @xmath58
. the time - dependent phase factors , arising from the energy difference in the idle state , can be removed from the eigenstates if all the calculations are performed in the interaction representation , with zero - order hamiltonian being the one at the idle point . in this way the information which is contained in the amplitudes of the qubit s states
is preserved .
however , there is a price for this simplification , namely the transformation to the interaction representation introduces additional time dependence in the hamiltonian during the operations .
thus a general 1-bit operation induced by switching at @xmath59 from @xmath45 to @xmath60 for some time @xmath41 is described by the unitary transformation @xmath61 this demonstrates that the effect of the operations depends not only on their time span but also on the moment when they start .
the transformation ( [ ugeneral ] ) is the elementary operation in the interaction picture in the sense that it can be performed in one step ( one voltage switching ) .
can we perform an arbitrary single - bit gate using such an operation ? to answer this question we note that all the unitary transformations of the two - state system form the group su(2 ) .
the group is three - dimensional , hence one needs in general three controllable parameters to obtain a given single - bit operation .
as is obvious from ( [ ugeneral ] ) , @xmath62 depends exactly on three parameters : @xmath60 , @xmath59 and @xmath41 .
it can be shown that _ any _ single - bit gate can be performed in one step if the voltage ( i.e. @xmath60 ) and the times @xmath59 , @xmath41 are chosen properly.- and @xmath26-rotations , as device - independent elementary single - qubit gates .
then , any single - bit gate in a quantum algorithm is described as a combination of three consecutive elementary rotations given by their angles ( e.g. euler angles
@xmath55 , @xmath63 , @xmath64 ) . to realize such a gate
@xmath65 we need to solve the equation @xmath66 to find the triple of @xmath60 , @xmath59 and @xmath41 .
since usually only a few different single - bit gates are used , corresponding triples can be calculated in advance . ] ( the dependence on @xmath59 is periodic with the period @xmath67 . hence the waiting time to a new operation is restricted to this period . ) as a final remark in this subsection , we notice that our choice of the logical basis is by no means unique . as follows from the preceding discussion , we can perform @xmath68- and @xmath26-rotations in this basis , which provides sufficient tools for any unitary operation . on the other hand ,
since we can perform _ any _ unitary transformation , we can as well perform @xmath68- and @xmath26-rotations in any basis of our two - dimensional system .
therefore , any basis can serve as the logical one .
the hamiltonian at the idle point is diagonal in the basis ( [ eigen_basis ] ) , while the controlled part of the hamiltonian , the charging energy favors the charge basis .
for this reason manipulation procedures for @xmath68- and @xmath26-rotations , in the interaction representation , are of comparable simplicity in all bases . the preparation procedure ( thermal relaxation at the idle point )
is easier described in the eigenbasis , while coupling to the meter ( see section [ section_measurement ] ) is diagonal in the charge basis .
so , there is no obvious choice for the logical states and it is a matter of convention . for quantum computation a register , consisting of a ( large ) number of qubits is needed , and pairs of qubits have to be coupled in a controlled way .
such two - bit operations ( gates ) are necessary , for instance , to create entangled states . for this purpose
we couple all qubits by one mutual inductor as shown in fig .
[ bits_ideal]b .
one can easily see that for @xmath69 the system reduces to a series of uncoupled qubits , while for @xmath70 they are coupled strongly .
finite values of @xmath71 introduce some retardation in the interaction .
the hamiltonian of this system , consisting of @xmath72 qubits and an oscillator formed by the inductance and the total capacitance of all qubits is @xcite @xmath73 here @xmath74 is the flux in the mutual inductor , @xmath75 is the flux quantum , and @xmath76 , the variable conjugated to @xmath74 , is related to the total charge on the gate capacitors of the qubits . for the oscillations in this @xmath77-circuit
the junction and gate capacitor of each qubit act in series , hence the relevant capacitance is @xmath78 .
the voltage oscillations of the @xmath77-circuit affect all the qubits equally , thus @xmath74 is coupled to each of the phases @xmath79 . the reduction factor
@xmath80 describes the screening of these voltage oscillations by the gate capacitors .
here we have chosen to express the coupling via a shift of the phase in the josephson coupling terms arising from the voltage oscillations .
this form is reminiscent of the usual description of squids , except that @xmath74 is a dynamic variable and its effect is reduced by the ratio of capacitances .
alternatively , we could have added the oscillating charges to the gate charges in the charging energy .
both forms are equivalent and related by a canonical transformation . this and a more detailed derivation of ( [ 2bit_hamiltonian_cosine ] ) has been presented in ref .
the oscillator described by the charge @xmath81 and flux @xmath74 with characteristic frequency @xmath82 produces an effective coupling between the qubits .
we choose parameters such that @xmath83 the first condition ( [ high_frequency_condition ] ) assures that the oscillator remains in its ground state at all relevant operation frequencies .
i.e. the logical operations on the qubits are not affected by excited states . the second condition ( [ small_fluctuation_assumption ] )
prevents the josephson coupling in ( [ 2bit_hamiltonian_cosine ] ) from being `` washed out '' by the fluctuations of @xmath74 .
since they are limited by @xmath71 , hence @xmath84 , this condition imposes only weak constraints on the parameters .
although the @xmath77-oscillator remains in its ground state , it provides an effective coupling between the qubits . to analyze this we expand in ( [ 2bit_hamiltonian_cosine ] ) the josephson coupling terms in @xmath74 and , because of ( [ small_fluctuation_assumption ] ) , neglect powers higher than linear .
the linear term is @xmath85 , where the current through the inductor is given by the sum of qubits contributions , @xmath86 the linear term and the unperturbed hamiltonian of the oscillator can be combined to a square , @xmath87 as long as the frequency of the oscillator is large compared to characteristic frequencies of the qubit s motion ( [ high_frequency_condition ] ) , one can use the adiabatic approximation and treat the slow qubit s variables ( @xmath79 ) as constant to find the energy levels of the oscillator . since the lowest level of the first two terms in the rhs of eq.([perturbedham ] ) , equal to @xmath88 , does not depend on @xmath89 , the correction to the ground state energy is given by the last term .
this term provides the effective coupling between the qubits @xmath90 where the energy scale is @xmath91 in the spin-@xmath20 notation @xmath92 and the interaction term becomes ( up to constant terms ) @xmath93 the ideal system would be one where the coupling between different qubits could be switched on and off , leaving the qubits uncoupled in the idle state and during 1-bit operations .
this option requires a more complicated design @xcite . with the present , simplest model the qubits
are coupled at all times .
but even in this case we can control the coupling in an approximate way by tuning the energies of the selected qubits in and out of resonance .
if @xmath94 is smaller or comparable to @xmath1 and the gates voltages @xmath95 are all different , such that no two qubits are near a degeneracy , the interaction ( [ 2bit_spin_coupling ] ) has only weak effects . in this case , the eigenstates of , say , a two - qubit system are approximately @xmath96 , @xmath97 , @xmath98 and @xmath99 , separated by energies larger than @xmath94 .
hence , the effect of the coupling is weak . if , however , a pair of these states is degenerate , even a weak coupling lifts the degeneracy , changing the eigenstates drastically .
for example , if @xmath100 , the states @xmath101 and @xmath102 are degenerate . in this case the correct eigenstates are : @xmath103 and @xmath104 with energy splitting @xmath94 between them . with the coupling ( [ 2bit_spin_coupling ] )
we are able to perform two - bit operations and create , e.g. , entangled states . in the idle state
we bias all qubits at different voltages .
then we suddenly switch the voltages of two selected qubits to be equal , bringing those two qubits for a time @xmath41 into resonance , and then we switch back .
the result is a two - bit operation , which is a rotation in the subspace spanned by @xmath105 : @xmath106 where @xmath107 .
the states @xmath108 merely acquire phase factors , relative to what they would acquire in the idle state . for all other qubits
the interaction remains a small perturbation .
similar to the situation with single - bit gates , the many - qubit states have time - dependent phase factors and the result of consecutive 2-bit gates depends on the waiting time between them .
again a transformation to the interaction representation makes this time dependence explicit .
one can show that any two - qubit operation can , in principle , be performed exactly .
however , the complexity of such exact manipulation grows with the number of qubits .
the improved design proposed in @xcite allows one to switch on and off the coupling between the qubits and to perform exact two - qubit operations in a simple way .
the two - bit gates ( [ 2bit_operation ] ) together with all one - bit gates , i.e. spin rotations ( [ 1bit_spin_flip ] ) and the phase shifts , constitute a _
universal set _ :
they are sufficient for all manipulations required for quantum computation . to check the experimental feasibility of the present proposal
it is necessary to estimate the time - span @xmath41 of the voltage pulses needed for typical single - bit operations .
we note that a reasonable value of @xmath1 is of the order of @xmath1091k .
it can not be chosen much lower , since the condition @xmath110 should be satisfied , and it should not be much larger since this would increase the technical difficulties associated with the time control .
the corresponding time - span is nevertheless short , @xmath111@xmath112ps , and is difficult to control in an experiment .
there exist , however , alternatives . on one hand , a controlled ramping of the coupling energy
provides a well defined , though non - trivial single - bit operation . since in general , any 1-bit gate can be performed with proper choice of 3 controlled parameters , a universal set of gates can be produced in this way .
another possibility is to follow the established procedures of spin resonance experiments .
e.g. a coherent spin rotation can be performed as follows : the system is moved adiabatically to the degeneracy point . then an ac voltage with frequency @xmath113
is applied . finally , the system is moved adiabatically back to the idle point .
the time - width of the ac - pulse needed , e.g. , for a total spin flip depends on the ac - amplitude , therefore it can be chosen much longer then @xmath114 .
unfortunately , in comparison to the sudden switching , the slow adiabatic approach allows only a reduced number of operations during the phase coherence time .
also the two - bit gates , instead of application of short voltage pulses , can be performed by moving the system adiabatically to a degeneracy point ( say @xmath115 ) , and then applying an ac voltage pulse in the antisymmetric channel @xmath116 . to improve
the performance of the device generalizations of the design of the qubits and their coupling may be useful .
we mention here one which has been described in ref . .
there we discuss a system where the josephson coupling can be tuned to zero as well .
this can be achieved by replacing each josephson junction by a squid with two junctions , and controlling the effective coupling by an applied magnetic flux . while this design is more complicated to fabricate , it has considerable advantages : ( i ) in the idle state
the hamiltonian can be tuned to zero ( @xmath117 ) , which removes the problem that the unitary transformations not only depend on the time span of their duration but also on the time @xmath59 when they are performed .
( ii ) the two - bit coupling is turned on only for those two qubits which have both @xmath118 .
( iii ) with only @xmath119 or @xmath120 non - zero , the unitary transformations depend in a simple way on the time integral of the corresponding hamiltonian . hence ramping the energies produces simple , well - defined gates . as a result of these improvements
the manipulations can be performed with much higher accuracy .
the idealized picture outlined above has to be extended to account for the possible dissipation mechanisms causing decoherence and relaxation .
we focus on the dissipation and fluctuations which originate from the circuit of the voltage sources . in fig .
[ bits_circuit ] the equivalent circuit of a qubit coupled to an impedance @xmath121 is shown .
the latter is characterized by intrinsic voltage fluctuations ( between its terminals when disconnected from the circuit ) , with spectrum @xmath122 when @xmath121 is embedded in a circuit , similar to that of fig .
[ bits_circuit ] but with @xmath123 , the voltage fluctuations between the terminals of @xmath121 are characterized by a modified spectrum : @xmath124 here @xmath125^{-1}$ ] is the total impedance between the terminals of @xmath121 .
= 12 0.8 cm following caldeira and leggett @xcite we model the dissipative element @xmath121 by a bath of harmonic oscillators described by the hamiltonian @xmath126 \ . \label{bath_hamiltonian_xp}\ ] ] it is assumed that the voltage between the terminals of @xmath121 is given by @xmath127 and the spectral function @xmath128 is chosen to reproduce the fluctuations spectrum ( [ dvdv_disconnected ] ) , @xmath129 .
then , we embed the element @xmath121 in the qubit s circuit and we use the kirchhof constraints to derive the hamiltonian of the whole system . taking into account the possibility of a time - dependent external voltage @xmath130
, we arrive at the resulting hamiltonian @xmath131 \sum_\alpha \tilde\lambda_\alpha \tilde x_\alpha \ , \label{disshamnewbath}\end{aligned}\ ] ] where the tilde - marking of the bath - related quantities reflects the fact that the bath has been modified and its spectral function corresponds now to the fluctuations spectrum ( [ dvdv_connected ] ) , @xmath132 . shifting the origins of all the oscillators @xmath133 , where @xmath134 is the time independent part of the external voltage , we get a more familiar form of the hamiltonian : @xmath135 \sum_\alpha \tilde\lambda_\alpha \tilde x_\alpha \ , \label{disshamusualform}\end{aligned}\ ] ] to be specific we will concentrate in the following on the fluctuations due to an ohmic resistor @xmath136 in the bias voltage circuit . in general ,
the environment has two effects : inelastic energy relaxation and dephasing , both characterized by their respective time scales , @xmath137 and @xmath138 . to illustrate this point we consider a qubit prepared in a superposition of eigenstates @xmath139 with non - vanishing coefficients @xmath140 and @xmath141
this corresponds to an initial density matrix @xmath142 . in this case
, one question is how fast the diagonal elements of the density matrix relax to their thermal equilibrium values .
this relaxation is determined by @xmath143 .
the second question is how fast the off - diagonal elements vanish , which is governed by the rate @xmath144 . in general , the two rates are not equal .
both inelastic transitions and dephasing were addressed in the context of spin - boson models , in particular for a two - level system with purely ohmic dissipation @xcite .
our system ( [ disshamusualform ] ) reduces to a spin - boson model in the limit where only two charge states need to be considered , with a harmonic oscillator bath coupling to @xmath145 . it still differs from an ohmic model due to the frequency dependence in the function @xmath146 .
however , for the relevant frequencies , @xmath147 , the physics is dominated by the resistor .
hence we can take over the results from the model calculations of refs . for purely ohmic environments .
the strength of the fluctuations and , hence , the relaxation and dephasing effects are controlled by the resistance of the circuit .
it turns out that it has to be measured in units of the quantum resistance @xmath148k@xmath149 .
hence for low circuit resistances in the range of @xmath150 we can expect a weak effect .
furthermore , in our system the effect of fluctuation is reduced due to the weak capacitive coupling of the qubit to the circuit . indeed , as is apparent from ( [ disshamusualform ] ) the coupling of @xmath151 to the qubit s charge @xmath9 involves the ratio @xmath152 .
thus , the relevant parameter characterizing the effect of the voltage fluctuations is @xmath153 . in ref .
only the unbiased case of the spin - boson model , corresponding to the qubit at the degeneracy point @xmath154 , has been analyzed rigorously in the limit of low dissipation ( the only limit relevant for the quantum computation ) .
the biased case of the spin - boson model at low dissipation ( i.e. the qubit out of the degeneracy ) was treated in ref . .
the two rates , expressed in terms of the mixing angle introduced in ( [ eigen_basis ] ) , are @xmath155 and @xmath156 we note that out of the degeneracy the dephasing rate acquires a component proportional to the temperature .
it should also be noted that the relaxation rate ( [ gamma_r_weiss ] ) may be obtained in a much simpler way @xcite using the so called `` @xmath157 '' theory discussed in refs .
@xcite .
thus , the diagonal and off - diagonal elements in the basis of the qubit s eigenstates relax with different rates described above , while the off - diagonal elements carry also an oscillating phase factor . the last is translated in the charge basis to the ` coherent charge oscillations ' , which are observed in the quantity @xmath158 . in the absence of dissipation this quantity oscillates coherently around some average value . with dissipation , this average value relaxes to equilibrium with rate @xmath143 , while the oscillations decay with rate @xmath159 @xcite . the factors @xmath160 and @xmath161 in eqs .
( [ gamma_r_weiss],[gamma_phi_weiss ] ) indicate the nature of different contributions to the relaxation and dephasing rates . in the basis of the eigenstates ( [ eigen_basis ] ) the qubit s
part of the hamiltonian ( [ disshamusualform ] ) is diagonal .
however , the circuit electromagnetic fluctuations couple to the charge .
hence , in the notation introduced after ( [ eigen_basis ] ) we have @xmath162 since the first term commutes with the qubit s hamiltonian , only the second one in ( [ fluctuation_hamiltonian ] ) contributes to the transitions between the eigenstates .
this explains the factor @xmath160 in ( [ gamma_r_weiss ] ) . the first term in ( [ fluctuation_hamiltonian ] ) , while ineffective for transitions , makes the energy difference between the qubit s levels fluctuate .
therefore , the off - diagonal elements of the density matrix acquire a random phase .
the last process is `` pure '' dephasing .
it occurs even when transitions are suppressed ( e.g. when @xmath163 ) .
this particular case ( @xmath163 , @xmath164 ) is quite simple and we can reproduce eq .
( [ gamma_phi_weiss ] ) in an exact calculation , examining also the zero temperature limit . to do this
, we note that in the two state approximation the off - diagonal element of the reduced density matrix is given by @xmath165 . to calculate it we perform a canonical transformation @xmath166 , @xmath167 ,
after which the hamiltonian ( [ disshamusualform ] ) becomes : @xmath168 \ .
\label{disshampuredephasing}\end{aligned}\ ] ] the josephson and the last terms are zero in our case and the evolution of \{@xmath169 } is decoupled from the bath . in the two state approximation ( @xmath170 ) the equations of motion read
: @xmath171 , @xmath172/dt = -i e_{\rm ch } ( v_{\rm qb,0 } ) \exp(-i\theta')$ ] , where @xmath173 was introduced in eq .
( [ magnetic_hamiltonian ] ) .
thus , introducing @xmath174 , we get @xmath175 ) \rangle \nonumber \\ & & = \langle \exp(i\phi(t))\ ,
\exp(-i\left[\theta'(0)+e_{\rm ch}(v_{\rm qb,0})\,t\right ] ) \rangle \nonumber \\ & & = \sigma_{0,1}(0)\,e^{-ie_{\rm ch}(v_{\rm qb,0})\,t}\ , \langle e^{i\phi(t)}\,e^{-i\phi(0 ) } \rangle \ .\end{aligned}\ ] ] the correlator @xmath176 was studied extensively by many authors @xcite .
it is equal to @xmath177 , where @xmath178 -i\sin(\omega t)\right ] \ . \end{aligned}\ ] ] since @xmath179 , we may roughly substitute the bath s spectrum by a purely ohmic one with a cut - off set at @xmath180 .
then , at non - zero temperature , @xmath181 for @xmath182 , and we reproduce eq .
( [ gamma_phi_weiss ] ) in the limit @xmath163 . at zero temperature @xmath183 for @xmath184 .
thus , the fact that the `` pure '' dephasing rate vanishes at zero temperature does not mean that there is no dephasing at all .
the decay of the off - diagonal element of the reduced density matrix is just non - exponential but rather algebraic ( see refs . ) .
we conclude that the dephasing of the initial quantum state of the qubit is caused by two different processes : the dissipative transitions between the eigenlevels , and the fluctuations of the energy difference between the levels .
the off - diagonal elements of the density matrix are suppressed by both of them .
while the first process survives at low temperature , the second one produces the dephasing rate proportional to the temperature .
these results should be taken into account when selecting the optimum idle state for the qubit .
if the temperature is low , @xmath185 , the best choice is obviously far from the degeneracy point ( @xmath186 ) .
also at higher temperatures this regime has a longer coherence time than the degeneracy point , although only by a numerical factor . in all cases
it is pure dephasing , rather than inelastic transitions which limits the coherence at the idle point . in summary ,
the dephasing rate is small if the effective resistance of the circuit is low compared to the quantum resistance , @xmath187 .
furthermore , a low gate capacitance @xmath3 reduces the coupling of the qubit to the environment .
hence , with suitable parameters ( @xmath188 ) at low temperatures the number of operations which can be performed before the environment destroys the phase coherence may be as large as @xmath189@xmath190 .
the value of the phase coherence time in this case is of the order of 10100 ns .
coherence times of this order of magnitude have been observed in experiments on quantum dots @xcite .
longer phase coherence times would be achieved if the coupling of the qubit to the noisy environment could be further reduced , without weakening the coupling to the common inductor @xmath71 which provides the 2-bit coupling .
the model discussed above is not ideal since the ohmic resistor with johnson - nyquist noise is assumed to be in the immediate vicinity of the qubit , reduced in its effect only by a low gate capacitor .
a suitable design with a combination of superconducting leads and filters can drastically improve the situation .
if the physical dephasing effects are reduced , the more serious will be the errors related to imperfections in the time control or imprecise parameters and coupling energies , e.g. , non - vanishing 2-bit couplings during the idle periods .
if the combination of all these effects is sufficiently reduced , allowing for @xmath190@xmath191 coherent manipulations steps , then eventually the remaining errors can be corrected by suitable codes @xcite .
it appears from our analysis that this goal can be reached with josephson junction qubits .
the read - out of the state of the qubit requires a quantum measurement process .
since the relevant quantum degree of freedom is the charge of the qubit island , the natural choice of measurement device is a single - electron transistor ( set ) .
this system is shown in fig .
[ circuit ] .
the left part is the qubit , with state characterized by the number of extra cooper pairs , @xmath9 , on the island , and controlled by its gate voltage , @xmath2 .
the right part shows a normal island between two normal leads , which form the set .
it s charging state is characterized by the number of extra single - electron charges on the middle island , @xmath72 .
it is controlled by gate and transport voltages , @xmath192 and @xmath193 , and further , due to the capacitive coupling to the qubit , by the state of the latter .
a similar setup has been studied in the experiments of refs . with the purpose to demonstrate that the ground state of a single cooper pair box is a coherent superposition of different charge states .
we will discuss the relation of these experiments to our proposal below . during the quantum manipulations of the qubit
the transport voltage @xmath194 across the set transistor is kept zero and the gate voltage of the set is chosen to tune the island away from degeneracy points . therefore no dissipative currents flow in the system , and the transistor merely modifies the capacitances of the system . to perform a measurement one tunes the set by @xmath192 to the vicinity of its degeneracy point and
applies a small transport voltage @xmath195 .
the resulting normal current through the transistor depends on the charge configuration of the qubit , since different charge states induce different voltages on the middle island of the set transistor . in order to investigate
whether the dissipative current through the set transistor allows us to resolve different quantum states of the qubit , we have to discuss various noise factors , including the shot noise associated with the tunneling current and the measurement induced transitions between the states of the qubit . for this purpose
we analyze the time evolution of the density matrix of the combined system .
we find that for suitable parameters , which can be realized experimentally , the dephasing by the passive set is weak .
when the transport voltage is turned on the dephasing is fast , and the current through the transistor after a transient period provides a measure of the state of the qubit . at still longer times
the dynamics of the set destroys the information of the quantum state to be measured .
= 18 the hamiltonian of the combined system @xmath196 contains the charging energy , the terms describing the microscopic degrees of freedom of the metal islands and electrodes , and the tunneling terms , including the josephson coupling .
the charging term is a quadratic form in the variables @xmath9 and @xmath72 .
i.e. for the system shown in fig .
[ circuit ] it is @xmath197 the charging energy scales @xmath14 , @xmath198 and @xmath199 are determined by the capacitances between all the islands . introducing @xmath200
we can write them as @xmath201 here we have assumed that the two junctions of the set have equal capacitances @xmath202 , and the approximate results refer to the limit @xmath203 , which we consider useful ( see below )
. the effective gate voltages @xmath204 and @xmath205 depend in general on all voltages @xmath2 , @xmath192 , and the voltages applied to both electrodes of the set . however , for a symmetric setup ( equal junction capacitances ) and symmetric distribution of the transport voltage , @xmath193 , between both electrodes of the set ( as shown in fig .
[ circuit ] ) , @xmath204 and @xmath205 are controlled only by the two gate voltages , @xmath206 the microscopic terms describe noninteracting electrons in the two leads and on the middle island of the set transistor @xmath207 the index @xmath208 labels transverse channels including the spin , while @xmath209 labels the wave vector within one channel .
similar terms describe the electrode and island of the qubit ; however , for the superconducting non - dissipative element the microscopic degrees of freedom can be integrated out @xcite , resulting in the `` macroscopic '' quantum description presented in sect .
ii . in this limit
the tunneling terms reduce to the josephson coupling @xmath210 , expressed in a collective variable describing the coherent transfer of cooper pairs in the qubit , @xmath211 .
the normal - electron tunneling in the set transistor is described by the standard tunneling hamiltonian , which couples the microscopic degrees of freedom , @xmath212 to make the charge transfer explicit , ( [ tunnel_hamiltonian ] ) displays two `` macroscopic '' operators , @xmath213 and @xmath214 .
the first one describes changes of the charge on the transistor island due to the tunneling : @xmath215 .
it may be treated as an independent degree of freedom if the total number of electrons on the island is large .
we further include the operator @xmath214 which describes the changes of the charge in the right lead .
it acts on @xmath216 , the number of electrons which have tunneled through the set transistor , @xmath217 .
since the chemical potential of the right lead is controlled , @xmath216 does not appear in any charging part of the hamiltonian . however , we have to keep track of it , since it is the measured quantity , related to the current through the set transistor . in equilibrium ,
i.e. when the set is kept in the state @xmath218 , the qubit s dynamics is described by the same hamiltonian as discussed in previous sections , @xmath219 , where @xmath220 .
we recall that in the limit where only the lowest energy charge states @xmath18 and @xmath19 are relevant , it reduces to a two state quantum system . in the basis of eigenstates ( [ eigen_basis ] ) , @xmath32 and @xmath33 , which are expressed in terms of the mixing angle @xmath60 , where @xmath221 , it becomes @xmath222 , where @xmath223 and @xmath224 . in this basis
the number operator @xmath9 becomes non - diagonal , @xmath225 for the following discussion we choose @xmath17 away from the degeneracy point , which combined with @xmath226 implies @xmath227 . the mixed term in ( [ charging_energy ] )
provides the interaction hamiltonian . with @xmath9 given by ( [ n_operator ] )
it becomes @xmath228 plus an extra term , @xmath229 , which together with further terms is collected in the hamiltonian of the set transistor , @xmath230 the transistor s gate charge became @xmath231 .
the total hamiltonian thus reads @xmath232 the total system composed of qubit and set is described by a total density matrix @xmath233 , which we can reduce , by taking a trace over the microscopic states of the left and right leads and of the island , to @xmath234 . in general , this reduced density matrix @xmath235 is a matrix in the index @xmath236 , which stand for the quantum states of the qubit @xmath32 or @xmath33 , in @xmath72 , and in @xmath216 . in the following
we will assume that initially as a result of previous quantum manipulations the qubit is prepared in the quantum state @xmath237 , and at time @xmath238 we switch on a transport voltage to the single - electron transistor .
we then proceed to further reduce the density matrix in two ways which provide complementary information about the measuring process .
the first , widely used procedure @xcite is to trace over @xmath72 and @xmath216 .
this yields a reduced density matrix of the qubit @xmath239 . just before the measurement
, it is in the state @xmath240 the questions then arise how fast the off - diagonal elements of @xmath241 vanish , i.e. how fast is the dephasing , and how fast the diagonal elements change their values after the set is switched to the dissipative state .
these questions are the same as in our discussion of fluctuation effects in section [ section_dephasing ] .
however , the coupling to the dissipative set in general shortens these times .
this description is enough when one is interested in the quantum properties of the measured system , i.e. the qubit only , and the measuring device serves merely as a source of dephasing @xcite .
it does not tell us much about the quantity measured in an experiment , namely the current flowing through the set .
the second procedure , pursued in the following , is to evaluate the probability distribution of the number of electrons @xmath216 which tunnel through the set during time @xmath242 , @xmath243 this distribution provides the information about the experimentally accessible quantity during the measurement process . at @xmath238
no electrons have tunneled , so @xmath244 .
then the peak of the distribution moves in positive @xmath216-direction and , simultaneously , it widens due to shot noise . since two states of the qubit correspond to different tunneling currents , and hence shift velocities in @xmath216-direction
, one may hope that after some time the peak splits into two . if after sufficient separation of the two peaks their weights are still close to @xmath245 and @xmath246 , a good quantum measurement can be performed by measuring @xmath216 .
after a longer time further processes destroy this idealized picture .
the two peaks transform into a broad plateau , since transitions between the qubit s states are induced by the measurement .
therefore , one should find an optimum time for the measurement , such that , on one hand , the two peaks are separate and , on the other hand , the induced transitions have not yet influenced the process . in order to describe this
we have to analyze the time evolution of the reduced density matrix quantitatively .
the time evolution of the density matrix leads to bloch - type master equations with coherent terms .
examples have recently been analyzed in contexts similar to the present @xcite . in ref .
a diagrammatic technique has been developed which provides a formally exact master equation as an expansion in the tunneling term @xmath247 , while all other terms constitute the zeroth order hamiltonian @xmath248 , which is treated exactly .
the time evolution of the reduced density matrix is given by @xmath249 .
the propagator @xmath250 can be expressed in a diagrammatic form and finally summed up in a way reminiscent of a dyson equation .
examples are shown in fig .
[ dyson ] . in contrasts
to ordinary many - body expansions , since the time dependence of the density matrix is described by a forward and a backward time - evolution operator , there are two propagators , which are represented by two horizontal lines ( keldysh contour ) .
the two bare lines describe the coherent time evolution of the system .
they are coupled due to the tunneling in the set , which is treated as a perturbation .
the sum of all distinct transitions defines a ` self - energy ' diagram @xmath251 . below we will present the rules how to calculate @xmath251 and present a suitable approximate form .
the dyson equation is equivalent to a ( generalized ) master equation for the density matrix , which reads @xmath252 = \int_0^t dt ' \hat{\sigma}(t ' ) \sigma(t - t ' ) \ .\ ] ] in the present problem the density matrix is a matrix @xmath253 in all three indices @xmath236 , @xmath72 , and @xmath216 , and the ( generalized ) transition rates due to single - electron tunneling processes ( in general of arbitrary order ) , @xmath254 , connect these diagonal and off - diagonal states .
= 3.5 0.8 cm the transition rates can be calculated diagrammatically in the framework of the real - time keldysh contour technique .
we briefly review the rules for their evaluation ; for more details including the discussion of higher order diagrams we refer to ref . .
typical diagrams , which will be analyzed below , are displayed in fig .
[ diagram_gain ] and [ diagram_loose ] .
again the horizontal lines describe the time evolution of the system governed by the zeroth order hamiltonian @xmath255 .
their properties will be discussed below .
the directed dashed lines stand for tunneling processes , in the example considered the tunneling takes place in the left junction .
according to the rules the dashed lines contribute the following factor to the self - energy @xmath251 , @xmath256 \over \sinh^2\left[{\pi k_{\rm b}t\over\hbar } \left(t - t ' \pm i\delta\right)\right ] } \ , \ ] ] where @xmath257 is the dimensionless tunneling conductance , @xmath258 is the electro - chemical potential of the left lead , and @xmath259 is the high - frequency cut - off , which is at most of order of the fermi energy .
the sign of the infinitesimal term @xmath260 depends on the time - direction of the dashed line .
it is negative if the direction of the line with respect to the keldysh contour coincides with its direction with respect to the absolute time ( from left to right ) , and positive otherwise .
for example the right part of fig .
[ diagram_gain ] should carry a minus sign , while the left part carries a plus sign .
furthermore , the sign in front of @xmath261 is negative ( positive ) , if the line goes forward ( backward ) with respect to the absolute time .
the first order diagrams are multiplied by @xmath262 if the dashed line connects two points on different branches of the keldysh contour .
= 5 = 5 the horizontal lines describe the time - evolution of the system between tunneling processes . for an isolated transistor island
they reduce to simple exponential factors @xmath263 , depending on the charging energy of the system . in the present case , however , where the island is coupled to the qubit we have to account for the nontrivial time evolution of the latter .
for instance , the upper line in the left part of fig .
[ diagram_gain ] corresponds to @xmath264 , while the lower line corresponds to @xmath265 . in principle
the density matrix is an arbitrary non - diagonal matrix in all three indices @xmath236 , @xmath72 , and @xmath216 .
but , as has been shown in ref .
, a closed set of equations can be derived , describing the time evolution of the system , which involves only the diagonal elements in @xmath72 .
the same is true for the matrix structure in @xmath216 .
therefore , we need to consider only the following elements of the density matrix @xmath266 . accordingly , of all the transition rates we need to calculate only those between the corresponding elements of the density matrix , i.e. @xmath267 . in the present problem
we further can assume that the tunneling conductance of the set is low compared to the inverse quantum resistance .
in this case , lowest order perturbation theory in the single electron tunneling , describing ` sequential tunneling processes ' , is sufficient .
the diagrams for @xmath251 can be split into two classes , depending on whether they provide expressions for off - diagonal ( @xmath268 ) or diagonal ( @xmath269 ) in @xmath72 elements of @xmath251 . in analogy to the scattering integrals in the boltzmann equation these can be labeled `` in '' and `` out '' terms , in the sense that they describe the increase or decrease of a given element @xmath266 of the density matrix due to transitions from or to other @xmath72-states .
examples for the `` in '' and `` out '' terms are shown in fig .
[ diagram_gain ] and fig .
[ diagram_loose ] , respectively .
we now are ready to evaluate the rates in fig .
[ diagram_gain ] and fig .
[ diagram_loose ] . as an example we consider an `` in '' tunneling process in the left junction : @xmath270 \over \sinh^2\left [ { \pi k_{\rm b}t\over\hbar}(\delta t + i\delta ) \right ] } \right . \\
& & \left .
+ { \exp\left [ -\frac{i}{\hbar } ( \tilde e^{n+1}_{n}+w^{n+1}_{n } ) \delta t \right ] \over \sinh^2\left [ { \pi k_{\rm b}t\over\hbar}(\delta t - i\delta ) \right ] } \right\}^{j'j}_{i'i } \nonumber \ , \end{aligned}\ ] ] where @xmath271 is the usual coulomb energy gain for the tunneling in the left junction in the absence of the qubit , and @xmath272 provide corrections to the energy gain sensitive to the state of the qubit . here
@xmath273 is the @xmath72-th block of the hamiltonian @xmath274 ( note that @xmath275 and @xmath276 are block - diagonal with respect to @xmath72 ) .
the indices @xmath277 and @xmath278 relate to the left and right side of the tensor product in ( [ w_n1_n2 ] ) correspondingly .
the form of the master equation ( [ master_equation ] ) suggests the use of the laplace transform , after which the last term in ( [ master_equation ] ) becomes @xmath279 .
we laplace transform ( [ left_rate_matrix ] ) in the regime @xmath280 , i.e. we assume the density matrix @xmath281 to change slowly on the time scale given by @xmath282 .
this assumption should be verified later for self - consistency .
these inequalities also mean that we choose the operation regime of the set far enough from the coulomb threshold .
therefore , the tunneling is either `` allowed '' for both states of the qubit or it is `` blocked '' for both of them . at low temperatures ( @xmath283 ) and for @xmath284
we obtain : @xmath285 \nonumber \\ & & - \alpha_{\rm l } d(\tilde e^{n}_{n+1 } ) \big [ 2s + \frac{i}{\hbar}(w^{n}_{n+1 } + w^{n+1}_{n } ) \big ] \bigg\}^{j'j}_{i'i } \ , \end{aligned}\ ] ] where @xmath286 and @xmath287 is euler s constant .
the first term of ( [ left_rate_laplace ] ) is the standard golden rule tunneling rate corresponding to the so - called orthodox theory of single - electron tunneling @xcite .
the rate depends strongly on the charging energy difference , @xmath288 , before and after the process , which in the present problem is modified according to the quantum state of the qubit ( the @xmath289 terms ) . at finite temperatures the step - function
is replaced by @xmath290^{-1}$ ] .
we denote the full matrix of such rates @xmath291 and we extensively exploit this matrix below .
if the temperature is low and the applied transport voltage not too high , the leading tunneling process in the set is the sequential tunneling involving only two adjacent charge states , say @xmath218 and @xmath292 .
we concentrate here on this case ( to avoid confusion with the states of the qubit we will keep using the notation @xmath72 and @xmath293 ) . the second , logarithmically diverging part of ( [ left_rate_laplace ] ) produces a `` coherent - like '' term in the rhs of the master equation ( [ master_equation ] ) .
these terms turn out to be unimportant in the first order of the perturbation theory .
indeed , for the left junction we obtain the following contribution to the rhs of ( [ master_equation ] ) : @xmath294 \right ) \ , \ ] ] where @xmath295 and @xmath296 is a matrix in @xmath72 and @xmath216 spaces .
the eigenvalues of the matrix @xmath296 are at most of order @xmath297 .
neglecting terms of order @xmath298 in eq.([left_divergent_term ] ) , we can replace @xmath299 by @xmath255 .
our analysis shows that these neglected `` coherent - like '' terms do not change the results as long as @xmath300 .
similar analysis can be made for the right tunnel junction of the set .
now we can transfer all the `` coherent - like '' terms into the lhs of the master equation , @xmath301 \right\ } = \frac{1}{\hbar}\hat{\gamma } \hat{\sigma}(t),\ ] ] and multiply eq .
( [ master_equation_corrected ] ) from the left by @xmath302 so that the corrections move back to the rhs .
since @xmath291 is itself linear in @xmath303 the corrections belong to the second order in @xmath55 s ( more accurately , they are small if @xmath304 for both junctions ) .
thus , we drop the `` coherent '' corrections and arrive at the final form of the master equation which we use below : @xmath305 = \frac{1}{\hbar}\hat{\gamma } \hat{\sigma}(t ) \ .\ ] ] to rewrite eq .
( [ master_equation_final ] ) in a matrix form we have to choose a particular basis for the qubit .
we choose the charge basis since in this basis the matrix @xmath306 has the simplest form .
the matrix structure in the @xmath216 variable simplifies considerably if we perform a fourier transform with respect to @xmath216 , @xmath307 . to shorten formulas
we introduce @xmath308 , @xmath309 , @xmath310 , and @xmath311 .
then , introducing a vector @xmath312 , we can rewrite ( [ master_equation_final ] ) as @xmath313 where @xmath314 is given by @xmath315 and @xmath316 \ ; , \nonumber \\
\gamma_r & \equiv & 2\pi\alpha_{\rm r } [ -\mu_{\rm r } + e_{\rm set}(1 - 2n_{\rm set } ) ] \ , \end{aligned}\ ] ] @xmath317 the shifts @xmath318 are proportional to the interaction energy between the qubit and set , @xmath319 as we will see , these shifts are responsible for the separation of the peaks of @xmath320 .
finally , the qubit s charging energies are given by @xmath321 we are interested in the limit @xmath322 . in this case the system ( [ me8by8 ] ) may be analyzed perturbatively in @xmath1 . in zeroth order
the system of equations ( [ me8by8 ] ) factorizes into three independent groups . the first one , @xmath323 has exponential solutions @xmath324 and eigenvalues @xmath325^{1\over 2}\right\ } \ .\ ] ] here the rate @xmath326 of single - electron tunneling through the set if the qubit is in the state @xmath32 is given by @xmath327 when @xmath209 is small the first eigenvalue , @xmath328 , has a large imaginary part and the corresponding eigenmode quickly dies out .
the second eigenvalue @xmath329 is small , and the corresponding eigenmode , with @xmath330 , survives .
this solution describes a wave packet propagating with the group velocity @xmath331 , which widens due to shot noise of the single electron tunneling , its width being given by @xmath332 .
the factor @xmath333 ( see e.g. ref . ) varies between @xmath334 in the symmetric situation ( @xmath335 ) and @xmath336 in the extremely asymmetric case ( @xmath337 much larger or much smaller than @xmath338 ) .
analogously the second group of equations for @xmath339 and @xmath340 describe a wave packet which moves in @xmath216-direction with the group velocity @xmath341 and the width growing as @xmath342 .
the two peaks correspond to the qubit in the states @xmath32 and @xmath33 , respectively .
they separate when their distance is larger than their widths , i.e.@xmath343 .
this means that after the time @xmath344 which we denote as the measurement time , the process can constitute a quantum measurement . to learn about the dephasing we analyze the last four equations of ( [ me8by8 ] ) at @xmath345 , which is equivalent to a trace over @xmath216 .
these four equations may be recombined into two complex ones : @xmath346 the analysis of this set shows that if @xmath347 the imaginary parts of the eigenvalues are @xmath348 and @xmath349 . in the opposite limit @xmath350 the imaginary parts are @xmath351 and @xmath352 .
the first limit is physically more relevant ( we assume parameters in this regime ) , although the second one is also possible if the tunneling is very weak or the coupling between the qubit and the set transistor is strong . in both limits the dephasing time , which is defined as the the longer of the two times , @xmath353^{-1 } , [ { \rm im}\,\omega_2]^{-1}\}\ ] ] is parametrically different from the measurement time ( [ measurement_time ] ) . in the first limit , @xmath354
, it is @xmath355 one can check that in the whole range of validity of our approach the measurement time exceeds the dephasing time , @xmath356 .
this is consistent with the fact that a `` good '' quantum measurement should completely dephase a quantum state .
the reason for the difference between the measurement time and the dephasing time is the entanglement of the qubit s state with additional microscopic states of the set , states which can not be characterized by the number of electrons which have tunneled through the transistor , @xmath216 , only .
imagine that we know ( with high probability ) that during some time from the beginning of the measurement process exactly one electron has tunneled through the set whatever the qubit s state was .
does it mean that with that high probability there was no dephasing ?
the answer is no .
the transport of electrons occurs via a real state of the island , @xmath293 .
the system may spend different times in this intermediate state , i.e. different phase shifts are acquired between the two states of the qubit . since the time spent in the state @xmath293 is actually random , some dephasing has occurred . next we can ask where the information about this phase uncertainty is stored , i.e. which states has the qubit become entangled with .
these may not be the states with different @xmath216 or @xmath72 since @xmath216 is equal to one and @xmath72 is again equal to zero irrespective of the state of the qubit .
the only possibility are the microscopic states of the middle island and/or leads , which were subject to different time evolutions during different histories of the tunneling process . to put it in the language of ref .
, the initial state of the system @xmath357 evolves into @xmath358 , where @xmath359 stands for the quantum state of the uncontrolled environment .
one may imagine a situation when @xmath360 , but @xmath361 and @xmath362 are orthogonal .
in this situation the dephasing has occurred but no measurement has been performed .
dephasing was also analyzed in refs . .
in these works a quantum point contact ( qpc ) measuring device was used only as a source of dephasing , i.e. the information on the current flowing in the measuring device was disregarded .
thus a distinction between measurement and dephasing was not made .
however the expressions for the dephasing time were given by expressions similar to ( [ measurement_time ] ) .
as it became clear later @xcite , the qpc does not involve an additional uncontrolled environment during the measurement process and , therefore , @xmath363 .
indeed , the tunneling in the qpc does not occur via a real intermediate state and the discussion above does not apply . in this sense
the qpc may be regarded as a @xmath364 efficient measuring device .
the additional environment in the set , which , as mentioned , reduces the efficiency of the measuring device , plays , actually , a positive role in the quantum measurement , provided it dephases the state of the qubit only when the system is driven out of equilibrium .
this is because the quicker dephasing suppresses the transitions between the states of the qubit ( such a suppression of the transitions due to a continuous observation of the quantum state of the system is called the zeno effect ) , while the initial probabilities are preserved .
it is worth noting that a set might also be used as a @xmath364 efficient device , if it was biased in the co - tunneling regime .
however , this possibility has other drawbacks and hence is not considered here further . finally , we analyze what happens if we take into account the mixing terms proportional to @xmath1 in the system ( [ me8by8 ] ) .
we consider @xmath345 and investigate the eigenvalues of the @xmath365 matrix ( [ matrix_8by8 ] ) .
note that in the discussion above we have calculated all the eight eigenvalues for @xmath163 ( the two eigenvalues of the complex system ( [ cd_subsystem ] ) are doubled when one considers it as a system of four real equations ) . in the diagonal part
there were two zeros , which corresponded to two conserved quantities ( for @xmath345 ) , @xmath366 and @xmath367 .
six other eigenvalues were large compared to the amplitudes of the mixing terms .
it is clear , that including the mixing , @xmath368 , changes only slightly the values of the six large eigenvalues .
a more pronounced effect may be expected in the subspace of the two degenerate eigenvectors with zero eigenvalues .
it turns out that this degeneracy is lifted in second order of perturbation theory .
one of the resulting eigenvalues remains zero .
this corresponds to the conservation of the total trace @xmath369 .
the second eigenvalue acquires now a small imaginary part and this gives the time scale of the mixing between the two states of the qubit . in the limit of ( our ) interest we obtain @xmath370 where @xmath371 , and @xmath372 . comparing eq .
( [ mixing_time ] ) and eq .
( [ measurement_time ] ) we see that both cases @xmath373 and @xmath374 are possible .
let us analyze a concrete physical situation .
we assume @xmath375 .
we choose @xmath376 far enough from the degeneracy point , which is @xmath377 , so that @xmath378 and the related coulomb blockade energy , @xmath379 , is of the order of @xmath380 . to satisfy the conditions for the golden rule ( see ( [ left_rate_laplace ] ) and the discussion thereafter ) we assume @xmath381 to be the largest energy scale of the system , @xmath382 , and the chemical potential of the left lead , @xmath383 , to exceed the coulomb blockade energy by an amount of the order of @xmath381 .
the transport voltage should not , however , exceed the value beyond which further charge states of the set transistor , e.g. @xmath384 and @xmath385 , become involved . thus @xmath386 and @xmath376 should be chosen far enough from zero as well .
thus we obtain @xmath387 } \ .\ ] ] the measurement time in the same regime is given approximately by @xmath388 for comparison we also give a rough value for the dephasing time , which is short when the measurement is performed .
we assume the regime in which eq .
( [ dephasing_time ] ) is valid .
then @xmath389 note that the limit @xmath390 is not allowed in ( [ dephase_time_approx ] ) , since the validity of ( [ dephasing_time ] ) eventually breaks down in this limit .
thus , @xmath391}\right]^2 \ .\ ] ] one recognizes two competing ratios here : @xmath392 , which is small , and @xmath393 $ ] , which is large .
the condition @xmath394 , thus , imposes an additional constraint on the parameters of the system . on the other hand , for low conductance barriers in the set ( @xmath395 but not too small ) ,
the dephasing time is always shorter than the measurement time .
let us summarize what has been done thus far : to measure the quantum state of the qubit we attach a set to the qubit and consider the time evolution of the whole system .
we provide two complimentary descriptions of the measurement process . in the first we trace over the set s degrees of freedom and
obtain a reduced density matrix of the qubit . from the time evolution of the last we deduce the dephasing time and the mixing time , i. e. the relaxation times for the off - diagonal and the diagonal elements of the reduced density matrix respectively .
this description does not , however , contain information about the measurable quantity , the current in the set .
therefore , in a second description , we calculate the probability distribution , @xmath320 , of the number of electrons @xmath216 which have passed through the set during time @xmath242 .
we will evaluate @xmath320 numerically for parameters to be specified in the next section .
the results , shown in fig . [ plot0.1 ] and [ plot0.3 ] , display the time evolution on the time scale of @xmath396 , relevant for the measurement process .
we observe that under appropriate conditions @xmath320 splits into two separate peaks , whose weights are given by the initial probabilities of the eigenstates of the qubit .
the splitting time is the minimum time after which one can distinguish between the states of the qubit and we call it , therefore , the measurement time , @xmath396 . the measurement time turns out to be longer than the dephasing time .
this fact indicates that the state of the qubit gets entangled not only with the transport degree of freedom in the set but also with other , microscopic degrees of freedom which we do not observe @xcite .
note , that the knowledge of @xmath320 does not provide immediately the value of the current flowing in the set at all times .
recently , attempts have been made to describe the measurement process in terms of the current in the measuring device @xcite .
however , further investigation of this problem is definitely needed .
to demonstrate that the constraints on the circuit parameters can be met by available technology , we summarize the constraints and suggest a suitable set .
the necessary conditions are : @xmath397 .
the temperature has to be chosen low to assure the initial thermalization , @xmath398 and @xmath399 , and to reduce the dephasing effects .
a good choice is @xmath400 since further cooling would not decrease the dephasing rate at the degeneracy point much further .
then , the dephasing rates during operations ( at the degeneracy point ) and in the idle state ( off degeneracy ) are close to each other .
depending on the parameters chosen , the number of qubits @xmath72 and the total computation time are restricted .
the dephasing times limits the number of operations to @xmath401 .
this is the only restriction for a single qubit . if several qubits are coupled , the dephasing for the whole system is faster .
since the sources of dissipation for different qubits are independent , the dephasing time gets @xmath72 times shorter . in addition
, the energy splittings of the qubits should fit into the range provided by eq .
( [ high_frequency_condition ] ) , and the levels within this range should be sufficiently different from each other to minimize the errors introduced by non - zero inter - qubit coupling between 2-bit operations ( cf . section [ section_qubits ] ) .
this provides a restriction on the number of qubits and the number of operations which can be performed . with circuit parameters discussed below and for
not too many qubits these restrictions are weaker than those provided by the dephasing .
moreover , we note that these restrictions are removed if the interaction is effectively turned off .
this can be achieved on the `` software '' level , by the use of refocusing voltage pulses . on the `` hardware '' level this
is achieved with the improved design suggested in ref . .
this allows for larger numbers of qubits and longer computations . as an example we suggest a system with the following parameters : + ( i ) we choose junctions with the capacitance @xmath402f , corresponding to the charging energy ( in temperature units ) @xmath403k , and a smaller gate capacitance @xmath404f to reduce the coupling to the environment .
thus at the working temperature of @xmath405mk the initial thermalization is assured .
the superconducting gap has to be slightly larger @xmath406 .
we further choose @xmath407mk , i.e. the time scale of one - qubit operations is @xmath408s .
+ ( ii ) we assume that the resistor in the gate voltages circuit has @xmath409 .
voltage fluctuations limit the dephasing time ( [ gamma_phi_weiss ] ) .
we thus arrive at an estimate of the decoherence rate which allows for @xmath410 coherent operations for a single bit .
+ ( iii ) to assure sufficiently fast 2-bit operations we choose @xmath411h .
then , the decoherence time is @xmath412 times longer than a 2-bit operation . in table
[ tab_alternative_sets ] we present alternative sets of parameters , assuming throughout that the resistance is fixed as @xmath413 .
the quantum measurement process introduces additional constraints on the parameters . in order to demonstrate that the conditions assumed in this paper are realistic we chose the charging energies @xmath198 , @xmath14 and @xmath199 as follows : the capacitance of the josephson junction is @xmath414f , the gate capacitance of the qubit @xmath415f , the capacitances of the normal tunnel junctions of the set @xmath416f , the gate capacitance of the set @xmath417f , and the capacitance between the set and the qubit @xmath418f . we obtain : @xmath419k , @xmath420k , @xmath421k .
taking @xmath422 , @xmath423 and @xmath424k we get @xmath425k , @xmath426k , and @xmath427k ( for definitions see subsection [ subsection_mixing_time ] ) .
we also assume @xmath428 .
the measurement time in this regime is @xmath429s . for this choice of parameters
we calculate @xmath430 using eq .
( [ mixing_time ] ) . assuming first @xmath431k we obtain @xmath432s .
thus @xmath433 and the separation of peaks should occur much earlier than the transitions happen .
indeed , the numerical simulation of the system ( [ me8by8 ] ) for those parameters given above shows almost ideal separation of peaks ( see fig .
[ plot0.1 ] ) . on the other hand , for @xmath434k , and we obtain @xmath435 . this is a marginal situation . the numerical simulation in this case ( see fig .
[ plot0.3 ] ) shows that the peaks , first , start to separate , but , later , the valley between the peaks is filled due to the mixing transitions .
these numbers demonstrate that the quantum manipulations of josephson junction qubits , as discussed in this paper , can be tested experimentally using the currently available lithographic and cryogenic techniques .
we have further demonstrated that the current through a single - electron transistor can serve as a measurement of the quantum state of the qubit , in the sense that in the case of a superposition of two eigenstates it gives one or the other result with the appropriate probabilities .
the demonstration that the qubit is in a superposition of eigenstates should be distinguished from another question , namely whether it is possible to verify that an eigenstate of a qubit is actually a superposition of two different charge states , which depends on the mixing angle @xmath60 as described by eq.([n_operator ] ) .
this question has been addressed in the experiments of ref . .
they used a setup similar to the one shown in fig .
[ circuit ] , a single - cooper - pair box coupled to a single - electron transistor .
they could demonstrate that the expectation value of the charge in the box varies continuously as a function of the applied gate voltage as follows from ( [ n_operator ] ) .
another experiment along the same lines has been performed by nakamura et al .
they demonstrated by spectroscopy that the energy of the ground state and of the first excited state have the expected gate - voltage dependence of superpositions of charge states .
our theory can also describe stationary - state measurements of ref . .
the measurement was performed in the whole range of voltages including the vicinity of the degeneracy point , and we need a different pertrubation expansion which is valid in this region @xcite . for this purpose
we have to analyze the rates in the master equation ( [ master_equation_final ] ) for general values of the mixing angle @xmath60 , relaxing the requirement @xmath436 . to this end we rewrite the master equation ( [ matrix_8by8 ] ) in the qubit s eigenbasis ( [ eigen_basis ] ) and consider the mixing terms as a perturbation .
then , the expansion parameter is @xmath437 which extends the region of the validity of the perturbative treatment .
it turns out that each eigenstate of the qubit , @xmath32 or @xmath33 , corresponds to a single , though @xmath60-dependent tunneling rate @xmath438 .
thus , if the qubit is prepared in one of its eigenstates , only one peak is observed .
this is the case even close to the degeneracy point @xmath439 where the eigenstates are superpositions of two charge states with substantial weight of both components . on the other hand a charge state ,
being a superposition of two eigenstates , would produce two peaks in the current distribution .
in this sense the measurement provides information about occupation probabilities of the eigenstates rather than charge states of the qubit .
the correction to the tunneling rate @xmath440 is proportional to the average charge in the corresponding eigenstate of the qubit , in accordance with the experiments of bouchiat et al .
close to the degeneracy it is more difficult ( takes longer ) to distinguish the eigenstates since the peaks get closer . in our solution of the master equation we neglected dissipative effects due to the environment discussed in section [ section_dephasing ] .
it is justified on time scales shorter than the environment - induced relaxation and dephasing times given by ( [ gamma_r_weiss ] ) , ( [ gamma_phi_weiss ] ) .
it is also justified at longer times as long as the environment - induced effects are weaker than the measurement - induced mixing , i.e. the rates ( [ gamma_r_weiss ] ) , ( [ gamma_phi_weiss ] ) are smaller than the mixing rate ( [ mixing_time ] ) . in this limit thermal relaxation is ineffective , and in the stationary regime ( @xmath441 ) the qubit is in the equally - weighted mixture of two states , corresponding to the infinite effective temperature .
the current in the set , @xmath442 , is insensitive to the gate voltage in the qubit s circuit . in the opposite limit of weaker mixing , which is relevant to the experiments of bouchiat et al .
@xcite , mixing is ineffective and thermal relaxation takes over . at low temperature @xmath443 dissipation
keeps the system in the ground state , and the stationary current value is @xmath444 .
it is also interesting to compare our proposal with the `` quantum jumps '' technique employed in quantum optics in general , and with the realizations of the qubits by trapped ions in particular ( for a review see ref . )
indeed , the concepts are very close in spirit : the state of the system is examined by an external nonequilibrium current ( electrons in our case and photons in the quantum jumps technique ) .
there is , however , an important difference . in the quantum jumps measurements only one of the logical states scatters photons
therefore , the efficiency of the measurement is limited by the ability to detect photons . in principle
we could realize this situation also in our system , if we bias the set transistor such that different states of the qubit switch the transistor between the off and on regimes .
then the efficiency of the measurement is determined by the ability to detect individual electron which is possible in single - electron devices , for instance by charging a single - electron box and the measurement time would be given by the time it takes the first electron to tunnel . however
, this mode of operation would require that the set transistor is kept near the switching point , where thermal fluctuations and higher order processes could modify the picture substantially .
therefore , we have concentrated here on a situation in which the set transistor conducts for both states of the qubit , and the measurement requires distinguishing large numbers of charges or macroscopic currents .
accordingly , the measurement time is limited by the shot noise .
to conclude , the fabrication of josephson junction qubits is possible with current technology . in these systems
fundamental features of macroscopic quantum - mechanical systems can be explored .
more elaborate designs as well as further progress of nano - technology , will provide longer coherence times and allow scaling to larger numbers of qubits .
the application of josephson junction systems as elements of a quantum computer , i.e. with a very large number of manipulations and large number of qubits , will remain a challenging issue , demanding in addition to the perfect control of time - dependent gate voltages a still longer phase coherence time .
we stress , however , that many aspects of quantum information processing can initially be tested on simple circuits as proposed here .
we have further shown that a single - electron transistor capacitively coupled to a qubit may serve as a quantum measuring device in an accessible range of parameters .
we have described the process of measurement by deriving the time evolution of the reduced density matrix .
we found that the dephasing time is shorter than the measurement time , and we have estimated the mixing time , i.e. the time scale on which the transitions induced by the measurement occur .
we thank e. ben - jacob , t. beth , c. bruder , l. dreher , y. gefen , s. gurvitz , z. hermon , j. knig , a. korotkov , y. levinson , j. e. mooij , t. pohjola , and h. schoeller for stimulating discussions .
this work has been supported by a graduiertenkolleg and the sfb 195 of the dfg , the a. v. humboldt foundation ( y.m . ) and the german israeli foundation ( contract g-464 - 247.07/95 ) ( a.s . ) .
s. lloyd , science * 261 * , 1589 ( 1993 ) ; c.h .
bennett , physics today * 48 * ( 10 ) , 24 ( 1995 ) ; a. barenco , contemp . phys . *
37 * , 375 ( 1996 ) ; d.p .
divincenzo , science * 269 * , 255 ( 1995 ) , and in _ mesoscopic electron transport _ , ed . l.l .
sohn et al . ,
kluwer , 1997 .
shor , phys .
rev . a , * 52 * , 2493 ( 1995 ) ; a. steane , proc .
soc . of london
a , * 452 * , 2551 ( 1996 ) ; e.h .
knill and r. laflamme phys .
a * 55 * , 900 , ( 1997 ) ; e. knill , r. laflamme and w. zurek , science , * 279 * , 342 ( 1998 ) . | several physical realizations of quantum bits have been proposed , including trapped ions , nmr systems and spins in nano - structures , quantum optical systems , and nano - electronic devices .
the latter appear most suitable for large - scale integration and potential applications .
we suggest to use low - capacitance josephson junctions , exploiting the coherence of tunneling in the superconducting state combined with the possibility to control individual charges by coulomb blockade effects .
these systems constitute quantum bits ( qubits ) , with logical states differing by one cooper - pair charge . single- and
two - bit operations can be performed by applying a sequence of gate voltages .
the phase coherence time is sufficiently long to allow a series of these steps .
the ease and precision of these manipulations depends on the specific design .
here we concentrate on a circuit which is most easily fabricated in an experiment .
in addition to the manipulation of qubits the resulting quantum state has to be read out . this can be accomplished by coupling a single - electron transistor capacitively to the qubit . to describe this quantum measurement process
we study the time evolution of the density matrix of the coupled system .
only when a transport voltage is turned on , the transistor destroys the phase coherence of the qubit ; in this case within a short time .
the measurement is accomplished after a longer time , when the signal resolves the different quantum states . at still longer times
the measurement process itself destroys the information about the initial state .
we present a suitable set of system parameters , which can be realized by present - day technology . 2 |
consider a random vector ( rv ) @xmath3 whose distribution function ( df ) is a copula @xmath4 , i.e. , each @xmath5 follows the uniform distribution on @xmath6 . the copula @xmath4 is said to be in the max - domain of attraction of an extreme value df ( evd ) @xmath7 on @xmath8 , denoted by @xmath9 , if @xmath10 the characteristic property of the df @xmath7 is its _ max - stability _ , precisely , @xmath11 see , e.g. , ( * ? ? ?
* section 4 ) .
let @xmath12 be independent copies of @xmath13 .
equation is equivalent with @xmath14 where @xmath15 and @xmath16 .
all operations on vectors such as @xmath17 are meant componentwise . from ( * ? ? ?
* corollary 2.2 ) we know that @xmath9 if and only if ( iff ) there exists a norm @xmath18 on @xmath8 such that the copula @xmath4 satisfies the expansion @xmath19 uniformly for @xmath20^d$ ] as @xmath21 , i.e. , @xmath22^d\setminus{{\mathopen{}\mathclose\bgroup\originalleft}\{{\bm{1}}{\aftergroup\egroup\originalright}\}}\\ { { \mathopen{}\mathclose\bgroup\originalleft}\vert{\bm{u}}-{\bm{1}}{\aftergroup\egroup\originalright}\vert}<t } } \frac{{{\mathopen{}\mathclose\bgroup\originalleft}\vertc({\bm{u}})-\bigl(1-{{\mathopen{}\mathclose\bgroup\originalleft}\vert{\bm{u}}-{\bm{1}}{\aftergroup\egroup\originalright}\vert}\bigr){\aftergroup\egroup\originalright}\vert}}{{{\mathopen{}\mathclose\bgroup\originalleft}\vert{\bm{u}}-{\bm{1}}{\aftergroup\egroup\originalright}\vert } } = 0.\ ] ] in this case the norm @xmath18 is called a _ @xmath23-norm _ and is commonly denoted by @xmath24 , where the additional character @xmath23 means dependence .
the corresponding evd @xmath7 is then given by @xmath25 we refer to ( * ? ? ?
* section 5.2 ) for further details .
we have in particular independence of the margins of @xmath7 iff the @xmath23-norm @xmath24 is the usual @xmath26-norm @xmath27 , and we have complete dependence iff @xmath24 is the maximum - norm @xmath28 .
if the copula @xmath4 satisfies @xmath29 for some @xmath30 , then we refer to it as a _ generalized pareto copula _ ( gpc ) .
the characteristic property of a gpc is its _ excursion stability _ : the rv @xmath3 follows a gpc iff there exists @xmath30 such that @xmath31,\ ] ] for all @xmath32 and each nonempty subset @xmath33 of @xmath34 , see ( * ? ? ?
* proposition 5.3.4 ) . based on this characterization
, @xcite investigated a test whether @xmath13 follows a gpc ; see also ( * ? ? ?
* section 5.8 ) .
it is by no means obvious to find a copula @xmath4 , which does _ not _ satisfy @xmath9 .
an example is given in @xcite .
a family of copulas , which are not in the max - domain of attraction of an evd but come arbitrary close to a gpc , is given in section [ sec : copula_not_in_domain_of_attraction ]
. parts of the simulations in section [ sec : simulations ] are based on this family .
if the remainder term in equation satisfies @xmath35 as @xmath36 for some @xmath37 , then the copula @xmath4 is said to be in the _ @xmath38-neighborhood _ of a gpc .
note that is already implied by and that @xmath39 for an arbitrary norm @xmath18 on @xmath8 .
the significance of such @xmath38-neighborhoods is outlined in section [ sec : delta - neighborhoods ] where we also give some prominent examples . in section [ sec : test_in_case_of_copula ] we will derive a @xmath1-goodness - of - fit test based on @xmath40 which checks whether the pertaining copula @xmath4 satisfies condition .
let @xmath41 be a rv with arbitrary df @xmath0 .
it is well - known @xcite that @xmath0 is in the max - domain of attraction of an evd iff this is true for the univariate margins of @xmath0 together with the condition that the copula @xmath42 corresponding to @xmath0 satisfies . while there are various tests which check for the univariate extreme value condition see , e.g. @xcite , @xcite as well as ( * ? ? ?
* section 5.3 ) much less has been done for the multivariate case . utilizing the _ empirical copula _ , we can modify the test statistic from section [ sec : test_in_case_of_copula ] and check , whether @xmath42 satisfies condition , based on independent copies @xmath43 .
this is the content of section [ sec : arbitrary_random_vector ] .
sections [ sec : test_gpcp ] and [ sec : test_processes_grid ] carry the results of section [ sec : test_multivariate ] over to function space .
the aim is to test whether the copula process of a given stochastic process in @xmath44 $ ] is in a @xmath38-neighborhood of a generalized pareto copula process ( gpcp ) . while section [ sec : test_gpcp ] deals with copula processes or general processes as a whole , respectively , section [ sec : test_processes_grid ] considers the case that the underlying processes are observed at a finite grid of points only . in order to demonstrate the performance of our test , section [ sec : simulations ] states the results of a simulation study .
since the results from the previous sections highly depend on a proper choice of some threshold , we also present a graphical tool that makes the decision , whether or not to reject the hypothesis , more comfortable .
the following result provides a one parametric family of bivariate rv , which are easy to simulate .
each member of this family has the property that its corresponding copula does not satisfy the extreme value condition .
however , as the parameter tends to zero , the copulas of interest come arbitrarily close to a gpc , which , in general , is in the domain of attraction of an evd .
[ lem : copula_not_in_domain_of_attraction ] let the rv @xmath45 have df @xmath46 where @xmath47 $ ] .
note that @xmath48 , @xmath49 and @xmath50 for @xmath51 .
furthermore let the rv @xmath52 be independent of @xmath45 and uniformly distributed on @xmath6 .
put @xmath53 .
then the copula @xmath54 corresponding to the bivariate rv @xmath55 ^ 2\ ] ] is not in the domain of attraction of a multivariate evd if @xmath56 , whereas @xmath57 is a gpc with corresponding @xmath23-norm @xmath58 for @xmath59 .
denote by @xmath60 the df of @xmath61 .
elementary computations yield that it is given by @xmath62 and , thus , @xmath60 is continuous and strictly increasing on @xmath63 $ ] .
we show that @xmath64 does not exist for @xmath47\setminus{{\mathopen{}\mathclose\bgroup\originalleft}\{0{\aftergroup\egroup\originalright}\}}$ ] . since @xmath54 coincides with the copula of @xmath65 we obtain setting @xmath66 , @xmath67 , @xmath68 the substitution @xmath69 yields @xmath70 where we have for each @xmath71 @xmath72 and @xmath73 which can be seen by applying integration by parts twice .
hence we obtain @xmath74 whose limit does not exist for @xmath75 if @xmath47\setminus{{\mathopen{}\mathclose\bgroup\originalleft}\{0{\aftergroup\egroup\originalright}\}}$ ] ; consider , e.g. , the sequences @xmath76 and @xmath77 as @xmath78 . on the other hand , elementary computations show for @xmath79 ^ 2\setminus{{\mathopen{}\mathclose\bgroup\originalleft}\{{\bm{0}}{\aftergroup\egroup\originalright}\}}$ ] @xmath80 the remaining assertion is thus implied by section 2 of @xcite .
similar results as in lemma [ lem : copula_not_in_domain_of_attraction ] can be obtained for different distributions of the rv @xmath81 in , which is still assumed to be independent of @xmath45 .
if @xmath82 , then @xmath83 implies @xmath84 , whereas @xmath85 , @xmath86 gives @xmath87 where @xmath88 are independent and uniformly distributed on @xmath89 $ ] .
however , if @xmath56 , then we obtain @xmath90 and @xmath91 respectively . note that both terms have no limit for @xmath75 ; consider , e.g. , the sequences @xmath92 and @xmath93 as in the proof of lemma [ lem : copula_not_in_domain_of_attraction ] .
the significance of the @xmath38-neighborhood of a gpc can be seen as follows .
denote by @xmath94^d:\,{{\mathopen{}\mathclose\bgroup\originalleft}\vert{\bm{t}}{\aftergroup\egroup\originalright}\vert}_1=\sum_{i=1}^d t_i=1{\aftergroup\egroup\originalright}\}}$ ] the unit sphere in @xmath95 with respect to the @xmath26-norm @xmath27 .
take an arbitrary copula @xmath4 on @xmath8 and put for @xmath96 @xmath97 then @xmath98 is a univariate df on @xmath63 $ ] and the copula @xmath4 is obviously determined by the family @xmath99 of univariate _
spectral df _ @xmath98 .
this family @xmath100 is the _ spectral decomposition _ of @xmath4
* section 5.4 ) .
a copula @xmath4 is , consequently , in @xmath101 with corresponding @xmath23-norm @xmath24 iff its spectral decomposition satisfies @xmath102 as @xmath103 .
the copula @xmath4 is in the @xmath38-neighborhood of the gpc @xmath104 with @xmath23-norm @xmath24 iff @xmath105 uniformly for @xmath106 as @xmath103 . in this case
we know from ( * ? ? ?
* theorem 5.5.5 ) that @xmath107^d}{{\mathopen{}\mathclose\bgroup\originalleft}\vertc^n{\mathopen{}\mathclose\bgroup\originalleft}({\bm{1}}+ \frac 1n{\bm{x}}{\aftergroup\egroup\originalright } ) - \exp{\mathopen{}\mathclose\bgroup\originalleft}(-{{\mathopen{}\mathclose\bgroup\originalleft}\vert{\bm{x}}{\aftergroup\egroup\originalright}\vert}_d{\aftergroup\egroup\originalright}){\aftergroup\egroup\originalright}\vert } = o{\mathopen{}\mathclose\bgroup\originalleft}(n^{-\delta}{\aftergroup\egroup\originalright}).\ ] ] under additional differentiability conditions on @xmath108 with respect to @xmath109 , also the reverse implication @xmath110 holds ; cf .
* theorem 5.5.5 ) .
thus the @xmath38-neighborhood of a gpc , roughly , collects those copula with a polynomial rate of convergence of maxima .
let @xmath111}{\mathopen{}\mathclose\bgroup\originalleft}(\sum_{k=1}^d\varphi(u_i){\aftergroup\egroup\originalright}),\qquad { \bm{u}}=(u_1,\dots , u_d)^{{\mathpalette{\mspace{-1mu}\raisebox{0.25ex}{$\intercal}$}}}\in[0,1]^d,\ ] ] be an archimedean copula with generator function @xmath112\to[0,\infty]$ ] and @xmath113}(t ) : = \inf\{u\in[0,1 ] : \varphi(u)\le t\}$ ] , @xmath114 .
the function @xmath115 is in particular strictly decreasing , continuous and satisfies @xmath116 ; for a complete characterization of the function @xmath115 we refer to @xcite .
suppose that @xmath115 is differentiable on @xmath117 $ ] for some @xmath118 with derivative satisfying @xmath119 for some @xmath37 as @xmath120 .
then @xmath4 is in the @xmath38-neighborhood of a gpc with @xmath23-norm given by @xmath121 , @xmath122 . the clayton family with @xmath123
, @xmath124 satisfies condition with @xmath125 if @xmath126 .
the parameter @xmath127 yields a gpc with corresponding @xmath23-norm @xmath27 . the gumbel - hougard family with @xmath128
, @xmath129 does not satisfy condition .
but for @xmath130 it is in the @xmath38-neighborhood with @xmath131 of a gpc having @xmath23-norm @xmath132 . for general results on the limiting distributions of archimedean copulas we refer to @xcite and @xcite .
let @xmath4 be a normal copula , i.e. , @xmath4 is the df of @xmath133 , where @xmath134 denotes the standard normal df and @xmath135 follows a multivariate normal distribution @xmath136 with mean vector @xmath137 and covariance matrix @xmath138 , where @xmath139 , @xmath140 .
if @xmath141 for @xmath142 , then @xmath4 is in the @xmath38-neighborhood of a gpc @xmath104 with @xmath143 and @xmath144 put @xmath145 .
then we have @xmath146 by the inclusion - exclusion theorem . by @xmath147
we denote in what follows a positive generic constant .
we have @xmath148 we will show that for all @xmath149 @xmath150 for @xmath151 , where @xmath152 is specified later .
this , obviously , implies the assertion .
equation is implied by the inequality @xmath153 for @xmath151 , which we will establish in the sequel .
fix @xmath149 . to ease the notation we put @xmath154 .
the covariance matrix of @xmath155 is @xmath156 , its inverse is @xmath157 and , hence , @xmath158 if @xmath159 ; recall that @xmath160 . from @xcite
( see also @xcite and @xcite ) we obtain the bound @xmath161 by the obvious inequality @xmath162 we obtain @xmath163 and , thus , equation implies @xmath164 from the fact that @xmath165 as @xmath166 , where @xmath167 denotes the standard normal density , we obtain for @xmath168 @xmath169 recall that @xmath160 .
this implies equation and , thus , the assertion .
this section deals with deriving a test for condition based on independent copies @xmath12 of the rv @xmath170 having df @xmath4 . put for @xmath171 @xmath172 which is a discrete version of the sojourn time that @xmath13 spends above the threshold @xmath173 ; see @xcite for details as well as section [ sec : test_gpcp ] . if @xmath4 satisfies condition , then we obtain with @xmath174 @xmath175 the constant @xmath176 , which is always between @xmath177 and @xmath178 , measures the tail dependence of the margins of @xmath4 .
it is the extremal coefficient @xcite and equal to one in case of complete dependence of the margins and equal to @xmath178 in case of independence @xcite ; we refer to ( * ? ? ?
* section 4.4 ) for further details . in order to test for condition
, we fit a grid in the upper tail of the copula @xmath4 and observe the exceedances with respect to this grid : choose @xmath179 and put for @xmath180 @xmath181 @xmath182 is the number of those rv @xmath183 among the independent copies @xmath40 of @xmath13 , whose sojourn times above @xmath184 are positive , i.e. , at least one component of @xmath183 exceeds the threshold @xmath185 . on the other hand , @xmath186}\bigl({\bm{u}}^{(i)}\bigr)\ ] ] is the number of those rv @xmath183 whose realizations are below the vector with constant entry @xmath185 . if @xmath4 satisfies condition , then each @xmath182 is binomial @xmath187-distributed with @xmath188 motivated by the usual @xmath1-goodness - of - fit test , we consider in what follows the test statistic @xmath189 which does not require the constant @xmath190 to be known . by @xmath191
we denote ordinary convergence in distribution as the sample size @xmath192 tends to infinity .
[ thm : limit_distribution_of_t_n ] suppose that @xmath4 satisfies condition with @xmath37 .
let @xmath193 satisfy @xmath194 , @xmath195 and @xmath196 as @xmath78 .
then we obtain @xmath197 where @xmath198 are independent and standard normal distributed rv and @xmath199 we have @xmath200 in case @xmath201 , and @xmath202 , @xmath203 in case @xmath204 . if the copula @xmath4 is a gpc , then the condition @xmath205 in the preceding result can be dropped .
the proof of theorem [ thm : limit_distribution_of_t_n ] shows that @xmath206 equals the distribution of @xmath207 , where @xmath208 is a standard brownian motion on @xmath209 .
computing the expected values , we obtain as a nice by - product the equation @xmath210 using characteristic functions it is straightforward to prove that @xmath211 as @xmath212 .
taking expectations on both sides motivates the well - known equality @xmath213 lindeberg s central limit theorem implies @xmath214 and , thus , @xmath215 yielding @xmath216 we , therefore , can substitute the denominator in the test statistic @xmath217 by @xmath218 , i.e. , @xmath217 is asymptotically equivalent with @xmath219 where @xmath220 with @xmath221 @xmath222 is the @xmath223 unit matrix and @xmath224 that @xmath223-matrix with constant entry 1 .
note that the matrix @xmath225 is a _ projection matrix _
, i.e. , @xmath226 , and that @xmath227 for every vector @xmath228 with constant entries .
the cramr - wold theorem and lindeberg s central limit theorem imply @xmath229 , where the @xmath223-covariance matrix @xmath230 is given by @xmath231 \\ & = \lim_{n\to\infty } \frac{ij}{c_nm_d } p{\mathopen{}\mathclose\bgroup\originalleft}(s_{{\bm{u}}}{\mathopen{}\mathclose\bgroup\originalleft}(-\frac{c_n}{\max(i , j)}{\aftergroup\egroup\originalright})>0{\aftergroup\egroup\originalright } ) \\ & = \frac{ij}{\max(i , j ) } = \min{\mathopen{}\mathclose\bgroup\originalleft}(i , j{\aftergroup\egroup\originalright}).\end{aligned}\ ] ] note that @xmath232 where the @xmath223-matrix @xmath233 is defined by @xmath234 altogether we obtain @xmath235 with a @xmath236-dimensional standard normal rv @xmath237 .
it is well - known that the eigenvalues of @xmath238 are @xmath239 see , for example , @xcite or @xcite , with corresponding orthonormal eigenvectors @xmath240 this implies @xmath241 as asserted . to evaluate the performance of the above test we consider in what follows @xmath192 independent copies @xmath242 of the rv @xmath13 , whose df @xmath4 satisfies for some @xmath37 the expansion @xmath243 as @xmath244 , uniformly for @xmath20^d$ ] , where @xmath245 is an arbitrary function on the set @xmath246 of directions in @xmath63^d$ ] .
the above condition specifies the remainder term in the @xmath38-neighborhood condition .
we obtain for @xmath247 and @xmath248 @xmath249 where @xmath250 . for @xmath182
as defined in we obtain by elementary arguments @xmath251 if @xmath194 , @xmath195 and @xmath252 as @xmath78 . repeating the arguments in the proof of theorem [ thm : limit_distribution_of_t_n ]
then yields for the test statistic @xmath253 defined in @xmath254 where @xmath255 note that each @xmath256 if @xmath257 . with @xmath258 , the fact that @xmath4 is not
a gpc is , therefore , detected at an arbitrary level - one error iff @xmath259 .
testing for a @xmath38-neighborhood requires however the rate @xmath260 . consider a rv @xmath41 whose df @xmath0 is continuous and the copula @xmath261 , @xmath262 , corresponding to @xmath0 satisfies condition .
now we will modify the test statistic @xmath217 from above to obtain a test which checks whether @xmath42 satisfies condition with @xmath263 unknown , @xmath264 .
we denote in what follows by @xmath265 , @xmath266 , the generalized inverse of an arbitrary univariate df @xmath267 .
let @xmath268 , @xmath269 , be independent copies of @xmath270 and fix @xmath271 .
it turns out that , contrary to , it is too ambitious to take all @xmath192 observations into account
. so choose an arbitrary subset @xmath272 of @xmath273 of size @xmath274 , and put for @xmath180 @xmath275}\bigl({\bm{x}}^{(i)}\bigr ) , \qquad 1\le j\le k,\end{aligned}\ ] ] which is the number of all rv @xmath276 exceeding the vector @xmath277 in at least one component .
in section [ sec : test_in_case_of_copula ] we have seen that one may choose @xmath278 if @xmath0 is a copula itself .
since @xmath0 is continuous , transforming each @xmath279 by its df @xmath280 does not alter the value of @xmath281 with probability one : @xmath282}\bigl({\bm{u}}^{(i)}\bigr)\ ] ] where @xmath283 , @xmath284 , and @xmath40 are iid with df @xmath42 .
as the margins of @xmath0 are typically unknown in applications , we now replace @xmath285 with their empirical counterparts @xmath286}(x_r)$ ] , @xmath287 , @xmath288 , and obtain analogously @xmath289 } \bigl({\bm{u}}^{(i)}\bigr)\end{aligned}\ ] ] with probability one .
note that @xmath290 where @xmath291 and @xmath292 denote the ordered values of @xmath293 for each @xmath294 .
thus @xmath295 and @xmath296 , @xmath297 , almost surely .
since @xmath40 are iid with df @xmath42 , the distribution of @xmath298 does not depend on the marginal df @xmath280 but only on the copula @xmath42 of the continuous df @xmath0 .
the following auxiliary result assures that we may actually consider @xmath299 instead of @xmath281 .
[ lem : crucial_approximation_of_n_j ] suppose that @xmath300 , @xmath301 as @xmath78 .
let @xmath302 satisfy @xmath194 , @xmath303 as @xmath304 .
then we obtain for @xmath305 @xmath306 we have almost surely @xmath307 } \bigl({\bm{u}}^{(i)}\bigr ) - 1_{{\mathopen{}\mathclose\bgroup\originalleft}[{\bm{0 } } , { \mathopen{}\mathclose\bgroup\originalleft}(1-\frac{c_n}{j}{\aftergroup\egroup\originalright}){\bm{1}}{\aftergroup\egroup\originalright}]}\bigl({\bm{u}}^{(i)}\bigr ) { \aftergroup\egroup\originalright } ) \\ & = \sum_{i\in m(n ) } 1_{\times_{r=1}^d{\mathopen{}\mathclose\bgroup\originalleft}[\vphantom{\frac{c_n}{j}}\smash{0,u_{\langle n(1-\frac{c_n}{j})\rangle : n , r}}{\aftergroup\egroup\originalright } ] } \bigl({\bm{u}}^{(i)}\bigr ) { \mathopen{}\mathclose\bgroup\originalleft}(1 - 1_{{\mathopen{}\mathclose\bgroup\originalleft}[{\bm{0 } } , { \mathopen{}\mathclose\bgroup\originalleft}(1-\frac{c_n}{j}{\aftergroup\egroup\originalright}){\bm{1}}{\aftergroup\egroup\originalright}]}\bigl({\bm{u}}^{(i)}\bigr ) { \aftergroup\egroup\originalright } ) \\ & \mathrel{\hphantom{= } } { } - \sum_{i\in m(n ) } 1_{{\mathopen{}\mathclose\bgroup\originalleft}[{\bm{0 } } , { \mathopen{}\mathclose\bgroup\originalleft}(1-\frac{c_n}{j}{\aftergroup\egroup\originalright}){\bm{1}}{\aftergroup\egroup\originalright}]}\bigl({\bm{u}}^{(i)}\bigr ) { \mathopen{}\mathclose\bgroup\originalleft}(1 - 1_{\times_{r=1}^d[0,u_{\langle n(1-\frac{c_n}{j})\rangle : n , r } ] } \bigl({\bm{u}}^{(i)}\bigr){\aftergroup\egroup\originalright } ) \\ & = : r_n - t_n.\end{aligned}\ ] ] in what follows we show @xmath308}\bigl(u_r^{(i)}\bigr){\aftergroup\egroup\originalright } ) = o(1)\ ] ] and thus @xmath309 ; note that @xmath310}\bigl(u_r^{(i)}\bigr).\end{aligned}\ ] ] put @xmath311 with @xmath312 .
then we have with @xmath313 @xmath314 where the first term is of order @xmath315 ; recall that @xmath316 is uniformly distributed on @xmath6 and @xmath317 as @xmath78 .
furthermore we deduce from ( * ? ? ?
* lemma 3.1.1 ) the exponential bound @xmath318 where @xmath319 and @xmath320 as well as @xmath321 as @xmath78 .
this implies @xmath322 repeating the above arguments shows that @xmath323 as well , which completes the proof of lemma [ lem : crucial_approximation_of_n_j ] .
the previous result suggests a modification of our test statistic in @xmath324 which does not depend on the margins but only on the copula of the underlying df @xmath0 . the following result is a consequence of theorem [ thm : limit_distribution_of_t_n ] and lemma [ lem : crucial_approximation_of_n_j ] .
[ thm : limit_distribution_of_test_statistic_general ] suppose that the df @xmath0 is continuous and that its copula @xmath42 satisfies expansion for some @xmath37 .
let @xmath325 satisfy @xmath300 , @xmath326 as @xmath78 , and let @xmath193 satisfy @xmath194 , @xmath303 , @xmath327 as @xmath78 .
then we obtain @xmath328 with @xmath329 and @xmath330 as in theorem [ thm : limit_distribution_of_t_n ] .
the condition @xmath327 can again be dropped if the copula @xmath42 is a gpc .
in this section we carry the results of section [ sec : test_multivariate ] over to function space , namely the space @xmath44 $ ] of continuous functions on @xmath89 $ ] . a stochastic process @xmath331}$ ] with sample paths in @xmath44 $ ]
is called a _ standard max - stable process _ ( smsp ) , if @xmath332 , @xmath333 , for each @xmath334 $ ] , and if the distribution of @xmath335 equals that of @xmath336 for each @xmath337 , where @xmath338 are independent copies of @xmath336 .
all operations on functions such as @xmath339 , @xmath340 , @xmath341 etc . are meant pointwise . to improve the readability we set stochastic processes such as @xmath336 or @xmath342 in bold font and deterministic functions like @xmath343 in default font . from @xcite
see also @xcite as well as @xcite we know that a stochastic process @xmath344 $ ] is an smsp iff there exists a _ generator process _
@xmath345}\in c[0,1]$ ] with @xmath346 , @xmath334 $ ] , for some @xmath347 , and @xmath348 , such that @xmath349}{\mathopen{}\mathclose\bgroup\originalleft}({{\mathopen{}\mathclose\bgroup\originalleft}\vertf(t){\aftergroup\egroup\originalright}\vert}z_t{\aftergroup\egroup\originalright}){\aftergroup\egroup\originalright}){\aftergroup\egroup\originalright}),\qquad f\in e^-[0,1].\ ] ] by @xmath350 $ ] we denote the set of those functions @xmath351\to{\mathbb{r}}$ ] , which are bounded and have only a finite number of discontinuities ; @xmath352 $ ] is the subset of those functions in @xmath350 $ ] that attain only non positive values .
note that @xmath353}{\mathopen{}\mathclose\bgroup\originalleft}({{\mathopen{}\mathclose\bgroup\originalleft}\vertf(t){\aftergroup\egroup\originalright}\vert}z_t{\aftergroup\egroup\originalright}){\aftergroup\egroup\originalright}),\qquad f\in e[0,1],\ ] ] defines a norm on @xmath350 $ ] .
let @xmath354}$ ] be a _ copula process _
, i.e. , each component @xmath355 is uniformly distributed on @xmath6 .
a copula process @xmath356 $ ] is said to be in the _ functional max - domain attraction _ of an smsp @xmath336 , denoted by @xmath357 , if @xmath358})\le f{\aftergroup\egroup\originalright})^n\to_{n\to\infty } p({\bm{\eta}}\le f)=\exp{\mathopen{}\mathclose\bgroup\originalleft}(-{{\mathopen{}\mathclose\bgroup\originalleft}\vertf{\aftergroup\egroup\originalright}\vert}_d{\aftergroup\egroup\originalright}),\qquad f\in e[0,1],\ ] ] where @xmath359}$ ] denotes the indicator function of the interval @xmath89 $ ] .
this is the functional version of .
we refer to @xcite for details .
a more restrictive definition of functional max - domain of attraction of stochastic processes in terms of usual weak convergence was introduced by @xcite . from ( * ? ? ?
* proposition 8) we know that condition is equivalent with the expansion @xmath360}+cf{\aftergroup\egroup\originalright})=1- c{{\mathopen{}\mathclose\bgroup\originalleft}\vertf{\aftergroup\egroup\originalright}\vert}_d+o(c ) , \qquad f\in e^-[0,1],\ ] ] as @xmath361 .
in particular we obtain in this case @xmath362}{\aftergroup\egroup\originalright})=1-c{\mathopen{}\mathclose\bgroup\originalleft}(m_d+r(-c){\aftergroup\egroup\originalright}),\ ] ] where @xmath363}z_t{\aftergroup\egroup\originalright } ) = { { \mathopen{}\mathclose\bgroup\originalleft}\vert1_{[0,1]}{\aftergroup\egroup\originalright}\vert}_d$ ] is the uniquely determined _ generator constant _ pertaining to @xmath336 , and the remainder term satisfies @xmath364 as @xmath361 .
a copula process @xmath365 $ ] is a _ generalized pareto copula process _ ( gpcp ) , if there exists @xmath366 such that @xmath367}+f{\aftergroup\egroup\originalright})=1- { { \mathopen{}\mathclose\bgroup\originalleft}\vertf{\aftergroup\egroup\originalright}\vert}_d , \qquad f\in e^-[0,1],\,{{\mathopen{}\mathclose\bgroup\originalleft}\vertf{\aftergroup\egroup\originalright}\vert}_\infty\le { \varepsilon}_0.\ ] ] a gpcp with a prescribed @xmath23-norm @xmath24 can easily be constructed ; cf .
* example 5 ) .
its characteristic property is its excursion stability ; cf .
@xcite . a copula process @xmath356 $ ] , consequently , satisfies @xmath357 iff there exists a gpcp @xmath368 such that @xmath369}+cf{\aftergroup\egroup\originalright})=p{\mathopen{}\mathclose\bgroup\originalleft}({\bm{v}}\le 1_{[0,1]}+cf{\aftergroup\egroup\originalright})+o(c ) , \qquad f\in e^-[0,1],\ ] ] as @xmath361 .
if the remainder term @xmath370 in expansion is in fact of order @xmath371 for some @xmath37 , then the copula process @xmath356 $ ] is said to be in the _ @xmath38-neighborhood _ of a gpcp ; cf . .
the test statistic @xmath217 investigated in section [ sec : test_in_case_of_copula ] carries over to function space @xmath44 $ ] , which enables us to check whether a given copula process @xmath354}$ ] is in a @xmath38-neighborhood of an smsp @xmath368 .
put for @xmath171 @xmath372,\ ] ] which is the sojourn time that the process @xmath13 spends above the threshold @xmath173 .
if @xmath357 , then we obtain from equation @xmath373}{\aftergroup\egroup\originalright})\nonumber\\ & = { { \mathopen{}\mathclose\bgroup\originalleft}\verts{\aftergroup\egroup\originalright}\vert}{\mathopen{}\mathclose\bgroup\originalleft}(m_d+ r(s){\aftergroup\egroup\originalright}).\end{aligned}\ ] ] choose again @xmath179 , @xmath374 , and put for @xmath375 and @xmath376 @xmath377}{\mathopen{}\mathclose\bgroup\originalleft}(s_{{\bm{u}}^{(i)}}{\mathopen{}\mathclose\bgroup\originalleft}(-\frac{c}j{\aftergroup\egroup\originalright}){\aftergroup\egroup\originalright})\ ] ] where @xmath40 are independent copies of @xmath13 . then @xmath182 is the number of those processes among @xmath40 , which exceed the threshold @xmath378 in at least one point .
if @xmath357 , then each @xmath182 is binomial @xmath187 distributed with @xmath379 put again @xmath380 repeating the arguments in the proof of theorem [ thm : limit_distribution_of_t_n ] , one shows that its assertion carries over to the functional space as well .
[ thm : limit_distribution_of_t_n_for_functional_data ] suppose that the copula process @xmath356 $ ] is in the @xmath38-neighborhood of a gpcp for some @xmath37 . in this case
the remainder term @xmath381 in expansion is of order @xmath382 as @xmath383 .
let @xmath193 satisfy @xmath194 , @xmath195 and @xmath260 as @xmath78 .
then we obtain @xmath197 with @xmath329 and @xmath330 as in theorem [ thm : limit_distribution_of_t_n ] . in what follows
we will extend theorem [ thm : limit_distribution_of_t_n_for_functional_data ] to the case when observing the underlying copula process is subject to a certain kind of nuisance .
let @xmath384}\in c[0,1]$ ] be a stochastic process with identical continuous univariate marginal df , i.e. , @xmath385 , @xmath334 $ ] , is a continuous function in @xmath287 .
@xmath270 is said to be in the functional max - domain of attraction of a max - stable process @xmath386}$ ] , if the copula process @xmath387}$ ] satisfies @xmath357 , where @xmath336 is a smsp , and the df @xmath0 satisfies the univariate extreme value condition ; for the univariate case we refer to ( * ? ? ?
* section 2.1 ) , among others .
let @xmath388 be independent copies of the process @xmath270 and denote the sample df pertaining to the univariate iid observations @xmath389 by @xmath390}\bigl(x_0^{(i)}\bigr)$ ] , @xmath287 .
as in section [ sec : arbitrary_random_vector ] , fix @xmath271 , choose an arbitrary subset @xmath272 of @xmath273 of size @xmath391 , and put for @xmath376 @xmath392}{\mathopen{}\mathclose\bgroup\originalleft}(\int_0 ^ 1 1_{(\gamma(c),\infty)}\bigl(x_t^{(i)}\bigr)\,dt { \aftergroup\egroup\originalright } ) \\ & \hphantom{:}= \sum_{i\in m(n ) } 1_{[0,1)}{\mathopen{}\mathclose\bgroup\originalleft}(\int_0 ^ 1 1_{(-\infty,\gamma(c)]}\bigl(x_t^{(i)}\bigr)\,dt { \aftergroup\egroup\originalright } ) \\ & \hphantom{:}= \sum_{i\in m(n ) } 1_{[0,1)}{\mathopen{}\mathclose\bgroup\originalleft}(\int_0 ^ 1 1_{{\mathopen{}\mathclose\bgroup\originalleft}[0,1-\frac{c}{j}{\aftergroup\egroup\originalright}]}\bigl(u_t^{(i)}\bigr)\,dt { \aftergroup\egroup\originalright})\\ & = \sum_{i\in m(n ) } 1_{{{\mathopen{}\mathclose\bgroup\originalleft}\{{\bm{u}}^{(i)}\nleq ( 1-\frac{c}{j})1_{[0,1]}{\aftergroup\egroup\originalright}\}}}\end{aligned}\ ] ] where @xmath393 and the next to last equation holds almost surely .
again , we replace the marginal df @xmath0 with its empirical counterpart and obtain analogously with @xmath394 @xmath395}{\mathopen{}\mathclose\bgroup\originalleft}(\int_0 ^ 1 1_{(\hat\gamma_n(c),\infty)}\bigl(x_t^{(i)}\bigr)\,dt { \aftergroup\egroup\originalright})\ ] ] thus the rv @xmath396 is the total number of processes @xmath397}$ ] among @xmath398 , which exceed the random threshold @xmath399 for some @xmath334 $ ] .
note that the distribution of the rv @xmath400 does not depend on @xmath0 but on the copula process @xmath13 since @xmath401}\bigl(x_t^{(i)}\bigr)\,dt { \aftergroup\egroup\originalright } ) \\ & \hphantom{:}= \sum_{i\in m(n ) } 1_{[0,1)}{\mathopen{}\mathclose\bgroup\originalleft}(\int_0 ^ 1 1_{[0,u_{\langle n(1-\frac{c}{j})\rangle : n}]}\bigl(u_t^{(i)}\bigr)\,dt { \aftergroup\egroup\originalright } ) \\ & \hphantom{:}= \sum_{i\in m(n ) } 1_{\bigl\{{\bm{u}}^{(i)}\nleq u_{\langle n(1-\frac{c}{j})\rangle : n}1_{[0,1]}\bigr\}}\end{aligned}\ ] ] with probability one , where @xmath402 denote the ordered values of @xmath403 .
the following auxiliary result is the extension of lemma [ lem : crucial_approximation_of_n_j ] to function space .
[ lem : crucial_approximation_function_space ] suppose that the copula process @xmath356 $ ] corresponding to @xmath270 is in the @xmath38-neighborhood of a gpcp for some @xmath37 . in this case
the remainder term @xmath381 in expansion is of order @xmath404 as @xmath103 .
choose @xmath405 and @xmath302 such that @xmath300 , @xmath406 , @xmath194 , @xmath303 and @xmath327 as @xmath304 .
then we obtain for @xmath407 @xmath408 we have with probability one @xmath409}{\aftergroup\egroup\originalright}\ } } } - 1_{\bigl\{{\bm{u}}^{(i)}\nleq u_{\langle n(1-\frac{c_n}{j})\rangle : n}1_{[0,1]}\bigr\ } } { \aftergroup\egroup\originalright } ) \\ & = \sum_{i\in m(n ) } 1_{\bigl\{{\bm{u}}^{(i)}\nleq ( 1-\frac{c_n}{j})1_{[0,1]},\ { \bm{u}}^{(i)}\le
u_{\langle n(1-\frac{c_n}{j})\rangle : n}1_{[0,1]}\bigr\ } } \\ & \mathrel{\hphantom{= } } { } - \sum_{i\in m(n ) } 1_{\bigl\{{\bm{u}}^{(i)}\nleq u_{\langle n(1-\frac{c_n}{j})\rangle : n}1_{[0,1]},\ { \bm{u}}^{(i)}\le ( 1-\frac{c_n}{j})1_{[0,1]}\bigr\ } } \\ & = : r_n - t_n.\end{aligned}\ ] ] we show in what follows that @xmath410 proceeding as in the proof of lemma [ lem : crucial_approximation_of_n_j ] : put @xmath311 with @xmath312 . note that @xmath411 satisfies @xmath412 for large values of @xmath192 as well as @xmath413 now we obtain by expansion , if @xmath192 is sufficiently large , @xmath414},\ { \bm{u}}^{(1)}\le u_{\langle n(1-\frac{c_n}{j})\rangle : n}1_{[0,1 ] } { \aftergroup\egroup\originalright } ) \\ & = p{\mathopen{}\mathclose\bgroup\originalleft } ( { \bm{u}}^{(1)}\le u_{\langle n(1-\frac{c_n}{j})\rangle : n}1_{[0,1 ] } { \aftergroup\egroup\originalright } ) - p\bigl ( { \bm{u}}^{(1)}\le \min\bigl\{1-\frac{c_n}{j } , u_{\langle n(1-\frac{c_n}{j})\rangle : n}\bigr\}1_{[0,1 ] } \bigr ) \\ & \le p{\mathopen{}\mathclose\bgroup\originalleft } ( { \bm{u}}^{(1)}\le ( \mu_n + { \varepsilon}_n)1_{[0,1 ] } { \aftergroup\egroup\originalright } ) + p{\mathopen{}\mathclose\bgroup\originalleft } ( u_{\langle n(1-\frac{c_n}{j})\rangle : n } \ge \mu_n+{\varepsilon}_n{\aftergroup\egroup\originalright } ) \\ & \mathrel{\hphantom{\le } } { } - p\bigl ( { \bm{u}}^{(1)}\le \min\bigl\{1-\frac{c_n}{j } , \mu_n-{\varepsilon}_n\bigr\}1_{[0,1 ] } \bigr ) + p{\mathopen{}\mathclose\bgroup\originalleft}(u_{\langle n(1-\frac{c_n}{j})\rangle : n } \le \mu_n-{\varepsilon}_n{\aftergroup\egroup\originalright } ) \\ & = p{\mathopen{}\mathclose\bgroup\originalleft}({{\mathopen{}\mathclose\bgroup\originalleft}\vertu_{\langle n(1-\frac{c_n}{j})\rangle : n } - \mu_n{\aftergroup\egroup\originalright}\vert}\ge{\varepsilon}_n{\aftergroup\egroup\originalright } ) - ( 1-\mu_n-{\varepsilon}_n){\mathopen{}\mathclose\bgroup\originalleft}(m_d+r(\mu_n+{\varepsilon}_n-1){\aftergroup\egroup\originalright } ) \\ & \mathrel{\hphantom{= } } { } + \max\bigl\{\frac{c_n}{j } , 1-\mu_n+{\varepsilon}_n\bigr\ } { \mathopen{}\mathclose\bgroup\originalleft}(m_d + r{\mathopen{}\mathclose\bgroup\originalleft}(-\max\bigl\{\frac{c_n}{j } , 1-\mu_n+{\varepsilon}_n\bigr\}{\aftergroup\egroup\originalright}){\aftergroup\egroup\originalright } ) \\ & \le p{\mathopen{}\mathclose\bgroup\originalleft}({{\mathopen{}\mathclose\bgroup\originalleft}\vertu_{\langle n(1-\frac{c_n}{j})\rangle : n } - \mu_n{\aftergroup\egroup\originalright}\vert}\ge{\varepsilon}_n{\aftergroup\egroup\originalright } ) - { \mathopen{}\mathclose\bgroup\originalleft}(\frac{n}{n+1 } \frac{c_n}{j } - { \varepsilon}_n{\aftergroup\egroup\originalright}){\mathopen{}\mathclose\bgroup\originalleft}(m_d + o{\mathopen{}\mathclose\bgroup\originalleft}(c_n^\delta{\aftergroup\egroup\originalright}){\aftergroup\egroup\originalright } ) \\ & \mathrel{\hphantom{= } } { } + { \mathopen{}\mathclose\bgroup\originalleft}(\frac1{n+1 } + \frac{n}{n+1}\frac{c_n}{j } + { \varepsilon}_n{\aftergroup\egroup\originalright } ) { \mathopen{}\mathclose\bgroup\originalleft}(m_d + o{\mathopen{}\mathclose\bgroup\originalleft}(c_n^\delta{\aftergroup\egroup\originalright}){\aftergroup\egroup\originalright } ) \\ & = p{\mathopen{}\mathclose\bgroup\originalleft}({{\mathopen{}\mathclose\bgroup\originalleft}\vertu_{\langle n(1-\frac{c_n}{j})\rangle : n } - \mu_n{\aftergroup\egroup\originalright}\vert}\ge{\varepsilon}_n{\aftergroup\egroup\originalright } ) + o{\mathopen{}\mathclose\bgroup\originalleft}(c_n^{1+\delta}{\aftergroup\egroup\originalright } ) + o{\mathopen{}\mathclose\bgroup\originalleft}(\frac1n + { \varepsilon}_n{\aftergroup\egroup\originalright}).\end{aligned}\ ] ] the arguments in the proof of lemma [ lem : crucial_approximation_of_n_j ] show @xmath415 as @xmath78 and , thus , ; recall @xmath416 and note that @xmath417 repeating the above arguments one shows that @xmath418 as @xmath78 as well , which completes the proof of lemma [ lem : crucial_approximation_function_space ] .
analogously to section [ sec : test_multivariate ] we now choose @xmath405 and replace @xmath182 in with @xmath299 and obtain @xmath419 by this statistic we can in particular check , whether the copula process @xmath387}$ ] pertaining to @xmath270 is in a @xmath38-neigborhood of some gpcp @xmath368 .
its distribution does not depend on the marginal df @xmath0 of @xmath270 but on the copula process @xmath13 .
the next result follows from lemma [ lem : crucial_approximation_function_space ] and the arguments in the proof of theorem [ thm : limit_distribution_of_t_n ] .
we have under the conditions of lemma [ lem : crucial_approximation_function_space ] @xmath420 with @xmath329 and @xmath330 as in theorem [ thm : limit_distribution_of_t_n ] .
let @xmath421 be two independent and standard negative exponential distributed rv . put for @xmath334 $ ] @xmath422 where the rv @xmath45 is independent of @xmath421 and follows the df @xmath423 defined in lemma [ lem : copula_not_in_domain_of_attraction ] with @xmath47 $ ] .
note that @xmath424 and , thus , the process @xmath384}$ ] has identical continuous marginal df . for @xmath82 ,
the process @xmath270 is a _ generalized pareto process _ , whose pertaining copula process is in the max - domain of attraction of a smsp , see @xcite . for @xmath425
this is not true : just consider the bivariate rv @xmath426 , where @xmath427 , @xmath428 and repeat the arguments in lemma [ lem : copula_not_in_domain_of_attraction ] .
observing a complete process on @xmath89 $ ] as in the preceding section might be a too restrictive assumption .
instead we will require in what follows that we observe stochastic processes with sample paths in @xmath44 $ ] only through an increasing grid of points in @xmath89 $ ] .
let @xmath365 $ ] be a gpcp with pertaining @xmath23-norm
@xmath429}({{\mathopen{}\mathclose\bgroup\originalleft}\vertf(t){\aftergroup\egroup\originalright}\vert}z_t){\aftergroup\egroup\originalright}),\qquad f\in e[0,1].\ ] ] choose a grid of points @xmath430 .
then the rv @xmath431 follows a gpc , whose corresponding @xmath23-norm is given by @xmath432 let now @xmath433 depend on @xmath192 .
if we require that @xmath434 then , by the continuity of @xmath345}$ ] , @xmath435 } z_t \qquad \mathrm{a.s.},\ ] ] and , thus , the sequence of generator constants converges : @xmath436}z_t{\aftergroup\egroup\originalright } ) = m_d.\ ] ] suppose we are given @xmath192 independent copies of a copula process @xmath13 .
the projection of each process onto the grid @xmath437 yields @xmath192 iid rv in @xmath438 , which follow a copula .
let @xmath182 as defined in be based on these rv .
note that @xmath182 depends on @xmath439 as well .
but in order not to overload our notation we suppress the dependence on the dimension .
moreover we require that @xmath440}+cf{\aftergroup\egroup\originalright } ) = 1- c{{\mathopen{}\mathclose\bgroup\originalleft}\vertf{\aftergroup\egroup\originalright}\vert}_d + o(c^{1+\delta})\ ] ] holds uniformly for all @xmath441 $ ] satisfying @xmath442 . again
a suitable version of the central limit theorem implies @xmath443 and thus @xmath444 yielding @xmath445 theorem [ thm : limit_distribution_of_t_n ] now carries over : [ thm : asymptotic_distribution_of_t_n_for_gpcp ] let @xmath13 be a copula process satisfying .
choose a grid of points @xmath430 with @xmath446 and @xmath447 as @xmath78 .
let @xmath253 as defined in be based on the projections of @xmath192 independent copies of @xmath13 onto this increasing grid of points .
let @xmath193 satisfy @xmath194 , @xmath195 and @xmath196 as @xmath78 .
then we obtain @xmath197 with @xmath329 and @xmath330 as in theorem [ thm : limit_distribution_of_t_n ] .
now we will extend theorem [ thm : asymptotic_distribution_of_t_n_for_gpcp ] to a general process @xmath384}\in c[0,1]$ ] with continuous marginal df @xmath448 , @xmath334 $ ] .
we want to test whether the copula process @xmath449}\in c[0,1]$ ] corresponding to @xmath270 satisfies . as before this will be done by projecting the process @xmath270 onto a grid of points @xmath450 with @xmath451 and and @xmath452 .
let @xmath388 be independent copies of @xmath270 and consider the @xmath192 iid rv of projections @xmath453 , @xmath269 .
fix @xmath271 , choose an arbitrary subset @xmath272 of @xmath273 of size @xmath274 and put for @xmath180 @xmath454}{\mathopen{}\mathclose\bgroup\originalleft}({\bm{x}}_{d_n}^{(i)}{\aftergroup\egroup\originalright})\end{aligned}\ ] ] which is the number of all rv @xmath455 exceeding the vector @xmath456 in at least one component .
clearly we have @xmath457}{\mathopen{}\mathclose\bgroup\originalleft}({\bm{u}}_{d_n}^{(i)}{\aftergroup\egroup\originalright})\ ] ] almost surely where @xmath458 and @xmath459 is the copula process of @xmath460 .
again we replace @xmath461 with @xmath462 where @xmath463 and @xmath464}{\mathopen{}\mathclose\bgroup\originalleft}(x_{t_r^{(d_n)}}^{(i)}{\aftergroup\egroup\originalright})$ ] , @xmath465 , yielding an estimator of @xmath182 : @xmath466}{\mathopen{}\mathclose\bgroup\originalleft}({\bm{x}}_{d_n}^{(i)}{\aftergroup\egroup\originalright}).\end{aligned}\ ] ] we have @xmath467 where @xmath468 denote the ordered values of @xmath469 for each @xmath470 and @xmath471 is again the right integer neighbor of @xmath472 .
since transforming each @xmath473 by its df @xmath474 does not alter the value of @xmath396 with probability one , we obtain @xmath475 } \bigl({\bm{u}}_{d_n}^{(i)}\bigr)\ ] ] almost surely where @xmath476 are the order statistics of @xmath477 . since @xmath478 are independent copies of the rv @xmath479 , the distribution of @xmath480 does not depend on the marginal df @xmath481 .
the following auxiliary result is crucial .
[ newlem : crucial_approximation_of_n_j ] suppose that @xmath482 .
let @xmath302 satisfy @xmath194 , @xmath303 and @xmath483 as @xmath304 .
if @xmath484 satisfies @xmath485 as @xmath78 , then we obtain for @xmath305 @xmath486 we have almost surely @xmath487 } \bigl({\bm{u}}_{d_n}^{(i)}\bigr ) - 1_{{\mathopen{}\mathclose\bgroup\originalleft}[{\bm{0 } } , { \mathopen{}\mathclose\bgroup\originalleft}(1-\frac{c_n}{j}{\aftergroup\egroup\originalright}){\bm{1}}{\aftergroup\egroup\originalright}]}\bigl({\bm{u}}_{d_n}^{(i)}\bigr ) { \aftergroup\egroup\originalright } ) \\ & = \sum_{i\in m(n ) } 1_{\times_{r=1}^{d_n}{\mathopen{}\mathclose\bgroup\originalleft}[\vphantom{\frac{c_n}{j}}\smash{0,u_{\langle n(1-\frac{c_n}{j})\rangle : n , r}}{\aftergroup\egroup\originalright } ] } \bigl({\bm{u}}_{d_n}^{(i)}\bigr ) { \mathopen{}\mathclose\bgroup\originalleft}(1 - 1_{{\mathopen{}\mathclose\bgroup\originalleft}[{\bm{0 } } , { \mathopen{}\mathclose\bgroup\originalleft}(1-\frac{c_n}{j}{\aftergroup\egroup\originalright}){\bm{1}}{\aftergroup\egroup\originalright}]}\bigl({\bm{u}}_{d_n}^{(i)}\bigr ) { \aftergroup\egroup\originalright } ) \\ & \mathrel{\hphantom{= } } { } - \sum_{i\in m(n ) } 1_{{\mathopen{}\mathclose\bgroup\originalleft}[{\bm{0 } } , { \mathopen{}\mathclose\bgroup\originalleft}(1-\frac{c_n}{j}{\aftergroup\egroup\originalright}){\bm{1}}{\aftergroup\egroup\originalright}]}\bigl({\bm{u}}_{d_n}^{(i)}\bigr ) { \mathopen{}\mathclose\bgroup\originalleft}(1 - 1_{\times_{r=1}^d[0,u_{\langle n(1-\frac{c_n}{j})\rangle : n , r } ] } \bigl({\bm{u}}_{d_n}^{(i)}\bigr){\aftergroup\egroup\originalright } ) \\ & = : r_n - t_n.\end{aligned}\ ] ] in what follows we show @xmath488}{\mathopen{}\mathclose\bgroup\originalleft}(u_{t_r^{(d_n)}}^{(i)}{\aftergroup\egroup\originalright}){\aftergroup\egroup\originalright } ) = o(1)\ ] ] and thus @xmath309 ; note that @xmath489}{\mathopen{}\mathclose\bgroup\originalleft}(u_{t_r^{(d_n)}}^{(i)}{\aftergroup\egroup\originalright}).\end{aligned}\ ] ] put @xmath311 with @xmath490 .
we have with @xmath491 @xmath492 where the first term is of order @xmath493 ; recall that @xmath494 is uniformly distributed on @xmath6 .
it , therefore , suffices to show that the second term is of order @xmath495 as well . as in the proof of lemma [ lem : crucial_approximation_of_n_j ]
we obtain for @xmath319 @xmath496 and @xmath497 .
we have , moreover , @xmath498 for large @xmath192 and , since @xmath499 as @xmath78 , @xmath500 by repeating the above arguments one shows that @xmath323 as well , which completes the proof of lemma [ newlem : crucial_approximation_of_n_j ] .
now we consider the modified test statistic @xmath324 which does not depend on the marginal df @xmath448 , @xmath334 $ ] , of the process @xmath270 .
the following result is a consequence of the arguments in the proof of theorem [ thm : limit_distribution_of_t_n ] and lemma [ newlem : crucial_approximation_of_n_j ] .
[ newthm : limit_distribution_of_test_statistic_general ] suppose that the process @xmath384}\in c[0,1]$ ] has continuous marginal df @xmath448 , @xmath501 $ ] and that the pertaining copula process @xmath502}$ ] satisfies .
choose a grid of points @xmath503 with @xmath484 and @xmath447 as @xmath78 .
let @xmath325 satisfy @xmath482 and let @xmath193 satisfy @xmath194 , @xmath303 , @xmath504 and @xmath483 as @xmath78 .
then we obtain @xmath328 with @xmath329 and @xmath330 as in theorem [ thm : limit_distribution_of_t_n ] .
in this section we provide some simulations , which indicate the performance of the test statistic @xmath217 from theorem [ thm : limit_distribution_of_t_n ] .
all computations were performed using the package written by pierre lafaye de micheaux and pierre duchesne .
we chose @xcite
@xcite method for computing the @xmath2-values of our test statistics ; cf .
@xcite for an overview of simulation techniques of quadratic forms in normal variables .
therefore we chose @xmath505 , @xmath506 and @xmath201 .
we generated @xmath507 independent realizations of @xmath217 that we denote by @xmath508 and we computed the asymptotic @xmath2-values @xmath509 , @xmath510 , where @xmath511 is the df of @xmath512 in theorem [ thm : limit_distribution_of_t_n ] .
the values @xmath513 are filed in increasing order , @xmath514 , and we plot the points @xmath515 this quantile plot is a discrete approximation of the quantile function of the @xmath2-value of @xmath217 , which visualizes the performance of the test statistic @xmath217 . if the underlying copula is in a @xmath38-neighborhood of a gpc , then the points @xmath516 , @xmath517 , should approximately lie along the line @xmath518 , @xmath519 $ ] , whereas otherwise @xmath520 should be significantly smaller than @xmath521 for many @xmath522 .
2[gpc ] .,title="fig:",width=6 ] .,width=6 ] 2[copula not in @xmath101 ] .,title="fig:",width=6 ] .,width=6 ] 2[normal copula with coefficient of correlation @xmath524 .,title="fig:",width=6 ] .,width=6 ] 2[clayton copula with parameter @xmath525 .,title="fig:",width=6 ] .,width=6 ] as can be seen in figures 18 , the test is quite reliable in detecting a gpc itself . however ,
if the underlying copula is _ not _ a gpc , the corresponding @xmath2-value is quite sensitive to the selection of @xmath147 .
e.g. , if we decrease the value of @xmath147 from @xmath526 to @xmath527 , a copula that is not even in the domain of attraction of an extreme value distribution can not be detected anymore , cf .
figure [ fig : copula_not_in_doa_0_2 ] and figure [ fig : copula_not_in_doa_0_01 ] . on the other hand , there are copulas satisfying the @xmath38-neighborhood condition that perform well with @xmath528 , such as the normal copula in figure [ fig : normal_copula_0_2 ] , and those that do not , such as the clayton copula in figure [ fig : clayton_copula_0_2 ] .
the aforementioned disadvantages can , however , be overcome by considering the @xmath2-value as a function of the threshold @xmath147 .
therefore we simulated a single data set of sample size @xmath505 and plotted the @xmath2-value for each @xmath147 of a some grid @xmath529 , see figures 912 .
it turns out that the @xmath2-value curve of the considered gpc is above the @xmath521-line for @xmath530 , roughly .
in contrast , the copula in figure [ fig : copula_not_in_doa ] has a peak for intermediate values of @xmath147 but , for shrinking @xmath147 , decreases again below the @xmath521-line .
finally , copulas in the @xmath38-neighborhood of a gpc behave similar to the gpc in figure [ fig : gpc ] except that the point of intersection with the @xmath521-line is notably smaller than @xmath531 . 2 , title="fig:",width=6 ] , width=6 ] 2 the shapes of the @xmath2-value plots in figures 912 seem to be a reliable tool for the decision whether or not to reject the hypothesis .
a great advantage of this approach is that a practitioner does not need to specify a suitable value of the threshold @xmath147 , which is a rather complicated task , but can make the decision based on a highly intelligible graphical tool .
a further analysis of these kind of @xmath2-value plots is part of future work .
the authors are grateful to kilani ghoudi for his hint to compute the asymptotic distribution of the above test statistics using @xcite s @xcite method . | a multivariate distribution function @xmath0 is in the max - domain of attraction of an extreme value distribution if and only if this is true for the copula corresponding to @xmath0 and its univariate margins .
@xcite have shown that a copula satisfies the extreme value condition if and only if the copula is tail equivalent to a generalized pareto copula ( gpc ) . in this paper
we propose a @xmath1-goodness - of - fit test in arbitrary dimension for testing whether a copula is in a certain neighborhood of a gpc .
the test can be applied to stochastic processes as well to check whether the corresponding copula process is close to a generalized pareto process .
since the @xmath2-value of the proposed test is highly sensitive to a proper selection of a certain threshold , we also present a graphical tool that makes the decision , whether or not to reject the hypothesis , more comfortable . |
feynman integrals in dimensional regularization in @xmath2 dimensions are invariant under poincar transformations . by applying these transformations at the integrand - level , one obtains nontrivial linear relations among different integrals , the integration - by - parts @xcite relations .
these relations can be used to reduce the large number of feynman integrals relevant to a particular process to a much smaller number of so - called master integrals .
this reduction is typically carried out using a lexicographic ordering of the integrals @xcite , implemented in computer algebra routines , for example in the codes @xcite or @xcite .
the type of master integrals that appear in a given process depends only on the external kinematics , and on possible internal propagator masses .
all two - loop five - parton amplitudes relate to a common set of master integrals : massless on - shell five - point functions at two loops .
these can be further classified into genuine five - point functions , four - point functions with one off - shell leg , three - point functions with up to two off - shell legs and off - shell two - point functions .
up to the four - point level , these functions appeared in the context of the derivation of the two - loop amplitude for @xmath3 jets @xcite and were already computed long ago @xcite .
the genuine five - point functions depend on five independent mandelstam invariants , @xmath4 where @xmath5 .
they are therefore considerably more complicated than the four - point functions , since the latter depend on three variables only .
we find in total 25 new integrals ( 10 planar and 15 nonplanar ) .
the planar integrals can be given in terms of four integral topologies , displayed in figure [ fig : pentagonintegrals ] .
there are @xmath6 , and @xmath7 master integrals for topologies ( a ) , ( b ) , ( c ) , and ( d ) , respectively . to compute the integrals ,
we derive differential equations for them in the @xmath8 .
the system of differential equations is then brought into a canonical form @xcite by means of a transformation of the basis of master integrals to integrals having unit leading singularities @xcite .
the canonical form we find is @xmath9 \vec{f}(v_i ; \epsilon ) \,,\end{aligned}\ ] ] where @xmath10 is the set of @xmath11 master integrals , the differential @xmath12 comprises partial derivatives w.r.t .
@xmath8 , and @xmath13 are constant ( kinematic and @xmath14-independent ) matrices .
the collection of letters @xmath15 specify the function alphabet @xmath16 .
the latter is given by @xmath17 and cyclic permutations thereof .
here , @xmath18 and the gram determinant @xmath19 , with @xmath20 .
it is interesting to note that @xmath21 $ ] and @xmath22 , where @xmath23 $ ] .
the full set of master integrals can be obtained by direct integration of the differential equations , order - by - order in @xmath14 , in terms of chen iterated integrals @xcite .
for a practical application of the latter to multivariable feynman integrals , including their numerical evaluation , see @xcite .
the boundary conditions are determined from consistency conditions , such as the absence of unphysical branch cuts .
this is particularly simple to implement in the canonical form ( [ decanonical ] ) of the differential equations , cf .
@xcite .
massless scattering is naturally parametrized using momentum twistor variables @xcite that solve both the on - shell as well as the momentum conservation constraints .
we find that in these variables the alphabet ( [ functionalphabet ] ) becomes rational .
this implies that , when expressed in terms of these variables , the chen iterated integrals degenerate to multiple polylogarithms @xcite , for which efficient and precise numerical representations exist @xcite . all subtopologies at four points and below are recomputed in terms of the momentum twistor variables , yielding agreement with earlier results @xcite .
we further validate our integrals by analytically computing all five - gluon amplitudes in @xmath24 super yang - mills at two loops .
their expression was initially conjectured in @xcite , tested numerically in @xcite , and proven in @xcite from a ward identity for dual conformal symmetry .
our calculation is the first direct analytical one , and we find complete agreement with the above references .
technical details on the determination of the five - point master integrals will be documented in a separate publication @xcite .
we consider the unrenormalized all - plus five - gluon amplitude at leading color : @xmath25 here @xmath26 denote all noncyclic rotations of five points , and @xcite @xmath27 since the amplitude vanishes at tree level , it is finite at the one - loop level @xcite , @xmath28 with @xmath29 and @xmath30 at two loops , the infrared and ultraviolet divergent terms can be predicted in terms of the one - loop result .
this motivates the definition of a finite remainder @xmath31 according to pole in eq .
( [ definitionremainder ] ) is due to an exact cancellation of ultraviolet and infrared terms in unrenormalized two - loop amplitudes that are finite at one loop .
after uv renormalization , the well - known infrared pole structure @xcite is recovered . ] @xmath32 + r \;f^{(2)}_{5 } + { \mathcal o}({\epsilon})\,.\end{aligned}\ ] ] we use the integral representation of @xcite and express it in terms of our basis of integrals . plugging in the solution for the @xmath14-expansion of the latter , we analytically verify the divergence structure of eq .
( [ definitionremainder ] ) . to define the finite remainder function ,
the expansion of ( [ oneloop ] ) to order @xmath33 is derived , which involves the one - loop massless pentagon integral to this order , computed from its differential equation . in the finite remainder ,
remarkably all chen iterated integrals of weight one , three and four cancel out .
we then express the remaining weight - two functions in terms of dilogarithms , and find the following expression for the finite remainder : @xmath34 } { ( v_2 + v_3 -v_5 ) } i_{23,5 } \right .
\nonumber \\ & \hspace{-0.7 cm } \left .
+ \frac{1}{6 } \frac { { \rm tr}\left [ ( 1+\gamma_{5 } ) \slashed{p}_{4 } \slashed{p}_{5 } \slashed{p}_{1 } \slashed{p}_{2 } \right]^2 } { v_1 v_4 } + \frac{10}{3 } v_1 v_2 + \frac{2}{3 } v_1 v_3 \right\ } \ , .
\end{aligned}\ ] ] where @xmath35 cyclically shifts all indices ( of @xmath36 , @xmath37 , and @xmath38 ) by @xmath39 , and where @xmath40 \nonumber \\ & - { \rm li}_{2}\left[\frac{v_5-v_3}{v_2}\right ] - { \rm li}_{2}\left[\frac{v_5-v_2}{v_3}\right ] \,.\end{aligned}\ ] ] note that eq .
( [ resultf5nice ] ) contains both parity odd and even terms .
we remark that the trace can also be written in a natural way using momentum twistors .
we compared our analytical result for the unrenormalized two - loop amplitude ( [ definitionremainder ] ) with the numerical values quoted in @xcite for specific phase space points in the euclidean region , finding full agreement . in the euclidean region , this expression is single - valued and real .
we note that eq .
( [ formulaonemassbox ] ) can be rewritten in a form where this is manifest , and that our result can straightforwardly be analytically continued to other kinematical regions .
the result above is for pure yang - mills theory .
we would like to mention that the full @xmath41 dependence can be reconstructed in a simple way : the @xmath42 terms only come from a restricted class of diagrams , and the remaining @xmath41 terms are fixed by supersymmetry @xcite .
scattering amplitudes have universal factorization properties in soft and collinear limits .
they serve as an important check of our result .
we take the @xmath43 collinear limit , without loss of generality . in the limit ,
one expects ( cf . fig .
[ fig : splitting ] ) @xmath44 where ` @xmath45 ' are splitting amplitudes @xcite .
the amplitudes appearing on the right hand side of eq .
( [ collinear ] ) can be found in @xcite .
taking the collinear limit of ( [ definitionremainder ] ) , we recover the structure predicted by ( [ collinear ] ) .
it is interesting to note in this context that the second line of eq .
( [ resultf5nice ] ) contains terms that behave as @xmath46/\langle 4 5\rangle$ ] in this limit ( and are amplified by @xmath47 from the overall factor @xmath48 ) .
the latter reproduces a contribution from the helicity - violating one - loop splitting function @xmath49 .
the simplicity of our result for the all - plus amplitude in qcd is reminiscent of similar results for six - gluon amplitudes in @xmath1 sym @xcite . in the latter case ,
the function alphabet is related to cluster algebras @xcite , and it would be interesting to know whether this is also true for our five - point function alphabet of eq .
( [ functionalphabet ] ) , or perhaps for a subset relevant for calculations up to finite parts of amplitudes .
it is interesting to investigate possible positivity properties @xcite of our result .
this is naturally done using momentum twistors . in qcd ,
the kinematics depends on the twistors @xmath50 ( @xmath51 ) , and on an infinity twistor @xmath52 .
it is tempting to treat the latter in the same way as the loop integration variable in the discussion of one - loop maximally helicity violating amplitudes in @xcite .
this defines a kinematical region where in particular @xmath53 and @xmath54<0 $ ] ( and cyclic ) .
interestingly , in this region , @xmath55 is positive ( and the same holds for its @xmath56-point generalization @xcite ) . at two loops , we find that all terms are positive , except for the terms involving @xmath57 , which are negative
. this can be seen by noting that @xmath58 is a one - loop one - mass box function in six dimensions .
it could be that the positivity properties at two loops are obscured by the infrared subtraction ; see eq .
( [ definitionremainder ] ) . however , one may speculate that the natural building blocks of definite sign that we found above point towards additional structure that is yet to be uncovered .
our result for the planar master integrals provides the maximal set of polylogarithmic functions that can appear in a generic planar massless five - particle scattering amplitude at two loops .
therefore we reduce the calculation of any such scattering amplitude to the determination of the algebraic coefficients accompanying the integral basis .
this can be envisioned using a variety of related methods @xcite . in the one - loop case ,
the knowledge of the integral basis triggered a revolution in our ability to compute amplitudes , and we expect the same to occur here .
we would also like to mention that our result for the functional basis provides the foundation for bootstrap techniques to be used , where one makes an ansatz for the remainder function of the type @xmath59 , where @xmath60 are members of the functional basis , and @xmath61 are certain kinematic - dependent factors .
see @xcite for applications of such a bootstrap approach to six - point amplitudes in @xmath24 sym . in the latter theory ,
the kinematic dependence of the @xmath61 is related , at least conjecturally , to leading singularities @xcite , simplifying the above ansatz .
we anticipate that the basis of pentagon functions that we provide will be useful for bootstrapping generic five - point amplitudes , especially when combined with an improved understanding of the possible algebraic coefficients that are allowed to appear in qcd .
an obvious next step consists in calculating the loop integrals in the nonplanar case ( the method for finding an appropriate integral basis @xcite does not rely on planarity ) and applying them to the all - plus loop integrand that was found very recently @xcite .
further generalizations include making one of the external legs off - shell , where we anticipate chen iterated integrals to play an even more important role than in the present work .
this will allow to describe processes such as the production of higgs boson plus jets at nnlo .
we would like to thank simon badger for numerous discussions and for providing the numerical routines of @xcite , which were used to validate our results . the expression ( [ resultf5nice ] ) for the finite remainder agrees with the very recent result of @xcite , which uncovered a typo in the original version of our manuscript .
this work was supported in part by the schweizer nationalfonds under grant 200020 - 162487 , as well as by the european commission through the erc advanced grant mc@nnlo " ( 340983 ) .
is supported in part by a gfk fellowship and by the prisma cluster of excellence at mainz university .
99 s. badger , h. frellesvig and y. zhang , jhep * 1312 * ( 2013 ) 045 .
s. badger , g. mogull , a. ochirov and d. oconnell , jhep * 1510 * ( 2015 ) 064 .
z. bern , l. j. dixon and d. a. kosower , jhep * 0001 * ( 2000 ) 027 .
j. m. drummond , j. m. henn and j. trnka , jhep * 1104 * , 083 ( 2011 ) .
l. j. dixon , j. m. drummond and j. m. henn , jhep * 1201 * , 024 ( 2012 ) .
g. heinrich , int .
j. mod .
a * 23 * ( 2008 ) 1457 ; s. borowka , g. heinrich , s. p. jones , m. kerner , j. schlenk and t. zirke , comput .
commun .
* 196 * ( 2015 ) 470 .
a. v. smirnov , v. a. smirnov and m. tentyukov , comput .
commun .
* 182 * ( 2011 ) 790 ; a. v. smirnov , comput .
commun .
* 185 * ( 2014 ) 2090 .
a. v. smirnov , jhep * 0810 * ( 2008 ) 107 ; a. v. smirnov , comput .
commun . * 185 * , 2090 ( 2014 ) . c. studerus , comput .
commun .
* 181 * ( 2010 ) 1293 ; a. von manteuffel and c. studerus , arxiv:1201.4330 [ hep - ph ] . c. anastasiou , z. bern , l. j. dixon and d. a. kosower , phys .
lett . *
91 * , 251602 ( 2003 ) .
f. cachazo , m. spradlin and a. volovich , phys .
d * 74 * , 045020 ( 2006 ) ; z. bern , m. czakon , d. a. kosower , r. roiban and v. a. smirnov , phys .
lett . * 97 * , 181601 ( 2006 ) .
z. bern , a. de freitas and l. j. dixon , jhep * 0203 * , 018 ( 2002 ) .
d. a. kosower and p. uwer , nucl .
b * 563 * ( 1999 ) 477 ; z. bern , v. del duca , w. b. kilgore and c. r. schmidt , phys .
d * 60 * ( 1999 ) 116001 ; z. bern , l. j. dixon and d. a. kosower , jhep * 0408 * ( 2004 ) 012 ; s. d. badger and e. w. n. glover , jhep * 0407 * ( 2004 ) 040 .
n. arkani - hamed , a. hodges and j. trnka , jhep * 1508 * , 030 ( 2015 ) .
z. bern , g. chalmers , l. j. dixon and d. a. kosower , phys . rev .
* 72 * , 2134 ( 1994 ) .
g. ossola , c. g. papadopoulos and r. pittau , nucl .
b * 763 * , 147 ( 2007 ) . | virtual two - loop corrections to scattering amplitudes are a key ingredient to precision physics at collider experiments . we compute the full set of planar master integrals relevant to five - point functions in massless qcd , and use these to derive an analytical expression for the two - loop five - gluon all - plus - helicity amplitude . after subtracting terms that are related to the universal infrared and ultraviolet pole structure ,
we obtain a remarkably simple and compact finite remainder function , consisting only of dilogarithms .
the precise theoretical description of scattering reactions of elementary particles relies on the perturbation theory expansion of the scattering amplitudes describing the process under consideration . in this expansion ,
higher perturbative orders correspond to more and more virtual particle loops . at present
, one - loop corrections can be computed to scattering amplitudes of arbitrary multiplicity , while two - loop corrections are known only for selected two - to - one annihilation or two - to - two scattering processes . for many experimental observables at higher multiplicity , a substantial increase in statistical precision
can be expected from the cern lhc in the near future .
perturbative predictions beyond one loop will be in demand for many precision applications of these data , for example in improved extractions of standard model parameters or in search for indirect signatures of new high - scale physics in precision observables .
progress on multiloop corrections to high - multiplicity amplitudes requires significant advances in two directions .
feynman - diagrammatic approaches to the computation of these amplitudes yield enormously large expressions that contain many thousands of different feynman integrals .
these integrals are related among each other through poincar invariance and symmetries , such that only a limited set of independent so - called master integrals will remain in the final answer for a scattering amplitude .
to express a generic two - loop multiparton amplitude in terms of the relevant master integrals ( ideally circumventing the large algebraic complexity at intermediate stages that is generated by working in terms of feynman diagrams ) is a yet outstanding problem .
a particular example where the reduction to a basis set of integrals was achieved @xcite is the two - loop five - gluon helicity amplitude with all helicities positive . in this case , the application of on - shell techniques led to a particularly compact integrand , which motivated a specific choice of basis integrals ( which do not necessarily form a minimal set in the sense of being master integrals ) . in @xcite
, these integrals were evaluated numerically for selected kinematical points .
although this specific helicity amplitude is not contributing to the second - order correction to the three - jet cross section ( due to its vanishing at tree level ) , it provides an ideal testing laboratory for new calculational concepts and methods that will carry over to the general helicity case , as previously in the case for the four - point two - loop amplitudes @xcite .
the other major challenge in the calculation of multileg multiloop amplitudes lies in the evaluation of the master integrals .
while the full set of feynman integrals at one loop is known for all configurations of internal masses and external kinematics , only specific integrals at low multiplicity ( typically four external legs ; see , however , @xcite ) are known at two loops and beyond . in principle
, these integrals can be evaluated using purely numerical methods , as for example iterated sector decomposition @xcite . in practice , these methods turn out to be too slow to allow for an efficient evaluation of multiloop integrals when sampling the multidimensional phase space , as required for the evaluation of scattering cross sections . on the other hand ,
analytical expressions for the integrals allow to uncover universal structures in the amplitudes and enable the study of limiting kinematical behaviour , thereby advancing our understanding of the high - order structure of perturbative quantum field theory .
the analytical understanding of the most basic amplitudes at a given multiplicity and loop order is moreover an important catalyst enabling further progress towards more complicated processes at the same multiplicity .
important examples are the reconstruction of amplitudes from constraints obtained in specific limits @xcite , the further development and validation of integrand reduction techniques , and methods for finding appropriate integral bases @xcite . in this letter
, we make use of advances in techniques for analytically evaluating loop integrals @xcite , and we compute for the first time the full set of planar master integrals relevant to massless two - loop five - point scattering .
we apply these integrals to the five - gluon all - plus helicity amplitude .
this is the first analytic result of a genuine @xmath0 two - loop amplitude in qcd .
we find that after subtraction of the universal infrared singular terms , our analytic formula is remarkably simple .
it can be written in terms of dilogarithms , with prefactors that are well - behaved in collinear limits .
this simplicity , which is reminiscent of results in @xmath1 super yang - mills ( sym ) , may help to uncover new structural properties of multileg multiloop amplitudes and lead to considerable simplifications in their calculation . |
entropic uncertainty relations for position and momentum , or other canonically conjugated variables have been derived a long time ago .
initial investigations @xcite were devoted to continuous shannon entropies .
further generalizations took into account , in a spirit of deutsch @xcite and maassen - uffink @xcite results , the accuracy of measuring devices @xcite and impurity of a quantum state @xcite .
recent 12th icssur / feynfest conference showed that the topic of entropic uncertainty relations including experimental accuracies is important in the task of entanglement detection in quantum optics @xcite .
this paper is organized as follows . in a further part of the present section
i point out some aspects related to entropic uncertainty relations , following the rephrased heisenberg sentence @xcite _ , , the more information we have about the position , the less information we can acquire about the momentum and vice versa _ and an observation made by peres @xcite _ , , the uncertainty relation such as @xmath1 is not a statement about the accuracy of our measuring instruments_. it the second section i generalize an approach presented in @xcite to the case of finite number of measuring devices ( detectors ) .
i would like to start with the famous bialynicki - birula - mycielski entropic uncertainty relation of the form @xcite @xmath2 where @xmath3 and @xmath4 are probability distributions in position and momentum spaces respectively .
wave functions in both spaces are related to each other by the fourier transform @xmath5 the introduction of @xcite starts as follows : _ , , the purpose of this paper is to derive a new stronger version of the heisenberg uncertainty relation in wave mechanics .
this new uncertainty relation has a simple interpretation in terms of information theory .
it is also closely related to newly discovered logarithmic sobolev inequalities .
_ information theory enters to the inequality ( [ bbm ] ) by the notion of continuous shannon entropies @xmath6 and @xmath7 .
connection with the logarithmic sobolev inequality can be recognized with the help of the reversed logarithmic sobolev inequality @xcite which , for the probability distribution function ( pdf ) @xmath8 defined on @xmath9 , reads @xmath10.\label{logsobol}\ ] ] the variance is defined as usual , @xmath11 .
we shall use the inequality ( [ logsobol ] ) independently for the position and momentum variables and obtain the stronger version of the heisenberg uncertainty relation , announced in @xcite @xmath12 it seems to be widely accepted that the continuous shannon entropy is also a good measure of information since it is a relative of shannon information entropy .
however , this statement is not completely true . to prove that let me recall the definition of the shannon entropy of a set of probabilities @xmath13 @xmath14 we have two important properties of the shannon entropy ( [ shannon ] ) : 1 .
since the probabilities @xmath15 are dimensionless the shannon entropy @xmath16 is also dimensionless .
2 . from the property @xmath17
we know that @xmath18 .
these two properties are essential for the information - like interpretation of the shannon entropy because information can be neither negative nor expressed in any physical units .
unfortunately the continuous shannon entropy does not possess these two properties .
for instance the unit of the entropy @xmath19 ( in si units ) is the logarithm of meter .
this makes impossible to check if the continuous shannon entropy is positive or negative . on the other hand
if we introduce some length ( momentum ) scale in order to measure @xmath19 ( @xmath20 ) then we still can not fulfill the second property since @xmath21 ( @xmath22 ) can be greater than @xmath23 and make the continuous entropy negative .
an idea that helps to overcome the difficulties which i have mentioned above ( related to the continuous entropies ) introduces a partition of the real line into bins of equal width ( see fig .
1 ) . [ fig1 ] as a result the continuous probability distribution @xmath24 is replaced by a discrete distribution @xmath25 , ( @xmath26 ) @xmath27 the bins are of equal width , thus @xmath28 .
this idea was for the first time introduced by partovi @xcite who noticed that @xmath29 shall be interpreted as a _ , , resolution of the measuring device_. the requirement that all bins have to have the same width @xmath29 is a strong constraint on @xmath30 .
in fact @xmath30 must be of the form @xmath31 so there is only one free parameter @xmath32 to be chosen .
one might expect that different values of the parameter @xmath32 are equivalent to different possible choices of the central point ( @xmath33 ) of the used coordinate .
in the first partovi s paper @xcite and later publications @xcite the easiest possible choice @xmath34 was made . in this choice the probability distributions and the shannon entropies are ( @xmath35 and @xmath36 are experimental accuracies for positions and momenta respectively ) : @xmath37 @xmath38 in this moment i would like to point out that the entropies defined in ( [ en1 ] , [ en2 ] ) are not correct measures of information ( because of the choice @xmath34 ) . in order to realize that one shall investigate the limit of large coarse graining ( large experimental accuracies ) @xmath39 or @xmath40 . in these limits performed measurements tell us nothing about the quantum state - our information is @xmath41 , thus , we shall expect @xmath42@xmath43 .
but in fact , we have : @xmath44 as a result @xmath45 ( @xmath46 ) have state - dependent values that vary between @xmath41 if the state is localized in @xmath47 or @xmath48 ( in positions or momenta ) and @xmath49 if the state is symmetric .
in @xcite we captured this ambiguity and performed the redefinition of the probability distributions ( [ en1 ] ) in the following way : @xmath50 it is equivalent to the choice @xmath51 in the construction ( [ con ] ) and means that the center of the coordinate @xmath33 lays in the middle of the central bin . in the previous choice the center of the coordinate was the border point between two bins .
similarly to ( [ bbm ] ) uncertainty relations for @xmath52 were found a long time @xcite .
the stronger one @xcite reads @xmath53 since the shannon entropies ( [ en2 ] ) have been correctly defined ( are positive and dimensionless ) it is obvious that @xmath54 .
thus , for @xmath55 the relation ( [ ibb ] ) becomes trivially satisfied and in fact does not give an optimal lower bound ( the bound is optimal , and saturated by a gaussian distribution , only in the limit @xmath56 and @xmath57 ) .
some attempts @xcite to find a better lower bound failed completely ( for details see @xcite - the comment on @xcite )
. recently @xcite we derived the bound ) .
red / dashed curve represents @xmath58 while green curve represents @xmath59 ; @xmath60.,title="fig:"][fig2 ] @xmath61,\label{bound new}\ ] ] which is always positive , but still not optimal ( cf .
function @xmath62}$ ] . ] is one of the radial prolate spheroidal wave functions of the first kind @xcite .
in the definition ( [ en2 ] ) the summation over @xmath63 ( @xmath64 ) goes from @xmath65 to @xmath66 . that property in a language of an experiment means that there are infinitely many measuring devices ( detectors ) covering an infinite number of bins
of course in realistic experiments only a finite number of bins can be covered , thus , we shall describe the measurement s result with the help of the finite shannon entropies , defined as follows : @xmath67 these definitions include @xmath68 detectors set to measure the probability distribution in positions and @xmath69 detectors for momenta .
we have the limits @xmath70 and @xmath71 . since all terms in the sums in ( [ en2 ] ) are non - negative we know that @xmath72 and @xmath73 .
thus , we can not expect that the uncertainty relations ( [ ibb ] ) or ( [ bound new ] ) will be satisfied for @xmath74
. moreover we can always find many states localized far away from the detectors ( with @xmath75 and @xmath76 ) , for example @xmath77,\;\longrightarrow\quad\rho\left(x\right)=\frac{1}{\sqrt{\pi}\sigma}\exp\left[-\frac{\left(x - x_{0}\right)^{2}}{\sigma^{2}}\right],\ ] ] with @xmath78 and @xmath79 .
this inconvenience is directly related to the translational invariance which was broken when we restricted the number of used bins .
of course an experimentalist knows where the measured state is and can properly choose the coordinate to assure that @xmath80 .
looking at fig .
2 we can realize that if @xmath81 then the previous lower bound ( [ ibb ] ) dominates .
i would like to work in this regime and , although the finite number of bins makes us unable to find a state independent lower bound for @xmath74 , i would like to find a state - dependent correction to the formula ( [ ibb ] ) .
i shall start in the same way as was done in @xcite and use the integral jensen inequality ( i will do the calculations only for the position variable since in the momentum case everything shall go the same way ) @xmath82\\ = & b_{x}+q_{\infty}\ln\left(\delta x\right)+\left(\int_{-\infty}^{-\left(m+1/2\right)\delta x}+\int_{\left(m+1/2\right)\delta x}^{\infty}\right)dx\rho\left(x\right)\ln\rho\left(x\right),\nonumber \end{aligned}\ ] ] where i have introduced the following notation : @xmath83 the shannon entropy , because of the logarithmic function , collects information from all moments of the probability distribution .
a convenient method to reduce the state - dependent input can be based on the reversed logarithmic sobolev inequality ( [ logsobol ] ) .
we shall choose @xmath84 and @xmath85-\infty,-\left(m+1/2\right)\delta x\right]\cup\left[\left(m+1/2\right)\delta x,\infty\right[$ ] and obtain @xmath86 since we expect that the state fulfills @xmath87 we can , without any significant loss , simplify the right hand side of ( [ 19 ] ) @xmath88 after this step we obtain the inequality @xmath89,\quad\lambda\geq0,\ ] ] where the @xmath90 function depends only on the , , @xmath41th moment ( the norm @xmath91 ) and the , , @xmath92nd moment ( @xmath93 ) of the function @xmath21 restricted to @xmath94 .
since the moments of the function are independent we can find @xmath95 keeping @xmath93 constant .
this step will reduce the state dependent input only to the one quantity @xmath93 . in order to find the minimum we shall calculate the following derivatives @xmath96 since for @xmath97 $ ] the second
derivative is positive we have only one global minimum @xmath98 . on the other hand when @xmath99 the minimum lays outside the domain of @xmath100 .
thus , since @xmath90 is a decreasing function on the interval @xmath101 $ ] we shall modify @xmath102 and write @xmath103 using this result we obtain that : @xmath104 if we assume that the numbers @xmath105 ( @xmath106 ) are sufficiently large , so that @xmath107 and @xmath108 we find the following uncertainty relation @xmath109 this bound gives a nontrivial limitation when @xmath110 , what it this case is equivalent to @xmath111 . in a general case
the uncertainty relation for the sum of the entropies is @xmath112 , where : @xmath113
i have briefly presented a current status in the topic of entropic uncertainty relations for position and momentum variables .
i have pointed out ambiguities appearing in the scientific literature .
finally i have investigated the case of finite number of detectors in an approach using discrete shannon entropies including experimental accuracies and derived nontrivial , state - dependent lower bound generalizing previous results .
i am specially indebted to iwo bialynicki - birula who inspired and supported all my efforts in the topic of entropic uncertainty relations .
i would like to thank stephen p. walborn , fabricio toscano , luiz davidovich and all other members of faculty of physics in federal university of rio de janeiro ( instituto de fsica in universidade federal do rio de janeiro ) for their hospitality after the 12th icssur / feynfest conference .
this research was partly supported by a grant from the polish ministry of science and higher education for the years 20102012 .
references i. bialynicki - birula and j. mycielski , _ commun .
phys . _ * 44 * , 129 ( 1975 ) .
d. deutsch , _ phys .
lett . _ * 50 * , 631 ( 1983 ) .
h. maassen and j.b.m .
uffink , _ phys .
lett_.*60 * , 1103 ( 1988 ) .
m. h. partovi , _ phys .
lett . _ * 50 * , 1883 ( 1983 )
. i. bialynicki - birula , _ phys . lett . _ * 103 a * , 253 ( 1984 ) .
i. bialynicki - birula , _ phys .
a _ * 74 * , 052101 ( 2006 ) . m. d. srinivas , _ pramana _ * 25 * , 369 ( 1985 ) .
i. bialynicki - birula and .
rudnicki , entropic uncertainty relations in quantum physics , arxiv:1001.4668 ( 2010 ) . v. v. dodonov , _ j. opt .
b : quantum semiclass . opt . _ * 4 * , 98 ( 2002 ) .
b. coutinho dos santos , k. dechoum , and a. z. khoury , _ phys .
lett . _ * 103 * , 230503 ( 2009 ) .
a. saboia , f. toscano , and s. p. walborn , _ phys .
a _ * 83 * , 032307 ( 2011 ) .
a. peres , quantum theory : concepts and methods .
kluwer , dordrecht ( 1995 ) .
d. chafa , _ gaussian maximum of entropy and reversed log - sobolev inequality _ , sminaire de probabilitis , strasbourg * 36 * , 194 - 200 ( 2002 ) . g. wilk and z. wodarczyk , _ phys .
a _ * 79 * , 062108 ( 2009 ) . i. bialynicki - birula and .
rudnicki , _ phys .
a _ * 81 * , 026101 ( 2010 ) .
rudnicki , uncertainty related to position and momentum localization of a quantum state , arxiv:1010.3269v1 ( 2010 ) .
rudnicki , s. p.
walborn and f. toscano , in preparation .
m. abramowitz and i.a .
stegun , _ handbook of mathematical functions_. dover , new york , ( 1964 ) . | this paper is prepared as a contribution to the proceedings after the 12th icssur / feynfest conference held in foz do iguau ( brazil ) from 2 to 6 may 2011 . in the first part
i briefly report the topic of entropic uncertainty relations for position and momentum variables .
then i investigate the discrete shannon entropies related to the case of finite number of detectors set to measure probability distributions in position and momentum spaces .
i derive an uncertainty relation for the sum of the shannon entropies which generalizes previous approaches [ _ phys .
lett . _ * 103 a * , 253 ( 1984 ) ] based on an infinite number of detectors ( bins ) . [ sh ] * * [ cols="^ " , ] * * * ukasz rudnicki@xmath0 * _ center for theoretical physics , polish academy of sciences _
+ _ aleja lotnikw 32/46 , pl-02 - 668 warsaw , poland _
@xmath0corresponding author e - mail : [email protected] + * keywords : * shannon entropy , entropic uncertainty relations , uncertainty of quantum measurements performed with finite accuracy and finite number of detectors |
since wigner s initial suggestion @xcite that the statistical properties of the eigenvalues of random hermitian matrices should provide a good description of the excited states of complex nuclei , random matrix theory ( rmt ) has become one of the prominent fields of research , at the boundary between atomic physics , solid state physics , statistical mechanics , statistics , probability theory , and number theory @xcite .
the two main models for hermitian random matrices ( which have been extensively studied in the literature , see @xcite for a review of rmt ) are : ( i ) the ensembles of wigner matrices which have independent ( up to symmetry ) and identically distributed entries ; ( ii ) the classical real , complex and quaternion _ invariant _ ensembles , which are respectively stable under the conjugation by the orthogonal , unitary , and symplectic groups .
the intersection between those two types of random matrices is actually reduced to the famous gaussian orthogonal , unitary and symplectic ensembles . in this paper
, we construct hermitian matrix diffusion processes @xmath11 , which evolve in _ non - confining _ potentials and which are invariant under rotation at all time @xmath12 .
the initial motivation for our work is to make sense of invariant ensembles of random matrices in non - confining potentials such as @xmath13 , which can not be defined in the usual way because of the divergence of the partition function .
the family @xmath14 of potentials we are looking at includes the cubic interaction ( @xmath15 ) and our results can be extended to a variety of non - confining polynomial potentials with the same ideas .
in particular , our method also permits us to construct a family of invariant ensembles in non - confining quartic potentials of the form @xmath16 where @xmath17 is a parameter called the coupling constant .
such potentials @xmath18 have already been considered in the paper @xcite , where brzin , itzykson , parisi and zuber use an enumerative formula established in @xcite for planar diagrams in terms of matrix integrals associated to the potential @xmath18 , to count planar diagrams .
our construction brings new lights on some of their results on the limiting spectral density as the dimension @xmath19 , in the case @xmath17 .
the idea of our construction is inspired by @xcite where halperin studies one dimensional diffusion processes in non - confining potentials .
we first simply let our diffusive matrix process evolve in the non - confining potential of interest , until the first explosion time of one of the coefficients of the matrix . to extend the trajectory after this explosion time
, we restart our matrix process at this time from a new position insuring continuity of the non - exploding coefficients , until the next explosion occurs , and so on . this procedure is explained in section [ cubic - hermitian - process ] . after some time
, we expect that this matrix process will reach an equilibrium , which is , in contrast with the classical case of dyson brownian motion in a quadratic potential , a _ dynamical steady state_. in particular , we establish in section [ stationary - flux ] that there is a stationary flux of particles ( eigenvalues ) flowing across @xmath8 in the steady state if the barrier of the potential is not too strong to confine the particles .
this feature appears to be new and leads to interesting further questions about the fluctuations of this flux around its leading order in the large @xmath20-limit .
we compute the leading behavior of the stationary flux in the large @xmath20 limit ( see formula [ leading - flux ] ) , which measures the number of eigenvalues crossing over the system per unit of time .
we believe that our construction defines an interesting model of interacting charged particles with a random flux , to be related with the asymmetric exclusion processes which have attracted attention in the last decade ( see e.g. @xcite ) . in the particular case of a cubic potential , we describe precisely the dynamics of the spectrum of our matrix process in the limit of large dimension @xmath20 in section [ dynamic - density ] .
we analyze the stationary state of this dynamic by computing explicitly the spectral density , which gives the global allocation of the eigenvalues .
we observe an interesting sharp phase transition for this spectral density at the critical value @xmath2 .
if @xmath6 , then the well of the potential @xmath4 is not deep enough to prevent a macroscopic proportion of the eigenvalues to explode and we find a stationary spectral density @xmath7 with full support in @xmath8 , flanked with heavy tails such that @xmath9 as @xmath10 . on the contrary , if @xmath21 , then the eigenvalues are confined in the well near @xmath22 of the potential @xmath4 such that the spectral density in the stationary state has compact support .
the underlying flux of particles displays also a phase transition at the critical value @xmath2 .
the current of particles , which allocate according to the stationary profile @xmath1 , is macroscopic compared to the total mass of the @xmath20 eigenvalues if @xmath23 whereas it is microscopic if @xmath3 .
we conclude in the last section [ conclusion ] with some open questions related to the statistics of the eigenvalues and their current , in the stationary state
. * acknowledgments .
* we are grateful to jol bun and antoine dahlqvist for interesting discussions on stieltjes transforms .
r.a . received funding from the european research council under the european union s seventh framework programme ( fp7/2007 - 2013 ) / erc grant agreement nr .
258237 and thanks the statslab in dpmms , cambridge for its hospitality .
the work of l.d . was supported by the engineering and physical sciences research council under grant ep/103372x/1 and l.d .
thanks the hospitality of the maths department of tu and the weierstrass institute in berlin .
analytic _ potential @xmath24 such that @xmath25 , we associate an invariant ensemble of random matrices in the space of ( real , complex or quaternion ) hermitian matrices by specifying their law as @xmath26 where @xmath24 is meant to act on hermitian matrices by holomorphic functional calculus , @xmath27 is the lebesgue measure on the space of hermitian matrices and @xmath28 is the partition function given by @xmath29 where the integral is over the corresponding space of hermitian matrices .
an assumption on the growth of @xmath30 when @xmath10 is necessary here @xcite , such that @xmath31 so that the partition function @xmath28 and the probability law @xmath32 are well defined . under this assumption ,
the eigenvalues of the matrix @xmath33 , which evolves in the stationary potential @xmath24 , are confined and do not explode .
the purpose of this article is to construct new invariant ensembles of random matrices in a family of potentials @xmath34 , which violates the key assumption . for simplicity , we reduce to the case of real symmetric matrices , although this discussion can be extended to the complex and quaternion cases .
the first step is to view the probability distribution @xmath32 defined in as the _ boltzmann equilibrium weight _ of the diffusive matrix process satisfying the langevin equation @xmath35 with @xmath36 at the initial time and where @xmath37 is a _ hermitian brownian motion _ , i.e. a real symmetric @xmath38 matrix process with entries given by independent ( up to symmetry ) real brownian motions ( bms ) , with an extra factor @xmath39 for the off diagonal entries of @xmath37 .
the dynamic preserves the invariance under rotation of the process @xmath40 , in the sense that the matrix @xmath33 is invariant in law under the conjugation of any orthonormal matrix : for all @xmath41 ( the orthonormal group ) and for any @xmath12 , we have @xmath42 where @xmath43 refers to the transpose of @xmath44 .
the interesting feature of this second approach is that one can still define a hermitian matrix process satisfying the langevin equation even if the potential @xmath24 does _ not _ fulfill the _ confining _ restriction .
in this paragraph , we consider the _ odd _ potential @xmath45 where @xmath46 is a given parameter .
this family of potential @xmath47 includes the cubic interaction ( case @xmath15 ) and we shall soon see that it is also natural to introduce a linear term in order to cover a wider variety of behaviours . besides , this family of potentials already appeared in various contexts : random matrices @xcite , random schrdinger operators and diffusions @xcite . if @xmath48 , the potential @xmath4 presents a local minimum in @xmath5 and a local maximum in @xmath49 ( see fig .
[ fig.potential ] ) .
the local minimum is then separated from the local maximum by a potential barrier of size @xmath50 . if @xmath51 however , the potential is fully non - confining .
for such a potential @xmath4 , the probability @xmath32 defined in does not exist because of the divergence of @xmath52 as @xmath53 which prevents the partition function @xmath28 to be finite . nevertheless , the langevin equation remains well defined up to a small time interval @xmath54 $ ] ( with overwhelming probability ) and there exists a diffusive matrix process @xmath55)$ ] in the space of real hermitian matrices such that , @xmath56 with @xmath36 at the initial time and where @xmath57 is a hermitian brownian motion ( defined above ) .
similar diffusion processes have also been thoroughly considered in dimension @xmath58 in @xcite .
we invite the reader to look at @xcite for a brief review on the dynamics of a one - dimensional diffusion in such a potential . as in the one - dimensional case ,
the main difficulty to define such a hermitian process on the whole positive half line @xmath12 , comes from the fact that , with probability one , the diffusive matrix process @xmath59 satisfying will eventually blow - up at some finite time @xmath60 defined as @xmath61 particles ( red dots ) carrying positive charges in the potential @xmath62 as a function of @xmath63 for @xmath64 . ] in @xcite , bloemendal and virg construct an hermitian diffusive process @xmath11 on the whole positive half line @xmath12 , satisfying a similar equation to off what they call the _ focal _ points ( see ( * ? ? ? * eq .
( 5.8 ) ) ) , which corresponds to the explosion times of the process @xmath59 .
they are interested in the case of a _ non stationary _ potentials similar to ours with an additional linear term @xmath65 in the drift ( @xmath66 ) .
our case would simply corresponds to the @xmath67 case .
the authors of @xcite use a matrix generalization of sturm oscillation theory , which goes back to the work of morse @xcite ( see also @xcite ) .
let us now explain how the eigenvalues and eigenvectors of @xmath33 evolve until the first explosion time @xmath60 and see how the trajectories of those processes can be extended after this time . for any @xmath12
, the real eigenvalues of the hermitian matrix @xmath33 will simply be denoted , in non increasing order , as @xmath68 .
the main point is that the symmetric matrix @xmath69 in is invariant under conjugation by an orthogonal matrix so that the usual derivation of _ dyson s brownian motion _
@xcite is easily extended to this case .
indeed , the authors of @xcite derive the stochastic differential system satisfied by the eigenvalues process using hadamart s variation formula , see ( * ? ? ? * theorem 5.4 ) , which is somehow the rigorous way of performing basic perturbation theory in the eigenvalues problem associated to the langevin equation .
either way , we eventually obtain the following stochastic differential system for the eigenvalues @xmath70 , @xmath71 where @xmath72 and @xmath73 are real independent brownian motions .
note that the cases @xmath74 may have also been covered using complex and quaternion hermitian brownian motions .
let us simply mention that for @xmath75 , the electrostatic repulsion is strong enough to prevent any collision between the eigenvalues so that the stochastic differential system has a well defined and continuous solution in the it s sense @xcite . towards a physical picture , we can see the process @xmath76 as a one dimensional repulsive coulomb gas of @xmath20 positively charged particles , subject to a thermal noise and lying in the _ non - confining cubic _ potential @xmath4 ( see fig . [ fig.potential ] ) . as a function of time @xmath77 for @xmath78 and @xmath79 .
the green horizontal dashed lines give the position of the well and the hill of the potential @xmath4 in @xmath80 . ]
the evolution of the ( orthonormal ) eigenvectors @xmath81 respectively associated to @xmath82 can also be derived using standard perturbation theory , or by the indirect method of ( * ? ? ?
* proof of theorem 4.3.2 ) , applied for the hermitian brownian motion , together with the eigenvalues .
note that the @xmath81 are all determined up to a sign @xmath83 .
up to an arbitrary choice at the initial time , we can prove , following ( * ? ? ?
* proof of theorem 4.3.2 ) ( see also @xcite ) , that there exists a continuous ( with respect to time ) version of the process @xmath84 which evolves according to @xmath85 where the real brownian motions @xmath86 are mutually independent and defined by symmetry @xmath87 for @xmath88
. moreover the @xmath89 are independent of the brownian motions @xmath90 driving the stochastic differential system of the eigenvalues .
this allows us to freeze the trajectories of the eigenvalues until the first explosion time @xmath60 and then to study the eigenvectors dynamics with this realization of the eigenvalues path .
now the main problem is to understand the behavior of the eigenvectors when we approach the explosion time @xmath60 at which @xmath91 as @xmath92 .
we can easily see that this singularity does not affect the eigenvectors in the sense that , for all @xmath93 , the trajectory of @xmath81 can be extended continuously when @xmath94 . indeed ,
all the terms of the form @xmath95 which appear in vanish at the explosion time @xmath60 , so that we can check the cauchy criterion @xmath96 when @xmath97 , which insures the existence of a limit for @xmath81 when @xmath94 for all @xmath98
. this remark suggests to extend the trajectory of the matrix process @xmath59 after each explosion times , which are labeled as @xmath99 , according to the following procedure . whenever an explosion occurs at some time @xmath100
, the ( exploding ) eigenvalue @xmath101 is immediately restarted at the explosion time @xmath102 in @xmath103 , while the trajectories of the other particles @xmath104 are extended in a continuous way ( note again from that for all @xmath105 , @xmath82 has a limit when @xmath94 ) . at each explosion @xmath100 , we re - label the eigenvalues according to the circular change of indexation @xmath106 now , in order to define the trajectory of the hermitian process @xmath59 for all time @xmath12 , we need to check that the sequence of explosion times @xmath107 , defined recursively for @xmath108 as @xmath109 has no accumulation points in @xmath110 .
this fact follows from ( * ? ? ?
* section 5 ) where the authors prove that the explosion times of the eigenvalues process correspond to the focal points , which are almost surely finitely many in compact sets of @xmath110 ( see in particular ( * ? ? ?
* proposition 5.1 ) ) . in this way , the trajectory of the hermitian process @xmath59 is defined for all time @xmath12 .
its eigenvalues process evolves according to the stochastic differential system with a circular re labeling at each explosion time @xmath100 , while the associated eigenvectors process follows with the same re labeling at time @xmath100 . because @xmath33 is invariant under rotation at all times , there is not much to say on the eigenvectors dynamics of the process @xmath40 .
in the next sections of this paper , we focus on the spectral statistics of @xmath40 .
[ rem - norestart1 ] one could have chosen a different dynamic where the process @xmath76 is still a solution of the stochastic differential system but with no restarting procedure after the explosions . instead the particles are killed in @xmath111 whenever they explode at some finite time ( in the sense that they are stuck forever in @xmath111 and do not interact anymore with the living particles , which have not yet exploded ) .
this interesting model seems more complicated to handle with our methods , see remark [ rem - norestart ] for more details .
we are mainly interested in the empirical measure of the eigenvalues of the hermitian process @xmath40 at time @xmath77 @xmath112 recall that the eigenvalues process @xmath113 satisfies the stochastic differential system on the intervals @xmath114 with the restarting and re - indexing procedures at each explosion time @xmath100 .
we denote by @xmath115 the space of probability measures on @xmath8 and by @xmath116 , \mathcal{p}({\mathbb{r}}))$ ] the space of continuous functions from @xmath117\to \mathcal{p}({\mathbb{r}})$ ] .
let us briefly recall the definition of the stieltjes transform of a measure .
let @xmath118 the upper half complex plane . if @xmath119 is a measure on @xmath8 , its stieltjes transform i.e. @xmath120 . ]
is the holomorphic function @xmath121 defined by @xmath122 we can recover the probability measure @xmath123 from the stieltjes inversion formula , which writes for @xmath124 , @xmath125 + \frac{\pi}{2}\big(\mu_t(\{y\ } ) - \mu_t(\{x\})\big)\ , .
\end{aligned}\ ] ] basic properties of stieltjes transforms , which shall be useful throughout the paper , are recalled in appendix [ stieltjes ] .
our main result in this section establishes the convergence of the continuous stochastic process @xmath126 when the dimension @xmath20 tends to @xmath127 .
[ main ] let @xmath128 and @xmath129 .
suppose that , at the initial time , the empirical density @xmath130 converges weakly as @xmath20 goes to infinity towards some @xmath131 .
then , @xmath132 converges almost surely in @xmath133,\mathcal{p}({\mathbb{r}}))$],\mathcal{p}({\mathbb{r}}))$ ] is a polish space as @xmath115 equipped with its weak topology is metrizable ( @xmath8 is a separable space ) ] .
its limit @xmath134 is the unique measure valued process such that @xmath135 and whose stieltjes transform @xmath136 satisfies the holomorphic equation @xmath137 \ , ds \ , .
\end{aligned}\ ] ] the proof of theorem [ main ] is deferred to section [ proof ] .
our approach is classical and follows the method introduced in @xcite ( see also @xcite ) .
it consists in writing an evolution equation for the stieltjes transform @xmath138 of the probability measure @xmath139 thanks to it s formula and the stochastic differential system satisfied by the @xmath140 in section [ evolution.gn ] .
we first prove the almost - sure pre - compactness of the family @xmath141 in the space @xmath116 , \mathcal{m}_{\le 1}({\mathbb{r}}))$ ] where @xmath142 is the space of measure with total mass smaller or equal to @xmath143 equipped with its weak-@xmath144 topology .
it turns out that it is sufficient for our purposes to prove pre - compactness in @xmath116 , \mathcal{m}_{\le 1}({\mathbb{r}}))$ ] ( instead of @xmath116 , \mathcal{p}({\mathbb{r}}))$ ] ) as we can show afterwards that the stieltjes transform solutions of are associated to probability measures .
this feature is original and contrasts with the classical case of the quadratic potential for which the solutions of the stieljes equation are not necessarily probability measures ( see @xcite where the authors prove a dynamical version of wigner s theorem ) .
finally we prove uniqueness of the solution in lemma [ unicity - lemma ] using the characteristic method .
we are interested in this paragraph in the convergence of the probability measure process @xmath123 when @xmath145 ( in the space of radon measures endowed with the topology of weak convergence ) .
to prove that @xmath123 converges weakly as @xmath145 to some @xmath146 , it is sufficient to show ( see @xcite ) that the stieltjes transform @xmath147 associated to @xmath123 converges point - wise for @xmath148 to the stieltjes transform @xmath149 of the _ probability measure _ @xmath150 .
although we were not able to prove it , it is natural to expect that the following convergence holds : [ conjecture1 ] let @xmath151 be the solution of the evolution equation .
then , the following limit exists for all @xmath148 , @xmath152 assuming that conjecture [ conjecture1 ] is true , we can prove that the limit is indeed the stieltjes transform of some probability measure and deduce the desired convergence : [ prop - conv - time ] the function @xmath153 is the stieltjes transform of a probability measure @xmath154 and is a stationary solution of the evolution equation .
consequently , the probability measure @xmath123 defined in theorem [ main ] converges weakly as @xmath155 to the probability measure @xmath156 .
the proof of this proposition is deferred to subsection [ proof - prop - conv - time ] . in the next section ,
we characterize @xmath157 as the unique function satisfying the following two properties : * @xmath153 is a stationary solution of ; * @xmath153 is the stieltjes transform of a ( probability ) measure @xmath154 .
moreover , we derive @xmath153 and @xmath154 explicitly in analytic forms ( see theorem [ theorem - statio ] and its proof below ) . the probability measure @xmath154 is the limiting empirical eigenvalue density of the matrix @xmath33 in the stationary state .
we will see that @xmath153 provides additional information on the current of particles in the system , when the well in @xmath158 of the potential @xmath4 is not too confining to retain all the particles inside ( see section [ stationary - flux ] for further details ) .
in this section , we compute the limiting ( i.e. when @xmath159 ) empirical eigenvalue density of the matrix @xmath33 in the stationary state ( i.e. after a long time @xmath145 ) . under the assumption that conjecture [ conjecture1 ] is true , we know that there exists a stieltjes transform @xmath153 associated to @xmath160 which is a stationary solution of the evolution equation . in this section ,
we prove the uniqueness of the analytic function @xmath153 which enjoys those properties and we compute it explicitly in terms of the roots of a cubic polynomial . using the stieltjes inversion formula recalled in ,
we derive the probability density of @xmath154 with respect to lebesgue measure and exhibit an interesting ( sharp ) phase transition displayed by @xmath154 at the critical value @xmath161 .
many details on the stationary probability measure @xmath154 are provided in this section .
we briefly summarize the main features in the following theorem .
[ theorem - statio ] for all @xmath129 , there exists a unique analytic function @xmath153 with the following two properties : * it is the stieltjes transform of a probability measure @xmath154 , or equivalently ( akhiezer s theorem @xcite ) @xmath153 is analytic on @xmath162 with @xmath163 and @xmath164 as @xmath165 ; * it is a stationary solution of the evolution equation .
moreover , the probability measure @xmath166 admits a density @xmath167 with respect to the lebesgue measure , computed explicitly ( see and ) in terms of the roots of the polynomial of degree three @xmath168 .
we distinguish two regimes depending on whether @xmath3 , or @xmath169 where @xmath170 is the critical value at which the probability density @xmath1 displays a sharp phase transition ( illustrated in fig .
[ fig.potential ] ) : * if @xmath3 , @xmath1 is supported on a compact interval .
* if @xmath23 , @xmath1 has full support in @xmath8 and is flanked with symmetric heavy tails as @xmath10 , @xmath171 where @xmath172 is an explicit constant ( see below ) . in the supercritical regime @xmath3
, the particles are all confined in the well of the local minimum of the potential @xmath4 .
the well is deep enough compared to the electrostatic repulsion between the eigenvalues , to keep the particles inside it .
the limiting density of particles has a classical shape in random matrix theory with a compact support and singularities of order @xmath173 at the upper and lower edges of the spectrum ( square root cancelations ) .
the critical density @xmath174 , still compactly supported , is of particular interest with an usual singularity at the lower edge ( see paragraph [ critical - regime ] where we compute it explicitly as function of @xmath175 ( see ) .
a sharp transition is observed at the critical value @xmath2 : the density @xmath1 is compactly supported for @xmath3 but has full support with heavy tails if @xmath23 . as we will see in the next section
, the equilibrium of the eigenvalues becomes very unusual in the subcritical regime , mainly due to the non - confining shape of the potential @xmath4 .
this observation appears to be new .
although the density profile of the eigenvalues has a stationary shape @xmath1 , the particles are still flowing across the system in the equilibrium state .
there is in fact a positive current of eigenvalues flowing from @xmath103 to @xmath111 in a stationary way : the number of particles per unit of time shifting from the right to the left ( counted algebraically ) at some given level @xmath176 is constant ( in time and in space ) .
this stationary current is also computed explicitly in the next section .
we provide an illustration of the variety of possible behaviors for the density @xmath1 in fig .
[ fig.eigenvalues-densities ] where we show the graphics of the limiting eigenvalues density @xmath177 as a function of @xmath63 , for particular values of @xmath178 in the three different regimes @xmath179 and @xmath180 .
we have also checked our result with numerical simulations with excellent agreement ( see fig .
[ fig.eigenvalues-densities ] ) .
the samples to construct the empirical densities were obtained by simulating the hermitian matrix process @xmath181 satisfying , with @xmath182 .
the method is usual and consists in discretizing time and diagonalizing the matrix @xmath33 at each time step .
we have introduced a cut off in order to deal with the explosions . whenever the lowest eigenvalue @xmath183 of the matrix @xmath33 gets smaller than the cut - off value , we re initialize the matrix @xmath33 according to the procedure described in section [ cubic - hermitian - process ] , using again a cut - off to approximate the value @xmath103 .
we let our algorithm run for a time @xmath184 with a time step @xmath185 and constructed the empirical densities in the respective cases using the eigenvalue samples at all time steps after time @xmath186 . we noted that both the convergences in dimension and time are extremely fast . as a function of @xmath63 , in the real case ( @xmath72 ) for @xmath15 , @xmath187 and @xmath188 corresponding respectively to the sub - critical , critical and super critical regimes .
the points on top of the straight line represent the empirical densities obtained from our simulated samples with @xmath182 . ]
the rest of this section is devoted to the proof of theorem [ theorem - statio ] .
additional informations on the probability measure @xmath154 are provided along this proof .
_ proof of theorem [ theorem - statio]_. from conjecture [ conjecture1 ] and proposition [ prop - conv - time ] , we know that there exists an analytic function @xmath153 which satisfies the two properties given in theorem [ theorem - statio ] .
it is therefore sufficient to prove the uniqueness of @xmath153 with those two properties . in the following , we work with the complex square root function @xmath189 defined for @xmath190 as @xmath191 . with this definition ,
the square root is analytic on the domain @xmath192 ( to @xmath162 ) but displays a discontinuity near the positive half line @xmath110 .
if @xmath153 is a stationary solution of , then there exists a constant @xmath193 such that for all @xmath148 , @xmath194 for any @xmath148 , we can solve the second degree equation for which we have two solutions , @xmath195 our analytic function @xmath196 is equal either to @xmath197 or to @xmath198 depending on the value of @xmath148 .
it is possible that both @xmath197 and @xmath199 belong to @xmath162 for certain values of @xmath200 , but for any @xmath148 , we will see that there will be one unique possible value of @xmath201 .
we now seek for the constants @xmath202 such that @xmath153 has the two required properties of theorem [ theorem - statio ] .
we shall in fact prove that there exists a unique such constant @xmath203 .
the main idea is that the analyticity of @xmath196 and the _ non- analyticity _ of the square root function in @xmath204 prevent the complex polynomial function @xmath205 inside the square root of to have any root with _ odd _
multiplicity , one or three , in @xmath162 .
additional information on the spatial locations of the roots of the polynomial @xmath32 is then provided by the zeroes of its derivative with degree three , @xmath206 mainly , it is well known ( _ gauss - lucas _ theorem ) that the roots of @xmath207 all lie within the _ convex hull _ of the roots of @xmath32 , that is the smallest convex polygon containing the roots of @xmath32 .
it is easy to check from _ cardan _ s formulas that @xmath207 has three real roots if and only if @xmath208 if @xmath23 , then @xmath207 has one real root and two roots in @xmath209 which are complex conjugate .
moreover cardan s formulas permit us to compute the three roots of @xmath207 analytically .
gathering those arguments , we can now compute the constant @xmath210 treating separately the two cases @xmath211 and @xmath212 . if @xmath211 , then @xmath207 has one ( unique ) root in @xmath162 so that , by gauss - lucas theorem , @xmath32 has at least one root in @xmath162 . by analyticity of @xmath213 on @xmath162 ,
any such root of @xmath32 is necessarily of multiplicity two , would have a root of multiplicity three . ] and therefore is equal to the unique root of @xmath207 in @xmath162 . denoting by @xmath214 this root and using the condition @xmath215 , we obtain @xmath216 and the polynomial @xmath217 is uniquely determined . note that we have the following expression for @xmath214 : @xmath218 where @xmath219 which permits us to check that @xmath220 thanks to elementary computations ( using the second expression for @xmath210 ) .
therefore @xmath153 can be extended analytically to a neighborhood of the real axis ( indeed @xmath221 implies that @xmath222 can not have real roots ) .
moreover @xmath153 takes the following expression for @xmath148 near the real axis : @xmath223 the polynomial @xmath222 can be factorized with elementary computations as @xmath224 where @xmath225 eq . characterizes the function @xmath153 uniquely .
the stieltjes inversion formula permits us to check that the probability measure @xmath154 admits a density @xmath1 with respect to the lebesgue measure given by for @xmath176 , @xmath226\ , .
\end{aligned}\ ] ] it is easy to check that @xmath1 has full support in @xmath8 .
recalling that @xmath227 , it is straightforward to derive the _
heavy _ tails of @xmath1 when @xmath10 , @xmath228 the function @xmath1 is integrable on @xmath8 as expected for a probability measure .
it would be interesting to check the normalization condition @xmath229 by direct integration . in order to prove the existence of the stieltjes
transform @xmath153 satisfying the two required properties of theorem [ theorem - statio ] _ without _ assuming that conjecture [ conjecture1 ] holds , one would need to prove that the stieltjes transform of the _ explicit _ probability density @xmath1 given in is indeed the analytic function @xmath153 characterized in .
we were not able to perform this integration , although the two formulas and look very similar .
in the super - critical regime ( see below ) , we can do this integration . if @xmath21 , then the derivative polynomial @xmath207 has three real roots , and the analyticity conditions on @xmath153 and the _ gauss - lucas _ theorem permit to show that all the roots of @xmath32 are real valued .
indeed , we have already seen that the polynomial @xmath32 can not have any root in @xmath162 ( otherwise , this root would be of multiplicity two and @xmath207 would have a root in @xmath162 , which leads to a contradiction ) .
the remaining scenario where @xmath32 has two distinct zeroes in @xmath8 with multiplicity one and two other zeroes ( counting multiplicity ) in @xmath230 is also excluded : the zeroes of @xmath207 would then lie on the frontier of the convex hull of the zeroes of @xmath32 and this is possible only if the two real zeroes of @xmath32 have multiplicity two , leading again to a contradiction .
we conclude that all the zeroes of the polynomial @xmath32 are real , so that @xmath32 has real coefficients .
in particular , @xmath231 .
the polynomial @xmath32 is now determined up to the real constant @xmath232 , which has the effect of translating vertically ( along the @xmath233-axis ) the graph @xmath234 .
the uniqueness of @xmath235 leading to the correct solution @xmath153 will come from the normalization condition @xmath236 . by the stieltjes inversion formula ,
we know that the measure @xmath154 is supported on the compact set @xmath237 is a union of intervals .
for instance , if @xmath32 has four distinct eigenvalues @xmath238 , then @xmath150 has a disconnected support of the form @xmath239 \cup [ x_3;x_4]$ ] . ] and has a density @xmath1 with respect to lebesgue , defined for @xmath240 as , @xmath241 the area @xmath242 under the graph of @xmath1 can thus be seen as a function of @xmath202 , @xmath243 .
it is obviously a continuous and strictly decreasing function of @xmath202 , and we have @xmath244 and @xmath245 ( for @xmath202 large enough , @xmath246 for all @xmath176 ) . by the intermediate value theorem , there exists a unique @xmath203 such that @xmath247 .
the uniqueness of @xmath203 is proved .
now we would like to determine the value of @xmath210 in the present case , @xmath21 .
to guess its value , let us notice that if @xmath32 has four distinct real roots in @xmath8 , then the measure @xmath154 has a disconnected compact support , union of two disjoint intervals , and this solution is not physically sound in view of the shape of the potential @xmath4 .
recalling the confining shape of the potential @xmath62 near the region @xmath248 when @xmath48 , we would rather expect the polynomial @xmath32 to have its smallest root of multiplicity two and then two other roots of multiplicity one near the confining zone of the potential .
the minimal root of @xmath32 , which will be denoted ( again ) by @xmath214 , would then also be the minimal root of @xmath207 .
finally , we can compute the real constant @xmath203 associated to this scenario using the condition @xmath215 , and we re obtain , with now @xmath249 , formulas for @xmath210 , for the polynomial @xmath222 and for @xmath201 when @xmath200 is near the real line . from the stieltjes inversion formula , we see that @xmath154 has a density @xmath1 with compact support @xmath250 $ ] , given by @xmath251 for @xmath252 $ ] .
reciprocally , we can check with an elementary integration that @xmath1 is a probability measure and ( using the residue theorem ) that its stieltjes transform is indeed the analytic function @xmath153 characterized for @xmath200 near the real line in .
we therefore do not need to assume conjecture [ conjecture1 ] to prove the existence of @xmath153 in this case .
we treat in the next paragraph the particular case when @xmath2 in more details .
we now analyze ( even more explicitly ) the probability density of the measure @xmath253 at the critical value .
a straightforward computation permits to factorize the derivative polynomial as @xmath254 its minimal root ( with multiplicity two ) is @xmath255 and from relation , we find @xmath256 the factorization of @xmath222 writes as @xmath257 we note in particular that @xmath214 is of multiplicity three for @xmath32 .
this behavior is rather natural at the transition : the root @xmath214 with multiplicity two for @xmath258 reaches multiplicity three for @xmath2 ( the first and second roots merge together ) and finally splits up into three non real roots in @xmath209 , leading to a constant @xmath259 .
we shall see in the next subsection that the probability density starts to flow at this point .
we finally derive the density @xmath174 , which has a compact support , @xmath260 we note the unusual order of the singularity @xmath261 of the density @xmath174 at its lower edge @xmath262 . at the critical value , the concavity changes near the lower edge in order to prepare the sharp phase transition when @xmath178 gets smaller than @xmath263 .
the probability will start to flow in the system as soon as @xmath169 with a constant profile ( allocation ) @xmath154 supported on the whole real line @xmath8 flanked with heavy tails .
it is interesting to note that such singularities at the edges of random matrices spectrum were already observed in @xcite in the context of a non - confining quartic potential .
we shall revisit some of the questions investigated in @xcite in section [ section - quartic - potential ] .
we are now interested in the so called flux of particles in the system , which measures the number of particles which shifts from the right to the left at some given level @xmath176 per unit of time .
it turns out that our method permits us to compute this flux explicitly in the stationary regime .
let us recall the evolution equation which may be rewritten for @xmath264 as @xmath265 where @xmath266 eq . is a continuity equation .
we can transform into an evolution equation on the measure @xmath123 thanks to the stieltjes inversion formula recalled in , so that the interpretation of this continuity equation becomes clearer .
taking imaginary part , integrating on the horizontal segment @xmath267 , x < y , \varepsilon > 0 $ ] and sending @xmath268 , we obtain @xmath269 - \mu_s[x;y ] = \frac{1}{\pi } \lim_{\varepsilon\to 0 } \int_{x+i\varepsilon}^{y+i\varepsilon } dz \int_s^t \partial_z \ , \im j(z , u ) \ , du\ , .
\end{aligned}\ ] ] for @xmath270 fixed , the fubini theorem permits us to exchange the order of integration over @xmath200 and @xmath271 in the right hand side .
we eventually obtain , for any @xmath272 , @xmath273 - \mu_s[x;y ] = \frac{1}{\pi } \lim_{\varepsilon\to 0 } \int_s^t \im \,j(y+i\varepsilon , u ) \ , du - \int_s^t \im \ , j(x+i\varepsilon , u ) \ , du \ , .
\end{aligned}\ ] ] we can interpret the probability measure @xmath123 as the electrostatic charge flowing across @xmath8 from @xmath103 to @xmath111 . therefore , the right hand side of may be seen as the amount of charge which enters the interval @xmath274 $ ] during the time interval @xmath275 $ ] . in order to further extend the present discussion ,
we admit that the stieltjes transform @xmath276 solution of has a _ continuous extension to @xmath277_. unfortunately , we are not able to prove this mathematical detail , although it is physically sound .
note that this continuous extension was proved in @xcite in the case of the complex burgers equation given in ( * ? ? ?
* introduction ) , which is the free analogue of the heat equation .
the fact that @xmath276 has a continuous extension to @xmath277 implies that the probability measure @xmath278 admits a density @xmath279 with respect to lebesgue measure , such that @xmath280 is smooth , and we have @xmath281 where @xmath282 stands for principal value . under this assumption , it is clear that the analytic function @xmath283 defined in has a continuous extension to @xmath277 .
in particular , we have @xmath284 where @xmath285 coming back to , we easily obtain using @xmath286 - \mu_s[x , y ] = \int_s^t ( j_u(y ) - j_u(x))\ , du\,,\end{aligned}\ ] ] which may be rewritten also as @xmath287 .
from , we have a clear physical interpretation of the quantity @xmath288 which is precisely the flux of probability density in @xmath63 at time @xmath77 , measuring the amount of probability density shifting from the right to the left of @xmath63 ( algebraically ) per unit of time at time @xmath77 .
the interesting feature of our matrix model is that the flux does not vanish identically in the stationary state of the scaling limit , as we will see .
if conjecture [ conjecture1 ] holds such that @xmath289 pointwise in @xmath162 , then it follows immediately that @xmath290 .
using now and montel s theorem , we get @xmath291 note that , as one may have expected , the flux of probability becomes independent of the position @xmath176 in the stationary state .
recall also from section [ equilibrium - density ] that @xmath292 is non zero if and only if @xmath6 .
in fact , if @xmath3 , the well of the potential @xmath4 is deep enough to confine all the particles .
the number of explosions per unit of time in the stationary state is negligible compared to the macroscopic mass of the confined particles . on the other hand , if @xmath169 , then the well in @xmath293 ( see fig . [ fig.potential ] ) is not strong enough to confine all the charges which repel each other with electrostatic interaction . if @xmath6 , we have the analytic expression for @xmath210 in terms of the unique root @xmath294 of @xmath207 , given in .
we end this section by establishing the link with the discrete setting where the @xmath20 eigenvalues of the matrix process @xmath59 defined in are diffusing in the potential @xmath4 ( see fig .
[ fig.potential ] ) . in this context , the probability ( or electrostatic charge for a physical analogy ) is carried by the eigenvalues @xmath82 satisfying the stochastic differential system .
each particle @xmath295 carries a proportion @xmath296 of the total probability . using
, we conclude that in the large @xmath20 limit and in the stationary state ( i.e. after a long time @xmath77 ) , the flux of probability density is @xmath297 . in other words ,
the numbers of particles @xmath298 which shift in @xmath63 from the right to the left at time @xmath77 per unit of time is , in the stationary state , proportional to @xmath20 with @xmath299 this formula was checked numerically with very good agreement for @xmath182 .
we have also noted that the convergence in time and in @xmath20 is very fast .
it would be very interesting to compute the fluctuations of this flux of particles around its typical value .
we leave this challenging problem for future research .
[ rem - norestart ] let us say a few words about the non - conservative system already mentioned in remark [ rem - norestart1 ] , where the particles are killed when they explodes instead of being restarted at @xmath103 .
one can easily adapt the proofs of theorem [ main ] , and prove that for any @xmath128 and @xmath129 , if the empirical density at the initial time converges weakly as @xmath20 goes to infinity towards some @xmath131 , then the limit points @xmath134 of the ( almost surely pre - compact ) sequence @xmath132 @xmath300,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] satisfy @xmath135 and their stieltjes transforms @xmath276 are solution of the holomorphic equation @xmath301 \ , ds \ , .
\end{aligned}\ ] ] the main difference with theorem [ main ] is that the solutions @xmath302 of evolve in the space of measures with a total mass @xmath303 decreasing over time .
there is no longer uniqueness of a solution @xmath134 to such that @xmath304 ( eq . depends itself of @xmath305 ) and eq .
does not characterize the limiting process @xmath306 .
it makes the analysis of this model much more complex than in the restarting case .
we conjecture that the limit @xmath307 of the continuum process @xmath302 corresponds to the _ metastable _ equilibrium of the finite-@xmath20 model ( note that for any finite @xmath20 , all the particles will explode in a finite time almost surely ) . in the case of the sub - critical regime ( @xmath169 )
, the system should lose some mass until it gets closer and closer to the critical regime .
the study of the stationary regime gives the relation @xmath308 between the critical point @xmath309 and the mass @xmath310 of the stationary measure . in the supercritical regime ,
the behaviour of the system should not depend on the choice of restarting or killing the particles as the explosions are too scarce to matter .
we now revisit a problem investigated in @xcite ( see also @xcite ) and related to the quartic potential @xmath18 such that @xmath311 where @xmath312 is usually referred as the coupling constant . for negative values of @xmath313 ,
the potential @xmath18 is non - confining with @xmath314 as @xmath10 , respectively . in @xcite ,
the authors consider an ensemble of random matrices with invariant law in the space of complex hermitian matrices given by @xmath315 although the probability distribution does not make sense for @xmath17 , the authors of @xcite are still able to derive ( analytically ) a density , which corresponds if @xmath316 to the limiting spectral density when @xmath19 of the random matrices in the ensemble .
the probability density they obtain still makes sense even for @xmath317 .
the purpose of this section is to bring new lights on this result .
we extend ( with more details ) their computations for @xmath318 to general values of @xmath319 . as a function of @xmath176 for @xmath320 . ]
the method described above with the non - confining cubic potential @xmath4 permits us to define a stationary ensemble of random matrices in the potential @xmath18 for negative values of @xmath313 . as before , we restrict ourselves to symmetric real random matrices although the complex and quaternion hermitian cases may be covered as well .
the idea is again to consider the symmetric matrix process @xmath59 such that @xmath321 with @xmath36 at the initial time and where @xmath37 is a @xmath38 symmetric brownian motion , until the first explosion time . to extend the trajectory of the hermitian process @xmath59 after this explosion time
, we follow the method explained in section [ cubic - hermitian - process ] .
the situation is very similar : the non exploding eigenvalues and eigenvectors trajectories are a.s .
continuous at the explosion times .
the exploding eigenvalue is immediately restarted from @xmath204 ( instead of @xmath103 in section [ cubic - hermitian - process ] ) .
there are now two different types of explosions , either on the right side in @xmath103 or on the left side in @xmath111 .
re starting the eigenvalues in @xmath204 permits us to preserve the symmetry with respect to @xmath322 and , for any @xmath12 , the equality in law @xmath323 holds . a sample path of the eigenvalues of the process @xmath40 is shown in fig .
[ fig.eigenvalues-path-quartic ] .
it is obtained with numerical simulations of the process @xmath40 following the restarting procedure at each explosion times described in this paragraph . as a function of time @xmath77 for @xmath78 , @xmath72 and @xmath324 .
the green horizontal dashed lines give the position of the two symmetric local maximum ( hills ) of the potential @xmath18 in @xmath325 . ]
the restarting procedure of the eigenvalues in @xmath204 leads to some new difficulties , compared to the previous case .
mainly , the stieltjes transform @xmath326 of the empirical measure of the eigenvalues is not continuous at the explosion times .
it is in fact right continuous in @xmath327 with left limit in @xmath328 , with a jump of size @xmath329 in @xmath100 . in order to obtain an evolution equation for the empirical measure in a differential form
, we can consider the set of smooth test functions @xmath330
if @xmath331 , then the function @xmath332 is continuous on @xmath110 , even at the explosions times , and we can use it s formula to obtain , for any @xmath333 , @xmath334 where the @xmath90 are independent real brownian motions . in order to prove convergence as @xmath335 of the empirical measure @xmath139 towards a probability measure @xmath123 and to characterize the evolution of the process
@xmath336 , we propose to follow the same steps as in section [ proof ] .
first , we show precompactness of the family of continuous processes @xmath337 in the space @xmath133,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] .
this proof is rather straightforward adapting the proof of lemma [ precompactness - lemma ] to the present quartic case .
then , taking @xmath19 , we obtain , as in subsection [ evolution.gn ] , the following evolution equation for the limiting process @xmath306 : for all @xmath338 $ ] and @xmath331 , @xmath339 it would remain to show the uniqueness of the continuous process @xmath134 and that this process takes values in the space of probability measures . as in the proof of theorem [ main ] , we only know _ a priori _ that the limit point ( along some subsequence ) @xmath134 belongs to @xmath133,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] ( instead of @xmath133,\mathcal{p}({\mathbb{r}}))$ ] ) .
this question seems more difficult and we do not have insights on the method to prove this .
we leave this problem as an open question .
we are now interested in the probability measures @xmath150 which are stationary solutions of the evolution equation , i.e. such that for any test functions @xmath331 , @xmath340 solving equation by choosing a particular family of functions @xmath331 is not an easy task and has led us to heavy computations .
another route like in @xcite is to fix @xmath148 and simply consider @xmath341 ( although @xmath342 ) . equation conveniently rewrites ( after a further integration ) in term of the stieltjes transform @xmath149 of the probability measure @xmath150 as @xmath343 where @xmath193 is an integration constant .
we repeat the same steps as in section to prove the uniqueness of the analytic function @xmath344 with the following two properties : * there exists @xmath345 such that @xmath346 satisfies the quadratic equation for all @xmath148 ; * @xmath346 is the stieltjes transform of a probability measure @xmath347 . in fact , assuming the existence of such a function @xmath346 , we show that there is a unique possible constant @xmath348 that we compute explicitly .
we can then determine @xmath346 uniquely by solving the quadratic equation .
note however that , in order to prove existence , we still have to check that the solution @xmath346 we obtain is indeed the stieltjes transform of a probability measure
. as we will see , such a stieltjes transform @xmath346 exists if and only if @xmath349 . for @xmath350
however , the method breaks down , leading to a measure which is not a probability measure . in this case
, there are no stieltjes transforms satisfying whatever the values of @xmath193 . for @xmath318 , we re obtain the critical value @xmath351 already obtained in @xcite .
we now explain the main steps of this computation . for any @xmath193 and all @xmath148 , there are two solutions to the quadratic equation @xmath352 the idea is again to compute the derivative polynomial @xmath353 of the discriminant of the quadratic equation , which writes as @xmath354 the quadratic equation @xmath355 has two roots @xmath356 therefore , if @xmath357 , all the zeroes of the polynomial @xmath358 are real and we denote by @xmath359 the minimal root of @xmath353 , given for @xmath17 , by @xmath360 using the gauss lucas theorem and the analyticity of @xmath346 , we can prove as before that all the zeroes of the polynomial @xmath361 are also real .
this implies that @xmath232 . with the same continuity argument of the area below the graph of the underlying probability density with respect to @xmath202
, we can prove that there is in fact a unique @xmath362 insuring the normalization constraint @xmath363 .
physical arguments lead us to the value of @xmath364 .
the six real roots of @xmath32 can not be all of multiplicity one otherwise the measure @xmath150 would have a disconnected support ( union of three disjoints intervals ) , which is counter intuitive .
moreover the polynomial @xmath32 is even so that there are two roots with multiplicity two and two roots with multiplicity one .
the zeroes with multiplicity two are symmetric and exterior while the zeroes with multiplicity one stand inside ( see fig .
[ polynomialp ] ) . for @xmath320 and @xmath72 .
] we can compute the constant @xmath364 from the condition @xmath365 .
we obtain @xmath366 the polynomial @xmath367 is characterized completely and we can factorize it as @xmath368 where @xmath369 the stieltjes transform @xmath346 is also determined : for @xmath370 near the real axis , we obtain the following ( explicit ) expression for @xmath371 after further ( elementary ) computations , @xmath372\ , . \end{aligned}\ ] ] from the stieltjes inversion formula , we can recover the probability density @xmath373 , which is supported on the compact interval @xmath374 $ ] where @xmath375 is given in . for @xmath376 $ ]
, we obtain @xmath377 reciprocally , we can check with an elementary integration that @xmath373 is a probability measure and ( using the residue theorem ) that its stieltjes transform is indeed the analytic function @xmath346 characterized for @xmath200 near the real line in . at the critical value @xmath378
, we find @xmath379 and @xmath380 for @xmath318 , we re obtain the solution found in ( * ? ? ?
* see their figure 1 ) . for @xmath381 and @xmath72 together with the semi - circle obtained for @xmath382 .
the red points represent the empirical density obtained from our simulated samples in the critical case . ] for @xmath383 , the probability density functions @xmath373 , given in for @xmath384 and for @xmath357 , are the limiting spectral density of the matrix @xmath33 which follows the langevin equation in the stationary state as @xmath19 . for @xmath350 ,
the situation is more complicated than that .
one can still try to find the unique constant @xmath364 such that there exists @xmath346 with the two required properties and compute the imaginary part @xmath385 for @xmath386 .
but , in this case , one can check ( at least numerically ) that the function @xmath387 is not a probability density .
its integral over @xmath8 is in fact strictly smaller than @xmath143 .
this simply means that there is not a constant @xmath193 such that the stieltjes transform of some ( probability ) measure satisfies the quadratic equation for all @xmath148 .
there is a rather clear physical interpretation to the non existence of such a constant @xmath202 . for the cubic potential ,
the constant @xmath259 was related to the flux of probability in the system .
the positive imaginary of @xmath210 was measuring the amount of probability density flowing from the right to the left of @xmath388 per unit of time in the stationary state . in the present case of the quartic potential with two sided exits
, there can not be such a stationary flux of charges as the particles can exit either to the right by exploding in @xmath103 or to the left in @xmath111 .
it would actually be very interesting to compute the limiting density @xmath373 of eigenvalues in the stationary state for values of @xmath389 , by analyzing directly .
we would expect some diverging terms coming in at the origin , where there is a birth process of new particles .
we now conclude this work with some open questions and further comments on possible extensions .
we note that the case of a cubic potential was also considered at the end of @xcite .
the authors study divergent matrix integrals of the form @xmath390 where @xmath391 viewed as power series and they recover the limiting spectral density when the confinement is strong enough so that the eigenvalues are localized in a compact support .
the ideas developed in this paper permit one to define invariant ensembles associated to the potential @xmath392 for @xmath316 .
the stationary spectral densities can also be obtained in explicit forms , as was done in @xcite for @xmath313 smaller than a critical value @xmath393 .
the novel contribution of our work is that we went further by computing the spectral density in the regime @xmath394 , where it has unbounded support with a stationary macroscopic current of particles across the system .
more generally , the construction explained in section [ cubic - hermitian - process ] permits one to define invariant ensembles of random matrices in general polynomial potentials @xmath30 of arbitrary odd degree @xmath395 , with possibly multiple wells .
the stieltjes transform method to study the large @xmath20 limit of the spectrum does not adapt straightforwardly though because unknown moments of the limiting spectral densities arise in the evolution equation satisfied by @xmath149 if @xmath396 .
those unknown moments have to be determined from the constraints and analyticity conditions but we expect the explosions to occur sufficiently fast for the tails of the spectral density to be light enough for those ( few ) moments to be finite .
we also conjecture that such potentials lead to stationary spectral densities with support on several disconnected intervals , in the semi - confining regimes .
a sharp phase transition should be observed as the confinement gets weaker , leading to spectral densities with unbounded support and associated to macroscopic stationary currents of particles in the system . in the semi - confining cases
, the eigenvalues would lie in the wells near the local minimums of the potential @xmath24 , leading to a more complex deformation towards the unbounded probabilities found in the fully non - confining regimes .
we are currently working on this problem .
it seems that other types of phase transitions for the spectral density at the critical value could be observed .
indeed , one can obtain , through an appropriate choice of the ( polynomial ) potential @xmath24 @xcite , a limiting spectral density which vanishes at the end point of the spectrum @xmath397 as @xmath398 for any integer @xmath399 . the potential @xmath24 is then called @xmath395 multi - critical .
a general potential usually leads to criticality @xmath400 , as in the case of the quadratic ( gaussian ) potential .
an interesting extension of our work would be to exhibit such ( polynomial ) potentials with odd degree which would cover general criticality parameters @xmath401 and lead to more complex phase transitions for the spectral density . in our work ,
the potential @xmath4 at the critical value @xmath2 is @xmath402 multi - critical , leading to a density vanishing at the lower edge as @xmath403 ( with @xmath404 ) .
other interesting questions would be to study the top eigenvalues statistics in the subcritical case @xmath23 when the density of eigenvalues has full support with tails @xmath405 .
in particular , the fluctuations of the top eigenvalue around its asymptotic value of order @xmath20 are of interest in this fully non - confining regime
. one may wonder whether those statistics are related to those of the top eigenvalues of heavy tailed wigner matrices , or if they are of a different nature .
in particular , the crossover for those statistics at the critical value @xmath2 is interesting .
the eigenvalues statistics at the edge of the spectrum in the critical @xmath2 and super critical @xmath258 cases have been extensively investigated since @xmath406 . in @xmath407 , bowick and brzin @xcite
have computed the spectral density at the edge of the spectrum for general multi - critical potentials , in the scaling region of width @xmath408 with @xmath409 , where @xmath395 is the criticality ( see also @xcite ) .
their results are in fact universal in the sense that the scaling shape of the density at the edges does not depend on the particular choice of the potential @xmath24 of given criticality @xmath395 . in particular , for @xmath400 their work is pioneer and provides the first description of the tracy - widom region . in @xmath410 , tracy and widom
@xcite proved the weak convergence of the top eigenvalue for the classical gaussian ensembles ( which correspond to the quadratic potential ) to the so called tracy - widom distributions .
a more precise description was later provided in @xcite where the authors prove that the joint convergence of the top eigenvalues of the gaussian ensembles to those of the stochastic airy operator .
the multi - critical cases @xmath411 were later considered in @xcite and the results of @xcite are conjectured to extend for those potentials as well ( see the end of @xcite ) .
we conclude this paper with a last open question on the fluctuations of the random flux of particles around its asymptotic value given in eq .
, in the stationary state and in the large @xmath20 limit . it would be very interesting to relate the order of the fluctuations , the shape of the fluctuations in the central regime or the large deviations regimes to other models of interacting particles with random flux .
recall the notations @xmath412 for the successive explosions times of the diffusive matrix process @xmath40 .
following @xcite , we look for an evolution equation for the stieltjes transform @xmath326 . by it s formula , we have for any @xmath148 and for any @xmath413 , @xmath414 at the explosion times @xmath415
, the eigenvalue which explodes in @xmath111 , is immediately restarted at @xmath103 , so that the stieltjes transform @xmath326 of the empirical measure is continuous in @xmath100 .
the evolution equation therefore holds for all @xmath12 .
we want to rewrite this equation in a more compact way as a function of @xmath416 . using symmetry and anti - symmetry properties
, we have @xmath417 the last expression conveniently rewrites in terms of @xmath416 as @xmath418 from and the previous computations , we have , for any @xmath148 and @xmath12 , @xmath419 ds \notag \\ & - \frac{1}{n^{3/2 } } \sum_{i=1}^n \int_0^t \frac{db_i(s)}{(\lambda_i(s)-z)^2}\ , .
\end{aligned}\ ] ] we aim at taking @xmath19 . for this
, we need to prove that the sequence of probability measures process @xmath420 is almost surely pre - compact in the space @xmath133,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] where @xmath142 is the space of borel measures @xmath119 on @xmath8 with total mass @xmath421 equipped with its weak-@xmath144 topology .
it is convenient to work with the weak-@xmath144 topology on @xmath142 ( where a sequence @xmath422 converges to @xmath150 iff @xmath423 converges to @xmath424 for all @xmath425 , the space of continuous and compactly supported functions on @xmath8 ) as this topology is metrizable on this bounded set and makes @xmath142 compact .
therefore , the topology on @xmath133,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] we consider is simply the topology of uniform convergence for this metric .
note that the weak-@xmath144 topology on the bounded set @xmath426 is equivalent to the vague topology topology is weaker than the usual topology of weak - convergence where the limit should hold for all continuous bounded functions . ]
( where @xmath422 converges to @xmath150 iff @xmath423 converges to @xmath424 for all @xmath427 , the continuous functions on @xmath8 which tends to @xmath204 when @xmath428 ) .
let us emphasize that , in contrast with the usual case handled in @xcite where the authors prove a dynamical version of wigner s theorem , we do not need to prove pre - compactness in the smaller space @xmath133,\mathcal{p}({\mathbb{r}}))$ ] as we will prove that the continuous limiting process ( along subsequences ) necessarily takes values in the space of probability measures .
this almost sure pre - compactness in @xmath133,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] is proved in lemma [ precompactness ] . from any subsequence of @xmath429
, we can now extract a sub - sub - sequence @xmath430 such that we have the following pointwise convergence in @xmath133,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] , @xmath431 where @xmath432,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] . to any limit point @xmath433 , we associate its stieltjes transform process @xmath434 such @xmath147 is the stieltjes transform of the measure @xmath123 for any @xmath12 .
we note that , for @xmath148 fixed , the function @xmath435 and its derivative @xmath436 are continuous and tends to @xmath204 when @xmath428 and therefore belongs to @xmath437 .
we deduce the following convergences ( along the subsequence @xmath438 ) for any @xmath439 $ ] , @xmath440 note in addition that the martingale term @xmath441 which appears in has a quadratic variation @xmath442 smaller then @xmath443 . by the burkholder - davis - gundy inequality ( * ? ? ?
* theorem h.8 ) and the chebyshev s inequality , we get that , for a universal constant @xmath444 , @xmath445 { \;\leqslant\;}\frac{c \ , t}{\varepsilon^2 n^2|\im z|^{4 } } \ , .
\end{aligned}\ ] ] by the borel - cantelli lemma , for any @xmath148 , we have @xmath446 as @xmath447 almost surely . eventually ,
from , we obtain the following equation satisfied by the stieltjes transform process @xmath448 of any limit point @xmath449 of the pre - compact family @xmath450 ( see lemma [ precompactness - lemma ] ) , @xmath451 \ , ds \ , , \end{aligned}\ ] ] which holds for all @xmath452 , z\in \h$ ] .
it remains to prove that the measure @xmath123 is indeed a probability measure for any @xmath12 .
let @xmath333 , @xmath270 and @xmath453 . from , we easily deduce that for all @xmath270 , @xmath454 \,.\end{aligned}\ ] ] recalling @xmath455
, we can check that the left hand side of is of order @xmath456 ( or even smaller ) as @xmath268 while the right hand side is equivalent to @xmath457 ( using in addition the continuity of the function @xmath458 ) .
therefore we have @xmath459 for every @xmath333 .
finally , the convergence of @xmath460 to @xmath461 follows by uniqueness of the solution of ( proved in lemma [ unicity - lemma ] ) .
theorem [ main ] is proved .
denote by @xmath142 the space of borel measures on @xmath8 with total mass smaller or equal to @xmath143 equipped with its weak-@xmath144 topology and recall that @xmath115 denotes the space of probability measures on @xmath8 .
[ precompactness - lemma ] the family @xmath463 of continuous process in @xmath115 is almost surely pre - compact in the space @xmath133,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] . _
_ we first describe a family of compact subsets of @xmath133,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] .
let @xmath464 be a sequence of bounded continuous functions dense in the space @xmath465 of continuous function on @xmath8 which converge to @xmath204 in @xmath466 and let @xmath467 be a family of compacts subsets of the space @xmath133,{\mathbb{r}})$ ] of continuous functions from @xmath468\to { \mathbb{r}}$ ] .
then , adapting the proof of lemma 4.3.13 in @xcite , we can prove that the set @xmath469 is a compact subset of @xmath133,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] .
this proof is straightforward showing that @xmath470 is closed and sequentially compact,\mathcal{m}_{\le 1}({\mathbb{r}}))$ ] is metrizable . ] with a diagonal extraction , noting in addition that @xmath142 is compact for the topology of weak-@xmath144 convergence .
we denote by @xmath471 the set of twice differentiable functions @xmath472 , such that @xmath473 and @xmath474 and such that , in addition , @xmath475 .
note that @xmath471 is dense in @xmath465 .
we need an estimate on the hlder norm of the function @xmath476 for any @xmath331 .
applying it s formula , we get for any @xmath477 , @xmath478 using symmetry , we have @xmath479 therefore , @xmath480 denote by @xmath481 . gathering the above estimates , it follows that , such that for any @xmath482 ^ 2 $ ] , @xmath483 is the martingale process defined as @xmath484 with the same proof as in ( * ? ? ?
* proof of lemma 4.3.14 , page 265 ) , we can prove that there exists a constant @xmath444 which depends only on @xmath485 such that , for any @xmath486 , @xmath487 { \;\leqslant\;}\frac{c \delta^{1/2}}{n^2 m^4 } ||f'||_\infty^2\,.\end{aligned}\ ] ] using , we deduce that for any @xmath488 small enough and @xmath489 , we have @xmath490 { \;\leqslant\;}\frac{c \delta^{1/2}}{n^2 m^4}||f'||_\infty^2\ , . \end{aligned}\ ] ] recall that , by the arzela - ascoli theorem , sets of the form @xmath491 , { \mathbb{r } } ) : \sup_{|t - s| { \;\leqslant\;}\delta_k \atop 0{\;\leqslant\;}s , t{\;\leqslant\;}t } |g(t)-g(s)| { \;\leqslant\;}\epsilon_k , \sup_{0{\;\leqslant\;}t { \;\leqslant\;}t } |g(t)| { \;\leqslant\;}1/\alpha_k \}\ , . \end{aligned}\ ] ] where @xmath492 are sequences of positive real numbers going to zero as @xmath395 goes to infinity , are compact in @xmath116 , { \mathbb{r}})$ ] . for @xmath331 and @xmath493
, we consider the subset of @xmath116 , \mathcal{m}_{\le 1}({\mathbb{r}}))$ ] , defined by @xmath494 , \mathcal{m}_{\le 1}({\mathbb{r } } ) ) : \sup_{|t - s| { \;\leqslant\;}k^{-4 } \atop 0{\;\leqslant\;}s , t{\;\leqslant\;}t } |\int_{\mathbb{r}}f(x ) \nu_t(dx ) -\int_{\mathbb{r}}f(x ) \nu_s(dx ) | { \;\leqslant\;}\frac{1}{\varepsilon \sqrt{k } } \}\ , .
\end{aligned}\ ] ] then , using , we have @xmath495 } \in c_t(f,\varepsilon)^c \right ] { \;\leqslant\;}\frac{c\varepsilon^4}{n^4}\ , .
\end{aligned}\ ] ] we now pick a dense family @xmath496 in @xmath465 of functions @xmath497 and setting @xmath498 , we define the subset @xmath499 , \mathcal{m}_{\le 1}({\mathbb{r}}))\,.\end{aligned}\ ] ] by the borel - cantelli lemma , we get @xmath500 } \in \mathcal{k } ] = 1\,,\end{aligned}\ ] ] and the lemma follows since @xmath470 is a compact set of @xmath116 , \mathcal{m}_{\le 1}({\mathbb{r}}))$ ] .
[ unicity - lemma]let @xmath146 .
there exists a unique ( deterministic ) process @xmath501 in the space of analytic function @xmath502 which enjoys the two following properties : * for any @xmath12 , @xmath147 is the stieltjes transform of a real probability measure ; * @xmath149 is a strong solution of the holomorphic partial differential equation on @xmath503 , @xmath504\ , , \end{aligned}\ ] ] with initial condition @xmath505 . _
_ we set @xmath506 it is straightforward to check that the analytic function @xmath507 satisfies the evolution equation for @xmath12 , @xmath508 it suffices to prove that there is at most one unique strong solution @xmath509 of the partial differential equation ( pde ) such that in addition @xmath510 is analytic for any @xmath12 .
following @xcite , we use the characteristic method . for @xmath511 , we consider the following cauchy problem @xmath512 for @xmath513 and any @xmath514 , we have for any @xmath12 , @xmath515 we deduce that the continuous function @xmath516 is _ locally _ in @xmath200 ( globally in @xmath77 ) lipschitz on @xmath517 with respect to the first variable @xmath200 .
more precisely , we mean that , for any compact set @xmath518 , there exists a constant @xmath489 such that for any @xmath12 and @xmath519 , we have @xmath520 , there exists a unique solution @xmath521 to the cauchy problem , defined up to a time @xmath270 small enough such that @xmath522 for all @xmath523 ( recall that @xmath507 is analytic on @xmath162 so that the solution has to remain in this domain for the differential equation in to be well defined ) .
let us now fix @xmath524 .
we can prove using the lipchitz condition and that we can pick @xmath525 small enough ( depending only on @xmath526 and on the constant @xmath527 ) such that _ any _ solution of the cauchy problem starting from any @xmath528 in the open ball @xmath529 with center @xmath526 and radius @xmath525 is defined up to a time @xmath270 ( which depends on @xmath527 and @xmath530 but which is _ independent _ of @xmath531 ) . differentiating with respect to @xmath77 and using also the pde
, we obtain a new _ explicit _ cauchy problem for the function @xmath200 , @xmath532 where @xmath207 is the polynomial we have already met in .
we now regard the cauchy problem as a holomorphic cauchy problem on the domain @xmath533 .
we know from the fundamental theorem @xcite that there exists a unique solution to defined for all @xmath534 and for any initial condition @xmath535 .
moreover , @xmath536 depends holomorphically on the initial conditions : there exists a holomorphic function @xmath537 such that , for any @xmath534 , @xmath538 by uniqueness , for any @xmath531 , the solution @xmath539 of the cauchy problem coincides with the one of the cauchy problem on the small interval @xmath540 .
we now fix @xmath541 $ ] and we consider the _ open and simply connected domain _
@xmath542 of the targets " . by definition of @xmath543 , we have @xmath544 .
besides , by uniqueness , the solutions of the cauchy problem can not coincide in @xmath162 at the given time @xmath77 , and @xmath545 is a conformal isomorphism ( i.e. analytic and bijective ) .
finally , for any @xmath546 $ ] and @xmath547 fixed , we can find a unique @xmath531 such that the solution of satisfies @xmath548 .
then the formula @xmath549 valid for any @xmath550 in the open and simply connected domain @xmath551 , characterizes uniquely the analytic function @xmath552 .
this argument is true for any @xmath553 so that @xmath554 is uniquely characterized for all @xmath148 and @xmath555 $ ] .
the same method ( starting from the same initial condition @xmath526 for example so that the same @xmath543 will work ) permits us to extend this characterization on the intervals @xmath556 $ ] , @xmath557 $ ] , ... uniqueness of @xmath149 is implied by the uniqueness of @xmath40 from .
we first prove the existence part of ( 1 ) .
it is easy to see that the solution @xmath149 of the evolution equation is lipchitz in @xmath77 locally uniformly in @xmath200 i.e. for all compact subset @xmath558 , there exists a constant @xmath527 such that for all @xmath12 and all @xmath559 , @xmath560 $ ] is lipchitz in @xmath77 locally uniformly in @xmath200 as well . the existence of the limit when @xmath307 of @xmath276 leads to the existence of the limit of the integral on the right hand side of . as the integrated function is uniformly continuous
, we deduce that @xmath153 is a stationary solution of , i.e. @xmath561 = 0\,,\end{aligned}\ ] ] so that there exists a constant @xmath193 such that for all @xmath148 , @xmath562 from , we already know that @xmath153 is bounded in the neighbourhood of @xmath127 . therefore , we deduce from that indeed @xmath563 .
we then know from ( * ? ? ?
* theorem 1 ) that @xmath153 is the stieltjes transform of a probability measure @xmath156 and that @xmath123 converges weakly to @xmath156 .
the proposition is proved .
let us check that the probability distribution @xmath32 defined in is a stationary measure of the stochastic differential system .
first notice that , if @xmath527 is a @xmath38 real matrix , then the gradient of the function @xmath564 with respect to the @xmath565 entries of the matrix @xmath527 , is the function @xmath566 .
thus , if @xmath40 is a hermitian matrix , we simply have @xmath567 it remains to check that the probability distribution @xmath32 is the unique stationary solution of the _ fokker planck _ equation satisfied by the ( stationary ) transition probability of the diffusion process @xmath59 , @xmath568 + \frac{1}{2n } \sum_{i , j=1 } \frac{1 + \delta_{i = j}}{2 } \frac{\partial^2}{\partial h_{ij}^2 } p(h)\ , .
\end{aligned}\ ] ] the reader may actually check that the function @xmath32 as defined in satisfies , for any hermitian matrix @xmath40 , the following conditions @xmath569 under which trivially holds .
the factor @xmath402 which appears in the second line is due to the symmetry of the matrix @xmath40 .
the stieltjes transform is frequently used in random matrix theory for the study of empirical spectral densities in the large @xmath20 limit . a measure @xmath150
is characterized by its stieltjes transform , which is an analytic function @xmath570 ( @xmath162 denotes the open upper half - plane ) , defined as @xmath571 we have the following inversion formula valid for any measure @xmath150 on @xmath8 , @xmath572 where @xmath573 denotes the imaginary part of @xmath574 .
if @xmath150 is a probability measure , its stieltjes transform @xmath576 behaves as @xmath577 when @xmath233 goes to @xmath103 .
reciprocally , akhiezer s theorem @xcite states a useful criterium characterizing stieltjes transforms of probability measure : @xmath149 is the stieltjes transform of a probability measure iff @xmath149 is analytic on @xmath162 with @xmath578 and @xmath579 as @xmath580 .
_ on the statistical distribution of the widths and spacings of nuclear resonance levels _ , math .
cambridge philos .
soc . , * 47 * , 790 - 798 xiii , 3 ( 1951 ) .
g. akemann , j. baik , and ph .
di francesco . _
the oxford handbook of random matrix theory _
( oxford university press , new york , 2011 ) .
anderson , a. guionnet , and o. zeitouni .
_ an introduction to random matrices _ , cambridge studies in advanced mathematics ( cambridge university press , cambridge , 2009 )
. z. bai and j. silverstein .
_ spectral analysis of large dimensional random matrices _
( springer , new york , 2010 ) , 2nd ed . m.l .
_ random matrices _ ( elsevier , new york , 2004 ) .
_ log gases and random matrices _ ( princeton university press , princeton , 2010 ) .
e. brezin , c. itzykson , g. parisi , and j. b. zuber .
_ planar diagrams _ , commun .
phys . * 59 * , 3551 ( 1978 ) .
g. t hooft .
_ a planar diagram theory for strong interactions_. nuclear physics b , * 72 * : 461?473 , ( 1974 )
. b. i. halperin .
_ green s functions for a particle in a one - dimensional random potential _ , phys
139 * a 104-a117 ( 1965 ) .
a. ferrari and l. r. g. fontes .
_ current fluctuations for the asymmetric simple exclusion process_. the annals of probability * 22 * ( 2 ) 820 - 832 ( 1994 )
. m. prhofer and h. spohn .
_ current fluctuations for the totally asymmetric simple exclusion process_. progress in probability , * 51 * 185 - 204 ( 2002 ) .
p. biane , r. speicher .
_ free diffusions , free entropy and free fischer information _ ann .
i. h. poincar , * 37 * 581 - 606 ( 2001 ) .
j. a. ramrez , b. rider and b. virg . _ beta ensembles , stochastic airy spectrum and a diffusion _ , j. amer .
* 24 * 919 - 944 ( 2011 ) .
l. dumaz and b. virg . _ the right tail exponent of the tracy - widom - beta distribution _ ,
h. poincar probab .
* 49 * , 4 , 915 - 933 , ( 2013 ) .
a. bloemendal and b. virg .
theory relat .
fields ( online first ) ( 2012 ) .
a. bloemendal and b. virg .
_ limits of spiked random matrices ii _ , arxiv:1109.3704 ( 2011 ) .
h. l. frisch and s. p. lloyd .
phys . rev . *
120 * 1175 - 1189 ( 1960 ) . h. p. mckean .
_ a limit law for the ground state of hill?s equation _ , j. stat
* 74 * , 1227 ( 1994 ) . c. texier .
_ individual energy level distributions for one - dimensional diagonal and off - diagonal disorder _ , j. phys .
a : math . gen . * 33 * 6095 ( 2000 ) .
r. allez and l. dumaz .
_ tracy - widom at high temperature _ , arxiv:1312.1283 ( 2013 )
k. it and h. p. mckean .
_ diffusion processes and their sample paths_. springer ( 1996 ) .
m. morse . _
the calculus of variations in the large _ , american mathematical society colloquium publications , * 18 * ( 1932 ) ( 1996 reprint of the original ) .
variational analysis : critical extremals and sturmian extensions , interscience publishers , john wiley and sons , inc .
( 1973 ) w. t. reid .
_ ordinary differential equations _ , john wiley and sons inc .
, new york ( 1971 ) .
w. t. reid .
_ riccati differential equations _
, academic press ( 1972 ) .
g. baur and w. kratz .
_ a general oscillation theorem for selfadjoint differential systems with applications to sturm - liouville eigenvalue problems and quadratic functionals _ , rend .
circ . mat .
palermo ( 2 ) * 38 * : 329 - 370 ( 1989 ) . f. j. dyson .
_ a brownian - motion model for the eigenvalues of a random matrix
_ , j. mathematical phys . * 3
* : 1191 - 1198 ( 1962 ) .
r. allez and a. guionnet .
_ a diffusive matrix model for invariant @xmath175-ensembles _ , electron . j. probab .
* 18 * 62 , 1 - 30 ( 2013 ) .
rogers and z. shi .
_ interacting brownian particles and the wigner law _ , probab .
theory relat .
fields * 95 * , 555 - 570 ( 1993 ) .
r. allez , j .-
bouchaud and a. guionnet .
_ invariant @xmath175-ensembles and the gauss - wigner crossover _ , phys .
lett . * 109 * , 094102 ( 2012 ) .
r. allez , j .-
bouchaud , s. n. majumdar , p. vivo .
_ invariant beta - wishart ensembles , crossover densities and asymptotic corrections to the marchenko - pastur law _ , j. phys .
* 46 * 015001 ( 2013 ) .
j. s.geronimo and t. p.hill .
_ necessary and sufficient condition that the limit of stieltjes transforms is a stieltjes transform _ ,
journal of approximation theory * 121 * 54 - 60 ( 2003 ) .
n. akhiezer .
_ the classical moment problem_. hafner , new york , ( 1965 ) .
_ on the free convolution with a semi - circular distribution
_ , indiana university mathematics journal .
* 46 * ( 3 ) 705 - 718 , ( 1997 ) .
m. douglas .
_ large n quantum field theory and matrix models _ in : d.v .
voiculescu , free probability theory , fields institute communications , * 12 * , 21 - 40 , ( 1997 ) .
h. neuberger .
_ regularized string and flow equations _ ,
nucl . phys .
b * 352 * 689 ( 1991 ) .
m. j. bowick and e. brzin .
_ universal scaling of the tail of the density of eigenvalues in random matrix models_. phys .
b * 268 * 21 - 28 ( 1991 ) .
p. j. forrester .
_ the spectrum edge of random matrix ensembles _ , nucl . phys .
b * 402 * , 709 - 728 ( 1993 ) . c. a. tracy and h. widom .
_ level - spacing distributions and the airy kernel _ , comm . math .
* 159 * ( 1 ) : 151 - 174 ( 1994 ) . c. a. tracy and h. widom .
_ on orthogonal and symplectic matrix ensembles _ , comm .
177*(3 ) : 727 - 754 ( 1996 ) .
t. claeys , i. krasovsky , a. its .
_ higher - order analogues of the tracy - widom distribution and the painlev ii _ communications on pure and applied mathematics * 63 * , 3 , 362 - 412 ( 2010 ) .
m. krishnapur , b. rider , b. virg .
_ universality of the stochastic airy operator _ , arxiv:1306.4832 ( 2013 ) . | we consider invariant matrix processes diffusing in non - confining cubic potentials of the form @xmath0 .
we construct the trajectories of such processes for all time by restarting them whenever an explosion occurs , from a new ( well chosen ) initial condition , insuring continuity of the eigenvectors and of the non exploding eigenvalues .
we characterize the dynamics of the spectrum in the limit of large dimension and analyze the stationary state of this evolution explicitly .
we exhibit a sharp phase transition for the limiting spectral density @xmath1 at a critical value @xmath2 .
if @xmath3 , then the potential @xmath4 presents a well near @xmath5 deep enough to confine all the particles inside , and the spectral density @xmath1 is supported on a compact interval . if @xmath6 however , the steady state is in fact dynamical with a macroscopic stationary flux of particles flowing across the system . in this regime ,
the eigenvalues allocate according to a stationary density profile @xmath7 with full support in @xmath8 , flanked with heavy tails such that @xmath9 as @xmath10 .
our method applies to other non - confining potentials and we further investigate a family of quartic potentials , which were already studied in @xcite to count planar diagrams . |
the performance of future long baseline neutrino oscillation experiments to unravel leptonic cp violation ( cpv ) and the neutrino mass hierarchy ( mh ) is crucially dependent on the achievable event rates and hence strongly on the size of @xmath3 @xcite .
recent t2k
@xcite and minos @xcite data have provided new information on @xmath0 , indicating that 0.01 @xmath4 0.1 at @xmath5 confidence level with a best - fit value of 0.051 ( 0.063 ) for normal ( inverted ) mass ordering @xcite .
it is expected that in the next few years , this hint will be validated with high significance with more informations coming from the t2k , minos and upcoming reactor and no@xmath6a experiments , opening up the possibility of observing cpv in the lepton sector if the dirac cp phase , @xmath7 , is not equal to @xmath8 or @xmath9 .
the prospects of the present generation long baseline experiments , t2k and no@xmath6a in probing cpv and the mass ordering are very limited even if @xmath0 is large @xcite .
the experiment class , which can ultimately address these unsolved issues with unprecedented sensitivity for both small and large @xmath0 , is the neutrino factory @xcite . in a neutrino factory ,
the neutrino beam is created from the decay of muons in flight in the straight sections of a storage ring .
the international neutrino factory and superbeam scoping study @xcite has laid the foundations for the currently ongoing design study for the neutrino factory ( ids - nf ) @xcite which has come up with a first - version baseline setup of a high energy neutrino factory ( henf ) with @xmath10 and two baselines @xmath11 and @xmath12 served by two racetrack - shaped storage rings @xcite .
the most promising avenue to explore cpv and @xmath13 at a neutrino factory is the sub - dominant @xmath14 oscillation channel which produces muons of the opposite charge ( wrong - sign muons ) to those stored in the storage ring and these can be detected with the help of the charge identification capability of a magnetized iron neutrino detector ( mind ) @xcite . as the design of the neutrino factory matures ,
more realistic detector simulation and analysis technique have been developed which improves the efficiency of mind at low energies and also lowers the threshold @xcite .
these new simulations provide the full response matrices for all the signal and background channels . in the most recent analysis @xcite ,
low energy neutrino signal events down to @xmath15 were selected with an efficiency plateau of @xmath1660% for @xmath17 and @xmath1680% for @xmath18 events starting at @xmath16@xmath19 , while maintaining the background level at or below 10@xmath20 .
for the neutral current background , the impact of migration is non - negligible and it is peaked at lower energies .
this feed - down is the strongest effect of migration and thus has potential impact on the energy optimization , since it penalizes neutrino flux at high energies , where there is little oscillation but a large increase in fed - down background .
now we discuss the simultaneous optimization of baseline and muon energy in a green - field neutrino factory scenario @xcite using the new migration matrices for mind detector @xcite .
( true ) for which cpv will be discovered at @xmath1 c.l . as a function of @xmath21 and @xmath22 for the single baseline neutrino factory .
the different panels correspond to different true values of @xmath23 , as given there . here
we consider @xmath24 useful muon decays in total with a 50 kt detector .
the optimal performance is marked by a dot : ( 2200,10.00 ) , ( 2288,13.62 ) , ( 3390,20.00 ) and ( 4345,22.08 ) with regard to their best reaches of the fraction of @xmath7(true ) at : 0.77 , 0.84 , 0.67 and 0.42.,width=491,height=377 ] [ fig : lvse ] here we are interested to optimize @xmath21 and @xmath22 for the single baseline neutrino factory to have cpv discovery at @xmath1 c.l . for different fractions of true values of @xmath7 assuming that the true value of @xmath23 is known .
the different panels of fig .
[ fig : lvse ] depict the same for four different true values of @xmath23 .
as expected , this figure clearly shows that the optimization strongly depends on the true value of @xmath23 chosen . in case of large @xmath25 ,
shorter baselines and lower energies are preferred .
even @xmath22 as low as 5 to @xmath26 at the fermilab - homestake baseline of about @xmath27 is quite close to the optimal choice , which means that the mind detector approaches the magnetized totally active scintillator detector performance of the low energy neutrino factory ( lenf ) @xcite .
note that compared to earlier analyses without feed - down effect due to background migration , now too high @xmath22 are in fact disfavored in the large @xmath23 case .
for the other extreme , @xmath28 , baselines between @xmath29 and @xmath30 km are preferred with @xmath31 gev , which corresponds more to the choice of @xmath32 for henf , such as the ids - nf baseline . including the other two panels
, the optimal region within each panel moves from the lower left on the plots to the upper right as the true value of @xmath23 decreases .
this means that , depending on the choice of true @xmath23 , the optimization results in the lenf , the henf , or an intermediate scenario , and that the low and high energy neutrino factories are just two versions of the same experiment optimized for different parts of the parameter space .
note that , all the figures presented in this talk have been taken from @xcite and all the simulations have been performed using the globes software @xcite .
confidence level is possible as a function of true @xmath23 . from left to right for cpv , mh and @xmath0 .
here we compare the performances of a single baseline lenf with @xmath33 gev , @xmath34 , and a 50 kt mind detector , and a two baseline henf with @xmath35 gev and a 50 kt mind detector at each baseline , and @xmath36 , and @xmath37.,scaledwidth=100.0% ] [ fig : lenf ] in this talk , we have discussed the optimization of the neutrino factory based on the most up - to - date analysis of the mind detector using response matrices for all the signal and background channels . the lower threshold and
higher efficiencies compared to earlier simulations indicate that the mind detector characteristics are getting more similar to the characteristics of the detectors proposed for the lenf ( e.g. , a magnetized tasd ) . for large @xmath23 ,
a single baseline neutrino factory with @xmath22 as low as 5 to 8 gev and a baseline as short as fermilab to homestake ( about @xmath27 ) might be sufficient . in the case of small @xmath38 , however , a two baseline neutrino factory with one baseline between about 2500 km and 5000 km , the other one at about the magic baseline 7500
km is ideal .
the additional information coming from the second baseline which is at the magic distance ( insensitive to @xmath7 ) helps to remove the degeneracies and it enhances the fraction of @xmath7 for which cpv discovery is possible .
[ fig : lenf ] nicely depicts this message where we compare the performance of the optimal single baseline lenf with the optimal two baseline henf for the same mind detector .
it is quite evident from this figure that for @xmath39 the low energy version can perform all of the required measurements , whereas for smaller values the henf is clearly better .
it suggests that this different optimization would be sufficient to compensate for the relative deterioration of performance at large @xmath23 observed in the traditional henf .
we conclude that the features of the new detector simulation that the backgrounds are typically reconstructed at lower energies and that the threshold is lower have significant impact on the optimization of @xmath22 .
the recent simulation results for the mind have reduced the performance margin between tasd and mind considerably .
therefore , we suggest that the low- and high - energy neutrino factory should not be regarded as separate options and they merely correspond to two extreme corners of a common parameter space .
i am grateful to the conveners of working group 1 for the invitation .
i would like to thank patrick huber , jian tang and walter winter for their collaboration .
i would also like to thank nita sinha and paul soler for insightful discussions during nufact11 .
i acknowledge the support from the european union under the european commission framework programme 07 design study euronu , project 212372 and the project consolider - ingenio cup . | recent results from long baseline neutrino oscillation experiments point towards a non - zero value of @xmath0 at around @xmath1 confidence level . in the coming years
, further ratification of this result with high significance will have crucial impact on the planning of the future long baseline neutrino factory setup aimed to explore leptonic cp violation and the neutrino mass ordering . in this talk
, we discuss the baseline and energy optimization of the neutrino factory including the latest simulation results on the magnetized iron neutrino detector ( mind ) in the light of both small and large @xmath0 .
we find that in case of small @xmath0 , baselines of about 2500 to 5000 km is the optimal choice for the cp violation measurement with @xmath2 as low as 12 gev can be considered .
however , for large @xmath0 , we show that the lower threshold and the backgrounds reconstructed at lower energies allow in fact for muon energies as low as 5 to 8 gev at considerably shorter baselines , such as fermilab to homestake .
this suggests that with the latest mind simulation , low- and high - energy versions of the neutrino factory are just two different forms of the same experiment optimized for different regions of the parameter space .
euronu - wp6 - 11 - 40 + ific/11 - 60 _ contribution to nufact 11 , xiiith international workshop on neutrino factories , super beams and beta beams , 1 - 6 august 2011 , cern and university of geneva ( submitted to iop conference series ) _ |
due to the advancement of cloud computing technologies , there has been an increased interest for individuals and business entities to move their data from traditional private data center to cloud servers . indeed , even popular storage service providers such as dropbox use third party cloud storage providers such as amazon s simple storage service ( s3 ) for data storage . with the wide adoption of cloud computing and storage technologies ,
it is important to consider data security and reliability issues that are strongly related to the underlying storage services .
though it is interesting to consider general data security issues in cloud computing environments , this paper will concentrate on the basic question of reliable data storage in the cloud and specifically the coding techniques for data storage in the cloud .
there has been extensive research in reliable data storage on disk drives .
for example , redundant array of independent disks ( raid ) techniques have been proposed and widely adopted to combine multiple disk drive components into a logical unit for better resilience , performance , and capacity . the well known solutions to address the data storage reliability are to add data redundancy to multiple drivers .
there are basically two ways to add the redundancy : data mirror ( e.g. , raid 1 ) and data stripping with erasure codes ( e.g. , raid 2 to raid 6 ) .
though data mirror ( or data replication ) provides the straightforward way for simple data management and repair of data on corrupted drives , it is very expensive to implement and deploy due to its high demand for redundancy .
in addition to data replication techniques , erasure codes can be used to achieve the required data reliability level with much less data redundancy .
note that though error correcting codes ( e.g. , reed - solomon codes ) could also be used for reliable data storage and correcting errors from failed disk drives , it is normally not used for data storage since it needs expensive computation for both encoding and decoding processes .
erasure codes that have been used for reliable data storage systems are mainly binary linear codes which are essentially xor - operation based codes . for example , flat xor codes are erasure codes in which parity disks are calculated as the xor of some subset of data disks . though it is desirable to have mds ( maximal distance separable ) flat xor codes , it may not be available for all scenarios .
non - mds codes have also been used in storage systems ( e.g. , the replicated raid configurations such as raid 10 , raid 50 , and raid 60 ) .
however , we have not seen any systematic research in designing non - mds codes with flat xor operations for storage systems . in order to achieve better fault tolerance with minimal redundancy in data storage systems
, there has also been active research in xor based codes which are not necessarily flat xor codes .
for example , blaum , brady , bruck , and menon @xcite proposed the array code evenodd for tolerating two disk faults and correcting one disk errors .
blaum , bruck , and vardy @xcite and huang @xcite have extended the construction of evenodd code to general codes for tolerating three disk faults .
other non - flat xor based codes include ( but are not limited to ) @xmath1 $ ] chain code , simple product code ( spc @xcite ) , row - diagonal parity ( rdp @xcite ) , and x - code @xcite .
the techniques that we have discussed above have been originally designed for data storage on disk drives and in storage area networks .
it may not be directly applicable to distributed storage services such as the storage cloud .
there have been many researches addressing the data storage reliability issues in distributed environments .
for example , weatherspoon and kubiatowicz @xcite have compared erasure coding based solutions and replication based solutions for reliable distributed data storage systems . based on the seminal work of low - density parity - check ( ldpc ) codes by gallager @xcite , several important techniques ( see , e.g. , luby , mitzenmacher , shokrollahi , spielman , and stemann @xcite )
have been developed for networked and communication systems such as digital fountain for content streaming .
there have also been extensive interests in applying digital fountain techniques ( such as lt codes ) to distributed storage systems ( see , e.g. , plank and thomason @xcite and cao , yu , yang , lou , and hou @xcite ) . however , it is not clear whether these applications of lt codes to distributed storage systems have advantages over other techniques that have been extensively adopted in disk drives and storage area networks such as flat xor codes or array xor codes .
plank and thomason s experimental results @xcite show that ldpc code performs well only asymptotically when the number of data packages increases and it has the worst performance for small number of data fragments ( e.g. , less than @xmath0 ) .
cao , yu , yang , lou , and hou @xcite proposed to use exhaustive search approach to find a deterministic lt code that could be used to decode the original data content correctly in distributed storage systems .
however , by plank and thomason s experimental results , it is not clear whether the exhaustive search approach will work efficiently or even correctly . in this paper , we carry out a theoretical analysis on the feasibility and performance issues for applying lt codes to distributed storage systems . by employing the underlying ideas of efficient belief propagation ( bp ) decoding process in lt codes ( @xcite ) , we introduce two classes of codes called flat bp - xor codes and array bp - xor codes .
the bp - xor codes and array bp - xor codes can be considered as a deterministic version of lt codes though flat bp - xor codes are different from lt codes .
edge - colored graph models were introduced by wang and desmedt in @xcite to model homogeneous faults in networks .
we will show the equivalence between the edge - colored graph model and degree - one - and - two encoding symbol based array bp - xor codes . using this equivalence result , we are able to design general array bp - xor codes using graph based results .
similarly , based on this equivalence result , we are able to get new results for edge - colored graph models using results from array bp - xor codes .
we have implemented an online software package for users to generate array bp - xor code with their own specification and to verify the validity of their array bp - xor codes ( see @xcite ) .
the structure of this paper is as follows . in section [ reviewtech ] ,
we briefly review several coding techniques proposed for distributed storage systems .
sections [ challengesec ] presents the challenges in applying lt codes to distributed storage systems .
section [ ltcodesec ] introduces flat bp - xor codes for distributed storage systems and investigates the necessary and sufficient bounds for the existence of the codes .
section [ arraybpxorsec ] introduces array bp - xor codes for distributed storage systems and establish the equivalence between edge - colored graph models and array bp - xor codes with degree one and two encoding symbols .
section [ arraybpxorexamplesec ] presents constructions of array bp - xor codes from graph based results ( e.g. , perfect one factorization of complete graphs ) , and section [ comparisonsec ] proves a theorem on the limitation of array bp - xor codes using only degree one and two encoding symbols .
in the seminal paper @xcite , rabin proposed the information dispersal algorithm ( ida ) to code a data file into @xmath2 pieces that will be stored among @xmath2 servers such that the recovery of the information is possible when there are at most @xmath3 failed servers ( inactive but not malicious byzantine style servers ) .
rabin s scheme is essentially a kind of reed - solomon codes and needs relatively expensive finite field operations for encoding and decoding .
krawczyk @xcite extended rabin s ida scheme to address the byzantine style malicious servers which may intentionally modify their pieces of the information .
krawczyk called his scheme as secure information dispersal algorithm ( sida ) .
there have been extensive interests in applying random linear coding techniques to distributed storage systems .
for example , dimakis , ramchandran , wu , and suh @xcite and dimakis , godfrey , wu , wainwright , and ramchandran @xcite used information flow graphs and random linear coding to achieve information theoretic minimum functional repair bandwidth @xmath4 . in another word , if we divide the file @xmath5 into @xmath6 pieces , store the encoded fragments in @xmath2 storage servers , and if one of these server fails , then a new comer storage server could functionally repair the system only if it could communicate @xmath7 bits from each of the @xmath8 surviving storage servers . inspired by network coding , acedanski , medard , and koetter @xcite proposed to use random linear coding for distributed networked storage with one centralized server and multiple storage servers .
dimakis , prabhakaran , and ramchandran @xcite considered the problem from a different approach : there are @xmath2 storage servers and many distributed data sources ( that is , data are not from a central location and there is no centralized server ) .
each data source node picks one out of the @xmath2 storage servers randomly , pre - routes its packet and repeat @xmath9 times .
each storage server multiplies what it receives with coefficients selected uniformly and independently from @xmath10 and stores the results together with the coefficients .
luby @xcite pointed out that one of the potential applications of lt codes is distributed data storage systems .
several authors have continued these ideas with fruitful outcomes .
for example , plank and thomason @xcite have considered the practical implementations of ldpc codes for peer - to - peer and distributed storage systems and several experimental results have been reported . in particular , the experimental results in @xcite show that ldpc _
codes display their worst performance for @xmath11 " _ where @xmath2 is the number of data fragments .
furthermore , their experiments show that _ generating good instances of the codes is a black art ... there is an opportunity for theoretical research on codes for small @xmath2 to have a very wide - reaching impact _ " .
recently , cao , yu , yang , lou , and hou @xcite ) proposed a lt code based secure cloud storage service ( ltcs ) . in the ltcs scheme , a data file @xmath5 is split into @xmath12 packets , each of which is @xmath13 bits .
lt coding process is then used to generate @xmath14 encoded packets , where @xmath15 .
these encoded packets are divided into @xmath2 groups and each of the @xmath2 storage servers receives @xmath16 packets .
one of the basic requirements from the paper @xcite is that the original @xmath12 data packets should always be recoverable from any @xmath6 healthy servers . for lt codes
, there is a small probability that one may not be able to reconstruct the original data packets from the @xmath6 servers . in order to address this challenge
, the authors in @xcite recommended an exhaustive search method to divide the encoded symbols into @xmath2 groups and check the decodability for each @xmath6 combinations of groups .
the process continues until one finds out one valid lt coding approach .
this approach is not efficient when @xmath6 and @xmath2 are relatively larger .
furthermore , there is no guarantee that the exhaustive search method will end with a valid lt code .
the authors did not give any analysis on how efficient this approach could be or any proof whether that is feasible .
for the case of @xmath17 , the coding scheme in @xcite is essentially a flat xor code that requires a belief propagation ( bp ) decoder ( @xcite ) .
the following example shows that the exhaustive search may not succeed for some cases with @xmath17 .
[ alpha1ex ] for @xmath18 and @xmath19 , the original data is divided into three fragments @xmath20 and coding symbols are stored in @xmath21 servers ( @xmath22 ) such that the original data could be recovered from any @xmath23 servers . in order for the belief propagation ( bp ) decoder to work , we must start from an original copy of @xmath24 or @xmath25 or @xmath26 . since there could be two erasure faulty servers , three servers have to store original copies of the data fragments . without loss of generality , assume that @xmath27 , and @xmath28 store @xmath29 , and @xmath26 respectively .
again , since there could be two erasure servers , each data fragment needs to be stored in at least three servers . thus both @xmath30 and @xmath31 need to store @xmath32 .
now if both @xmath33 and @xmath34 are faulty , neither @xmath24 nor @xmath25 could be recovered .
@xmath35 for the case of @xmath36 , the coding scheme in @xcite is a kind of array xor codes ( not flat xor codes ) .
in this case , the robust soliton distribution will be used to generate the @xmath14 encoding symbols . in order for the analysis and bounds of the lt code to work ,
the numbers @xmath37 , and @xmath12 in @xcite have to be sufficiently large ( this has been confirmed by the experiments in @xcite ) . for smaller values
, these bounds may not work and the exhaustive search methods in @xcite may never end with a successful code .
however , for large enough @xmath37 , and @xmath12 , the exhaustive search method will be inefficient and may be infeasible .
the experiment results from @xcite and the potential challenges in the scheme @xcite show that it is necessary and important to systematically study the encoding symbol generation problems for applying lt code to distributed storage systems . in the following sections
, we will show what we could achieve and what we could not achieve with lt codes when applied to distributed storage systems .
in this section , we introduce a class of codes called flat bp - xor codes . in short ,
flat bp - xor codes are flat xor codes that could be decoded with the belief propagation ( bp ) algorithm for erasure codes .
the bp algorithm for binary symmetric channels is present in gallager @xcite and is also used in artificial intelligence community @xcite . in our paper , we use the bp algorithm for binary erasure channels ( see @xcite ) .
let @xmath38 be the message symbol set .
the length @xmath39 could be any number and it does not have impact on the coding .
an @xmath40 $ ] flat bp - xor code is a binary linear code determined by a @xmath41 zero - one valued generator matrix @xmath42 such that for a given message vector @xmath43 , the corresponding code @xmath44 is computed as @xmath45 where the addition of two strings in @xmath46 is defined as the xor on bits .
furthermore , a flat @xmath40 $ ] bp - xor code requires that if at most @xmath47 components in @xmath48 are missing , then @xmath49 could be recovered from the remaining components of @xmath48 with the belief propagation algorithm , it is easy to see that each flat bp - xor code is a flat xor code , but the other direction may not hold . we can consider flat bp - xor codes as one kind of applications of lt codes to reliable storage system design with deterministic decoding .
example [ alpha1ex ] shows that flat bp - xor version of the lt code may not be applicable to threshold based distributed storage systems as proposed in @xcite . in the following , we first mention the folklore fact to support our arguments . [ singluarlemma ]
let @xmath50 , @xmath51 , and @xmath52 .
then there is no @xmath40 $ ] bp - xor code .
the fact could be easily proved by the following observation : let @xmath53 $ ] be an @xmath54 parity check matrix .
if every @xmath55 columns in the matrix @xmath56 $ ] are linearly independent , then @xmath57 , where @xmath58 is the hamming weight .
thus for @xmath50 , there is neither binary linear @xmath40 $ ] code nor @xmath40 $ ] bp - xor code .
fact [ singluarlemma ] shows the impossibility of designing flat @xmath40 $ ] bp - xor codes for @xmath50 and @xmath52 . since flat bp - xor codes are extremely efficient for encoding and decoding in practice , we are also interested in flat bp - xor codes that are not mds ( maximal distance separable ) . in the following we show theoretical bounds designing flat bp - xor codes for distributed data storage systems . for an mds
@xmath40 $ ] code with @xmath52 , we can tolerate @xmath47 erasure faults .
the question that we are interested in is : for given @xmath50 , what is best distance @xmath8 we could achieve for a flat @xmath40 $ ] bp - xor code ?
fact [ singluarlemma ] shows that @xmath8 must be less than @xmath59 .
* tolerating one erasure fault * : let @xmath60 .
the generator matrix @xmath61 $ ] corresponds to the mds flat @xmath62 $ ] bp - xor code that could tolerate one erasure fault .
* tolerating two erasure faults * : fact [ singluarlemma ] shows that two parity check servers are not sufficient to tolerate two erasure faults for flat bp - xor codes . in order to tolerate two erasures
, we have to consider codes with @xmath63 . for @xmath64 ,
the following generator matrices show the existence of flat @xmath65 $ ] , @xmath66 $ ] , and @xmath67 $ ] bp - xor codes for tolerating two erasure faults .
\left [ \begin{array}{c|c } i_3 & \begin{array}{ccc } 0&1&1\\ 1&0&1\\ 1&1&1 \end{array } \end{array}\right ] \left [ \begin{array}{c|c } i_4 & \begin{array}{ccc } 0&1&1\\ 1&0&1\\ 1&1&0\\ 1&1&1 \end{array } \end{array}\right]\ ] ] indeed , the above three codes are the only flat @xmath69 $ ] bp - xor codes tolerating two erasure faults with three redundancy columns .
[ 2erasurefaults ] for @xmath63 and @xmath70 , there exists a flat @xmath71 $ ] bp - xor code if and only if @xmath72 . _
proof_. let @xmath73 $ ] be an @xmath54 parity check matrix .
the code determined by @xmath74 has minimum distance @xmath23 if and only if every @xmath75 columns in @xmath74 are linearly independent .
this implies that @xmath74 is the parity check matrix of a flat @xmath71 $ ] bp - xor code if and only if for every @xmath76 , we have @xmath77 where @xmath58 is the hamming weight . by the fact that @xmath78 it follows that there exists a flat @xmath71$]bp - xor code if and only if @xmath72 .
@xmath35 * note * : it should be noted that the codes we have constructed in theorem [ 2erasurefaults ] is the well known hamming code when @xmath79 . for @xmath80
, it is a truncated version of the hamming code . by theorem [ 2erasurefaults ]
, there is no flat @xmath81 $ ] bp - xor code for @xmath82 .
table [ table1 ] lists the required redundancy for tolerating two erasure faults when the value of @xmath6 changes .
.redundancy for flat bp - xor @xmath71 $ ] codes [ cols="^,^,^",options="header " , ] the values in table [ necessn ] show that for degree one and two encoding symbols based array bp - xor codes , if we want to recover the information symbols from more than three columns ( i.e. , @xmath83 ) of encoding symbols , then we could only have one column redundancy for @xmath84 . combining theorem [ necessaryarraybpxor ] and values in table [ necessn ]
, we get the following results for edge - colored graphs .
[ edgetheorem ] for a color set @xmath85 with @xmath86 , if we want to design an edge - colored graph @xmath87 ( or a network with more than @xmath88 kinds of homogeneous devices ) with minimum cost , then the edge - colored graph is robust against at most one color failures ( or one brand of homogeneous devices failures ) .
_ based on the results in theorems [ graphtobpxorthm ] , [ necessaryarraybpxor ] and values in table [ necessn ] , we can have the following conclusion : given integers @xmath89 , @xmath90 and @xmath2 , an edge - colored graph @xmath87 with @xmath91 , @xmath92 , and @xmath93 , @xmath87 is @xmath94-color connected only @xmath95 .
thus the theorem follows .
based on the bp ( belief propagation ) decoding process and the edge - colored graph model @xcite , we introduced flat bp - xor codes and array bp - xor codes .
we have established the equivalence between edge - colored graphs and degree one and two based array bp - xor codes . in particular , we used results in array bp - xor codes to get new results in edge - colored graphs . for array bp - xor codes with higher degree encoding symbols , we do not have general results yet .
it would be interesting to have a compelete characterization of the existence and bounds for array bp - xor codes with higher degree encoding symbols .
these characterizations may be used to design more efficient lt codes or digital fountain techniques .
we have implemented an online software package for users to generate array bp - xor codes with their own specification and to verify the validity of their array bp - xor codes ( see @xcite ) .
i would like to thank duan qi for some discussion on hamming code and theorem [ 2erasurefaults ] and thank prof .
doug stinson and yvo desmedt , for some discussions on edge - colored graphs , hamiltonian circuit , and factorization of complete graphs . | lt codes and digital fountain techniques have received significant attention from both academics and industry in the past few years .
there have also been extensive interests in applying lt code techniques to distributed storage systems such as cloud data storage in recent years .
however , plank and thomason s experimental results show that ldpc code performs well only asymptotically when the number of data fragments increases and it has the worst performance for small number of data fragments ( e.g. , less than @xmath0 ) . in their infocom 2012 paper , cao ,
yu , yang , lou , and hou proposed to use exhaustive search approach to find a deterministic lt code that could be used to decode the original data content correctly in distributed storage systems .
however , by plank and thomason s experimental results , it is not clear whether the exhaustive search approach will work efficiently or even correctly .
this paper carries out the theoretical analysis on the feasibility and performance issues for applying lt codes to distributed storage systems . by employing the underlying ideas of efficient belief propagation ( bp ) decoding process in lt codes ,
this paper introduces two classes of codes called flat bp - xor codes and array bp - xor codes ( which can be considered as a deterministic version of lt codes ) .
we will show the equivalence between the edge - colored graph model and degree - one - and - two encoding symbols based array bp - xor codes . using this equivalence result , we are able to design general array bp - xor codes using graph based results . similarly , based on this equivalence result , we are able to get new results for edge - colored graph models using results from array bp - xor codes . |
systems interacting with a thermal reservoir are frequently used in descriptions of a variety of physical phenomena . for such systems ,
the reservoir plays roles not only to keep systems in an equilibrium state , but also to make systems relaxed to an equilibrium state , and even to cause nonequilibrium steady states by interactions with different reservoirs .
depending on the type of reservoirs , various types of thermodynamic quantities are kept to be fixed in the systems , for example , a temperature ( by the heat reservoir ) , a chemical potential ( by the particle reservoir ) and a pressure .
because of the importance of interactions with a thermal reservoir in the foundation and applications of thermodynamics and statistical mechanics , many effective dynamical approaches to interactions with a thermal reservoir have been proposed , such as the langevin or the fokker - planck equations @xcite , hamiltonian dynamics with a linear coupling with infinite number of harmonic oscillators ( e.g. the caldeira - leggett model ) @xcite , the deterministic thermostat dynamics ( e.g. the nose - hoover thermostat and the isokinetic thermostat ) @xcite , the anderson thermostat @xcite and the local monte carlo method @xcite .
many efforts have been done to check justifications of these approaches as thermostats , via calculations of various statistical - mechanical properties like the ergodicity and momentum distributions , leading to their improvements @xcite .
the stochastic boundary approach is a dynamical method to take into account of effects of interactions with a thermal reservoir as boundary conditions involving stochastic processes . in this approach , for example
, interactions of a system with a thermal reservoir on a boundary @xmath0 are described by the boundary condition on @xmath0 in which the momentum distribution of particles coming from the boundary @xmath0 is given by the equilibrium distribution corresponding to the temperature of the heat reservoir @xcite .
systems with such stochastic boundary conditions as a coupling with a single heat reservoir relaxes to an equilibrium state with the temperature of the reservoir , and dissipations in the systems occur on the boundary @xmath0 . in systems with the stochastic boundary conditions , some statistical - mechanical features , such as consistencies with the kinetic theory @xcite , the local equilibrium hypothesis @xcite , the specific heat @xcite and spatial energy distributions @xcite , have been investigated . in the sense of using boundary conditions to describe coupling with a thermal reservoir , the methods using the master equation @xcite , numerical evaluations of boundary conditions @xcite or quantum scattering techniques @xcite , etc .
, are also regarded as stochastic boundary approaches . in general , these methods introduce interactions with a thermal reservoir without explicit degrees of freedom of particles inside the reservoir , leading to a simple and effective treatment of interactions with the reservoir . this approach can be used to describe nonequilibrium properties and phenomena , such as onsager s reciprocal relations @xcite , heat currents @xcite , electric currents @xcite and thermoelectric effects @xcite .
the principal aim of this paper is to discuss the stochastic boundary conditions for many - particle systems coupling to a particle reservoir with particle - particle interactions . in order to describe effects of coupling to a particle reservoir
, we introduce the particle injection rate to the system from the reservoir via the boundary @xmath0 in terms of the pressure of the reservoir which specifies a momentum transport via @xmath0 , while the momentum distribution of injected particles on @xmath0 is given by the equilibrium distribution corresponding to the temperature of the reservoir .
since the pressure and the momentum distribution of the injected particles on @xmath0 are given by the temperature @xmath1 and chemical potential @xmath2 of the particle reservoir , our particle - injecting stochastic boundary ( pisb ) conditions are specified by these thermodynamic quantities @xmath1 and @xmath2 as external parameters . as applications of this formulation to equilibrium many - particle systems , we consider ideal gases and many - hard - disk systems .
because of a simplicity of hard - core interactions , hard - disk ( or hard - sphere ) systems , as typical examples of many - particle systems with particle - particle interactions , have been widely used to investigate statistical - mechanical properties by numerical methods @xcite and analytical methods @xcite . since equilibrium
statistical - mechanical quantities not only of ideal gases but also of many - hard - disk systems in low - density cases can be calculated analytically based on the grand - canonical distribution , we use these systems to show that the systems with the pisb conditions can reproduce correctly equilibrium properties of the systems coupled to a particle reservoir .
furthermore , we apply our pisb conditions to particle - escape phenomena @xcite , as a typical example of nonequilibrium phenomena .
we consider the systems in which hard disks escaping from a particle reservoir into one end of a tube leave from another end of the tube , and produce a steady particle current inside the tube after a long time . by using the pisb approach , we analyze statistical - mechanical properties of hard disks inside the tube , such as a dependence of the current with respect to the chemical potential of the reservoir , deviations from local equilibrium behaviors , the particle density and the local pressure of the system . the outline of this paper is as follows . in sec .
[ stochasticlboundaryconditions ] , pisb conditions for coupling to a particle reservoir are introduced in terms of the properties of particles injected from the reservoir . in sec .
[ equilibriumpropertiesharddisksystems ] , the pisb approach are checked by comparing various equilibrium statistical - mechanical quantities calculated by using the pisb conditions in equilibrium ideal gases and hard - disk systems , with those calculated analytically based on their corresponding grand - canonical distributions . in sec .
[ applicationsparticlecurrents ] , we apply the pisb conditions to nonequilibrium systems which consist of hard disks with a steady particle current inside a tube connected to a particle reservoir . finally , we give conclusions and remarks in sec .
[ conclusionsremarks ] .
we consider classical particle systems coupled to a particle reservoir with a temperature @xmath1 and a chemical potential @xmath2 . here
, the coupling of a system to a particle reservoir allows particles to come from and to go to the reservoir via the boundary @xmath0 between the system and the reservoir . for simplicity
, we assume that the boundary @xmath0 is flat with a constant unit vector @xmath3 on @xmath0 , in which @xmath3 is perpendicular to @xmath0 and is directed toward the system . since particles inside the reservoir are in an equilibrium state with the temperature @xmath1 , the momentum probability density function @xmath4 of injected particles from the reservoir on the boundary @xmath0 is given by @xcite @xmath5 with boltzmann constant @xmath6 , the mass @xmath7 of particles and the spatial dimension @xmath8 of the system . here , since @xmath9 is the momentum of a particle injected to the system , the inequality @xmath10 must be satisfied , i.e. @xmath11 , while other components of the momentum @xmath9 except for @xmath12 , if any , can take any real value in @xmath13 .
equation ( [ injecmomendistr1 ] ) comes from a count of the number of particles reaching the boundary @xmath0 par unit time from the particle reservoir whose momentum distribution function is given by the maxwell momentum distribution function @xcite .
it may also be noted that the momentum probability density function @xmath4 of injected particles on the boundary @xmath0 is not the maxwell momentum distribution function itself @xcite . in order to describe how particles are injected into the system from the particle reservoir , we also have to specify how frequently such injected particles reach the boundary @xmath0 between the system and the reservoir .
we assume that successive injections of particles are uncorrelated , and the probability density function @xmath14 of time intervals @xmath15 ( @xmath16 ) between two successive injections of particles via the boundary @xmath0 is of the form @xmath17 with a constant frequency @xmath18 as a positive coupling strength between the system and the particle reservoir
. the exponential probability density function like in eq .
( [ timeinterdistr1 ] ) are also discussed in dynamics with coupling with a thermal reservoir in refs .
@xcite . by using eqs .
( [ injecmomendistr1 ] ) and ( [ timeinterdistr1 ] ) for systems with the spatial dimension @xmath19 , we estimate the average momentum injection @xmath20 par unit area of the boundary @xmath0 with the area @xmath21 par unit time as @xmath22 with @xmath23 and @xmath24 . by introducing the average outgoing momentum @xmath25 from the system par unit area of the boundary @xmath0 par unit time ,
the pressure @xmath26 on the boundary @xmath0 is given by the magnitude of the total momentum transfer @xmath27 , i.e. , @xmath28 .
using the fact that at equilibrium states the average outgoing momentum @xmath25 should be given by @xmath29 as a balance between the injecting momenta and outgoing momenta via the boundary @xmath0 , the pressure @xmath30 at an equilibrium state should be given by @xmath31 equations ( [ momentrans1 ] ) and ( [ press1 ] ) lead to @xmath32 as the relation among the system - reservoir coupling strength @xmath18 , the equilibrium pressure @xmath33 , and the temperature @xmath1 .
here , we note that the equilibrium pressure @xmath33 can be represented as the pressure of the particle reservoir as a function of the temperature @xmath1 and the chemical potential @xmath2 : @xmath34 , based on the gibbs - duhem relation , so eq . ( [ nupress1 ] )
is regarded as an expression of the coupling strength @xmath18 as a function of @xmath1 and @xmath2 .
as the first application of the pisb conditions discussed in sec . [ stochasticlboundaryconditions ] , we consider two - dimensional ideal gases without any potential energy inside the systems .
this type of systems is chosen , not only because of its simplicity , but also to show that the thermalization can occur by the pisb conditions themselves rather than by complex ( chaotic ) dynamics caused by particle - particle interactions , etc . by the equilibrium statistical mechanics with the grand - canonical distribution ,
the grand potential @xmath35 of the two - dimensional ideal gases in an equilibrium state is given by @xmath36 with the spatial area @xmath37 of the system , the inverse temperature @xmath38 and planck s constant @xmath39 .
( a brief derivation of eq .
( [ grandpotenideal1 ] ) from the corresponding grand canonical distribution is shown in appendix [ idelagasgrand ] . ) by the grand potential ( [ grandpotenideal1 ] ) we obtain the pressure @xmath40 and the average @xmath41 of the number of particles as @xmath42 respectively . by eqs .
( [ pressideal1 ] ) and ( [ partinumbeideal1 ] ) we obtain the equation of state @xmath43 for the ideal gases . the grand potential @xmath44 as a function of @xmath1 , @xmath37 and @xmath2 also leads to the equilibrium generation function @xmath45 of the number @xmath46 of particles as a function of @xmath47 , which is defined by the equilibrium average of @xmath48 , as @xmath49\right\ } .
\label{generpartinumbe2}\ ] ] as shown in appendix [ analyticalequilibriumproperties ] .
inserting eq .
( [ grandpotenideal1 ] ) into eq .
( [ generpartinumbe2 ] ) we obtain @xmath50 with @xmath51 defined by @xmath52 the generation function ( [ generpartiideal2 ] ) leads to the probability distribution @xmath53 of the number @xmath46 of particles as the poisson distribution @xmath54 for the two - dimensional ideal gases .
( another derivation of eq .
( [ partinumbedistrideal1 ] ) is shown in appendix [ idelagasgrand ] . ) from this distribution we can also derive the average @xmath55 of the number @xmath46 of particles , i.e. , eq .
( [ partinumbeideal1 ] ) . by inserting eq .
( [ pressideal1 ] ) into eq .
( [ nupress1 ] ) we obtain the system - reservoir coupling strength @xmath18 for the ideal gases as @xmath56 concretely .
particle injections from a particle reservoir in cases without particle - particle interactions are also discussed in refs .
@xcite . in order to apply the pisb conditions to the ideal gases concretely
, we further assume that the two - dimensional ideal gases consist of point particles inside a square area with the side length @xmath57 , so @xmath58 and @xmath59 .
we choose a single side of this square as the boundary @xmath0 with the length @xmath57 , and its other three sides are simply hard walls with which particles collide elastically .
particles are injected into the square area via the boundary @xmath0 with the probability density function ( [ timeinterdistr1 ] ) of the time intervals @xmath15 between two successive injections of particles with the frequency ( [ nupress2 ] ) , and with the momentum distribution ( [ injecmomendistr1 ] ) .
positions of such injected particles are chosen randomly and uniformly on the boundary @xmath0 . any particle which reaches the boundary @xmath0 from the square area
is removed immediately from the system . here and hereafter , for two - dimensional models with a flat boundary @xmath0 we take the @xmath60-axis as that perpendicular to the boundary @xmath0 , and the @xmath61-axis along @xmath0 . for such square systems consisting of point particles , it would be meaningful to show that the relation ( [ nupress2 ] ) among the coupling strength @xmath18 , the temperature @xmath1 and the chemical potential @xmath2 can also be derived from another argument .
we note that for the square systems consisting of point particles the average @xmath62 of the time interval for a particle injected from the reservoir at a point of the boundary @xmath0 to return back to @xmath0 can be estimated by the average of @xmath63 , i.e. , @xmath64 f({\boldsymbol{p } } ) = l \sqrt{2\pi m\beta}$ ] . with this time interval @xmath62 , the average @xmath65 of the number of particles inside the square systems
could be estimated as the time interval @xmath62 divided by the average time interval @xmath66 between two successive injections of particles on @xmath0 , namely @xmath67 by eqs .
( [ partinumbeideal1 ] ) , ( [ partinumbeideal2 ] ) and @xmath68 we obtain eq .
( [ nupress2 ] ) again .
now , we discuss results of numerical calculations of equilibrium quantities of the ideal gases with the pisb conditions , and compare them with the corresponding analytical expressions given in sec .
[ equilibriumpropertiesidealgases ] . for such numerical calculations , we chose the parameter values as @xmath69 , @xmath70 , @xmath71 and @xmath72 , and used the data over the time interval @xmath73 to calculate various time - average quantities and distributions .
we took this time interval @xmath74 after calculations of particle orbits at an early stage , so that in the time - interval @xmath74 these quantities and distributions can be regarded to be stationary in time , supposing that the systems would be in equilibrium states . and @xmath75 of the components @xmath76 and @xmath77 of the particle momentum @xmath78 , respectively , in ideal gases with the pisb conditions , for the chemical potentials @xmath79 , @xmath80 and @xmath81 .
all these probability density functions of the momentum components are almost indistinguishable with each other .
the black thin line is the corresponding maxwell momentum distribution . here , and in all figures hereafter , we use dimensionless units with @xmath69 , @xmath70 and @xmath71 . ] in fig .
[ fig1momendistrideal ] we show the probability density functions @xmath82 and @xmath75 of the components @xmath76 and @xmath77 of the particle momentum @xmath78 , respectively , for the ideal gases with the pisb conditions , for the chemical potentials @xmath79 , @xmath80 and @xmath81 .
( here , the steps in these graphs of @xmath82 and @xmath75 come from the momentum width in which probabilities of the momentum components are calculated discretely .
we use this type of presentations for momentum distributions in this paper . ) the black thin line in fig .
[ fig1momendistrideal ] is the corresponding maxwell momentum distribution . in this figure , all these probability density functions of the momentum components are almost indistinguishable with each other , and the probability density functions @xmath82 and @xmath75 almost coincide with the maxwell momentum distribution @xmath83 which is derived from the corresponding grand canonical distribution and is independent of @xmath2 . concerning the fact that the maxwell momentum distribution ( [ maxwedistr1 ] ) for particles inside the system is generated from the momentum distribution ( [ injecmomendistr1 ] ) for injected particles on the boundary @xmath0 , one may notice that values of the probability density of the momentum component @xmath76 of particles injected from the reservoir on the boundary @xmath0 , given from eq .
( [ injecmomendistr1 ] ) , are small for small absolute values of @xmath76 , but injected particles with such a small absolute value of @xmath76 stay for a long time inside the square area , so that the probability density @xmath82 of the momentum component @xmath76 of particles is large for small absolute values of @xmath76 .
since the number of particles injected from the reservoir on the boundary @xmath0 with small absolute values of @xmath76 are small relatively , it could take a long time for the maxwell momentum distribution ( [ maxwedistr1 ] ) for @xmath82 to be realized for particles inside the square area , for some initial conditions such as the one with no particle initially . of the number @xmath46 of particles in ideal gases with the pisb conditions , for the chemical potentials @xmath79 ( the red solid line ) , @xmath80 ( the blue dashed line ) and @xmath81 ( the green dotted line ) .
the poisson distribution ( [ partinumbedistrideal1 ] ) is also shown as the black thin line in each chemical potential case , which is almost indistinguishable from the corresponding numerical results . ] in fig .
[ fig2numbedistrideal ] we show the probabilities @xmath53 of the numbers @xmath46 of particles in the systems with the pisb conditions for the chemical potentials @xmath79 ( the red solid line ) , @xmath80 ( the blue dashed line ) and @xmath81 ( the green dotted line ) .
the corresponding graphs of eq .
( [ partinumbedistrideal1 ] ) , given from the grand canonical distribution of the ideal gases , are also shown in this figure as the black thin lines , which show an excellent agreement with the ones calculated numerically by using the pisb approach .
this also suggests that eq .
( [ partinumbeideal1 ] ) derived from eq .
( [ partinumbedistrideal1 ] ) is satisfied for the ideal gases with the pisb conditions . of ideal gases with the pisb conditions as a function of the chemical potential @xmath2 ( the red circles ) .
the black line is the pressure ( [ pressideal1 ] ) based on the corresponding grand canonical distribution . ]
the pressure of the ideal gases can be calculated numerically as the absolute value of the time - average impulses which particles give to a wall by their collisions par unit time and par unit length of the wall .
as such a wall for actual numerical calculations of the pressure , we chose a wall perpendicular to the boundary @xmath0 . this wall was chosen to analyze the pressure in the direction perpendicular to the one for the pressure from which we introduced the system - reservoir coupling strength @xmath18 as eq .
( [ nupress1 ] ) . in fig .
[ fig3pressideal ] we show such a pressure @xmath26 of the ideal gases with the pisb conditions as a function of the chemical potential @xmath2 ( the red circles ) . here
, the pressure ( [ pressideal1 ] ) calculated analytically by the corresponding grand - canonical distribution is also shown as the black line , and its values are almost indistinguishable from the ones calculated numerically by the pisb approach .
the results shown in figs . [
fig2numbedistrideal ] and [ fig3pressideal ] also suggest that the ideal gases in equilibrium states with the pisb conditions satisfy the equation of state @xmath43 .
now we discuss applications of the pisb conditions to systems consisting of hard disks , as an example of systems with particle - particle interactions .
for the hard - disk systems discussed in this paper , we assume that there is not a potential energy other than the one corresponding to the particle - particle interactions inside the systems . in this paper
, we use the grand potential @xmath35 of the two - dimensional equilibrium systems consisting of identical hard disks with the radius @xmath84 as @xmath85 in low - density cases , as a function of the temperature @xmath1 @xmath86 $ ] , the spatial area @xmath37 and the chemical potential @xmath2 .
note that this grand potential @xmath35 of hard - disk systems , which is proportional to @xmath37 , has an extensive property .
a derivation of eq .
( [ grandpotendisk1 ] ) and specific conditions for it are discussed in appendix [ harddiskgrand ] , based on the corresponding grand canonical distribution .
the grand potential ( [ grandpotendisk1 ] ) leads to the pressure @xmath87 , the average @xmath41 of numbers of particles , and the variance @xmath88 of numbers of particles as @xmath89 @xmath90 @xmath91 respectively . by eqs .
( [ pressdisk1 ] ) and ( [ numdisk1 ] ) we obtain the relation @xmath92 approximately , neglecting the terms including the order of @xmath93 , leading van der waal s equation of state as @xmath94 for two - dimensional imperfect gases without any attractive particle - particle interaction . by eqs .
( [ nupress1 ] ) and ( [ pressdisk1 ] ) the system - reservoir coupling strength @xmath18 for the hard - disk systems is given by @xmath95 concretely .
( [ nupress3 ] ) means that the frequency @xmath18 in the probability density function ( [ timeinterdistr1 ] ) of time intervals @xmath15 between two successive injections of particles decreases as the particle size @xmath96 increases .
now , we discuss statistical - mechanical properties of hard - disk systems with the pisb conditions .
we consider hard disks with the radius @xmath84 inside a rectangular area with the side lengths @xmath97 and @xmath98 , and one of four sides of the rectangular area with the side length @xmath98 becomes the boundary @xmath0 for disks to be injected from a particle reservoir with the temperature @xmath1 and the chemical potential @xmath2 , while the other three sides are hard walls . for this type of systems , we introduce the length @xmath21 of the boundary @xmath0 and the spatial area @xmath37 of the rectangular area as them for the centers of particles to move , so that they are given by @xmath99 and @xmath100 since the centers of particles can not be closer to their hard walls than the disk radius @xmath84 .
the centers of injected hard disks with the momentum probability density function ( [ injecmomendistr1 ] ) and the time - interval probability density function ( [ timeinterdistr1 ] ) appear randomly on the boundary @xmath0 as far as such injected hard disks are not overlapped with other hard disks inside the systems @xcite .
any hard disk whose center reaches the boundary @xmath0 from the inside of the systems is removed immediately from the system .
we investigated numerically equilibrium quantities and distributions of the hard - disk systems with @xmath101 , @xmath102 , @xmath70 , @xmath103 , @xmath104 and @xmath105 ( so @xmath106 and @xmath107 ) for various values of the chemical potential @xmath2 . for these numerical calculations we used the data over the time - interval @xmath108 in which these quantities and distributions would be stationary in time .
our numerical results show that the probability density functions @xmath82 and @xmath75 of the components @xmath76 and @xmath77 of the particle momentum @xmath78 of the hard - disk systems , respectively , satisfy the maxwell momentum distributions ( [ maxwedistr1 ] ) .
( a figure to show this point for the hard - disk systems is quite similar to fig .
[ fig1momendistrideal ] for ideal gases , so we omit to show such a figure explicitly in this paper . ) of disks ( red circles ) in hard - disk systems with the pisb conditions , as well as the graph of eq .
( [ numdisk1 ] ) ( a black solid line ) , as a function of the chemical potential @xmath2 . here , the graph of eq .
( [ partinumbeideal1 ] ) for the corresponding ideal gases is also shown by a blue dashed line .
the inset : the deviation @xmath109 ( red circles ) of the average number @xmath65 from its ideal - gas case , as well as the graph of eq .
( [ numdiskdevia1 ] ) ( a black solid line ) , as functions of @xmath2 . ] the main figure in fig .
[ fig4numbedistrdisks ] is the average number @xmath65 of disks ( red circles ) in the hard - disk systems with the pisb conditions , as well as the graph of eq .
( [ numdisk1 ] ) ( a black solid line ) , as a function of the chemical potential @xmath2 . here ,
for a comparison , we also show the graph of eq .
( [ partinumbeideal1 ] ) for the corresponding ideal gases by a blue dashed line . in low - density cases , a difference between eqs .
( [ partinumbeideal1 ] ) and ( [ numdisk1 ] ) is small because of weak effects of disk - disk collisions , so in the inset of fig .
[ fig4numbedistrdisks ] as a quantity to show such collision effects explicitly we plot the deviation @xmath109 ( red circles ) of the average number @xmath65 from the one of the corresponding ideal gas given by @xmath110 , i.e. , eq .
( [ partinumbeideal1 ] ) , as well as the graph of @xmath111 given from eq .
( [ numdisk1 ] ) ( a black solid line ) , as a function of @xmath2 .
figure [ fig4numbedistrdisks ] shows that the average number @xmath65 of disks in the hard - disk systems with the pisb conditions almost coincide with eq .
( [ numdisk1 ] ) based on the corresponding grand - canonical distribution for @xmath112 . of numbers of particles ( red circles ) as a function of the chemical potential @xmath2 in hard - disk systems with the pisb conditions . here , the graph of eq .
( [ variadisk1 ] ) ( a black solid line ) , as well as the graph of the variance of numbers of particles in the corresponding ideal - gas cases ( a blue dashed line ) , given from the corresponding grand canonical distributions , are also shown .
the inset : the deviation @xmath113 ( red circles ) and the graph of eq .
( [ variadiskdevia1 ] ) ( a black solid line ) as functions of @xmath2 . ] in the main figure of fig .
[ fig5numbevariadisks ] we show the variance @xmath114 of numbers of particles ( red circles ) as a function of the chemical potential @xmath2 in the hard - disk systems with the pisb conditions . here , the graph of eq .
( [ variadisk1 ] ) ( a black solid line ) , as well as the graph of the variance of numbers of particles in the corresponding ideal - gas cases ( a blue dashed line ) , both of which are given by the corresponding grand - canonical distributions , are also shown .
the inset of this figure is the deviation @xmath113 of the variance of numbers of particles ( red circles ) from the one of the corresponding ideal gas given by @xmath115 , as well as the graph of @xmath116 given from eq .
( [ variadisk1 ] ) ( a black solid line ) , as functions of @xmath2 .
figure [ fig5numbevariadisks ] shows that eq .
( [ variadisk1 ] ) based on the corresponding grand canonical distribution gives a good approximation of the variance @xmath114 of numbers of particles for the hard - disk systems with the pisb conditions , although it slightly underestimates the values of the quantity @xmath117 .
actually , the absolute values of all relative errors of the numerical values of @xmath114 in fig .
[ fig5numbevariadisks ] from eq .
( [ variadisk1 ] ) are less than 1 percent . of hard - disk systems with the pisb conditions ( red circles ) , and the graphs of eq .
( [ pressdisk1 ] ) ( a black solid line ) and eq .
( [ pressideal1 ] ) ( a blue dashed line ) , as functions of the chemical potential @xmath2 .
the inset : the difference @xmath118 between the pressures @xmath26 and @xmath119 ( red circles ) , and the graph of eq .
( [ pressdiskderiv1 ] ) , as functions of @xmath2 .
] we calculated the pressure @xmath26 of the hard - disk systems as the absolute value of the time - average impulses by particle collisions with a wall perpendicular to the boundary @xmath0 par unit time and par unit length of the wall . here , in order to calculate impulses par unit length of the wall , we divide total impulses by the effective side length @xmath120 for the centers of disks to move in the @xmath60-direction , not by the actual side length @xmath97 .
the main figure of fig .
[ fig6pressdisks ] is such a pressure @xmath26 ( red circles ) of the hard - disk systems with the pisb conditions , as well as eq .
( [ pressdisk1 ] ) ( a black solid line ) and eq .
( [ pressideal1 ] ) ( a blue dashed line ) given from the corresponding grand canonical distributions , as functions of the chemical potential @xmath2 .
the difference @xmath118 between the pressures @xmath26 and @xmath121 for a hard - disk system and the corresponding ideal gas , respectively , is small for the range of the chemical potential @xmath2 presented in the main figure of fig .
[ fig6pressdisks ] , so we plotted this difference @xmath122 itself with @xmath123 , i.e. eq . ( [ pressideal1 ] ) , in the inset of fig .
[ fig6pressdisks ] , for the hard - disk systems with the pisb conditions ( red circles ) , as well as the graph of @xmath124 given from eq .
( [ pressdisk1 ] ) ( a black solid line ) .
these graphs show that the pressure ( [ pressdisk1 ] ) by the corresponding grand canonical distribution is reproduced extremely well by the one of the hard - disk systems with the pisb conditions discussed in sec .
[ stochasticlboundaryconditions ] .
our results in figs .
[ fig4numbedistrdisks ] and [ fig6pressdisks ] also indicate that the equation of state of the hard - disk systems with the pisb conditions is given by eq .
( [ vanderwaals1 ] ) , i.e. van der waal s equation of state , in a good approximation , although we omit a figure to show it explicitly in this paper .
in this section , we discuss an application of the pisb approach to nonequilibrium phenomena involving particle currents . in an open - end tube with the length @xmath97 and the width @xmath98 .
the left end of the tube is connected to a particle reservoir ( a grey region ) with the temperature @xmath1 and the chemical potential @xmath2 via the boundary @xmath0 , and the right end of the tube opens .
hard disks injected from the particle reservoir can leave not only from the left end , but also from the right end after moving inside the tube . ] as simple nonequilibrium phenomena driven by a particle reservoir , we consider currents of particles escaping from a particle reservoir via a tube .
the systems , which we consider in this section , consist of hard disks with the radius @xmath84 in an open - end two - dimensional tube whose length and width are @xmath97 and @xmath98 , respectively .
an end of this tube ( the end @xmath125 ) is connected to a particle reservoir with the temperature @xmath1 and the chemical potential @xmath2 , and the comings and goings of hard disks at this end as the boundary @xmath0 are described as these in sec . [ harddiskstochasticlboundaryconditions ] with the pisb conditions .
another end of this tube ( the end @xmath126 ) opens so that wherever the center of any hard disk reaches this end its hard disk are removed immediately from the system . in this type of systems , some injected particles from the reservoir
can leave the systems from the end @xmath126 of the tube , as escaping particles from the particle reservoir via the tube , and a nonequilibrium steady state with a nonzero particle current from the end @xmath125 to the end @xmath126 is realized after a long time . a schematic illustration of the systems is given in fig .
[ fig7escapedisk ] . for later uses ,
we introduce the @xmath60-axis directing from the end @xmath125 to the end @xmath126 , and put the origin at a point in the end @xmath125 of the tube . in this section
we analyze nonequilibrium statistical - mechanical properties of the systems numerically with the parameter values of @xmath69 , @xmath70 , @xmath101 , @xmath127 , @xmath128 and @xmath103 . here
, we use a long tube with @xmath129 so that many disk - disk collisions can occur inside the tube .
disk - disk collisions allow some disks injected from the reservoir to go back to the reservoir after moving inside the tube , so they have an effect to suppress the particle current from the end @xmath125 to the end @xmath126 .
( the red circles ) of hard disks in an open tube connected to a particle reservoir with the chemical potential @xmath2 and the temperature @xmath1 , and the injected current density @xmath130 ( the black solid line ) from the reservoir , as functions of @xmath2 .
the inset : the ratio @xmath131 between the particle current density @xmath132 and the injected current density @xmath130 as a function of @xmath2 . ] in the main figure of fig .
[ fig8current ] we show the particle current density @xmath132 from the end @xmath125 to the end @xmath126 as a function of the chemical potential @xmath2 .
here , we calculated numerically the current density @xmath132 as @xmath133 with the number @xmath134 of injected disks from the particle reservoir and the number @xmath135 of disks leaving for the particle reservoir from the end @xmath125 of the tube over the time interval @xmath73 in which the systems could be regarded as being in steady states . in this figure , we also show the injected current density @xmath136 from the particle reservoir as a comparison , so that the difference @xmath137 gives the particle current density leaving for the particle reservoir from the end @xmath125 of the tube , caused by disk - disk collisions .
figure [ fig8current ] shows that not only the injected current density @xmath130 given from eq .
( [ nupress3 ] ) , but also the particle current density @xmath132 itself is a monotonically increasing function of the chemical potential @xmath2 in the region of @xmath2 presented in this figure . to discuss an effect of particle - particle interactions
we also plot the ratio @xmath131 ( as the quantity taking the value @xmath138 for the corresponding ideal - gas systems ) between the particle current density @xmath132 and the injected current density @xmath130 as a function of the chemical potential @xmath2 , in the inset of fig .
[ fig8current ] .
this inset suggests that the ratio @xmath131 is a monotonically decreasing function of @xmath2 .
this behavior of @xmath131 could be explained by the fact that an increase of the chemical potential @xmath2 makes the number of disks ( and the number of disk - disk collisions par unit time ) increase inside the tube , so that the probability for a disk injected from the particle reservoir to return back to the same reservoir also increases .
now , we proceed to discuss statistical - mechanical properties of the systems consisting of hard disks escaping from a particle reservoir via a tube .
since the systems are in nonequilibrium states , various quantities of the systems depend on spacial positions . to discuss those position - depending properties of the system , we introduce the region @xmath139 inside the tube as that satisfying the inequality @xmath140 with @xmath141 , and consider statistical - mechanical properties of hard disks in each of the evenly divided regions @xmath139 of the tube , for @xmath142 with @xmath143 .
( the red solid line ) , @xmath144 ( the blue dashed line ) and @xmath145 ( the green dotted line ) of @xmath60-components of momentum vectors of hard disks in the regions @xmath146 , @xmath147 and @xmath148 , respectively , in the model with hard disks in an open tube connected to a particle reservoir with the chemical potential @xmath2 and the temperature @xmath1 .
the corresponding maxwell momentum distribution @xmath82 with the temperature @xmath1 , and the gaussian distribution function @xmath149 with the average and the variance given from the data for @xmath144 are also shown as a black thin dashed - dotted line and the black thin dashed line , respectively . ]
we introduce the probability distribution functions @xmath150 and @xmath151 of @xmath60-components and @xmath61-components of momentum vectors of hard disks in the region @xmath139 , respectively , for @xmath152 , and discuss local momentum distributions of hard disks in the tube by these probability distribution functions in the case of @xmath153 . in fig .
[ fig9momenpxdistrinoneqil ] we show the probability density functions @xmath154 ( the red solid line ) , @xmath144 ( the blue dashed line ) and @xmath145 ( the green dotted line ) for hard disks in the open tube connected to a particle reservoir with the chemical potential @xmath2 and the temperature @xmath1 . for comparisons , we also plotted in this figure the maxwell momentum distribution @xmath82 ( the black thin dashed - dotted line ) given by eq .
( [ maxwedistr1 ] ) with the temperature @xmath1 .
we can easily see in fig .
[ fig9momenpxdistrinoneqil ] that the peak positions of @xmath150 , @xmath155 are sifted to positive values from the zero value for @xmath82 because of the escaping current of hard disks in the positive direction of the @xmath60 axis . in fig .
[ fig9momenpxdistrinoneqil ] we can also recognize asymmetric shapes in the peak region of @xmath154 and in the tail regions of @xmath145 , meaning that the probability density function @xmath154 ( @xmath145 ) in the end area @xmath146 ( @xmath148 ) of the tube is not a gaussian distribution . on the other hand ,
the probability density function @xmath144 in a middle region @xmath147 of the tube looks to be symmetric . to discuss this point further , we plotted in fig .
( [ fig9momenpxdistrinoneqil ] ) the gaussian distribution @xmath156 $ ] ( the black thin dashed line ) with the average @xmath157 and the variance @xmath158 of @xmath76 calculated numerically from the data giving the function @xmath144 . the function @xmath159 is fitted well for the function @xmath144 in fig .
[ fig9momenpxdistrinoneqil ] , implying that the probability density function @xmath144 is well approximated by a gaussian distribution .
( the red solid line ) , @xmath160 ( the blue dashed line ) and @xmath161 ( the green dotted line ) of @xmath61-components of momentum vectors of hard disks in the regions @xmath146 , @xmath147 and @xmath148 , respectively , in the model with hard disks in an open tube connected to a particle reservoir with the chemical potential @xmath2 and the temperature @xmath1 .
the corresponding maxwell momentum distribution @xmath75 with the temperature @xmath1 is shown as a black thin dashed - dotted line .
the gaussian distribution functions @xmath162 , @xmath163 and @xmath164 with the average and the variance given from the data for @xmath154 , @xmath144 and @xmath145 are also shown as a black thin solid line , the black thin dashed line and the black thin dotted line , respectively . ] in fig . [ fig10momenpydistrinonequil ]
we show the graphs of the probability density functions @xmath165 ( the red solid line ) , @xmath160 ( the blue dashed line ) and @xmath161 ( the green dotted line ) of @xmath61-components of momentum vectors of hard disks in the regions @xmath146 , @xmath147 and @xmath148 , respectively , in the model with hard disks in an open tube connected to a particle reservoir . in this figure ,
the peak positions of the probability density functions @xmath165 , @xmath160 and @xmath161 of @xmath61-components of momentum vectors of hard disks are at @xmath166 .
besides , the variances of @xmath77 in them are small for their positions far from the boundary @xmath0 with the particle reservoir with the chemical potential @xmath2 and the temperature @xmath1 .
( for a comparison , the maxwell momentum distribution @xmath75 with the temperature @xmath1 is also shown as a black thin dashed - dotted line in fig .
[ fig10momenpydistrinonequil ] .
the variances of @xmath77 in the probability density functions @xmath165 , @xmath160 and @xmath161 are smaller than that in @xmath75 . ) in fig .
[ fig10momenpydistrinonequil ] we also show the gaussian distribution functions @xmath162 , @xmath163 and @xmath164 with the average and the variance given from the data for @xmath154 , @xmath144 and @xmath145 as a black thin solid line , the black thin dashed line and the black thin dotted line , respectively .
these functions are fitted well for the corresponding probability density functions @xmath165 , @xmath160 and @xmath161 , respectively , suggesting that the probability density functions of @xmath61-components of momentum vectors of hard disks are well described by gaussian distributions locally .
( the red solid line ) and the variance @xmath167 ( the blue dashed line ) of the momentum component @xmath76 , and the average @xmath168 ( the green dotted line ) and the variance @xmath169 ( the purple dashed - dotted line ) of the momentum component @xmath77 , in the region @xmath139 for @xmath152 as functions of @xmath60 , in the model with hard disks in an open tube connected to a particle reservoir . ] now , we discuss the average @xmath170 and the variance @xmath171 ( the average @xmath168 and the variance @xmath172 ) of the momentum component @xmath76 ( the momentum component @xmath77 ) of hard disks in the region @xmath139 for @xmath152 .
figure [ fig11pxpyaveflunoneq ] is the graphs of @xmath170 ( the red solid line ) , @xmath171 ( the blue dashed line ) , @xmath168 ( the green dotted line ) and @xmath172 ( the purple dashed - dotted line ) as functions of @xmath60 .
( in this paper , any quantity @xmath173 defined in the region @xmath139 for @xmath152 is plotted as a function of @xmath60 so that it takes a constant value @xmath173 in @xmath174 for each value of @xmath175 . )
it is shown in this figure that the average @xmath170 of the momentum component @xmath76 increases as the position goes away from the boundary with the particle reservoir , while the average @xmath168 of the momentum component @xmath77 is zero at any position @xmath60 .
it may be noted that for the ideal gases consisting of point particles realized in the limit of @xmath176 of the hard - disk systems , the average @xmath177 of the momentum component @xmath76 of particles inside the tube is given by @xmath178= \sqrt{2 m k_{b}t/\pi}$ ] , which takes the value @xmath179 in the case of @xmath102 and @xmath103 .
figure [ fig11pxpyaveflunoneq ] with this fact shows that the average momentum @xmath170 in the @xmath60-direction is smaller than the value @xmath177 near the reservoir ( for a small @xmath60 ) , and it is larger than @xmath177 far from the reservoir ( for a large @xmath60 ) .
this feature suggests that averagely speaking , disks are accelerated in the @xmath60-direction by collisions with other disks injected from the particle reservoir , although the average momentum @xmath170 near the reservoir is smaller than that of the corresponding ideal gas because there would be more number of disks moving toward the boundary @xmath0 by disk - disk collisions in regions closer to @xmath0 .
the variances of the momentum components also depend on the @xmath60-component of position inside the tube .
we discuss this property by using the quantities @xmath180 and @xmath181 defined by @xmath182 and @xmath183 , for the variances of the momentum components in the @xmath60- and @xmath61-directions , respectively . here
, the quantity @xmath184 ( @xmath185 ) can be regarded as the average kinetic energy in the @xmath60-direction ( the @xmath61-direction ) in the frame moving with the average velocity of the particle current ( with the velocity @xmath186 in the region @xmath139 ) , so we call the quantity @xmath180 ( @xmath181 ) the kinetic temperature in the @xmath60-direction ( the @xmath61-direction ) in this paper . with the kinetic temperatures ,
the averages @xmath187 and @xmath188 of the kinetic energy @xmath189 and @xmath190 in the region @xmath139 are given by @xmath191 and @xmath192 , respectively .
in terms of these kinetic temperatures , fig . [ fig11pxpyaveflunoneq ]
shows that the kinetic temperature @xmath181 in the @xmath61-direction decreases as the position goes away from the boundary @xmath0 , and the kinetic temperature @xmath180 in the @xmath60-direction is also a decreasing function of @xmath60 for a large value of @xmath60 .
( it may be noted in fig .
[ fig11pxpyaveflunoneq ] that the kinetic temperature @xmath180 is rather an increasing function of @xmath60 for the regions of @xmath146 and @xmath193 in which the average @xmath170 of the momentum component @xmath76 is smaller than the value of @xmath177 . )
these kinetic temperatures , especially @xmath180 , are lower than the temperature @xmath1 of the particle reservoir , even in the region @xmath146 closest to the reservoir .
this gap between the temperatures @xmath180 and @xmath1 occurs even in the ideal - gas cases .
actually , in the ideal gases realized in the limit of @xmath176 of the hard - disk systems , the corresponding kinetic temperature @xmath194 in the @xmath60-direction are given by @xmath195 /(k_{b}m)=
[ 1-(2/\pi)]t$ ] , which is lower than the temperature @xmath1 of the reservoir . in this case
, the inequality @xmath196 indicates that the average kinetic energy by particle movements in the @xmath60-direction consists not only of the energy @xmath197 by the kinetic temperature , but also of the energy @xmath198 by the average particle current .
it is also important to notice the inequality @xmath199 shown in fig .
[ fig11pxpyaveflunoneq ] , suggesting violations of the equipartition law and the local equilibrium hypothesis in the moving frame .
this feature would imply that values of the particle density are at least not large enough ( and there are not enough numbers of particle - particle collisions par unit time ) to thermalize particles in local regions . of the number @xmath46 of particles in the region @xmath139 for @xmath152 as a function of @xmath46 and @xmath60 , in the model with hard disks in an open tube connected to a particle reservoir . ]
( the red solid line ) and the variance @xmath200 ( the blue dashed line ) of the number @xmath46 of particles in the region @xmath139 for @xmath152 as functions of @xmath60 , in the model with hard disks in an open tube connected to a particle reservoir . ] we proceed to discuss local properties related to numbers of disks in our nonequilibrium model for a particle current escaping from a particle reservoir . in fig .
[ fig12parnumdistr ] we show the graph of the probability @xmath201 of the number @xmath46 of particles in the region @xmath139 for @xmath152 as a function of @xmath46 and @xmath60 .
this figure shows that the probabilities @xmath202 and @xmath203 are increasing functions of @xmath60 , while @xmath204 are decreasing functions of @xmath60 .
the average @xmath205 and the variance @xmath200 of the number of particles calculated by the probability @xmath201 in the region @xmath139 for @xmath152 are shown in fig .
[ fig13parnumaveflu ] as functions of @xmath60 . in this figure
we see that the average and the variance of the number of particles decrease as the position goes away from the boundary @xmath0 .
moreover , the value of @xmath206 is smaller than that of @xmath205 for any region of @xmath139 in this figure , meaning that the probability @xmath201 of the number @xmath46 of particles is not a poisson distribution ( with the average equal to the variance ) locally in space . on the wall of the region @xmath139 of the tube for @xmath152 as a function of @xmath60 , in the model with hard disks in an open tube connected to a particle reservoir . ]
we calculate the pressure @xmath207 as the absolute value of the time - average impulses by particle collisions with a wall of the region @xmath139 of the tube par unit time and par unit length of the wall for @xmath152 , and show them in fig .
[ fig14pressnonequil ] as a function of @xmath60 .
it is shown in this figure that the pressure on the wall of the tube decreases as the position goes away from the boundary @xmath0 . especially , the pressure @xmath208 near the end @xmath126 of the tube can be less than half as much as the pressure @xmath209 near the end @xmath125 as the boundary @xmath0 , and this decrease ratio of the pressure @xmath207 from the left end @xmath125 to the right end @xmath126 of the tube is larger than that of the locally average number @xmath205 of particles shown in fig .
[ fig13parnumaveflu ] . in low - density cases ,
the average number density of particles moving to the negative @xmath60-direction is much smaller than that of particles moving to the opposite direction inside the tube because the right end of the tube is open , while particles inside the particle reservoir should move in these both directions with an equal probability .
this implies that the particle number density inside the tube , as well as the pressure on walls of the tube , are much smaller than those inside the particle reservoir , even in a region close to the reservoir .
for example , in the limit of @xmath210 , i.e. , in ideal - gas cases , there is no particle moving to the negative @xmath60-direction , so that the particle number density and the pressure by particles inside the tube are half as much as those inside the reservoir . on the other hand , in such ideal - gas cases , the local momentum probability distributions , the local particle - number density and the local pressure , etc .
, are independent of @xmath60 . in this sense , @xmath60-dependences of local quantities such as @xmath211 , @xmath212 , @xmath205 and @xmath207 in the cases of @xmath213 , as shown in this subsection [ statisticalmechanicalpropertiesescapingcurrents ] , appear by the particle - particle interactions .
in this paper , we discussed particle - injecting stochastic boundary ( pisb ) conditions for systems coupling to a particle reservoir with the temperature @xmath1 and the chemical potential @xmath2 . in order to describe states of particles injected from a particle reservoir
, we imposed the boundary condition that particles injected from the particle reservoir via the boundary @xmath0 have the momentum distribution function ( [ injecmomendistr1 ] ) produced by particles with the maxwell momentum distribution function ( [ maxwedistr1 ] ) with the temperature @xmath1 .
we also assumed that particle injections occur on the boundary @xmath0 stochastically so that the probability distribution function of time interval between two successive injections of particles via the boundary @xmath0 is given by the exponential function ( [ timeinterdistr1 ] ) .
furthermore , we took into account that the average momentum transfer on the boundary @xmath0 by injected particles or leaving particles via @xmath0 is given from the pressure in an equilibrium state with the temperature @xmath1 and the chemical potential @xmath2 by eqs .
( [ momentrans1 ] ) and ( [ press1 ] ) , leading to the direct connection ( [ nupress1 ] ) between the frequency of the distribution ( [ timeinterdistr1 ] ) and the pressure of the particle reservoir . based these arguments
, we constructed the pisb conditions with the external parameters @xmath1 and @xmath2 in many - particle systems . in order to check accuracy of the pisb conditions
, we applied them to two different systems whose equilibrium quantities can be calculated analytically based on grand canonical distributions : two - dimensional ideal gases and low - density hard - disk systems .
we showed that in these systems the numerical results for the momentum distribution functions @xmath82 and @xmath75 , the average number @xmath65 of particles , the pressure @xmath26 and so on by using the pisb conditions , almost coincide with those calculated analytically from the corresponding grand canonical distributions .
the well - known equilibrium equations of state , @xmath214 for ideal gases and @xmath94 for low - density hard - disk systems , were also shown numerically by using the pisb conditions .
an advantage of using stochastic boundary conditions for the systems coupled to thermal reservoirs is their applicability to nonequilibrium phenomena .
to show this feature in the pisb conditions , we applied these conditions to a nonequilibrium model with a particle current escaping from a particle reservoir in an open tube . in this model
, hard disks can leave from the tube , not only via the boundary @xmath0 located in an end of the tube but also via another end of the tube with an open boundary condition , and a steady particle current escaping from the particle reservoir to the open boundary of the tube is sustained after a long time .
we discussed statistical - mechanical properties of various quantities in this nonequilibrium model , such as the particle current density as a function of the chemical potential @xmath2 , the probability density functions of momentum components of particles , and the particle number density and the pressure as functions of the spatial position in the direction parallel to the tube .
many local quantities in this model depend on the spatial position in the tube , differently from equilibrium states , and this model also shows deviations from properties based on the local equilibrium hypothesis .
it is important to note that the chemical potential @xmath2 in the pisb conditions is introduced via the @xmath2-dependence of the pressure of the particle reservoir , as represented in eq .
( [ nupress1 ] ) . for ideal gases
the analytical form of the pressure as a function of the temperature @xmath1 and the chemical potential @xmath2 is given by eq .
( [ pressideal1 ] ) , and such a pressure for hard - disk systems is given by eq .
( [ pressdisk1 ] ) in low - density cases .
however , even in low - density hard - disk cases , there is still a problem to clarify how wide range of values of the chemical potential @xmath2 and the temperature @xmath1 the form ( [ pressideal1 ] ) of the pressure is applicable to . in the numerical calculations whose
results were shown in this paper , we restricted values of the chemical potential @xmath2 to those satisfying the inequality ( [ restrchemic1 ] ) , but it is only the restriction of @xmath2 in the derivation of eq .
( [ pressdisk1 ] ) shown in appendix [ partiinter ] .
it would also be interesting to investigate statistical - mechanical properties of other many - particle systems with particle - particle interactions , including hard - disk systems in high - density cases , with the pisb conditions expressed by their pressures .
although pressures of many - particle systems in high - density cases have already discussed by using canonical distributions @xcite , it may be noted that in order to use pressures in the pisb conditions they need to be expressed as functions of @xmath1 and @xmath2 based on grand - canonical distributions . in the pisb conditions applied to hard - disk systems , we assumed in this paper that particles are injected randomly via the boundary @xmath0 from a particle reservoir , and there is not a spatial correlation between particles entering to the systems via the boundary @xmath0 and other particles inside the system , except for the condition in which any injected disk does not overlap with other disks .
however , spatial correlations of particles by particle - particle interactions would not be negligible in the cases of high - density particles in general .
as another point , for many - particle systems with soft - core interaction potentials ( or an external potential slope ) , it has to be taken into account of additional potential effects for the spatial distribution of injected particles on the boundary @xmath0 , differently from hard - disk systems discussed in this paper .
the formulations of pisb conditions with effects of such spatial correlations and potential energies of injecting particles remain as interesting but unsettled problems .
one of the authors ( t. t. ) thanks c. b. mcrae and g. p. morriss for discussing nonequilibrium phenomena with thermal boundary conditions in hard - disk systems .
in this appendix , we discuss the analytical expressions of statistical - mechanical quantities of equilibrium systems coupled to a particle reservoir , which are used in the main text of this paper , based on grand canonical distributions .
we consider @xmath8-dimensional systems described by the hamiltonian @xmath215 for @xmath46 identical particles as a function of the phase space vector @xmath216 , where @xmath217 and @xmath218 are the position vector and the momentum vector of the @xmath46 number of particles , respectively , @xmath7 is the mass of particles , and @xmath219 is the potential energy as a function of @xmath217 . the grand partition function of the systems in an equilibrium state with the temperature @xmath1 and the chemical potential @xmath2 is represented as @xmath220 with the partition function @xmath221 of @xmath46 particles , which is given by @xmath222 for @xmath223 and @xmath224^{{\mathcal{n } } } \left(1+\frac{a_{{\mathcal{n}}}}{v_{d}^{{\mathcal{n}}}}\right ) \label{partifunct1}\ ] ] for @xmath225 , with the volume @xmath226 of the spatial area for particles to exist . here , the quantity @xmath227 is defined by @xmath228 for @xmath223 and @xmath229 \label{functa1}\ ] ] for @xmath225 , and includes effects of the potential energy @xmath230 in the grand partition function @xmath231 . by eqs .
( [ hamilnpar1 ] ) , ( [ grandparti1 ] ) and ( [ partifunct1 ] ) we obtain @xmath232^{{\mathcal{n } } } \left(1+\frac{a_{{\mathcal{n}}}}{v_{d}^{{\mathcal{n}}}}\right ) . \nonumber\\ \label{grandparti2}\ ] ] with the grand partition function @xmath231 , the grand potential @xmath35 is represented as @xmath233 , which leads to the entropy @xmath234 , the pressure @xmath235 and the average @xmath236 of the number @xmath46 of particles , based on the first law of thermodynamics @xmath237 . using the grand partition function @xmath231 ,
the grand canonical distribution is represented as @xmath238\}$ ] . for the following arguments , for any quantity @xmath239 in the systems coupled to a particle reservoir we use the notation @xmath240 for the average of the quantity @xmath239 based on this grand canonical distribution . by using the grand partition function @xmath231 or the grand potential @xmath241 as a function of @xmath1 , @xmath226 and @xmath2 , the generation function @xmath45 of the number @xmath46 of particles is represented as @xmath242\right\ } \nonumber \\ & = & \frac{\left.\xi\right|_{\mu\rightarrow \mu + \beta^{-1}\ln z } } { \xi } = \exp\left\ { -\beta \left[j\!\left(t , v_{d},\mu + \frac{\ln z}{\beta}\right ) - j(t , v_{d},\mu)\right]\right\}. \label{generpartinumbe1}\ ] ] we obtain eq .
( [ generpartinumbe2 ] ) as eq .
( [ generpartinumbe1 ] ) in the case of @xmath243 and @xmath244 . by this generation function @xmath45 , the probability distribution @xmath53 of the number @xmath46 of particles
is given from @xmath245 , because of the relation @xmath246 .
it may be noted that the probability distribution @xmath53 of the number @xmath46 of particles is also given by @xmath247\right\ } \nonumber \\ & = & \frac{\theta_{{\mathcal{n}}}}{\xi}e^{\beta\mu { \mathcal{n } } } .
\label{generpartinumbe4}\ ] ] directly from the grand canonical distribution @xcite .
we consider the case of ideal gases with no potential energy @xmath248 . in this case , we obtain @xmath249 in eq .
( [ functa1 ] ) , leading to the grand partition function ( [ grandparti2 ] ) as @xmath250 . \label{grandparti3}\ ] ] by eq .
( [ grandparti3 ] ) we obtain the grand potential ( [ grandpotenideal1 ] ) and the pressure ( [ pressideal1 ] ) in the case of @xmath243 and @xmath244 .
( see ref .
@xcite for the grand potential of ideal gases in the case of @xmath251 . ) noting that by eqs .
( [ partifunct1 ] ) and ( [ grandparti3 ] ) the partition function @xmath221 and the grand partition function @xmath231 satisfy the relations @xmath252 and @xmath253 , respectively , with @xmath254 defined by @xmath255 [ so @xmath256 for @xmath51 given in eq .
( [ lambda2d1 ] ) ] , eq .
( [ generpartinumbe4 ] ) leads to eq .
( [ partinumbedistrideal1 ] ) for the two - dimensional ideal gases .
now , we consider systems with weak two - body interactions .
we assume that the potential energy @xmath230 is represented as @xmath248 for @xmath257 and @xmath258 for @xmath259 with a function @xmath260 of a single variable @xmath60 , where @xmath261 is the @xmath8-dimensional position vector of the @xmath175-th particle for @xmath262 so @xmath263 . here , the quantity @xmath264 is the potential energy by the interaction between the @xmath175-th particle and the @xmath265-th particle , depending only on the distance @xmath266 between these two particles , for @xmath267 , @xmath262 and @xmath268 .
we also assume that particle densities of the systems are so low that any simultaneous interaction involving more than three particles is negligible . under this assumption , and by using the potential energy @xmath230 defined by @xmath248 for @xmath257 and eq .
( [ potenpartiinter1 ] ) for @xmath259 , the quantity @xmath227 defined by @xmath269 for @xmath223 and eq .
( [ functa1 ] ) for @xmath270 is given by @xmath249 for @xmath257 and is approximated by @xmath271\right]-1\right\ } \nonumber \\ & \approx & \displaystyle \frac{{\mathcal{n}}({\mathcal{n}}-1)}{2}v_{d}^{{\mathcal{n}}-2}\int d{\boldsymbol{q}}^{(j ) } \int d{\boldsymbol{q}}^{(k ) } \ ; \left\{\exp\left[-\beta u\!\left(\left|{\boldsymbol{q}}^{(j)}-{\boldsymbol{q}}^{(k)}\right|\right ) \right]-1\right\ } \nonumber \\ & \approx & \displaystyle \frac{{\mathcal{n}}({\mathcal{n}}-1)}{2}v_{d}^{{\mathcal{n}}-1}\int d{\boldsymbol{x}}^{(j , k ) } \ ; \left\{\exp\left[-\beta u\!\left(\left|{\boldsymbol{x}}^{(j , k)}\right|\right ) \right]-1\right\ } \nonumber \\ & = & \displaystyle - { \mathcal{n}}({\mathcal{n}}-1)v_{d}^{{\mathcal{n}}-1 } b \label{parama1}\ ] ] for @xmath259 , with @xmath272 defined by @xmath273 \label{interparamb1}\ ] ] as a parameter describing a magnitude of particle - particle interactions . here , the factor @xmath274 is the number of particle pairs chosen from @xmath46 particles , and we introduced @xmath275 by @xmath276 , and approximated @xmath277 as @xmath278 for @xmath279 .
an approximation similar to that used to obtain eq .
( [ parama1 ] ) is used to describe nonideal gases based on the canonical distribution @xcite . by using the approximation ( [ parama1 ] ) and @xmath254 defined by eq .
( [ lambdaddim1 ] ) , the grand partition function ( [ grandparti2 ] ) is represented as @xmath280 approximately .
we further assume that particle - particle interactions are so weak that for the quantity defined by @xmath281 the condition @xmath282 is satisfied . under the condition ( [ weakinter1 ] ) , the grand partition function ( [ grandparti4a ] )
is given approximately by @xmath283 where we used the approximation @xmath284 for any integer @xmath46 under the condition ( [ weakinter1 ] ) .
it may be noted that eq .
( [ grandparti4a ] ) can also be transformed as @xmath285 suggesting the necessary condition @xmath286 for the derivation of eq .
( [ grandparti5 ] ) because of the positivity @xmath287 of the grand partition function @xmath231 . by the grand partition function ( [ grandparti4 ] ) and eqs . ( [ lambdaddim1 ] ) and ( [ weakinter0 ] ) ,
the grand potential @xmath233 is represented by @xmath288 \nonumber \\
\label{grandpoten1}\end{aligned}\ ] ] approximately . for two - dimensional systems consisting of hard disks with the radius @xmath84
, the potential energy @xmath260 for two - disk interactions is expressed as @xmath289 for this potential with @xmath243 , the quantity ( [ interparamb1 ] ) is given by @xmath290 \nonumber \\ & = & 2 \pi r^{2 } , \label{interparambdisk1}\ ] ] so that the quantity b is twice as large as the area of the disk . inserting eqs .
( [ interparambdisk1 ] ) , @xmath243 and @xmath291 into eq .
( [ grandpoten1 ] ) we obtain the grand potential ( [ grandpotendisk1 ] ) for the hard - disk systems in low - density cases .
h. risken , _ the fokker - planck equation : methods of solution and applications _ ( springer - verlag , berlin , 1989 ) .
n. g. van kampen , _ stochastic processes in physics and chemistry _
( elsevier , amsterdam , 1992 ) .
w. t. coffey and y. p. kalmykov , _ the langevin equation : with applications to stochastic problems in physics , chemistry and electrical engineering _ ( world scientific , singapore , 2012 ) .
j. m. thijssen , _ computational physics _
( cambridge university press , cambridge , 1999 ) .
d. frenkel and b. smit , _ understanding molecular simulation : from algorithms and to applications _
( academic press , san diego , 2002 ) .
d. j. evans and g. morriss , _ statistical mechanics of nonequilibrium liquids _
( cambridge university press , cambridge , 2008 ) . m. e. tuckerman , _ statistical mechanics : theory and molecular simulation _ ( oxford university press , oxford , 2010 )
. b. leimkuhler and c. matthews , _ molecular dynamics : with deterministic and stochastic numerical methods _
( springer international publishing switzerland , cham , 2015 ) . h. c. andersen , j. chem . phys . * 72 * , 2384 ( 1980 ) .
a. papadopoulou , e. d. becker , m. lupkowski , and f. van swol , j. chem , phys .
* 98 * , 4897 ( 1993 ) .
g. s. heffelfinger and f. van swol , j. chem . phys . *
100 * , 7548 ( 1994 ) .
g. ciccotti and a. tenenbaum , j. stat . phys .
* 23 * , 767 , ( 1980 ) .
a. tenenbaum , g. ciccotti , and r. gallico , phys .
a**25 * * , 2778 ( 1982 ) .
m. mareschal and e. kestemont , phys .
rev . a**30 * * , 1158 ( 1984 ) .
m. mareschal , e. kestemont , f. baras , e. clementi , and g. nicolis , phys .
rev . a**35 * * , 3883 ( 1987 ) .
d. k. bhattacharya and g. c. lie , phys .
rev . a**43 * * , 761 ( 1991 ) .
r. tehver , f. toigo , j. koplik , and j. r. banavar , phys .
rev . e**57 * * , r17 ( 1998 ) .
s. datta , _ electronic transport in mesoscopic systems _
( cambridge university press , cambridge , 1995 ) . y. imry , _ introduction to mesoscopic physics _ ( oxford university press , new york , 1997 ) . m. lundstrom and c. jeong , _ near - equilibrium transport : fundamentals and applications ( lessons from nanoscience : a lecture note series , vol . 2 ) _ ( world scientific , singapore , 2013 ) m. p. allen and d. j. tildesley , _ computer simulation of liquids _
( oxford university press , oxford , 1987 ) .
j. m. haile , _ molecular dynamics simulation : elementary methods _
( wiley - interscience , new york , 1992 ) .
c. g. gray and k. e. gubbins , _ theory of molecular fluids , volume 1 : fundamentals _ ( oxford university press , oxford , 1984 ) .
a. mulero ( ed . ) , _ theory and simulation of hard - sphere fluids and related systems _
( springer - verlag , berlin , 2008 ) . j .-
hansen and i. r. mcdonald , _ theory of simple liquids : with applications to soft matter _
( academic press , oxford , 2013 ) .
j. r. solana , _ perturbation theories for the thermodynamic properties of fluids and solids _ ( crc press , boca raton , 2013 ) .
w. bauer and g. f. bertsch , phys .
lett . * 65 * , 2213 ( 1990 ) ; o. legrand and d. sornette , phys .
* 66 * , 2172 ( 1991 ) ; w. bauer and g. f. bertsch , phys .
66 * , 2173 ( 1991 ) . t. taniguchi and s. sawada , phys .
e * 83 * , 026208 ( 2011 ) .
e. g. altmann , j. s. e. portela , and t. tl , rev .
phys . * 85 * , 869 ( 2013 ) .
t. taniguchi , h. murata , and s. sawada , phys .
e * 90 * , 052923 ( 2014 ) . it may be noted that the position distribution of a hard disk in equilibrium states of many - hard - disk systems is not uniform in general and it has a high probability near hard walls of the systems @xcite . in this sense ,
the assumption of an uniform position - distribution of particles injected from the reservoir on the boundary @xmath0 would imply , for example , that the area of the particle reservoir is large enough so that the walls constructing the boundary of the reservoir , which are not perpendicular to the unit vector @xmath292 , are far from the boundary @xmath0
. z. zheng , g. hu , and j. zhang , phys .
rev . e**53 * * , 3246 ( 1996 ) .
w. greiner , l. neise , and h. stcker , _ thermodynamics and statistical mechanics _
( springer - verlag , new york , 1995 ) .
f. reif , _ fundamentals of statistical and thermal physics _
( mcgraw - hill , boston , 1965 ) . l. d. landau and e. m. lifshitz , _ statistical physics , part 1 : course of theoretical physics , volume 5 _ ( butterworth - heinemann , oxford , 1980 ) . | stochastic boundary conditions for interactions with a particle reservoir are discussed in many - particle systems .
we introduce the boundary conditions in which the injection rate and the momentum distribution of particles coming from a particle reservoir via a boundary are described with the pressure and the temperature of the reservoir , respectively .
this method is applied to ideal gases and hard - disk systems , showing that these systems with the stochastic boundary conditions reproduce equilibrium properties based on the corresponding grand - canonical distributions
. we also apply it to hard - disk currents escaping from a particle reservoir in a tube , and discuss their nonequilibrium properties such as a chemical potential dependence of the currents and deviations from the local equilibrium hypothesis . |
the dama @xcite and cogent @xcite dark matter detection experiments have reported events in excess of the expected background , which would be compatible with a wimp mass of a few gev . also the cdms - ii experiment @xcite has reported two events , which could be explained by a wimp mass of @xmath6 gev ( or background ) . on the other hand , exclusion limits from the xenon10 @xcite , xenon100 @xcite and cdms - si @xcite experiments set upper bounds on the spin independent detection cross sections for this mass range of a wimp , which seem incompatible with the reported hints for a signal . in any case
, it is important to know whether specific models for dark matter with a wimp mass of a few gev can produce direct detection cross sections compatible with the reported excesses , and/or whether regions in the parameter space of such models can be tested by present and future exclusion limits .
supersymmetric ( susy ) extensions of the standard model are popular , amongst others , since they predict naturally ( for unbroken r - parity , and for a neutral lightest supersymmetric particle ( lsp ) ) a candidate for dark matter , with a relic density compatible with wmap constraints @xcite . within the minimal supersymmetric extension of the standard model ( mssm )
four neutral fermions ( neutralinos @xmath7 , @xmath8 ) exist , which are composed of the bino ( superpartner of the @xmath9 gauge boson ) , the wino ( superpartner of the @xmath10 gauge boson ) and two higgsinos ( superpartners of neutral higgs bosons ) .
these states mix , and the lightest neutralino @xmath11 , which is the lightest eigenvalue of the @xmath12 mass matrix , will be the lsp ( leaving aside the possibility of a sneutrino lsp ) . often the lsp is dominantly bino - like , whose mass @xmath13 is approximately given by the soft susy breaking gaugino mass @xmath14 . assuming unification of the three gaugino masses for the bino ( @xmath15 ) , the winos ( @xmath16 ) and the gluino ( @xmath17 ) at the scale of grand unification , @xmath15 is naturally the smallest among these mass terms at the electroweak scale .
however , due to the lower bound of @xmath18 gev on @xmath16 from the lower bound on chargino masses , the assumption of unification of the three gaugino masses implies @xmath19 gev and a similar lower bound on the mass of the lsp .
the assumption of unification of the three gaugino masses can be dropped , however , in that case @xmath15 and hence the lsp mass @xmath13 can be arbitrarily small .
then , on the other hand it can become difficult to satisfy the wmap constraint on the dark matter relic density i.e. , to reduce the dark matter relic density after the big bang to an acceptable value compatible with this constraint . to this end , dark matter annihilation processes have to be sufficiently effective . for a lsp mass @xmath20 gev
the following pair annihilation processes can be relevant : exchange of susy partners of fermions ( sfermions , in particular sleptons ) in the t - channel , and z - exchange or higgs - exchange in the s - channel ( if the lsp has a sufficiently large higgsino component ) .
in addition , neutralinos can co - annihilate with other sparticles if they have similar masses , but co - annihilation processes will not be relevant for a light lsp as considered here . in the mssm ,
sufficiently effective dark matter annihilation processes impose constraints on a light lsp : considering lsp annihilation via slepton exchange in the t - channel , a lower bound @xmath21 gev was derived in @xcite from the lower bound of @xmath22 gev on the slepton masses .
( relaxing this bound to @xmath23 gev for stau masses , one obtains @xmath24 gev @xcite , unless the lsp mass is very small corresponding to hot dark matter @xcite . ) allowing for lsp annihilation via cp - odd higgs ( @xmath25 ) exchange in the s - channel , a lower limit @xmath26 gev was given in @xcite from @xmath27 gev for large values of @xmath28 .
however , as noted in @xcite , this region of the parameter space of the mssm is now strongly constrained by the bounds on @xmath29 .
a lsp with a mass in the @xmath30 gev range in the mssm has been considered in @xcite without , however , asking for a correct relic density . in the next - to - minimal supersymmetric standard model ( nmssm ,
see @xcite for recent reviews ) , which can solve the @xmath31-problem of the mssm , the higgs and neutralino sectors are extended by gauge singlet states .
as noticed in @xcite , the mass of the lsp can be considerably smaller in the nmssm than in the mssm and can still be compatible with the wmap constraint on the relic density .
this is a consequence of a light cp - odd higgs boson in the spectrum ( on top of the cp - odd higgs boson of the mssm ) , which can be mostly singlet - like and which is not ruled out by lep - constraints .
then , sufficiently large lsp annihilation cross sections via the exchange of this additional cp - odd higgs boson in the s - channel may be possible even for a light lsp with mass of a few gev .
a light lsp in the nmssm could be a ( dominantly ) singlet - like state ; in this case , however , its direct detection cross sections would be tiny .
on the other hand , as in the mssm , a light lsp in the nmssm can originate from a small value of @xmath15 in which case it will be dominantly bino - like and can have larger direct detection cross sections .
these have been estimated in @xcite , where also constraints on the corresponding parameter space from b - physics , lep and @xmath32-physics were discussed .
however , the corresponding points in the parameter space given as examples in @xcite suffer from a negative effective @xmath31-parameter ( which is in conflict with the measured anomalous magnetic moment of the muon ) , and not all experimental constraints considered below are taken into account . in view of the interest in a light lsp with a mass in the @xmath0 gev range
, we find it appropriate to study upper bounds on its direct detection cross sections in the nmssm .
direct detection cross sections in the nmssm including wmap constraints have been studied before in @xcite , but not for the lsp mass range considered here .
apart from wmap constraints , we take care of a lengthy list of experimental constraints from b - physics ( important for large @xmath33 and/or relatively light charged and cp - odd higgs bosons as relevant here ) , @xmath32-physics and lep - constraints on neutralino production . among the latter
, opal limits on @xmath34 ( @xmath2 ) turn out to be very important . since these are also relevant for the mssm , but
have hardly been discussed before ( a notable exception is @xcite ) , we study their consequences in some detail . for the numerical analysis we use the code nmssmtools @xcite
coupled to micromegas @xcite . as a result
we obtain upper bounds on the spin - independent and spin - dependent lsp - nucleon cross sections of @xmath35 @xmath4 and @xmath36 @xmath4 , varying somewhat with the lsp mass in the @xmath0 gev range .
the maximal value for @xmath37 is indeed near the estimate given in @xcite . in the next section ( 2 ) we present the relevant parameters of the nmssm and their impact on the lsp cross sections . in section ( 3 )
, we discuss the relevant experimental constraints .
the consequences of the opal constraints on @xmath38 ( @xmath2 ) on the parameter space ( implying a lower bound on @xmath39 ) are estimated in an analytic approximation , which reproduces well the full numerical results .
section ( 4 ) is devoted to our results and conclusions .
in the nmssm the @xmath31 parameter of the mssm is replaced by a yukawa coupling @xmath40 to a gauge singlet ( super- ) field @xmath41 .
then , the vacuum expectation value ( vev ) @xmath42 of the real scalar component of @xmath41 generates an effective @xmath31-term [ eq:1 ] _ eff = s .
most of the time one studies the nmssm with a scale invariant superpotential @xmath43 which contains , apart from the yukawa coupling of @xmath41 to the mssm - like higgs doublet fields @xmath44 and @xmath45 , a trilinear term @xmath46 .
hence the higgs mass term @xmath47 in @xmath48 is replaced by [ eq:2 ] w_nmssm = s h_u h_d + s^3 + .
( occasionally one considers the so - called nmssm @xcite without the trilinear coupling @xmath49 , which is replaced by a tadpole - term @xmath50 . )
compared to the mssm , the gauge singlet superfield @xmath41 adds additional degrees of freedom to the cp - even and cp - odd higgs sectors as well as to the neutralino sector .
hence the spectrum contains * 3 cp - even neutral higgs bosons @xmath51
, @xmath52 , which mix in general ; * 2 cp - odd neutral higgs bosons @xmath53 and @xmath54 ; * one charged higgs boson @xmath55 ; * five neutralinos @xmath7 , @xmath56 , which are mixtures of the bino , the neutral wino , the neutral higgsinos and the singlino ; * two charginos which are mixtures of the charged winos and the charged higgsinos . apart from the susy generalisations of the standard - model - like gauge and yukawa couplings and the superpotential in eq .
( [ eq:2 ] ) , the lagrangian of the nmssm contains soft susy breaking terms in the form of gaugino masses @xmath15 , @xmath16 and @xmath17 for the bino , the winos and the gluino , respectively , mass terms for all scalars ( squarks , sleptons , higgs bosons including the singlet @xmath41 ) as well as trilinear scalar self - couplings as @xmath57 , which reflect the trilinear couplings among the superfields in the superpotential .
expressions for the mass matrices for all higgs- and neutralino states can be found in @xcite ; below we confine ourselves to those which are of relevance subsequently . dropping the goldstone mode
, the @xmath58 mass matrix for the cp - odd higgs bosons @xmath59 in the basis ( @xmath60 ) has the elements _
p,11 ^ 2 & = & m_a^2 , + m_p,22 ^ 2 & = & (a_/s+4)v_u v_d -3a _ s , + m_p,12 ^ 2 & = & (a_- 2s ) v [ eq:3 ] where @xmath61 , @xmath62 denote the vevs of @xmath44 , @xmath45 , respectively , @xmath63 gev and , as usual , @xmath64 . the matrix element @xmath65 would be the mass squared of the mssm - like cp - odd scalar @xmath66 , if the singlet sector were absent ;
subsequently we will denote it simply by @xmath67 .
( this parameter can replace the parameter @xmath68 . )
for any ( possibly large ) value of @xmath67 , @xmath59 can have another small eigenvalue corresponding to an additional light cp - odd higgs boson @xmath53 which is mostly singlet - like .
this state will be relevant for the lsp annihilation cross section below .
the mass of the charged higgs scalar is given by [ eq:4 ] m_h^^2 = m_a^2 + v^2 ( - ^2 ) ; note that it decreases with increasing @xmath40 .
as is well known , too small values of @xmath69 can cause disagreements between measurements and corresponding contributions to b - physics - observables as @xmath70 ; this will be of importance below .
notably for large @xmath71 , one of the 3 cp - even higgs bosons will have a mass close to @xmath71 . in the mssm
, the corresponding cp - even state is denoted by @xmath72 , and we will maintain this denomination .
the spin - independent lsp - nucleon cross section will be dominated by the exchange of this cp - even scalar @xmath72 , since its couplings to down - type quarks ( particularly the strange quark ) are enhanced for large values of @xmath33 .
also , the mass of the charged higgs scalar is close to @xmath71 for large @xmath71 ; then the states @xmath72 , @xmath66 and @xmath55 form a nearly degenerate su(2 ) doublet .
in fact this approximate degeneracy holds down to fairly low values of @xmath73 gev . in the neutralino sector , the bino @xmath74 and the neutral wino @xmath75 mix with the neutral higgsinos @xmath76 and the singlino @xmath77 , and generate a symmetric @xmath78 mass matrix @xmath79 .
in the basis @xmath80 , @xmath81 reads [ eq:5 ] m_0 = ( m_1 & 0 & - & & 0 + & m_2 & & - & 0 + & & 0 & - _ & -v_u + & & & 0 & -v_d + & & & & 2 s ) .
it can be diagonalized by an orthogonal real matrix @xmath82 such that the physical masses @xmath83 ordered in @xmath84 are real ( but not necessarily positive ) .
denoting the 5 eigenstates by @xmath85 , we have [ eq:6 ] ^0_i = n_ij ^0_j .
finally , the chargino masses are described by a @xmath58 mass matrix containing @xmath16 and @xmath39 as diagonal entries .
the lower bound of @xmath86 gev on the lightest chargino implies at least the constraint [ eq:7 ] min\{m_2 , next , we discuss the dominant contribution to the spin - independent lsp - nucleon cross section @xmath37 .
leaving aside scenarios with light squark masses of @xmath18 gev ( which are difficult to reconcile with tevatron constraints ) , @xmath37 is dominated by the exchange of cp - even higgs bosons , which couple mostly to the strange quark sea . among the cp - even higgs bosons , the coupling of the state @xmath72 to down - type quarks ( as the strange quark ) increases with @xmath33 .
hence , although its mass is generally larger than the mass of the standard - model - like higgs boson @xmath87 , @xmath72-exchange provides the leading contribution to @xmath37 for large values of @xmath33 .
then , the dominant component of @xmath72 is given by @xmath45 .
the dominant coupling of @xmath88 to the lsp is induced by the bino - higgsino - higgs vertex @xmath89 and hence proportional to @xmath90 , where @xmath91 denotes the bino- and @xmath92 the @xmath93-higgsino - component of the lsp .
all in all one finds [ eq:8 ] ^si ~n_11 ^ 2 n_13 ^ 2 , which shows that the largest values of @xmath37 are obtained for a large product @xmath94 , large @xmath33 and low values of @xmath95 . the dominant contribution to the spin - dependent lsp - nucleon cross section @xmath96 originates , as in the mssm , from @xmath97-exchange .
at first sight one could imagine that , for a light cp - odd higgs boson @xmath53 , its exchange could also give important contributions to @xmath96 .
however , a light @xmath53 is dominantly singlet - like and , moreover , the coupling of its doublet component to strange quarks is always tiny compared to the @xmath97-boson coupling . the coupling of the @xmath97-boson to the lsp originates from the gauge couplings of the higgsino components @xmath98 and @xmath93 .
since no additional free parameters intervene , the spin - dependent cross section @xmath96 is proportional to [ eq:9 ] ^sd ~(n_13 ^ 2 - n_14 ^ 2)^2 . finally the lsp annihilation cross section @xmath99 is dominated , for the lsp mass range @xmath0 gev under consideration , by the exchange of a light @xmath53 in the s - channel .
the dominant contribution to the @xmath100 coupling is induced by the doublet component of @xmath53 and the bino - higgsino components of @xmath11 as in the case of the @xmath101 coupling above ; the singlet components of @xmath53 and @xmath11 play a minor rle here . in any case one has ( neglecting the finite width of @xmath53 and the velocity of @xmath11 near the freeze - out temperature ) [ eq:10 ] _ ann ~ , and hence @xmath99 can be sufficiently large for suitable values of @xmath102 , the lightest eigenvalue of @xmath103 in eq .
( [ eq:3 ] ) .
in this section we discuss various constraints on the parameters of the nmssm , notably ( but not exclusively ) from lep and b - physics , separately in various subsections . as we have seen in eq .
( [ eq:8 ] ) , a large spin - independent detection cross section @xmath37 requires bino - components @xmath91 and higgsino - components @xmath92 of the lsp . for small @xmath15
such that @xmath13 is in the @xmath0 gev range , the bino component of @xmath11 is automatically large .
however , a large higgsino component of @xmath11 require relatively small values for @xmath39 ( below @xmath104 gev ) in the mass matrix @xmath81 in eq .
( [ eq:5 ] ) . consequently the neutralino states @xmath105 and @xmath106 ( for @xmath107 ) are higgsino - like with masses of the order of @xmath39 .
then , the production process @xmath108 ( @xmath109 ) was kinematically possible at lep2 , and corresponding limits from delphi @xcite and opal @xcite have to be taken into account .
the strongest limits come from opal at 208 gev , where we can assume 100% @xmath110 branching ratios for the @xmath7 decays ( see fig .
10 in @xcite ) .
upper bounds on the cross section are given in 5 gev - wide bins of @xmath111 .
since we will find @xmath112 gev , the bounds apply for @xmath105 and @xmath106 separately . for @xmath113 gev , at least one of the @xmath105 or @xmath106 production cross sections ( in association with @xmath11 ) is bounded from above by @xmath114 pb . in principle , both @xmath110-exchange in the s - channel and selectron exchange in the t - channel contribute to this cross section .
however , the interference between these channels is positive , hence the most conservative bounds on the parameters are obtained by assuming heavy selectrons and that @xmath115 originates from @xmath110-exchange only .
the expression for @xmath116 is given , e.g. , in @xcite and can be written as _
z(e^+ e^- _ 1 ^ 0_i^0 ) & = & ( n_13n_i3-n_14n_i4)^2 ( -^2_w+2 ^4_w ) + & & ( e_1 e_i + - m__1 ^ 0 m__i^0 ) [ eq:11 ] ( note the different basis for the neutralinos in @xcite ) with [ eq:12 ] ( s ) = s^2+m__1 ^ 0 ^ 4+m__i^0 ^ 4 - 2 s ( m__1 ^ 0 ^ 2 + m__i^0 ^ 2 ) - 2 m__1 ^ 0 ^ 2 m__i^0 ^ 2 . in order to obtain an approximate expression for the resulting constraints on the parameters , we first neglect @xmath13 everywhere in ( [ eq:11 ] ) . using numerical values for the gauge couplings , ( [ eq:11 ] )
simplifies to [ eq:13 ] _
z(e^+ e^- _ 1 ^ 0_i^0 ) [ ] 4.910 ^ 4 ( 1 + ) ( n_13n_i3-n_14n_i4)^2 with @xmath42 and the masses in gev .
next we look for approximations for the relevant neutralino mixing parameters @xmath82 . for simplification
we assume @xmath16 , @xmath117 such that the wino- and singlino - sectors in the mass matrix @xmath81 in eq .
( [ eq:5 ] ) decouple .
( the wino- and singlino - components of the lsp hardly contribute to the spin - independent cross section . ) assuming , in addition , large @xmath33 such that @xmath118 , @xmath81 can be diagonalised analytically with the results ( we define @xmath119 gev and write @xmath120 ) & & n_11 ~ , n_13 ~ , n_14 ~0 + & & n_21 ~- n_31 ~ , n_23 ~- n_33 ~ , n_24 ~n_34 ~
. [ eq:14 ] replacing these expressions into ( [ eq:13 ] ) , using the numerical values for @xmath42 and @xmath121 in the denominator and , notably , approximating @xmath122 , one ends up with [ eq:15 ] _
z(e^+ e^- _ 1 ^ 0_i^0 ) [ ] 8.310 ^ -7 ( 1 + ) , where @xmath42 , @xmath31 and @xmath123 are in gev .
then the upper opal bound on @xmath124 of @xmath114 pb becomes a lower bound on @xmath125 ( @xmath126 ) , [ eq:16 ] |_eff| 111 . a somewhat stronger version of the opal bound ( @xmath127 pb ) is implemented in the default version of nmssmtools @xcite .
we replace it by the published value of @xmath114 pb @xcite for our numerical analysis . from this , without any approximations , we obtain @xmath128 gev ( varying somewhat with @xmath16 and @xmath129 ) for small values of @xmath13 in good agreement with the previous estimation .
we remark that , within the approximations used in eqs .
( [ eq:14 ] ) , this implies an upper bound on @xmath92 of @xmath130 .
next , we consider constraints from the upper bound on the invisible @xmath97 decay width , to which the decay @xmath131 would contribute . from @xcite
we obtain @xmath132 mev ( a value slightly above the one used in @xcite , but below the value used in @xcite ) .
the expression for the contribution to @xmath133 from @xmath11 reads [ eq:17 ] _
z^inv = ( n_13 ^ 2-n_14 ^ 2)^2 ( 1-)^3/2 ~0.165 ( n_13 ^ 2-n_14 ^ 2)^2 , where the last expression holds for small @xmath13 .
then the upper bound on @xmath133 implies [ eq:18 ] |n_13 ^ 2 - n_14 ^ 2|
< 0.11 . for large @xmath33 , where @xmath134 , this bound on @xmath135 is very similar to the bound obtained above from the opal limits .
according to the numerical analysis without approximations we find that the constraints on the parameter space from @xmath136 are mostly somewhat stronger than those from @xmath137 ; from ( [ eq:8 ] ) and ( [ eq:9 ] ) it should be clear , that these constraints are relevant for upper bounds on the spin - independent and spin - dependent lsp - nucleon cross sections . for the chargino masses we require a lower bound of @xmath138 gev @xcite , which implies lower limits on combinations of the parameters @xmath16 and @xmath39 .
in the neutral higgs sector we apply the various constraints from @xcite .
since the lightest cp - even higgs boson @xmath87 is mostly standard - model - like in our case , these constraints reduce to the well - known bound @xmath139 gev . on the other hand the constraints from b - physics , as described below ,
will imply charged higgs masses above @xmath140 gev , hence additional bounds from direct charged higgs production are not required .
relevant constraints from b - physics originate from bounds on @xmath141 @xcite , @xmath142 ps@xmath143 @xcite and @xmath144 ps@xmath143 @xcite , and the branching ratios @xmath145 @xcite ( which was recently improved to @xmath146 at 95% c.l .
@xcite ) and @xmath147 @xcite .
these constraints are implemented in nmssmtools as described in @xcite , to which we refer for the corresponding contributions to these observables in the nmssm .
it should be noted that charged higgs boson exchange contributes to @xmath148 and @xmath149 , hence the corresponding limits impose lower bounds on @xmath150 . on the other hand
, susy diagrams also contribute to these observables which depend on parameters like @xmath16 , @xmath39 , @xmath151 and @xmath152 @xcite . for specific choices of these parameters ( notably not too large positive values of @xmath152 ) ,
the charged higgs boson contributions can be partially cancelled .
this will be relevant below , since the spin - independent lsp - nucleon cross section ( [ eq:8 ] ) is maximal for small @xmath95 and , as noted above , @xmath153 . at large @xmath129
, the observables @xmath154 , @xmath155 and @xmath156 can receive large contributions from a light cp - odd higgs boson @xmath53 @xcite which , in turn , plays an important rle for the lsp annihilation cross section ( [ eq:10 ] ) for a small lsp mass @xmath13 .
again , additional susy contributions ( box diagrams ) exist , leading to a complicated combination of constraints in the parameter space .
we find that , for a small lsp mass ( light @xmath53 ) , practically all these observables impose bounds on various corners in the parameter space . on the dark matter relic density
we impose the @xmath157 wmap bound @xcite [ eq:19 ] 0.091 < h^2 < 0.129 , which requires a sufficiently large lsp annihilation rate ( [ eq:10 ] ) .
a light cp - odd higgs boson @xmath53 with a mass below @xmath158 gev can appear in radiative @xmath159 decays , on which cleo @xcite and babar @xcite have obtained upper bounds .
these can be translated into the parameter space ( couplings of @xmath53 ) of the nmssm @xcite and are implemented , together with constraints from possible @xmath160 mixing effects @xcite , in nmssmtools .
we find that these constraints are so strong ( imposing , essentially , strong upper bounds on the @xmath161 coupling for @xmath162 gev ) that it becomes very difficult to obtain a lsp annihilation rate compatible with ( [ eq:19 ] ) for @xmath163 gev .
finally we require that the susy contributions to the anomalous magnetic moment of the muon ( see @xcite for such contributions in the nmssm ) improve the disagreement between the result of the e821 experiment @xcite and the standard model ; as in the mssm , this implies a positive value for @xmath39 .
before we turn to our results , we discuss the range of parameters used to maximise the direct detection cross sections respecting the experimental constraints above .
first , for @xmath40 we choose a small value @xmath164 such that its negative effect on @xmath165 as in eq .
( [ eq:4 ] ) remains negligible while a non - zero doublet component of @xmath53 induced by the off - diagonal term in eq .
( [ eq:3 ] ) .
a large value for @xmath166 makes the singlino ( and the singlet - like cp - even higgs state ) heavy such that perturbing mixing effects in these sectors are avoided . for the susy breaking squark and slepton masses we use 1 tev such that sleptons hardly contribute to @xmath115 .
@xmath152 varies from 300 to 650 gev where @xmath55-induced and susy - induced contributions to @xmath148 tend to cancel @xcite .
the susy breaking gluino and the wino masses are chosen as @xmath167 gev and @xmath168 gev , respectively .
( these parameters appear in the loop - induced flavour changing @xmath53-quark vertices @xcite , which should be small in order to allow for a light @xmath53 consistent with the constraints from @xmath169 . )
although eq .
( [ eq:8 ] ) suggests that @xmath37 is maximised for very large values of @xmath33 , the best compromise with b - physics is obtained for reasonable values of @xmath170 .
likewise , eq . ( [ eq:14 ] ) suggests that @xmath39 should be as small as possible in order to maximise @xmath171 , but we find that the best compromise in parameter space is obtained for @xmath172 gev .
( [ eq:8 ] ) also suggests that @xmath37 is maximised for @xmath95 as small as possible .
however , we recall that @xmath173 and that @xmath150 is bounded from below by several b - physics processes .
we choose @xmath71 as an input parameter of the nmssm ( instead of @xmath68 ) and find the largest values of @xmath37 for @xmath174 gev , implying @xmath175 gev and @xmath176 gev where the larger values correspond to lower masses of @xmath13 below .
the susy breaking bino mass term @xmath15 determines @xmath13 , with @xmath177 gev for @xmath178 gev and @xmath179 gev for @xmath180 gev .
finally @xmath181 is chosen in the range @xmath182 gev , which determines @xmath183 such that @xmath11 pair annihilation via @xmath53 exchange in the s - channel gives the correct relic density in agreement with the wmap bound in eq .
( [ eq:19 ] ) . due to the relatively large couplings involved in the @xmath11 pair annihilation process
, @xmath53 must actually be off - shell and hence @xmath183 substantially larger than @xmath184 ; otherwise the relic density is too small . in fig .
[ fig:1 ] we show the result for @xmath183 as function of @xmath13 . for @xmath185 gev ( @xmath186 gev )
the constraints from @xmath156 ( where @xmath53 appears in the s - channel ) become particularly strong and require a somewhat smaller doublet component of @xmath53 . denoting its doublet component by @xmath187
, we have @xmath188 for @xmath189 gev , but @xmath190 for @xmath191 gev .
we note that for @xmath185 gev the value of @xmath181 has to be chosen within a precision less than @xmath192 such that the relic density of @xmath11 is below the wmap bound ( possibly smaller ) , but @xmath193 ; hence this region in the parameter space requires considerable fine tuning . for @xmath194 gev ( @xmath163 gev )
the constraints from cleo and babar become so strong that @xmath187 must be much smaller requiring an even stronger fine tuning of parameters , therefore we will not consider this range of parameters subsequently .
the components of @xmath11 ( the coefficients @xmath195 , see eq .
( [ eq:6 ] ) ) hardly change in the range @xmath196 gev considered here , once we maximise the product @xmath171 in order to maximise @xmath37 .
we have [ eq:20 ] & & n_11 ~-0.94 , n_12 ~0.01 0.03 , n_13 ~-0.32 -0.34 , + & & n_14 ~0.013 0.06 , n_15 ~0.001 .
the masses of the mostly higgsino - like states @xmath105 and @xmath106 are @xmath197 and @xmath198 gev , respectively , and hence as stated before , the limits on @xmath199 are relevant .
the scattering rates of @xmath11 depend somewhat on astrophysical parameters as the escape velocity @xmath200 and the dark matter density @xmath201 near the sun and , more importantly , on nuclear form factors ( quark matrix elements ) as the pion - nucleon sigma term @xmath202 and the size of @xmath203 symmetry breaking parametrized by @xmath204 .
( the difference @xmath205 is proportional to the strange quark matrix element . ) for the astrophysical parameters we use the default values of micromegas @xmath206 km / s and @xmath207 gev/@xmath208 @xcite .
the default values in micromegas for @xmath202 and @xmath204 are @xmath209 mev and @xmath210 mev .
the corresponding results for the upper bound on the spin - independent cross section of @xmath11 off protons @xmath211 in the nmssm are shown in fig .
[ fig:2 ] as a function of @xmath13 as a full red line .
( the spin - independent cross section off neutrons is nearly the same . ) in order to indicate the variation of this upper bound with @xmath202 and @xmath204 , we show a red dashed line as the upper bound on @xmath211 for @xmath212 mev and @xmath213 mev , which would correspond to a larger strange quark matrix element and hence an increase of @xmath37 by a factor @xmath214 . also shown in fig .
[ fig:2 ] are the regions compatible with the excesses of events reported by dama @xcite ( without channeling ( dark blue ) and with channeling ( light blue ) ) , cogent @xcite ( light green ) and a fit to the two events observed by cdms - ii @xcite ( denoted as cdms-09 fit surrounded in dashed green ; these events are also compatible with background ) .
exclusion limits are shown from xenon10 @xcite ( violet ) , xenon100 @xcite ( black ) and cdms - ii @xcite ( magenta , assuming that the two observed events originate from background ) .
* it seems difficult to explain the excesses of events reported by dama and cogent within the general nmssm ( without unification constraints on @xmath15 ) .
hence , as stated in @xcite , significant modifications of parameters like a larger local dark matter density @xmath201 would be required to this end . on the other hand , the two events observed by cdms - ii ( within the contour denoted as cdms-09 fit )
could be explained in the nmssm .
* actual limits of xenon10 , xenon100 and cdms - ii on spin - independent cross sections of wimps in the @xmath215 gev mass range test regions of the parameter space of the nmssm . for completeness
we have also considered the spin - dependent cross section @xmath96 in the nmssm , which is maximal for @xmath216 ( such that @xmath134 in eq .
( [ eq:9 ] ) ) , large values of @xmath71 ( since @xmath95 is irrelevant here ) , and @xmath217 gev . in fig .
[ fig:3 ] we show the maximum of the spin - dependent cross section off protons @xmath218 for the same range of @xmath196 gev .
note that @xmath96 originates from @xmath97-exchange , hence the spin - dependent cross section off neutrons @xmath219 is given by @xmath220 .
the actual experimental upper limits on @xmath96 are one to two orders of magnitude larger @xcite than the upper bounds in the nmssm and not shown in fig .
[ fig:3 ] . to conclude , we have performed a detailed analysis of the parameter space of the nmssm for general values of @xmath15 , which allows for wimp masses in the 2 - 20 gev range .
in contrast to the mssm , light bino - like wimps can have a relic density compatible with wmap constraints due to a light nmssm - specific cp - odd higgs state which can be exchanged in the s - channel . due to reported excesses of events compatible with wimp masses below 20 gev ,
this region is of particular interest .
we have studied in detail the constraints on this region of the parameter space of the nmssm from lep and b - physics , and the regions of parameter space which give rise to maximal direct detection cross sections while not contradicting experimental limits .
the resulting upper bounds on @xmath35 @xmath221 pb make it difficult to explain the excesses of events reported by dama and cogent within the nmssm for small values of @xmath15 . on the other hand
, the two events observed by cdms - ii could be explained in the nmssm .
notably the xenon10 limits @xcite on @xmath37 for wimp masses below 20 gev start to test corresponding regions of the nmssm parameter space
. future results from xenon100 could confirm the presence of a light wimp compatible with the nmssm , or impose further constraints on its parameter space .
[ aleph collaboration , delphi collaboration , l3 collaboration , opal collaboration , sld collaboration , lep electroweak working group , sld electroweak , heavy flavour groups ] , phys .
* 427 * ( 2006 ) 257 [ arxiv : hep - ex/0509008 ] . | in the next - to - minimal supersymmetric standard model , a bino - like lsp can be as light as a few gev and satisfy wmap constraints on the dark matter relic density in the presence of a light cp - odd higgs scalar .
we study upper bounds on the direct detection cross sections for such a light lsp in the mass range @xmath0 gev in the nmssm , respecting all constraints from b - physics and lep .
the opal constraints on @xmath1 ( @xmath2 ) play an important rle and are discussed in some detail .
the resulting upper bounds on the spin - independent and spin - dependent nucleon cross sections are @xmath3 @xmath4 and @xmath5 @xmath4 , respectively .
hence the upper bound on the spin - independent cross section is below the dama and cogent regions , but could be compatible with the two events observed by cdms - ii .
lpt orsay 10 - 53 + + * debottam das and ulrich ellwanger * + _ laboratoire de physique thorique , umr 8627 , cnrs and universit de paris sud , + _ bt . 210 , 91405 orsay , france + _ _ |
a fundamental statistical problem is shrinkage estimation of a multivariate normal mean .
see , for example , the february 2012 issue of _ statistical science _ for a broad range of theory , methods , and applications .
let @xmath0 be multivariate normal with _ unknown _ mean vector @xmath1 and _ known _ variance matrix @xmath2 .
consider the problem of estimating @xmath3 by an estimator @xmath4 under the loss @xmath5 , where @xmath6 is a _ known _ positive definite , symmetric matrix .
the risk of @xmath7 is @xmath8 .
the general problem can be transformed into a canonical form such that @xmath2 is diagonal and @xmath9 , the identity matrix ( e.g. , lehmann and casella @xcite , problem 5.5.11 ) . for simplicity , assume except in section [ sec3.2 ] that @xmath2 is @xmath10 and @xmath11 , where @xmath12 for a column vector @xmath13 .
the letter @xmath14 is substituted for @xmath2 to emphasize that it is diagonal . for this problem , we aim to develop shrinkage estimators that are both minimax and capable of effective risk reduction over the usual estimator @xmath15 even in the heteroscedastic case ( i.e. , @xmath16 are not equal ) .
an estimator of @xmath3 is minimax if and only if , _ regardless of _ @xmath17 , its risk is always no greater than @xmath18 , the risk of @xmath19 . for @xmath20 ,
minimax estimators different from and hence dominating @xmath21 are first discovered in the homoscedastic case where @xmath22 ( i.e. , @xmath23 ) . james and stein @xcite showed that @xmath24 is minimax provided @xmath25 .
stein @xcite suggested the positive - part estimator @xmath26 , which dominates @xmath27 .
throughout , @xmath28 .
shrinkage estimation has since been developed into a general methodology with various approaches , including empirical bayes ( efron and morris @xcite ; morris @xcite ) and hierarchical bayes ( strawderman @xcite ; berger and robert @xcite ) . while these approaches are prescriptive for constructing shrinkage estimators , minimaxity is not automatically achieved but needs to be checked separately . for the heteroscedastic case ,
there remain challenging issues on how much observations with different variances should be shrunk relatively to each other ( e.g. , casella @xcite , morris @xcite ) . for the empirical bayes approach ( efron and morris @xcite ) ,
the coordinates of @xmath29 are shrunk directly in proportion to their variances .
but the existing estimators are , in general , non - minimax ( i.e. , may have a greater risk than the usual estimator @xmath21 ) . on the other hand , berger @xcite proposed minimax estimators , including admissible minimax estimators , such that the coordinates of @xmath29 are shrunk inversely in proportion to their variances .
but the risk reduction achieved over @xmath21 is insubstantial unless all the observations have similar variances . to address the foregoing issues , we develop novel minimax estimators for multivariate normal means under heteroscedasticity .
there are two central ideas in our approach .
the first is to develop a class of minimax estimators by generalizing a geometric argument essentially in stein @xcite ( see also brandwein and strawderman @xcite ) . for the homoscedastic case , the argument shows that @xmath27 can be derived as an approximation to the best linear estimator of the form @xmath30 , where @xmath31 is a scalar .
in fact , the optimal choice of @xmath31 in minimizing the risk is @xmath32 .
replacing @xmath33 by @xmath34 leads to @xmath35 with @xmath36 .
this derivation is highly informative , even though it does not yield the optimal value @xmath37 .
our class of minimax estimators are of the linear form @xmath38 , where @xmath39 is a nonnegative definite , diagonal matrix indicating the direction of shrinkage and @xmath31 is a scalar indicating the magnitude of shrinkage .
the matrix @xmath39 is open to specification , depending on the variance matrix @xmath14 but _ not _ on the data @xmath29 . for a fixed @xmath39 , the scalar @xmath31
is determined to achieve minimaxity , depending on both @xmath14 and @xmath29 .
@xcite minimax estimator corresponds to the special choice @xmath40 , thereby leading to the unusual pattern of shrinkage discussed above .
the second idea of our approach is to choose @xmath39 by approximately minimizing the bayes risk with a normal prior in our class of minimax estimators .
the bayes risk is used to measure average risk reduction for @xmath3 in an elliptical region as in berger @xcite .
it turns out that the solution of @xmath39 obtained by our approximation strategy has an interesting simple form .
in fact , the coordinates of @xmath29 are automatically segmented into two groups , based on their bayes `` importance '' ( berger @xcite ) , which is of the same order as the coordinate variances when the specified prior is homoscedastic .
the coordinates of high bayes `` importance '' are shrunk inversely in proportion to their variances , whereas the remaining coordinates are shrunk in the direction of the bayes rule .
this shrinkage pattern may appear paradoxical : it may be expected that the coordinates of high bayes `` importance '' are to be shrunk in the direction of the bayes rule .
but that scheme is inherently aimed at reducing the bayes risk under the specified prior and , in general , fails to achieve minimaxity ( i.e. , it may lead to even a greater risk than the usual estimator @xmath21 ) .
in addition to simplicity and minimaxity , we further show that the proposed estimator is scale adaptive in reducing the bayes risk : it achieves close to the minimum bayes risk , with the difference no greater than the sum of the 4 highest bayes `` importance '' of the coordinates of @xmath29 , simultaneously over a scale class of normal priors ( including the specified prior ) . to our knowledge
, the proposed estimator seems to be the first one with such a property in the general heteroscedastic case .
previously , in the homoscedastic case , @xmath41 is known to achieve the minimum bayes risk up to the sum of 2 ( equal - valued ) bayes `` importance '' of the coordinates over the scale class of homoscedastic normal priors ( efron and morris @xcite ) .
the rest of this article is organized as follows .
section [ sec2 ] gives a review of existing estimators .
section [ sec3 ] develops the new approach and studies risk properties of the proposed estimator .
section [ sec4 ] presents a simulation study .
section [ sec5 ] provides concluding remarks .
all proofs are collected in the .
we describe a number of existing shrinkage estimators . see lehmann and casella @xcite for a textbook account and strawderman @xcite and morris and lysy @xcite for recent reviews . throughout
, @xmath42 denotes the trace and @xmath43 denotes the largest eigenvalue
. then @xmath44 and @xmath45 . for a bayes approach ,
assume the prior distribution : @xmath46 , where @xmath47 is the prior variance .
the bayes rule is given componentwise by @xmath48 .
then the greater @xmath49 is , the more @xmath50 is shrunk whether @xmath47 is fixed or estimated from the data . for the empirical bayes approach of efron and morris @xcite , @xmath47
is estimated by the maximum likelihood estimator @xmath51 such that @xmath52 morris @xcite suggested the modified estimator @xmath53 in our implementation , the right - hand side of ( [ eb - iter ] ) is computed to update @xmath51 from the initial guess , @xmath54 , for up to 100 iterations until the successive absolute difference in @xmath51 is @xmath55@xmath56 , or @xmath51 is set to @xmath57 so that @xmath58 otherwise . alternatively , xie _ et al . _
@xcite proposed empirical bayes - type estimators based on minimizing stein s @xcite unbiased risk estimate ( sure ) under heteroscedasticity .
their basic estimator is defined componentwise by @xmath59 where @xmath60 is obtained by minimizing the sure of @xmath61 , that is , @xmath62 . in general , the two types of empirical bayes estimators , @xmath63 and @xmath64 , are non - minimax , as shown in section [ sec4 ] . for a direct extension of @xmath27 ,
consider the estimator @xmath65 and , more generally , @xmath66 , where @xmath67 is a scalar constant and @xmath68 a scalar function .
see lehmann and casella @xcite , theorem 5.7 , although there are some typos .
both @xmath69 and @xmath70 are spherically symmetric .
the estimator @xmath69 is minimax provided @xmath71 and @xmath70 is minimax provided @xmath72 .
no such @xmath73 exists unless @xmath74 , which restricts how much @xmath75 can differ from each other .
for example , condition ( [ s - cond ] ) fails when @xmath76 and @xmath77 because @xmath78 and @xmath79 .
berger @xcite proposed estimators of the form @xmath80 and @xmath81 , where @xmath67 is a scalar constant and @xmath68 a scalar function .
then @xmath82 is minimax provided @xmath25 , and @xmath83 is minimax provided @xmath84 , regardless of differences between @xmath75 .
however , a striking feature of @xmath82 and @xmath83 , compared with @xmath63 and @xmath64 , is that the smaller @xmath49 is , the more @xmath50 is shrunk . for example ( [ example ] ) , under @xmath82 ,
the coordinates @xmath85 are shrunk only slightly , whereas @xmath86 are shrunk as if they were shrunk as a 7-dimensional vector under @xmath27 .
the associated risk reduction is insubstantial , because the risk of estimating @xmath87 is a small fraction of the overall risk of estimating @xmath3 .
define the positive - part version of @xmath82 componentwise as @xmath88 the estimator @xmath89 dominates @xmath82 by baranchik @xcite , section 2.5 .
berger @xcite , equation ( 5.32 ) , stated a different positive - part estimator , @xmath83 with @xmath90 , but the @xmath91th component may not be of the same sign as @xmath50 . given a prior @xmath92 , berger @xcite suggested an approximation of berger s
@xcite robust generalized bayes estimator as @xmath93 x.\end{aligned}\ ] ] the estimator is expected to provide significant risk reduction over @xmath15 if the prior is correct and be robust to misspecification of the prior , but it is , in general , non - minimax . in the case of @xmath94 , @xmath95 becomes @xmath96 , in the form of spherically symmetric estimators @xmath97 , where @xmath68 is a scalar function ( bock @xcite , brown @xcite ) .
the estimator @xmath98 is minimax provided @xmath99 and @xmath68 is nondecreasing .
moreover , if @xmath100 , then @xmath98 is non - minimax unless @xmath101 . to overcome the non - minimaxity of @xmath95 ,
berger @xcite developed a minimax estimator @xmath102 by combining @xmath103 , @xmath95 , and a minimax estimator of bhattacharya @xcite .
suppose that @xmath104 and the indices are sorted such that @xmath105 , where @xmath106 .
define @xmath102 componentwise as @xmath107\frac{d_j}{d_j+\gamma_j } x_j,\end{aligned}\ ] ] where @xmath108 . in the case of @xmath94
, @xmath102 reduces to the original estimator of bhattacharya @xcite .
the factor @xmath109 is replaced by @xmath110 in berger s
@xcite original definition of @xmath102 , corresponding to replacing @xmath111 by @xmath112 in @xmath95 . in our simulations ,
the two versions of @xmath102 somehow yield rather different risk curves , and so do the corresponding versions of other estimators .
but there has been limited theory supporting one version over the other .
therefore , we focus on comparisons of only the corresponding versions of @xmath102 and other estimators .
we develop a useful approach for shrinkage estimation under heteroscedasticity , by making explicit how different coordinates are shrunk differently .
the approach not only sheds new light on existing results , but also lead to new minimax estimators .
assume that @xmath113 ( diagonal ) and @xmath9 .
consider estimators of the linear form @xmath114 where @xmath39 is a nonnegative definite , diagonal matrix indicating the _ direction _ of shrinkage and @xmath31 is a scalar indicating the _ magnitude _ of shrinkage .
both @xmath39 and @xmath31 are to be determined .
a sketch of our approach is as follows .
a. for a fixed @xmath39 , the optimal choice of @xmath31 in minimizing the risk is @xmath115 b. for a fixed @xmath39 and a scalar constant @xmath116 , consider the estimator @xmath117 by theorem [ th1 ] , an upper bound on the risk function of @xmath118 is @xmath119,\end{aligned}\ ] ] where @xmath120 .
requiring the second term to be no greater than 0 shows that if @xmath121 , then @xmath118 is minimax provided @xmath122 if @xmath121 , then the upper bound ( [ upper - bound ] ) has a minimum at @xmath123 . c. by taking @xmath123 in @xmath118 ,
consider the estimator @xmath124 subject to @xmath125 , so that @xmath126 is minimax by step ( ii ) .
a positive - part estimator dominating @xmath126 is defined componentwise by @xmath127 where @xmath128 are the diagonal elements of @xmath39 .
the upper bound ( [ upper - bound ] ) on the risk functions of @xmath129 and @xmath130 , subject to @xmath125 , gives @xmath131 we propose to choose @xmath39 based on some optimality criterion , such as minimizing the bayes risk with a normal prior centered at 0 ( berger @xcite ) .
further discussions of steps ( i)(iii ) are provided in sections [ sec3.2][sec3.3 ] .
we first develop steps ( i)(ii ) for the general problem where neither @xmath2 nor @xmath6 may be diagonal .
the results can be as concisely stated as those just presented for the canonical problem where @xmath132 is diagonal and @xmath9 .
such a unification adds to the attractiveness of the proposed approach .
consider estimators of the form ( [ delta - form ] ) , where @xmath39 is not necessarily diagonal , but @xmath133 condition ( [ a - cond ] ) is invariant under a linear transformation . to see this ,
let @xmath134 be a nonsingular matrix and @xmath135 and @xmath136 .
for the transformed problem of estimating @xmath137 based on @xmath138 with variance matrix @xmath139 , the transformed estimator from ( [ delta - form ] ) is @xmath140 .
the application of condition ( [ a - cond ] ) to @xmath141 says that @xmath142 is nonnegative definite and therefore is equivalent to ( [ a - cond ] ) itself . for the canonical problem where @xmath113 ( diagonal ) , condition ( [ a - cond ] ) only requires that @xmath143 is nonnegative definite , allowing @xmath39 to be non - diagonal . on the other hand
, it seems intuitively appropriate to restrict @xmath39 to be diagonal
. then condition ( [ a - cond ] ) is equivalent to saying that @xmath39 is nonnegative definite ( and diagonal ) , which is the condition introduced on @xmath39 in the sketch in section [ sec3.1 ] .
the risk of an estimator of the form ( [ delta - form ] ) is @xmath144 for a fixed @xmath39 , the optimal @xmath31 in minimizing the risk is @xmath145 replacing @xmath146 by @xmath147 and @xmath148 by a scalar constant @xmath149 leads to the estimator @xmath150 for a generalization , replacing @xmath67 by @xmath151 with a scalar function @xmath152 leads to the estimator @xmath153 we provide in theorem [ th1 ] an upper bound on the risk function of @xmath154 .
[ th1 ] assume that @xmath68 almost differentiable ( stein @xcite ) . if ( [ a - cond ] ) holds and @xmath155 is nondecreasing , then for each @xmath3 , @xmath156,\end{aligned}\ ] ] where @xmath157 and @xmath158 . taking @xmath159 in ( [ r - upper - bound ] )
gives an upper bound on @xmath160 . requiring the second term in the risk upper bound ( [ r - upper - bound ] ) to be no greater than 0 leads to a sufficient condition for @xmath154 to be minimax .
[ cor1 ] if ( [ a - cond ] ) holds and @xmath161 , then @xmath154 is minimax provided @xmath162 particularly , @xmath118 is minimax provided @xmath163 . for the canonical problem , inequality ( [ r - upper - bound ] ) and condition ( [ tan - cond2 ] ) for @xmath118 give respectively ( [ upper - bound ] ) and ( [ tan - cond ] ) .
these results generalize the corresponding ones for @xmath69 and @xmath82 in section [ sec2 ] , by the specific choices @xmath164 or @xmath165 .
the generalization also holds if @xmath67 is replaced by a scalar function @xmath166 .
in fact , condition ( [ tan - cond2 ] ) reduces to baranchik s @xcite condition in the homoscedastic case . if @xmath167 , then the risk upper bound ( [ r - upper - bound ] ) has a minimum at @xmath168 . as a result , consider the estimator @xmath169 which is minimax provided @xmath167 . if @xmath170 ( berger @xcite ) , then @xmath171 and , by the proof of theorem [ th1 ] in the , the risk upper bound ( [ r - upper - bound ] ) becomes exact for @xmath118 .
therefore , for @xmath170 , the estimator @xmath172 is uniformly best in the class @xmath118 , in agreement with the result that @xmath41 is uniformly best among @xmath27 in the homoscedastic case .
the estimator @xmath126 has desirable properties of invariance .
first , @xmath126 is easily shown to be invariant under a multiplicative transformation @xmath173 for a scalar @xmath174 .
second , @xmath126 is invariant under a linear transformation of the inference problem .
similarly as discussed below ( [ a - cond ] ) , let @xmath134 be a nonsingular matrix and @xmath175 , @xmath176 , and @xmath136 .
for the transformed problem of estimating @xmath137 based on @xmath138 , the transformed estimator from @xmath126 is @xmath177 , whereas the application of @xmath126 is @xmath178 .
the two estimators are identical because @xmath179 , @xmath180 , and hence @xmath181 . finally , we present a positive - part estimator dominating @xmath126 in the case where both @xmath182 and @xmath183 are symmetric , that is , @xmath184 similarly to ( [ a - cond ] ) , it is easy to see that this condition is invariant under a linear transformation . condition ( [ a - cond2 ] ) is trivially true if @xmath2 , @xmath6 , and @xmath39 are diagonal . in the
, we show that ( [ a - cond2 ] ) holds if and only if there exists a nonsingular matrix @xmath134 such that @xmath185 , @xmath186 , and @xmath187 , where @xmath14 and @xmath188 are diagonal and the diagonal elements of @xmath14 or @xmath188 are , respectively , the eigenvalues of @xmath189 or @xmath39 . in the foregoing notation , @xmath190 and @xmath191 . for the problem of estimating @xmath192 based on @xmath138 , consider the estimator @xmath193 and the positive - part estimator @xmath194 with the @xmath91th component , @xmath195 where @xmath196 are the diagonal elements of @xmath188 .
the estimator @xmath194 dominates @xmath197 by a simple extension of baranchik @xcite , section 2.5 . by a transformation back to the original problem , @xmath197 yields @xmath126 , whereas @xmath198 yields @xmath199 b x.\end{aligned}\ ] ] then @xmath200 dominates @xmath126 .
therefore , ( [ r - upper - bound ] ) also gives an upper bound on the risk of @xmath201 , with @xmath202 , even though @xmath201 is not of the form @xmath154 .
in practice , a matrix @xmath39 satisfying ( [ a - cond2 ] ) can be specified in two steps .
first , find a nonsingular matrix @xmath134 such that @xmath185 and @xmath203 , where @xmath14 is diagonal .
second , pick a diagonal matrix @xmath188 and define @xmath204 .
the first step is always feasible by taking @xmath205 , where @xmath206 is a nonsingular matrix such that @xmath207 and @xmath208 is an orthogonal matrix @xmath208 such that @xmath209 is diagonal . given @xmath210 and @xmath14 , it can be shown that @xmath39 and @xmath200 depend on the choice of @xmath188 , but not on that of @xmath134 , provided that @xmath211 if @xmath212 for any @xmath213 . in the canonical case
where @xmath113 and @xmath9 , this condition amounts to saying that any coordinates of @xmath29 with the same variances should be shrunk in the same way .
different choices of @xmath39 lead to different estimators @xmath126 and @xmath200 .
we study how to choose @xmath39 , depending on @xmath214 but _ not _ on @xmath29 , to approximately optimize risk reduction while preserving minimaxity for @xmath126 .
the estimator @xmath201 provides even greater risk reduction than @xmath126 .
we focus on the canonical problem where @xmath113 ( diagonal ) and @xmath9 .
further , we restrict @xmath39 to be diagonal and nonnegative definite . as discussed in berger @xcite , any estimator can have significantly smaller risk than @xmath15 only for @xmath3 in a specific region .
berger @xcite considered the situation where significant risk reduction is desired for an elliptical region @xmath215 with @xmath216 and @xmath217 the prior mean and prior variance matrix .
see @xmath95 and @xmath102 reviewed in section [ sec2 ] .
to measure average risk reduction for @xmath3 in region ( [ region ] ) , berger @xcite used the bayes risk with the normal prior @xmath218 . for simplicity , assume throughout that @xmath219 and @xmath220 is diagonal .
we adopt berger s @xcite ideas of specifying an elliptical region and using the bayes risk to quantify average risk reduction in this region .
we aim to find @xmath39 , subject to @xmath125 , minimizing the bayes risk of @xmath126 with the prior @xmath221 , @xmath222 , @xmath223 where @xmath224 denotes the expectation with respect to the prior @xmath221 .
given @xmath39 , the risk @xmath225 can be numerically evaluated .
a simple monte carlo method is to repeatedly draw @xmath92 and @xmath226 and then take the average of @xmath227 .
but it seems difficult to literally implement the foregoing optimization .
alternatively , we develop a simple method for choosing @xmath39 by two approximations .
first , if @xmath125 , then taking the expectation of both sides of ( [ point - bound ] ) with respect to the prior @xmath221 gives an upper bound on the bayes risk of @xmath126 : @xmath228 where @xmath229 denotes the expectation with respect to the marginal distribution of @xmath29 in the bayes model , that is , @xmath230 .
an approximation strategy for choosing @xmath39 is to minimize the upper bound ( [ bayes - bound ] ) on the bayes risk or to maximize the second term .
the expectation @xmath231 can be evaluated as a 1-dimensional integral by results on inverse moments of quadratic forms in normal variables ( e.g. , jones @xcite ) .
but the required optimization problem remains difficult .
second , approximations can be made to the distribution of the quadratic form @xmath232 .
suppose that @xmath232 is approximated with the same mean by @xmath233 , where @xmath234 is a chi - squared variable with @xmath235 degrees of freedom .
then @xmath231 is approximated by @xmath236 .
we show in the that this approximation gives a valid lower bound : @xmath237 a direct application of jensen s inequality shows that @xmath238 .
but the lower bound ( [ bayes - bound2 ] ) is strictly tighter and becomes exact when @xmath239 .
no simple bounds such as ( [ bayes - bound2 ] ) seem to hold if more complicated approximations ( e.g. , satterthwaite @xcite ) are used .
combining ( [ bayes - bound ] ) and ( [ bayes - bound2 ] ) shows that if @xmath125 , then @xmath240 notice that @xmath126 is invariant under a multiplicative transformation @xmath241 for a scalar @xmath174 , and so is the upper bound ( [ bayes - bound3 ] ) . our strategy for choosing @xmath39 is to minimize the upper bound ( [ bayes - bound3 ] ) subject to @xmath125 or , equivalently , to solve the constrained optimization problem : @xmath242\\[-8pt ] & & \quad \mbox{subject to } \quad \sum_{j=1}^p ( d_j+ \gamma_j ) a_j^2 = \mbox{fixed}. \nonumber\end{aligned}\ ] ] the condition @xmath125 is dropped , because for @xmath20 , the achieved maximum is at least @xmath243 for some scalar @xmath244 . in spite of the approximations used in our approach ,
theorem [ th2 ] shows that not only the problem ( [ opt ] ) admits a non - iterative solution , but also the solution has a very interesting interpretation . for convenience ,
assume thereafter that the indices are sorted such that @xmath245 .
[ th2 ] assume that @xmath20 , @xmath246 with @xmath247 and @xmath248 with @xmath249 ( @xmath250 ) .
for problem ( [ opt ] ) , assume that @xmath251 with @xmath252 ( @xmath253 ) and @xmath254 , satisfied by @xmath255
. then the following results hold .
a. there exists a _
unique _ solution , @xmath256 , to problem ( [ opt ] ) .
b. let @xmath257 be the largest index such that @xmath258 .
then @xmath259 , @xmath260 for @xmath261 , and @xmath262 where @xmath263 and @xmath264 the achieved maximum value , @xmath265 , is @xmath266 . c. the resulting estimator @xmath267 is minimax .
we emphasize that , although @xmath39 can be considered a tuning parameter , the solution @xmath268 is _ data independent _ , so that @xmath269 is automatically minimax .
if a data - dependent choice of @xmath39 were used , minimaxity would not necessarily hold .
this result is achieved both because each estimator @xmath126 with @xmath125 is minimax and because a global criterion ( such as the bayes risk ) is used , instead of a pointwise criterion ( such as the frequentist risk at the unknown @xmath3 ) , to select @xmath39 . by these considerations ,
our approach differs from the usual exercise of selecting a tuning parameter in a data - dependent manner for a class of candidate estimators .
there is a remarkable property of monotonicity for the sequence @xmath270 , which underlies the uniqueness of @xmath257 and @xmath268 .
[ cor2 ] the sequence @xmath271 is nonincreasing : for @xmath272 , @xmath273 , where the equality holds if and only if @xmath274 the condition @xmath275 is equivalent to saying that the left side is greater than the right - hand side in the above expression for @xmath276 . therefore , @xmath257 is the smallest index @xmath272 with this property , and @xmath277 .
the estimator @xmath267 is invariant under scale transformations of @xmath268 .
therefore , the constant @xmath278 can be dropped from the expression of @xmath268 in theorem [ th1 ] .
[ cor3 ] the solution @xmath279 can be rescaled such that @xmath280 then @xmath281 .
moreover , it holds that @xmath282 the estimator @xmath267 can be expressed as @xmath283 the foregoing results lead to a simple algorithm for solving problem ( [ opt ] ) : a. sort the indices such that @xmath284 .
b. take @xmath257 to be the smallest index @xmath285 ( corresponding to the largest @xmath286 ) such that @xmath272 and @xmath287 or take @xmath288 if there exists no such @xmath285 . c. compute @xmath289 by ( [ sol1])([sol2 ] ) .
this algorithm is guaranteed to find the ( unique ) solution to problem ( [ opt ] ) by a fixed number of numerical operations .
no iteration or convergence diagnosis is required .
therefore , the algorithm is exact and non - iterative , in contrast with usual iterative algorithms for nonlinear , constrained optimization .
the estimator @xmath267 has an interesting interpretation . by ( [ sol1])([sol2 ] )
, there is a dichotomous segmentation in the shrinkage direction of the coordinates of @xmath29 based on @xmath290 .
this quantity @xmath291 is said to reflect the bayes `` importance '' of @xmath292 , that is , the amount of reduction in bayes risk obtainable in estimating @xmath292 in berger @xcite .
the coordinates with high @xmath291 are shrunk inversely in proportion to their variances @xmath49 as in berger s
@xcite estimator @xmath82 , whereas the coordinates with low @xmath291 are shrunk in the direction of the bayes rule .
therefore , @xmath267 mimics the bayes rule to reduce the bayes risk , except that @xmath267 mimics @xmath82 for some coordinates of highest bayes `` importance '' in order to achieve minimaxity .
in fact , by inequality ( [ sol - ineq ] ) , the relative shrinkage , @xmath293 , of each @xmath50 ( @xmath294 ) in @xmath267 versus the bayes rule is always no greater than that of @xmath295 ( @xmath296 ) .
the expression ( [ delta - a ] ) suggests that there is a close relationship in beyond the shrinkage direction between @xmath297 and the bayes rule under the bayes model , @xmath298 . in this case , @xmath299 , and hence @xmath297 behaves similarly to @xmath300 .
therefore , _ on average _ under the bayes model , the coordinates of @xmath29 are shrunk in @xmath269 the same as in the bayes rule , except that some coordinates of highest bayes `` importance '' are shrunk no greater than in the bayes rule .
while this discussion seems heuristic , we provide in section [ sec3.4 ] a rigorous analysis of the bayes risk of @xmath267 , compared with that of the bayes rule .
we now examine @xmath267 for two types of priors : @xmath301 and @xmath302 ( @xmath303 ) , referred to as the homoscedastic and heteroscedastic priors . for both types ,
@xmath304 are of the same order as the variances @xmath305 .
recall that @xmath126 is invariant under a multiplicative transformation of @xmath39 .
for both the homoscedastic prior with @xmath306 and the heteroscedastic prior _ regardless _ of @xmath307 , the solution @xmath308 can be rescaled such that @xmath309 denote by @xmath310 this rescaled matrix @xmath268 , corresponding to @xmath94 .
then coordinates with high variances are shrunk inversely in proportion to their variances , whereas coordinates with low variances are shrunk symmetrically . for @xmath94 ,
the proposed method has a purely frequentist interpretation : it seeks to minimize the upper bound ( [ bayes - bound3 ] ) on the pointwise risk of @xmath126 at . for the homoscedastic prior with @xmath311 ,
the proposed method is then to minimize the upper bound ( [ bayes - bound3 ] ) on the bayes risk of @xmath126 with an extremely flat , homoscedastic prior . as @xmath311
, the solution @xmath268 can be rescaled such that @xmath312 denote by @xmath313 this rescaled matrix @xmath268 .
then coordinates with low ( or high ) variances are shrunk directly ( or inversely ) in proportion to their variances . the direction @xmath313 can also be obtained by using a fixed prior in the form @xmath314 ( @xmath250 ) for arbitrary @xmath315 , where @xmath316 .
finally , in the homoscedastic case ( @xmath23 ) , if the prior is also homoscedastic ( @xmath317 ) , then @xmath288 , @xmath318 , and @xmath269 reduces to the james stein estimator @xmath319 , _ regardless _ of @xmath320 and @xmath47 . the estimator @xmath267 is constructed by minimizing the upper bound ( [ bayes - bound3 ] ) on the bayes risk subject to minimaxity .
in addition to simplicity , interpretability , and minimaxity demonstrated for @xmath267 , it remains important to further study risk properties of @xmath267 and show that @xmath269 can provide effective risk reduction over @xmath15 .
write @xmath321 whenever needed to make explicit the dependency of @xmath268 on @xmath217 .
first , we study how close the bayes risk of @xmath322 can be to that of the bayes rule , which is the smallest possible among _ all _ estimators including non - minimax ones , under the prior @xmath221 , @xmath323 .
the bayes rule @xmath324 is given componentwise by @xmath325 , with the bayes risk @xmath326 where @xmath327 , indicating the bayes `` importance '' of @xmath292 ( berger @xcite ) .
the upper bound ( [ bayes - bound3 ] ) on the bayes risk of @xmath328 gives @xmath329 because @xmath330 and hence @xmath331 by corollary [ cor3 ] .
it appears that the difference between @xmath332 and @xmath333 tends to be large if @xmath257 is large .
but @xmath334 can not differ too much from each other because by corollary [ cor1 ] , @xmath335 then the difference between @xmath336 and @xmath337 should be limited even if @xmath257 is large .
a careful analysis using these ideas leads to the following result .
[ th3 ] suppose that the prior is @xmath222 .
if @xmath338 , then @xmath339 if @xmath340 , then @xmath341 throughout , an empty summation is 0 .
there are interesting implications of theorem 3 . by ( [ loose - bound1 ] ) and ( [ loose - bound2 ] ) , @xmath342 then @xmath343 achieves almost the minimum bayes risk if @xmath344 . in terms of bayes
risk reduction , the bound ( [ bayes - close ] ) shows that @xmath345 therefore , @xmath343 achieves bayes risk reduction within a negligible factor of that achieved by the bayes rule if @xmath346 . in the homoscedastic case
where both @xmath22 and @xmath347 , @xmath267 reduces to @xmath319 , regardless of @xmath307 ( section [ sec3.3 ] ) .
then the bounds ( [ tight - bound1 ] ) and ( [ tight - bound2 ] ) become exact and give efron and morris s @xcite result that @xmath348 or equivalently @xmath349 .
it is interesting to compare the bayes risk bound of @xmath350 with that of the following simpler version of berger s
@xcite estimator @xmath102 : @xmath351 by berger @xcite , @xmath352 is minimax and @xmath353 there seems to be no definite comparison between the bounds ( [ tight - bound1 ] ) and ( [ tight - bound2 ] ) on @xmath354 and the exact expression ( [ mb - bayes1 ] ) for @xmath355 , although the simple bounds ( [ loose - bound1 ] ) and ( [ loose - bound2 ] ) is slightly higher , by at most @xmath356 , than the bound ( [ mb - bayes2 ] ) .
of course , each risk upper bound gives a conservative estimate of the actual performance , and comparison of two upper bounds should be interpreted with caution .
in fact , the positive - part estimator @xmath357 yields lower risks than those of the non - simplified estimator @xmath102 in our simulation study ( section [ sec4 ] ) .
the simplicity of @xmath267 and @xmath357 makes it easy to further study them in other ways than using the bayes ( or average ) risk .
no similar result to the following theorem [ th4 ] has been established for @xmath102 or @xmath358 .
corresponding to the prior @xmath359 , consider the worst - case ( or maximum ) risk @xmath360 over the hyper - rectangle @xmath361 ( e.g. , donoho _ et al . _
@xcite ) . applying jensen s inequality to ( [ point - bound ] ) shows that if @xmath362 , then @xmath363 which immediately leads to @xmath364 by the discussion after ( [ bayes - bound2 ] ) , a direct application of jensen s inequality to ( [ bayes - bound ] ) shows that the bayes risk @xmath225 is also no greater than the right - hand side of ( [ minimax - bound ] ) , whereas inequality ( [ bayes - bound2 ] ) leads to a strictly tighter bound ( [ bayes - bound3 ] ) .
nevertheless , the upper bound ( [ minimax - bound ] ) on the worst - case risk of @xmath343 gives @xmath365 similarly as how ( [ bayes - bound3 ] ) leads to ( [ bayes - bound4 ] ) on the bayes risk of @xmath343 .
therefore , the following result holds by the same proof of theorem 3 .
[ th4 ] suppose that @xmath366 .
if @xmath338 , then @xmath367 if @xmath340 , then @xmath368 there are similar implications of theorem [ th4 ] to those of theorem [ th3 ] . by donoho
@xcite , the minimax linear risk over @xmath369 , @xmath370 , coincides with the minimum bayes risk @xmath337 , and is no greater than @xmath371 times the minimax risk over @xmath372 , @xmath373 .
these results are originally obtained in the homoscedastic case ( @xmath374 ) , but they remain valid in the heteroscedastic case by the independence of the observations @xmath50 and the separate constraints on @xmath292 . therefore , a similar result to ( [ bayes - close ] ) holds : @xmath375 if @xmath344 , then @xmath269 achieves almost the minimax linear risk ( or the minimax risk up to a factor of @xmath371 ) over the hyper - rectangle @xmath369 , in addition to being globally minimax with @xmath3 unrestricted .
the foregoing results might be considered non - adaptive in that @xmath328 is evaluated with respect to the prior @xmath376 or the parameter set @xmath372 with the same @xmath217 used to construct @xmath343 .
but , by the invariance of @xmath126 under scale transformations of @xmath39 , @xmath328 is identical to the estimator , @xmath377 , that would be obtained if @xmath217 is replaced by @xmath378 for any scalar @xmath379 such that the diagonal matrix @xmath380 is nonnegative definite . by theorems 34 , this observation leads directly to the following adaptive result .
in contrast , no adaptive result seems possible for @xmath102 .
[ cor4 ] let @xmath378 and @xmath381 .
then for each @xmath382 , @xmath383 & \le & r\bigl ( \delta^{\mathrm{bayes}}_{\gamma_\alpha } , \pi_{\gamma_\alpha}\bigr ) + \alpha^{-1 } \bigl(d_1^*+d_2^*+d_3^*+d_4^ * \bigr ) \\ & = & r^l(\mathcal h_{\gamma_\alpha})+ \alpha^{-1 } \bigl ( d_1^*+d_2^*+d_3^*+d_4^*\bigr),\end{aligned}\ ] ] where @xmath384 . for fixed @xmath217
, @xmath343 can achieve close to the minimum bayes risk or the minimax linear risk with respect to each prior in the class @xmath385 or each parameter set in the class @xmath386 under mild conditions . for illustration , consider the case of a heteroscedastic prior with @xmath387 .
then @xmath388 can be reparameterized as @xmath389 . by corollary [ cor4 ] , for each @xmath307 , @xmath390 where @xmath391 and @xmath392 . therefore , if @xmath393 , then @xmath394 achieves the minimum bayes risk , within a negligible factor , under the prior @xmath395 for each @xmath396 .
this can be seen as an extension of the result that in the homoscedastic case , @xmath41 asymptotically achieves the minimum bayes risk under the prior @xmath397 for each @xmath396 as @xmath398 .
finally , we compare the estimator @xmath267 with a block shrinkage estimator , suggested by the differentiation in the shrinkage of low- and high - variance coordinates by @xmath267 .
consider the estimator @xmath399 where @xmath400 is a cutoff index , and @xmath401 if @xmath402 is of dimension 1 or 2 .
the index @xmath400 can be selected such that the coordinate variances are relatively homogeneous in each block .
alternatively , a specific strategy for selecting @xmath400 is to minimize an upper bound on the bayes risk of @xmath403 , similarly as in the development of @xmath267 . applying ( [ bayes - bound3 ] ) with @xmath40 to @xmath404 in the two blocks shows that @xmath405 , where @xmath406 the first ( or second ) term in @xmath407 is set to 0 if @xmath408 ( or @xmath409 )
. then @xmath400 can be defined as the smallest index such that @xmath410 .
but the upper bound ( [ bayes - bound4 ] ) on @xmath411 is likely to be smaller than the corresponding bound on @xmath412 , because @xmath413 for each @xmath414 by the cauchy schwarz inequality @xmath415 .
therefore , @xmath267 tends to yield greater risk reduction than @xmath403 .
this analysis also indicates that @xmath267 can be advantageous over @xmath403 extended to multiple blocks .
the rationale of forming blocks in @xmath267 and @xmath416 differs from that in existing block shrinkage estimators ( e.g. , brown and zhao @xcite ) . as discussed in cai @xcite ,
block shrinkage has been developed mainly in the homoscedastic case as a technique for pooling information : the coordinate means are likely to be similar to each other within a block .
nevertheless , it is possible to both deal with heterogeneity among coordinate variances and exploit homogeneity among coordinate means within individual blocks in our approach using a block - homoscedastic prior ( i.e. , the prior variances are equal within each block ) .
this topic can be pursued in future work .
we conduct a simulation study to compare the following 8 estimators , a. non - minimax estimators : @xmath63 by ( [ eb ] ) , @xmath64 by ( [ xkb ] ) , @xmath95 by ( [ rb ] ) with @xmath94 ; b. minimax estimators : @xmath417 by ( [ b+ ] ) , @xmath102 by ( [ mb ] ) with @xmath94 or @xmath418 for some large @xmath47 , @xmath419 by ( [ a+ ] ) with @xmath420 and @xmath313 .
recall that @xmath421 corresponds to @xmath94 or @xmath422 and @xmath423 corresponds to @xmath424 with @xmath311 .
in contrast , letting the diagonal elements of @xmath217 tend to @xmath57 in any direction in @xmath95 and @xmath102 leads to @xmath15 . setting @xmath217 to 0 or @xmath57
is used here to specify the relevant estimators , rather than to restrict the prior on @xmath3 . for completeness
, we also study the following estimators : @xmath425 by ( [ b+ ] ) , @xmath95 with @xmath111 replaced by @xmath112 in ( [ rb ] ) , @xmath102 with @xmath109 replaced by @xmath110 in ( [ mb ] ) , and @xmath419 with @xmath426 replaced by @xmath427 in ( [ a+ ] ) , referred to as the alternative versions of @xmath428 , @xmath95 , @xmath102 , and @xmath429 respectively .
the usual choices of the factors , @xmath111 , @xmath109 , and @xmath426 , are motivated to minimize the risks of the non - positive - part estimators , but may not be the most desirable for the positive - part estimators . as seen below , the alternative choices @xmath112 , @xmath110 , and @xmath430 can lead to risk curves for the positive - part estimators rather different from those based on the usual choices @xmath431 , @xmath109 , and @xmath426 .
therefore , we compare the estimators @xmath432 , @xmath95 , @xmath102 , and @xmath419 and , separately , their alternative versions .
each estimator @xmath7 is evaluated by the pointwise risk function @xmath433 as @xmath3 moves in a certain direction or the bayes risk function @xmath434 as @xmath435 varies in a set of priors on @xmath3 .
consider the homoscedastic prior @xmath436 or the heteroscedastic prior @xmath437 for @xmath438 .
as discussed in section [ sec3.3 ] , the bayes risk with the first or second prior is meant to measure average risk reduction over the region @xmath439 or @xmath440 .
corresponding to the two priors , consider the direction along @xmath441 or @xmath442 , where @xmath197 gives the euclidean distance from 0 to the point indexed by @xmath197 .
the two directions are referred to as the homoscedastic and heteroscedastic directions .
we investigate several configurations for @xmath14 , including ( [ example ] ) and @xmath443 where @xmath444 is a chi - squared variable with @xmath285 degrees of freedom . in the last case ,
@xmath445 can be considered a typical sample from a scaled inverse chi - squared distribution , which is the conjugate distribution for normal variances . in the case ( [ example - group3 ] ) , the coordinates may be segmented intuitively into three groups with relatively homogeneous variances . in the case ( [ example - group22 ] ) , there is no clear intuition about how the coordinates should be segmented into groups . for fixed @xmath14 ,
the pointwise risk @xmath433 is computed by repeatedly drawing @xmath446 and then taking the average of @xmath447 .
the bayes risk is computed by repeatedly drawing @xmath92 and @xmath448 and then taking the average of @xmath447 .
each monte carlo sample size is set to @xmath449 .
axis ( third row ) in the case ( [ example - group3 ] ) . left : non - minimax estimators @xmath63 ( @xmath450 ) , @xmath95 ( @xmath451 ) , @xmath64 ( @xmath452 ) .
right : minimax estimators @xmath417 ( @xmath453 ) , @xmath102 with @xmath94 ( @xmath454 ) and @xmath455 ( @xmath456 ) , @xmath419 with @xmath420 ( @xmath457 ) and @xmath458 ( @xmath459 ) . ]
axis ( third row ) in the case ( [ example - group3 ] ) , with the same legend as in figure [ fig1 ] .
the alternative versions of @xmath428 , @xmath460 , @xmath102 , and @xmath419 are used . ] the relative performances of the estimators are found to be consistent across different configurations of @xmath14 studied . moreover , the bayes risk curves under the homoscedastic prior are similar to the pointwise risk curves along the homoscedastic direction .
the bayes risk curves under the heteroscedastic prior are similar to the pointwise risk curves along the heteroscedastic direction .
figure [ fig1 ] shows the pointwise risks of the estimators with the usual versions of @xmath428 , @xmath95 , @xmath102 , and @xmath419 and figure [ fig2 ] shows those of the estimators with the alternative versions of @xmath428 , @xmath95 , @xmath102 , and @xmath419 for the case ( [ example - group3 ] ) , with roughly three groups of coordinate variances , which might be considered unfavorable to our approach . for both @xmath310 and @xmath313 , the cutoff index @xmath257 is found to be 3 .
see the supplementary material ( tan @xcite ) for the bayes risk curves of all these estimators for the case ( [ example - group3 ] ) and the results for other configurations of @xmath14 .
a number of observations can be drawn from figures [ fig1][fig2 ] .
first , @xmath461 , @xmath64 , and @xmath95 have among the lowest risk curves along the homoscedastic direction . but along the heteroscedastic direction , the risk curves of @xmath63 and @xmath64 rise quickly above the constant risk of @xmath29 as @xmath197 increases .
moreover , all the risk curves of @xmath63 , @xmath64 , and @xmath95 along the @xmath462 axis exceed the constant risk of @xmath29 as @xmath463 increases .
therefore , @xmath63 , @xmath64 , and @xmath460 fail to be minimax , as mentioned in section [ sec2 ] .
second , @xmath417 or @xmath464 has among the highest risk curve , except where the risk curves of @xmath63 and @xmath64 exceed the constant risk of @xmath29 along the heteroscedastic direction .
the poor performance is expected for @xmath465 or @xmath464 , because there are considerable differences between the coordinate variances in ( [ example - group3 ] ) .
third , among the minimax estimators , @xmath419 with @xmath420 or @xmath313 has the lowest risk curve along various directions , whether the usual versions of @xmath428 , @xmath466 , and @xmath419 are compared ( figure [ fig1 ] ) or the alternative versions are compared ( figure [ fig2 ] ) .
fourth , the risk curve of @xmath419 with @xmath420 is similar to that of @xmath419 with @xmath467 along the heteroscedastic direction .
but the former is noticeably higher than the latter along the homoscedastic direction as @xmath197 increases , whereas is noticeably lower than the latter along the @xmath462 axis as @xmath463 increases .
these results agree with the construction of @xmath310 using a heteroscedastic prior and @xmath313 using a flat , homoscedastic prior .
their relative performances depend on the direction in which the risks are evaluated .
fifth , @xmath102 with @xmath94 has risk curves below that of @xmath417 or @xmath464 , but either above or crossing those of @xmath419 with @xmath420 and @xmath458 .
moreover , @xmath102 with @xmath455 has elevated , almost flat risk curves for @xmath197 from 0 to 16 .
this seems to indicate an undesirable consequence of using a non - degenerate prior for @xmath102 in that the risk tends to increase for @xmath3 near 0 , and remains high for @xmath3 far away from 0 .
the foregoing discussion involves the comparison of the risk curves as @xmath3 moves away from 0 between @xmath102 and @xmath468 specified with fixed priors .
alternatively , we compare the pointwise risks at @xmath469 or @xmath442 and the bayes risks under the prior @xmath470 or @xmath471 between @xmath102 and @xmath357 specified with the prior @xmath470 for a range of @xmath197 .
the homoscedastic prior used in the specification of @xmath102 and @xmath357 can be considered correctly specified or misspecified , when the bayes risks are evaluated under , respectively , the homoscedastic or heteroscedastic prior or when the pointwise risks are evaluated along the homoscedastic or heteroscedastic direction .
for each situation , @xmath357 has lower pointwise or bayes risks than @xmath102
. see figure a2 in the supplementary material ( tan @xcite ) .
the estimator @xmath267 and its positive - part version @xmath472 are not only minimax and but also have desirable properties including simplicity , interpretability , and effectiveness in risk reduction .
in fact , @xmath267 is defined by taking @xmath473 in a class of minimax estimators @xmath126 .
the simplicity of @xmath267 holds because @xmath126 is of the linear form @xmath474 , with @xmath39 and @xmath31 indicating the direction and magnitude of shrinkage .
the interpretability of @xmath269 holds because the form of @xmath268 indicates that one group of coordinates are shrunk in the direction of berger s
@xcite minimax estimator whereas the remaining coordinates are shrunk in the direction of the bayes rule .
the effectiveness of @xmath267 in risk reduction is supported , in theory , by showing that @xmath269 can achieve close to the minimum bayes risk simultaneously over a scale class of normal priors ( corollary [ cor4 ] ) . for various scenarios in our numerical study ,
the estimators @xmath475 with extreme priors yield more substantial risk reduction than existing minimax estimators .
it is interesting to discuss a special feature of @xmath154 and hence of @xmath118 and @xmath126 among linear , shrinkage estimators of the form @xmath476 where @xmath39 and @xmath134 are nonnegative definite matrices and @xmath477 is a scalar function .
the estimator @xmath154 corresponds to the choice @xmath478 , which is motivated by the form of the optimal @xmath31 in minimizing the risk of @xmath479 for fixed @xmath39 .
on the other hand , berger and srinivasan @xcite showed that under certain regularity conditions on @xmath477 , an estimator ( [ general ] ) can be generalized bayes or admissible only if @xmath480 .
this condition is incompatible with @xmath481 , unless @xmath482 as in berger s @xcite estimator .
therefore , @xmath126 including @xmath267 is , in general , not generalized bayes or admissible .
this conclusion , however , does not apply directly to the positive - part estimator @xmath200 , which is no longer of the linear form @xmath474 .
there are various topics that can be further studied .
first , the prior on @xmath3 is fixed , independently of data in the current paper .
a useful extension is to allow the prior to be estimated within a certain class , for example , homoscedastic priors @xmath483 , from the data , in the spirit of empirical bayes estimation ( e.g. , efron and morris @xcite ) .
second , the bayes risk with a normal prior is used to measure average risk reduction in an elliptical region ( section [ sec3.3 ] ) .
it is interesting to study how our approach can be extended when using a non - normal prior on @xmath3 , corresponding to a non - elliptical region in which risk reduction is desired .
the following extends stein s @xcite lemma for computing the expectation of the inner product of @xmath484 and a vector of functions of @xmath29 .
[ lem1 ] let @xmath485 be multivariate normal with mean @xmath3 and variance matrix @xmath2 .
assume that @xmath486 is almost differentiable stein @xcite with @xmath487 for @xmath488 , where @xmath489
. then @xmath490,\ ] ] where @xmath491 is the matrix with @xmath492th element @xmath493 .
a direct generalization of lemma 2 in stein @xcite to a normal random vector with non - identity variance matrix gives @xmath494 where @xmath495 is the row vector with @xmath91th element @xmath496 .
taking the @xmath497th element of both sides of the equation gives @xmath498 where @xmath499 is the @xmath492th element of @xmath2 .
summing both sides of the preceding equation over @xmath497 gives the desired result .
proof of theorem [ th1 ] by direct calculation , the risk of @xmath154 is @xmath500 by lemma [ lem1 ] and the fact that @xmath501 , the third term after the minus sign in @xmath502 is @xmath503 by condition ( [ a - cond ] ) , @xmath504 is nonnegative definite . by section 21.14 and
exercise 21.32 in harville @xcite , @xmath505 for @xmath506 .
then the preceding expression is bounded from below by @xmath507 which leads immediately to the upper bound on @xmath508 .
proof for condition ( [ a - cond2 ] ) we show that if condition ( [ a - cond2 ] ) holds , then there exists a nonsingular matrix @xmath134 with the claimed properties .
the converse is trivially true .
let @xmath509 be the unique symmetric , positive definite matrix such that @xmath510 .
then @xmath511 is symmetric , that is , @xmath512 , because @xmath513 .
moreover , @xmath514 and @xmath511 commute , that is , @xmath515 , because @xmath516 and @xmath511 is symmetric .
therefore , @xmath517 and @xmath511 are simultaneously diagonalizable ( harville @xcite , section 21.13 ) .
there exists an orthogonal matrix @xmath208 such that @xmath518 and @xmath519 for some diagonal matrices @xmath14 and @xmath188 .
then @xmath520 satisfies the claimed properties .
proof of inequality ( [ bayes - bound2 ] ) we show that if @xmath521 are independent standard normal variables , then @xmath522 .
. then @xmath524 and @xmath525 are independent , @xmath526 , and @xmath527 .
the claimed inequality follows because @xmath528 , @xmath529 , and @xmath530 @xmath531 by jensen s inequality .
proofs of theorem [ th2 ] and corollary [ cor2 ] consider the transformation @xmath532 and @xmath533 , so that @xmath534 and @xmath535 .
problem ( [ opt ] ) is then transformed to @xmath536 , subject to @xmath537 ( @xmath538 ) and @xmath539 , which is of the form of the special case of ( [ opt ] ) with @xmath540 ( @xmath250 ) .
but it is easy to verify that if the claimed results hold for the transformed problem , then the results hold for original problem ( [ opt ] ) .
therefore , assume in the rest of proof that @xmath540 ( @xmath250 )
. there exists at least a solution , @xmath268 , to problem ( [ opt ] ) by boundedness of the constraint set .
let @xmath541 and @xmath542 .
a key of the proof is to exploit the fact that , by the setup of problem ( [ opt ] ) , @xmath289 is automatically a solution to the problem @xmath543 the karush
tucker condition for this problem gives @xmath544 where @xmath31 , @xmath545 ( @xmath546 ) , and @xmath547 satisfying @xmath548 ( @xmath250 ) are lagrange multipliers .
first , we show that @xmath549 and hence @xmath550 for @xmath253 .
if @xmath551 , then either @xmath549 for @xmath250 , or @xmath552 .
the latter case is infeasible by the constraint @xmath553 .
suppose @xmath554 . by ( [ a2 ] ) , @xmath549 for each @xmath555 .
then @xmath556 for each @xmath557 because @xmath558 .
second , we show that @xmath259 . if @xmath551 , then @xmath559 .
. then @xmath560 by ( [ a2 ] ) . summing ( [ a3 ] ) over @xmath561 and ( [ a4 ] ) shows that @xmath562 .
therefore , @xmath563 or equivalently @xmath259 .
third , we show that @xmath564 and @xmath565 . for each @xmath566 and @xmath567 ,
@xmath568 by ( [ a2])([a3 ] ) and then @xmath569 because @xmath558
. the inequalities also hold for @xmath276 , by application of the argument to problem ( [ a1 ] ) with @xmath257 replaced by some @xmath566 .
then @xmath570 because @xmath571 for each @xmath555 , @xmath572 , and @xmath257 is the largest element in @xmath573 .
fourth , we show the expressions for @xmath574 and the achieved maximum value . by the definition of @xmath573 ,
@xmath575 for @xmath576 . by ( [ a2 ] ) , @xmath577 for @xmath578 .
let @xmath579 and @xmath580 .
then @xmath581 is a solution to the problem @xmath582 by the definition of @xmath573 , @xmath583 and hence @xmath584 lies off the boundary in the constraint set .
then @xmath581 is a solution to the foregoing problem with the constraint @xmath585 removed .
the problem is of the form of maximizing a linear function of @xmath586 subject to an elliptical constraint .
straightforward calculation shows that @xmath587 and the achieved maximum value is @xmath588 , where @xmath589 @xmath590 .
finally , we show that the sequence @xmath591 is nonincreasing : @xmath273 , where the equality holds if and only if @xmath592 . because @xmath593 or @xmath594 , this result implies that @xmath595 and hence @xmath268 is a unique solution to ( [ opt ] ) .
let @xmath596 so that @xmath597 . by the identity
@xmath598 and simple calculation , @xmath599 \frac { d_{k+1}^{-1}}{\sum_{j=1}^{k+1 } d_j^{-1 } } \nonumber \\[-6pt ] \\[-10pt ] & = & d_{k+1 } \frac { \ { r_k - ( k-2 ) \}^2}{r_k ( r_k+1)},\nonumber\end{aligned}\ ] ] where @xmath600 .
therefore , @xmath601 .
moreover , @xmath602 if and only if @xmath603 , that is , @xmath604 .
proof of corollary [ cor3 ] it suffices to show ( [ sol - ineq ] ) . by corollary [ cor2 ] , @xmath605 and hence @xmath606 .
then for @xmath294 .
@xmath607 because @xmath608 for @xmath609 .
proof of theorem [ th3 ] let @xmath610 so that @xmath611 , similarly as in the proof of theorem [ th2 ] . by equation ( [ a5 ] ) with @xmath612 and @xmath613 replaced by @xmath614 , @xmath615 by the relationship @xmath616 and simple calculation ,
@xmath617 if @xmath340 , combining the two preceding equation gives @xmath618 the first inequality follows because @xmath619 for @xmath620 and @xmath621 is increasing for @xmath622 with a maximum at @xmath623 .
the second inequality follows because @xmath624 .
therefore , if @xmath340 then @xmath625 if @xmath338 , then @xmath626 and hence @xmath627 this completes the proof .
the author thanks bill strawderman and cunhui zhang for helpful discussions . | consider the problem of estimating a multivariate normal mean with a known variance matrix , which is not necessarily proportional to the identity matrix .
the coordinates are shrunk directly in proportion to their variances in efron and morris ( _ j .
amer .
statist .
assoc . _
* 68 * ( 1973 ) 117130 ) empirical bayes approach , whereas inversely in proportion to their variances in berger s ( _ ann .
statist .
_ * 4 * ( 1976 ) 223226 ) minimax estimators .
we propose a new minimax estimator , by approximately minimizing the bayes risk with a normal prior among a class of minimax estimators where the shrinkage direction is open to specification and the shrinkage magnitude is determined to achieve minimaxity .
the proposed estimator has an interesting simple form such that one group of coordinates are shrunk in the direction of berger s estimator and the remaining coordinates are shrunk in the direction of the bayes rule .
moreover , the proposed estimator is scale adaptive : it can achieve close to the minimum bayes risk simultaneously over a scale class of normal priors ( including the specified prior ) and achieve close to the minimax linear risk over a corresponding scale class of hyper - rectangles .
for various scenarios in our numerical study , the proposed estimators with extreme priors yield more substantial risk reduction than existing minimax estimators . |
theories which violate lorentz symmetry can result in a modified photon dispersion relation which manifests itself in an energy ( or frequency ) dependent speed of light in vacuum @xcite . at high energies , this vacuum speed of light could depend , to leading order , linearly or quadratically upon photon energy , and this energy dependence could either decrease or increase with energy .
such modified dispersion relations are common in theories of quantum gravity @xcite . as an experimental consequence ,
simultaneously emitted photons of different energies traveling over a long baseline will become temporally separated .
the energy dependence of the arrival time of photons from gamma ray bursts ( grbs ) has been used to place limits on the energy scale of quantum gravity or the scale at which possible liv may occur @xcite .
grbs are bright , violent bursts of high energy photons lasting on the order of fractions to hundreds of seconds , making them an apt source for study ; indeed , their brightness makes observation at high redshift possible . from grb 080916c , the observation of @xmath4(gev ) photons probes liv effects at the planck scale @xcite . though the matter density of space is low , it is not truly a vacuum .
matter comprises nearly thirty percent of the universe s energy budget @xcite , and over large scales , this matter density is relatively uniform @xcite .
the overwhelming majority of this matter is some yet unknown dark matter ( dm ) . through its gravitational interaction , the existence of dm
is established from a concordance of astronomical observations from different cosmological epochs @xcite .
little is known about dm except for the fact that it is not relativistic , must be stable on the scale of billions of years , and interacts very little through non - gravitational channels . since the universe does contain matter , when taken _ en masse _ , this engenders the cosmos with an index of refraction .
rather generically , this index of refraction is dispersive ; that is , the phase velocity for light traveling through the medium is frequency dependent .
similar to liv effects , light traveling over long baselines through a dispersive medium will suffer a frequency - dependent time lag .
these matter effects could confound potential claims of liv gleaned from grb photon arrival times .
herein , we assess the dispersive effects of the matter in the cosmos .
since dm comprises the bulk of the matter in the universe , we will consider it in detail . at the particle level
, the index of refraction of a bulk medium arises from the forward scattering of photons on the particles that comprise the medium .
photons directly couple to charged particles , yet the refractive index for neutral particles is also typically nonzero .
we emphasize that a dark - matter particle need not have an electric charge to scatter a photon .
for example , it can be electrically neutral but possess an electric or magnetic dipole moment , but such moments are not generic properties of dm candidates . indeed
, majorana fermions can not have permanent dipole moments , and the existence of a permanent electric dipole moment requires time - reversal and parity violation .
for the case in which the dm has no permanent dipole moments , the dm medium can still have a nonzero polarizability and , hence , index of refraction , provided that the dm particle couples to other electromagnetically charged particles . classically , the polarizability of a substance is a measure of an external electric field s ability to induce dipole moments in the molecules that comprise the substance . in the simplest case ,
say for a spherically symmetric atom , an external electric field exerts opposite forces on the protons and electrons , resulting in a nonzero dipole moment parallel to the external field . at leading order ,
this substance s polarizability is proportional to the applied electric field ; however , higher order contributions depend quadratically , etc .
, on the field strength .
if a structureless neutral particle couples to electrically charged particles , then an electric field can induce an analogous effective dipole moment , despite the fact that the neutral particle is not composed of charged constituents .
many dm candidates do couple to charged particles .
in fact , some indirect searches for dm involve the detection of high energy photons produced in a dm - dm annihilation event .
since the compton amplitude can be related by crossing symmetry to the amplitude for dark - matter annihilation into two photons , then any dark - matter model which gives rise to an indirect detection signal via @xmath5 @xcite can also drive the index of refraction of light from unity . for a given dm model
, one must compute the forward compton amplitude to determine the particle s contribution to the refractive index ; however , in a low photon - energy expansion , some model independent statements can be made about the amplitude .
if charge conjugation , parity , and time - reversal symmetries are conserved , then the leading order contributions to the forward scattering amplitude are model independent , attributable to the charge , mass , and magnetic dipole moment of the scatterer @xcite .
should these discrete symmetries be violated , then additional contributions to the forward scattering amplitude arise . by including parity- and time - reversal violating terms in the hamiltonian , a more general form for
the scattering amplitude has been explored in ref .
@xcite . in general , these new terms are model dependent , and it turns out that only parity violating interactions contribute to the amplitude in the forward limit .
such parity - violating terms can account for permanent electric dipole moments and anapole interactions . in ref .
@xcite , we considered the leading order behavior of the forward compton amplitude and used limits from the non - observation of dispersion attributable to charged matter to place constraints on the charge to mass ratio of sub - ev dark matter particles .
herein , we consider higher order contributions to the index of refraction .
in particular , we compute higher order contributions to the index from millicharged matter , namely its quasi - static polarizability so called because a charged particle s polarizability diverges in the ( static ) limit of vanishing frequency @xcite .
we also consider two models of neutral dark matter the boehm - fayet ( bf ) scalar dm model @xcite and the neutralino of the minimal supersymmetric standard model ( mssm ) .
we compute the full forward compton amplitude to first order in perturbation theory for both dm candidates ; this will result in the polarizability of the dm medium . at low photon energies ,
we find that the explicit computations yield the form expected from purely analytic considerations .
the electromagnetic form factors of dm have been considered elsewhere .
stringent bounds on the fractional charge of more massive ( mev to gev ) dark matter candidates have been established by requiring that the charge not be so large as to affect the density perturbations at recombination from the cmb spectrum @xcite .
additionally , bounds on the electric and magnetic dipole moments have been established through both theoretical and experimental considerations including dm relic abundance , non - observation of dm - nuclei interactions in direct detection experiments , precision measurements , and non - production at colliders @xcite .
others have determined bounds on the polarizability of dm which sets the scale for the refractive index in the static limit , i.e. , for low photon energies . in ref .
@xcite , the authors used direct detection searches to limit the polarizability of weakly interacting massive particles with masses on the order of 100 gev . additionally , in ref .
@xcite , the authors take an effective - field - theory approach to constrain the interaction between dark matter and the electroweak gauge bosons .
their constraints on the effective coupling for a photon - dm contact term can be mapped to the polarizability of the dm particle . as for these polarizability constraints ,
the leading order term only sets the scale for the refractive index s deviation from unity because it is independent of omega . to assess , the dispersive nature of the dm medium , higher order contributions to the polarizability
are needed , which we compute below .
for neutral matter , dispersion rather generically depends quadratically upon photon energy . because of collider constraints on the existence of heavy of charged , exotic particles ,
the dispersion due to neutral dm is extremely small , and grb photon arrival time currently provides no meaningful constraints on dm models .
the redshift dependence for time lags due to a dm medium is distinct from those due to liv effects , yet it is possible that matter effects could confound claims of the existence of a modified photon dispersion relation that is quadratic in energy .
herein , we determine the energy scale at which matter effects become comparable in size with such liv effects . as for dispersive matter effects which scale linearly with energy , we find that these arise if there are parity violating terms in the forward scattering amplitude .
these terms can produce a circularly birefringent medium ; that is , the index of refraction is different for right- and left - handed polarized light .
these effects could potentially be confounded with a modified photon dispersion relation which scales linearly with energy , but based upon one model dependent calculation , these matters effects appear to be small .
the relationship between the index of refraction @xmath6 and the coherent forward scattering amplitude @xmath7 for light of frequency @xmath2 and polarization @xmath8 is given by the lorentz relation @xcite @xmath9 where @xmath10 denotes the number density of scatterers , and we work in units of @xmath11 throughout .
the relationship can be derived by considering the transmission of a wave through a thin slab of material at rest @xcite , where the deviation of @xmath12 from unity arises from the interference between the incident and scattered waves . to relate the completely coherent quantum mechanical forward scattering amplitude , @xmath7 , to the analog in quantum field theory , @xmath13 ,
we compute the differential cross section for a generic forward scattering event in the scatterer s rest frame and match it with @xmath14 . using the conventions of ref .
@xcite , we determine @xmath15 , where the overall phase is fixed by demanding that the field theoretic amplitude obey the optical theorem .
we thus have @xmath16 with @xmath17 the mass density of the scatterers .
note that the amplitude is evaluated in the dark - matter rest frame so that @xmath18 and @xmath19 where @xmath20 points in the direction of light propagation . to determine the index of refraction for a specific dark matter model
, one must compute the forward compton amplitude for that model ; however , there are significant model - independent features of the index of refraction that we wish to highlight . assuming lorentz invariance and conservation of the discrete symmetries
@xmath21 , @xmath22 , and @xmath23 , we can factor the forward compton amplitude as follows @xcite @xmath24 where @xmath25 is the spin operator associated with the dark - matter particle and @xmath26 ( @xmath27 ) is the polarization vector associated with the photon in its initial ( final ) state .
we note that the photon is transverse , so that @xmath28 .
if these discrete symmetries are violated , then additional terms may appear in the forward compton amplitude . in ref .
@xcite , we find the general structure of the compton amplitude for processes which violate time reversal symmetry .
for real , transverse photons , @xmath23-violating processes do not contribute to the amplitude in the forward limit .
on the other hand , parity violating interactions do contribute . from refs .
@xcite , we find @xmath29 at leading order in a perturbative calculation of the forward compton amplitude , there exists a nonzero threshold photon energy @xmath30 at which virtual particles in the scattering process go on mass shell . below this inelastic threshold ,
the functions @xmath31 and @xmath32 are shown to be both real and even and odd , respectively , under the map @xmath33 . additionally , in a power series expansion of these functions about @xmath2 , it is shown that their leading order behavior is determined by the static properties of the scatterer , namely its charge , mass , and magnetic moment .
in particular , we have @xmath34 where @xmath35 is the charge of the scatterer ( in units of @xmath36 ) and @xmath37 is its magnetic moment .
this low energy behavior holds even if the scatterers are loosely bound composite particles @xcite . for the @xmath22-violating terms in the amplitude , the leading order behavior of the functions @xmath38 and
@xmath39 are model dependent . as an example , for photon - nucleon scattering , the leading order behavior for these functions , in the center - of - mass frame , is @xcite @xmath40 here , @xmath41 is the nucleon - pion coupling constant ; @xmath42 is the @xmath22-violating isovector pion - nucleon coupling ; @xmath43 is the nucleon mass ; @xmath44 is the pion decay constant ; and @xmath45 is the neutron magnetic moment in units of nuclear magnetons . in an expansion about @xmath2 , the leading order contribution to
the scattering amplitude takes the same form in both the center - of - mass frame and the scatterer s rest frame ( though this is _ not _ true for higher order terms in the expansion ) .
this low - energy expansion is valid for photon energies well below the inelastic pion threshold , @xmath46 mev .
we note that the terms in the forward amplitude , eqs .
( [ for_amp],[pviol_amp ] ) , which depend on the cross product of the polarization vectors , @xmath47 , change the polarization of the incident photon .
if the incident electromagnetic wave is linearly polarized , then scattered waves of this ilk can not completely coherently interfere with the incident wave , which is why , per eq .
( [ qft_n ] ) , these terms do not contribute to the index of refraction .
however , we can extend our considerations to include optically active media . in this case , the @xmath32 and @xmath48 terms result in birefringence for circularly polarized light ; that is , the index of refraction , @xmath49 , for postive - helicity photons differs from that , @xmath50 , for negative - helicity photons .
let us suppose that the electromagnetic wave is traveling in the direction @xmath51 ; then , it can have linear polarization @xmath52 or @xmath53 .
but , we could equally as well consider states of circular polarization @xmath54 . in this basis , the @xmath22-violating forward compton amplitude is @xmath55 with @xmath56 .
hence , the phase speed of the photons differs according to helicity . for the moment , we focus upon the amplitude which conserves the discrete symmetries , eq .
( [ for_amp ] ) . as for @xmath31 , by imposing the additional assumption of causality and using the analyticity and unitarity of the scattering matrix , dispersion relations emerge for @xmath31 and @xmath32 @xcite @xmath57 where we use the optical theorem to replace @xmath58 with the unpolarized cross section @xmath59 .
we define @xmath60 . where @xmath61 ( @xmath62 ) is the cross section for parallel ( antiparallel ) photon and scatterer spins .
the low energy theorem sets @xmath63 and @xmath64 .
recall , at leading order in perturbation theory , @xmath31 and @xmath32 are real for @xmath65 , so that the lower limit of this integral is merely formal , since @xmath66 below the inelastic threshold ( at the same order in perturbation theory ) . as such ,
the integral actually commences at @xmath67 so that it is well - posed at @xmath68 .
the relationship for @xmath31 is the quantum field theoretic analogue of the kramers - kronig @xcite relationship of optics . in the limit @xmath69 , we can expand @xmath31 , eq .
( [ disp_rel_g ] ) , as a series of even powers of @xmath2 with _ positive _ coefficients @xmath70 , we thus have @xmath71 where @xmath72 and @xmath73 . the terms in @xmath74 and
higher are associated with the polarizability of the dark - matter candidate .
this is the expected refractive index for a medium which is not optically active .
again , for @xmath75 , we can expand @xmath32 , eq .
( [ disp_rel_h ] ) , as a power series in @xmath2 . below threshold , this yields a series of odd powers in @xmath2 . inserting this into the index formula
, we see that we can achieve a birefringent term that scales as @xmath76 ; this term is associated with the scatterer s anomalous magnetic dipole moment .
higher order terms can result in birefringence linear @xmath2 . additionally , the leading term in the parity - violating contribution from nucleons , eq .
( [ hslash_lot ] ) , yields an odd power of @xmath2 in the refractive index ; in this case , the dependence is linear .
thus , we see that in optically active media there can be terms in the index of refraction which are linear in @xmath2 , similar to some quantum gravity models . the assumptions that lead to the model - independent form of the index of refraction in eq .
( [ n_expans ] ) are actually too restrictive for our purposes as even a standard model background of particles is not symmetric under the discrete symmetries . as an example , consider a background of neutrinos and antineutrinos . at the lagrangian level of this system , we can , to a good approximation ,
take @xmath77 as conserved ; however , given the chiral nature of the weak interactions , parity is maximally violated .
if the number density of neutrinos differs from that of the antineutrinos in the background , then this medium will be birefringent and odd powers of @xmath2 will appear in the index of refraction @xcite . such a background is asymmetric under a @xmath77 transformation .
if this asymmetry is removed by equalizing the neutrino and antineutrino number density , then the medium is no longer optically active and , below threshold , only even powers of @xmath2 are present in the refractive index , consistent with eq .
( [ n_expans ] ) . in general
, it was shown in ref .
@xcite that both @xmath22 _ and _ @xmath77 asymmetry must be present in either the lagrangian or background ( or both ) in order for the index of refraction to be birefringent with odd powers of @xmath2 ; failing that , eq . ( [ n_expans ] ) will hold . from a sm standpoint , it is easy to conceive that the background matter is @xmath77 asymmetric
after all we live in a universe predominantly comprised of _ matter _
antimatter_. in order to achieve an optically active medium , we merely need either @xmath22 violating processes in the light - matter interactions as in eq .
( [ pviol_amp ] ) or a mechanism to make the background @xmath22 asymmetric .
an example of such a background would be a polarized medium which could occur , for instance , in a region containing a background magnetic field @xcite . in that case , one would expect leading order contribution to @xmath32 , attributable to the particle s magnetic dipole moment , to induce birefringence .
we now turn to the manner in which we can realize observational constraints on the cosmic index of refraction .
the refractive index generated by the completely coherent portion of the forward compton amplitude is generic to all matter , so we will confine our attention , for the moment , with the observation procedure one may use to realized constraints on the coefficients @xmath78 .
we will return to a discussion of an optically active medium later .
the index of refraction fixes the phase velocity @xmath79 in the medium ; it is @xmath80 with @xmath81 .
observable dispersive effects in light propagation are controlled , rather , by the group velocity @xmath82 , for which @xmath83 .
thus the light emitted from a source at a particular time at a distance @xmath84 away possesses a frequency - dependent arrival time @xmath85 , namely @xmath86 , or @xmath87 for sources at cosmological distances , we must account for the impact of an expanding universe on the arrival time @xcite .
as we look back to a light source at redshift @xmath88 , we note that the dark - matter density accrues a scale factor of @xmath89 , whereas the photon energy is blue shifted by a factor of @xmath90 relative to its present - day value @xmath91 @xcite .
thus we have the arrival time @xmath92 for light of observed angular frequency @xmath91 from a source with red shift @xmath88 : @xmath93 \label{cosmictlag}\ ] ] with the hubble rate @xmath94 .
the cosmological parameters determined through the combined analysis of wmap nine - year data in the @xmath95cdm model with distance measurements from type ia supernovae ( sn ) and with baryon acoustic oscillation information from the distribution of galaxies can be found in ref .
the hubble constant today is @xmath96 , whereas the fraction of the energy density in matter relative to the critical density today is @xmath97 and the corresponding fraction of the energy density in the cosmological constant @xmath95 is @xmath98 @xcite .
not only do distant astrophysical sources provide an increased baseline over which frequency - dependent time lags can develop ; the photon frequency and dark matter density s dependence upon redshift provide an additional cosmological lever arm .
for the coefficient @xmath78 in eq .
( [ cosmictlag ] ) , the integral over redshift is given by @xmath99 so that higher order @xmath100 have more factors of @xmath90 in the numerator .
various strategies must be employed to isolate the coefficients @xmath78 in eq .
( [ cosmictlag ] ) . to start we note that the @xmath101 term has no frequency dependence , so that its effects are unobservable with our method , though its consequences have been explored in ref .
the remaining terms can be constrained by dispersive effects .
the term that scales as @xmath102 is best constrained through the arrival time difference of a gamma - ray pulse from a grb and its radio afterglow @xcite .
the terms in positive powers of @xmath91 are best constrained through arrival time differences between optical and gamma - ray pulses of varying energy .
though the observational program can be carried out independent of a particular dark matter model , we compute the forward compton amplitude for three explicit particles : a millicharged fermion , the bf scalar dm particle , and mssm s neutralino .
these explicit models will demonstrate that the generic analytic behavior of the refractive index holds .
millicharged particles arise naturally in models where the sm photon kinetically mixes with a new massless gauge boson @xcite . as discussed earlier , there are limits on heavy millicharged dm @xcite .
a summary on the existence of particles with fractional electric charge can be found in ref .
@xcite . in ref .
@xcite , we constrain the electric charge of light dark matter through the non - observation of dispersion due to charged matter . for matter of charge @xmath103 and mass @xmath104 ,
the leading order contribution to the forward compton amplitude is set by the low energy theorem to be @xmath105 where we define @xmath106 .
this is the result upon evaluating the feynman diagrams for forward scattering on a charged fermion , fig .
[ fig1 ] . in this figure ,
the second diagram is the crossed - photon " version of the first diagram ; we will omit the crossed " diagrams in the figures that follow .
contribution to the forward compton amplitude for a charged fermion @xmath107 .
the second diagram is the crossed - photon " version of the first diagram .
[ fig1],width=288 ] higher order contributions to the amplitude will yield terms which are dependent upon the photon energy .
such contributions are model dependent , requiring input from the structure of the scatterer and its coupling to other particles .
we make no assumptions of structure or couplings and only calculate the nave @xmath108 contribution to the forward scattering amplitude , @xmath109 .
a word of caution is in order because the low energy expansion of the forward amplitude , @xmath110 , is not relevant at @xmath108 for a structureless charged particle .
this is because the inelastic threshold for a photon scattering on a charge particle commences at @xmath111 at this order in perturbation theory @xcite .
this follows from a simple application of the optical theorem .
the optical theorem relates the total cross section for a scattering process to the imaginary part of the forward amplitude @xmath112 where @xmath113 is the energy in the center of momentum frame and @xmath114 is the momentum of one the particles in this frame . at low photon energies ,
the thomson cross section is valid for this scattering process @xmath115 . using this cross section
, we would expect that the imaginary part of the forward amplitude , at low photon energies , to be @xmath116 . since @xmath117 is proportional to @xmath2 in the limit of vanishing photon energy ,
then it is clear that the inelastic threshold for photon scattering is @xmath111 at this order in perturbation theory . in fig .
[ fig2 ] , we present the feynman graphs needed to determine @xmath109 ; note , the graphs with crossed " photons are not shown .
the infrared divergences generated by the internal photon lines are regulated by giving the photon a small fictitious mass @xmath37 ; however , upon summing all the diagrams , this photon mass vanishes in the forward limit so that the physical process is independent of @xmath37 .
the ultraviolet divergences generated by the vertex and propagator corrections are regulated , upon renormalization , by the vertex and propagator counterterms .
we use form @xcite to aid our algebraic manipulations .
contribution to the forward compton amplitude for a charged fermion @xmath107 . the graphs with
crossed " photons are not shown .
[ fig2],width=288 ] to set the notation , let the fermion four momentum be @xmath118 with @xmath119 and the photon four momentum be @xmath120 with @xmath121 .
the real part of the @xmath109 contribution to the forward compton amplitude is @xmath122 \label{rech}\end{aligned}\ ] ] with the usual mandelstam variables @xmath123 and @xmath124 .
we employ the definition of the dilogarithm found in ref .
@xcite @xmath125 for completeness , we include the imaginary part of the amplitude @xmath126.\ ] ] as a check , we reproduce the thomson cross section via the optical theorem .
working in the scatterer s rest frame with @xmath127 and @xmath128 , we find that , in the limit of vanishing photon energy @xmath129 , the imaginary part of the amplitude behaves as @xmath130 , which agrees with our expectations . returning to the real part of the amplitude , we will focus upon the leading order contributions to the polarizability of the fermion ; in particular , in an expansion about @xmath2 , we find @xmath131 . \label{quasipol}\ ] ] the expression in eq .
( [ quasipol ] ) which is proportional to @xmath1 can only define a quasi - static polarizability due to the presence of the @xmath132 term which leads to a divergence in the static limit @xcite ; we note that our expression for the quasi - static polarizability extracted from eq .
( [ rech ] ) is equivalent to the analogous expression in refs .
@xcite . calculating the local group speed for light traveling in the charged medium
, we have to @xmath108 @xmath133 to assess the effect of the correction upon the group velocity , we estimate its size separately for two regimes , small and large @xmath2 . for @xmath134 ,
one has @xmath135.\ ] ] for small @xmath2 , we see that a logarithmic term modifies the usual dispersion of light in the medium .
the variation in the logarithm is rather slow relative to the leading @xmath136 dispersion ; furthermore , the logarithm term is suppressed by the additional factor @xmath137 relative to the leading order term in the group velocity .
this correction will have little impact upon the arrival time of radio frequency waves from a distant grb .
recalling the analysis from ref .
@xcite , we sought a frequency dependent time lag between the arrival of high energy photons from a grb and its radio afterglow . from the non - observation of such a time lag ( consistent with millicharged dark matter )
, we limit the charge - to - mass ratio of dark matter to be @xmath138 ev@xmath139 . to assess the size of the correction to the leading order dispersive factor , consider a 1 ghz radio wave which corresponds to a photon energy of @xmath140 ev ; then , the next term is negligible , representing a percent correction of , at most , @xmath141 multiplied by the logarithm .
we now consider the other extreme of large photon energies , namely @xmath142 . in this limit
, the mandelstam variables behave as @xmath143 and @xmath144 .
we recall the behavior of the dilogarithms from ref . @xcite . for @xmath145 ,
one has the limit @xmath146 and for @xmath147 , one finds @xmath148 using eqs .
( [ diloglim1],[diloglim2 ] ) in the amplitude eq .
( [ rech ] ) , the correction to the group velocity in the high energy limit behaves as @xmath149 .
\label{hi_e_charged}\ ] ] for sufficiently large @xmath2 , this correction term will dominate the leading order term in the dispersion relation .
crudely , we may estimate that for photon energies @xmath150 $ ] the correction @xmath151 will be the dominant source of dispersion .
again , from ref .
@xcite , we limit the charge - to - mass ratio of dark matter to be @xmath138 ev@xmath139 . to assess the smallest value of @xmath2 where the dispersive corrections will dominate , we consider the unrealistic possibility of a particle with unit electric charge @xmath152 , requiring @xmath153 ev .
then , the correction to the @xmath0 dispersion dominates for a photon energy of @xmath154 ev ; this is a prohibitively high energy , orders of magnitude greater than the gzk cutoff @xcite .
alternatively , we consider the size of the correction for more reasonable photon energies , @xmath155 tev . for @xmath156 and @xmath157 ev , we find the correction to the usual @xmath0 behavior to be on the order of a few parts in @xmath158 . as such , for millicharged dm , these higher order corrections are not appreciable for the photon energies under consideration .
this higher order process in perturbation theory is nave in that we assume a theory with only a charged fermion and a photon , i.e. , qed . if there were additional fields in the theory , then the high energy behavior of the dispersion would certainly be modified .
for instance , if the lagrangian contained an interaction term like @xmath159 which coupled the charged fermion @xmath107 to some other charged massive fermion @xmath160 and scalar @xmath161 , then photon scattering mediated by the new fermion and scalar would contribute terms dispersive terms of order @xmath162 . for these processes , the arrival time formula in eq .
( [ cosmictlag ] ) is valid as the inelastic threshold of this process is set by the mass of the particles @xmath161 and @xmath160 ; these computations will mirror the results for the neutral dark matter considered below .
alternatively , we could imagine that our charged fermion @xmath107 is a composite , rather than point - like , particle ; its composite nature would also impact the particle s polarizability .
these results can be applied to standard model particles as well . the baryonic energy density relative to the critical density in the universe today
is @xmath163 @xcite , roughly 15% of the entire matter in the universe , and the overwhelming majority of this matter consists of atomic hydrogen . in the long wavelength limit , electromagnetic radiation can not resolve the individual charged constituents of a neutral atom , so neutral hydrogen will not appreciably contribute to the refractive index for charged particles in the low photon energy limit . in the high energy limit , well beyond the ionization energy ,
neutral hydrogen can be considered as a plasma of protons and electrons . from eqs .
( [ vg_correction ] ) and ( [ hi_e_charged ] ) , we see that the less massive electrons will contribute more to dispersion than protons .
but , as before , for reasonable photon energies , the correction to the @xmath164 term in the refractive index is of little consequence . since the protons are composite particles , their interactions with photons are richer .
indeed , the parity - violating terms , eq . ( [ hslash_lot ] ) , in the forward compton amplitude can result in an @xmath165 term in the index of refraction . using the helicity basis for the photon polarization eq .
( [ pviol_amp_circ ] ) , we can write the @xmath22-violating contribution to the proton index of refraction as @xmath166.\ ] ] if the proton medium is unpolarized then the spin - dependent terms will average to zero @xmath167 . from the low - energy expansion in eq .
( [ hslash_lot ] ) , we thus have @xmath168 from ref .
@xcite , we find the values @xmath169 , @xmath170 mev , @xmath171 , @xmath172 . for photon energies below the pion threshold , @xmath46 mev
, we have @xmath173 as a point of comparison , we note that relic neutrinos can render the cosmos optically active , though on a vastly different scale . for relic neutrinos at a temperature of 2 k ,
the difference between the index of refraction for positive and negative helicity photons is @xmath174 @xcite . at sufficiently high energy , the @xmath165 term in the index of refraction
should dominate the @xmath175 term , present for charged particles , in eq .
( [ n_expans ] ) .
for the explicit example considered here , the @xmath165 term for the proton is only valid below the pion inelastic threshold . for photon energies below this threshold , the @xmath175 term dominates ; as such , the explicit terms linear in @xmath2 can not be observed via the time - lag procedure . considering a broader class of particles
, we do expect that parity - violating terms in the forward compton amplitude can result in circularly birefringent medium arising from a dispersive term in the index of refraction that is linear in @xmath2 .
for instance , the interaction between a photon and the electron s anapole moment violates parity @xcite , and this should result in such a term for photons with energy below the weak scale . as an example of neutral dark matter
, we consider the @xmath176 susy - inspired model of light ( mev ) scalar dark matter proposed initially in ref .
@xcite by boehm and fayet and expounded upon in ref .
this model was introduced as a possible explanation for excessive 511 kev radiation , seen by the integral satellite , originating from our galactic center @xcite .
a comprehensive review of this topic can be found in ref .
regardless of the motivation for this model , we choose to study it in the present context because a medium consisting of scalar particles is optically simple , producing no birefringent effects .
additionally , this model requires only a few extra particles beyond the sm , so it is a natural starting point before considering the panoply of particles in the mssm .
the relevant new particles contributing to the index of refraction are the scalar dark - matter particle @xmath177 and a new massive fermion @xmath178 , both of which couple to a sm fermion such as an electron or quark @xmath179 .
the interaction term in the lagrangian is @xmath180 in ref .
@xcite , dm annihilation into two photons is considered via an effective four - point interaction by integrating out the heavy fermion @xmath181 @xmath182 where @xmath183 is the mass of the heavy fermion @xmath181 .
using this effective theory to explore the crossed process in which the photon forward scatters on the dm is fruitless as the amplitude is identically zero .
this is not surprising when one considers the low energy theorem for the forward compton amplitude . for a neutral candidate
, we expect the leading order contribution to be @xmath162 . as the dm and photon four - momenta , @xmath118 and @xmath184 , respectively , are the only external vectors involved in the scattering process
, terms which yield factors of @xmath2 must come from the lorentz - invariant product @xmath185 .
if the four point interaction in @xmath186 is used for the forward scattering process , then momentum flow through the electron loop does not involve @xmath118 ( at least , not in a non - trivial manner ) . as a result
, the effective theory can only render the @xmath187 term , which vanishes for a neutral scatterer .
to go beyond this approximation , we need to consider the full theory at the one - loop level , eq .
( [ boehml ] ) .
this involves calculating six one - loop feynman diagrams , including both circulations of the fermion loops .
three of the diagrams are pictured in fig .
( [ fig3 ] ) with clockwise circulation ; the remaining graphs are found by crossing the photon lines . ) .
the dashed lines represent the scalar dm ; the single fermion line represents the sm particle ; and the double fermion line represents @xmath181.,width=288 ] each closed fermion loop results in a trace over a product of fermion propagators and vertex operators , and we use the usual dirac trace technology to simplify these .
simple power counting shows that each individual diagram is logarithmically divergent . as such
, we use dimensional regularization to calculate its contribution to the amplitude . upon summing all terms
, the result is finite .
each graph yields a four - point integral over loop momentum ; however , in the forward limit , we can utilize various tricks , such as partial fraction decomposition , to express the integrals in terms of only three feynman parameters .
we work in the rest frame of the dm particle with @xmath188 and @xmath189 the four - momenta of the dm and photon , respectively ; we let @xmath190 be the mass of the sm fermion and @xmath183 the mass of the new heavy fermion .
the amplitude can be naturally grouped into terms which are either proportional to @xmath191 or @xmath192 .
we will decouple these accordingly in the amplitude @xmath193 .
we note that terms proportional to @xmath194 vanish identically for each diagram , and terms proportional to @xmath195 vanish when both orientations of fermion flow are considered . performing all but the last parameter integral , we find the following contributions to the scattering amplitude @xmath196 , \nonumber \\ \label{ampa } \\
\mathcal{m}_+ & = & -8 e^2 ( g^ * g + h^ * h ) \frac{1}{(4\pi)^2}\bigg\ { \int_0 ^ 1\mathrm{d}x \frac{1}{2 } \bigg [ \frac{1}{x}\log \bigg ( \frac{s_1(x)}{p_1(x ) } \bigg ) + \frac{1}{x}\log \bigg ( \frac{u_1(x)}{p_1(x ) } \bigg ) + \frac{1}{x}\log \bigg ( \frac{s_2(x)}{p_2(x ) } \bigg ) + \frac{1}{x}\log \bigg ( \frac{u_2(x)}{p_2(x ) } \bigg ) \bigg ] \nonumber \\ & & + [ m_f^2 -p^2 + m_f^2 ] \int_0 ^ 1 \mathrm{d}x \bigg [ \frac{-1}{p_1(x ) } + \frac{s}{2(p \cdot q)^2 } \log \bigg ( \frac{s_1(x)}{p_1(x ) } \bigg ) + \frac{u}{2(p \cdot q)^2 } \log \bigg ( \frac{u_1(x)}{p_1(x ) } \bigg ) \nonumber \\ & & + \frac{m_f^2 -s + m_f^2}{4(p \cdot q)^2}\bigg ( \frac{1}{x } \log\bigg ( \frac{s_1(x)}{p_1(x ) } \bigg ) + \frac{1}{x } \log\bigg ( \frac{s_2(x)}{p_2(x ) } \bigg ) \bigg ) + \frac{m_f^2 -u + m_f^2}{4(p \cdot q)^2}\bigg ( \frac{1}{x } \log\bigg ( \frac{u_1(x)}{p_1(x ) } \bigg ) + \frac{1}{x } \log \bigg ( \frac{u_2(x)}{p_2(x ) } \bigg)\bigg ) \bigg]\bigg\}. \nonumber \\ \label{ampc}\end{aligned}\ ] ] we define the polynomials @xmath197 where the usual mandelstam variables are @xmath198 and @xmath199 .
the polynomials @xmath200 , @xmath201 , and @xmath202 are defined in the same manner after swapping the masses of the charged fermions @xmath203 .
closed form solutions to the integrals are discussed in the appendix .
central to the low energy behavior of this amplitude is the claim that the leading order behavior is @xmath162 . to prove this
we shall focus upon @xmath204 ; the result for @xmath205 follows rather trivially from this . in eq .
( [ ampa ] ) , there are terms with prefactors of @xmath76 and @xmath0 , and one term completely independent of @xmath2 . after expanding the integrand as a power series in @xmath2 , we find that the @xmath76 and @xmath0 terms vanish but a nonzero @xmath187 term remains @xmath206 \int_0 ^ 1 \mathrm{d}x \bigg\ { \frac{4x^2 -4x + 1}{p_1(x)}- \frac{x(1-x)[2p^2x(1-x ) + ( m_f^2 -p^2 + m_f^2)]}{[p_1(x)]^2 } \bigg\ } + \mathcal{o}(\omega^2).\label{loint}\ ] ] however , the integral in eq .
( [ loint ] ) is exactly zero as the integrand is the derivative of a function which vanishes on the boundary @xmath207 \int_0 ^ 1 \mathrm{d}x \frac{\mathrm{d}}{\mathrm{d}x}\bigg(\frac{x-3x^2 + 2x^3}{p_1(x)}\bigg ) + \mathcal{o}(\omega^2).\ ] ] to assess the leading order contribution to the dispersive index of refraction , we make further approximations . the dm candidate of boehm _
et al_. has a mass of mev scale , and the new charged fermion @xmath181 has evaded detection in collider searches , such as lep @xcite , setting a lower bound on its mass @xmath208 gev .
given these masses , it is appropriate to consider the limit @xmath209 .
we approximate the amplitude in the limit @xmath210 @xmath211 + \mathcal{o}(\omega^6 ) \\
\mathcal{m}_+ & \approx & \frac{8 e^2 } { ( 4\pi)^2}(g^ * g + h^ * h)\bigg [ \frac{m_\text{dm}^2 \omega^2}{m_f^4 } \bigg(\frac{4}{3}\log \frac{m_f^2}{m_f^2 } -2 \bigg)+ \frac{m_\text{dm}^4 \omega^4}{m_f^8}\bigg(\frac{44}{15}\log \frac{m_f^2}{m_f^2 } -\frac{92}{9 } \bigg)\bigg ] + \mathcal{o}(\omega^6 ) \label{ampcapprox},\end{aligned}\ ] ] where , for terms which are @xmath212 , we only keep terms up to @xmath213 .
we note that the @xmath162 term of the forward compton amplitude , @xmath205 , for the bf model is similar in structure to the analogous term in photon - neutrino scattering @xcite .
photon - neutrino scattering is effected , at the leading order in perturbation theory , by one - loop diagrams involving virtual @xmath214 bosons and electrons . per ref .
@xcite , neglecting terms of order @xmath215 , we find @xmath216.\ ] ] though the neutrino is a spin-1/2 particle , the leading order behavior of the forward compton amplitude should exhibit a structure similar to the results from the bf model .
this is a useful check for our calculations .
returning to the bf model , relative to @xmath205 , the @xmath204 portion of the amplitude is suppressed by an additional factor of @xmath217 . inserting this amplitude into the local index of refraction governing the phase speed , eq .
( [ n_expans ] ) , we see the dark matter polarizability scales as @xmath218 , and the leading dispersive term of @xmath162 has a coefficient proportional to @xmath219 , highly suppressing dispersion . at the one - loop level , the index of refraction is real for @xmath220 . with the assumption @xmath209 ,
the photon energy needed to approach this threshold energy is @xmath221 . as a point of reference , for @xmath222 gev and @xmath223 mev , this photon threshold energy is around a 36 tev . given this , our @xmath162 approximation of the refractive index is quite good for the photon energies accessible by the fermi telescope @xcite . from the full expression for the amplitude , eq .
( [ ampc ] ) , we plot in fig .
[ fig4 ] the deviation of the real part of the index of refraction from unity for various values of dm mass with @xmath222 gev and @xmath224 .
to be definite , we set @xmath225 and @xmath226 .
we employ the usual value for the average dark matter density @xmath227 gev/@xmath228 .
the curves in fig .
[ fig4 ] exhibit typical behavior for the index of refraction around a resonance .
the cusps in the curves occur at the inelastic threshold at this order in perturbation theory . above this threshold energy ,
the imaginary part of the refractive index becomes nonzero .
below the threshold energy , the real part of the refractive index increases with photon energy , and beyond the threshold the medium exhibits anomalous dispersion in which the real part of the refractive index decreases as the photon energy increases .
mev ; the dashed ( red ) line employs @xmath229 mev ; the dotted ( blue ) line employs @xmath230 mev . for all curves @xmath222 gev and @xmath224 .
the peaks in each curve signal the appearance of the inelastic threshold at this order in perturbation theory . at this threshold energy ,
the imaginary part of the index of refraction is first nonzero , and the real part exhibits anomalous dispersion beyond this energy .
[ fig4],width=288 ] though the travel time over cosmological scales will be enhanced due to density and blue - shift factors , the timing resolution needed to detect these time lags is prohibitive . as a best case scenario ,
we calculate the time difference between the arrival of a low energy and 1 tev photon for a source at the distant redshift @xmath231 .
we use the group velocity with the approximate @xmath232 amplitude in eq .
( [ ampcapprox ] ) and the parameters indicated above with @xmath223 mev .
namely , we calculate @xmath233 where @xmath234 we note that @xmath235 mpc and @xmath236 gev@xmath237 ; the resulting time lag is prohibitively small , @xmath238 s. as is apparent in fig .
[ fig4 ] and eq .
( [ ampcapprox ] ) , the scale of the dispersive effects is set by the polarizability term which is , for the present limits , independent of @xmath239 and controlled primarily by a term of the form @xmath218 .
there is an additional logarithmic enhancement of the polarizability dependent upon the ratio @xmath240 ; however , this enhancement is relatively small . in a supersymmetric theory , if r - parity is conserved , then the lightest supersymmetric particle ( lsp ) is a natural dark matter candidate since it is stable to decay .
a popular candidate for the lsp is the lightest neutralino , a majorana fermion .
a neutralino @xmath241 is a linear combination of the four neutral gauginos : the bino @xmath242 , wino @xmath243 , and higgsinos @xmath244 , the superpartners to the sm neutral bosons . in fact , the neutralinos are the mass eigenstates of these gauginos / higgsinos . using the notation of ref .
@xcite , the matrix , @xmath245 , relating the mass eigenstates to the gauginos / higgsinos is defined by @xmath246 the lightest neutralino corresponds to @xmath247 ; however , we shall drop the subscript and denote the lightest neutralino as @xmath248 .
since this dark matter candidate is spin-1/2 , this admits the possibility of a birefringent medium .
recalling the requirements for birefringence @xcite , both @xmath22 and @xmath77 asymmetry must be present in the photon - neutralino interaction and/or in the neutralino medium itself . since majorana fermions are their own antiparticles , the only way for the neutralino medium to be @xmath22 or @xmath77 asymmetric is for the medium to be polarized .
if we assume an unpolarized medium , then the conditions necessary for birefringence must be satisfied by the photon - neutralino interaction ; however , in the forward scattering limit , the process must be @xmath23 ( or @xmath77 ) symmetric . in what follows , we will only consider an unpolarized medium so that the refractive index will have the form of eq .
( [ n_expans ] ) .
a full one - loop calculation of the amplitude for neutralino annihilation into two photons has been reported in refs .
the amplitude in these papers is computed in the rest frame of neutralinos so that we can not simply make use of crossing symmetry to determine the forward compton amplitude .
here , we will compute , in steps , the full one - loop forward compton amplitude in the neutralino rest frame .
first , we will consider the contribution to the amplitude for processes involving virtual fermion and sfermions in the loop , and we include , in appendix b , the remaining contributions which involve virtual @xmath249 bosons , charginos , and higgs particles . since the neutralino is a majorana particle , the usual feynman rules for dirac fermions must be adapted ; we follow the procedure outlined in refs .
@xcite by denner _
et al._. alternate rules for dealing with majorana fermions exist , e.g. , ref .
@xcite , but either set of rules will yield equivalent results .
evaluating an individual diagram will result in a scalar product which depends upon the initial and final spinor states of the neutralino .
since we are considering completely coherent forward scattering , evaluation of an individual diagram can take the form @xmath250 , for positive - energy states @xmath251 of spin @xmath252 with @xmath253 some product of dirac gamma functions .
since we are considering an unpolarized neutralino medium , we would like to average over the possible spin states . by averaging over spin states
, we can turn our scalar products into traces @xmath254 , \ ] ] where we have used the completeness relation @xmath255 . at this point , we can employ the usual dirac trace technology to simplify the traces . *
fermion - sfermion contribution * some of the feynman diagrams for neutralino - photon forward scattering which involve virtual fermion and sfermions are shown in fig . [ fig5 ] .
namely , the dashed lines represent scalar particles ( sfermions ) ; the solid lines with arrows represent dirac fermions ; and the wavy lines represent spin-1 bosons ( photons ) .
[ fig5],title="fig:",width=288 ] in addition to the graphs pictured , we need to consider both orientations of the virtual fermion line @xmath256 and the graphs with the crossed " photons .
we adopt the notation of ref .
@xcite for the coupling constants .
in particular , the neutralino - sfermion - fermion vertex is expressed as @xmath257 , where the charged fermion s momentum flows away from the vertex and @xmath258 . as before ,
we represent the photon four - momentum as @xmath184 with @xmath259 , and the neutralino momentum is @xmath118 with @xmath260 .
we denote the fermion mass as @xmath190 and the sfermion mass as @xmath261 ; the charge of this fermion is @xmath262 .
the contribution from the fermion - sfermion graphs to the forward compton amplitude is @xmath263 \nonumber \\ & & + \frac{1}{p\cdot q } \bigg [ [ m_{\tilde{f}}^2- m_{\tilde\chi^0}^2 + m_f^2 ] \bigg ( \frac{1}{s_1(x ) } -\frac{1}{u_1(x ) } \bigg ) - \frac{1}{2x } \bigg ( \log\bigg ( \frac{s_1(x)}{u_1(x ) } \bigg ) + \log\bigg ( \frac{s_2(x)}{u_2(x ) } \bigg ) \bigg ) \bigg ] \nonumber\\ & & - \frac{1}{(p\cdot q)^2 } \bigg [ \frac{1}{4 } [ m_{\tilde{f}}^2 - ( m_f + m_{\tilde\chi^0})^2 ] [ m_{\tilde{f}}^2 - ( m_f -m_{\tilde\chi^0})^2 ] \bigg ( \frac{1}{s_1(x ) } - \frac{2}{p_1(x ) } + \frac{1}{u_1(x)}\bigg ) \nonumber\\ & & -[m_{\tilde{f}}^2 - m_{\tilde\chi^0}^2 + m_f^2 ] \frac{1}{4x } \bigg ( \log\bigg ( \frac{s_1(x)}{p_1(x ) } \bigg ) + \log\bigg ( \frac{u_1(x)}{p_1(x ) } \bigg)+\log\bigg ( \frac{s_2(x)}{p_2(x ) } \bigg ) + \log\bigg ( \frac{u_2(x)}{p_2(x ) } \bigg ) \bigg ) \bigg ] \bigg \rgroup \bigg\ } \nonumber\\ & & - \frac{2 e_f^2 e^2}{(4 \pi)^2 } \int_0 ^ 1 \mathrm{d}x ( g^{r*}_{\tilde{f } f1 } g^l_{\tilde{f } f1 } + g^{l*}_{\tilde{f } f1 } g^r_{\tilde{f } f1})2 m_{\tilde\chi^0 } m_f \bigg\ { \bigg [ \frac{1}{s_1(x ) } + \frac{1}{u_1(x ) } + \frac{1}{p_1(x ) } \bigg ] \nonumber \\ & & - \frac{1}{p\cdot q } \bigg [ [ m_{\tilde{f}}^2 -m_{\tilde\chi^0}^2 + m_f^2 ] \bigg ( \frac{1}{s_1(x ) } - \frac{1}{u_1(x ) } \bigg ) -\frac{1}{2x } \bigg ( \log \bigg ( \frac{s_1(x)}{u_1(x)}\bigg ) + \log \bigg ( \frac{s_2(x)}{u_2(x ) } \bigg ) \bigg ) \bigg ] \nonumber \\ & & + \frac{1}{(p\cdot q)^2 } \bigg [ \frac{1}{4 } [ m_{\tilde{f}}^2- ( m_f + m_{\tilde\chi^0})^2 ] [ m_{\tilde{f}}^2- ( m_f - m_{\tilde\chi^0})^2 ] \bigg ( \frac{1}{s_1(x ) } -\frac{2}{p_1(x ) } + \frac{1}{u_1(x ) } \bigg ) \nonumber \\ & & -[m_{\tilde{f}}^2 -m_{\tilde\chi^0}^2 + m_f^2 ] \frac{1}{4x } \bigg ( \log\bigg ( \frac{s_1(x)}{p_1(x ) } \bigg ) + \log\bigg ( \frac{u_1(x)}{p_1(x ) } \bigg)+\log\bigg ( \frac{s_2(x)}{p_2(x ) } \bigg ) + \log\bigg ( \frac{u_2(x)}{p_2(x ) } \bigg ) \bigg ) \bigg ] \bigg\}. \label{ampff } \end{aligned}\ ] ] we adapt the notation from the previous section and set @xmath264 with @xmath265 and @xmath266 .
the polynomials @xmath267 , @xmath268 , and @xmath269 are defined as above under the interchange @xmath270 . using the symbolic manipulation program form @xcite , the leading order term in an expansion about @xmath2
is shown to be @xmath162 ; we find @xmath271 \nonumber \\ & & + ( g^{r*}_{\tilde{f } f1 } g^l_{\tilde{f } f1 } + g^{l*}_{\tilde{f } f1 } g^r_{\tilde{f } f1 } ) m_{\tilde\chi^0 } m_f \frac{4}{3 } \frac{x^2(1-x)^2 } { ( p_1(x))^3 } \bigg[1 + ( q \cdot p)^2 \frac{8}{5 } \frac{x^2(1-x)^2 } { ( p_1(x))^2 } \bigg ] \bigg\ } + \mathcal{o}(\omega^6 ) .
\label{nulinopol}\end{aligned}\ ] ] * numerics * the mssm contains well over a hundred parameters . models which attempt to describe a mechanism of symmetry breaking at the gut scale introduce additional assumptions which pare down the list of free parameters within the particular model ; however , more general considerations can still produce viable susy models . in refs .
@xcite , the authors explore p(henomenological)mssm , searching a large swath of susy parameter space with no regard to the method of symmetry breaking .
their searches employ 19 real parameters subject to collider and cosmological constraints as well as some basic theoretical constraints .
the models are cp - conserving with minimal flavor violation ; the first two generations of sfermions are degenerate ; and the lsp is a conventional thermal relic .
germane to our present work are the masses of the supersymmetric particles in viable models . in comparing eq .
( [ nulinopol ] ) with eq .
( [ wloop_lot ] ) in appendix b , we see that the leading order contributions to the forward compton amplitude coming from quark - squark , @xmath214-chargino , and higgs - chargino box diagrams have similar functional forms .
recalling the formula for the index of refraction , eq .
( [ qft_n ] ) , we see that the scale at which the index deviates from unity is set by the masses of virtual particles in figs . [ fig5 ] and [ fig8 ] .
in particular , the dominant mass particle in the loop sets the scale ; this fact was also borne out in the bf dark matter model in sec . [ boehm_sect ] . generally speaking , lighter masses of particles in the loop will result in a greater contribution to the refractive index
. given this , we will focus only upon the contributions coming from the fermion - sfermion loops since the sm quarks ( aside from the top ) and leptons have masses much smaller than the @xmath214 boson , higgs , and any susy charged particles .
versus @xmath272 .
the solid ( black ) curve corresponds to @xmath273 ; the dashed ( red ) curve has @xmath274 ; and the dotted ( blue ) curve has @xmath275 .
[ fig6],width=288 ] focusing upon the fermion - sfermion loops , we make some approximations as a means to understand which susy parameters might maximize the size of dispersive effects in eq .
( [ nulinopol ] ) . as most of the sm fermions entering the loop will have masses at a relatively lower scale , the term proportional to @xmath276 in eq .
( [ nulinopol ] ) should dominate ; we analyze this piece as a function of the input masses .
to do so , we define the dimensionless function @xmath277 as @xmath278\ ] ] with the ratio of masses @xmath272 and @xmath279 . the actual dispersive part of the index of refraction is governed by the terms which are @xmath232 in the amplitude ; we define the dimensionless @xmath280 to be @xmath281.\ ] ] in the limit that @xmath282 , i.e. , @xmath283 , one may evaluate the integrals rather easily @xmath284 and @xmath285 indeed , it is now clear that a small value of @xmath190 relative to @xmath261 can lead to a slight logarithmic enhancement of the dispersive term . versus @xmath272 .
the solid ( black ) curve corresponds to @xmath273 ; the dashed ( red ) curve has @xmath274 ; and the dotted ( blue ) curve has @xmath275 .
[ fig7],width=288 ] in figs .
[ fig6 ] and [ fig7 ] , we plot @xmath286 and @xmath287 for various values of @xmath288 as a function of @xmath289 .
we only plot values of @xmath289 up to unity since we assume the lightest neutralino to be the lsp . in general , we see that both functions are largest whenever @xmath290 and @xmath291 . given the prefactor of @xmath292 and @xmath293 , we also note that the polarizability and dispersive effects will be largest when @xmath261 is smallest . in refs .
@xcite , the scan of susy parameter space admits models which have third generation squarks with masses on the order of hundreds of gev .
additionally , a large fraction of these models have a nearly pure wino lsp . to estimate the size of dispersion from a viable neutralino dark matter candidate
, we will assume a pure wino lsp of mass 100 gev and a sbottom of mass 200 gev .
for a pure wino , we have in eq .
( [ nulino ] ) @xmath294 with @xmath295 otherwise ; as a result , the only nonzero sbottom - bottom - wino coupling is @xmath296 with @xmath297 the usual weak coupling constant . with this notation ,
the leading order deviation of the local refractive index from unity for these fermion - sfermion loops is @xmath298 where the arguments of @xmath299 are @xmath300 and @xmath301 so that @xmath302 and @xmath303 . in terms of the time lag
accrued between simultaneously emitted high ( energy @xmath2 ) and low energy photons from a distant grb , we have @xmath304 as a best case scenario for detecting such a time lag , we assume a high energy photon @xmath305 gev can be detected from a very distant grb at @xmath231 ; this yields a time lag @xmath306 s.
gamma ray bursts are a useful tool by which we may realize constraints on the coefficients @xmath307 in the index of refraction expansion , eq .
( [ n_expans ] ) , because they are extremely bright emissions of high energy photons of a rather short duration . in order to assess reliable limits on the refractive index from the study of photon arrival times , we must separate propagation effects from intrinsic source effects . on average ,
we expect time delays intrinsic to the source to be independent of @xmath88 , and the time delay from propagation to depend on @xmath88 and @xmath2 as in eq .
( [ cosmictlag ] ) .
thus , from a large sample of grbs occurring at various redshifts @xmath88 , the photon energy and redshift dependence of time lags due to photon - matter interactions should be discernible from intrinsic time delays associated with grbs .
such notions have been previously employed in searches for lorentz invariance violation ( liv ) @xcite . in this case ,
frequency dependent time lags for high energy photons are also expected due to a modified dispersion relation , and observations from distant grbs _ do _ provide meaningful constraints on the energy scale at which lorentz invariance violation ( liv ) can occur .
following ref .
@xcite , liv effects can modify the dispersion relation for photons at an energy scale @xmath308 @xmath309 for some integer @xmath310 .
if the modification in the dispersion relation in eq .
( [ disp_rel_g ] ) comes with the minus sign ( rather than the plus sign ) , then high energy photons will lag simultaneously emitted lower energy photons . for a ( present day ) photon energy @xmath2 emitted at redshift @xmath88 ,
this time lag is approximately @xmath311 to place constraints on a cosmic refractive index , we can look to the recent observations of high energy photons by the fermi telescope @xcite . in ref .
@xcite , several multi - gev photons were detected from grb 080916c which is located at a redshift @xmath312 @xcite .
the highest energy photon , with energy @xmath313 gev , arrived 16.54 s after initial @xmath314 photons from the grb .
if this time lag were attributable to interactions with dark matter , we can assess the magnitude of the coefficient @xmath315 in the index of refraction expansion .
we recall eq .
( [ lag_w2 ] ) which expresses the time lag engendered by a high energy photon in a neutral dark - matter medium @xmath316 as before , we assume the @xmath317 term for the mev photons is relatively negligible . from the redshift of grb 080916c , we calculate @xmath318 mpc .
the limits on @xmath315 and the dark - matter mass @xmath104 are thus @xmath319 an even higher energy photon was observed from grb 090510 @xcite .
the redshift of this grb was measured to be @xmath320 .
the highest energy photon detected is @xmath321 gev some 0.829 s after the trigger of the grb monitor .
in extracting their limits on lorentz invariance , the authors of ref .
@xcite conservatively use the 1-@xmath59 lower bound on the photon energy , @xmath322 gev , a time lag of 0.859 s , and the 1-@xmath59 lower bound on the redshift , @xmath323 . using these values
, we find the limit @xmath324 despite a higher energy photon and smaller time lag , this limit is slightly weaker than that achieved from grb 080916c .
this is due to the fact that grb 080916c occurred at a much higher redshift . in the context of the mssm neutralino calculation from sect .
[ nulino_sect ] , these experimental limits do not result in any meaningful constraints on the lsp .
it is interesting to explore under what circumstances dispersive effects due to dark matter might confound measurements designed to ascertain liv . we see that for @xmath325 the liv time lag depends on energy as @xmath1 , the same dependence as for neutral dark - matter dispersive effects . for both the bf and mssm models , the index of refraction has a similar structure .
if the charged particles in the loop are dominated by a particle of mass @xmath326 , then we can approximate the dispersive term in the forward scattering amplitude via @xmath327 for a dm candidate mass @xmath104 .
this structure follows from an eft analysis of the forward compton amplitude for a neutral particle . for photon energies below threshold
, the amplitude can be approximated as @xmath328 in the scatterer s rest frame . in this frame , the only way that @xmath2 can make an appearance in the amplitude is through the lorentz invariant quantity @xmath329 .
since the amplitude is itself dimensionless , we find balancing factors of @xmath326 in the denominator , since this mass dominates the loop process .
a scaling argument then indicates @xmath330 and @xmath331 . combining this with the index of refraction formula , eq .
( [ qft_n ] ) , we see that the @xmath332 prefactor which multiplies the amplitude is cancelled by the factor of @xmath333 common to all the amplitude terms . hence , with this naive analysis
, low - mass scatterers will not necessarily boost the size of the dispersive terms . taking the dispersive term as @xmath327 ,
the time lag is then approximately @xmath334 though the integrands in eqs .
( [ t_liv],[approx_tau ] ) carry different powers of @xmath335 , they only differ by factor of order @xmath336 and can be neglected in an order of magnitude calculation .
thus , we see that dark - matter dispersive effects could confound @xmath325 liv effects whenever @xmath337 weakly interacting massive particles ( wimps ) are a favored dm candidate because weak - scale interactions and dm masses naturally produce the correct relic density @xmath338 ; this is the so - called wimp miracle . however , it was shown in ref .
@xcite that alternative dm models can still achieve the desired relic density if @xmath339 this is referred to as the wimpless miracle . invoking this relation ,
we define the dm coupling in terms of the dm mass , @xmath340 . inserting this into the inequality in eq .
( [ ineq ] ) , we find that dm dispersive effects are important whenever the dm mass satisfies @xmath341 futhermore , stability of the dm candidate requires @xmath342 , so that we have @xmath343 to estimate the size of the relevant dm mass , let us suppose that liv effects occur at the planck scale @xmath344 gev .
setting @xmath345 and @xmath346 gev , we find that dm dispersive effects could confound liv effects whenever @xmath347 mev .
however , to saturate this inequality , the dominant mass particle , @xmath326 , to which the dm couples must be on the same mev scale , that is @xmath348 .
this presents two difficulties .
first , collider constraints limit unit - charged exotic particles to have masses well beyond this , @xmath349 gev .
second , if this could somehow be evaded , an mev scale mass @xmath326 would limit the inelastic threshold for the photon - dm interaction to also be on the mev scale .
this would result in a dispersive medium only below the mev threshold with no relevant dispersive effects above this energy .
given this , if liv effects occur near the planck scale , then frequency dependent time - lags that scale as @xmath1 can not be confused with dm dispersive effects because the matter effects are not appreciable .
turning this on its head , suppose we evade collider constraints on the mass of exotic charged particles by setting the dm mass and dominant mass to be @xmath350 gev .
then the inequality , eq .
( [ m_liv ] ) , indicates that matter effects are competitive with @xmath325 liv effects at an liv energy scale @xmath351 gev well beyond the gzk cutoff @xcite . below this energy ,
an observation of time lags that correlate with photon energy as @xmath1 can not be due to dispersive matter effects for the ilk of neutral dm models considered herein .
the same grbs above can be used to constrain the liv scale for an @xmath352 modification of the photon dispersion relation , eq .
( [ disp_rel ] ) .
grb 080916c indicates @xmath353 gev @xcite , and grb 090510 indicates @xmath354 gev @xcite . recalling from above
, an @xmath165 term can appear in the index of refraction if there is both @xmath22 and @xmath77 asymmetry present in the photon - matter interactions or the medium itself .
generally , this term results in circular birefringence of the medium , but no model independent expression exists for the @xmath165 term in the index of refraction , making it difficult to make generic remarks about the energy scale at which matter effects could compete with those attributable to liv .
however , one class of quantum gravity models stringy models of spacetime foam @xcite results in a modified dispersion relation which _ can _ be distinguished from matter effects . using an eft approach
, one can induce liv effects by including five- and six - dimensional operators in a qed lagrangian @xcite .
the resulting @xmath165 modifications to the photon dispersion relation will necessarily result in vacuum birefringence . to contrast , spacetime foam models move beyond the eft paradigm and can result in a modified photon dispersion relation which is linear in photon energy but _ not _ birefringent . since matter effects will always result in a birefringent medium for @xmath165 energies
, photon time lags that scale linearly with photon energy _ and _ are polarization independent signal liv due to a qg effect like spacetime foam .
from some rather basic assumptions , we determine the index of refraction for light traveling through particulate matter . in previous work , we investigated the optical consequences of a dm particle with fractional electric charge @xcite , where the leading order contribution to the refractive index scales with photon energy as @xmath0 . herein , we considered higher order contributions coming from the @xmath355 corrections to the forward compton amplitude .
the dominant term in this correction is a quasi - static polarizability which results in a term in the index of refraction that scales as @xmath356 at low photon energies ; however , this term is dominated by the leading order term in perturbation theory . at higher energies , likewise , the corrections are negligible for photon energies below the gzk cutoff . for neutral dm candidates ,
the leading order contribution to the forward compton amplitude comes from the polarizability of the particle for photon energies below the inelastic threshold . as a consequence ,
the leading order term in the refractive index is frequency independent , and the largest dispersive term is quadratic in photon energy . at the one - loop level in perturbation theory , we explicitly confirmed this behavior for a neutral scalar dm candidate as well as for mssm s neutralino . for both models , we considered the situation in which one of the charged particles in the loop had a dominant mass @xmath326 .
for this situation , the dispersive term in the index of refraction scaled as @xmath357 for dm mass @xmath104 and energy density @xmath358 .
given the collider constraints @xmath359 gev , dispersion from even a distant grb results in a negligible time lag between simultaneously emitted high and low energy photons .
if lighter charged exotic particles did in fact exist , say , on the mev scale , then the cosmos could have measurable dispersion .
since neutral dark matter can result in time lags which scale with photon energy as @xmath1 , we compared the size of photon time lags arising from a dispersive medium with the @xmath162 time lags that can result from lorentz invariance violation .
for the models considered herein , in which the neutral dm candidate couples to a massive charged exotic particle , we find @xmath162 time lags due to liv will dominate matter effects below the gzk cutoff .
the following integrals appear in the forward - scattering box diagrams @xmath360 , \\
i_3(s , m_1 ^ 2,m_2 ^ 2 ) & = & \int_0 ^ 1 \mathrm{d}x \frac{1}{x } \log \bigg [ \frac{p(s , m_1 ^ 2,m_2 ^ 2;x)}{m_1 ^ 2}\bigg],\end{aligned}\ ] ] with the polynomial defined to be @xmath361 . here , the masses of the virtual particles are @xmath362 and @xmath363 ; the four - momentum of the scatterer is @xmath118 with @xmath119 ; and the four momentum of the photon is @xmath184 with @xmath259 .
the relevant mandelstam variables are @xmath364 and @xmath199 .
we define the threshold energy @xmath365 and mass difference @xmath366 .
we assume that the particle is stable to decay so that @xmath367 .
the first argument of the functions @xmath368 will be either @xmath252 , @xmath369 , or @xmath370 .
we consider the case where @xmath252 is the first argument of @xmath368 .
the integrals can be divided into three different regions @xmath371 , @xmath372 , and @xmath373 .
all three integrals are real except when @xmath373 .
we note that @xmath374 for @xmath375 ( but _ not _ @xmath376 ) .
this is true because @xmath377 so that , after a change of variables in the second integral , we have @xmath378 = \int_0 ^ 1 \mathrm{d}x f[p(s , m_2 ^ 2,m_1 ^ 2;x)]\ ] ] for some function @xmath256 whose argument is the polynomial @xmath22 .
we begin with @xmath379 and @xmath380 and discuss @xmath381 later .
for @xmath382 , we have @xmath383 for @xmath372 , we have @xmath384 , \\ i_2(s , m_1 ^ 2,m_2 ^ 2 ) & = & \log(m_1 m_2 ) + \frac{m_2 ^ 2 - m_1 ^ 2}{2s } \log\bigg ( \frac{m_2 ^ 2}{m_1 ^ 2 } \bigg ) -2
\nonumber \\ & & + \frac{\sqrt{(e_\text{th}^2 - s)(s-\delta^2)}}{s } \bigg [ \tan^{-1 } \bigg ( \frac{m_2 ^ 2-m_1 ^ 2+s } { \sqrt{(e_\text{th}^2-s)(s-\delta^2 ) } } \bigg ) - \tan^{-1 } \bigg ( \frac{m_2 ^ 2-m_1 ^ 2- s } { \sqrt{(e_\text{th}^2-s)(s-\delta^2 ) } } \bigg ) \bigg ] .\end{aligned}\ ] ] for @xmath373 , we have @xmath385 to deal with @xmath381 , we first factor the polynomial in the following manner @xmath386 with @xmath387 with this decomposition , the integral can be evaluated in terms of dilogarithms @xmath388 in certain situations , the @xmath381 integrals come in pairs as the sum @xmath389
. this second integral can be evaluated in exactly the same way @xmath390 with the analogous definition @xmath391 with these integrals in pairs , it turns out that the dilogarithms in the amplitude may be completely eliminated via the identity from ref .
@xcite @xmath392 valid for @xmath393 .
this expression is operative as @xmath394 to prove this , we recall the relation @xmath395 . employing the factorization from eq .
( [ factor ] ) , one has @xmath396 each of the two roots of the polynomial on the lhs corresponds to one of the roots of the polynomial on the rhs yielding the identities in eq .
( [ i d ] ) . as before ,
we express the real part of this equation for the three different intervals of @xmath252 . for @xmath371
, we have @xmath397.\end{aligned}\ ] ] for @xmath372 , we have @xmath398 ^ 2\ ] ] with @xmath399 the heaviside function . finally , when @xmath400 , we have @xmath401.\end{aligned}\ ] ] when evaluating integrals of the form @xmath402 and @xmath403 , one only needs to substitute @xmath370 or @xmath369 for @xmath252 in the above forumlae .
we note that if the dark matter particle is stable to decay then @xmath367 so that one need not consider the intervals where @xmath404 as this is not physically possible .
in this appendix , we include the remaining contributions to the forward compton amplitude of the neutralino .
these calculations involve @xmath214-chargino and higgs - chargino loops .
* @xmath214-chargino contribution * as discussed in ref .
@xcite , opting to work in a non - linear @xmath405 gauge greatly simplifies the calculations .
proposed by fujikawa @xcite , this choice of gauge eliminates vertices between photons , @xmath214 bosons , and the unphysical higgs bosons @xmath406 , and it simplifies the @xmath214 boson propagators .
a detailed list of the feynman rules in this framework can be found in ref .
@xcite ; in our calculations , we set @xmath407 .
figure [ fig8 ] contains the feynman diagrams needed to compute the contribution to the forward comptom amplitude coming from the @xmath214-chargino loops .
there are four such charginos @xmath408 indexed by @xmath100 , and the coupling between it , the @xmath214 boson , and the lightest neutralino is denoted by @xmath409 , depending on whether we project onto the left- or right - handed state .
[ h ] boson and charginos contributing to the forward compton amplitude for a neutralino .
[ fig8],title="fig:",width=288 ] the forward scattering amplitude for these processes follows @xmath410 \nonumber \\ & & + \frac{1}{p\cdot q } \bigg [ 2 [ m_w^2- m_{\tilde\chi^0}^2 + m_{\tilde\chi^+}^2 ] \bigg ( \frac{1}{s_1(x ) } -\frac{1}{u_1(x ) } \bigg ) - \frac{1}{x } \bigg ( \log\bigg ( \frac{s_1(x)}{u_1(x ) } \bigg ) + \log\bigg ( \frac{s_2(x)}{u_2(x ) } \bigg ) \bigg ) \bigg ] \nonumber\\ & & - \frac{1}{(p\cdot q)^2 } \bigg [ \frac{1}{2 } [ m_w^2 - ( m_{\tilde\chi^+ } + m_{\tilde\chi^0})^2 ] [ m_w^2 - ( m_{\tilde\chi^+ } -m_{\tilde\chi^0})^2 ] \bigg ( \frac{1}{s_1(x ) } - \frac{2}{p_1(x ) } + \frac{1}{u_1(x)}\bigg ) \nonumber\\ & & -[m_w^2 - m_{\tilde\chi^0}^2 + m_{\tilde\chi^+}^2 ] \frac{1}{2x } \bigg ( \log\bigg ( \frac{s_1(x)}{p_1(x ) } \bigg ) + \log\bigg ( \frac{u_1(x)}{p_1(x ) } \bigg)+\log\bigg ( \frac{s_2(x)}{p_2(x ) } \bigg ) + \log\bigg ( \frac{u_2(x)}{p_2(x ) } \bigg ) \bigg ) \bigg ] \bigg \rgroup \nonumber\\ & & -\frac{1}{m_w^2 } ( m_w^2-(m_{\tilde\chi^0}+m_{\tilde\chi^+})^2 ) ( m_w^2-(m_{\tilde\chi^0}-m_{\tilde\chi^+})^2 ) \bigg [ \frac{1}{s_1(x ) } -\frac{2}{p_1(x)}+\frac{1}{u_1(x)}\bigg ] \nonumber \\ & & + \frac{4 p \cdot q}{m_w^2 } ( m_w^2 - m_{\tilde\chi^0}^2+m_{\tilde\chi^+}^2 ) \bigg [ \frac{1}{s_1(x ) } -\frac{1}{u_1(x)}\bigg ] -\frac{4(p \cdot q)^2}{m_w^2 } \bigg [ \frac{1}{s_1(x ) } + \frac{1}{u_1(x)}\bigg ] \bigg\ } \nonumber\\ & & - \frac{2 e^2}{(4 \pi)^2 } \int_0 ^ 1 \mathrm{d}x ( g^{r*}_{w1j } g^l_{w1j } + g^{l*}_{w1j } g^r_{w1j})2 m_{\tilde\chi^0 } m_{\tilde\chi^+ } \bigg\ { -4 \bigg [ \frac{1}{s_1(x ) } + \frac{1}{u_1(x ) } + \frac{1}{p_1(x ) } \bigg ] \nonumber \\ & & + \frac{1}{p\cdot q } \bigg [ 4 [ m_w^2 -m_{\tilde\chi^0}^2 + m_{\tilde\chi^+}^2 ] \bigg ( \frac{1}{s_1(x ) } - \frac{1}{u_1(x ) } \bigg ) -\frac{2}{x } \bigg ( \log \bigg ( \frac{s_1(x)}{u_1(x)}\bigg ) + \log \bigg ( \frac{s_2(x)}{u_2(x ) } \bigg ) \bigg ) \bigg ] \nonumber \\ & & + \frac{1}{(p\cdot q)^2 } \bigg [ - [ m_w^2- ( m_{\tilde\chi^+ } + m_{\tilde\chi^0})^2 ] [ m_w^2- ( m_{\tilde\chi^+ } - m_{\tilde\chi^0})^2 ] \bigg ( \frac{1}{s_1(x ) } -\frac{2}{p_1(x ) } + \frac{1}{u_1(x ) } \bigg ) \nonumber \\ & & + [ m_w^2 -m_{\tilde\chi^0}^2 + m_{\tilde\chi^+}^2 ] \frac{1}{x } \bigg ( \log\bigg ( \frac{s_1(x)}{p_1(x ) } \bigg ) + \log\bigg ( \frac{u_1(x)}{p_1(x ) } \bigg)+\log\bigg ( \frac{s_2(x)}{p_2(x ) } \bigg ) + \log\bigg ( \frac{u_2(x)}{p_2(x ) } \bigg ) \bigg ) \bigg ] \bigg\}. \end{aligned}\ ] ] the polynomials @xmath411 , @xmath412 , and @xmath413 which appear in this amplitude are defined in eq .
( [ polynomials_ff ] ) after substituting @xmath414 and @xmath415 ; as before , the polynomials @xmath267 , @xmath268 , and @xmath269 can be obtained from these by switching the masses @xmath416 . using form
@xcite to expand the integrands in a power series of @xmath2 , we find the leading order behavior of the amplitude is @xmath162 as expected ; we compute the expansion to include the leading order dispersive effects @xmath417 \nonumber \\ & & - ( g^{r*}_{w1j } g^l_{w1j } + g^{l*}_{w1j } g^r_{w1j } ) m_{\tilde \chi^0 } m_{\tilde \chi^+ } \frac{16}{3}\frac { x^2 ( 1- x)^2}{(p_1(x))^3 } \left [ 1 + ( p \cdot q)^2 \frac{8}{5 } \frac { x^2 ( 1- x)^2}{(p_1(x))^2 } \right]\bigg\ } + \mathcal{o}(\omega^6 ) .
\label{wloop_lot}\end{aligned}\ ] ] * higgs - chargino contribution * two contributions to the forward compton amplitude remain .
the first is due to box diagrams involving the higgs boson @xmath418 and charginos , and the second , a consequence of our gauge choice , is due to the box diagrams involving the unphysical higgs boson @xmath419 and charginos . in both cases , the box diagrams are identical to those in fig .
[ fig5 ] as long as we identify the internal scalar lines with the ( un)physical higgs bosons and the internal fermion line with the charginos . given this similarity
, the ( un)physical higgs - chargino amplitude will take the same form as eqs .
( [ ampff],[nulinopol ] ) .
for the physical higgs - chargino diagrams , we merely need to substitute the mass @xmath420 , @xmath415 and couplings @xmath421 for the @xmath100th chargino . for the unphysical higgs loop ,
we must substitute @xmath422 and @xmath423 ; explicit expressions for these couplings can be found in ref .
the author is grateful for useful comments from susan gardner who helped shape the trajectory of this work .
this work is supported , in part , by the u.s .
department of energy under contract de fg0296er40989 .
r. c. myers and m. pospelov , phys .
lett . * 90 * , 211601 ( 2003 ) . v. a. kostelecky and m. mewes , astrophys . j. * 689 * , l1 ( 2008 ) . v. a. kostelecky and m. mewes , phys .
d * 80 * , 015020 ( 2009 ) [ arxiv:0905.0031 [ hep - ph ] ] .
j. r. ellis , n. e. mavromatos and d. v. nanopoulos , gen .
grav . * 32 * , 127 ( 2000 ) .
j. alfaro , h. a. morales - tecotl and l. f. urrutia , phys .
d * 65 * , 103509 ( 2002 ) .
j. r. ellis , n. e. mavromatos , d. v. nanopoulos and g. volkov , gen .
grav . * 32 * , 1777 ( 2000 ) .
j. ellis , n. e. mavromatos and d. v. nanopoulos , phys .
d * 63 * , 024024 ( 2000 ) .
j. r. ellis , n. e. mavromatos and m. westmuckett , phys .
d * 70 * , 044036 ( 2004 ) .
j. r. ellis , n. e. mavromatos , d. v. nanopoulos , a. s. sakharov , and e. k. g. sarkisyan , astropart .
phys . * 25 * , 402 ( 2006 ) [ erratum - ibid . astropart .
* 29 * , 158 ( 2008 ) ] .
u. jacob and t. piran , jcap * 0801 * , 031 ( 2008 ) .
j. r. ellis , n. e. mavromatos , and d. v. nanopoulos , phys .
b * 665 * , 412 ( 2008 ) .
g. amelino - camelia and l. smolin , phys .
d * 80 * , 084017 ( 2009 ) .
v. vasileiou , a. jacholkowska , f. piron , j. bolmont , c. couturier , j. granot , f. w. stecker and j. cohen - tanugi _ et al .
_ , phys .
d * 87*,122001 ( 2013 ) .
a. a. abdo _ et al .
_ [ fermi lat collaboration ] , nature * 462 * , 331 ( 2009 ) .
m. scrimgeour , t. davis , c. blake , j. b. james , g. poole , l. staveley - smith , s. brough and m. colless _ et al .
_ , mon . not .
soc . * 425 * , 116 ( 2012 ) .
d. n. spergel _ et al . _
[ wmap collaboration ] , astrophys .
j. suppl .
* 148 * , 175 ( 2003 ) .
m. tegmark _ et al . _
[ sdss collaboration ] , phys .
d * 69 * , 103501 ( 2004 ) ; j. h. heo , phys .
b * 702 * , 205 ( 2011 ) .
v. barger , w. -y .
keung and d. marfatia , phys .
b * 696 * , 74 ( 2011 ) .
v. barger , w. -y .
keung , d. marfatia and p. -y .
tseng , phys .
b * 717 * , 219 ( 2012 ) .
m. pospelov and t. ter veldhuis , phys .
b * 480 * , 181 ( 2000 ) [ hep - ph/0003010 ] .
r. c. cotta , j. l. hewett , m. p. le and t. g. rizzo , `` bounds on dark matter interactions with electroweak gauge bosons , '' arxiv:1210.0525 [ hep - ph ] .
n. prantzos , c. boehm , a. m. bykov , r. diehl , k. ferriere , n. guessoum , p. jean and j. knoedlseder _ et al .
_ , arxiv:1009.4620 [ astro-ph.he ] .
r. akers _ et al .
_ [ opal collaboration ] , z. phys .
c * 67 * , 203 ( 1995 ) ; p. abreu _ et al .
_ [ delphi collaboration ] , phys .
* b396 * , 315 ( 1997 ) .
l. bergstrom and p. ullio , nucl .
b * 504 * , 27 ( 1997 )
. z. bern , p. gondolo and m. perelstein , phys .
b * 411 * , 86 ( 1997 ) .
a. denner , h. eck , o. hahn and j. kublbeck , phys .
b * 291 * , 278 ( 1992 ) . c. f. berger , j. s. gainer , j. l. hewett and t. g. rizzo , jhep * 0902 * , 023 ( 2009 ) .
r. c. cotta , j. s. gainer , j. l. hewett and t. g. rizzo , new j. phys . * 11 * , 105026 ( 2009 ) ; m. w. cahill - rowley , j. l. hewett , s. hoeche , a. ismail and t. g. rizzo , eur .
j. c * 72 * , 2156 ( 2012 ) [ arxiv:1206.4321 [ hep - ph ] ] .
m. w. cahill - rowley , j. l. hewett , a. ismail and t. g. rizzo , phys .
d * 86 * , 075015 ( 2012 ) [ arxiv:1206.5800 [ hep - ph ] ] .
m. w. cahill - rowley , j. l. hewett , a. ismail and t. g. rizzo , arxiv:1211.1981 [ hep - ph ] .
a. a. abdo _ et al .
_ [ fermi lat and fermi gbm collaborations ] , science * 323 * , 1688 ( 2009 ) .
j. greiner _ et al .
_ , arxiv:0902.0761 [ astro-ph.he ] . | searches for dispersive effects in the propagation of light at cosmological distances have been touted as sensitive probes of lorentz invariance violation ( liv ) and of theories of quantum gravity .
frequency - dependent time lags between simultaneously emitted pulses of light can signal a modification in the photon dispersion relation ; however , matter engenders the cosmos with a dispersive index of refraction to similar effect .
we construct a theoretical framework for the analysis of such effects , contrasting these dispersive terms with those from liv models .
we consider all matter , both luminous and dark . though the only known mode of interaction for dark matter ( dm ) is gravitational
, most models of dark matter also allow for electromagnetic interactions , if only at the one - loop level in perturbation theory .
generically , the leading order dispersive effects due to matter scale with photon energy as @xmath0 for a charged dm candidate and @xmath1 for a neutral dm candidate .
terms linear in @xmath2 can arise in the index of refraction if parity and charge - parity asymmetries are present at the lagrangian or system level .
herein , we compute the index of refraction for a millicharged dark matter candidate at the one - loop level , a neutral scalar dm candidate introduced by boehm and fayet @xcite , and the mssm s neutralino . for a neutral dm candidate
, we determine that matter effects can compete with liv effects that depend quadratically on energy whenever the photon energy is beyond @xmath3 gev , well beyond the gzk cutoff .
the dispersive matter effects that scale linearly with @xmath2 are model dependent , and their existence results in circular birefringence . |
in the very high energy ( vhe ) region above 100 gev , positive detections of radiation at a high confidence level have been reported from the direction of the crab , b1706 - 44 , and vela pulsars ( bowden et al .
1993 ; nel et al . 1993 ; edwards et al . 1994 ; yoshikoshi et al . 1997
; see also kifune 1996 for a review ) , by virtue of the technique of imaging cerenkov light from extensive air showers .
however , as for _ pulsed _ tev radiation , only the upper limits have been obtained from these pulsars .
if the vhe emission originates the pulsar magnetosphere , rather than the extended nebula , significant fractions of them are expected to show a pulsation . therefore , the lack of _ pulsed _ tev emissions provides a severe constraint on the modeling of particle acceleration zones in pulsar magnetospheres . in the picture of cheng , ho , and ruderman ( 1986a ,
b ) , who developed a version of an outer magnetosphere @xmath0-ray emission zone , the magnetosphere should be optically thick for pair production in order that the tev flux is absorbed to be unobservable .
this , in turn , requires very high luminosities of infrared photons , which are generally orders of magnitude larger than the observed values ( usov 1994 ) .
we are therefore motivated by the need to contrive an outer gap model which produces less tev emission with a moderate infrared luminosity .
more recently , romani ( 1996 ) showed that the pulsed tev flux can be somewhat less than @xmath1 of the pulsed gev flux , assuming the functional form of the acceleration field as @xmath2 , where @xmath3 denotes the distance from the star .
subsequently , hirotani ( 1999 , paper iv ) revealed semi - analytically that the tev fluxes from outer gaps are unobservable , by considering a gap closure condition so that a single pair produces copious @xmath0-rays one of which materialize as a pair on average ( see also zhang & cheng 1998 for another kind of closure condition ) . in this letter , instead of imposing the closure condition , we investigate the spatial distribution of @xmath4 , by solving the poisson equation and the boltzmann equations simultaneously .
this method was first developed by beskin et al .
( 1992 ) for black - hole magnetospheres and applied to pulsar magnetospheres by hirotani & shibata ( 1999a , b , c ; hereafter paper i , ii , iii ) .
we formulate basic equations in 2 ,
demonstrate the invisibility of pulsed tev fluxes in 3 , and discuss the validity of assumptions in 4 .
to simplify the geometry , let us introduce a rectilinear coordinates for a region around the null surface .
suppose that the magnetic field lines are straight lines parallel to the @xmath5 axis , where @xmath5 increases outwardly .
we approximate that the null surface ( @xmath6 ) is perpendicular to the @xmath5 axis .
it is convenient to non - dimensionalize the length scales by a typical debey scale length @xmath7 ( see the discussion in 3.1 in paper i for details ) , where @xmath8 @xmath9 refers to the magnitude of the charge on the electron , @xmath10 the angular frequency ( @xmath11 ) of the neutron star in @xmath12 unit , @xmath13 the velocity of light , and @xmath14 the magnetic field strength ( @xmath15 ) at @xmath6 in @xmath16 g unit . then the dimensionless coordinate variable becomes @xmath17 assuming the trans - field thickness is comparable to or greater than the longitudinal width , we fourier - analyze the poisson equation in the trans - field directions to obtain ( paper i , ii , iii ) @xmath18 @xmath19 where @xmath20 and @xmath21 ; @xmath22 is a typical gap thickness in the trans - field directions and @xmath23 the electrostatic potential .
the particle densities are normalized in terms of a typical value of the goldreich
julian density as @xmath24 where @xmath25 are the spatial number density of @xmath26 s .
we evaluate it with the newtonian value as @xmath27 } , \label{eq : rhogj}\ ] ] where @xmath28 refers to the inclination angle of the magnetic moment , @xmath29 the clatitude angle of the position at which @xmath30 is measured , @xmath31 the distance from the rotation axis , and @xmath32 the light - cylinder radius .
let us first introduce the following dimensionless @xmath0-ray densities in the dimensionless energy interval between @xmath33 and @xmath34 : @xmath35 where @xmath36 are the distribution functions of @xmath0-ray photons having momentum @xmath37 in this letter , we set @xmath38 ; therefore , the lowest @xmath0-ray energy considered is @xmath39 . in this letter
, we divide the @xmath0-ray spectra into @xmath40 energy bins as follows : @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 , @xmath53 .
assuming for simplicity that both electrostatic and curvature - radiation - reaction forces cancel out each other , we obtain the following continuity equations for particles ( eqs . [ 31 ] and [ 32 ] in paper iii ) : @xmath54 , \label{eq : basic-3}\ ] ] @xmath55 the factor @xmath56 denotes the projection effect due to the three - dimensional motion of the particles , which was not considered in papers i , ii , and iii .
since the electric field is assumed to be positive in the gap , @xmath57 s ( or @xmath58 s ) migrate outwards ( or inwards ) .
the x - ray field affects the gap structure through the pair - production rate @xmath59 .
putting @xmath60 , we can evaluate @xmath59 as @xmath61 where @xmath62 denotes the number density of the x - rays between energies @xmath63 and @xmath64 ; @xmath65 is the pair production cross section when two photons collide with an angle @xmath66 ( berestetskii et al . 1989 ; eqs .
[ 21]-[23 ] in paper ii ) . for surface blackbody x - rays ,
we obtain @xmath67 , where @xmath68 is the aberration angle . for power - law , magnetospheric x - rays , we assume @xmath69 , where @xmath70 is the full gap width , for both inwardly and outwardly propagating @xmath0-rays . as the lower bound of the integral , the threshold energy , @xmath71 , for pair production appears .
a combination of the two continuity equations ( [ eq : basic-3 ] ) gives the current density @xmath72 , which is constant for @xmath73 .
when @xmath74 , the current density equals the typical goldreich julian value , @xmath75 .
we next consider the boltzmann equations for @xmath0-rays .
integrating the @xmath0-ray boltzmann equations between @xmath33 and @xmath34 , we obtain ( eq . [ 39 ] in paper iii ) @xmath76 where @xmath77 .
the @xmath0-ray production rate via curvature radiation , @xmath78 , are defined by equation ( 40 ) in paper iii .
the particle motion is assumed to be mono - energetic and the lorentz factor saturates in a balance between electrostatic acceleration and the curvature drag force ( see eq . [ 19 ] in paper iii ) . in the same manner as in papers i , ii , and iii ,
we impose the boundary conditions at the inner ( @xmath79 ) and outer ( @xmath80 ) boundaries as follows : @xmath81 @xmath82 @xmath83 where @xmath84 .
we have totally @xmath85 boundary conditions ( [ eq : bd-1])([eq : bd-3 ] ) for @xmath86 unknown functions @xmath23 , @xmath87 , @xmath88 , @xmath89 .
thus two extra boundary conditions must be compensated by making the positions of the boundaries @xmath90 and @xmath91 be free .
the two free boundaries appear because @xmath92 is imposed at the _ both _ boundaries and because @xmath93 is externally imposed .
the pair production cascade is primarily described by the curvature - radiated , gev @xmath0-rays .
therefore , the tev emission due to compton upscatterings is energetically negligible and can be computed passively from the electrodynamic structure of the gap .
assuming that the azimuthal gap width is @xmath94 rad , we can evaluate the ratio between the tev flux and the infrared one as ( see 2.7 in paper v ) @xmath95 where @xmath96 refers to the solid angle in which photons with energy @xmath97 are emitted ; @xmath98 is assumed for simplicity . to estimate the upper bound , we take @xmath99 .
the inequality comes from the fact that the scattered photon energy can not exceed the particle energy .
it is noteworthy that it is the infrared photons with energy @xmath100 ev that contribute most effectively as the target photons of comptonization .
neither the higher energy photons like surface blackbody x - rays nor the lower energy photons like polar - cap radio emission contribute as the target photons , because they have either too small cross sections or too small energy transfer when they are scattered .
substituting the x - ray fields , @xmath11 and @xmath101 of individual pulsars into the differential equations , we can solve @xmath102 and hence @xmath103 ; equation ( [ eq : tev_flux_2 ] ) then gives @xmath104 as a function of @xmath105 and @xmath93 .
in this section , we apply the theory to seven rotation - powered pulsars .
we first present the observed x - ray properties of individual pulsars as input parameters in order of spin - down luminosity in table 1 .
the x - ray spectrum of each pulsar is approximated by separating into a soft blackbody component with temperature @xmath106 and emitting area @xmath107 , a hard blackbody one with @xmath108 and @xmath109 , and a power - law one with @xmath110 ( @xmath111 ) , where @xmath112 and @xmath113 are assumed for homogeneous discussion .
[ cols="<,^,^,^,^,^,^,^,^,^ , < " , ] @xmath114 @xmath70 indicates the full gap width .
+ @xmath115 @xmath116 .
in summary , we have developed a one - dimensional model for an outer - gap accelerator .
as the inclination angle ( @xmath105 ) increases , the gap center ( where @xmath117 vanishes ) shifts starwards to be illuminated by a strong x - ray field , which reduces the pair production mean free path and hence the gap longitudinal width , @xmath70 . at the same time , the starwardly shifted gap has a strong magnetic field , which increases the acceleration field ( see eq . [ [ eq : basic-2 ] ] ) . as a result
, the tev flux increases with increasing @xmath105 , even though @xmath70 decreases .
however , it is difficult to detect the tev flux , even when @xmath105 is as large as @xmath118 .
we can therefore conclude that the difficulty of excessive tev emission , which appears in the picture of cheng , ho , and ruderman ( 1986a , b ) , does not arise in the present outer gap model .
it is seemingly possible to obtain observable tev fluxes , if @xmath105 exceeds @xmath118 . in this case , the outer gap is located close to the polar cap ; the distance from the star is within @xmath119 of @xmath120 .
the magnetic field there is so strong that the synchrotron ( or synchro - curvature ) process becomes important in general ; this could be verified by computing the pitch angle evolution of particles created in the gap ( see 5.3 in paper i ) .
such a polar - gap - like outer gap will be discussed in a separate paper . for middle - age pulsars
b0656 + 14 and geminga , there is no stationary solution that satisfies the boundary conditions presented in [ sec : bd ] .
however , if we allow an external electric current flowing into the gap at the outer boundary , there exists a branch of solutions . in this case , the gap is no longer located around the null surface but shifts starwards .
interestingly , when the external current density approaches the goldreich - julian value , the outer gap is found to shift to the polar cap and have a similar electrodynamic structure of scharlemann , arons , & fawley ( 1978 ) .
such a unification of the outer gap and polar cap accelerators will be discussed in separate papers .
this research owes much to the helpful comments of dr .
s. shibata .
the author wishes to express his gratitude to drs .
y. saito and a. harding for valuable advice .
he also thanks the astronomical data analysis center of national astronomical observatory , japan for the use of workstations .
berestetskii , v. b. , lifshitz , e. m. & pitaevskii , l. p. , 1989
, quantum electrodynamics 3rd ed .
beskin , v. s. , istomin , ya .
n. , & parev , v. i. 1992 , sov .
36(6 ) , 642 bowden , c. c. g. et al .
1993 , proc . of 23rd int
. cosmic ray conf .
( calgary ) , 1 , 294 caswell , j. l. , murray , j. d. , roger , r. s. , cole , d. j. , cooke , d. j. 1975 , a & a 45 , 239 cheng , k. s. , ho , c. , ruderman , m. , 1986a apj , 300 , 500 cheng , k. s. , ho , c. , ruderman , m. , 1986b apj , 300 , 522 edwards , p. g. , et al . 1994 a & a 291 , 468 greiveldinger , c. et al . 1996 , apj 465 , l35 hirotani , k. 1999 ( paper iv ) , submitted to apj hirotani , k. okamoto , i. 1998 , apj 497 , 563 hirotani , k. shibata , s. , 1999a ( paper i ) , mnras 308 , 54 hirotani , k. shibata , s. , 1999b ( paper ii ) , mnras 308 , 76 hirotani , k. shibata , s. , 1999c ( paper iii ) , pasj 51 , 683 kifune , t. 1996 , space science reviews 75 , 31 knight f. k. 1982 , apj 260 , 538 nel , h. i. et al .
1993 , apj 418 , 836 @xmath121gelman , h. , finley , j. p. , zimmermann , h. u. 1993 , nature 361 , 136 romani , r. w. 1996 , apj , 470 , 469 safi - harb , s. @xmath121gelman , h. 1995 , apj 439 , 722 safi - harb , s. @xmath121gelman , h. 1995 , apj 439 , 722 saito , y. 1998 , ph.d .
thesis scharlemann , e. t. , arons , j. , & fawley , w. m. 1978 , apj , 222 , 297 torii , k. et al .
1998 , apj 494 , l207 usov , v. v. 1994 , apj 427 , 394 yoshikoshi , t. , et al .
1997 , apj 487 , l65 zhang , l. cheng , k. s. 1998 , mnras 294 , 177 | we study the structure of an outer - magnetospheric gap around a rotating neutron star .
migratory electrons and positrons are accelerated by the electric field exerted in the gap and radiate copious gamma - rays via curvature process .
some of the gamma - rays materialize by colliding with the x - rays illuminating the gap , leading to a pair production cascade .
the replenished charges partially screen the original acceleration field , which is self - consistently solved from the poisson equation , together with the boltzmann equations for gamma - rays and the continuity equations for particles .
we demonstrate that it is difficult to detect the tev emission due to compton upscatterings in the gap , by the current ground - based telescopes . |
the number of rotation - powered pulsars with detected emission at x - ray and @xmath0-ray energies has been steadily growing ( thompson 2001 , becker & aschenbach 2002 ) .
although the theory of pulsar acceleration and high - energy emission has been studied for over twenty - five years , the origin of the pulsed non - thermal emission is a question that remains unsettled .
the observations to date have not been able to clearly distinguish between an emission site at the magnetic poles ( daugherty & harding 1996 ) and emission in the outer magnetosphere ( cheng , ho & ruderman 1986 , hirotani & shibata 2001 ) . in the case of polar cap ( pc ) models , while the energetics and pair - cascade spectrum have had success in reproducing the observations , the predicted beam size of radiation emitted near the neutron star ( ns ) surface is too small to produce the wide pulse profiles that are observed ( thompson et al .
however , arons ( 1983 ) first noted the possibility of a high - altitude acceleration region or slot gap " ( near the pc rim , based on the finding of arons & scharlemann ( 1979 , hereafter as79 ) that the pair formation front , above which the accelerating field is screened , occurs at increasingly higher altitude as the magnetic colatitude approaches the last open field line where the electric field vanishes .
we have re - examined the slot - gap model with the inclusion of two new features : 1 ) the acceleration due to inertial - frame dragging ( muslimov & tsygan 1992 ) and 2 ) the additional decrease in the electric field near the boundary at the edge of the polar cap due to the narrowness of the slot gap .
these two effects combine to enable acceleration to altitudes of several stellar radii in the slot gap at all azimuthal angles around the polar cap .
these features result in the production of a larger high - energy emission beam at smaller magnetic inclination angles , both necessary for widely spaced double - peaked pulse profiles .
in the space - charge limited flow acceleration model ( arons & scharlemann 1979 ) , electrons are freely emitted from the neutron star surface near the magnetic poles and accelerated in steady - state .
when the electrons achieve a sufficient lorentz factor they radiate curvature and inverse compton photons that can produce electron - positron pairs in the strong magnetic field .
some of the positrons turn around and accelerate downward toward the neutron star , while the electrons accelerate upward .
this pair polarization screens the electric field above a pair formation front ( pff ) .
these models assume a boundary condition that the accelerating electric field and potential are zero at the last open field line .
since near the boundary , a larger distance is required for the electrons to accelerate to the lorentz factor needed to radiate photons energetic enough to produce pairs , the pff occurs at higher and higher altitudes as the boundary is approached .
the pff thus curves upward , approaching infinity and becoming asymptotically tangent to the last open field line .
if the electric field is effectively screened above the pff , then a narrow slot surrounded by two conducting walls is formed ( see figure 1 ) . within this slot gap , the electric field
is further screened by the presence of the second conducting boundary and acceleration occurs at a reduced rate .
pair production and pair cascades therefore do not take place near the neutron star surface in the slot gap , as do the pair cascades along field lines closer to the magneic pole ( core ) , but occur at much higher altitudes .
there are several important differences between our revised slot - gap model and the original slot - gap model of arons & sharlemann ( 1979 ) .
the inclusion of general relativistic frame dragging enables particle acceleration on both favorably " and unfavorable " curved field lines and also at all inclination angles .
we also consider the radiation from pair cascades occurring along the interior edge of the slot gap .
the cascade radiation emission beam from the slot gap is thus a full hollow cone centered on the magnetic axis .
a narrower emission beam from field lines interior to the slot gap will form a core component of pairs and high - energy emission .
the electrodynamics of the slot gap is primarily dependent on a single parameter , the slot gap width , @xmath1 .
the ratio of the electric field in the slot gap to the electric field in the core region of the pc is @xmath2 and the luminosity of particles accelerated in the slot gap is @xmath3 where @xmath4 is the spin - down luminosity of the pulsar ( for full details see muslimov & harding 2003 ) .
one can estimate the width of the slot gap as defined by the magnetic colatitude where the variation of the curvature radiation pff becomes comparable to a fraction @xmath5 of the stellar radius , or [ sgwidth ] _ _ sg \ { lr 0.2 p_0.1(b_12)^-4/7i_45 ^ -3/7 .
[ deltaxi1 ] & b 0.1 b_cr + 0.1 p_0.1(b_12 ^ 3/4)^-4/7 i_45 ^ -3/7 & b > ~0.1 b_cr . where @xmath6 s , @xmath7 g , and @xmath8 , are the ns rotation period , surface magnetic field and moment of inertia .
the emission solid angle of radiation from the slot gap can be estimated by integrating over the thin annulus defined by the slot gap width ( eqn [ [ sgwidth ] ] ) .
[ omega ] _ _ 0 ^ 2 _ _ sg ster , _ _ . where @xmath9 is the dimensionless radius of emission and @xmath10 is the pc half - angle . based on the luminosity of the primary electrons and the above solid angle estimate
, we may derive the quantity , [ l / omega1 ] l _ ( _ ) = _ l_prim = 310 ^ 34 _
l_sd , 35 ^ 3/7 p_0.1 ^ 5/7 r_6 ^ 17/7 ( _ _ ) erg s^-1 ster ^-1 . where @xmath11 cm is the neutron star radius , @xmath12 is the radiation efficiency and @xmath13 is the spin - down luminosity .
the above expressions for @xmath14 are equivalent to the observed quantity @xmath15 , where @xmath16 is the high - energy bolometric flux observed at the earth , and @xmath17 is the distance to the pulsar .
figure 2 shows the pulsar empirical ( solid circles with error bars ) and theoretical ( upside - down triangles ) values of @xmath18 as a function of spin - down luminosity , @xmath19 .
the theoretical values are calculated for the parameters @xmath20 and @xmath21 ( see eq .
[ [ l / omega1 ] ] ) .
note that parameter @xmath22 can range from 0.2 to 0.5 in cascade simulations , and @xmath23 . in figure 2
the dashed line represents the absolute upper limit , where the spin - down luminosity is radiated into the unit solid angle , i.e. where @xmath24 .
one can see that there is good agreement for most high - energy pulsars except several of the pulsars , geminga and psr b0656 + 14 , having low @xmath19 , and for j0218 + 4232 , which is a millisecond pulsar .
these pulsars are near or below the death line for pairs from curvature radiation ( see harding & muslimov 2002 , and next section ) , and therefore have either very wide slot gaps or no slot gaps at all .
all other high - energy pulsars depicted in figure 2 are above their curvature radiation death lines and are expected to have slot gaps .
in order to explore the characteristics of high - energy emission produced by particles accelerated in the slot gap , we have carried out monte carlo simulations of the pair cascades initiated by primary electrons accelerated by the @xmath25 in the gap .
these pair cascade simulations are based on those of daugherty & harding ( 1996 ) , but using the electric field of the slot gap ( see mulsimov & harding 2003 ) . to illustrate the features of the sg acceleration and cascades , we use the period ( @xmath26 ) and surface magnetic field ( @xmath27 ) of the crab pulsar for the case @xmath28 . a stellar radius of 16 km is assumed .
figure 3 shows examples of pulse profiles for different inclination angles @xmath29 and seen at various viewing angles @xmath30 to the rotation axis .
it is evident that widely spaced double - peaked profiles can result at relatively small inclination by slicing through the middle of the hollow cone .
profiles with two pulses separated by more than 100 degrees , as are seen from the crab , vela and geminga , can be formed when the inclination angle is comparable to the beam opening angle ( about 20 degrees in this model for the crab ) .
broad single pulses , as seen from psr b1509 - 58 , may be produced when the observer cuts along the edge of the cone .
as the inclination angle increases , the maximum pulse separation decreases and the pulses themselves also become narrower . bridge emission as seen in vela and geminga high - energy profiles ( and in the crab profile around 1 mev ) is not produced in this calculation since we have modeled only cascades and emission from the slot gap , which produces a pure hollow cone beam .
the core cascades will produce additional emission on field lines interior to the slot gap which will fill in the profile between the two pulses and produce bridge emission .
the revised picture of acceleration and radiation from pulsar slot gaps makes several observational predictions that seem to agree with some of the properties of the known high - energy pulsars . although the total luminosity produced by the slot gap is lower than that from the whole polar cap , the small solid angle of the slot gap allows for fluxes that are in agreement with those observed .
the opening angle of the high - energy emission beam from the slot gap is much larger than that of emission from near the neutron star surface .
this naturally allows for wider , double - peaked pulse profiles as observed in several @xmath0-ray pulsars .
however , this type of profile still requires small magnetic inclination angles that are comparable to the emission beam opening angle . even for the fast pulsars ,
the slot gap beam opening angles are only @xmath31 . applying this model to the observed @xmath0-ray pulsars
would therefore require that these , and probably most , pulsars have inclination angles smaller than @xmath32 and our viewing angle can not be much larger .
this may be problematic in light of the chandra x - ray image of the crab nebula ; if the plane of the torus is really orthogonal to the spin axis , then our viewing angle is @xmath33 .
however , a number of other , more recently discovered , young pulsars seem to have small magnetic inclination ( harding et al .
2003 ) and slot - gap radiation may be visible at x - ray and @xmath0-ray energies . | the characteristics of the high - energy emission from polar cap accelerators will be discussed .
particles accelerated in the slot gap " near the polar cap rim will reach altitudes of several stellar radii before initiating pair cascades , producing a wide hollow cone of emission in young pulsars and some millisecond pulsars .
model x - ray and gamma - ray spectra and pulse profiles , based on monte - carlo simulations of polar cap pair cascades , will be presented .
# 1 # 2 _ astron .
astrophys . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
lett . _ * # 1 * , l#2 # 1 # 2 _ astron .
astrophys .
rev . _ * # 1 * , # 2 # 1 # 2 _ astron .
astrophys .
suppl .
ser . _ * # 1 * , # 2 # 1 # 2 _ astron .
j. _ * # 1 * , # 2 # 1 # 2 _ ann .
rev .
astron .
astrophys . _
* # 1 * , # 2 # 1 # 2 _ astrophys .
j. _ * # 1 * , # 2 # 1 # 2 _ astrophys .
j. lett .
_ * # 1 * , l#2 # 1 # 2 _ astrophys .
j. suppl .
_ * # 1 * , # 2 # 1 # 2 _ astrophys .
space sci .
_ * # 1 * , # 2 # 1 # 2 _ adv .
space res .
_ * # 1 * , # 2 # 1 # 2 _ bull .
astron .
inst .
czechosl .
_ * # 1 * , # 2 # 1 # 2 _ j. quant .
spectrosc .
radiat .
transfer _ * # 1 * , # 2 # 1 # 2 _ mon . not .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ mem .
r. astr .
soc . _ * # 1 * , # 2 # 1 # 2 _ phys .
lett .
rev . _ * # 1 * , # 2 # 1 # 2 _ publ .
astron .
. japan _ * # 1 * , # 2 # 1 # 2 _ publ .
astr .
soc .
pacific _ * # 1 * , # 2 # 1 # 2 _ nature _ * # 1 * , # 2 # 1 # 2 _ mem .
soc .
astron .
it . _ * # 1 * , # 2 # 1 # 2 _ the messenger _ * # 1 * , # 2 # 1 # 2 _ astron .
nach . _ * # 1 * , # 2 epsf.sty |
blazars are radio - loud active galactic nuclei ( agn ) viewed at small angles between the jet axis and the line of sight . due to relativistic beaming
, they are the brightest and most variable high energy sources among agns . bl lacertae objects , a subclass of blazars @xcite ,
are completely dominated by nonthermal emission and show only faint or no emission lines at all .
their spectral energy distribution ( sed ) extends over more than 15 orders of magnitude and exhibits two pronounced peaks when plotted as @xmath1 versus @xmath2 .
high - peaked bl lacertae objects ( hbls ) are characterized by the first peak being located in the uv to x - ray range , whereas the second peak appears at @xmath3-ray energies ( gev to tev ) .
the non - thermal , beamed emission detected from hbls is commonly ascribed to synchrotron and inverse - compton radiation from ultrarelativistic charged particles , accelerated at shocks in jets moving with relativistic bulk speed .
progress in understanding the roles of shocks and plasma turbulence for particle acceleration and the interplay with the dynamics of the jet expansion is expected from modelling the sed of blazars . at present , it is unclear whether the observed emission is entirely due to accelerated electrons , as in synchrotron - self - compton ( ssc ) models ( e.g. @xcite , @xcite , @xcite ) , or whether the high - energy emission is due to pions produced by accelerated protons and ions and subsequent pion decay ( e.g. @xcite , @xcite ) .
owing to their strong spectral and flux variability at all wavebands down to minutes ( see e.g. @xcite ) , simultaneous multi - wavelength ( mwl ) campaigns are essential to study the sed of hbls . of paramount importance are hereby the x - ray and @xmath3-ray regimes , where the two sed peaks are located for hbls . in former mwl campaigns ,
the minimum time scale of truly simultaneous observations was restricted by the low sensitivity of the participating vhe telescopes , enabling the investigation of transient events only in high flux states . with the advent of a new generation of imaging atmospheric cherenkov telescopes with improved sensitivity and lowered energy threshold such as magic , h.e.s.s . and veritas ,
the detection of hbls in a few hours becomes feasible also during low flux states . in this proceedings we present magic ( @xcite , @xcite ) observations of the well - known tev blazars mrk421 and mrk501 in 2006 .
for individual nights , these were conducted ( quasi-)simultaneously with h.e.s.s .
@xcite , suzaku @xcite , _ xmm - newton _
@xcite , integral @xcite and kva @xcite .
[ cols="^,^,^,^,^,^,^,^,^ " , ] the number * _ nnnn _ * can be found from the conference web page after login as participant ( http://icrc2009.uni.lodz.pl/ ) .
this is the i d number of the submitted abstract converted to four ( 4 ) digits ( e.g. , the paper corresponding to the abstract number 23 translates into icrc0023.pdf ) .
+ these files must be archived to a * single compressed file * for submission in gzipped tar or zipped format .
the archive should be named according to the scheme : + * _ icrcnnnn_bw.tar.gz _ * or * _ icrcnnnn_bw.zip _ * , + where * _ bw _ * denotes black white version .
+ if you prefer to have your paper on cd rom in colour , you should prepare also additional coloured version of your figures and pdf file of the paper , and send us two archives : one with black - white files ( for printed version of proceedings ) and the second one with files in colour ( for pre - conference cd rom ) .
both archives should be constructed as described above . + the archive with coloured version should be named according to the scheme : * _ icrcnnnn_colour.tar.gz _ * or * _
icrcnnnn_colour.zip_*. + if version in colour will not be submitted , then automatically black white version of the paper will be included in the cd rom .
before submitting the final version of the paper please remember to : + * manually equalize the lengths of the two columns on the last page of your paper ( see @xcite ) . *
+ please send the files + * _ icrcnnnn_bw.tar.gz _ * or * _ icrcnnnn_bw.zip _ * + ( or additionally + * _ icrcnnnn_colour.tar.gz _ * or * _ icrcnnnn_colour.zip _ * ) + as e
mail attachment to : + + ( please note the difference of addresses and do not send files to [email protected] . ) + the subject of e - mail should contain the name of the author presenting the paper and the i d number of corresponding abstract . +
examples of figures : simple ( fig . [ simp_fig ] ) , double column using two subfigures ( fig . [ double_fig ] ) , and double column using one big figure ( fig . [ wide_fig ] ) table [ table_simple ] is an example of a simple table . note that footnotes inside tables must be created by a pair of commands : ` \footnotemark ` inside ` table ` enviroment and ` \footnotetext{text } ` outside it .
command ` \footnote ` has no effect .
example of itemized list : * d the name of the city * d example of itemized list with castomized labels : * d * d example of enumerated list : 1 . lotnisko airport 2 .
bus 175 from airport to warsaw train station example of list with sublists : * dzie dobry
good morning * * dziekuj thanks * * prosz here you are ; please * do widzenia good bye * dobranoc good night * instytut prawa law department ( icrc host ) 1 .
d fabryczna
train station from / to warsaw 2 .
warszawa centralna
train station in warsaw * instytut fizyki
institute of physics one can use ieee list style , i.e. ` \begin{itemize } ` ` [ \ieeesetlabelwidth{z } ] ` , but it is not necessary .
footnotes can be created with the ` \footnote ` + ` { this is footnote } ` command .
equation [ simp_eqn ] is an example of a simple equation @xmath4 an example of multilines equation , with lines numbered or not : @xmath5 the array environment can be used for expression of the following type : @xmath6 when you find not correct hyphenation or you would like to force latex to split the word , then place the word with suggested hyphenation in the list of ` \hyphenation ` command which is before the title of the paper in the source _
template.tex _ file .
e.g. abcdefghijklmnoprstuwxyz .
do not use ` \bibtex ` .
as a base of 31@xmath7 icrc proceedings we decided to use ieee latex style , _ ieeetran _ version v1.7a .
usage of it is described in @xcite in details ( we include it with this template package ) .
99 m. shell , _ how to use the ieeetran latex class _ http://tug.ctan.org/get/macros/latex/contrib/ieeetran/ieeetran_howto.pdf t. oetiker h. partl , i. hyna , e. schlegl , _ a ( not so ) short introduction to latex2e _ http://www.ctan.org/tex-archive/info/lshort/ | original abstract the magic telescope observed in 2006 several vhe blazars in the course of multi - frequency campaigns , comprising measurements in the optical , x - ray and vhe regime . amongst others ,
the well - know tev blazar mrk 421 has been studied in detail .
magic detected mrk 421 each day of these campaigns with high significance , establishing once more flux variability on nightly scales for this object .
for certain nights , the integral flux exceeded the one of crab significantly .
the individual data sets have been examined for intra - night variability and were used to search for a correlation between the spectral index and the flux . in this contribution
, the results of the magic observations of different blazars will be presented in detail . in 2006
the magic telescope observed the well known very high energy ( vhe , @xmath080gev ) blazars mrk421 and mrk501 in the course of multi - wavelength campaigns , comprising measurements in the optical , x - ray and vhe regime .
magic performed additional snapshot observations on mrk421 around the mwl campaigns and detected the source each night with high significance , establishing once more flux variability on nightly scales for this object . for certain nights ,
the integral flux exceeded the one of crab significantly , whereas the truly simultaneous observations have been conducted in a rather low flux state .
the magic observations contemporaneous to _ xmm - newton _ revealed clear intra - night variability .
no significant correlation between the spectral index and the flux could be found for the nine days of observations .
the vhe observations of mrk501 have been conducted during one of the lowest flux states ever measured by magic for this object .
the vhe and optical light curves do not show significant variability , whereas the flux in x - rays increased by about 50% . in this contribution
, the results of the magic observations will be presented in detail .
vhe mwl blazar |
in the present paper we consider overdetermined problems for the fractional laplacian in unbounded exterior sets or bounded annular sets .
different cases will be taken into account , but the results obtained will lead in any case to the classification of the solution and of the domain , that will be shown to possess rotational symmetry . the notation used in this paper will be the following . given an open set whose boundary is of class @xmath0 , we denote by @xmath1 the inner unit normal vector and for any @xmath2 on the boundary of such set , we use the notation @xmath3 of course , when writing such limit , we always assume that the limit indeed exists and we call the above quantity the _ inner normal @xmath4-derivative _ of @xmath5 at @xmath2 .
the parameter @xmath4 above ranges in the interval @xmath6 and it corresponds to the fractional laplacian @xmath7 .
a brief summary of the basic theory of the fractional laplacian will be provided in section [ sec : preliminaries ] : for the moment , we just remark that @xmath7 reduces to ( minus ) the classical laplacian as @xmath8 , and the quantity in becomes in this case the classical neumann condition along the boundary . with this setting
, we are ready to state our results . for the sake of clarity , we first give some simplified versions of our results which are `` easy to read '' and deal with `` concrete '' situations .
these results will be indeed just particular cases of more general theorems that will be presented later in section [ ffgg ] .
more precisely , the results we present are basically of two types . the first type deals with _ exterior sets_. in this case , the equation is assumed to hold in @xmath9 , where @xmath10 is a non - empty open bounded set of @xmath11 , not necessarily connected , whose boundary is of class @xmath0 , and @xmath12 denotes the closure of @xmath10 .
we sometimes split @xmath10 into its connected components by writing @xmath13 where any @xmath14 is a bounded , open and connected set of class @xmath0 , and , to avoid pathological situations , we suppose that @xmath15 is finite . notice that @xmath16 if @xmath17 .
then , the prototype of our results for exterior sets is the following .
[ thm : example ] let us assume that there exists @xmath18 such that @xmath19 then @xmath10 is a ball , and @xmath5 is radially symmetric and radially decreasing with respect to the centre of @xmath10 .
the second case dealt with in this paper is the one of _
annular sets_. namely , in this case the equation is supposed to hold in @xmath20 , where @xmath21 is a bounded set that contains @xmath10 and such that @xmath20 is of class @xmath0 .
then , the prototype of our results for annular sets is the following .
[ thm : example 2 ] let us assume that there exists @xmath18 such that @xmath22 where @xmath23 denotes the inner ( with respect to @xmath20 ) normal @xmath4-derivative of @xmath5 . then @xmath10 and @xmath21 are concentric balls , and @xmath5 is radially symmetric and radially decreasing with respect to the centre of @xmath10 .
we stress that , here and in the following , we do not assume that @xmath9 ( resp .
@xmath20 ) is connected ; this is why we do not use the common terminology _ exterior domain _
( resp . _ annular domain _ ) , which indeed refers usually to an exterior connected set ( resp .
annular connected set ) .
we also stress that , while the dirichlet boundary datum has to be the same for all the connected components of @xmath10 , the neumann boundary datum can vary . overdetermined elliptic problems have a long history , which begins with the seminal paper by j. serrin @xcite .
a complete review of the results which have been obtained since then goes beyond the aims of this work . in what follows , we only review the contributions regarding exterior or annular domains in the local case , and the few results which are available about overdetermined problems for the fractional laplacian
. overdetermined problems for the standard laplacian in exterior domains have been firstly studied by w. reichel in @xcite , where he assumed that both @xmath10 and @xmath9 are connected .
in such a situation , w. reichel proved that if there exists @xmath24 ( i.e. of class @xmath0 up to the boundary of the exterior domain ) such that @xmath25 where @xmath26 denotes the usual normal derivative , and @xmath27 is a locally lipschitz function , non - increasing for non - negative and small values of @xmath28 , then @xmath10 has to be a ball , and @xmath5 is radially symmetric and radially decreasing with respect to the centre of @xmath10 .
the proof is based upon the moving planes method . with a different approach , based upon the moving spheres ,
a. aftalion and j. busca @xcite addressed the same problem when @xmath29 is not necessarily non - increasing for small positive values of its argument .
in particular , they could treat the interesting case @xmath30 for @xmath31 .
afterwards , b. sirakov @xcite proved that the result obtained in @xcite holds without the assumption @xmath32 , and for possibly multi - connected sets @xmath10 .
moreover , he allowed different boundary conditions on the different components of @xmath10 : that is , @xmath33 and @xmath34 on @xmath35 can be replaced by @xmath36 and @xmath37 on @xmath38 , with @xmath39 and @xmath40 depending on @xmath41 .
his method works also in the setting considered in @xcite .
we point out that in @xcite a quasi - linear regular strongly elliptic operator has been considered instead of the laplacian .
concerning quasi - linear but possibly degenerate operators , we refer to @xcite .
as far as overdetermined problems in annular domains is concerned , we refer the reader to @xcite . in @xcite , g. alessandrini proved the local counterpart of theorem [ thm : example 2 ] for quasi - linear , possibly degenerate , operators .
this enhanced the results in @xcite . in @xcite
, w. reichel considered inhomogeneous equations for the laplace operator in domains with one cavity . regarding the nonlocal framework ,
the natural counterpart of the j. serrin s problem for the fractional @xmath4-laplacian ( @xmath42 ) has been recently studied by m. m. fall and s. jarohs in @xcite . in such contribution the authors introduced the main tools for dealing with nonlocal overdetermined problems , such as comparison principles for anti - symmetric functions and a nonlocal version of the serrin s corner lemma .
such results will be used in our work .
we refer also to @xcite , where the authors considered a similar problem in dimension @xmath43 , and for @xmath44 .
as already mentioned , theorems [ thm : example ] and [ thm : example 2 ] are just simplified versions of more general results that we obtained
. next section will present the results obtained in full generality .
now we present our results in full generality .
for this , first we consider the overdetermined problem in an exterior set @xmath9 , namely : @xmath45 note that the @xmath4-neumann boundary datum can depend on @xmath46 .
our first main result in this framework is the following .
[ thm : main 1 ] let @xmath27 be a locally lipschitz function , non - increasing for nonnegative and small values of @xmath28 , and let @xmath47 for every @xmath46 . if there exists a weak solution @xmath18 of , satisfying @xmath48 then @xmath10 is a ball , and @xmath5 is radially symmetric and radially decreasing with respect to the centre of @xmath10 .
the concept of weak solution and its basic properties will be recalled in section [ sec : preliminaries ] for the facility of the reader .
we point out that under additional assumptions on @xmath29 , the assumption @xmath49 in theorem [ thm : main 1 ] can be dropped , and the condition @xmath50 can be relaxed .
[ thm : main 1 prime ]
let @xmath29 be a locally lipschitz function , non - increasing in the whole interval @xmath51 $ ] .
if there exists a weak solution @xmath18 of satisfying @xmath52 then @xmath10 is a ball , and @xmath5 is radially symmetric and radially decreasing with respect to the centre of @xmath10 . other results in the same direction are the following .
[ corol : main 1 ] under the assumptions of theorem [ thm : main 1 ] , let us assume that @xmath53 ; then the request @xmath54 in the statement of theorem [ thm : main 1 ] is not necessary , and condition @xmath55 can be replaced by @xmath56 .
[ corol : subharmonicity ] if under the assumptions of theorem [ thm : main 1 ] , we suppose that @xmath57 for @xmath58 , then the request @xmath54 is not necessary , and condition @xmath55 can be replaced by @xmath59 .
analogously , if under the assumptions of theorem [ thm : main 1 prime ] , we suppose that @xmath57 for @xmath58 , then condition @xmath56 can be replaced by @xmath59 . the proof of the corollaries is based upon simple comparison arguments . when @xmath60 , in the same way one could show that assumption @xmath59 is not necessary , obtaining in particular theorem [ thm : example ] .
it is worth to notice that the regularity assumption @xmath18 is natural in our framework , see theorem [ thm : regularity ] in the appendix at the end of the paper : what we mean is that each bounded weak solution of the first two equations in is of class @xmath61 .
all the previous results ( and the forthcoming ones ) could have stated for bounded weak solutions , without any regularity assumption .
we preferred to assume from the beginning that @xmath18 , since in this way the condition on @xmath23 makes immediately sense , without further observations .
the @xmath61 regularity is optimal , as shown by the simple example @xmath62 which has the explicit solution @xmath63 , where @xmath64 denotes the positive part of @xmath65 , and @xmath66 is a normalization constant depending on @xmath67 and on @xmath4 .
also , eigenfunctions are not better than @xmath68 , see e.g. @xcite .
[ rem : on regularity ] in the local setting @xcite , see also @xcite , the solution @xmath5 of is supposed to be of class @xmath0 up to the boundary . in this way , the authors could avoid to assume that @xmath69 : indeed , in proposition 1 in @xcite , as well as in step 4 in the proof of the main results in @xcite , the authors computed the second derivatives of @xmath5 on the boundary of @xmath10 . in our context , it seems not natural to ask that @xmath5 has better regularity than @xmath70 , and this is the main reason for which we need to suppose @xmath71 in theorem [ thm : main 1 ] .
we think that it is worth to point out that there exist solutions of satisfying all the assumptions of the above statements when @xmath10 is a ball .
such existence results will be stated and proved at the end of the paper , in section [ sec : existence ] , for the sake of completeness . as already pointed out ,
the fact that @xmath9 is not supposed to be connected marks a difference with respect to the local case . the same difference , which arises also in @xcite , is related to the non - local nature of both the fractional laplacian and the boundary conditions .
now we present our results in the general form for an annular set @xmath20 ( recall that in this case @xmath21 is bounded , and , by the regularity assumed , @xmath10 can not be internally tangent to @xmath72 ) .
recall that @xmath21 can be multi - connected but , in this case , we assume that @xmath21 has a finite number of connected components .
the natural counterpart of the overdetermined problem for annular sets is given by @xmath73 where the notation @xmath74 is used for the inner normal @xmath4-derivative in @xmath75 . in this
setting we have : [ thm : main 2 ] let @xmath29 be locally lipschitz .
let us assume that there exists a weak solution @xmath18 of satisfying @xmath76 then @xmath21 and @xmath10 are concentric balls , and @xmath5 is radially symmetric and radially decreasing with respect to their centre .
note that in this case no assumption on the monotonicity of @xmath29 , or on the sign of @xmath40 and @xmath77 , is needed .
as for the problem in exterior sets , the condition @xmath78 can be relaxed under additional assumptions on @xmath29 .
[ thm : main 2 prime ] let @xmath29 be a locally lipschitz function , non - increasing in @xmath51 $ ] . if there exists a weak solution @xmath18 of satisfying @xmath79 then both @xmath21 and @xmath10 are concentric balls , and @xmath5 is radially symmetric and radially decreasing with respect to their centre .
[ corol : main 2 ] under the assumptions of theorem [ thm : main 2 ] , let us assume that @xmath53 ; then the condition @xmath78 can be replaced by @xmath80 in @xmath11 .
analogously , if @xmath81 , then the condition @xmath78 can be replaced by @xmath50 in @xmath11 . clearly , if both @xmath53 and @xmath81 , we obtain the thesis for @xmath82 and @xmath56 , and then we also obtain theorem [ thm : example 2 ] as a particular case .
the proof of corollary [ corol : main 2 ] is analogue to that of corollary [ corol : main 1 ] ( and thus will be omitted ) .
one could also state a counterpart of corollary [ corol : subharmonicity ] in the present setting . at last
, we observe that when @xmath10 ( or @xmath20 ) is a priori supposed to be radial , our method permits to deduce the radial symmetry of the solutions of the dirichlet problem . in the local framework , this type of results have been proved in @xcite .
[ thm : radial 1 ] let @xmath83 be a ball , and let @xmath18 be a weak solution of @xmath84 such that @xmath85 if @xmath27 is a locally lipschitz function , non - increasing for nonnegative and small values of @xmath28 , and @xmath5 satisfies @xmath86 then @xmath5 is radially symmetric and radially decreasing with respect to @xmath2 . when compared with the local results , condition seems not to be natural . on the other hand ,
for the reasons already explained in remark [ rem : on regularity ] we could not omit it in general .
nevertheless , under additional assumptions on @xmath29 it can be dropped .
[ corol : radial ] if under the assumptions of theorem [ thm : radial 1 ] we suppose that @xmath53 , then can be omitted , and condition @xmath87 can be replaced by @xmath56 .
if moreover @xmath57 for every @xmath88 , then condition @xmath55 can be replaced by @xmath59 .
it is clear that similar symmetry results hold for dirichlet problems in annuli .
an interesting limit case takes place when @xmath89 is a single point of @xmath11 .
the reader can easily check that the proof of theorem [ thm : radial 1 ] works also in this setting . with some extra work ,
we can actually obtain a better result .
[ thm : radial point ] let us assume that there exists a bounded weak solution of @xmath90 if @xmath27 is a locally lipschitz functions , non - increasing for nonnegative and small values of @xmath28 , and @xmath5 satisfies @xmath91 then @xmath5 is radially symmetric and radially decreasing with respect to @xmath2 .
notice that no condition on the @xmath4-normal derivative is needed .
for this reason , we omitted the assumption @xmath18 , which anyway , by theorem [ thm : regularity ] , would be natural also in this context . as a straightforward corollary , we obtain a variant of the gidas - ni - nirenberg symmetry result for the fractional laplacian .
let @xmath5 be a nonnegative bounded weak solution of @xmath92 in @xmath11 .
if @xmath27 is a locally lipschitz functions , non - increasing for nonnegative and small values of @xmath28 , and @xmath5 satisfies @xmath93 then @xmath5 is radially symmetric and radially decreasing with respect to a point of @xmath11 .
analogue symmetry results have been proved in @xcite , for a different class of nonlinearities @xmath29 ( having non - empty intersection with the one considered here ) .
we point out that , with respect to @xcite , we do not require any condition at infinity on the decay of @xmath5 .
the basic technical definitions needed in this paper will be recalled in section [ sec : preliminaries ] . in section [ sec : over exterior ] we consider overdetermined problems in exterior sets , proving theorems [ thm : main 1 ] , [ thm : main 1 prime ] and corollaries [ corol : main 1 ] , [ corol : subharmonicity ] . section [ sec : over annular ] is devoted to overdetermined problems in annular sets . in section [ sec : radial ] we study the symmetry of the solutions when the domain is a priori supposed to be radial . in section [ sec : existence ] we present some existence results .
finally , in a brief appendix we discuss the regularity of bounded weak solutions of dirichlet fractional problems in unbounded sets .
we collect in this section some definitions and results which will be used in the proofs of the main theorems .
let @xmath94 and @xmath95 . for a function @xmath96 , the fractional @xmath4-laplacian
is defined by @xmath97 where @xmath98 is a normalization constant , and @xmath99 stays for principal value " . in the rest of the paper ,
to simplify the notation we will always omit both @xmath98 and @xmath99 .
the bilinear form associated to the fractional laplacian is @xmath100 it can be proved that @xmath101 defines a scalar product , and we denote by @xmath102 the completion of @xmath103 with respect to the norm induced by @xmath101 .
we also introduce , for an arbitrary open set @xmath104 , the space @xmath105 .
it is a hilbert space with respect to the scalar product @xmath106 , where @xmath107 stays for the scalar product in @xmath108 .
the case @xmath109 is admissible .
we write that @xmath110 if @xmath111 for every compact set @xmath112 .
a function @xmath113 is a weak supersolution of @xmath114 if @xmath115 and @xmath116 if the opposite inequality holds , we write that @xmath113 is a weak subsolution . if @xmath117 and equality holds in for every @xmath118
, then we write that @xmath113 is a weak solution of @xmath119 since we will always consider weak solutions ( supersolutions , subsolutions ) , the adjective weak will sometimes be omitted .
let @xmath120 be a weak solution of @xmath121 with @xmath29 locally lipschitz continuous . by theorem [ thm : regularity ] in the appendix , we know that * @xmath122 for some @xmath123 ( _ interior regularity _ ) ; * @xmath18 , and in particular @xmath124 for some @xmath125 , where @xmath126 denotes the distance from the boundary of @xmath10 ( _ boundary regularity _ ) . since any weak solution @xmath18 of such that @xmath127 as @xmath128 is in @xmath129 ( and also in @xmath102 , by definition of weak solution )
, the previous regularity results will be used throughout the rest of the paper .
we recall a strong maximum principle and a hopf s lemma for anti - symmetric functions ( * ? ? ?
* proposition 3.3 and corollary 3.4 ) . in the quoted paper ,
the strong maximum principle is stated under the assumption that @xmath21 is bounded , but for the proof this is not necessary . as a result ,
the following holds .
[ strong ] let @xmath130 be a half - space , and let @xmath131 ( not necessarily bounded ) .
let @xmath132 , and let @xmath113 satisfy @xmath133 where @xmath134 denotes the reflection of @xmath135 with respect to @xmath136 .
then either @xmath137 in @xmath21 , or @xmath138 in @xmath139 .
furthermore , if @xmath140 and @xmath141 , then @xmath142 , where @xmath143 is the outer unit normal vector of @xmath21 in @xmath2 .
if @xmath131 shares part of its boundary with the hyperplane @xmath144 , and @xmath145 , then we can not apply the hopf s lemma in @xmath2 , since it is necessary to suppose that it lies on the boundary of a ball compactly contained in @xmath139 .
this assumption is used in the proof in @xcite . in the proof of theorem [ thm : main 1 ]
, we shall need a version of the hopf s lemma allowing to deal with points of @xmath146 . to be more precise ,
let @xmath147 be a @xmath0 set in @xmath11 , symmetric with respect to the hyperplane @xmath148 , and let @xmath139 be a half - space such that @xmath149 .
let @xmath150 , and let us assume that @xmath151 satisfies @xmath152 where @xmath153 and @xmath134 denotes the reflection of @xmath135 with respect to @xmath148 .
we note that @xmath154 is divided into two parts : a regular part of hausdorff dimension @xmath155 , which is a relatively open set in @xmath148 , and a singular part of hausdorff dimension @xmath156 , which is @xmath157 .
we also note that , by anti - symmetry , @xmath158 for every @xmath159 . [ prop : new hopf ] in the previous setting , if @xmath2 is a point in the regular part of @xmath160 , then @xmath161 where @xmath162 is the outer unit normal vector to @xmath149 in @xmath2 . at a first glance
it could be surprising that a boundary lemma involving a fractional problem gives a result on the full outer normal derivative of the function .
namely , functions satisfying a fractional equation of order @xmath163 are usually not better than @xmath61 at the boundary , so the first order incremental quotient in is in general out of control . but in our case , if we look at the picture more carefully , we realize that since we are assuming that @xmath113 is an anti - symmetric solution in the whole @xmath147 ( which contains both @xmath21 and its reflection ) , any @xmath2 on the regular part of @xmath160 is actually an interior point for @xmath113 , and hence it is natural to expect some extra regularity . without loss of generality , we can assume that @xmath164 so that @xmath165 .
let @xmath166 be such that @xmath167 .
if necessary replacing @xmath168 with a smaller quantity , we can suppose that @xmath169 and @xmath170 are both compactly contained in @xmath147 .
now we follow the strategy of lemma 4.4 in @xcite : for @xmath171 to be determined in the sequel , we consider the barrier @xmath172 where @xmath173 is the positive solution of @xmath174 and @xmath175 are the truncated distance functions from the boundary of @xmath176 and @xmath177 , respectively . with this definition
, we can compute @xmath178 exactly as in lemma 4.4 in @xcite , proving that @xmath179 provided @xmath180 is sufficiently large . by continuity and recalling that @xmath137 in @xmath21 , we deduce that @xmath181 in @xmath176 .
this permits to choose a positive constant @xmath182 such that @xmath183 in @xmath184 , and hence @xmath185 in @xmath186 .
therefore , the weak maximum principle ( proposition 3.1 in @xcite ) implies that @xmath183 in @xmath21 , and in particular @xmath187 which gives the desired result .
as far as overdetermined problem in bounded exterior sets , namely theorem [ thm : main 2 ] , we shall make use of a maximum principle in domain of small measure , proved in ( * ? ? ?
* proposition 2.4 ) in a parabolic setting . in our context
, the result reads as follows .
[ small ] let @xmath188 be a half - space and let @xmath189 .
there exists @xmath190 such that if @xmath191 with @xmath192 , and @xmath5 satisfies in a weak sense @xmath193 with @xmath194 , then @xmath195 in @xmath196 .
in the first part of this section , we prove theorem [ thm : main 1 ] .
we follow the same sketch used by reichel in @xcite , applying the moving planes method to show that for any direction @xmath197 there exists @xmath198 such that both the @xmath10 and the solution @xmath5 are symmetric with respect to a hyperplane @xmath199 where @xmath200 denotes the euclidean scalar product in @xmath11 . in the following we fix the direction @xmath201 and use the notation @xmath202 for points of @xmath11 . for @xmath203
, we set @xmath204 it is known that , for @xmath205 a little smaller than @xmath206 , the reflection of @xmath207 with respect to @xmath208 lies inside @xmath14 , namely @xmath209 in addition , @xmath210 for every @xmath211 ( we recall that @xmath1 denotes the outer normal vector on @xmath38 , thus directed directed towards the interior of @xmath212 ) ; this remains true for decreasing values of @xmath205 up to a limiting position @xmath213 such that one of the following alternatives takes place : * _ internal tangency _ : the reflection @xmath214 becomes internally tangent to @xmath38 ; * _ orthogonality condition _ : @xmath215 for some @xmath216 .
let @xmath217 . for @xmath218 ,
it is clear that inclusion holds for every @xmath46 .
since @xmath16 for every @xmath17 , it follows straightforwardly that @xmath219 furthermore , @xmath220 can be characterized as @xmath221 this simple observation permits to treat the case of multi - connected interior sets @xmath10 essentially as if they were consisting of only one domains .
two preliminary results regarding the geometry of the reduced half - space are contained in the following statement .
[ lem : geom remark ] for every @xmath222 , the following properties hold : * @xmath223 is convex in the @xmath224-direction ; * the set @xmath225 is connected . for property ( @xmath46 )
, we show that for any @xmath222 , if a point @xmath226 belongs to @xmath227 , then also @xmath228 for every @xmath229 .
if this is not true , then there exist @xmath230 and @xmath231 with @xmath232 .
for @xmath233 , we have that @xmath234 , but @xmath231 , which is in contradiction with the fact that , since @xmath235 , the reflection of @xmath10 with respect to @xmath236 does not exit @xmath12 itself .
as far as property ( @xmath237 ) is concerned , given two points @xmath238 , @xmath239 , we fix a large @xmath240 and consider the two vertical segments @xmath241 , with @xmath242 $ ] . by point ( @xmath46 ) , these segments lie in @xmath243 .
each segment connects @xmath244 with @xmath245 .
then , since @xmath10 is bounded , if @xmath246 is large we can connect @xmath247 and @xmath248 with a horizontal segment @xmath249 lying well outside @xmath10 . in this way , by considering the two vertical segments and the horizontal one as a single polygonal , we have joined @xmath238 and @xmath250 by a continuous path that lies in @xmath243 .
this shows that @xmath243 is connected .
we define @xmath251 .
notice that @xmath252 where @xmath253 is in @xmath129 since @xmath5 is bounded and @xmath29 is locally lipschitz continuous .
we aim at proving that the set @xmath254 coincides with the interval @xmath255 , that @xmath256 for every @xmath257 , and that @xmath258 in @xmath259 . from this
we deduce that @xmath5 is symmetric with respect to @xmath260 , and non - increasing in the @xmath224 direction in the half - space @xmath261 .
furthermore , we shall deduce that @xmath21 ( and hence also @xmath10 ) is convex in the @xmath224 direction , and symmetric with respect to @xmath260 . as a product of the convexity and of the fact that @xmath262 in @xmath263 for @xmath264 , it is not difficult to deduce that @xmath5 is strictly decreasing in @xmath265 in @xmath266 . repeating the same argument for all the directions @xmath197
, we shall deduce the thesis .
although the strategy of the proof is similar to that of theorem 1.1 in @xcite , its intermediate steps will differs substantially .
we write that the hyperplane @xmath208 moves , and reaches a position @xmath267 , if @xmath268 in @xmath263 for every @xmath269 . with this terminology ,
the first step in the previous argument consists in showing that the movement of the hyperplane can start .
[ lem : moving initial ] there exists @xmath270 sufficiently large such that @xmath268 in @xmath263 for every @xmath271 .
we argue by contradiction , assuming that for a sequence @xmath272 there exists @xmath273 such that @xmath274 .
since @xmath275 on @xmath276 , and @xmath277 as @xmath278 tends to infinity , we can suppose that each @xmath279 is an interior minimum point of @xmath280 in @xmath281 .
notice that @xmath282 indeed , on one side @xmath283 , on the other side since @xmath284 we have both @xmath285 and @xmath286 , and since @xmath29 is monotone non - increasing for small value of its argument , we deduce by that @xmath287 .
now we show that @xmath288 for this , we consider the sets @xmath289 and @xmath290 .
notice that , by the minimality property of @xmath279 in @xmath291 , we have that @xmath292 .
therefore any integral in @xmath11 may be decomposed as the sum of four integrals , namely the ones over @xmath196 , @xmath293 , @xmath294 and @xmath295 . using this and the fact that @xmath296 outside @xmath196 , we have that @xmath297 also , if @xmath298 we have that latexmath:[$|x_k - y|\ge latexmath:[\[\int_{u}\frac{w_{\lambda_k}(x_k)-w_{\lambda_k}(y)}{|x_k - y|^{n+2s}}\,dy \le \int_{u}\frac{w_{\lambda_k}(x_k)-w_{\lambda_k}(y ) } { is positive in @xmath196 . by changing variable @xmath300 in the latter integral
, we obtain @xmath301 ( and renaming the last variable of integration ) , we can write as @xmath302 with strict inequality if @xmath303 since @xmath274 . this proves .
now we claim that @xmath304 indeed , we already know that both the quantities above are non - positive , due to and .
if at least one of them were strictly negative , their sum would be strictly negative too , and this is in contradiction with .
having established , we use it to observe that @xmath280 must be constant .
indeed , we firstly notice that @xmath305 ( otherwise , as already observed , in we would have a strict inequality ) .
this means that @xmath306 in the whole @xmath11 , and if the set @xmath307 would have positive measure , this would imply that @xmath308 . thus we conclude that @xmath309 , in contradiction with its anti - symmetry with respect to @xmath310 .
thanks to the above statement , the value @xmath311 is a real number .
we aim at proving that the hyperplane @xmath208 reaches the position @xmath220 , i.e. @xmath312 . in this perspective , a crucial intermediate result is the following .
[ lem : condizione per simmetria ] let @xmath222 . if @xmath313 for some @xmath314 , then @xmath10 is symmetric with respect to @xmath208 .
in particular , if @xmath268 in @xmath263 and @xmath315 , then @xmath262 in @xmath263 . by the strong maximum principle , proposition [ strong ] , we have that if @xmath313 for a point @xmath316 , then @xmath317 in @xmath261 .
let us assume by contradiction that @xmath10 is not symmetric with respect to @xmath208 .
at first , it is easy to check that @xmath318 hence , having assumed that @xmath10 is not symmetric we have @xmath319 ; since for @xmath218 the inclusion holds , this implies that @xmath320 with strict inclusion .
let @xmath321 ; on the right , the one for @xmath322 with @xmath10 not symmetric with respect to @xmath260.,title="fig:",height=188 ] ; on the right , the one for @xmath322 with @xmath10 not symmetric with respect to @xmath260.,title="fig:",height=188 ] for every @xmath323 , we have @xmath324 and hence @xmath325 , a contradiction .
so far we showed that if @xmath222 and @xmath326 vanishes somewhere in @xmath263 , then @xmath10 is symmetric with respect to @xmath208 .
but , by definition of @xmath220 , this can not be the case for @xmath315 , and hence for any such @xmath205 if we can prove that @xmath268 in @xmath263 , we immediately deduce that @xmath262 there . in the proof we used in a crucial way both the nonlocal boundary conditions , and the nonlocal strong maximum principle .
in particular , we point out that if @xmath313 at one point @xmath316 , then @xmath317 in the whole half - space @xmath261 . in the following
we shall make use the fact that from the sign of the @xmath4-derivative of a function @xmath5 in a given direction we can infer information about the monotonicity of @xmath5 itself . [
lem : monotonia ] let @xmath327 and @xmath151 .
if @xmath328 then @xmath113 is monotone decreasing in the direction @xmath143 in a neighborhood of @xmath2 . if the conclusion were false , there would exists a sequence @xmath329 for which @xmath330 .
then @xmath331 which is a contradiction .
note that the limit does exist , since we are assuming that @xmath151 .
we are finally ready to prove that the hyperplane @xmath208 reaches the critical position @xmath220 .
[ lem : moving continuation ] there holds @xmath332 . by contradiction
, we suppose that @xmath333 . by continuity
, we have @xmath334 in @xmath335 , and hence lemma [ lem : condizione per simmetria ] yields @xmath262 in @xmath263 .
since @xmath336 , there exist sequences @xmath337 , @xmath338 , and @xmath273 , such that @xmath339 . since @xmath340 on @xmath341 and @xmath277 as @xmath128 , it is not restrictive to assume that @xmath279 are interior minimum point of @xmath280 in @xmath281 .
if @xmath284 , we obtain a contradiction as in lemma [ lem : moving initial ] .
if @xmath342 is bounded , there exists @xmath343 such that , up to a subsequence , @xmath344 .
concerning the pre - compactness of the sequence @xmath345 , we recall that by definition @xmath346 the function @xmath5 is of class @xmath61 , and hence for every compact @xmath347 there exists @xmath348 such that @xmath349 therefore , the sequence @xmath280 is convergent to @xmath350 in @xmath351 for any @xmath352 .
the uniform convergence entails @xmath353 , and by lemma [ lem : condizione per simmetria ] this implies that @xmath354 is a boundary point . to continue the proof
, we have to distinguish among three different possibilities , and in each of them we have to find a contradiction .
_ case 1 ) @xmath134 lies on the regular part of @xmath355 .
_ we note that @xmath356 for every @xmath15 sufficiently large , where @xmath357 denotes the interior of a set .
since , by interior regularity ( see subsection [ sub : regularity ] ) , we know that @xmath122 for some @xmath123 , by definition @xmath358 is uniformly bounded in @xmath359
. in particular , since by interior minimality ( and interior regularity ) we have @xmath360 and @xmath361 in @xmath362 , we deduce that @xmath363 .
this is in contradiction with proposition [ prop : new hopf ] , where we have showed that @xmath364 _ case 2 ) @xmath365 . _ since @xmath366 , we know that @xmath367 ( otherwise we have to be in a critical position of internal tangency ) .
having assumed that @xmath55 in @xmath21 , we deduce that @xmath368 , and hence @xmath369 in contradiction with the fact that by convergence @xmath370 .
_ case 3 ) @xmath371 .
_ we observe that @xmath372 , and since @xmath373 , the outer ( with respect to @xmath10 ) unit normal vector @xmath374 is such that @xmath375 .
as @xmath5 is constant on @xmath376 , we have @xmath377 for every @xmath143 which is tangent to @xmath35 in @xmath134 . recalling that @xmath378 , and recalling lemma [ lem : monotonia ]
, we see that @xmath379 in a neighborhood @xmath380 .
let @xmath381 denote the reflection of @xmath279 with respect to @xmath310 .
since both @xmath382 , at least for @xmath15 sufficiently large the whole segment connecting @xmath279 with @xmath381 is contained in @xmath383 .
recalling that @xmath379 , this implies that @xmath5 is monotone decreasing along the segment @xmath384 $ ] , that is @xmath385 in contradiction with the fact that @xmath339 . for every @xmath222 , we have @xmath268 in @xmath263 .
if @xmath315 , then the strict inequality holds , and it remains to show that @xmath258 in @xmath259 .
to this aim , we argue by contradiction assuming that @xmath386 in @xmath259 , and we distinguish two cases .
the following argument is adapted by @xcite .
_ case 1 ) @xmath220 is a critical value of internal tangency .
_ there exists @xmath387 .
clearly we have @xmath388 , so that @xmath389 , and by proposition [ strong ] we deduce that @xmath390 , where @xmath1 denotes the outer unit normal vector to @xmath35 in @xmath2 .
we observe that it can not be @xmath391 and @xmath392 with @xmath17 .
this follows from the definition of @xmath220 and the fact that @xmath16 , see ( * ? ? ?
* lemma 2.1 ) for a detailed proof .
hence , having assumed that @xmath393 on @xmath38 , and observing that by internal tangency @xmath394 , we find also @xmath395 which is a contradiction .
_ case 2 ) @xmath220 is a critical value where the orthogonality condition is satisfied .
_ let @xmath396 , and let @xmath397 denote the distance function from the boundary of @xmath14 .
up to rigid motions , it is possible to suppose that @xmath398 , @xmath162 is a vector of the orthonormal basis , say @xmath399 , and @xmath400 is a diagonal matrix ; to ensure that the function @xmath126 is twice differentiable , we used the @xmath0 regularity of @xmath10 .
let @xmath401 . adapting step by step the proof of lemma 4.3 in @xcite , it is possible to deduce that @xmath402 as @xmath403 . in this step
we need to recall that @xmath124 , see subsection [ sub : regularity ] . on the contrary , thanks to the nonlocal version of the serrin s corner lemma ( lemma 4.4 in @xcite ; the result is stated therein in a bounded domain , but the reader can check that this is not use in the proof ) we also infer that @xmath404 for @xmath28 positive and small , for some constant @xmath348 .
this gives a contradiction .
we proved that @xmath258 . by lemma [ lem : condizione per simmetria ] , this implies that @xmath10 , and hence @xmath9 , are symmetric with respect to the hyperplane @xmath260 . in principle
both @xmath9 and @xmath10 could have several connected components .
but , as proved in lemma [ lem : geom remark ] , @xmath266 is connected , which implies by symmetry that @xmath405 is in turn connected . therefore
, if @xmath9 is not connected , necessarily @xmath10 contains a neighborhood of the hyperplane @xmath260 , which is not possible since @xmath10 is bounded .
as far as the connectedness of @xmath10 is concerned , we firstly observe that by property ( @xmath46 ) of lemma [ lem : geom remark ] and by symmetry , @xmath10 is convex in the @xmath224-direction .
let us assume by contradiction that there exists at least two connected components @xmath406 and @xmath407 of @xmath10 .
it is not possible that @xmath406 and @xmath407 meet at boundary points , since we assumed that @xmath10 is of class @xmath0 .
since @xmath10 is convex , there exists a hyperplane @xmath408 not parallel to @xmath260 separating @xmath406 and @xmath407 .
let @xmath409 be an orthogonal direction to @xmath408 .
defining @xmath410 without loss of generality we can suppose that @xmath411 , while @xmath412 .
in the same way as we defined @xmath220 for the direction @xmath224 , we can now define @xmath413 for @xmath409 , and prove that @xmath10 is symmetric with respect to @xmath414 .
but this gives clearly a contradiction , since by definition @xmath415 , while @xmath416 .
the proof of theorem [ thm : main 1 prime ] follows a different sketch , being based upon the following known result ( we refer to the appendix in @xcite for a detailed proof ) .
[ prop : criterion ] let @xmath417 be continuous and such that @xmath5 has a limit when @xmath128
. then the following statements are equivalent : * @xmath5 is radially symmetric and radially non - increasing with respect to a point of @xmath11 ; * for every half - space @xmath139 of @xmath11 we have that either @xmath418 in @xmath139 , or @xmath419 in @xmath139 , where @xmath420 denotes the reflection with respect to the boundary @xmath136 .
hence , to prove the radial symmetry of @xmath5 we aim at showing that condition ( @xmath237 ) in proposition [ prop : criterion ] is satisfied .
let us consider at first all the half - space @xmath139 such that @xmath136 is orthogonal to the @xmath224 direction . using the notation introduced at the beginning of this section , and recalling the definition of @xmath421
, we see that @xmath422 in @xmath11 for every @xmath205 .
thus , it is not difficult to adapt the proof of lemma [ lem : moving initial ] , using the fact that for @xmath222 we have @xmath268 in @xmath423 , to deduce that @xmath424 on the contrary , we point out that now we can not immediately conclude that @xmath262 in @xmath263 for @xmath315 , since in the proof of lemma [ lem : condizione per simmetria ] we have used both the assumptions @xmath71 and @xmath32 in @xmath9 .
nevertheless , we can prove that @xmath425 arguing exactly as in the conclusion of the proof of theorem [ thm : main 1 ] .
this line of reasoning can be used for all the direction @xmath197 . to be more explicit ,
we introduce the following notation : for a direction @xmath409 and a real number @xmath205 , we set @xmath426 as in and , for every @xmath197 there hold @xmath427 and @xmath428 we show that this implies that condition ( @xmath237 ) in proposition [ prop : criterion ] .
the following lemma has been implicitly used in the proof of theorem 5.1 in @xcite ; here we prefer to include a detailed proof for the sake of completeness . [ 090 ] let @xmath197 , and let us assume that for @xmath429 it results @xmath430 in @xmath431 , and @xmath432 in @xmath433 .
then @xmath434 notice that in principle @xmath435 , and hence the result is not immediate . only to fix our minds
, we consider @xmath201 , and for the sake of simplicity we omit the dependence on @xmath224 in the notation previously introduced . for @xmath222 there is nothing to prove .
let @xmath436 .
we fix @xmath437 , so that @xmath220 is the medium point between @xmath205 and @xmath267 .
in this way we have @xmath438 where we used the fact that @xmath439 for every @xmath440 .
now it is sufficient to observe that if @xmath441 , then @xmath442 that is , @xmath443 .
therefore , using the fact that @xmath334 in @xmath444 and is anti - symmetric , we conclude that @xmath445 in @xmath261 for every @xmath446 .
the result in lemma [ 090 ] means that for every @xmath197 and @xmath203 we have that either @xmath447 in @xmath431 , or else @xmath448 in @xmath431 .
since the @xmath431 are all the possible half - spaces of @xmath11 , by proposition [ prop : criterion ] we infer that @xmath5 is radially symmetric and radially non - increasing with respect to some point of @xmath11 .
this still does not prove that @xmath10 is radially symmetric , but it is sufficient to ensure that @xmath449 is the complement of a ball @xmath176 or a certain radius @xmath168 .
up to a translation , it is not restrictive to assume that the centre of @xmath176 is the origin . to complete the proof of theorem [ thm : main 1 prime ] , we have to show that @xmath450 .
before , we point out that since @xmath5 is radial and non - constant , if @xmath451 in @xmath431 , then necessarily @xmath452
. indeed , if this were not true , then there exists @xmath453 , and for any such @xmath135 we have @xmath454 a contradiction .
therefore , the strong maximum principle together with imply that @xmath455 in @xmath456 for every @xmath197 and @xmath457 , which in particular proves the radial strict monotonicity of @xmath5 outside @xmath458 .
now we show that @xmath450 .
we argue by contradiction , noting that there are two possibilities : either @xmath459 , or the intersection is empty , which means that @xmath10 is an annular region surrounding @xmath460 . ; on the right , @xmath459.,title="fig:",height=170 ] ; on the right , @xmath459.,title="fig:",height=188 ] if the latter alternative takes place , then there exists @xmath197 and @xmath457 such that @xmath461 is not convex .
this is in contradiction with lemma [ lem : geom remark ] .
it remains to show that also @xmath462 can not occur . to this aim
, we observe that in such a situation there exists a direction @xmath197 , a small interval @xmath463 $ ] with @xmath464 , and a small ball @xmath465 , such that @xmath466 while @xmath467 for every @xmath468 $ ] .
let @xmath469 .
then the reflection @xmath470 , and hence therein we have @xmath471 .
but on the other hand , since @xmath472 , we have also that by the strong maximum principle @xmath473 in @xmath456 , a contradiction ( recall that the only @xmath267 such that @xmath474 in @xmath475 is @xmath476 ) . in this way
, we have shown that @xmath477 is a ball , and the function @xmath5 is radial with respect to the centre of @xmath10 and radially decreasing in @xmath9 , which is the desired result . to show that @xmath56 and @xmath53 imply @xmath478 in @xmath11
, we use a comparison argument , as in the local case ( see ( * ? ? ?
* corollary 1 ) ) .
let us set @xmath479 then , recalling that @xmath480 , we have @xmath481 in @xmath9 .
we claim that this implies @xmath32 in @xmath9 . indeed , if this were not true that there would exists a point @xmath482 for @xmath5 with @xmath483 .
but then @xmath484 , in contradiction with the fact that @xmath485 here the strict inequality holds since @xmath480 in @xmath11 , and @xmath32 in a set of positive measure by the boundary condition @xmath127 as @xmath128 .
now it remains to show that @xmath54 for every @xmath46 .
this is a direct consequence of the hopf s lemma for non - negative supersolutions proved in @xcite , proposition 3.3 plus remark 3.5 therein .
indeed , we have already checked that for @xmath486 defined in , it results that @xmath487 this implies @xmath488 on @xmath35 , that is @xmath489 on @xmath35 , where @xmath1 denotes the unit normal vector to @xmath35 directed inwards @xmath9 . if there exists @xmath490 such that @xmath491 , then by the boundary conditions @xmath5 has an interior maximum point @xmath492 .
therefore @xmath493 since @xmath494 in @xmath11 , this forces @xmath495 , and in turn @xmath496 for every @xmath440 , in contradiction with the fact that @xmath127 as @xmath128 .
the strategy of the proof of theorem [ thm : main 2 ] is similar to that of theorem [ thm : main 1 ] .
we apply the moving planes method to show that for any direction @xmath197 there exists @xmath198 such that both the sets @xmath10 and @xmath21 , and the solution @xmath5 , are symmetric with respect to the hyperplane @xmath497 .
we fix at first @xmath201 and , for @xmath203 , we let @xmath498 be defined as in . we only modify the definitions of @xmath263 in the following way : @xmath499 furthermore , instead of @xmath500 and @xmath220 we define @xmath501 note that @xmath502 and @xmath503 are the critical positions for @xmath10 and @xmath21 , respectively , and @xmath220 can be considered as a critical position for @xmath20 . as in the previous section ,
we start with a simple geometric observation .
[ lem : geom remark bdd ] the following properties hold : * for @xmath504 , the set @xmath223 is convex in the @xmath224 direction ; * for @xmath505 , the set @xmath506 is convex in the @xmath224 direction .
the proof is analogue to that of lemma [ lem : geom remark ] , and thus is omitted . for @xmath251
, we have that @xmath507 exactly as in the previous section ( we refer to for the definition of @xmath421 ) . in the first part of the proof
, we aim at showing that the set @xmath508 defined by @xmath509 coincides with the interval @xmath510 , that @xmath256 for every @xmath257 , and that @xmath258 in @xmath259 . since we are not assuming that @xmath29 is monotone ( not even for small value of its argument ) , the argument in lemma [ lem : moving initial ] does not work .
nevertheless , we can take advantage of the boundedness of @xmath21 to apply the maximum principle in domain of small measure .
[ lem : moving initial bdd ] there exists @xmath182 such that @xmath511 in @xmath263 for every @xmath512 .
since @xmath29 is lipschitz continuous and @xmath5 is bounded , there exists @xmath189 independent of @xmath205 such that @xmath513 .
then , it is well defined , and independent on @xmath205 , the value @xmath514 as in proposition [ small ] . for @xmath205
a little smaller than @xmath515 , the measure of @xmath263 is smaller than @xmath126 , and the function @xmath326 satisfies @xmath516 as a consequence , by proposition [ small ] we deduce that @xmath268 in @xmath263 .
this means that the hyperplane @xmath208 moves and reaches a position @xmath517 .
we aim at showing that @xmath312 .
this is the object of the next two lemmas .
[ lem : condizione simmetria bdd ] let @xmath222 . if @xmath518 for @xmath314 , then both @xmath10 and @xmath21 are symmetric with respect to @xmath208 .
in particular , if @xmath315 , then @xmath268 in @xmath263 implies @xmath262 therein . by the strong maximum principle , if @xmath518 for @xmath314 , then @xmath317 in @xmath261 . as in the proof of lemma [ lem : condizione per simmetria ]
, this implies that @xmath10 is symmetric with respect to @xmath208 , that is , @xmath519 .
it remains to show that @xmath205 is also equal to @xmath503 , and @xmath21 is symmetric with respect to @xmath208 . to this aim
, we observe that if this is not the case , then @xmath520 therefore , if @xmath521 , we have @xmath522 and hence @xmath325 , a contradiction . then also @xmath21 is symmetric with respect to @xmath208 , which forces @xmath523 . [ 4.4 ] there holds @xmath524 . by contradiction , suppose that @xmath525 .
differently from the previous section , we use the again maximum principle in sets of small measure . let @xmath126 as in proposition [ small ] ( we have already observed in the proof of lemma [ lem : moving initial bdd ] that @xmath126 can be chosen independently of @xmath205 ) . by lemma [ lem : condizione simmetria bdd ] ,
@xmath526 in @xmath335 .
thus , there exists a compact set @xmath527 such that @xmath528 where we have used the boundedness of @xmath335 .
furthermore , observing that by continuity @xmath529 as @xmath530 , we can suppose that @xmath531 provided @xmath532 for some @xmath533 sufficiently small .
if necessary replacing @xmath534 with a smaller quantity , by continuity again it follows that @xmath535 whenever @xmath532 . for such values of @xmath205 , in the remaining part @xmath536 we have @xmath537 which means that we are in position to apply proposition [ small ] , deducing that @xmath268 in @xmath261 .
in particular , for @xmath538 we obtain @xmath539 in @xmath263 thanks to lemma [ lem : condizione simmetria bdd ] , in contradiction with the minimality of @xmath267 .
we proved in lemma [ 4.4 ] that @xmath540 .
hence , by lemma [ lem : condizione simmetria bdd ] , to obtain the symmetry of @xmath10 and of @xmath21 , it is sufficient to check that @xmath258 in @xmath541 . as in the proof of theorem [ thm : main 1 ] ,
we argue by contradiction assuming that @xmath542 in @xmath259 .
note that the critical position @xmath220 can be reached for four possible reasons : internal tangency for @xmath10 , internal tangency for @xmath21 , orthogonality condition for @xmath10 , orthogonality condition for @xmath21 . in all such cases
we can reach a contradiction exactly as in the conclusion of the proof of theorem [ thm : main 1 ] .
this proves that both @xmath21 and @xmath10 are symmetric with respect to @xmath260 , and by lemma [ lem : geom remark bdd ] , they are also convex in the @xmath224 direction .
if @xmath21 has several two connected components @xmath543 and @xmath544 , then by convexity the only possibility is that @xmath543 and @xmath544 are aligned along @xmath260 .
but in this case we can obtain a contradiction as in the conclusion of the proof of theorem [ thm : main 1 ] .
on @xmath10 we can argue exactly in the same way .
the proof differs only for some details from that of theorem [ thm : main 1 prime ] , and thus it is only sketched .
first of all , by monotonicity @xmath422 in @xmath11 for every @xmath205 , and hence by proposition 3.1 in @xcite ( weak maximum principle for anti - symmetric functions ) we have that @xmath268 in @xmath263 for every @xmath222 .
moreover , as in the conclusion of the proof of theorem [ thm : main 2 ] , @xmath258 in @xmath541 . repeating the same argument for any direction @xmath197 , we deduce by proposition [ prop : criterion ] that @xmath5 is radially symmetric and radially non - increasing in @xmath11 , which implies that @xmath545 and @xmath546 are concentric balls , and @xmath547 . by monotonicity , @xmath548 and @xmath549 . arguing as in the proof of theorem [ thm : main 1 prime ]
, we deduce that @xmath550 and @xmath551 .
we briefly describe how the proof of theorem [ thm : main 1 ] can be adapted to obtain theorem [ thm : radial 1 ] . without loss of generality
, we suppose that @xmath2 , the centre of the cavity , is @xmath552 . using the same notation introduced in section [ sec : over exterior ] , see , we observe that for any direction @xmath197 the critical position @xmath553 is reached for @xmath554 .
let us fix @xmath201 , and let us introduce @xmath555 we aim at proving that @xmath556 , and that @xmath557 in @xmath263 for every @xmath457 .
once that this is proved , we can repeat the argument with @xmath558 .
since the critical position for @xmath224 and @xmath559 is the same , we have @xmath560 from which we infer that @xmath5 is symmetric with respect to @xmath561 , and strictly decreasing in the @xmath265 variable outside @xmath562 . symmetry and monotonicity in all the other directions can be obtained in the same way . as in lemma
[ lem : moving initial ] , we can show that @xmath563 .
once that this is done , as in lemmas [ lem : condizione per simmetria ] and [ lem : moving continuation ] , we can show that @xmath556 .
this completes the proof .
notice that assumption [ cond neumann ] is used in lemma [ lem : moving continuation ] , case 3 ) .
when @xmath21 is bounded , by using the method in section [ sec : over annular ] , we see that the conclusion of theorem [ thm : radial 1 ] remains true if we assume that only one between @xmath21 and @xmath10 is a ball , and prescribe constant @xmath4-neumann boundary condition on the other component . for the same result in the local case
, we refer to ( * ? ? ?
* theorem 5 ) . without loss of generality
, we suppose that @xmath398 . using the same notation
introduce in section [ sec : over exterior ] , we fix @xmath201 and observe that @xmath564 and @xmath565 is the critical position for the hyperplane @xmath208 .
we slightly modify the definition of @xmath508 in the following way : @xmath566 where we recall that @xmath567 satisfies the equation @xmath507 and @xmath421 has been defined in .
we aim at showing that @xmath556 .
this is the object of the next three lemmas .
there exists @xmath270 sufficiently large such that @xmath262 in @xmath263 for every @xmath568 .
exactly as in lemma [ lem : moving initial ] , it is possible to show that @xmath268 in @xmath263 . by the strong maximum principle , proposition [ strong ] , either @xmath262 in @xmath263 or @xmath317 in @xmath261 . if @xmath317 in @xmath261 , then @xmath569 . on the other hand , since @xmath127 as @xmath128 we have that for @xmath205 very large @xmath570 in the whole half - space @xmath261 , a contradiction .
thus , the quantity @xmath571 is a real number .
[ lem : monot ] the function @xmath5 is monotone strictly decreasing in @xmath265 in the half - space @xmath572 .
let @xmath573 with @xmath574 .
we aim at showing that @xmath575 . for @xmath576
, we claim that @xmath577 .
once that this is shown , the desired conclusion simply follows by the fact that @xmath578 as @xmath579 . since @xmath580 , if @xmath581 , then necessarily @xmath582 .
this means that @xmath583 , in contradiction with the fact that @xmath584
. we are ready to complete the proof of theorem [ thm : radial point ] by showing that @xmath476 .
it holds @xmath556 . by contradiction ,
let @xmath585 .
at first , by continuity @xmath334 in @xmath335 .
thus , by the strong maximum principle we have that either @xmath526 in @xmath335 , or @xmath586 in @xmath444 . to rule out the latter alternative , we observe that having assumed @xmath585
, we obtain @xmath587 .
thanks to the previous lemma , we infer that @xmath588 whenever @xmath589 , in contradiction with the maximality of @xmath590 .
thus , it remains to reach a contradiction when @xmath591 in @xmath335 . by the definition of @xmath592 ,
there exist sequences @xmath593 and @xmath273 such that @xmath338 and @xmath274 . since @xmath340 in @xmath594 and tends to @xmath552 as @xmath128 , it is not restrictive to assume that @xmath279 is an interior minimum points for @xmath280 in @xmath281 .
if @xmath284 , we obtain a contradiction as in lemma [ lem : moving initial ] . hence , up to a subsequence @xmath595 . notice that by uniform convergence @xmath353 , which forces @xmath596 .
if @xmath597 , this means that @xmath587 , and by lemma [ lem : monot ] we obtain a contradiction with the fact that @xmath480 in @xmath11 .
therefore @xmath598 .
this means that all the points @xmath279 , and also @xmath134 , are interior points for the anti - symmetric functions @xmath280 and @xmath599 in the sets @xmath600 respectively . as a consequence
, we can argue as in case 1 ) of the proof of lemma [ lem : moving continuation ] , deducing that for some @xmath601 the sequence @xmath358 is uniformly bounded in @xmath602 .
this entails @xmath603 convergence in @xmath604 , and by minimality @xmath605 , in contradiction with proposition [ prop : new hopf ] .
this section is devoted to the proof of the existence of a solution to @xmath606 satisfying all the assumptions of theorem [ thm : main 1 prime ] . to this aim , we recall that the critical exponent for the embedding @xmath607 is defined as @xmath608 .
let @xmath609 be the space of functions in @xmath610 that are @xmath5 radial and radially decreasing with respect to the origin .
we point out that , if @xmath611 , the decay estimate latexmath:[\[\label{ddcc } holds ( see e.g. lemma 2.4 in @xcite for a simple proof ) , and this ensures that @xmath127 as @xmath128 .
moreover , it is known in @xmath613 and fixed @xmath614 , we find an @xmath615-net for @xmath616 .
that is , first we use to say that for any @xmath617 , we have that @xmath618 if @xmath619 is chosen suitably large ( in dependence of @xmath615 )
. then we use local compact embeddings ( see e.g. corollary 7.2 in @xcite ) for the compactness in @xmath620 : accordingly , we find @xmath621 such that for any @xmath617 there exists @xmath622 such that @xmath623 ( of course @xmath246 may also depend on @xmath615 ) . we extend @xmath624 as zero outside @xmath625 , and we found an @xmath615-net since in this way @xmath626 .
this shows the compactness that we need . for a more general and comprehensive treatment of this topic
, we refer to @xcite and theorem 7.1 in @xcite . ]
that @xmath613 compactly embeds into @xmath627 for every @xmath628 .
let @xmath629 for some @xmath630 continuous , odd , and such that @xmath631 then there exists a radially symmetric and radially decreasing solution @xmath18 of problem , satisfying the additional condition @xmath56 in @xmath11 .
we denote @xmath632 , and set @xmath633 let @xmath634 be defined by @xmath635 where @xmath10 denotes a primitive of @xmath636 .
it is not difficult to check that if @xmath637 is a minimizer for @xmath638 , then @xmath5 solves , and hence in the following we aim at proving the existence of such a minimizer .
let @xmath639 . since @xmath640 by assumption , we have that @xmath641 which implies that @xmath638 is coercive on @xmath642 .
let @xmath643 be a minimizing sequence for @xmath644 .
since @xmath636 is odd we can suppose that @xmath645 for every @xmath646 ( recall that , if @xmath647 , also @xmath648 ) , and thanks to the fractional polya - szego inequality ( see @xcite ) it is not restrictive to assume that each @xmath649 is radially symmetric and radially non - increasing with respect to @xmath552 . thus @xmath643 is a bounded sequence in @xmath613 , so that by compactness we can extract a subsequence of @xmath643 ( still denoted @xmath643 ) , and find a function @xmath611 , such that @xmath650 weakly in @xmath642 and a.e . in @xmath11 .
notice that @xmath637 .
now by weak lower semi - continuity we infer @xmath651 namely @xmath5 is a minimizer for @xmath644 , and hence a solution of . by convergence , it is radially symmetric and radially non - increasing with respect to @xmath552 , and is nonnegative .
moreover , @xmath652 which , as in the proof of corollary [ corol : subharmonicity ] , implies @xmath480 in @xmath11 . finally , by theorem [ thm : regularity ] it results @xmath18 .
in this section we discuss the regularity of weak solutions of @xmath653 where @xmath21 is an arbitrary open set of @xmath11 , neither necessarily bounded , nor necessarily connected . when @xmath21 is bounded domain , the regularity of weak solutions has been studied in @xcite , and in what follows we show how to adapt the arguments therein to deal with the general case considered here . notice that if @xmath5 is a bounded solution of the first two equations in , then the difference @xmath654 solves a problem of type .
[ thm : regularity ] let @xmath21 be an arbitrary open set of class @xmath655 , and let @xmath656 . if @xmath657 is a bounded weak solution of , then @xmath658 for some @xmath123 .
moreover , if @xmath126 denotes the distance from the boundary of @xmath21 , then @xmath659 for some @xmath125 . by proposition 5 in @xcite , @xmath661 ,
and therefore it remains to rule out the possibility that @xmath113 has a discontinuity on @xmath662 . to this aim , we argue by contradiction assuming that there exists a point of discontinuity @xmath663 for @xmath113 .
let @xmath664 be such that @xmath665 for some @xmath348 .
the existence of @xmath666 has been proved in ( * ? ? ?
* lemma 2.6 ) . since @xmath21 satisfies an exterior sphere condition ,
there exists a ball @xmath667 which is tangent to @xmath21 in @xmath2 . by scaling and translating @xmath666
, we find an upper barrier for @xmath113 in @xmath668 , vanishing in @xmath669 ( here we use the fact that both @xmath636 and @xmath113 are in @xmath129 ) .
arguing in the same way on @xmath670 , we deduce that @xmath671 in @xmath672 , yielding in particular @xmath673 as @xmath674 .
this implies that @xmath113 is continuous in @xmath2 , a contradiction .
w. reichel . radial symmetry by moving planes for semilinear elliptic bvps on annuli and other non - convex domains . in _ elliptic and parabolic problems ( pont - - mousson , 1994 ) _ , volume 325 of _ pitman res .
notes math .
_ , pages 164182 .
longman sci .
tech . , harlow , 1995 . | we consider a fractional elliptic equation in an unbounded set with both dirichlet and fractional normal derivative datum prescribed .
we prove that the domain and the solution are necessarily radially symmetric .
the extension of the result in bounded non - convex regions is also studied , as well as the radial symmetry of the solution when the set is a priori supposed to be rotationally symmetric . |
fluctuation - dissipation ( fd ) relations are commonly used in equilibrium systems , derived from the notion that small perturbations and brownian fluctuations produce the same response in a given system @xcite .
mobility , the constant of proportionality between a particles drift speed and a constant external force , is extracted from velocity statistics of particles in a given system .
diffusivity , calculated from fluctuation displacements of particles in a system over time , represent the brownian motion .
the temperature of a system in thermodynamic equilibrium can be extracted from a fd relation , defined as the ratio of diffusivity and mobility , as is commonly used in the einstein relation . in equilibrium
this temperature is taken to be the bath temperature .
as studies in granular matter have grown more important within the environmental and industrial fields , the need to establish a scientific framework that accurately predicts granular system responses on the continuum level , beyond merely geometrical features , has also escalated .
granular matter , when condensed to sufficiently high volume fractions , undergoes a ` jamming ' transition to the jammed state .
the jammed state is defined as the condition when a many - body system is blocked in a configuration far from equilibrium , such that relaxation can not occur within a measurable time - scale . for granular matter
, the jammed state indicates a transition between a solid - like behavior , and a liquid - like behavior . at high volume fractions ,
the physical size of the constituent grains inhibits particle motion , thereby rendering the system out of equilibrium , and the granular system behaves more like a solid .
thermal motion does not govern the exploration of states in jammed granular matter .
theories proposed by edwards and collaborators @xcite propose a statistical mechanics for granular matter based on jamming the constituent grains at a fixed total volume such that all microscopic jammed states are equally probable and exhibit ergodicity .
the exploration of reversible jammed states is achieved via an external perturbation such as tapping or shear , not brownian motion as in thermal systems .
there is an important difference between reversible jammed states , and states that are only mechanically stable within certain limits of perturbation magnitude .
for example , pouring grains into a container results in a pile at a particular angle of repose .
this mechanical equilibrium configuration is jammed regularly but not reversibly jammed because in response to an external perturbation , the constituent particles will irreversibly rearrange , approaching a truly jammed configuration .
studying an ensemble of truly jammed , reversible states is thereby suitable for a plausible application of statistical mechanics under the present theory .
these ensembles , inherently non - equilibrium systems , will not be governed by the commonly used parameters of equilibrium statistical mechanics , such as a bath temperature . in recent studies
theoretical mean - field models of glasses @xcite have introduced the concept of an `` effective temperature '' as extracted from the fd relations in non - equilibrium systems . while not equivalent to the equilibrium bath temperature , the effective temperature reflects a change in the relaxation time - scale of the system .
these non - equilibrium systems extend beyond glasses , and into granular media , where physical size of the constituent grains inhibits motion , allowing for jammed systems far from equilibrium .
this concept has been furthered by computer simulations of granular media and other non - equilibrium soft - matter systems @xcite .
it remains a question whether or not granular media can be characterized by an effective temperature , thus revealing a dynamic counterpart to the static `` compactivity '' as proposed by edwards @xcite .
athermal systems require the input of energy by an external source to explore the effective temperature @xcite .
one proposed method of calculating the effective temperature of a jammed granular system is a slow shearing procedure @xcite , leading to the design of the experiment we present in @xcite .
slow shearing , at the quasi - static limit , allows for extrapolation towards an effective temperature of jammed , static , systems .
the jammed system of interest is one of identical , spherical grains , confined between the two cylindrical walls of a 3d couette cell .
the grains are further confined by an external pressure in the vertical direction .
the inner cylinder of the couette cell is slowly rotated to induce shearing in the system .
tracer particles are inserted in the system , and their trajectories recorded via multiple cameras surrounding the system .
the couette cell is partially filled with a refractive index matched fluid to allow for system transparency .
the cylindrical walls are roughened by gluing grains , identical to those of the bulk , such that crystallization is avoided .
the cameras record tracer particle trajectories throughout the bulk , recording data in cylindrical coordinates , @xmath0 .
distributions of tracer particle displacements are measured in each direction . as gravity is the external force applicable to the mobility calculation in the current formalism , only displacements in the z - direction
are applied to the fd relation .
additionally , average velocity profiles are calculated for each direction , first with constant shear rate , @xmath9 , and further studied to determine shear rate dependence .
displacement measurements are further limited to the `` constant mobility and diffusivity '' ( cmd ) region , defined as the narrow range of radial coordinates such that the average vertical velocity is roughly independent on radial distance .
the pdf of displacement distributions for each direction is presented .
further , fluctuations in displacement are determined for each cylindrical direction , and studied as a function of time .
radial displacement fluctuation is found to be sub - diffusive , while angular displacement fluctuation is found to be super - diffusive .
vertical displacement fluctuation is purely diffusive within the time scales of the experiment , allowing for the validity of the fd relations used herein .
all displacement fluctuations are reduced to functions of angular displacement , and the results are presented .
such relationships permit scaling of the pdf curves with varying angular displacements due to changing shear rates . utilizing the fd relations presented above ,
the diffusivity and mobility in the z - direction are extracted from the tracer particle trajectories and the effective temperature is realized .
this effective temperature is found to be independent of tracer particle properties , as shown in @xcite , and further independent of the slow - shear rate .
moreover , the effective temperature may then be considered a physical variable that characterizes the jammed granular system , with respect to the generalization of the equilibrium statistical mechanics of boltzmann , as applied to non - equilibrium systems .
we further study the limits within which this effective temperature may be a valid physical variable , as we determine mobility and diffusivity as a function of shear rate .
while diffusivity appears independent of shear rate , even somewhat above the ` slow ' regime , mobility shows a clear decrease in magnitude as we explore shear rates above the slow regime , resulting in an increase in the effective temperature as a function of shear rate . in this paper , we will further report the experimental detail of particle dynamics in @xcite .
the outline is as follows :
the experiment is performed using a three - dimensional ( 3d ) couette cell , as shown in fig .
[ setup1 ] , [ setup2 ] and [ setup3 ] .
the grains are confined between two cylinders of height @xmath10 cm .
the inner cylinder is rotated via a motor , while the outer cylinder remains fixed .
the walls of the cylinders , in contact with the grains , are roughened by means of a glued layer of identical granular material , thereby minimizing wall slip .
the walls of the inner and outer cylinders are roughened by acrylic beads with diameter @xmath11 and @xmath12 mm , respectively . testing the experiment with a rough inner wall and a smooth outer wall resulted in packing crystallization .
the grains are compacted by an external pressure of a specific value ( typically 386 pa ) , introduced by a moving piston at the top of the granular material , acting in the negative z - direction .
observation techniques are used to monitor the granular packing evolution as it explores the available jammed configurations .
the couette cell is sheared at the quasi - static limit , with slow frequencies @xmath13 mhz defining the external shear - rate @xmath14 @xmath15 , where @xmath16 cm and @xmath17 cm are the radius of the inner and outer cylinders , respectively , and @xmath18 is the angular velocity of the inner cylinder ( notice that @xmath19 and @xmath20 are measured after the walls are roughened by a glued layer of beads ) . the experiment is designed to measure the diffusivity and mobility of tracer particles @xcite , as opposed to tracking the motion of all constituent grains .
the distance between the inner and outer cylinder is less than 10 grain diameters to prevent bulk shear band formation @xcite that may interfere with the experimental measurements by altering the diffusivity . a refractive index matching suspending solution is employed in order to create a transparent sample .
the suspending solution is also density matched to the grains in order to eliminate pressure gradients derived from gravity in the vertical direction , circumventing problems seen in previous experiments of compactivity @xcite and other effects such as convection and size segregation such as the brazil nut effect inside the cell @xcite .
the solution used in this experiment is approximately 74% weight fraction of cyclohexyl bromide and 26% decalin @xcite .
these steps avoids problems encountered in previous tests of compactivity .
the granular system is a bidisperse , 1:1 by mass , mixture of spherical , transparent poly - methyl methacrylate ( acrylic ) particles , with density @xmath21 and index of refraction @xmath22 .
the bidisperse mixture is used in an effort to inhibit crystallization of the system .
the respective particle diameters are either @xmath23 mm and @xmath11 mm ( packing 1 ) or @xmath11 mm and @xmath24 mm ( packing 2 ) . the approximate same size ratio of each bidisperse packing leads to approximately the same value of volume fraction for both , being @xmath25 before shearing and @xmath26 during shearing . a negative consequence of utilizing a suspending solution includes possible modification of the friction coefficient between the grains . while this can not be completely avoided , it is important to note that the liquid only partially fills the cell ( see fig .
[ setup1 ] ) , such that the pressure of the piston is transmitted to the granular material exclusively , not to the fluid .
additionally , hydrodynamic effects from partial cell filling are avoided by the extremely slow rotational speeds applied to our system .
the system remains very closely packed , such that particles are not free to float in the fluid .
therefore , the random motion of the particles is controlled by the ` jamming ' forces exerted by the contacts between neighboring grains , not fluid mechanics .
cylindrical coordinates , @xmath0 , of tracer particles are obtained by analyzing images acquired by four digital cameras surrounding the couette cell . for systems in thermal equilibrium ,
a fluctuation - dissipation ( fd ) relation may be utilized in an effort to calculate the bath temperature of the system .
this method may be extended to non - equilibrium systems , such as jammed granular systems presented in this study .
the fd relation is defined as follows : @xmath27 ^ 2\rangle \sim 2 d \delta t,\ ] ] @xmath28 @xmath27 ^ 2\rangle = ~ 2 t_{\mbox{\scriptsize eff } } ~ \frac{\langle x(t+\delta t ) - x(t)\rangle } { f}. \label{fdt}\ ] ] the tracer particles must experience a constant force , @xmath29 , in order to calculate the mobility as defined above .
the most convenient constant , external , force , is gravity in the z - direction .
if the effective temperature is to be regarded as an intensive variable of the non - equilibrium system , it requires independence from the tracer particles properties , and we present data in favor of this result .
however , we acknowledge that temperature measures from multiple observables would be necessary to analyze the underlying thermodynamic meaning of @xmath8 .
tracer particles added to the bulk must have properties unique from the grains comprising the bulk .
however , tracer particles too small , or too large , with respect to the acrylic grains described previously , would result in erroneous measurements .
dynamics of tracer particles that were too small would be dominated by `` percolation effects '' @xcite , resulting in larger than expected tracer particle displacements .
those too large would require shear rates above the quasi - static limit we propose to study , or possibly have no dynamics at all due to size limitations . with these notions in mind ,
two different types of tracers , nylon ( @xmath30 ) and delrin ( @xmath31 ) , are employed , which result in different external forces , @xmath32 , where @xmath33 and @xmath34 are the densities of the acrylic particles and the tracers , respectively , @xmath35 is the volume of the tracer particle and @xmath36 the gravitational acceleration
. variations in tracer particle diameter and density allow us to study dynamical changes due to a change in constant external force , while we remain within a range appropriate to achieve results expected to be governed by the effective temperature .
four digital cameras symmetrically surround the shear cell to track the tracer particles with frame rate @xmath37 5 frame / s , as shown in fig .
[ setup3 ] .
the outer cylinder is made of the same material as the grains ( acrylic , @xmath38 ) .
the refractive index is matched by the fluid such that the system can be regarded as an optical whole , i.e. , the light scattered from tracers refracts only once at the outer surface of the outer cylinder , as shown in fig .
[ setup2 ] .
the determination of the 3d tracer position is achieved by a simple calculation considering both system geometry and 2d projections captured by two adjacent cameras .
camera calibration and determination of relative position is important as a minimal asymmetry will result in a large calculation error of the tracer particles coordinates , @xmath0 .
as opposed to directly measuring relative positions of the cameras by physical devices , we utilize computer programming . in order to simplify the calculation
, we assume the camera to be a pinhole , meaning all light coming into the camera coincides at a single focal point . before each experiment , we record the images of a piece of grid paper attached to the surface of the outer cylinder , acting as the 2d projections of the outer cylinder for each camera .
next , we adjust the positions of four cameras until each camera can give the approximately same 2d projections of the outer cylinder .
then we use the computer program to generate a virtual cylinder , along with four virtual cameras , according to the geometry of the shear cell , in other words , we build a virtual space of the entire experimental setup and the respective geometrical relations between its elements . from the previous calibration procedure
, we have the relative positions of the four cameras to the shear cylinder with sufficient accuracy . in order to further calibrate and know the exact position of cameras , we adjust the relative position of cameras in our virtual space until the virtual 2d projection of the cylinder to the cameras coincides exactly with the actual projection , being the grid paper attached to the outer cylinder .
when this procedure is accomplished , the virtual space exactly coincides with the real experimental setup space .
therefore the virtual relative position of cameras are also the real positions .
furthermore , in our virtual space , any point with 3d coordinates , @xmath39 , we can calculate its 2d coordinates in four virtual 2d projections , @xmath40 , by considering the geometry relation to cameras .
oppositely , for any tracer particle , if we know its 2d coordinates in four 2d projections , @xmath40 , we can exactly locate its 3d positions , @xmath39 , since the virtual space is equal to the actual one .
the resulting vertical trajectories of the tracers @xmath41 are depicted in fig .
[ trajectory1 ] showing that the nylon tracers not only diffuse , but also move with a constant average velocity to the top of the cell .
[ trajectory2 ] shows a typical trajectory of tracer particle in 3d plotting .
we first study the velocity profiles for a fixed shear rate , @xmath9 , followed by a study on the shear rate dependence in the next section .
the average velocity profiles in the angular direction , @xmath42 , in the vertical direction , @xmath43 , and in the radial direction , @xmath44 , are obtained by averaging the velocities of all tracer particles over all times at each radius @xmath45 , as shown in fig .
[ v_z_t_r ] .
as observed in previous work @xcite , we find that @xmath42 can be expressed in the exponential form demonstrated in fig .
[ v_z_t_r]a : @xmath46 where @xmath47 and @xmath48 are constants independent of shear rate , tracer size and tracer type , depending only on the type of packings and geometry of the shear cell .
we find that @xmath49 and @xmath50 , and @xmath51 and @xmath52 for packing 1 and packing 2 , respectively .
when @xmath53 , @xmath54 being the angular velocity of the first layer of grains closest to the sheared inner cylinder with angular velocity @xmath18 .
therefore @xmath47 ( @xmath55 ) can be taken as the efficiency of shearing , describing the amount of slip between the inner rotating cylinder and the first layer of grains it contacts .
packing 1 has a higher value of @xmath47 than packing 2 , as the smaller grains follow the rotating inner cylinder more easily . on the other hand , when @xmath56 , we find @xmath57 , the velocity of the last layer of grains closest to the static outer cylinder .
this velocity is non - zero , so that the shear band is located right at the outer cylinder , avoiding the formation of shear bands in the bulk . in order to avoid the tracer particles sticking to the outer cylinder surface and forcing its velocity to zero
, we glue smaller size particles to roughen the outer cylinder .
this roughens the surface of the outer cylinder and avoids crystallization .
further , this allows slipping of the bulk particles at the outer cylinder , forcing the shear band to be located exactly at the outer cylinder , not in the bulk .
the glued particles are @xmath12 mm , smaller than the sheared granular material .
the mean angular and vertical velocity , @xmath58 and @xmath59 , of tracer particles do not decay to zero even if the tracers come close to outer cylinder surface .
( see fig .
[ vtheta_r_shear ] and fig .
[ vz_r_shear ] at @xmath56 ) .
the exponential decay of @xmath42 results in a local shear rate , @xmath60 , dependent on radial distance . in the region near the outer cylinder
, @xmath42 decays slowly with increasing @xmath45 which leads to weak dependence of @xmath60 on the radial distance @xmath45 . if we do taylor expansion at @xmath56 , the average angular velocity of the tracers , @xmath42 , can be approximated to a linear function of @xmath45 , i.e. , @xmath61 with constant local shear rate @xmath62 s@xmath63cm@xmath63 .
the diffusivity and mobility of the tracer particles strongly depend on the local rearrangement of the grains . a constant shear rate results in homogenous local rearrangement of the packings ensuring that the diffusivity and mobility of tracers , dependent on local shear rate , remain approximately independent of @xmath45 .
as shown in fig .
[ v_z_t_r]b , we find a plateau in the vertical velocity profile which can be further seen in fig .
[ vz_r_shear ] .
similar behavior is observed in the vertical diffusivity profile , @xmath65 as shown in fig .
[ d_r_shear ] , which we will discuss in detail in section [ teff ] .
we denote this the `` constant mobility and diffusivity region '' , i.e. , cmd region , @xmath66 cm @xmath67 cm .
contrary to prior work @xcite on sheared granular matter in the couette cell , our experiment focuses only a narrow gap , @xmath68 mm , of couette cell .
the cmd region allows us to well define the diffusivity and mobility of the tracer particles , such that we can calculate the average vertical velocity , @xmath59 , and the average vertical diffusivity , @xmath69 , by averaging the velocities of all tracers over all times in the cmd region , significantly improving the statistics . in this study , the statistical average and the measurements of tracer fluctuations
will be confined only to the cmd region .
we find @xmath44 to be flat for different types of tracer particles and for different packings except when the tracers are close to the inner and outer cylinder , i.e. , @xmath53 and @xmath56 , as shown in fig .
[ v_z_t_r]c .
@xmath70 is negative and @xmath71 is positive , indicating the inner and outer cylinder walls can slightly attract the tracers .
it should be noted that the statistics presented in this study does not incorporate data from the regions close to the inner and outer cylinder to avoid these boundary effects .
next , we study the dependence of the particle velocity on the external shear rate . according to eq .
( [ vtheta_exp ] ) , the velocity profile in the angular direction , @xmath42 , is proportional to the external shear rate @xmath72 .
we can collapse @xmath42 by scaling the shear rate .
the results are shown in the inset of fig .
[ vtheta_r_shear ] for packing 2 .
the collapsing of @xmath73 shows a periodic shape superimposed to exponential decay with a very small amplitude , also found in the velocity profile of @xmath44 ( see fig .
[ vr_r_shear ] ) .
the periodic length is roughly equal to the grain particle size and reflects the different layers of grains in the radial direction .
this periodicity is weaker in packing 1 than packing 2 , since the particle size of packing 1 is smaller than that of packing 2 .
the collapsing method can be further applied to @xmath43 , as seen in fig .
[ vz_r_shear ] . after scaling by the shear rate
, @xmath74 also shows a flat plateau indicating the cmd region .
[ pz]a shows the results of the probability distribution of the displacements @xmath75 in the vertical direction for a given time interval @xmath76 .
the data corresponds to the @xmath23 mm delrin tracers in packing 1 .
usually , 20 tracers are used for calculations .
tracer trajectories are split into sub - trajectories confined in two regions , ( i ) : @xmath77 cm @xmath78 cm , close to the inner rotating cylinder , and ( ii ) : @xmath66 cm @xmath79 cm , i.e. , cmd region , close to the outer cylinder .
we compare the calculated @xmath80 by using the sub - trajectories from the regions of ( i ) and ( ii ) respectively , which are plotted as black triangle and black circle in the fig .
the data in the inner region ( i ) clearly display an asymmetric tail for @xmath81 .
this extra spreading is similar to the phenomena of the taylor dispersion @xcite .
taylor dispersion appears when diffusion couples with the gradient of flow giving rise to a larger dispersion along the flowing direction ( see for instance @xcite for a study of taylor dispersion in granular materials ) . in the present experiment , the shear rate of granular flow in the angular direction exhibits exponential decay , as shown in fig .
. the larger shear rate in the inner region ( i ) results in larger packing rearrangement , which gives rise to a larger dispersion in the vertical direction . in this case
it is not possible to extract the bare diffusion constant . on the contrary , for the region ( ii ) , i.e. , cmd region , as we mentioned , the gradient of the flow , i.e.
, the shear rate is approximately constant , giving rise to a gaussian diffusion , as shown in fig .
[ pz]a . by measuring the width and the mean value of this gaussian distribution of @xmath82 \\
& \sim\exp[{-\frac{(\delta z - m_z f \delta
t)^2}{4d_z \delta t } } ] , \end{split}\ ] ] we can define the diffusivity and mobility , @xmath69 and @xmath83 , which lead to the effective temperature of the granular packing discussed in the following section . in the fig .
[ pz]b , we define a new scaled variable @xmath84 and plot @xmath85 for different @xmath76 , all the curves are found to collapse into a single curve @xmath86 in this experiment , we will focus our measurements in the region away from the inner boundary ( region ( ii ) , i.e. , cmd region ) , where the mobility is a constant ( as shown a plateau in the inset of fig . [ vz_r_shear ] ) and taylor dispersion effects are absent .
we find exponential fluctuations for the probability distributions of the tracer particles in the radial direction as shown in fig .
[ pr ] , @xmath87 where @xmath88 is a function of @xmath76 . the symmetric shape for @xmath89 indicates the absence of a shear induced segregation , as observed with multiple sizes of grains , as there is no net flow of the tracer particles towards either cylindrical wall within the time - scales of the experiment .
we also observe no average motion of the tracer particles towards the center of the couette cell except within a small range of radial distance , around @xmath90 cm , close to both walls where particles experience a slight attraction to boundaries .
these features are shown in fig .
[ v_z_t_r]c and fig .
[ vr_r_shear ] .
the analysis of the radial displacement fluctuation reveals a power law , sub - diffusive , process : @xmath91 as shown in figure 8 , where @xmath92 for both the delrin and nylon tracers .
the data taken for the angular displacement is in the direction of the flow , and affected by taylor dispersion as shown in the non - gaussian tail of the displacement distribution @xmath93 in fig .
[ ptheta ] .
this leads to a power law , super - diffusive process , illustrated by @xmath94 as seen in the analysis of the fluctuations of @xmath95 shown in fig .
[ ptheta ] , where @xmath96 and @xmath97 for packing 1 and packing 2 respectively .
we further study how shear rate affects the displacement probability distribution .
we find that for small shear rate , the probability distributions of displacements in the three cylindrical coordinates are independent of the shear rate , depending only on the sheared displacement , i.e. , the external rotating displacement , defined as @xmath98 where @xmath99 is the rotating displacement of the inner cylinder .
this result is expected .
since we shear the couette cell very slowly , the diffusion of the tracers depends only the number of granular packing configurations sampled by the couette cell , which depends only on the sheared displacement . as emphasized in the previous text , the statistical average and the measurements of the tracer fluctuations
is confined to the cmd region , such as the @xmath69 shown in fig .
[ d_m_t]a .
we calculate the @xmath69 by measuring the width and the mean value of the gaussian distribution of @xmath80 , and obtain the @xmath100 by averaging the displacement fluctuations of all tracers over all time in the cmd region .
next , we apply a different method to reveal how @xmath65 depends on the radial distance @xmath45 , as shown in the fig .
[ d_r_shear ] .
we first obtain @xmath101 for a certain radial distance @xmath45 , then we calculate @xmath65 by measuring the width and the mean value of the gaussian distribution of @xmath101 . in fig .
[ d_r_shear ] , we see that the tracer particles have higher diffusivity close to the inner cylinder than the outer . since @xmath102 , we can collapse all the @xmath65 for various shear rates , as shown in the inset of fig .
[ d_r_shear ] .
the collapse of @xmath103 shows a plateau close to the outer cylinder , consistent with our previous discussion of the cmd region .
[ fluctuations ] @xmath104 ^ 2\rangle & \sim\delta\theta_e \\ \langle[r(t+\delta t)-r(t)]^2\rangle & \sim{\delta\theta_e}^\alpha \\
\langle[\theta(t+\delta t)-\theta(t)]^2\rangle & \sim{\delta\theta_e}^\beta\end{aligned}\ ] ] eq . (
[ fluctuations ] ) implies that one can collapse the probability distribution of the displacements , @xmath80 , @xmath89 and @xmath105 for different shear rates and time intervals by scaling @xmath75 , @xmath106 and @xmath107 respectively to @xmath108 , @xmath109 and @xmath110 .
the results are presented in fig . [ pz_collapsing ] , [ pr_collapsing ] and [ ptheta_collapsing ] .
tracer&@xmath111 `` & @xmath33 & @xmath69 & @xmath83 & @xmath112 + acrylic&@xmath23&@xmath113&@xmath114 & & + delrin&@xmath23&@xmath115&@xmath116&@xmath117&@xmath118 + delrin&@xmath11&@xmath115&@xmath119&@xmath120&@xmath121 + nylon&@xmath11&@xmath122&@xmath123&@xmath124&@xmath125 + ceramic&@xmath11&@xmath126&&@xmath127 & + brass&@xmath11&@xmath128&&@xmath129 & + + tracer&@xmath111 `` & @xmath33 & @xmath69 & @xmath83 & @xmath112 + nylon&@xmath11&@xmath122&@xmath130&@xmath131&@xmath132 + nylon&@xmath24&@xmath122&@xmath133&@xmath134&@xmath135 + we present results for the diffusivity in the @xmath136 direction , the only direction where the effective temperature can be calculated due to the vertically acting external force . the gaussian distribution in @xmath80 allows us to apply the fd relation to the particle displacements , as the diffusivity is proportional to the variance of a gaussian distribution in displacements .
exponential fluctuations do not possess this same property , but it is important to note that the radial direction has no constant applied external force . it remains a possibility that a well - defined effective temperature for displacements in the radial direction could exist . to test whether the effective temperature is isotropic , as done in @xcite , may be of great interest in future studies .
a common method of performing a time average to measure transport coefficients is employed ( see chapter 5.3 in @xcite ) by dividing the trajectory of a single tracer particle into a series of trajectories , having evenly spaced start times , separated by time interval @xmath76 .
the diffusion constant is obtained by averaging over the aggregate of tracers and over the initial time intervals , allowing for the use of merely 20 tracer particles in this particular system .
correlations between measurements are ensured to have decayed almost to zero , rendering time - translational invariance valid in this system , without any measurable``aging '' , since under shearing , system reaches the `` stationary state '' @xcite .
furthermore , doubling the number of tracer particles leaves @xmath69 unchanged , indicating independence of the diffusion constant from the number of tracers that explore the jammed configurations of this non - equilibrium system .
analysis of the vertical particle displacements in the cmd region reveals a gaussian distribution , broadening over time , as seen in fig .
[ pz ] and fig .
[ pz_collapsing ] . for sufficiently long times period , the mean square fluctuations grow linearly ( see fig .
[ d_m_t]a ) : @xmath137 ^ 2\rangle \sim 2d_z\delta t,\ ] ] where @xmath69 is the self - diffusion constant in the vertical direction . for the both nylon and delrin @xmath11 mm tracers in packing 1
we obtain @xmath138 m@xmath139/s .
figure [ d_m_t]b shows mean value tracer particle positions , extracted from the peak of the gaussian distribution , as a function of time .
the mobility in the vertical direction , @xmath83 , is defined as @xmath140 the applied force on the tracers , @xmath141 , is the gravitational force due to density mismatch where @xmath33 and @xmath34 are the densities of the acrylic particles and the tracers , respectively , @xmath35 is the volume of the tracer particle and @xmath36 the gravitational acceleration .
the value of the mobility for the both nylon and delrin @xmath11 mm tracers in packing 1 is @xmath142 s / kg .
[ d_m_t]a further reveals a downward curvature of the mean - square fluctuations , for sufficiently long times period . additionally , an apparent cut - off time for the tracer particles fluctuation measurements is shown .
these effects are due to the finite size effect imposed upon the tracers by the finite trajectories and should be inversely proportional to the tracer particles velocities .
tracer particles with larger mobility will have larger mean velocities and take a shorter time to complete its trajectory in the cell .
the cut - off discussed in reference to fig .
[ d_m_t]a is prominently displayed in the @xmath23 mm delrin tracers of packing 1 , having the largest mobility , hence increased mean velocities , as shown in fig .
[ d_m_t]b .
the larger mobility results in the shortest cut - off time for the diffusivity .
conversely , @xmath11 mm delrin tracers of packing 1 have a smaller mobility , hence a longer cut - off time for the diffusivity .
it is important to note that for all tracer particles studied here , the cut - off is observed for distances larger than a few particles diameters , ensuring that the study examines the structural motion of the grains and not internal motion inside of `` cages '' . according to a fluctuation - dissipation relation , we calculate @xmath8 : @xmath143 ^ 2\rangle}{2\langle z(t+\delta t ) - z(t)\rangle } .
\label{tcalc}\ ] ] fig .
[ d_m_t]c shows a parametric plot of fluctuations and responses , with @xmath76 , as the parameter , as extracted from fig .
[ d_m_t]a and fig .
[ d_m_t]b .
a linear relationship exists between diffusivity and mobility , with a slope of @xmath8 .
we obtain for the both nylon and delrin @xmath11 mm tracers in packing 1 , @xmath144 .
if the effective temperature is to be regarded as an intensive thermodynamic quantity , changing the tracer particle size should give rise to a different diffusion and mobility yet result in the same measurement of effective temperature .
the above calculation is repeated for delrin tracers of @xmath23 mm in packing 1 .
we find that while the mobility and diffusivity change dramatically with respect to tracers of @xmath11 mm , ( @xmath145 m@xmath139/s and @xmath146
s / kg ) as shown in table [ table ] , due to the change in tracer size , their ratio remains unchanged . in all cases @xmath69 and @xmath83 are inversely proportional to the size of the tracers , but the effective temperature remains approximately the same , as seen in fig .
[ d_m_t]c , with an average value over all tracers of @xmath147 though this effective temperature is high with respect to the bath temperature , we note that a plausible scale for the system energy @xcite , is @xmath148 , the gravitational potential energy to move a nylon tracer particle one particle diameter , @xmath111 .
a corresponding temperature would arise from the conversion of this energy into a temperature via the boltzmann constant , @xmath149 , is @xmath150 at room temperature ( t = 300k ) .
this specific value serves as a coarse - grained estimate , since the tracer size and density clearly shift its value , and we focus on the order of magnitude .
this large value is expected @xcite , and agrees with computer simulation estimates for an athermal granular system @xcite .
therefore , our calculated value for @xmath8 in a sheared granular system appears reasonable within the boundaries of the present theory . in an effort to further test the concept of the effective temperature as an intensive quantity ,
a linear response regime in the system is of great interest .
such a regime would imply that mobility and diffusivity are independent of the external gravitational force as @xmath151 .
the external force is varied by changing the density of the tracers of the same size .
this is realized experimentally in packing 1 by the introduction of delrin ( @xmath31 ) tracers of @xmath11 mm diameter , the density of which is higher than that of nylon ( @xmath30 ) .
analysis of the trajectories reveals that the mobility is approximately the same for both the delrin and nylon tracers with the same diameter and is thereby independent of the external force , as shown in fig .
[ d_m_t]b and fig . [ m_f ] .
further , the external force should have no effect on the diffusivity . by calculating the diffusivity of the non - tracer particles via dying acrylic tracers and analyzing their trajectories , as shown in fig .
[ d_m_t]a , the diffusion of the acrylic tracers of size @xmath23 mm ( for which no external force is applied ) is the same as the diffusion of the delrin tracers of the same size ( for which the gravitational force is applied ) .
a further example of this property would be using two different tracers with different sizes , but having the same external force applied .
one would calculate two different values of the diffusivity , due to the variation in tracer size , without having any variation in external force .
nonlinear effects appear for tracers heavier than delrin , implying that mobility depends on the external force for large enough forces .
we find that for a @xmath11 mm ceramic tracer @xmath152 in packing 1 the mobility is @xmath153 s / kg and for a brass tracer @xmath154 , @xmath155 s / kg as shown in table [ table ] , smaller than the mobility of the nylon and delrin tracers of the same size .
this behavior is expected since if a linear regime exists in the system , it will be valid only within certain limits , i.e. , @xmath83 remains a constant for small value of external force , @xmath29 , as shown in fig .
it is here that our experiments approach the boundaries presented above for estimated of energy scales for the sheared granular system .
our effective temperature measurements are therefore limited to those tracer particles for which we experience a linear regime with respect to both mobility and diffusivity .
lastly , the experiment is again repeated for a different packing of spherical particles , noted earlier as packing 2 . having nearly the same volume fraction of particles being both packings of spherical particles , one would expect @xmath8 to remain unchanged , as it is a measure of how dense the particulate packing is ( i.e. a large @xmath8 implies a loose configuration , e.g. random loose packing , while a reduced @xmath8 implies a more compact structure , e.g. random close packing ) .
it is found that although differences exist between the two packings with respect to mobility and diffusivity , as shown in fig .
[ d_m_t]a , b , the effective temperature remains approximately the same , as shown in fig .
[ d_m_t]c .
it should be noted that both packings are composed of spherical particles and the statement regarding effective temperature as a measure of particulate packing density would not be true if the packings are composed of particles of , for instance , different shapes , even if they have the same volume fraction .
we further explore the effective temperature as an intensive quantity by analyzing diffusivity and mobility as a function of the shear rate .
we show in fig .
[ d_m_shear ] that the effective temperature seems to become approximately constant , as long as the particulate motion is slow enough such that the system is very close to jamming .
we find that @xmath156 @xmath157 while @xmath158 remains approximately constant for sufficiently small @xmath72 .
it is within this quasi - static range where the effective temperature could be identified with exploration of the jammed configurations . as it remains an important assumption of this study that the system is being continuously jammed , shear rates high enough to impact the effective temperature measurement imply systems that are not continuously exploring jammed configurations . as we study the nature of the jammed granular packings , it is logical to presume that quasi - static shearing will provide systems of interest .
the limit of @xmath8 as @xmath159 may result in an effective temperature for the static jammed configuration .
the quasi - static shear rate regime observed could be analogous to the shear - rate independent regime observed in the behavior of shear stress in slowly sheared granular materials @xcite . this solid friction - like behavior has been previously studied @xcite and occurs when frictional forces and enduring contacts dominate the dynamics .
this regime has been also observed in recent computer simulations of the effective temperature of sheared granular materials @xcite .
our calculations of @xmath160 for systems close to jamming exclude the systems outside of the quasi - static range , in accordance with prior studies .
in contrast to measurements of slow mode temperatures , exemplified by @xmath8 , we also measure the temperature of the fast modes as given by the root mean square ( rms ) fluctuations of the velocity of the particles .
it should be noted that these velocities are not instantaneous , as the time necessary to obtain an instantaneous velocity is much smaller than the time between measurements .
nevertheless , we can obtain an estimate of the kinetic granular temperature , @xmath161 , from @xmath162 , where @xmath163 with @xmath164 the average kinetic energy of the grains .
we obtain @xmath165 , or @xmath166 and @xmath167 , or @xmath168 , for @xmath23 mm and @xmath11 mm delrin tracers in packing 1 , respectively . here , @xmath169 , the room temperature , and @xmath170 .
this kinetic granular temperature is smaller than @xmath160 and differs for each type of tracer indicating that it is not governed by the same statistics .
similar results have been obtained in experiments of vibrated granular gases @xcite .
the significance of this result is that fast modes of relaxation are governed by a different temperature .
this result is analogous to what is found in models of glasses and computer simulations of molecular glasses ( see for instance refs .
@xcite ) . in the glassy phase of these models
, the bath temperature is found to control the fast modes of relaxation and a different , larger , effective temperature is found to control the slow modes of relaxation .
similarly , we find a granular bath temperature for the fast modes and a larger effective temperature for slow modes of relaxation .
it is possible to identify @xmath8 as the property of the system governing the exploration of jammed configurations .
as this particular non - equilibrium system remains athermal , the bath temperature in which the grains exist is immaterial , as shown above .
particle diffusion is of the order of several particle diameters over the time scale of the experiment ( see fig .
[ trajectory1 ] and [ d_m_t]a ) implying that exploration of the available jammed configurations occurs via rearrangements of the particles outside their `` cages '' , suggesting that the trajectory of the system can be mapped onto successive jammed configurations explored by the system . incorporating certain experimental conditions of reversibility , and ergodicity
, a statistical mechanics formulation may well describe a jammed granular system @xcite . under the primary assumption that different jammed configurations are taken to have equal statistical weight ,
observables can be calculated by `` flat '' averages over the jammed configurational space @xcite .
this assumption , advocated by edwards and collaborators , has been thoroughly debated in the literature ( see for instance @xcite ) .
existing work suggests the effective temperature obtained by applying a fluctuation - dissipation theory to non - equilibrium systems is analogous to performing a `` flat '' average over the jammed configurational space , at least for frictionless systems @xcite .
additionally , the effective temperature can be identified with the compactivity introduced in @xcite , resulting from entropic calculations of the granular packing @xcite . experimentally testing these ideas
is difficult as the entropy of the jammed configurations is not easily measured , and it is not possible to obtain the compactivity from entropic considerations in the present study .
the exploration of reversible jammed states in granular matter bears similarity to that of inherent structures in glasses .
inherent structures form a network of attractive basins within an energy landscape , and the system explores these basins as governed by their stability over the slow - relaxation time of the glass
. it should be noted , however , that there exists a crucial difference between glasses and grains . in liquids energy
remains conserved , while energy is dissipated in granular systems through frictional contact and path dependent forces between grains .
thus , a driven granular system will quickly come to a mechanically stable , or jammed , state after the removal of the driving forces . by its nature
, energy is not conserved in a granular system . as energy conservation is the crucial property used to define an energy ensemble in statistical mechanics
, the use of energy to characterize granular systems is questionable .
thus , while @xmath8 seems to imply the exploration of reversible jammed states within an energy ensemble , with @xmath171 describing the nature of the exploration , the validity of the energy ensemble to describe granular matter in the absence of energy conservation remains an open question .
here , we set the analogous boltzmann constant for grains equal to unity for simplicity . noting the drastic difference between the bath temperature and the effective temperature in a granular system , we are inspired towards a more careful analysis of the energy ensemble in slowly driven granular systems .
the work of edwards has promoted the concept of a volume ensemble , where the free volume per grain in a static granular system replaces the energy as the conserved quantity of the non - equilibrium system , at a particular volume fraction @xcite . the basis for using the volume ensemble stems from the ability to conserve volume in a given packing and additivity of volume per grain .
further , it is possible to explore the configuration of states at a fixed volume , via experiment or simulation .
the statistical mechanics is then derived using methods similar to boltzmann statistics for equilibrium systems . from these methods , one can obtain the compactivity , @xmath172 , as a derivative of the entropy with respect to the volume , enabling the calculation of an equation of state in the volume ensemble as follows .
@xmath173 the compactivity , @xmath172 , is thereby assumed to be an equilibrium measure of a system within the framework of the volume ensemble , much like the bath temperature of the energy ensemble .
this assumption can be realized by performing an abc experiment and testing a zero - th law of thermodynamics for volumes @xcite .
according to the zero - th law of thermodynamics , if system a and c are in thermal equilibrium with system b respectively , then a and c are in thermal equilibrium with equal temperature .
in granular system , such an experiment would require two granular systems with distinct volumes , @xmath174 and @xmath175 , with the same @xmath172 .
bringing these two systems together should result in a granular system of volume @xmath176 , at the same @xmath172 , if the assumption is valid .
this experiment is feasible due to the fact that it is always possible to prepare a system at a given volume fraction and will be the subject of future study and experiment , facilitated by recent theoretical findings @xcite .
similar to the conservation of volume , boundary stress may also be a conserved quantity in jammed granular systems , and edwards statistical mechanics for volume distributions could be applied analogously to the distribution of boundary stresses , @xmath177 , or forces , referred to as the force ensemble @xcite .
the angoricity , @xmath178 , is calculated as the derivative of the entropy with respect to the boundary stress , and an additional equation of state is thereby achieved as follows : @xmath179 this result can be combined with that of the volume ensemble in an effort to accurately define the statistical mechanics of static jammed granular matter . such an approach remains a topic of ongoing research .
however , slowly driven granular systems introduce yet another ensemble , the energy ensemble , from which the above defined @xmath8 is derived .
while the above results reveal that @xmath8 does not tend to zero as the magnitude of the driving force decreases , indicating extrapolation to a non - zero static quantity , it remains unclear how the effective temperature may relate to the compactivity and angoricity as defined by edwards statistics .
are we defining a new static quantity by determining the static limit of @xmath8 , or are we expanding the statistical mechanics of jammed granular matter to include dynamic systems by relating @xmath8 , @xmath172 and @xmath178 ?
@xmath8 is obtained in the quasi - static limit @xmath180 , while the volume and force ensembles correspond to @xmath181 , exactly .
is it possible that a relation between @xmath8 , @xmath172 and @xmath178 can be expected ? there exists the further requirement of energy conservation for the validity of a boltzmann approach that would guarantee : @xmath182 as discussed above , energy is constantly dissipated in a driven granular system , through coulomb friction and path - dependent tangential forces between grains .
however , the input of energy by the external driving force brings the system to a steady state where the average energy is constant over the time - scale of the experiment .
this steady state energy could be likened to the conserved variable in a statistical formalism depicted in eq .
[ effective ] , thereby introducing a thermodynamic meaning for @xmath8 . in a compressed emulsion system
the absence of coulomb friction and inter - particle tangential forces greatly simplifies the formalism @xcite . jamming occurs due to osmotic pressure , and
the system remains athermal as a result of the large particle size .
a well defined potential energy exists due to the absence of tangential forces , corresponding to the deformation of the particles at the inter - particle contact points .
therefore , a restriction to use the energy ensemble in an effort to describe a jammed system is lifted , as frictional tangential forces no longer hinder energy conservation .
computer simulations of frictionless emulsion droplets @xcite incorporate a simulated annealing method employing an auxiliary temperature to sample the available jammed configurations .
the simulated annealing method assumes a boltzmann distribution , or a flat average assumption , over the jammed states of the emulsion .
the @xmath8 obtained by eq .
[ effective ] with simulated annealing methods @xcite is very close in value to the @xmath160 obtained via the fdt calculations as in the present work .
such a result could indicate that ergodicity holds in this frictionless system , a further justification of the methods presented herein .
therefore , a firmer basis for the validity of using the @xmath8 obtained in the quasi - static limit to describe the statistical mechanics of the same system at the static limit is achieved .
further , @xmath8 remains approximately constant with varying tracer particle size , implying a zero - th law of thermodynamics for slowly sheared jammed granular systems .
these statements further provoke the necessity for an abc experiment , to test the zero - th law for the effective temperature , as well as similar experiments for the compacitivity and angoricity .
such experiments may enlighten us to understand under what conditions @xmath183 , @xmath184 and @xmath185 may be valid in describing the statistics of the jammed and nearly jammed granular systems . at this point , we believe that the most prominent direction is the exploration of the volume and pressure ensembles .
our understanding is that these ensembles may be sufficient to characterize the jammed state of granular matter , while the energy ensemble may be necessary for slowly moving granular systems .
these are open questions at the present time .
recent papers in the theory and simulation front suggest that the compactivity characterizes the system into a phase diagram at the isostatic point , while the angoricity will be necessary to describe the pressure ensemble of compressible granular matter @xcite .
in summary , this study focuses on the dynamics of slowly sheared granular matter in a 3d couette cell . a mixture of spherical , transparent and bi - disperse grains are confined between two cylinders , having walls roughened by glued identical grains , with the inner cylinder rotated via motor .
we compact the grains by means of an external pressure in the negative z - direction .
fluid matching the density and refractive index of the grains partially fills the cell , allowing tracking of tracer particle trajectories as a function of time .
tracers of varying density and size are used .
multiple cameras track the tracer particle positions relative to the cylinders .
we find that the angular velocity of the tracer particles , @xmath42 , follows an exponential relation with @xmath45 , defined by the type of packing and geometry of the couette cell .
the velocity of the last layer of grains is non - zero , such that the shear band is located at the outer cylinder and ensures no formation of shear bands in the bulk . near
the outer cylinder @xmath42 decays slowly with increasing @xmath45 , such that @xmath186 can be approximated linearly with a constant local shear rate .
the constant local shear rate ensures that the mobility and diffusivity of tracers , dependent on local shear rate , remain approximately independent of @xmath45 .
we define this region the `` constant mobility and diffusivity region '' , or the cmd region .
an `` effective temperature '' , @xmath8 , is realized by a fluctuation - dissipation relation generalized to granular materials .
statistical measurements are confined exclusively to the cmd region .
the mobility in the vertical direction , @xmath83 , is found to be proportional to the shear rate , @xmath72 , for small enough values of @xmath72 . as @xmath69 is also found proportional to shear rate , collapsing all the @xmath65 for various shear rates shows a plateau in the cmd region .
an approximately constant effective temperature is obtained from measurements of the mobility and diffusivity , under a constant external applied force , and with sufficiently small shear rates .
this effective temperature is calculated by an analogous equation used in equilibrium statistical mechanics .
we find this effective temperature to be independent of the tracer particle properties , and dependent only on the packing density of the system .
while this result describes an intensive property of the system , it remains an important future study to test the effective temperature against the laws of thermodynamics .
more specifically , a test of the zeroth - law of thermodynamics with respect to these non - equilibrium jammed systems could expand the scope of @xmath8 beyond that of an intensive quantity of a particular system .
a well defined effective temperature in the radial direction may exist , though its existence would require a constant external force applied in the radial direction .
the probability distribution of the displacements in the radial direction , @xmath89 , reveals exponential fluctuations .
the analysis of the fluctuations reveals a power law , sub - diffusive , process , @xmath187 , with @xmath5 less than unity . a similar analysis for fluctuations in the angular direction reveal a super - diffusive process , @xmath188 , with @xmath6 greater than unity .
lastly , the probability distribution of the displacements in the vertical direction are found have a gaussian distribution such that @xmath189 .
it is this linearity that defines vertical displacement as a diffusive process , and allows for the use of the fluctuation - dissipation relation to calculate the diffusivity in the vertical direction .
we further discover a linear relationship between angular displacement and the time between measurements @xmath190 , such that all mean square fluctuations can be defined in terms of @xmath191 for the small shear rates of our experiments . in the cmd region ,
the linear approximation of @xmath42 proportional to approximately constant external shear rate , @xmath72 , allows for the collapsing of all tracer particle velocity curves via dividing @xmath42 by the shear rate .
this collapse reveals a periodic shape with a small amplitude and periodic length roughly equal to the grain size .
the effect is shown to be weaker in packings with smaller size grains .
we further apply this remarkable scaling feature @xmath43 and @xmath44 , achieving similar results .
it is important to note that the effective temperature , defined in this study for small shear rates , does not remain constant as the shear rate increases .
while previous studies have discovered an an increasing effective temperature via simulations , we have measured diffusivity and mobility separately in an effort to calculate @xmath8 through a fluctuation - dissipation relation .
we find the diffusivity in the z - direction remains approximately constant throughout the range of shear rates used in this experiment , while the mobility in the z - direction approaches a plateau . exclusively increasing @xmath160 .
such an effect in the radial direction would be of great interest for future studies in sheared granular dynamics .
the nearly constant value of @xmath8 with respect to varying tracer particle size indicates that a zero - th law of thermodynamics for slowly sheared jammed granular systems could be valid and prompts one to perform an abc experiment , fully testing the zero - th law for the effective temperature .
as we work towards a more complete description of the statistical mechanics of jammed granular matter , we strive to incorporate the varied statistical ensembles into one fundamental picture .
these ensembles include the energy ensemble , as described herein , along with the volume and force ensembles , as proposed by edwards .
such an incorporation may link static quantities of compacitivity and angoricity , describing volume and force ensembles , respectively , to the dynamic effective temperature presented in this study , derived from the energy ensemble .
the exact nature of the relation between such quantities remains an open topic .
ultimately , these quantities will help to develop a thorough statistical description for jammed granular matter and reveal an equation of state . a deeper topic of concern is the formation of a clear definition of energy in jammed granular matter .
energy is not conserved in frictional systems and it remains open to debate as to how one would incorporate energy into the statistical mechanics .
one possible approach to describe the energy of jammed systems is to consider the similarities between the inherent structure formalism of glasses and the exploration of jammed states in granular matter , at least for the case of frictionless granular systems .
inherent structures probe a network of potential energy basins within an energy landscape .
such an approach toward the jammed states of granular matter may assist in understanding exactly what is meant by energy within the framework of a non - equilibrium system . | we report experimental measurements of particle dynamics on slowly sheared granular matter in a three - dimensional ( 3d ) couette cell .
a closely - packed ensemble of transparent spherical beads is confined by an external pressure and filled with fluid to match both the density and refractive index of the beads .
this allows us to track tracer particles embedded in the system and obtain three - dimensional trajectories , @xmath0 , as a function of time .
we study the pdf of the vertical and radial displacements , finding gaussian and exponential distributions , respectively . for slow shear rates ,
the mean - square fluctuations in all three directions are found to be dependent only on the angular displacement of the couette cell , @xmath1 , @xmath2 , @xmath3 , @xmath4 , where @xmath5 and @xmath6 are constants . with @xmath1
proportional to the time between measurements , the values of the constants , @xmath5 and @xmath6 , are found to be sub - diffusive and super - diffusive , respectively .
the linear relation between @xmath7 and angular displacement implies a diffusive process , from which we can calculate an `` effective temperature '' , @xmath8 , in the vertical direction , through a fluctuation - dissipation relation .
it is of interest to determine whether these systems can be described by analogous equilibrium statistical mechanics concepts such as `` effective temperature '' and `` compactivity '' . by studying the dynamics of tracer particles , we find the effective temperature defined by the stokes - einstein relation to be independent of the tracer particle characteristic features , such as density and size , and dependent only on the packing density of the system . for slow shear rate ,
both the diffusivity and mobility of tracer particles are proportional to the shear rate , giving rise to a constant effective temperature , characteristic of the jammed system .
we finally discuss the significance of the existence of @xmath8 for a statistical mechanics formulation of granular matter . |
it has been well known for a long time that the disks of spiral galaxies show negative radial abundance gradients , in the sense that the abundance is higher at the centre and decreases with galactocentric distance @xcite .
it is a common practice to describe the nebular abundance distribution across the disk of a galaxy by the relation between oxygen abundance o / h and galactocentric distance @xmath3 and to specify this distribution by the characteristic abundance ( the abundance at a given galactocentric distance , e.g. , abundance at the centre of the disk ) and by the radial abundance gradient .
relations of this type were determined for many galaxies by different authors ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* among many others ) .
such relations are based on the assumption that the abundance distribution across the disk is axisymmetric .
the azimuthal abundance variations across the disk of a galaxy were discussed in a number of papers .
the dispersion in abundance at fixed radius ( the scatter around the general o / h @xmath4 trend , residuals ) is used as the measure of the azimuthal abundance variations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the number of data points in such investigations is usually small .
the two - dimensional abundance distribution was analyzed for the galaxy ngc 628 @xcite .
the observations obtained by the califa survey ( calar alto legacy integral field area survey ; * ? ? ?
* ) provide the possibility to construct abundance maps for disk galaxies .
this allows one to investigate the distribution of abundances across the disk in detail , in particular the global azimuthal asymmetry of abundance in the disks of galaxies .
we define the global azimuthal asymmetry in abundance in the following way .
we divide the target galaxy into two semicircles and determine the difference between the arithmetic means of the residuals for the opposite semicircles .
the differences are determined for the different position angles of the dividing line .
the maximum absolute value of this difference is adopted as a measure of the global azimuthal asymmetry of abundance in the disk of a galaxy . in the present work
, we will determine the values of the global azimuthal abundance asymmetry in the disks of the califa galaxies and compare them with the level of azimuthal abundance variations .
thus , the goal of this investigation is to construct maps of the oxygen abundance distribution in the disks of the selected califa galaxies and to use those maps to explore the presence ( or lack ) of a global azimuthal asymmetry in the oxygen abundance as well the existence of a break in the radial oxygen abundance distribution , e.g. , a flattening of the radial abundance gradient at the central part of a galaxy .
evidence for a decrease of the oxygen abundances in the central parts of a number of the califa galaxies was found by @xcite .
we also examine whether the geometric parameters of a galaxy ( coordinates of the center , inclination and position angle of the major axis ) can be estimated from the analysis of the abundance map .
the geometric parameters of a galaxy are usually determined from the analysis of the photometric or / and velocity map of the galaxy under the assumption that the surface brightness of the galaxy ( or the velocity field ) is a function of the radius only , i.e. , that there is no azimuthal asymmetry in brightness ( or velocity ) . since the metallicity in the disk is also a function of the galactocentric distance one may expect that the abundance map can also be used for the determination of the geometric parameters of a galaxy .
we will determine the `` chemical '' ( abundance - map - based ) geometrical parameters of our galaxies and compare them to their `` photometric '' ( canonical , photometry - map - based ) geometrical parameters .
the paper is organized in the following way .
the algorithm employed in our study is described in the section 2 .
the results and discussion are given in section 3 .
section 4 is a brief summary . throughout the paper
, we will use the following notations for the line fluxes : @xmath5 = @xmath6 \lambda 3727 + \lambda 3729 } /i_{\rm { \rm h}\beta } $ ] , @xmath7 = @xmath8 \lambda 6548 + \lambda 6584 } /i_{\rm { \rm h}\beta } $ ] , @xmath9 = @xmath10 \lambda 6717 + \lambda 6731 } /i_{\rm { \rm h}\beta } $ ] , @xmath11 = @xmath12 } \lambda 4959 + \lambda 5007 } /i_{\rm
{ \rm h}\beta } $ ] .
the units of the wavelengths are angstroms .
we used publicly available spectra from the integral field spectroscopic califa survey data release 2 ( dr2 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) based on observations with the pmas / ppak integral field spectrophotometer mounted on the calar alto 3.5-meter telescope .
califa dr2 provides wide - field ifu data for 200 galaxies . in fig .
[ figure : images ] we present an example of _ sdss _ images of six galaxies . the data for each galaxy consist of two spectral datacubes , which cover the spectral regions of 43007000 at a spectral resolution of @xmath13 ( setup v500 ) and of 37005000 at @xmath14 ( setup v1200 ) .
the spectrum of each spaxel from the califa dr2 datacubes ( setup v500 ) is processed into two steps .
first , the stellar background in all spaxels is fitted using the public version of the starlight code @xcite .
we use a set of 45 synthetic simple stellar population ( ssp ) spectra with metallicities @xmath15 , 0.02 , and 0.05 , and 15 ages from 1 myr up to 13 gyr from the evolutionary synthesis models of @xcite and the reddening law of @xcite with @xmath16 .
the resulting stellar radiation contribution is subtracted from the measured spectrum in order to find the nebular emission spectrum .
second , the emission lines are fitted by gaussians .
the widths of the individual lines of the doublets [ oiii]@xmath17@xmath174959,5007 , [ nii]@xmath17@xmath176548,6584 , and [ sii]@xmath17@xmath176717,6731 are set to be equal to each other for each doublet . since only the low resolution ( v500 setup )
spectra are used the [ oii]@xmath17@xmath173727,3729 doublet is fitted by a single gaussian . applying @xmath18 minimization ,
the estimation of the flux error relies on the @xmath18 quadratic approximation in the neighborhood of the @xmath18 minimum based on a hessian matrix . for each spectrum , we measure the fluxes of the [ oii]@xmath173727+@xmath173729 , h@xmath19 , [ oiii]@xmath174959 , [ oiii]@xmath175007 , [ nii]@xmath176548 , h@xmath20 , [ nii]@xmath176584 , [ sii]@xmath176717 , [ sii]@xmath176731 lines . the measured line fluxes are corrected for interstellar reddening using the theoretical h@xmath20 to h@xmath19 ratio ( i.e. , the standard value of h@xmath20/h@xmath19 = 2.86 ) and the analytical approximation of the whitford interstellar reddening law from @xcite .
when the measured value of h@xmath20/h@xmath19 is lower than 2.86 the reddening is adopted to be zero .
the [ oiii]@xmath175007 and @xmath174959 lines originate from transitions from the same energy level , so their flux ratio is determined only by the transition probability ratio , which is very close to 3 @xcite .
the stronger line [ oiii]@xmath175007 is usually measured with higher precision than the weaker line [ oiii]@xmath174959 .
therefore , the value of @xmath11 is estimated as @xmath21 [ oiii]@xmath175007 but not as a sum of the line fluxes .
similarly , the [ nii]@xmath176584 and @xmath176548 lines also originate from transitions from the same energy level and the transition probability ratio for those lines is again close to 3 @xcite . the value of @xmath7 is therefore estimated as @xmath22 [ nii]@xmath176584 . thus , the lines h@xmath19 , [ oiii]@xmath175007 , h@xmath20 , [ nii]@xmath176584 , [ sii]@xmath176717 , and [ sii]@xmath176731 are used for the dereddening and the abundance determinations .
the precision of the line flux is specified by the ratio of the flux to the flux error ( parameter @xmath23 ) .
we select spectra where the parameter @xmath24 for each of those lines .
if weak auroral lines such as such as [ oiii]@xmath174363 are detected in the spectrum of an hii region the oxygen abundance can be derived using the direct @xmath25 method , which is considered to yield the most reliable nebular oxygen abundance determinations ( e.g. , * ? ? ?
this method requires the measurement of reddening - corrected diagnostic line ratios and knowledge of the local physical conditions ( i.e. , the effective tempeature of the gas , @xmath25 , and its electron density ) .
@xmath25 is then usually determined from the intensity ratio of the auroral to nebular lines of doubly ionized oxygen ; in other words , _ directly _ from the observed line ratios ( hence the name `` direct @xmath25 method '' ) . a short description of the required steps is given in @xcite .
more details can be found in @xcite and @xcite . in our current study
, the oxygen abundances will be determined through a strong - line method ( the @xmath26 method @xcite ) since the auroral lines are undetectable in the majority of the califa spectra @xcite and , consequently , the @xmath25 method can not be applied .
the @xmath26 method is based on the assumption that hii regions with a similar combination of intensities of metallicity - sensitive strong emission lines have similar abundances . when the strong lines [ oiii]@xmath174959,5007 , [ nii]@xmath176548,6584 , and [ sii]@xmath176717,6731 are measured in the spectrum of an hii region the oxygen ( o / h)@xmath27 abundance can be determined .
when the strong lines [ oii]@xmath173727,3729 , [ oiii]@xmath174959,5007 , and [ nii]@xmath176548,6584 are measured in the spectrum then we can determine the oxygen ( o / h)@xmath28 abundance .
since the oxygen line [ oii]@xmath173727+@xmath173729 is not available or noisy in many spectra of the califa - v500 setup the @xmath29 variant of the @xmath26 method is used for the abundance determination in the hii regions . it should be emphasized that our analysis is restricted to the v500 setup data . in order to improve the accuracy of the @xmath26 method
we have updated our sample of reference hii regions that are used for the abundance determination with this method .
new spectral measurements of hii regions with detected auroral lines were added to our collection @xcite .
the abundances in those hii regions were determined through the @xmath30 method .
the realization of the @xmath30 method , i.e. , the equations of the @xmath30 method that serve to convert the values of the line fluxes to electron temperatures and abundances , are described in numerous papers ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the equations adopted in our current study were derived from the five - level atom solution with recent atomic data ( einstein coefficients for spontaneous transitions , energy level data , and effective cross sections for electron impact ) in @xcite . using the collected data with @xmath25-based abundances
we select a sample of reference hii regions following @xcite , i.e. , we select a sample of reference hii regions for which the absolute differences of the oxygen abundances ( o / h)@xmath28 ( o / h)@xmath31 and ( o / h)@xmath27 ( o / h)@xmath31 and of the nitrogen abundances ( n / h)@xmath28 ( n / h)@xmath31 and ( n / h)@xmath27 ( n / h)@xmath31 are less than 0.1 dex .
this sample of reference hii regions contains 313 objects and will be used for the abundance determinations in the present paper .
this sample of hii regions has been also used as calibration data in the construction of a new calibration for abundance determinations in hii regions @xcite .
the good agreement between the calibration - based and @xmath30-based abundances suggests that our reference sample does indeed contain hii regions with reliable oxygen abundances ( o / h)@xmath31 . we determine the oxygen abundance in the target object by comparing it with 30 counterpart reference hii regions in order to decrease the influence of the uncertainties in the oxygen abundances of the individual reference hii regions on the resulting abundance of the target objects .
since the abundances produced by the @xmath26 method and @xmath30 method are close to each other our approach is very similar to using the empirical metallicity scale defined by hii regions with oxygen abundances derived through the direct method ( @xmath25 method ) .
this conclusion has been confirmed in our recent studies @xcite .
it has been well known for a long time that the oxygen abundances produced by the calibrations based on hii regions with @xmath30-based measurements and calibrations based on photoionization models can differ by up to 0.7 dex ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
there are different types of objects with emission - line spectra , e.g. , hii region - like objects , agn - like objects , and liner - like objects .
the intensities of the strong lines can be used to separate different types of emission - line objects according to their main excitation mechanism .
@xcite proposed a diagnostic diagram ( the so - called bpt classification diagram ) where the excitation properties of hii regions are established from the [ nii]@xmath176584/h@xmath20 and [ oiii]@xmath175007/h@xmath19 line ratios .
the location of the different classes of the emission - line objects in the [ nii]@xmath176584/h@xmath20 and [ oiii]@xmath175007/h@xmath19 and other diagrams were investigated in many studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the exact location of the dividing line between hii regions and agns is still under debate . the demarcation line of @xcite in the [ nii]@xmath176584/h@xmath20 vs. [ oiii]@xmath175007/h@xmath19 diagram seems to be the most stringent condition to select true hii regions @xcite .
however , an appreciable fraction of hii regions is located to the right ( thus on the `` wrong '' side ) of this demarcation line @xcite .
the goal of the current study is to construct metallicity maps of galaxies .
thus we are interested in using as many points as possible .
some hii regions are lost when the demarcation line of @xcite is adopted .
can one avoid the loss of data points with reliable @xmath26-based abundances ?
panel @xmath32 of fig .
[ figure : f3811 ] shows the [ nii]@xmath176584/h@xmath20 vs. [ oiii]@xmath175007/h@xmath19 diagram for the spectra of individual spaxels in the galaxy ngc 3811 .
solid and dashed curves mark the boundary between agns and hii regions defined by @xcite and @xcite , respectively .
the light - blue filled circles show the large sample of emission - line sdss galaxies studied by @xcite .
the red points indicate agn - like objects in ngc3811 according to the dividing line of @xcite .
the black points represent hii - region - like objects in ngc3811 according to the dividing line of @xcite .
the dark - blue points are `` intermediate '' objects in ngc 3811 located between the dividing lines of @xcite and @xcite .
panel @xmath33 shows the locations of the agn - like , hii - region - like , and `` intermediate '' objects in ngc 3811 in the deprojected face - on image of the galaxy .
panel @xmath34 shows the radial distribution of oxygen abundances in the disk of the galaxy ngc 3811 traced by the hii - region - like and `` intermediate '' objects .
[ figure : f2410 ] shows similar diagrams for ngc 2410 .
the bpt diagram for ngc 2410 differs from that for ngc 3811 .
there is a well - defined agn branch in the case of ngc 2410 .
panels @xmath34 in fig .
[ figure : f3811 ] and fig .
[ figure : f2410 ] clearly show that the abundances of the intermediate objects closely follow the general radial oxygen abundance trends traced by the hii regions selected with the dividing line of @xcite .
therefore the demarcation line of @xcite is adopted here .
it should be emphasized that we do not claim to establish the exact locus of the boundary between the hii region - like and other classes of emission line objects .
we only illustrate that the @xmath26-based abundances in the objects located left of ( or below ) the demarcation line of @xcite and the objects located between the demarcation lines of @xcite and @xcite show the same radial oxygen abundance trend in the disks of our target galaxies .
the galaxy inclination and position angle of the major axis are estimated from the analysis of the surface brightness map in the @xmath1 band of the sloan digital sky survey ( sdss ) .
we constructed radial surface brightness profiles in the sdss @xmath0 and @xmath1 bands using the photometric maps of sdss data release 9 @xcite in the same way as in @xcite .
the optical isophotal radius @xmath35 of a galaxy is determined from the analysis of the surface brightness profiles in the sdss @xmath0 and @xmath1 bands converted to the vega @xmath36 band .
we use the spaxel coordinates of the califa plates .
the location of the center of the galaxy is at @xmath37 and @xmath38 .
the galactocentric @xmath39-coordinate is the right ascension offset with opposite sign .
the galactocentric @xmath40-coordinate is the declination offset .
since the size of a spaxel is equal to one arcsec the offset in spaxels is equal to the offset in arcsec .
we also use fractional radii , i.e. , radii normalized to the optical isophotal radius @xmath35 .
the radial oxygen abundance distribution within the optical isophotal radius of the disk , @xmath35 , is fitted by the following expression , which will be referred to as the o / h @xmath41 relation : @xmath42 where 12 + log(o / h)@xmath43 is the oxygen abundance at @xmath44 , i.e. , the extrapolated central oxygen abundance , @xmath45 is the slope of the oxygen abundance gradient expressed in terms of dex @xmath46 , and @xmath47/@xmath48 is the fractional radius ( the galactocentric distance normalized to the disk s isophotal radius @xmath35 ) . if there is an abundance depletion in the central part of the disk then this part is excluded from the gradient estimation ( for more details see section [ bends ] ) .
the radial abundance gradient obtained using the geometrical parameters of a galaxy from an analysis of the surface brightness map will be referred as case @xmath49 .
the radial abundance gradient obtained using the geometrical parameters based on an analysis of the abundance map will be referred as case @xmath26 ( see below ) .
we examine the global azimuthal asymmetry in the oxygen abundance distribution across the disk of a galaxy .
we divide the galaxy into two semicircles by a dividing line at a position angle @xmath49 . for a fixed value of the angle @xmath49
, we determine the arithmetic mean of the deviations @xmath50 from the o / h @xmath51 relation for spaxels with azimuthal angles from @xmath49 to @xmath52 and the mean deviation @xmath53 for spaxels with azimuthal angles from @xmath52 to @xmath54 .
the absolute value of the difference @xmath55 is determined for different positions of @xmath49 .
the position of @xmath49 is counted counterclockwise from the @xmath40-axis ( dashed line in panel @xmath32 of fig .
[ figure : n6132ac ] ) and is changed with a step size of 3@xmath56 in the computations in both the @xmath49 and the @xmath26 cases .
the maximum absolute value of the difference @xmath57 is used to specify the global azimuthal asymmetry in the abundance distribution across the galaxy .
the position angle of the dividing line marking the maximum absolute value of the difference @xmath57 will be referred to as the azimuthal asymmetry angle .
accurate geometric parameters of a galaxy ( position of the center of a galaxy , inclination angle , and the position angle of the major axis ) should be available for the determination of a reliable galactocentric distance of an hii region ( spaxel ) . in the present study ,
the position of the center of the galaxy is described by the coordinates @xmath58 and @xmath59 of the center of the galaxy in the ccd images .
it is a canonical way to determine the geometric parameters of a galaxy from the analysis of the photometric or / and velocity map of the galaxy .
it is assumed that the surface brightness distribution ( or velocity field ) is axisymmetric . since the metallicity in the disk is a function of the galactocentric distance one can expect that the abundance map can also be used for the determination of the geometric parameters of a galaxy .
here we examine whether the geometric parameters of a galaxy can be estimated from the analysis of the abundance map .
hence , in addition we obtain the geometric parameters of our galaxies from the analysis of the abundance map .
we search for a set of four parameters : the position of the galaxy center on the califa plate ( the @xmath58 and @xmath59 coordinates in spaxels ) , the inclination @xmath60 of the galaxy , and the position angle @xmath61 of the major axis .
we determine those parameters from the requirement that the correlation coefficient between oxygen abundance and galactocentric distance is maximum or the scatter around o / h @xmath4 relation is minimum .
it should be emphasized that both conditions result in the same values of the geometrical parameters .
we use the data points within the optical radius of the galaxy .
if there is a flattening of the metallicity distribution in the central part of the galaxy then this area is excluded from the analysis .
the optical radius of a galaxy @xmath35 is fixed and comes from the analysis of the photometric map .
[ figure : n6132ac ] illustrates the application of the described algorithm .
the galaxy ngc 6132 is a spiral _ sab _ galaxy ( morphological type @xmath62 ) at a distance of 76.5 mpc ( leda , ned ) .
our analysis of the @xmath1-band sdss map results in an inclination angle @xmath63 with a position angle of the major axis of @xmath64 .
the angular optical radius of the galaxy is 0.56 arcmin .
this results in a linear optical radius @xmath65 kpc for the adopted distance .
the left - column panels of fig .
[ figure : n6132ac ] show the properties of the abundance distribution across the disk of ngc 6132 where the deprojected radii of the hii regions ( spaxels ) were computed using the geometric parameters of the galaxy obtained from the photometric map , case @xmath49 . from the analysis of the abundance map
, we obtained the position of the `` chemical center '' of the galaxy at @xmath66 spaxels , @xmath67 spaxels , a position angle of the major axis of @xmath68 , and an inclination angle @xmath69 .
the panels in the right column of fig .
[ figure : n6132ac ] show the properties of the abundance distribution across the disk of ngc 6132 where the deprojected radii of the hii regions were computed using the geometric parameters of case @xmath26 .
the abundance - map - based geometrical parameters will be estimated for each califa galaxy of our sample below .
the comparison between the photometry - map - based and abundance - map - based geometrical parameters can tell us something about the credibility of the abundance - map - based geometrical parameters of galaxies .
we consider late - type galaxies from the califa dr2 list .
we include in our sample galaxies with morphological type sa and later types for which the following conditions are fulfilled : a ) the number of spaxels with measured h@xmath19 , [ oiii]@xmath175007 , h@xmath20 , [ nii]@xmath176584 , [ sii]@xmath176717 , and [ sii]@xmath176731 emission lines is larger than a few hundreds ; b ) those spaxels are distributed along the radius and with the azimuthal angle in such way that the radial abundance trend and azimuthal variations in the oxygen abundances can be investigated .
the number of the data points is usually larger than three hundred for each of the selected galaxies , but we also included three galaxies with 213 , 299 , and 176 data points , respectively .
of course , the procedure of including / rejecting galaxies in our sample is somewhat arbitrary .
the @xmath26-based abundances in some galaxies ( e.g. , ugc 312 , ugc 7012 ) show unusually large scatter .
the difference between maximum and minimum oxygen abundances at a given galactocentric distance can be as large as up to an order of magnitude .
this prevents a reliable determination of the radial abundance trend and the azimuthal variations in oxygen abundances .
when we consider only the spaxels where the flux to the flux error @xmath70 for each line ( instead of @xmath71 ) then the scatter significantly decreases .
this suggests that the large scatter in the oxygen abundances in those galaxies seems to be artificial and can be attributed to the uncertainty in the flux measurements or / and to the uncertainty in the oxygen abundance determinations through the @xmath26 method .
those galaxies ( eight galaxies ) were excluded from further consideration .
we also excluded three galaxies ( ngc 825 , ngc 2906 , and ugc 3995 ) with very flat ( @xmath72 dex ) radial abundance gradients
where we can not estimate the geometrical parameters of the galaxy from the abundance map .
our final sample contains 88 califa galaxies .
table [ table : sample ] lists the general characteristics of each galaxy : morphological type , isophotal radius @xmath35 , and distance .
table [ table : property ] lists the obtained properties of the oxygen abundance distributions in the disks of our sample of califa galaxies for both the photometric and abundance - map derived geometric parameters .
figure [ figure : gistprop ] summarizes the properties of our sample of califa galaxies .
figure [ figure : gistprop ] shows histograms of the distances ( panel @xmath32 ) , optical radii @xmath35 ( panel @xmath33 ) , morphological @xmath73 types ( panel @xmath34 ) , and central oxygen abundances 12 + log(o / h)@xmath43 ( panel @xmath74 ) for our galaxy sample .
the optical radii of our galaxies range from @xmath75 kpc to @xmath76 kpc , but galaxies with radii between 10 and 16 kpc occur most frequently .
the central oxygen abundances of most of the galaxies have a value between 12 + log(o / h)@xmath77 and 12 + log(o / h)@xmath78 .
a number of galaxies of our sample have a large inclination angle , @xmath79 .
can one obtain a reliable deprojected abundance map for such galaxies ?
a satisfactory agreement between the photometry - map - based values and abundance - map - based geometrical parameters of such galaxies ( as well as between the values of the radial abundance gradients obtained with photometry - map - based and abundance - map - based geometrical parameters ; table [ table : property ] ) can be considered as evidence in favor that reliable deprojected abundance maps can be obtained from the califa measurements even for galaxies with an inclination up to about @xmath80 .
figure [ figure : u11792ac ] shows the results for the galaxy ugc 11792 which is an example of a galaxy with high inclination , @xmath81 .
panel @xmath82 of fig .
[ figure : grad - dx ] shows the differences between the coordinates of the center obtained from the analysis of the photometric maps and the coordinates obtained from the analysis of the abundance maps ( the differences in the @xmath37 coordinates are presented by circles and the difference in the @xmath38 coordinates by the plus signs ) as a function of the radial abundance gradient across the optical disk of a galaxy .
panel @xmath83 of fig .
[ figure : grad - dx ] shows the differences between the coordinates of the center as a function of the azimuthal asymmetry in the abundance within the optical disk of a galaxy .
the photometry - map - based radial abundance gradient and asymmetry value are plotted along the @xmath39-axis .
inspection of fig .
[ figure : grad - dx ] and table [ table : property ] show that the photometry - map - based and abundance - map - based coordinates of the centers of galaxies usually agree within 2 3 spaxels , i.e. , the inferred photometric and chemical centers of a galaxy are usually close to each other .
however , there is a large ( more than 5 spaxels ) disagreement between the photometry - map - based and the abundance - map - based coordinates of the centers for some galaxies with flat radial abundance gradients .
panels @xmath84 and @xmath85 of fig .
[ figure : grad - dx ] show the comparison between the inclination angle @xmath86 obtained from the analysis of the photometric maps and the inclination angle @xmath87 obtained from the analysis of the abundance maps
. panels @xmath88 and @xmath89 of fig .
[ figure : grad - dx ] show the comparison between the position angle of the major axis @xmath90 obtained from the analysis of the photometric maps and the position angle of the major axis @xmath91 obtained from the analysis of the abundance maps .
inspection of fig .
[ figure : grad - dx ] shows that the inclination angles and the position angle of the major axis derived from the analysis of the abundance maps are rather close to those obtained from the analysis of the photometric maps for the majority of our galaxies .
again , there is a large ( more than 10 ) disagreement between the photometry - map - based and the abundance - map - based inclination angles for some galaxies with flat radial abundance gradients .
a flat radial abundance gradient ( i.e. , small variations of the abundance with radius ) makes the accurate determination of the geometrical parameters of a galaxy from the analysis of the abundance map difficult .
the galaxy ngc 171 , which has a relatively steep radial abundance gradient , @xmath92 dex @xmath93 , also shows a large disagreement between the photometry - map - based and the abundance - map - based values of the inclination angle @xmath94 ( panel @xmath84 of fig .
[ figure : grad - dx ] ) and between the values of the position angle of the major axis @xmath95 ( panel @xmath88 of fig .
[ figure : grad - dx ] ) .
the inclination angle of the galaxy ngc 171 is small ; @xmath96 .
therefore , the determined radial abundance gradient is not very sensitive to the values of ( and error in ) the values of the inclination angle and the position angle of the major axis .
hence , both the photometry - map - based and the abundance - map - based values of the inclination angle and of the position angle of the major axis can be rather uncertain .
panels @xmath97 and @xmath98 show the difference between the central oxygen abundances of the disk 12+log(o / h)@xmath43 obtained with photometry - map - based and abundance - map - based geometrical parameters of a galaxy as a function of the radial abundance gradient ( panel @xmath97 ) and asymmetry parameter ( panel @xmath98 ) .
the difference in the central oxygen abundance is usually small , less than 0.02 dex .
panels @xmath99 and @xmath100 show the difference between radial abundance gradients across the disk obtained with photometry - map - based and abundance - map - based geometrical parameters of a galaxy as a function of the radial abundance gradient ( panel @xmath99 ) and asymmetry parameter ( panel @xmath100 ) .
the difference is small ( less than 0.05 dex @xmath93 ) but for some galaxies it can be up to 0.1 dex @xmath93 .
thus , the geometrical parameters of a galaxy can be estimated from the analysis of the abundance map .
[ figure : n6132ac ] shows the galaxy ngc 6132 , which is a galaxy with possible azimuthal asymmetry in the oxygen abundance in the disk . indeed , the upper left panel of fig .
[ figure : n6132ac ] shows that the points with positive oxygen abundance deviations ( brown points ) from the general radial abundance trend are usually located in the opposite semicircle than the points with negative oxygen abundance deviations ( blue points ) .
the difference between the means of the residuals in two semicircles divided by the line with an angle @xmath49 = 147@xmath56 is @xmath1010.046 dex .
the scatter @xmath102 in oxygen abundances around the o / h @xmath4 relation is 0.051 dex .
let us clarify how significant this value of azimuthal asymmetry is .
panel @xmath32 of fig .
[ figure : sigasym ] shows the value of azimuthal asymmetry in the oxygen abundance in the disk of a galaxy as a function of the scatter @xmath102 in oxygen abundances around the o / h @xmath103 relation for our sample of galaxies for the case @xmath49 .
the values of the global azimuthal asymmetry are small and exceed 0.04 dex for a few galaxies only .
inspection of panel @xmath32 of fig .
[ figure : sigasym ] shows that there is a correlation between the value of azimuthal asymmetry and the value of the azimuthal variation ( scatter @xmath102 around of the o / h @xmath103 trend ) in oxygen abundance for the case @xmath49 . on the one hand
, this correlation may suggest that the azimuthal asymmetry makes a significant contribution to the scatter around the o / h @xmath4 relation .
this is the case if the global azimuthal asymmetry is real . on the other hand ,
the obtained azimuthal asymmetry may be artificial and can be caused by the uncertainties in the geometrical parameters of a galaxy or / and by the azimuthal asymmetry of the error in oxygen abundance . since the geometric parameters for case @xmath26 are derived by minimizing the scatter around the o / h @xmath103 relation while the geometric parameters for case @xmath49 are derived from the analysis of the photometric maps , the comparison of the azimuthal asymmetry and scatter for case @xmath49 and case @xmath26 allows us to check whether the obtained value of the azimuthal asymmetry for case @xmath49 is real .
the values of the scatter @xmath102 in oxygen abundances around the o / h @xmath103 relation for case @xmath49 are close to the ones for case @xmath26 ( panel @xmath33 of fig .
[ figure : sigasym ] ) .
in contrast , the values of azimuthal asymmetry in the oxygen abundance for case @xmath26 are significantly lower than those for case @xmath49 ( panel @xmath34 of fig .
[ figure : sigasym ] ) .
the values of the global azimuthal asymmetry are significantly smaller than the values of the scatter @xmath102 in oxygen abundances around the o / h @xmath103 relation for case @xmath26 ( panel @xmath74 of fig .
[ figure : sigasym ] ) .
thus , the panels @xmath33 , @xmath34 , and @xmath74 of fig .
[ figure : sigasym ] suggest that the scatter @xmath102 in oxygen abundances around the o / h @xmath103 relation can not be attributed to the global azimuthal asymmetry in the oxygen abundance distribution .
the azimuthal asymmetry obtained for case @xmath49 ( with photometry - map - based geometrical parameters ) seems to be artificial and may be caused by the uncertainties in the geometrical parameters of a galaxy at least in some galaxies . to verify whether the values of the global azimuthal asymmetry for the case @xmath49 can be caused by uncertainties in the geometrical parameters of a galaxy obtained from the analysis of the photometric map
, the following numerical experiment was performed .
a disk model including 400 points randomly distributed across the disk was constructed .
the oxygen abundance in each point was determined from the adopted radial abundance gradient .
then the random error was added to the oxygen abundance in each point in a such way that the random abundance errors follow a gaussian with an adopted scatter @xmath102 .
two values of the abundance gradient , @xmath104 and @xmath105 dex @xmath46 , and four values of the scatter in oxygen abundances , @xmath102 = 0.02 , 0.03 , 0.04 , 0.05 dex , were considered .
thus , eight disk models were examined .
we considered three values of the initial inclination of a galaxy , @xmath106 , and @xmath107 .
the initial position angle of the major axis of the galaxy was fixed at @xmath108 .
these models do not assume any cause of the appearance of global azimuthal asymmetry , except randomly distributed errors in the oxygen abundances caused by the finite number of points in the abundance map ( discussed below ) .
formally we can consider this case as the case @xmath26 .
we introduced a random deviation of the position of the galaxy center @xmath109 and @xmath110 , the galaxy inclination @xmath111 , and the position of the major axis @xmath112 .
the maximal values of those deviations were chosen to be @xmath113 spaxels , @xmath114 , and @xmath115 .
those deviations cover the range of the discrepancy between geometrical parameters in the cases @xmath49 and @xmath26 for the bulk of galaxies from our sample ( see fig . [
figure : grad - dx ] ) .
models with non - zero deviations of the geometric parameters can be regarded as the case @xmath49 .
the azimuthal asymmetry parameters were calculated for a set of 160 models .
[ figure : sigasym - sim ] shows the results of our simulations .
the panels and symbols in fig .
[ figure : sigasym - sim ] are the same as in fig .
[ figure : sigasym ] .
the discontinuity of the distribution of scatter is attributed to the low number of the initial values of the scatter ; only four initial values of scatter in the oxygen abundances were considered .
the panels @xmath32 and @xmath74 of fig .
[ figure : sigasym - sim ] , where the lowest asymmetries of case @xmath49 for a given scatter value are in close agreement with the asymmetries of the case @xmath26 , confirm that minimizing the scatter in oxygen abundance and deviations of the geometric parameters results in minimizing the azimuthal asymmetry .
comparison between fig .
[ figure : sigasym ] and fig .
[ figure : sigasym - sim ] shows clearly that our simulation reproduces the general behavior of the scatter and global azimuthal asymmetry obtained for our sample of the califa galaxies .
it can be considered as evidence favouring that the values of the global azimuthal asymmetry for the case @xmath49 can be caused by the rather small uncertainties in the geometrical parameters of a galaxy obtained from the analysis of the photometric map .
the obtained range of the global azimuthal asymmetry for case @xmath26 can be attributed to the azimuthal asymmetry in the randomly distributed errors in the oxygen abundances . to illustrate how uncertainties in the position of the galactic center affected to the global azimuthal asymmetry we consider ngc 5520 and ngc 6132 .
we shift the position of the galaxy s center along the @xmath39 or @xmath40 axis ( while other geometrical parameters are fixed ) and determine the value of the global azimuthal asymmetry and its azimuthal angle .
the left panel of fig .
[ figure : dxy - as ] shows the value of the global azimuthal asymmetry as a function of shift along the @xmath39 or @xmath40 axis .
this figure demonstrates that the range of the values of the global azimuthal asymmetry obtained for case @xmath49 can be reproduced by the uncertainties in the geometrical parameters of the galaxies .
the right panel of fig .
[ figure : dxy - as ] shows the azimuthal angle of the global azimuthal asymmetry as a function of the shift along the @xmath39 or @xmath40 axis .
this figure shows that in case of a small global azimuthal asymmetry , small uncertainties in the position of the galactic center can lead to large uncertainties in the angle of the global azimuthal asymmetry .
azimuthal variations of the oxygen abundance across galactic disks were discussed in the literature for several galaxies .
the scatter in oxygen abundance at a given radius was usually attributed to azimuthal variations in the oxygen abundance distribution . @xcite
have considered the dispersion in abundance at fixed radius in the disk of the galaxy m101 using 41 hii regions .
they found that there is evidence in favor of a non - axisymmetric abundance distribution .
however , they noted that more data are needed to test whether this asymmetry is real .
@xcite obtained and analyzed the radial oxygen abundance gradient in the disk of m101 using a sample of 79 hii regions .
they found no evidence for significant azimuthal variations of the oxygen abundance across the entire disk of this galaxy .
@xcite found scatter around the o / h @xmath4 relation of 0.128 dex in their fiber - to - fiber sample and of 0.070 dex in their hii region catalog of ngc 628 .
they concluded that the physical conditions and the star formation history of different symmetric regions of the galaxy ngc 628 have evolved in a different manner .
@xcite detected the auroral lines and measured direct abundances in 45 hii regions in the disk of the galaxy ngc 628 .
they found that the o / h abundances have a large intrinsic dispersion of @xmath116 dex .
they posit that this dispersion represents an upper limit to the true dispersion in oxygen abundance at a fixed galactocentric distance and that some of that dispersion is caused by systematic uncertainties in the temperature measurements .
@xcite found that there is substantial scatter of 0.11 dex in the metallicity at any given radius in the disk of the galaxy m33 .
@xcite found that the oxygen abundance gradient in the inner 2 kpc of the m33 disk measured from the detection of the [ oiii]@xmath174363 auroral line displays a scatter of approximately 0.06 dex . a large sample of hii regions assembled from the literature results in a comparably small scatter ( 0.05 0.07 dex ) over the whole optical disk of m33 .
@xcite noted that this dispersion can be explained simply by the measurement uncertainties .
he concluded that no evidence is therefore found for significant azimuthal variations in the present - day metallicity of the interstellar medium of m33 on spatial scales from @xmath117 pc to a few kpc .
@xcite have measured direct gas - phase abundances in 29 individual hii regions in the disk of the galaxy ngc5194 ( m51 ) .
they found an oxygen abundance gradient with very little scatter ( @xmath118 dex ) .
they concluded that most of this scatter can be attributed to random errors and is not caused by an intrinsic dispersion .
@xcite selected and investigated 396 hii regions in the galaxy ngc 6754 .
they found evidence of an azimuthal variation in the oxygen abundance of about 0.05 dex , which may be related to radial migration .
the scatter around the o / h @xmath4 relation obtained here for the califa galaxies is in the range of @xmath119 to @xmath120 dex , which is lower than the scatter in galaxies ( from @xmath121 to @xmath116 dex ) determined in the above quoted studies .
this discrepancy may be caused by the fact that we select the spectra included in the construction and analysis of the metallicity maps , i.e. , we use only spectra where the ratio of the flux to the flux error is higher than three for each of the lines used in the abundance determinations .
can the number of data points influence the obtained value of the global azimuthal asymmetry in a galaxy ?
we consider the galaxy ngc 5520 to examine this problem .
the abundance map of ngc 5520 consists of 2316 spaxels and has a global azimuthal asymmetry of 0.011 dex in case @xmath49 .
we construct a set of 1000 abundance maps of the galaxy ngc 5520 with a reduced number of spaxels , randomly selected from the full map of 2316 spaxels .
the value of the global azimuthal asymmetry was determined for each constructed abundance map .
[ figure : n - as ] shows the value of the global azimuthal asymmetry as a function of the number of the spaxels in the map .
this figure shows that a global azimuthal asymmetry larger than 0.02 dex appears only when the number of spaxels decreases to @xmath122 .
since the number of spaxels in our target galaxies are usually higher than 200 ( see table [ table : property ] ) , the obtained values of the global azimuthal asymmetry in the target galaxies can not be attributed to too small a number of data points in the maps .
thus , there is no significant global azimuthal asymmetry for our sample of the califa galaxies .
the values of the global azimuthal asymmetry are small and can be attributed to uncertainties in the geometrical parameters of our galaxies .
we have found that the radial abundance distribution across the disks of the majority of the galaxies of our sample can be well fitted by a single relation . in a subset of galaxies
a bend in the radial abundance distribution across the disk may exist within the optical @xmath35 radius .
[ figure : u3253ac ] shows ugc 3253 , which is a galaxy with a possible bend in the radial abundance distribution within the optical radius .
the oxygen abundances in the central part of the galaxy are systematically lower as compared to the general radial abundance trend
. possible bends in the radial abundance distributions in other galaxies ( a flattening or lowering in the central part ) are reported in table [ table : property ] ( column 11 ) .
the decrease of the oxygen abundances in the central parts of a number of the califa galaxies was first revealed by @xcite .
the nearly flat distribution in the innermost ( @xmath1230.2@xmath35 ) part of the galaxy ngc 628 was reported by @xcite .
it should be noted that it is difficult to establish the exact value of the bend point due to the scatter in oxygen abundance values at any fixed radius and because the overall decrease of the oxygen abundance in the central part of a galaxy is small , within @xmath124 dex .
therefore , the break radius reported in table [ table : property ] is an indicator that there is evidence for a decrease in the oxygen abundance in the central part of the galaxy rather than the exact value of the bend point .
the arguments for or / and against the existence of the break in the radial abundance gradients in the disks of spiral galaxies were discussed in many papers ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* among others ) there is consensus is that the change in the radial abundance gradient can occur near the isophotal radius of a galaxy .
the depressed gas metallicity in the central part of some califa galaxies noticed by @xcite is confirmed here .
further investigations will strengthen or reject this picture .
thus , there is evidence for a small change in the slope of the gradient ( a flattening or decrease in the central part ) in the disks of a number of galaxies .
we construct maps of the oxygen abundances across the disks of 88 califa galaxies .
the oxygen abundances were determined through the @xmath26 method using califa dr2 spectra .
hence , here we use the empirical metallicity scale defined by hii regions with oxygen abundances derived through the direct method ( @xmath25 method ) .
the position of the center of a galaxy ( coordinates on the plate ) were taken from the califa dr2 .
the galaxy inclination and position angle of the major axis were determined from our analysis of @xmath1 band photometric maps of sdss data release 9 .
the optical radii were determined from the radial surface brightness profiles in the sdss @xmath0 and @xmath1 bands constructed on the basis of the photometric maps and converted to the vega @xmath36 band .
we examine the global azimuthal asymmetry of the abundances in the disks of our target galaxies .
the arithmetic mean of the deviations from the o / h @xmath51 relation @xmath50 for spaxels with azimuthal angles from @xmath49 to @xmath52 and mean deviation @xmath53 for spaxels with azimuthal angles from @xmath52 to @xmath54 are determined for different positions of @xmath49 .
the maximum absolute value of the difference @xmath125 is used to specify the global azimuthal asymmetry in the abundance distribution across the galaxy .
the scatter around the o / h @xmath103 relation for our sample of califa galaxies is in the range of @xmath119 to @xmath120 dex .
there is no significant global azimuthal asymmetry for our califa sample .
the values of the global azimuthal asymmetry are small , less than 0.05 dex for the bulk of target galaxies .
these values can be attributed to the uncertainties in the photometry - map - based geometrical parameters of the galaxies , i.e. , the uncertainties in the photometry - map - based geometrical parameters of a galaxy can make an appreciable ( and possibly dominant ) contribution to the obtained values of the azimuthal asymmetry .
we have found that the radial abundance distribution across the disks of the majority of the galaxies of our sample can be well fitted by a single relation .
however , in eighteen of the galaxies in our sample , the oxygen abundances in the central part of the galaxies are systematically lower as compared to the general radial abundance trend .
although the decrease is rather well defined , its value is small , within @xmath1010.1 dex , and can be questioned taking into account the uncertainties of the abundances in the reference hii regions .
we note that the decrease of the oxygen abundances in the central parts of a number of the califa galaxies was first revealed by @xcite and recently confirmed by @xcite .
we estimated the geometric parameters of our galaxies ( coordinates of the center , inclination and position angle of the major axis ) from the analysis of the abundance map .
the geometrical parameters are determined on the condition that the coefficient of the correlation between oxygen abundance and galactocentric distance is maximized or the scatter around the o / h @xmath4 relation is minimized .
both these conditions result in the same values of the geometrical parameters .
the photometry - map - based and the abundance - map - based geometrical parameters are relatively close to each other for the majority of our galaxies but the discrepancy is large for a few galaxies with a flat radial abundance gradient .
l.s.p . , e.k.g . , and i.a.z .
acknowledge support within the framework of sonderforschungsbereich sfb 881 on `` the milky way system '' ( especially subproject a5 ) , which is funded by the german research foundation ( dfg ) .
+ l.s.p . and i.a.z .
thank for the hospitality of the astronomisches rechen - institut at heidelberg university where part of this investigation was carried out .
+ l.s.p . and i.a.z .
acknowledge the support of the volkswagen foundation under the trilateral partnerships grant no .
90411 . + i.a.z
. also acknowledges the special support by the nasu under the main astronomical observatory grid / gpu computing cluster golowood project . +
this work was partly funded by a subsidy allocated to the kazan federal university for the state assignment in the sphere of scientific activities ( l.s.p . ) .
+ s.f.s .
thanks the conacyt-125180 and dgapa - ia100815 projects for providing him support in this study .
+ this study uses data provided by the calar alto legacy integral field area ( califa ) survey ( http://califa.caha.es/ ) . based on observations collected at the centro astronmico hispano alemn ( caha ) at calar alto , operated jointly by the max - planck - institut fr astronomie and the instituto de astrofsica de andaluca ( csic ) .
+ this research made use of montage , funded by the national aeronautics and space administration s earth science technology office , computational technnologies project , under cooperative agreement number ncc5 - 626 between nasa and the california institute of technology .
the code is maintained by the nasa / ipac infrared science archive .
+ funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. ahn c.p .
et al . , 2012 ,
apjs , 203 , 21 aller , l.h . , 1984 , physics of thermal gaseous nebulae , reidel , dordrecht annibali f. , tosi m. , pasquali a. , aloisi a. , mignoli m. , romano d. , 2015 , astro - ph 1505.05545 asari n. v. , cid fernandes r. , stasiska g. , et al .
2007 , mnras , 381 , 263 baldwin j.a . ,
phillips m.m . ,
& terlevich r. 1981 , pasp , 93 , 5 belfiore f. , maiolino r. , bundy k. , et al . , 2015 , mnras , 449 , 867 berg d.a .
, skillman e.d . ,
marble a.r . , 2012 , apj , 754 , 98 berg d.a . , skillman e.d . , garnett d.r .
, croxall k.v . ,
marble a.r . ,
smith j.d . ,
gordon k. , kennicutt r.c .
, 2013 , apj , 775 , 128 berg d.a .
, skillman e.d .
, croxall k.v . ,
pogge r.w . , moustakas j. , johnson - groh m. , 2015 , apj , 806 , 16 bresolin f. , ryan - weber e. , kennicutt r.c . , goddard q. , 2009 , apj , 695 , 580 bresolin f. , 2011 , apj , 730 , 129 bresolin f. , kennicutt r.c .
, ryan - weber e. , 2012 , apj , 750 , 122 brown j.s .
, ceoxall k.v . ,
pogge r.w . , 2014 ,
apj , 792 , 140 g. , charlot s. , 2003 , mnras , 344 , 1000 j. a. , clayton g. c. , mathis j. s. , 1989 , apj , 345 , 245 r. , mateus a. , sodr l. , stasiska g. , gomes j. m. , 2005 , mnras , 358 , 363 croxall k.v .
, pogge r.w . , berg d. , skillman e.d . , moustakas j. , 2015 , apj , 808 , 42 de vaucouleurs , g. , de vaucouleurs , a. , corvin , h. g. , et al .
1991 , third reference catalog of bright galaxies , new york : springer verlag ( rc3 ) dinerstein , h. l. , 1990 , in thronson h.a . , shull j.m .
2nd teton conf . , the interstellar medium in galaxies , kluwer academic publishers , dordrecht , p. 257
esteban c. , garca - rojas j. , carigi l. , peimbert m. , bresolin f. , lpez - snchez a.r .
, mesa - delgado a. , 2014 , mnras , 443 , 624 r. et al .
, 2014 , a&a , 576 , a135 goddard q. , bresolin f. , kennicutt r.c .
, ryan - weber e.v .
, rosales - ortega f.f . , 2011 , mnras , 412 , 1246 gusev a.s . , pilyugin l.s . , sakhibov f. , dodonov s.n . , ezhkova o.v . ,
khramtsova m.s . , 2012 , mnras , 424 , 1930 haurberg n.c . , rosenberg j. , salzer j.j . , 2013 ,
apj , 765 , 66 haurberg n.c . ,
salzer j.j . , cannon j.m . ,
marshall m.v . , 2015 , apj , 800 , 121 izotov y.i . ,
thuan t.x . ,
lipovetsky v.a . , 1994 ,
apj , 435 , 647 izotov y.i . , stasiska g. , meynet g. , guseva n.g . , thuan t.x .
2006 , a&a , 448 , 955 izotov y.i .
, thuan t.x .
, guseva n.g . , 2012 ,
a&a , 546 , a122 kauffmann g. , heckman t.m .
, tremonti c. , et al .
2003 , mnras , 346 , 1055 kennicutt r.c .
, garnett d.r . , 1996 , apj , 456 , 504 kewley l.j . ,
dopita m.a . ,
sutherland r.s . ,
heisler c.a .
, trevena j. 2001 apj , 556 , 121 kewley l.j . ,
ellison s.l . , 200 , apj , 681 , 1183 li y. , bresolin f. , kennicutt r.c . , 2013 ,
apj , 766 , 17 lpez - snchez .r . ,
esteban c. , 2010 , a&a , 517 , a85 marino r.a .
, gil de paz a. , castillo - morales a. , et al . , 2012 ,
apj , 754 , 61 marino r.a .
, rosales - ortega f.f . ,
sanchez s.f . , et al .
, 2013 , a&a , 559 , 114 marino r.a .
, gil de paz a. , sanchez s.f . , et al . , 2016 , a&a , 585 , a47 martin p. , belley j. , 1996 , apj , 468 , 598 a. , sodr l. , cid fernandes r. , stasiska g. , schoenell w. , gomes j. m. , 2006 , mnras , 370 , 721 moustakas j. , kennicutt r.c . , tremonti c.a . ,
dale d.a . , smith j .- d.t . ,
calzetti d. , 2010 , apjs , 190 , 233 nicholls d.c .
, dopita m.a . ,
sutherland r.s . , jerjen h. , kewley l.j . , basurah h. , 2014 , apj , 786 , 155 osterbrock , d.e . , 1989 , astrophysics of gaseous nebulae and active galactic nuclei , university science books , mill valley , ca pagel b.e.j . ,
simonson e.a . ,
terlevich r.j . , edmunds m.g .
1992 , mnras , 255 , 325 pilyugin l.s . , 2001 , a&a , 373 , 56 pilyugin l.s . , 2003a , a&a , 397 , 109 pilyugin l.s . , 2003 ,
a&a , 399 , 1003 pilyugin l.s .
, vlchez j.m . , contini t. , 2004 , a&a , 425 , 849 pilyugin l.s . , thuan t.x .
, vlchez j.m . , 2006 , mnras , 367 , 1139 pilyugin l.s . , thuan t.x .
, vlchez j.m . , 2007 , mnras , 376 , 353 pilyugin l.s .
, vlchez j.m . , thuan t.x . , 2010 ,
apj , 720 , 1738 pilyugin l.s . ,
grebel e.k . , & mattsson l. 2012 , mnras , 424 , 2316 pilyugin l.s . ,
lara - lpez m.a .
, grebel e.k . , et .
2013 , mnras , 432 , 1217 pilyugin l.s .
, grebel e.k .
, kniazev a.y . , 2014a , aj , 147 , 131 pilyugin l.s . , grebel e.k . , zinchenko i.a . ,
kniazev a.y .
, 2014b , aj , 148 , 134 pilyugin l.s .
, grebel e.k . ,
zinchenko i.a . , 2015 ,
mnras , 450 , 3254 pilyugin l.s .
, grebel e.k . , 2016 , mnras , 457 , 3678 rosales - ortega f.f . , daz a.i . ,
kennicutt r.c . ,
snchez s.f . , 2011 , mnras , 415 , 2439 rosolowsky e. , simon j.d . , 2008 ,
apj , 675 , 1213 rupke d.s.n .
, kewley l.j . , chien l .- h . , 2010 ,
apj , 723 , 1255 snchez , s. f. , kennicutt , r. c. , gil de paz , a. , et al .
2012 , a&a , 538 , a8 snchez s.f .
, rosales - ortega f.f . , iglesias - pramo j. , et al . 2014 , a&a , 563 , 49 snchez s.f .
, galbany l. , prez e. , et al .
2015a , a&a , 573 , 105 snchez s.f .
, prez e. , rosales - ortega f.f . , et al .
2015b , a&a , 574 , 47 snchez - menguiano l. , snchez s.f .
, prez i. , et al .
2016 , a&a , in press , eprint arxiv:1601.01542 searle l. 1971 , apj , 168 , 327 singh r. , van de ven g. , jahnke k. , et al .
2013 , a&a , 558 , 43 smith h.e .
1975 , apj , 199 , 591 skillman e.d .
, salzer j.j . ,
berg d.a . , et al . , 2013 ,
aj , 146 , 3 storey p.j . , zeippen c.j . , 2000 ,
mnras , 312 , 813 stasiska g. , cid fernandes r. , mateus a. , sodr l. , asari n.v . 2006 , mnras , 371 , 972 thuan t.x .
, pilyugin l.s . , &
zinchenko i.a .
2010 , apj , 712 , 1029 vila - costas m.b . , edmunds m.g .
1992 , mnras , 259 , 121 vilchez j.m .
, pagel b.e.j . ,
daz a.i . , terlevich e. , edmunds m.g . , 1988 , mnras , 235 , 633 vilchez j.m . , esteban c. , 1996 , mnras , 280 , 720 vogt f.p.a . ,
dopita m.a . ,
kewley l.j . ,
sutherland r.s . , scharwaechter j. , basurah h.m . , ali a. , amer m.a . , 2014 ,
apj , 793 , 127 c. j. et al .
, 2014 , a&a , 569 , a1 zaritsky d. , kennicutt r.c .
, huchra j.p . , 1994 , apj , 420 , 87 zinchenko i.a . , kniazev a.y . , grebel e.k . , pilyugin l.s . , 2015 , a&a , 582 , a35 zurita a. , bresolin f. , 2012 , mnras , 427 , 1463
table [ table : sample ] lists the general characteristics of each galaxy .
+ column 1 gives its name .
we have used the most widely used name for each galaxy .
the galaxies are listed in the order of the name category , with the following categories in descending order : + ngc new general catalogue , + ic index catalogue , + ugc
uppsala general catalog of galaxies , + pgc catalogue of principal galaxies . + columns 2 and 3 report the morphological type of the galaxy and the morphological type code @xmath73 from leda .
+ column 4 lists the isophotal radius @xmath35 in arcmin of each galaxy .
we determined the isophotal radius from the photometric maps in the sdss @xmath0 and @xmath1 bands .
+ column 5 gives the isophotal radius in kpc , estimated from the data in columns 4 and 6 .
+ column 6 reports the ned distances using flow corrections for virgo , the great attractor , and shapley supercluster infall .
+ table [ table : property ] lists the obtained properties of the oxygen abundance distributions in the disks of our sample of califa galaxies .
the inferred properties of each galaxy are given in two consecutive rows .
in the first row , we report the geometrical parameters of the galaxies obtained from the analysis of the photometric map and the properties of the abundance distribution in the disk of the galaxies determined with deprojected galactocentric distances of spaxels for those geometrical parameters ( case @xmath49 in the text and figures ) . in the second row ,
we report the geometrical parameters of the galaxies obtained from the analysis of the abundance map and the properties of the abundance distribution in the disk of the galaxy determined with deprojected galactocentric distances of spaxels for those geometrical parameters ( case @xmath26 in the text ) .
column 1 gives the name of the galaxy .
we have used the most widely used name for each galaxy .
the galaxies are listed in the same order as in table [ table : sample ] . + columns 2 and 3 report the position of the galaxy center on the califa exposure ( the @xmath37 and @xmath38 coordinates in spaxels ) .
+ columns 4 and 5 give the galaxies inclination and the position angle of the major axes .
+ columns 6 and 7 list the extrapolated central 12 + log(o / h)@xmath126 oxygen abundance and the radial oxygen abundance gradient expressed in terms of dex @xmath46 .
+ column 8 reports the scatter of oxygen abundances around the general radial oxygen abundance trend within the optical @xmath35 radius of a galaxy . +
columns 9 and 10 give the global azimuthal asymmetry ( maximum difference between the arithmetic means of the deviations from the o / h @xmath127 relation for the opposite semicircle sectors ) and the position of the dividing line ( see panels @xmath74 in fig . [
figure : n6132ac ] ) . + column
11 lists the break radius if a bend in the radial abundance distribution exists .
+ column 12 provides the number of the points in the abundance map for a galaxy . | we construct maps of the oxygen abundance distribution across the disks of 88 galaxies using califa data release 2 ( dr2 ) spectra .
the position of the center of a galaxy ( coordinates on the plate ) were also taken from the califa dr2 .
the galaxy inclination , the position angle of the major axis , and the optical radius were determined from the analysis of the surface brightnesses in the sdss @xmath0 and @xmath1 bands of the photometric maps of sdss data release 9 .
we explore the global azimuthal abundance asymmetry in the disks of the califa galaxies and the presence of a break in the radial oxygen abundance distribution .
we found that there is no significant global azimuthal asymmetry for our sample of galaxies , i.e. , the asymmetry is small , usually lower than 0.05 dex .
the scatter in oxygen abundances around the abundance gradient has a comparable value , @xmath2 dex .
a significant ( possibly dominant ) fraction of the asymmetry can be attributed to the uncertainties in the geometrical parameters of these galaxies .
there is evidence for a flattening of the radial abundance gradient in the central part of 18 galaxies .
we also estimated the geometric parameters ( coordinates of the center , the galaxy inclination and the position angle of the major axis ) of our galaxies from the analysis of the abundance map .
the photometry - map - based and the abundance - map - based geometrical parameters are relatively close to each other for the majority of the galaxies but the discrepancy is large for a few galaxies with a flat radial abundance gradient .
galaxies : abundances ism : abundances hii regions |
consider solving the elliptic partial differential equation@xmath6 with the dirichlet boundary condition @xmath7 assume @xmath2 .
assume @xmath0 is an open , simply - connected , and bounded region in @xmath1 , and assume that its boundary @xmath3 is several times continuously differentiable .
similarly , assume the functions @xmath8
@xmath9 @xmath10 are several times continuously differentiable over @xmath11 . as usual , assume the matrix @xmath12 $ ] is symmetric , and also assume it satisfies the strong ellipticity condition,@xmath13 with @xmath14 .
also assume @xmath15 , @xmath16 .
in [ spec_method ] we consider the special region @xmath17 , the open unit ball in @xmath1 .
we define a galerkin method for ( [ e1])-([e2 ] ) with a special finite - dimensional subspace of polynomials , and we give an error analysis that shows rapid convergence of the method . in [ elliptic_transf ]
we discuss the use of a transformation from a general region @xmath0 to the unit ball @xmath5 , showing that the transformed equation is again elliptic over @xmath5 .
implementation issues are discussed in [ implementation ] for problems in @xmath18 and @xmath19 .
we conclude in [ num_example ] with numerical examples in @xmath18 and @xmath19 .
the methods of this paper generalize to the equation@xmath20 which contains first order derivative terms , provided the operator @xmath21 is strongly elliptic . to do so , use the results of brenner and scott @xcite , combined with the methods of the present paper .
we have chosen to restrict our work to the more standard symmetric problem ( [ e1 ] ) .
there is a rich literature on spectral methods for solving partial differential equations . from the more recent literature ,
we cite @xcite , @xcite , and @xcite .
their bibliographies contain references to earlier papers on spectral methods .
our approach is somewhat different than the standard approaches , as we are converting the partial differential equation to an equivalent problem on the unit disk or unit ball , and in the process we are required to work with a more complicated equation .
our approach is reminiscent of the use of conformal mappings for planar problems .
conformal mappings can be used with our approach when working on planar problems , although having a conformal mapping is not necessary .
the dirichlet problem ( [ e1])-([e2 ] ) has the following variational reformulation : find @xmath22 such that@xmath23{r}{\displaystyle\int\nolimits_{\omega } } \left [ { \displaystyle\sum\limits_{i , j=1}^{d } } a_{i , j}(\mathbf{s})\dfrac{\partial u(\mathbf{s})}{\partial s_{j}}\dfrac{\partial v(\mathbf{s})}{\partial s_{i}}+\gamma(\mathbf{s})u(\mathbf{s})v(\mathbf{s})\right ] d\mathbf{s}\quad\quad\quad\quad\medskip\\ = { \displaystyle\int_{\omega } } f(\mathbf{s})v(\mathbf{s})\,d\mathbf{s},\quad\quad\forall v\in h_{0}^{1}\left ( \omega\right ) \end{array } \label{e20}\ ] ] we define a spectral galerkin method in this section for the special region @xmath17 . in [ elliptic_transf ] we discuss the transformation of ( [ e1 ] ) from a general @xmath11 to an equivalent equation over the unit ball @xmath24 , a transformation that retains the ellipticity of the problem . in the remainder of this section
, we replace @xmath0 with @xmath5 .
introduce the bilinear form@xmath25
d\mathbf{x},\quad v , w\in h_{0}^{1}\left ( b\right ) \label{e21}\ ] ] and the bounded linear functional@xmath26 the variational problem ( [ e20 ] ) can now be written as follows : find @xmath27 for which@xmath28 it is straightforward to show @xmath29 is bounded,@xmath23{c}\left\vert \mathcal{a}\left ( v , w\right ) \right\vert \leq c_{\mathcal{a}}\left\vert v\right\vert _ { 1}\left\vert w\right\vert _ { 1}\medskip\\ c_{\mathcal{a}}=\max\limits_{\mathbf{x}\in\overline{b}}\left\vert a(\mathbf{x})\right\vert _ { 2}+\left\vert \gamma\right\vert
_ { \infty}\end{array}\ ] ] with @xmath30 the norm of @xmath31 and @xmath32 the matrix 2-norm of the matrix @xmath33 .
in addition , we assume @xmath34 this follows generally from ( [ e3 ] ) and the size of the function @xmath35 over @xmath24 ; when @xmath36 , @xmath37 . under standard assumptions on @xmath29 ,
including the strong ellipticity in ( [ e23 ] ) , the lax - milgram theorem implies the existence of a unique solution @xmath38 to ( [ e22 ] ) with @xmath39 denote by @xmath40 the space of polynomials in @xmath41 variables that are of degree @xmath42 : @xmath43 if it has the form@xmath44 with @xmath45 a multi - integer , @xmath46 , and @xmath47 .
let @xmath48 denote our approximation subspace,@xmath49 with @xmath50 .
the subspaces @xmath40 and @xmath48 have dimension@xmath51 [ lemma1]let @xmath52 denote the laplacian operator in @xmath1 .
then@xmath53 for a short proof ,
see @xcite . the galerkin method for obtaining an approximate solution to ( [ e22 ] ) is as follows : find @xmath54 for which@xmath55 the lax - milgram theorem ( cf .
@xcite , brenner - scott ) implies the existence of @xmath56 for all @xmath57 . for the error in this galerkin method , cea s lemma ( cf .
@xcite , @xcite ) implies the convergence of @xmath56 to @xmath38 , and moreover,@xmath58 it remains to bound the best approximation error on the right side of this inequality .
given an arbitrary @xmath59 , define @xmath60 .
then @xmath61 and @xmath38 satisfies the boundary value problem @xmath62 it follows that @xmath63 for @xmath18 and @xmath19 , the green s function is defined as follows.@xmath23{cl}d=2 : & g(p , q)=\dfrac{1}{2\pi}\log\dfrac{\left\vert p - q\right\vert } { \left\vert \mathcal{t}(p)-q\right\vert } , \quad\quad\medskip\\ d=3 : & g(p , q)=-\dfrac{1}{4\pi}\left\ { \dfrac{1}{\left\vert p - q\right\vert } -\dfrac{1}{\left\vert p\right\vert } \dfrac{1}{\left\vert \mathcal{t}(p)-q\right\vert } \right\ } \end{array } \label{e42}\ ] ] for @xmath64 @xmath65 @xmath66 .
@xmath67 denotes the inverse point for @xmath68 with respect to the unit sphere @xmath69,@xmath70 differentiate ( [ e41 ] ) to obtain@xmath71 w(q)\,dq,\quad\quad p\in\overline{b } \label{e43}\ ] ] note that @xmath72 is absolutely integrable over @xmath24 , for all @xmath66 .
let @xmath73 be an approximation of @xmath74 , say in the norm of either @xmath75 or @xmath76 , and let @xmath77 we can show @xmath78 .
this follows from lemma [ lemma1 ] and noting that the mapping in ( [ e41 ] ) is the inverse of ( [ e28 ] )
. then we have @xmath79 \,dq,\quad\quad p\in\overline{b}\]]@xmath80 = \int_{b}\left
[ \nabla _ { p}g(p , q)\right ] \left [ w(q)-w_{n}(q)\right ] \,dq,\quad\quad p\in \overline{b}\ ] ] the integral operators on the right side are weakly singular compact integral operators on @xmath76 to @xmath76 ( * ? ? ?
this implies@xmath81 by letting @xmath82 be the orthogonal projection of @xmath74 into @xmath40 , the right side will go to zero since the polynomials are dense in @xmath83 . in turn , this implies convergence in the @xmath84 norm for the right side in ( [ e31 ] ) provided @xmath59 .
the result @xmath85 can be extended to any @xmath86 .
it basically follows from the denseness of @xmath87 in @xmath88 .
let @xmath86 .
we need to find a sequence of polynomials @xmath89 for which @xmath90 .
we know @xmath87 is dense in @xmath88 . given any @xmath91 , choose @xmath92 with @xmath93 . then choose a polynomial @xmath94 for which we have the corresponding polynomial @xmath95 satisfying @xmath96 , based on ( [ e47 ] ) .
[ regarding the earlier notation , @xmath95 need not be of degree @xmath97 .
] then @xmath98 . to obtain orders of convergence , use ( [ e47 ] ) and results on best multivariate polynomial approximation over the unit disk .
for example , use results of ragozin ( * ? ? ? * thm 3.4 ) or yuan xu @xcite .
from @xcite we have the following .
[ theorem1]assume @xmath99 for some @xmath91 .
then there is a polynomial @xmath78 for which@xmath100 in this , @xmath101@xmath102@xmath103
consider the differential operator@xmath104 which satisfies the ellipticity condition ( [ e3 ] ) with @xmath14 . the operator @xmath105 is said to be elliptic on @xmath106 .
we want to transform the operator @xmath105 to one acting on functions @xmath107 with @xmath5 the unit ball in @xmath1
. assume the existence of a twice - differentiable mapping@xmath108 and let @xmath109 .
let@xmath110 _ { i , j=1}^{d},\quad\quad\mathbf{x}\in\overline{b}\subseteq\mathbb{r}^{d}\ ] ] denote the jacobian of the transformation . as usual
we assume @xmath111 is nonsingular on @xmath24 , and furthermore@xmath112 similarly , let @xmath113 denote the jacobian of @xmath114 over @xmath11 . by differentiating the components of the equation@xmath115 we obtain@xmath116
this general approach is reminiscent of the coordinate transformations in ( * ? ? ?
2 ) in which the mapping function is used in generating a mesh on a region @xmath0 . for @xmath117 , let@xmath118 and conversely,@xmath119 then @xmath120 \nabla_{\mathbf{s}}v\end{aligned}\ ] ] with the gradient @xmath121 a column vector evaluated at @xmath122 . more concisely,@xmath123 similarly,@xmath124 [ theorem2]assume the transformation @xmath125 satisfies ( [ e61 ] ) or ( [ e64 ] ) .
then for @xmath126,@xmath127@xmath128 _ { i , j=1}^{d}\nonumber\end{aligned}\ ] ] let @xmath129 .
then @xmath130 on the other hand , using integration by parts we have@xmath131 using ( [ e67]),@xmath23{l}{\displaystyle\sum\limits_{i , j=1}^{d } } a_{i , j}(\phi\left ( \mathbf{x}\right ) ) \dfrac{\partial v(\phi\left ( \mathbf{x}\right ) ) } { \partial s_{j}}\dfrac{\partial w(\phi\left ( \mathbf{x}\right ) ) } { \partial s_{i}}=\left [ \nabla_{\mathbf{s}}w(\phi\left ( \mathbf{x}\right ) ) \right ] ^{\text{t}}a\left ( \phi\left ( \mathbf{x}\right ) \right ) \left [ \nabla_{\mathbf{s}}v(\phi\left ( \mathbf{x}\right ) ) \right ] \medskip\\ \quad\quad\quad\quad\quad\quad=\left [ \nabla_{\mathbf{x}}\widetilde { w}(\mathbf{x})\right ] ^{\text{t}}k(\phi\left ( \mathbf{x}\right ) ) a\left ( \phi\left ( \mathbf{x}\right ) \right ) k(\phi\left ( \mathbf{x}\right ) ) ^{\text{t}}\left [ \nabla_{\mathbf{x}}\widetilde{v}(\mathbf{x})\right ] \medskip\\ \quad\quad\quad\quad\quad\quad=\left [ \nabla_{\mathbf{x}}\widetilde { w}(\mathbf{x})\right ] ^{\text{t}}\widetilde{a}\left ( \mathbf{x}\right ) \left [ \nabla_{\mathbf{x}}\widetilde{v}(\mathbf{x})\right ] \end{array}\ ] ] using this to continue ( [ e71]),@xmath132 ^{\text{t}}\widetilde{a}\left ( \mathbf{x}\right ) \left [ \nabla_{\mathbf{x}}\widetilde{v}(\mathbf{x})\right ] \det\left ( j(\mathbf{x})\right ) \,d\mathbf{x}\nonumber\\ & = \int_{\omega}\sum_{i , j=1}^{d}\widetilde{a}_{i , j}(\mathbf{s})\frac { \partial\widetilde{v}(\mathbf{x})}{\partial x_{j}}\frac{\partial\widetilde { w}(\mathbf{x})}{\partial x_{i}}\det\left ( j(\mathbf{x})\right ) \,d\mathbf{x}\nonumber\\ & = -\int_{\omega}\sum_{i , j=1}^{d}\frac{\partial}{\partial x_{i}}\left ( \det\left ( j(\mathbf{x})\right ) \widetilde{a}_{i , j}(\mathbf{x})\frac{\partial\widetilde{v}(\mathbf{x})}{\partial x_{j}}\right ) \widetilde{w}(\mathbf{x})\,d\mathbf{x } \label{e73}\ ] ] comparing ( [ e70 ] ) and ( [ e73 ] ) , and noting that @xmath133 is arbitrary , we have@xmath134 which proves ( [ e68 ] ) . with this transformation , we can solve the dirichlet problem over a general region @xmath0 by transforming it to an equivalent problem over the unit ball @xmath5 .
we can apply the galerkin method to ( [ e1 ] ) by means of the transformation ( [ e68 ] ) .
we convert ( [ e1 ] ) to the equation@xmath23{r}-{\displaystyle\sum\limits_{i , j=1}^{d } } \dfrac{\partial}{\partial x_{i}}\left ( \det\left ( j(\mathbf{x})\right ) \widetilde{a}_{i , j}(\mathbf{x})\dfrac{\partial\widetilde{v}(\mathbf{x})}{\partial x_{j}}\right ) + \det\left ( j(\mathbf{x})\right ) \gamma ( \phi\left ( \mathbf{x}\right ) ) \widetilde{v}(\mathbf{x})\quad\quad\quad\\ = \det\left ( j(\mathbf{x})\right ) f\left ( \phi\left ( \mathbf{x}\right ) \right ) \end{array } \label{e74}\ ] ] this system is also strongly elliptic .
[ theorem3]assume @xmath135 , @xmath136 , satisfies ( [ e3 ] ) ; and without loss of generality , assume @xmath137 recall @xmath138 as defined by ( [ e69 ] ) .
then @xmath138 satisfies the strong ellipticity condition@xmath139 with @xmath140 the smallest eigenvalue of @xmath141 ( which equals the reciprocal of the largest eigenvalue of @xmath142 ) .
@xmath143 in addition , @xmath144@xmath145 with @xmath146 the smallest eigenvalue of @xmath147 ; cf .
consider the implementation of the galerkin method of [ spec_method ] for the elliptic problem ( [ e22 ] ) over the unit ball @xmath5 .
we are to find the function @xmath54 satisfying ( [ e30 ] ) .
to do so , we begin by selecting an orthonormal basis for @xmath40 , denoting it by @xmath149 , with @xmath150 . choosing
an orthonormal basis is an attempt to have the linear system in ( [ e30 ] ) be better conditioned .
next , let @xmath151 to form a basis for @xmath48 .
we seek@xmath152 then ( [ e30 ] ) becomes@xmath23{r}{\displaystyle\sum\limits_{k=1}^{n_{n } } } \alpha_{k}{\displaystyle\int_{b } } \left [ { \displaystyle\sum\limits_{i , j=1}^{d } } a_{i , j}(\mathbf{x})\dfrac{\partial\psi_{k}(\mathbf{x})}{\partial x_{j}}\dfrac{\partial\psi_{\ell}(\mathbf{x})}{\partial x_{i}}+\gamma(\mathbf{x})\psi_{k}(\mathbf{x})\psi_{\ell}(\mathbf{x})\right ] d\mathbf{x}\quad \quad\quad\medskip\\ = { \displaystyle\int_{b } } f\left ( \mathbf{x}\right ) \psi_{\ell}\left ( \mathbf{x}\right ) \,d\mathbf{x},\quad\quad\ell=1,\dots ,
n \end{array } \label{e78}\ ] ] we need to calculate the orthonormal polynomials and their first partial derivatives ; and we also need to approximate the integrals in the linear system . for an introduction to the topic of multivariate orthogonal polynomials , see dunkl and xu @xcite and xu @xcite . for multivariate quadrature over the unit ball in @xmath1 , see stroud @xcite .
the dimension of @xmath40 is @xmath153 for notation , we replace @xmath154 with @xmath155
. how do we choose the orthonormal basis @xmath156 for @xmath40 ?
unlike the situation for the single variable case , there are many possible orthonormal bases over @xmath157 , the unit disk in @xmath18 .
we have chosen one that is particularly convenient for our computations .
these are the `` ridge polynomials '' introduced by logan and shepp @xcite for solving an image reconstruction problem .
we summarize here the results needed for our work .
let @xmath158 the polynomials of degree @xmath57 that are orthogonal to all elements of @xmath159 .
then the dimension of @xmath160 is @xmath161 ; moreover,@xmath162 it is standard to construct orthonormal bases of each @xmath160 and to then combine them to form an orthonormal basis of @xmath40 using the latter decomposition . as an orthonormal basis of @xmath160 we use@xmath163 for @xmath164 .
the function @xmath165 is the chebyshev polynomial of the second kind of degree @xmath57:@xmath166 the family @xmath167 is an orthonormal basis of @xmath160 . as a basis of @xmath40 , we order @xmath168 lexicographically based on the ordering in ( [ e101 ] ) and ( [ e100]):@xmath169 returning to ( [ e75 ] ) , we define@xmath170 to calculate the first order partial derivatives of @xmath171 , we need @xmath172 .
the values of @xmath173 and @xmath172 are evaluated using the standard triple recursion relations@xmath174 for the numerical approximation of the integrals in ( [ e78 ] ) , which are over @xmath5 being the unit disk , we use the formula@xmath175 here the numbers @xmath176 are the weights of the @xmath177-point gauss - legendre quadrature formula on @xmath178 $ ] . note that @xmath179 for all single - variable polynomials @xmath180 with @xmath181 . the formula ( [ e106 ] )
uses the trapezoidal rule with @xmath182 subdivisions for the integration over @xmath24 in the azimuthal variable .
this quadrature is exact for all polynomials @xmath183 .
this formula is also the basis of the hyperinterpolation formula discussed in @xcite . in @xmath19 ,
the dimension of @xmath40 is @xmath184 here we choose orthonormal polynomials on the unit ball as described in @xcite , @xmath185 here @xmath186 is a constant , and @xmath187 , @xmath188 , are the normalized jabobi polynomials which are orthonormal on @xmath189 $ ] with respect to the inner product @xmath190 see for example @xcite , @xcite .
the functions @xmath191 are spherical harmonic functions , and they are given in spherical coordinates by @xmath192{ll}\cos(\frac{\beta}{2}\phi)t_{k}^{\frac{\beta}{2}}(\cos\theta),\medskip & \beta\text { even}\\ \sin(\frac{\beta+1}{2}\phi)t_{k}^{\frac{\beta+1}{2}}(\cos\theta ) , & \beta\text { odd}\end{array } \right.\ ] ] the constant @xmath193 is chosen in such a way that the functions are orthonormal on the unit sphere @xmath194 in @xmath19 : @xmath195 the functions @xmath196 are the associated legendre polynomials , see @xcite , @xcite . according to ( [ e75 ] ) we define the basis for our space of trial functions by @xmath197 and we can order the basis lexicographically . to calculate all of the above functions we can use recursive algorithms similar to the one used for the chebyshev polynomials .
these algorithms also allow the calculation of the derivatives of each of these functions , see @xcite , @xcite for the numerical approximation of the integrals in ( [ e78 ] ) we use a quadrature formula for the unit ball @xmath5 @xmath198\medskip\nonumber\\ q_{q}[g ] & : = \sum_{i=1}^{2q}\sum_{j=1}^{q}\sum_{k=1}^{q}\frac{\pi}{q}\,\omega_{j}\,\nu_{k}\widetilde{g}\left ( \frac{\zeta_{k}+1}{2},\frac { \pi\;i}{2q},\arccos(\xi_{j})\right ) \label{e1004}\ ] ] here @xmath199 is the representation of @xmath200 in spherical coordinates . for the @xmath201 integration
we use the trapezoidal rule , because the function is @xmath202periodic in @xmath201 .
for the @xmath203 direction we use the transformation @xmath204 where the @xmath205 and @xmath206 are the weights and the nodes of the gauss quadrature with @xmath207 nodes on @xmath189 $ ] with respect to the inner product @xmath208 the weights and nodes also depend on @xmath207 but we omit this index . for the @xmath209 direction we use the transformation @xmath210 where the @xmath211 and @xmath212 are the nodes and weights for the gauss
legendre quadrature on @xmath189 $ ] . for more information on this quadrature rule on the unit ball in @xmath19 ,
see @xcite .
finally we need the gradient in cartesian coordinates to approximate the integral in ( [ e78 ] ) , but the function @xmath213 in ( [ e1001 ] ) is given in spherical coordinates .
here we simply use the chain rule , with @xmath214 , @xmath215 and similarly for @xmath216 and @xmath217 .
[ tb ] g8_fig.eps [ c]|c|c|c|c||c|c|c|c|@xmath57 & @xmath218 & @xmath219 & _ cond _ & @xmath57 & @xmath218 & @xmath219 & _ cond _
+ @xmath220 & @xmath221 & @xmath222 & @xmath223 & @xmath224 & @xmath225 & @xmath226 & @xmath227 + @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath232 & @xmath233 & @xmath234 & @xmath235 + @xmath236 & @xmath232 & @xmath237 & @xmath238 & @xmath239 & @xmath240 & @xmath241 & @xmath242 + @xmath243 & @xmath244 & @xmath245 & @xmath246 & @xmath247 & @xmath248 & @xmath249 & @xmath250 + @xmath221 & @xmath251 & @xmath252 & @xmath253 & @xmath254 & @xmath255 & @xmath256 & @xmath257 + @xmath258 & @xmath259 & @xmath260 & @xmath261 & @xmath262 & @xmath263 & @xmath264 & @xmath265 + @xmath266 & @xmath267 & @xmath268 & @xmath269 & @xmath270 & @xmath271 & @xmath272 & @xmath273 + @xmath274 & @xmath275 & @xmath276 & @xmath277 & @xmath244 & @xmath278 & @xmath279 & @xmath280 + @xmath229 & @xmath281 & @xmath282 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & @xmath287 + @xmath288 & @xmath289 & @xmath290 & @xmath291 & @xmath292 & @xmath293 & @xmath294 & @xmath295 + @xmath296 & @xmath297 & @xmath298 & @xmath299 & @xmath300 & @xmath301 & @xmath302 & @xmath303 + @xmath304 & @xmath305 & @xmath306 & @xmath307 & @xmath308 & @xmath309 & @xmath310 & @xmath311 +
our programs are written in matlab and can be obtained from the authors .
our transformations have been so chosen that we can invert explicitly the mapping @xmath125 , to be able to better construct our test examples .
this is not needed when applying the method ; but it simplified the construction of our test cases .
the elliptic equation being solved is @xmath312 which corresponds to choosing @xmath313
. then we need to calculate@xmath23{c}\widetilde{a}\left ( \mathbf{x}\right ) = k\left ( \phi\left ( \mathbf{x}\right ) \right ) k\left ( \phi\left ( \mathbf{x}\right ) \right ) ^{\text{t}}\smallskip\\ k\left ( \phi\left ( \mathbf{x}\right ) \right ) = j\left ( \mathbf{x}\right ) ^{-1}\end{array } \label{e131}\ ] ] for our variables , we replace @xmath314 with @xmath155 , and we replace @xmath16 with @xmath315 .
define the mapping @xmath316 by @xmath317,@xmath23{l}s = x - y+ax^{2}\\ t = x+y \end{array } \label{e132}\ ] ] with @xmath318 .
it can be shown that @xmath125 is a 1 - 1 mapping from the unit disk @xmath24 .
in particular , the inverse mapping @xmath319 is given by@xmath23{l}x=\dfrac{1}{a}\left [ -1+\sqrt{1+a\left ( s+t\right ) } \right ] \medskip\\ y=\dfrac{1}{a}\left [ at-\left ( -1+\sqrt{1+a\left ( s+t\right ) } \right ) \right ] \end{array } \label{e133}\ ] ] in figure [ fig1 ] , we give the images in @xmath11 of the circles @xmath320 , @xmath321 and the azimuthal lines @xmath322 , @xmath323 . the following information is needed when implementing the transformation from @xmath324 on @xmath0 to a new equation on @xmath5:@xmath325{cc}1 + 2ax & -1\\ 1 & 1 \end{array } \right)\]]@xmath326@xmath327{cc}1 & 1\\ -1 & 1 + 2ax \end{array } \right)\]]@xmath328{cc}1 & ax\\ ax & 2a^{2}x^{2}+2ax+1 \end{array } \right)\]]@xmath329{cc}1 & ax\\ ax & 2a^{2}x^{2}+2ax+1 \end{array } \right)\ ] ] the latter are the coefficients for the transformed elliptic operator over @xmath330 , given in ( [ e68 ] ) .
we give numerical results for solving the equation@xmath331 as a test case , we choose@xmath332 with @xmath155 replaced using ( [ e133 ] ) .
the solution is pictured in figure [ true_soln ] . to find @xmath333 , we use ( [ e136 ] ) and ( [ e138 ] ) .
we use the domain parameter @xmath334 , with @xmath0 pictured in figure [ fig1 ] .
numerical results are given in table [ table1 ] .
the integrations in ( [ e78 ] ) were performed with ( [ e106 ] ) ; and the integration parameter @xmath207 ranged from @xmath229 to @xmath335 .
we give the condition numbers of the linear system ( [ e78 ] ) as produced in matlab .
to calculate the error , we evaluate the the numerical solution and the error on the grid @xmath336 the results are shown graphically in figure [ error_planar ] .
the use of a semi - log scale demonstrates the exponential convergence of the method as the degree increases . to examine experimentally the behaviour of the condition numbers for the linear system ( [ e78 ] ) , we have graphed the condition numbers from table [ table1 ] in figure [ cond_planar ] .
note that we are graphing @xmath218 vs. the condition number of the associated linear system .
the graph seems to indicate that the condition number of the system ( [ e78 ] ) is directly proportional to the order of the system , with the order given in ( [ e79 ] ) .
here we define the mapping @xmath316 by @xmath337 , @xmath338 @xmath339 , which is an extension of the mapping defined in ( [ e132 ] ) . the inverse mapping @xmath340 is given by @xmath341 \\
y & = \dfrac{1}{a}\left [ at-\left ( -1+\sqrt{1+a\left ( s+t\right ) } \right ) \right ] \\ z & = \dfrac{1}{b}\left [ -1+\sqrt{1+bu}\right]\end{aligned}\ ] ] in figure [ fig5 ] we show the image of the surface of @xmath24 under @xmath125 . as in the planar case
, we also need @xmath342{ccc}1 + 2ax & -1 & 0\\ 1 & 1 & 0\\ 0 & 0 & 2 + 2bz \end{array } \right)\]]@xmath343 and @xmath344{ccc}\dfrac{1}{2(1+ax)^{2 } } & \dfrac{ax}{2(1+ax)^{2 } } & 0\\ \dfrac{ax}{2(1+ax)^{2 } } & \dfrac{1+ax+2a^{2}x^{2}}{2(1+ax)^{2 } } & 0\\ 0 & 0 & \dfrac{1}{4(1+bz)^{2}}\end{array } \right)\end{aligned}\ ] ] again , these are the coefficients for the second order term for the transformed equation on @xmath24 , given in ( [ e68 ] ) .
we give numerical results for solving the equation @xmath345 and for our test case we choose @xmath346 where the second term guarantees the dirichlet boundary conditions on @xmath11 .
the integrations in ( [ e78 ] ) were performed with ( [ e1004 ] ) ; and the integration parameter @xmath207 was chosen as @xmath347 .
numerical experiments indicate that a larger @xmath207 does not change the results significantly .
the condition numbers for the system ( [ e78 ] ) were again calculated with matlab .
an estimation for the error in the maximum norm was calculated on the grid given by @xmath348{r}x_{i , j , k}\\ y_{i , j , k}\\ z_{i , j , k}\end{array } \right ) = \left ( \begin{array } [ c]{l}\frac{i}{21}\sin\left ( \frac{k}{21}\pi\right ) \cos\left ( \frac{2j}{20}\pi\right ) \\[2mm]\frac{i}{21}\sin\left ( \frac{k}{21}\pi\right ) \sin\left ( \frac{2j}{20}\pi\right ) \\[2mm]\frac{i}{21}\cos\left ( \frac{k}{21}\pi\right ) \end{array } \right ) , \quad i , k=1,\ldots,20,\quad j=1,\ldots,40.\ ] ] the error for the galerkin method is shown in figure [ error_spatial ] and the development of the condition number is shown in figure [ cond_spatial ] .
again the numerical experiment seems to indicate an exponential convergence of the method and a linear growth of the condition numbers with respect to the number of degrees of freedom @xmath218 of the linear system ( [ e78 ] ) .
additional remarks .
we present and study a spectral method for the neumann problem@xmath349 in a forthcoming paper .
we are also investigating the behaviour of the condition number for the linear system ( [ e78 ] ) associated with our spectral method , attempting to prove that it has size @xmath350 , consistent with the numbers shown in tables [ table1 ] and [ table2 ] .
our earlier numerical examples use given @xmath351 chosen to be nontrivial and illustrative .
in general , however , when given a smooth mapping@xmath352 it may not be clear as to how to extend @xmath353 to @xmath125 over @xmath5 . in some cases
there is an obvious choice , as when @xmath0 is an ellipsoid .
we are investigating schemes to produce continuously differentiable extensions @xmath125 which satisfy@xmath354 and for which @xmath111 is easily computable .
o. hansen , k. atkinson , and d. chien , on the norm of the hyperinterpolation operator on the unit disk and its use for the solution of the nonlinear poisson equation , _
i m a j. numer .
_ , doi : 10.1093/imanum / drm052 , 2008 . | let @xmath0 be an open , simply connected , and bounded region in @xmath1 , @xmath2 , and assume its boundary @xmath3 is smooth .
consider solving an elliptic partial differential equation @xmath4 over @xmath0 with zero dirichlet boundary values .
the problem is converted to an equivalent elliptic problem over the unit ball @xmath5 , and then a spectral method is given that uses a special polynomial basis .
with sufficiently smooth problem parameters , the method is shown to have very fast convergence .
numerical examples illustrate exponential convergence . |
trace inequalities are useful in many areas such as the multiple input multiple output ( mimo ) systems in control theory and communications .
the trace inequality derived in this paper is used to prove the sufficient condition that guarantees the uniqueness of a nash equilibrium in certain mimo communications games with @xmath8 players ( see @xcite for details ) . to be more specific , the diagonally strict concavity condition of rosen @xcite is proven to be satisfied in the scenario of @xcite .
the trace inequality under discussion has already been proven in two special cases : in @xcite for @xmath9 and in @xcite for @xmath10 . in what follows
, we will provide the proof for the general case where @xmath1 is arbitrary .
[ lemma_k ] let @xmath11 be a strictly positive integer , @xmath3 , @xmath4 be any positive definite matrices and @xmath12 , @xmath5 , @xmath6 be any positive semidefinite matrices . then @xmath13 \right\ } \geq 0,\ ] ] where @xmath2 denotes the matrix trace operator .
in order to prove theorem [ lemma_k ] , we will use the following auxiliary results .
@xcite [ lemma_3 ] let @xmath14 , @xmath15 be two positive definite matrices and @xmath16 , @xmath17 be two positive semidefinite matrices and @xmath18 a hermitian matrix .
then @xmath19 the proof can be found in @xcite .
[ lemma_4 ] let @xmath14 , @xmath15 be two positive definite matrices , @xmath16 , @xmath17 , two positive semi - definite matrices .
then @xmath20 to prove this result , let us define @xmath21 as follows : @xmath22\right\}\ ] ] we observe that @xmath21 can be written in two different ways : @xmath23({\mathbf}{a}+{\mathbf}{c})^{-1}\right\ } \\ & = & \mathrm{tr } \left\{({\mathbf}{c}-{\mathbf}{d})({\mathbf}{b}+{\mathbf}{d})^{-1}({\mathbf}{c}-{\mathbf}{d})({\mathbf}{a}+{\mathbf}{c})^{-1}\right\ } + \\ & & \mathrm{tr } \left\{({\mathbf}{c}-{\mathbf}{d})({\mathbf}{b}+{\mathbf}{d})^{-1}({\mathbf}{a}-{\mathbf}{b})({\mathbf}{a}+{\mathbf}{c})^{-1}\right\ } , \end{array}\ ] ] and @xmath24({\mathbf}{b}+{\mathbf}{d})^{-1}\right\ } \\ & = & \mathrm{tr } \left\{({\mathbf}{c}-{\mathbf}{d})({\mathbf}{a}+{\mathbf}{c})^{-1}({\mathbf}{c}-{\mathbf}{d})({\mathbf}{b}+{\mathbf}{d})^{-1}\right\ } + \\ & & \mathrm{tr } \left\{({\mathbf}{c}-{\mathbf}{d})({\mathbf}{a}+{\mathbf}{c})^{-1}({\mathbf}{a}-{\mathbf}{b})({\mathbf}{b}+{\mathbf}{d})^{-1}\right\}. \end{array}\ ] ] using this fact and the commutative property of the trace of a matrix product , the desired result follows directly .
the only thing left to be proven is that @xmath21 is real . to this end , if we denote by @xmath25 , we observe that @xmath26 .
therefore , we obtain that @xmath27 and also @xmath28 .
define , for all @xmath29 , @xmath30 and @xmath31 .
notice that @xmath32 and @xmath33 are positive definite matrices .
we observe that @xmath34 can be re - written recursively as follows : @xmath35 we proceed in two steps .
first , we find a lower bound for @xmath34 and then we prove that this bound is positive .
we start by proving that , for all @xmath1 : @xmath36 to this end we proceed by induction on @xmath11 .
for all @xmath1 , define the proposition : @xmath37 it is easy to check that , for @xmath9 , @xmath38 is true : @xmath39 now , let us assume that @xmath40 is true and prove that @xmath41 is also true .
we have that : the last step of the proof of theorem [ lemma_k ] reduces to showing that the term on the right side of the equality ( c ) is positive .
this can be easily checked by observing that all the terms of the form @xmath47 , with @xmath18 a hermitian matrix , @xmath14 and @xmath15 two positive definite matrices , can be re - written as @xmath48 with @xmath49 .
thus , for all @xmath8 , @xmath50 and the desired result has been proven .
e. v. belmega , s. lasaulce , and m. debbah , `` a trace inequality for positive definite matrices '' , _ journal of inequalities in pure and applied mathematics ( jipam ) _ , vol .
1 , article 5 , pp . 14 , ( 2009 ) .
e. v. belmega , s. lasaulce , m. debbah , m. jungers , and j. dumont , `` power allocation games in wireless networks of multi - antenna terminals '' , _
springer telecommunications systems journal , valuetools special issue _ , issn 1018 - 4864 ( print ) 1572 - 9451 ( online ) , doi 10.1007/s11235 - 010 - 9305 - 3 , ( 2010 ) .
j. rosen , `` existence and uniqueness of equilibrium points for concave n - person games '' , _ econometrica _ , vol .
33 , pp . 520534 , ( 1965 ) .
k. m. abadir and j. r. magnus , `` matrix algebra '' , _ cambridge university press , new york , usa _ , pp . 329 , pp . 338 , ( 2005 ) . | in this note , we provide the generalization of the trace inequality derived in @xcite . more precisely , we prove that @xmath0 \right\ } \geq 0 $ ] , for arbitrary @xmath1 where @xmath2 denotes the matrix trace operator , @xmath3 , @xmath4 are any positive definite matrices and @xmath5 , @xmath6 , for all @xmath7 , are any positive semidefinite matrices . |
since the 11-year period of the solar cycle was discovered by schwabe ( 1844 ) , the long - term activity of the solar cycle has become a hot issue in the field of solar physics .
sunspots are taken as the most famous and typical indicators of solar activity .
sunspot activity has complex spatial and temporal behavior .
carrington ( 1858 , 1859 ) investigated a drift latitude of sunspot motion towards the equator and a variation of the rotation rate of the sun .
hathaway et al .
( 2003 ) examined the drift of the centroid of the sunspot area toward the equator in each hemisphere from 1874 to 2002 and found that the drift rate slows as the centroid approaches the equator .
the distribution of sunspots and flares in a solar cycle exhibits a butterfly diagram " ( carrington , 1858 , maunder 1904 , 1913 ; garcia 1990 ; li et al .
2003 ) and the cycle appears uniformly in both hemispheres on average ( newton & milson 1955 ; white & trotter 1977 ) .
sunspot activity is described by sunspot numbers and sunspot areas in general .
consequently , the sunspot numbers and sunspot areas are used to investigate long - term solar activity ( li et al .
because of the periodicity of sunspot activity ( sunspot numbers and sunspot areas ) , the associated eruptions ( flares , cmes ) appear to have the similar periodicity to that of sunspot activity ( storini & hofer 1999 ) .
in addition , the soft x - ray flares were significantly delayed with respect to sunspot numbers , with a time lag of two to three years between the peak times in solar cycle 21 ( wagner 1988 ; aschwanden 1994 , bromund et al .
. however , wilson ( 1993 ) did not find evidence for a time lag between the maxima of the rates of optical flares and x - ray background flux in solar cycle 22 .
wheatland & litvinenko ( 2001 ) found that there is an average delay of about 6 months considering cycles 21 and 22 together by using correlation analysis .
for the cycle 23 , tan ( 2010 ) compared the sunspot numbers with the distributions of the appearance rate of solar goes flares .
they found that there is a time lag between the maximum value of sunspot numbers and the maximum values of the annual numbers of c - class , m - class , and x - class flares .
the annual averaged relative sunspot number reaches its maximum in about 2000 and the annual numbers of c - class , m - class , and x - class flares reach their maxima in about 2001 , 2001 , and 2002 respectively .
temmer et al .
( 2003 ) found that there is a characteristic time lag between flare activity and sunspot activity in the range of 10 - 15 months for solar cycles 19 , 21 , and 23 by using the number of the flares in each month . in order to investigate all kinds of long - term flare activity , the flare index calculated by t. atac and a. ozguc from bogazici university kandilli observatory and the daily soft x - ray flare index defined by antalov@xmath0 ( 1996 ) are two main flare indices in the present research .
for instance , li et al . (
2010 ) found that the northern - hemispheric flare activity should lead the southern - hemispheric flare activity for low - frequency components by using the former flare index .
joshi et al .
( 2004 ) reported a real north - south asymmetry during solar minimum and obtained the significant periods of approximately 28.26 days , 550.3 days , and 3.72 years by using the latter flare index . in this paper
, we focus on the three solar cycles ( 21 , 22 , and 23 ) to investigate the phase relation between sunspot numbers and flare activity ( c - class , m - class , and x - class flares ) by using a flare index , which is similar to antalov@xmath0 ( 1996 ) .
the data used in this paper are as follows : \1 .
the monthly mean northern and southern hemispheric sunspot numbers from january 1945 to december 2004 were compiled by temmer et al .
( 2006)(@xmath1 ) .
the monthly mean northern and southern hemispheric sunspot numbers from january 1992 to december 2008 can be downloaded from noaa s national geophysical data center ( ngdc ) ( @xmath2 ) .
there is an overlapping time span from january 1992 through december 2004 with the first sunspot data .
the first group of data generally matches the second group ( for details , see temmer et al .
2006 ) . in this paper
, we extracted the sunspot numbers from may 1976 to may 2008 as our samples .
the soft x - ray flares were downloaded from ngdc ( @xmath3 ) . in this paper , we adopted the peak fluxes of soft x - ray observed by goes 1 - 8 as the defined flare flux when one flare occurred .
afterwards , we summed monthly peak fluxes of the defined flares ( c - class , m - class , or x - class flares ) as a flare index .
the equations are as follows : @xmath4 @xmath5 @xmath6 in the above equations , @xmath7 , @xmath8 , @xmath9 indicate all sub - levels of c - class , m - class , x - class flares ( e.g. , for c - class flares , the sub - level flares ( @xmath7 ) are c1.0 , c1.1 , c1.2 , @xmath10 , and so on . ) .
@xmath11 " is given as a float number with one decimal ( prior to 1980 april 27 ,
@xmath11 " is listed as an integer . ) . the value of
@xmath11 " is 1 , 1.1 , 1.2 @xmath10 , @xmath12 .
for c - class and m - class flares , the value of
@xmath11 " is no more than 9.9 .
@xmath13 denotes the monthly numbers of sub - level flares ( e.g. , @xmath14 denotes the monthly numbers of @xmath15 , @xmath16 or @xmath17 . ) .
@xmath18 , @xmath19 , and @xmath20 indicate the monthly peak fluxes of c - class , m - class , and x - class flares .
this method is the same as the flare index ( fi ) defined by antalov@xmath0 ( 1996 ) and used by joshi et al .
the unit of the peak flux is erg@xmath21@xmath22@xmath21@xmath23 . as weak soft x - ray flares can not be detected during times of high activity ( feldman et al . 1997 ; veronig et al .
2002 ) , we discarded the classification of soft x - ray flares lower than c1.0 to remove the effect of the variation of soft x - ray background radiation .
figure 1 shows the distribution of the monthly numbers of three types of flares from may 1976 to may 2008 .
the period covers all of solar cycle 21 ( may 1976- july 1986 ) , cycle 22 ( august 1986- march 1996 ) , and cycle 23 ( april 1996- may 2008 ) .
the monthly flare numbers of c - class , m - class , and x - class flares have obvious periodical variability as well as sunspot numbers ( see fig .
however , different flares have different peak intensities of soft x - ray flux . for instance
, the peak flux of an x3.0 flare is 100 times larger than that of c3.0 flares . for m - class flares ,
the difference of the peak flux is about 10 times from m1.0 to m9.9 . considering the physics , we summed the monthly peak fluxes of all the sub - class flares of the same type flares .
detailed calculations can be seen from the equations in section 2 . from table 1 ,
the numbers of c - class flares are larger than that of m - class and x - class flares in each cycle
. however , the total peak fluxes of the c - class flares are lower than those of m - class flares in cycles 21 and 22 .
2a , b , c , d show that the monthly peak fluxes of the c - class , m - class , x - class flares and total peak fluxes of the three types of flares ( blue lines ) from may 1976 to may 2008 and their 13-point smoothed values superimposed ( red lines ) .
the monthly peak fluxes of c - class , m - class , and x - class flares have obvious periodical variability as well as the change of monthly sunspot numbers ( fig .
2e ) . to show the systematic time - lag ( or time - lead ) between the monthly peak fluxes of c - class , m - class , x - class flares and the monthly sunspot numbers in a cycle , we have done a correlation analysis of the two time series .
in order to eliminate the impact that large , active flare - productive regions exert simultaneously to sunspot and flare numbers , we adopted the 13-point smoothed monthly peak fluxes of c - class , m - class , x - class flares and sunspot numbers to calculate the correlation coefficient and phase relation between them .
the smoothed monthly peak fluxes of c - class , m - class and x - class flares have positive correlation coefficients with the smoothed monthly sunspot numbers ( seen from table 2 ) . the correlation coefficients of m - class and
x - class flares and sunspot numbers are higher in cycle 22 than those in cycles 21 and 23 .
furthermore , the correlation coefficients of the smoothed peak fluxes between the three kinds of soft x - ray flares in cycle 22 are higher than those in * 21 * and 23 .
the correlation coefficients between the smoothed monthly sunspot numbers and the smoothed monthly peak fluxes of c - class , m - class , and x - class flares can be seen in figure 3 .
the abscissa indicates the shift of the smoothed monthly sunspot numbers pertaining to the smoothed monthly peak fluxes of c - class , m - class and x - class flares , with negative values representing backwards shifts .
it is found that the monthly peak fluxes of c - class , m - class , and x - class flares have a very noticeable time lag of 13 , 8 , and 8 months respectively in cycle 21 with respect to sunspot numbers .
there is no time lag between the sunspot numbers and m - class flares in cycles 22 .
however , there is a one - month time lag for c - class flare and a one - month time lead for x - class flare with respect to sunspot numbers in cycle 22 .
moreover , the smoothed monthly peak fluxes of c - class , m - class , and x - class flares have a very noticeable time lag of one month , 5 months , and 21 months respectively with respect to the smoothed monthly sunspot numbers in cycle 23 .
if we take the three types of flares together , there is an obvious time lag of 9 months in cycle 21 , no time lag in cycle 22 and a characteristic time lag of 5 months in cycle 23 with respect to sunspot numbers .
in this paper , we present a statistical study on three types of soft x - ray flares from may 1976 to may 2008 .
we use the data of the smoothed monthly peak fluxes of c - class , m - class , and x - class flares and the smoothed monthly sunspot numbers .
the main results are as follows : \1 .
the smoothed monthly peak fluxes of c - class , m - class , and x - class flares have a very noticeable time lag of 13 , 8 , and 8 months respectively with respect to the smoothed monthly sunspot numbers in cycle 21 .
there is a one - month time lag for the smoothed monthly peak fluxes of c - class flares , a one - month time lead for the smoothed monthly peak fluxes of x - class flares and no time lag for the smoothed monthly peak fluxes of m - class flares with respect to the smoothed monthly sunspot numbers in cycle 22 .
the smoothed monthly peak fluxes of c - class , m - class , and x - class flares have a very noticeable time lag of one month , 5 months , and 21 months in cycle 23 respectively with respect to the smoothed monthly sunspot numbers .
if we take the three types of flares together , we find that the soft x - ray flares have an obvious time lag of 9 months in cycle 21 , no time lag in cycle 22 and a characteristic time lag of 5 months in cycle 23 with respect to the smoothed monthly sunspot numbers . in this paper
, we adopted the method developed by antalov@xmath0 ( 1996 ) and used the smoothed monthly peak fluxes of soft x - ray flares .
we divided the soft x - ray flares into three groups according to the classification of soft x - ray flare , which is slightly different from temmer et al .
when the numbers of flares are multiplied by the peak fluxes , there is a great change in value pertaining to the original flare numbers . from table 1
, one can see that the numbers of m - class , x - class flares are small but they have large monthly peak fluxes .
the peak fluxes of flares can be approximately taken as the energy released by flares .
as m - class and x - class flares have released more energy than that of b - class and c - class flares , we think that the peak fluxes of flares taken as the flare index may be closer to the energy released by the flares .
that is why we used this method to analyze the phase relation .
the activity of the c - class , m - class and x - class flares exhibited an obvious time lag behind sunspot activity in odd - numbered cycles , while in even - numbered cycles m - class flares do not .
also , the c - class and x - class flares show reversed behavior in the even cycle .
if we consider the three types of flares together , we find that there is an obvious time lag in the odd solar cycles and no time lag in the even solar cycle .
though different sorts of flares show different time lags or time leads with respect to sunspot numbers , there is a regular time delay in the odd solar cycles and no time delay in the even solar cycle between soft x - ray flares and sunspot numbers after we sum the three types of flares together .
the value of summing the monthly peak fluxes of three types of flares approximately denotes the total energy released by flares .
these results imply that the flare activities are related to the 22-year magnetic cycle of the sun .
we need forthcoming data to confirm these results and model the process of flare energy storage and dissipative mechanism .
table 1 + the numbers and peak fluxes of three types of soft x - ray flares ( c - class , m - class , and x - class flares ) in cycles 21 , 22 , and 23 .
the values in parentheses denote the peak fluxes of the defined flares .
total " denotes the total numbers and the total peak fluxes in the selected period from may 1976 to may 2008 .
the unit of flare peak flux is erg@xmath21@xmath22@xmath21@xmath23 .
+ llllllll [ tbl-1 ] flare class & 21 cycle & 22 cycle & 23 cycle & total + c & 14576 ( 46.76)&12430(41.79 ) & 13148 ( 39.39)&40154(127.94 ) + m & 2175(50.22)&2019(48.15)&1439(34.37)&5633(132.74 ) + x&165(37.59)&152(35.46 ) & 126(31.08)&443(104.13 ) + table 2 + correlation coefficients between the smoothed monthly peak fluxes of c - class flares , m - class flares , x - class flares and the smoothed monthly sunspot numbers in cycle 21 ( the values in parentheses denote correlation coefficients in cycles 22 and 23 . ) .
ssn " , sfc " , sfm " , and sfx " indicate the smoothed monthly sunspot numbers , the smoothed monthly peak fluxes of c - class , m - class , and x - class flares respectively .
+ the authors thank the anonymous referee for her / his careful reading of the manuscript and constructive comments that improved the original version .
the authors thank the staff of all the web sites that provide the data for the public to download .
this work is supported by the national science foundation of china ( nsfc ) under grant numbers 10903027 , 10673031 and 40636031 , 11003041 , yunnan science foundation of china under grant number 2009cd120 , and the national basic research program of china 973 under grant number . | to better understand long - term flare activity , we present a statistical study on soft x - ray flares from may 1976 to may 2008 . it is found that the smoothed monthly peak fluxes of c - class , m - class , and x - class flares have a very noticeable time lag of 13 , 8 , and 8 months in cycle 21 respectively with respect to the smoothed monthly sunspot numbers .
there is no time lag between the sunspot numbers and m - class flares in cycle 22 .
however , there is a one - month time lag for c - class flares and a one - month time lead for x - class flares with regard to sunspot numbers in cycle 22 . for cycle 23 ,
the smoothed monthly peak fluxes of c - class , m - class , and x - class flares have a very noticeable time lag of one month , 5 months , and 21 months respectively with respect to sunspot numbers .
if we take the three types of flares together , the smoothed monthly peak fluxes of soft x - ray flares have a time lag of 9 months in cycle 21 , no time lag in cycle 22 and a characteristic time lag of 5 months in cycle 23 with respect to the smoothed monthly sunspot numbers .
furthermore , the correlation coefficients of the smoothed monthly peak fluxes of m - class and x - class flares and the smoothed monthly sunspot numbers are higher in cycle 22 than those in cycles 21 and 23 .
the correlation coefficients between the three kinds of soft x - ray flares in cycle 22 are higher than those in cycles 21 and 23
. these findings may be instructive in predicting c - class , m - class , and x - class flares regarding sunspot numbers in the next cycle and the physical processes of energy storage and dissipation in the corona . |
this paper is a theoretical extension of our recent letter @xcite on nanoscale stratification of local field and atomic dipole excitation in low - dimensional lattices driven by a laser at the frequency of the resonant atomic transition .
we focus here on the most fundamental case of a one - dimensional ( 1d ) array of resonant atoms , and construct a detailed theory of both linear and nonlinear interactions in the system , resulting in such phenomena as sub - wavelength spatial modulation ( stratification ) of polarization and local field , long - wave and short - wave stratification , size - related resonances and large field enhancement , `` magic '' numbers , ferromagnetic- and antiferromagneticlike atomic polarization , optical bistability and hysteresises , etc .
in addition to the results @xcite and their derivations , we also present new results ( i ) on the size - related resonances using a many - body approximation involving interactions of each atom with all other atoms in the array , beyond the ising - like approximation whereby atoms only interact with their nearest neighbors , ( ii ) on traveling locsiton waves and their dissipation , as well as an estimate of the maximum size of the 1d array to support most of the effects discussed here , ( iii ) on hysteresises and optical bistability in arbitrarily long arrays , and ( iv ) general mathematical consideration of dispersion relation in the self - interacting arrays for dipole approximation and beyond it .
it is well known that optical properties of sufficiently dense materials are substantially affected by the near - field interactions between neighboring particles at the frequency of the incident field , in particular , quasi - static ( non - radiative ) dipole interactions . the best known manifestation of this fact is the _ local vs incident field _ phenomenon and related lorentz lorenz or clausius mossotti relations @xcite for dielectric constant as a nonlinear function of the number density .
the microscopic ( or _ local _ ) field ( lf ) , @xmath0 , acting upon atoms or molecules becomes then different from both the applied and average macroscopic fields because of inter - particle interaction . in the most basic case ,
that relation between @xmath0 and the average field @xmath1 is @xmath2 , where @xmath3 is the dielectric constant of the material at the laser frequency @xmath4 .
it is worth noting that we are not interested here in the relation between @xmath3 and the number density of the material , since in the case of small 1d or 2d arrays the issue is moot .
for the same reason , it also makes sense to deal directly with the incident field , @xmath5 , instead of the averaged field . in most theories of those interactions a traditional standard ( and
at that often implicit ) assumption , reflected also in the above formula , is that the local field and polarization are uniform in the near neighborhood of each particle at least at the distances shorter than the wavelength of light , @xmath6 .
this essentially amounts to the so called `` mean - field approximation '' .
it was shown by us @xcite , however , that if the _ local uniformity _ is not presumed , then , under certain conditions on the particle density and their dipole strengths , the system of interacting particles is bound to exhibit periodic spatial variations of polarization and local - field amplitude
. these variations result in sub - wavelength strata with a nanoscale period much shorter than @xmath6 . under certain conditions ,
the system may exhibit an ultimate non - uniformity , whereby each pair of neighboring atoms in 1d arrays has their dipoles counter - oscillating with respect to each other , i.e. their excitations and thus local fields have opposite signs . to a certain extent
this is reminiscent of the situation in magnetic materials with _
ferromagnetic vs antiferromagnetic _ effects . indeed , the mean - field approximation which is at the root of the curie
weiss theory of _ ferromagnetism _
@xcite , is based on the assumption of uniform polarization of all the neighboring magnetic dipoles even without external magnetic field , when taking into account their interaction with each other .
contrary to that , the ising theory @xcite , which does not make the assumption of uniformity , showed that even in the near - neighbor approximation for that interaction , the excitation may result in counter - polarization of neighboring atoms in 1d arrays , and thus in a new phenomenon of _ antiferromagnetism_. note here that in these effects there is no notion of `` local '' _ vs _ `` external '' field phenomenon : in a `` pure '' case of either of the magnetic effects , no external field is applied ; the effects here are the result of self - organization of permanent , `` hard '' , atomic dipoles with pre - existing _ dc _ dipole fields , without any `` help from outside '' . in this lies a profound difference between _ dc _ magnetic material phenomena ( ferro- and antiferromagnetism ) on the one hand and the effects considered by us here and in @xcite on the other hand ,
all of which are based on the _ optical _ ( or , in general , any other quantum or classical resonance ) excitation of atoms ( or other small particles , e.g. quantum dots , clusters , small - particle plasmons , etc . ) .
while magnetic dipoles in ferromagnetics are nonzero even in the absence of an external field ( we may call them `` hard '' dipoles ) , the oscillating dipoles ( in the linear case ) can be induced only by the driving field at the near - resonant frequency , so they can be called `` soft '' dipoles ; without such a driving their polarization vanishes .
because in dense material the atoms actually are acted upon by _ local _ field , the response of each one of them may differ from the others by phase and amplitude ( or even direction ) , with some of the dipoles fully suppressed while others fully excited . thus the effects considered here are induced by the interplay of external and local fields , which put the entire phenomenon squarely into the domain of relations between the local and incident fields . because of that , since the phenomenon depends strongly on the characteristics of the incident field ( polarization , frequency , and , in the nonlinear case , intensity ) , the spatial modulation of the dipole excitation and local field can vary substantially .
this results in a wealth of different patterns , some of them reminiscent of the ferromagnetic , other of antiferromagnetic , but the most of them forming all kinds of hybrid patterns .
the complete cross - over from ferromagneticlike to antiferromagneticlike state of the system with all the intermediate states can be attained then by simply tuning laser frequency .
another significant difference here is that the system size is small .
provided there is sufficiently strong interparticle interaction , the new phenomenon can occur in the vicinity of boundaries , lattice defects , impurities or in sufficiently small group of atoms ; recent advances in technology allow fabrication of nanoscale structures with small numbers of atoms . thus , our theory emphasizes phenomena in relatively small ordered arrays of interacting atoms , in contrast to , e.g. , microscopic models of ferromagnetism that mostly focus on averaged , `` thermodynamic '' perspective on sufficiently large systems .
this brings forward a new set of nanoscale phenomena . harking back to ferromagnetic systems ,
this new emphasis may reveal similar phenomena for nanoscale magnetic systems , which could be an exciting topic for a separate study .
our choice of 1d and 2d dielectric systems based on the arrays or lattices of atoms , quantum dots , clusters , molecules , etc .
, allows to control anisotropy of near - field interaction .
it also eliminates the issues of electromagnetic ( em ) propagation being modified by the effects as the em wave propagates through the structure ( especially if it propagates normally to the lattice ) .
if _ local uniformity _ is broken by any perturbation , the system may exhibit near - periodic spatial sub-@xmath6 patterns ( strata ) of polarization .
in general , two major modes of the strata transpire : _ short - wave _ ( sw ) , with the period up to four interatomic spacings , @xmath7 , and _ long - wave _ ( lw ) strata .
the strata are standing waves of elementary lf excitations ( called _ locsitons _ in @xcite ) having a near - field , electrostatic , nature and low group velocity . in the first approximation ,
the phenomenon is linear in the driving field , and the locsitons may be excited within a spectral band much broader than the atomic linewidth .
it can be viewed as a rabi broadening of an atomic line by interatomic interactions .
the strata are controlled by laser polarization and the strength of atom coupling , @xmath8 , _ via _ atomic density , dipole moments , relaxation , and detuning .
once @xmath9 , the lf uniformity can be broken by boundaries , impurities , vacancies in the lattice , etc .
a striking manifestation of the effect is large field resonances due to locsiton eigenmodes in finite lattices , and at certain , `` magic '' numbers of atoms in the lattice
almost complete cancellation of field suppression at the atomic resonance ; saturation nonlinearity results in hysteresises and optical bistability .
the paper is structured as follows . in section [ sec2 ]
we derive the main equations for self - interacting atomic lattices of arbitrary dimensions using two - level ( nonlinear in general ) model for atomic resonances and dipole
dipole interaction between atoms , while section [ sec3 ] is on specific equations for linear infinite and finite 1d arrays . in section
[ sec4 ] we develop the general theory of locsitons and derive the dispersion relation . in section [ sec5 ]
we study locsiton band formation , size - related resonances due to standing waves of locsitons ( strata ) and local - field enhancement . in section [ sec6 ]
we concentrate on detailed theory of resonances beyond the near - neighbor approximation , including evanescent solutions ( see also section [ sec4 ] ) .
magic numbers are considered in section [ sec7 ] . in section [ sec8 ] ,
we study the effects of losses on locsiton excitation , depth of penetration , and traveling locsiton waves .
section [ sec9 ] is on nonlinear locsiton modes , in particular , optical bistability and hysteresis .
section [ sec10 ] addresses potential applications of locsitons and their analogies in other physical systems . in conclusions ,
section [ sec11 ] , we summarize our results . appendix [ app ] is on general mathematical aspects of dispersion relations for 1d arrays .
our model is based on the near - field dipole atomic interactions , with the incident frequency @xmath4 being nearly resonant to an atomic transition with a dipole moment @xmath10 at the frequency @xmath11 . in the linear case ,
i.e. when the laser intensity is significantly lower than that for the quantum transition saturation ( see below ) , the result of this model coincides with that of a classical harmonic oscillator formed by an electron in a harmonic potential with the same resonant frequency @xmath11 and with the dipole moment @xmath12 where @xmath13 is the compton wavelength of electron . in a standard lf situation , @xmath14 , where @xmath15 , the field of an elementary dipole with the polarization @xmath16 is dominated in its near vicinity by a non - radiative , quasi - static ( and only electric ) component , which is strongly anisotropic in space .
this dominant term in the near - field area , @xmath17 , attenuates as @xmath18 ( see e.g. @xcite , section 72 ) . at @xmath19 the amplitude of the field of an oscillating dipole located at @xmath20 induces a field at the point of observation @xmath21 with the amplitude coinciding with that of an elementary static dipole with the same polarization @xmath22 : @xmath23 where @xmath24 is a unit vector in the direction of observation , and @xmath3 is a background dielectric constant .
we will model a resonant atomic transition by a basic two - level atom in a steady - state mode under the action of a field , @xmath25 , with the amplitude @xmath1 and frequency @xmath4 , and assume that @xmath26 , so that the wave functions of neighboring atoms do not overlap . using a semi - classical approach standard in lf theory of resonant atoms
@xcite , we can now find the atomic polarization as : @xmath27 where @xmath16 is the polarization amplitude [ the full polarization is then @xmath28 ; @xmath29 is the population difference , with @xmath30 and @xmath31 being atomic populations at respective ground and excited levels ( @xmath32 ) , @xmath33 is a dimensionless detuning from the resonant frequency @xmath11 of the two - level atom , @xmath34 is a _ transverse _ relaxation time ( the time of polarization relaxation ) and @xmath35 is the ( homogeneous ) linewidth of the linear resonance .
is comprised of two contributions @xcite : one is the so called natural radiation linewidth @xmath36 , which is due to spontaneous radiation at the resonant frequency of the respective quantum transition of a single atom in vacuum , and the other one , @xmath37 , is due to all other processes , mostly phase broadening , non - radiative transitions , and radiative ones to non - resonant quantum levels .
it is worth noting that when the atoms are very close to each other , @xmath38 , due to atomic interactions the radiation linewidth , @xmath36 , of each atom can be broadened up if the neighboring dipoles move in phase ( as in long - wave locsitons ) , or narrowed down if those dipoles move in counter - phase ( as in short - wave locsitons ) . in general case
this effect should be taken into consideration too .
however , we assume here that , as is usual for atoms in a dense - material environment ( e.g. when they are positioned at the surface of some material or embedded in it ) , @xmath39 , and we ignore here the natural linewidth , @xmath36 and its broadening . ] in turn , the steady - state population difference is @xcite @xmath40 where @xmath41 is a _ longitudinal _ relaxation time ( life - time of the excited atom ) , and @xmath42 is an equilibrium population difference at the system temperature due to boltzmann s distribution ; in optics one can usually assume @xmath43 , so that @xmath44^{-1 } \equiv f_\mathrm{nl}(|e|^2 ) , \nonumber \\ e_\mathrm{sat}^2 & = & \frac{\hbar^2}{|d_a|^2 } \cdot \frac{1}{\tau t } , \label{eq2.5}\end{aligned}\ ] ] where @xmath45 is the saturation intensity , and @xmath46 is nonlinearity due to saturation . substituting ( [ eq2.5 ] ) into ( [ eq2.3 ] ) , one obtains a closed - form expression for the polarization , @xmath47 for a classical harmonic ( linear ) oscillator we have : @xmath48 where @xmath49 is the mass of electron .
the local field @xmath50 at each atom is the incident laser field , @xmath5 , plus the sum of the near - fields , @xmath51 ( [ eq2.2 ] ) , induced by all the surrounding dipoles at @xmath20 acted upon by the respective local fields @xmath52 , i.e. @xmath53 where @xmath54 denotes summation over the entire lattice or array . to obtain a closed - form master equation , e.g. for @xmath50 alone , we use ( [ eq2.6 ] ) to write : @xmath55 - { \mathbf{e}}_\mathrm{l}({\mathbf{r } } ' ) \right\ } f_\mathrm{nl } [ |{\mathbf{e}}_\mathrm{l } ( { \mathbf{r}}')|^2 ] , \label{eq2.9}\end{aligned}\ ] ] where @xmath56 are local fields only at the locations of atoms in the lattice and not at any other points inside or outside it ; @xmath57 is a tuning - dependent strength of dipole
dipole interaction , and the maximum absolute strength , @xmath58 , is @xmath59 where @xmath60 is the fine - structure constant , and nonlinear factor @xmath61 is as in ( [ eq2.5 ] ) . for a classical harmonic oscillator , ( [ eq2.1 ] ) , we have @xmath62 where @xmath63 nm is the classical radius of electron . eqs . ( [ eq2.8 ] ) , ( [ eq2.9 ] )
reflect many - body nature of the interaction . a conventional approach to local fields within the lorentz
lorenz theory is to look for a self - consistent solution for the fields in this interaction , with an assumption , however , that they are _ uniform _ ( the mean - field theory ) , i.e.to set @xmath64 and also use an encapsulating sphere around the observation point .
these assumptions effectively shut out any strong spatial variations of the atomic excitations and local field that may exist at the inter - atomic scale .
that is where we depart from the lorentz
lorenz theory ; _ none of those assumptions are used here _ , and our approach is to use general expressions ( [ eq2.8 ] ) or ( [ eq2.9 ] ) and seek straightforward solution for them .
we will see below that the major critical condition for the phenomenon to exist and be observable at least at other optimal conditions is that @xmath58 exceeds some critical value , @xmath65 .
three parameters are critical in this respect : atomic dipole moment @xmath66 , the spacing between atoms , @xmath7 , and the atomic linewidth , @xmath35 , since @xmath67 .
to get an idea of whether the above critical condition is realistic , let us look first at the case of a gas - like collection of atoms , with the relaxed requirement on the spacing @xmath7 .
large dipole moments and narrow resonances in , e.g. , _ alkali _ vapors @xcite or co@xmath68 gas @xcite , in solids @xcite , quantum wells and clusters may greatly enhance the phenomenon and allow for @xmath7 from sub - nanometer to a few tens of nanometers .
considering an example with @xmath69 , corresponding to the volume density of @xmath70 @xmath71 , @xmath72 , @xmath73 @xmath74 , @xmath75 , and @xmath76 , all of which are reasonable data , we obtain @xmath77 , which provides a margin large enough to see all the effects discussed here .
it is also of interest to roughly estimate what is the upper limit for @xmath58 . to that end , consider the extreme situation of @xmath78 ( solid - state- or liquid - like packing of participating atoms ) , in which case we have the ceiling for @xmath58 as @xmath79 even taking into consideration significant line broadening , @xmath35 , @xmath80 may exceed unity by many order of magnitude , thus providing huge margin for the existence and observation of locsitons and related effects .
we consider here the most basic model of a 1d array of @xmath81 atoms lined up along the @xmath82 axis , spaced by @xmath7 , and driven by a laser propagating normally to the array and having an arbitrary polarization ( fig .
[ fig1 ] ) . in the linear case , @xmath83 , when in ( [ eq2.9 ] ) @xmath84 ,
it is sufficient to consider effects caused by linearly polarized light with either one of the two mutually orthogonal polarizations ; any other polarization ( e.g. a circular one ) can be treated as a linear combination of those two . in the case of a 1d array , natural choices for these two basic configurations
are : * the incident field @xmath5 is parallel to the @xmath82 axis , @xmath85 ( and the dipoles line up `` head - to - tail '' ; we will call it @xmath86-configuration ) , and * @xmath5 is normal to the @xmath82 axis , @xmath87 ( `` side - by - side '' lineup ; @xmath88-configuration ) .
the general solution will be a linear vectorial superposition of these two .
this choice of the basic configurations is dictated by the simplicity of the resulting polarization of the local field .
indeed , in both cases , it follows that the polarization of local field is parallel to that of the incident field , @xmath89 , so we can use scalar equations for all fields . using the dimensionless notation @xmath90_{(p)}$ ] , where @xmath91 denotes polarization , @xmath92 or @xmath93 , and recalling that now @xmath94 , we write ( [ eq2.9 ] ) for both configuration as @xmath95 where @xmath96 is a form - factor due to polarization configuration : @xmath97 and @xmath98 . if the 1d array is infinite , @xmath99 , or sufficiently long , @xmath100 , it is instructive to rewrite ( [ eq3.1 ] ) in the form @xmath101 where @xmath102 . the sums over @xmath103 in ( [ eq3.1 ] ) , ( [ eq3.2 ] ) converge rather fast ,
hence @xmath104 is not too large , see also appendix [ app ] below .
( [ eq3.1 ] ) and ( [ eq3.2 ] ) , the same as master equation ( [ eq2.9 ] ) , represent the case of _ fully - interacting arrays _
( fia ) , whereby each atom `` talks '' to all the other atoms in the array , which presents a challenge to an analytical treatment .
of course , a linear equation ( [ eq3.1 ] ) for @xmath105 is solved analytically using a standard linear algebra approach with matrices . however , analyzing the results for large - size arrays , @xmath100 , in particular analytically finding all the resonances in @xmath106 , can only be done by using numerical matrix solver , even if we neglect the dissipation . thus , there is a need for a simple approximation that would preserve most of the qualitative features of the phenomenon , yet could be easily analyzed analytically .
this can be done by using the near - neighbor approximation ( nna ) , similar to that of the ising model of ( anti)ferromagnetism , in which the full sum in ( [ eq3.1 ] ) or ( [ eq3.2 ] ) is replaced by the sum over the nearest neighbors , @xmath107 in the ultimate two - atom case , @xmath108 @xcite , the two approaches merge .
the further two - near - neighbors approximation ( 2-nna ) and even three - near - neighbors approximation are considered in section [ sec6 ] below .
we found , however , that in general , a full summation ( fia ) in ( [ eq3.1 ] ) on the one hand , and nna ( [ eq3.3 ] ) as well as 2-nna on the other hand , produce qualitatively similar results that differ by a factor of @xmath109 . since effects discussed here are most pronounced in relatively small systems or in the small vicinity of perturbations in large lattices , it is natural to stipulate that the _ incident _ field within the array is uniform , unless otherwise is stated ; however , this condition can readily be arranged even for an array larger than @xmath6
one of the solutions for the local field ( and atomic excitation ) in the _ infinite _ 1d array ( or a sufficiently long one , whereby we can neglect edge effects ) is also uniform
. we will call it the `` lorentz '' solution , @xmath110 , to be found from ( [ eq3.2 ] ) by setting @xmath111 as : @xmath112 } , \nonumber \\ & & ( \delta_\mathrm{ll})_{(p ) } = - s q_a f_{(p ) } , \label{eq3.4}\end{aligned}\ ] ] where we introduced polarization - dependent parameter @xmath113 , which determines lorentz lorenz shift at @xmath114 ( see below ) ; @xmath115 is a measure of the polarization - related strength of interaction .
( [ eq3.4 ] ) may be viewed as a 1d counterpart of the _ lorentz lorenz _ relation for local field ; notice , however , that the field @xmath110 is strongly _ anisotropic _ with respect to the polarization .
the spectral behavior of @xmath116 is depicted in fig .
[ fig2 ] with thicker dashed curves in all the graphs . ) in the near - neighbor approximation with @xmath117 : ( a ) locsiton resonances with 12 atoms in the array ; ( b ) the same with 13 ( magic number ) atoms in the array ; ( c ) merging and damping of locsiton resonances for a large number of atoms , @xmath118 . the curves show the amplitudes of local fields for size - related resonances , local fields due to lorentz lorenz theory , and lower and upper amplitude envelopes of the resonances.,width=325 ] if @xmath119 , it shows , as one may expect , a deep dip at the atomic resonance frequency , i.e. at @xmath120 , @xmath121 ( so that at the atomic resonance the local field is suppressed , as if it is pushed out of the array ) , and a strong new resonant peak appears at the shifted frequency @xmath114 due to the lorentz
lorenz effect : @xmath122 whose nature is essentially similar to e.g. lorentz
lorenz resonance observed experimentally in alkali vapors @xcite .
however , in the 1d case considered here , the lorentz lorenz shift and even its sign are polarization - dependent ; in particular , @xmath123 i.e. the lorentz
lorenz resonance is red - shifted for @xmath86-polarization of laser , and blue - shifted for @xmath88-polarization .
the lorentz field @xmath110 and lorentz lorenz shift in the near - neighbor approximation , eq .
( [ eq3.3 ] ) , are determined by the same equations ( [ eq3.4])([eq3.7 ] ) , where one has to set @xmath124 .
we look now for solution of eq .
( [ eq3.2 ] ) as the sum of _ uniform _ lf , @xmath110 , eq .
( [ eq3.4 ] ) , and oscillating ansatz @xmath125 , where @xmath126 is an ( unknown ) wavenumber , similarly , e.g. to the phonon theory @xcite , with the difference being that we have here an excitation of bound electrons , and not atomic vibrations .
essentially , the locsitons may be classified as _ frenkel _
excitons @xcite because of their no - electron - exchange nature .
the wavenumbers @xmath126 are found _ via _ the dispersion relation @xmath127 the behavior of @xmath128 in the lossless case , @xmath129 , is depicted in fig .
[ fig3 ] by the solid curve . _ vs _ the normalized wavenumber , @xmath130 .
curves show the dispersion for near - neighbor approximation , nna ( [ eq4.2 ] ) , two - near - neighbors approximation , 2-nna ( [ eq6.6 ] ) , and for fully interacting array ( fia ) , whereby each atom interacts with all the other atoms ( [ eq4.1 ] ) , which coincides with the analytical fit ( [ eqa.10]).,width=325 ] within nna , we have to set @xmath124 and replace the sum in ( [ eq4.1 ] ) by its first term : @xmath131 see fig .
[ fig3 ] , fine - dashed curve .
distinct oscillations emerge in the area between the two edges of the locsiton band .
their wavenumbers are determined from ( [ eq4.1 ] ) or ( [ eq4.2 ] ) by neglecting the dissipation , i.e. by assuming real @xmath126 and @xmath132 .
one of the band edges corresponds to the maximum ( and positive ) @xmath133_\mathrm{max } = 1 $ ] ( same as for nna , @xmath134_\mathrm{max } = d(0 ) = 1 $ ] ) , and thus to the lorentz lorenz shift , @xmath135 .
the other , `` anti - lorentz '' , edge is at the opposite side of the atomic resonance , and is determined by the minimum ( and negative ) @xmath136 .
thus we have : @xmath137 which can be evaluated using an amazingly simple relation for the sum ( [ eq4.1 ] ) , which is actually valid for a more general sum and an arbitrary exponent @xmath138 : @xmath139 this can be readily proven by an appropriate rearrangement of the terms in the sums in ( [ eq4.4 ] ) , see below ( [ eqa.11 ] ) . in the case of @xmath140 and infinite array , ( [ eq4.1 ] ) , we have @xmath141 .
hence , the locsiton band is determined as : @xmath142 with well developed locsitons at @xmath143 .
indeed , if the dissipation is neglected and the wavenumbers @xmath126 are real , there is an infinite number of solutions for them within the limits ( [ eq4.5 ] ) ; in this case the meaningful positive solutions , within the first brillouin zone , are @xmath144 .
the plots of the functions in the l.h.p . of ( [ eq4.1 ] ) and ( [ eq4.2 ] ) are shown as @xmath145 vs @xmath126 in fig .
[ fig3 ] . based on ( [ eq4.5 ] ) ,
the total width of the locsiton band in terms of @xmath146 is thus @xmath147 , if one accounts for the interactions of each atom with _ all _ the rest of atoms in the infinite array , whereas it is @xmath148 in the near - neighbor approximation . to gauge the dipole
dipole interaction in the lattice , one can also introduce its rabi - energy as : @xmath149 it brings about a locsiton energy band @xmath150 ( if @xmath143 ) akin to those in solid - state @xcite , photonic crystals @xcite , and electronic band - pass filters .
one of the most interesting effects due to locsitons is a wide spectrum of the standing waves , strata , formed by them , see fig .
[ fig4 ] .
as we already noted , they can range from the very long ones , long - wave ( lw ) strata , with the maximum spatial period being double of the whole length of a 1d array , similarly to the main mode of oscillation in a violin string , to short - wave ( sw ) strata whereby each dipole oscillates in counter - phase to each of its nearest neighbors .
the latter mode is the strongest manifestation of the fact that the array of @xmath81 atoms is a discrete - element resonator , akin a string of beads connected to each other , with the beads capable of the same kind of motion , when each individual bead is oscillating in counter - phase with its neighbor .
because in both cases the excitation is of dynamic nature and is due to external driving , the mechanical analogy of sw locsiton modes is more adequate than that of a static ferromagnetic configuration _ vs _
antiferromagnetic one ; see also appendix [ app ] below .
it can be immediately found , e.g. from consideration of the plots at fig .
[ fig4 ] , upper curve , that the lw strata emerge at the laser tuning very near to the lorentz
lorenz resonance , i.e. in the limit @xmath151 . in this case , in both fia ( [ eq4.1 ] ) and nna ( [ eq4.2 ] ) , the wave number @xmath152 and the respective spatial wavelength @xmath153 are as : @xmath154 in the infinite array , the longest @xmath155 is up to @xmath156 , when @xmath157 ; the locsitons with longer wavelength get significantly suppressed . the opposite limit , or the anti - lorentz side of the locsiton band
, @xmath158 defines sw locsitons , with @xmath159 i.e. @xmath160 is the finest grain of locsiton structure , as one would expect from the `` bead string '' analogy . however , an incommensurability , i.e. mismatch between @xmath160 and the lattice spacing @xmath7 , whose ratio is in general an _ irrational _ number , results in a strong spatial modulation of the sw , giving rise to a _
coarse _ lw - like structure .
in the nna , this coarse structure of sw mode has the half - period roughly the same as for a pure lw mode , @xmath161 .
this can be readily understood in terms of `` beating '' between the locsiton wavelength , @xmath162 , and the spatial scale of the discrete structure of the system , which is the normalized spacing between atoms , 1 . indeed , near the anti - lorentz edge , @xmath163 , we can write for a sw wavenumber : @xmath164 , with @xmath165 , and find the spatial oscillations as @xmath166 with @xmath167 , which shows alternating , counter - phase motion of the neighboring atoms , @xmath168 , modulated by a slow envelope , @xmath169 .
both the fine grain and the coarse modulation may be well pronounced ( fig .
[ fig4 ] , middle curve ) . at @xmath170 ,
the lw and sw periods converge to the same scale , @xmath171 , see ( [ eq4.12 ] ) below . using the phonon analogy , the lw locsitons may be viewed to a certain extent as counterparts to acoustic , and sw to optical phonons . between those limiting points
lw locsitons at the lorentz end of the locsiton spectrum , @xmath172 , and sw locsitons at the anti - lorentz end , @xmath173 , there are all kinds of locsitons making chaotic looking strata ( due to the above mentioned incommensurability , i.e.irrational ratio between a locsiton wavelength , @xmath155 , and the array spacing , @xmath7 ) .
however , same as in the chaotic motion , there are small islands of well ordered wave patterns , located at the spectral points where the ratio @xmath174 ( or @xmath130 ) is a rational number , provided that the system is at a _ resonance _ , see ( [ eq5.2 ] ) and ( [ eq5.3 ] ) below .
one can think of them as sort of hybrids of ferromagneticlike and antiferromagneticlike behavior .
indeed , the purely antiferromagneticlike sw locsiton at @xmath170 is formed by the atoms with alternating polarizations , @xmath175 this happens at @xmath173 within nna and @xmath176 for fia .
the simplest hybrid pattern is formed at @xmath177 [ which is at @xmath120 within nna , and @xmath178 for a fully interacting array ( [ eq4.1 ] ) ] , whereby each second atom is _ non - excited _ , while the other atoms alternate their polarization : @xmath179 examples of other simplest hybrid states within nna are : @xmath180 , @xmath181 , etc . ; in general , periodic patterns exist if @xmath130 is a rational number , i.e. for all the eigen - resonances in finite 1d arrays within nna , see eq .
( [ eq5.3 ] ) below . finally , it is worth noting that the strata , albeit fast - decaying , exist even _ beyond _ the locsiton band , ( [ eq4.5 ] ) .
they are not , however , propagating locsitons , and their amplitudes exponentially decay with the distance .
consider the simplest case , nna , with the losses negligibly small , @xmath182 in ( [ eq4.2 ] ) . in this case , @xmath183 in ( [ eq4.2 ] ) , which indicates that the wavenumber @xmath126 must be complex . indeed , writing @xmath184 for the `` lorentz '' end of the band , @xmath185 , and @xmath186 for the anti - lorentz end , @xmath187 ( see also section [ sec6 ] below ) , we have the solution for @xmath188 as @xmath189 \label{eq4.13}\ ] ] and for the local fields as @xmath190 for the lorentz and anti - lorentz ends of the band , respectively . here , in the case of a semi - infinite array , the sign in ( [ eq4.13 ] ) has to be chosen such that the amplitude of the field vanishes at @xmath191 .
these modes can be viewed as _ evanescent waves / locsitons _ , see also below , section [ sec6 ] .
one can note though that they still bear the signature of the respective locsitons on either side of the band : long - wave , almost synchronous oscillations on the lorentz end , and short - wave , phase - alternating oscillations on the anti - lorentz end .
in general , the same patterns hold in the fia case .
due to boundary conditions in ( [ eq3.1 ] ) [ or ( [ eq3.3 ] ) within nna ] , the array of @xmath81 atoms is a discrete - element resonator .
this should result in locsiton quantization within the locsiton band ( [ eq4.5 ] ) and corresponding size - related resonances of the local field . in a 1d array with @xmath81 atoms
, we have @xmath81 coupled oscillators with the same individual atomic resonance frequency , and therefore , one should expect the original atomic line to be split into @xmath81 lines at most , with the collective broadened band ( [ eq4.5 ] ) being replaced with those lines .
of course , when the dissipation ( or finite linewidth of each individual line ) is taken into account , those split lines will merge into one continuous locsiton band ( [ eq4.5 ] ) if the array is sufficiently large , @xmath192 the simplest result for the resonant line positions is obtained within nna . using the boundary condition ( [ eq3.3 ] ) , i.e.@xmath193 , we find that the longest locsiton half - wave , corresponding to the fundamental mode , is @xmath194 thus , @xmath195 is the distance between nodes where lf zeros out , whereas the wavelengths @xmath196 of eigen - locsitons and their eigen - frequencies @xmath197 , with the quantum number @xmath198 , are respectively @xmath199 [ note that the first eq .
( 7 ) in @xcite , which corresponds to the second eq .
( [ eq5.3 ] ) here , contained a typo ( an extraneous @xmath200 in the cos argument ) which we corrected here . ] from these , only the resonances with odd @xmath201 will be realized for a symmetric driving laser profile , in particular , the uniform one , @xmath202 ( which is the most common case here ) , and with even @xmath201 for an anti - symmetric one , say , @xmath203 , where @xmath204 is the sequence number of an atom in the array . in all other cases , a full set of @xmath81 resonances
will be realized .
in essence , the size - related locsiton resonances in discrete arrays are , to a limited extent , similar to any eigen - resonances in regular continuous , i.e.non-discrete , 1d system .
examples can be found both in classical setting , e.g. a violin string , a fabri
perot resonator ( as e.g. in a laser ) , and in quantum mechanics , from the resonances in a quantum well with infinitely - high walls , to electron gas in a finite layer @xcite , electrons in long molecules @xcite , etc .
the major difference here is that the number of eigenmodes , or resonances , in an array with @xmath81 elements is limited to @xmath81 , in contrast to the theoretically infinite number of eigenmodes in continuous finite 1d systems .
the resonances for uniform driving within nna are shown in fig .
[ fig2 ] for @xmath117 in the case of @xmath205 [ fig .
[ fig2](a ) ] , @xmath206 [ fig .
[ fig2](b ) ] , and @xmath118 [ fig . [ fig2](c ) ] .
one can readily find out that the lower amplitude envelope is @xmath207 while the upper envelope of the resonant peaks within nna for a uniform driving is @xmath208 where @xmath209 .
as @xmath81 increases , the resonances merge and are suppressed at @xmath210 , see e.g. fig .
[ fig2](c ) .
however , even then @xmath211 still exceeds the uniform field @xmath116 ( [ eq3.4 ] ) by a factor of 2 . for @xmath212 ( @xmath201 is a natural number ) , lf amplitude dips below the lower envelope @xmath213 at @xmath214 . at that frequency , within nna , @xmath215 , @xmath216 , and the sw period @xmath217 is an integer of the atomic spacing , so only fine sw structure remains , resulting in an _ anti - resonance _ and in the strongest inhibition of the locsiton .
as we mentioned above , while the quantization of locsitons in finite arrays can be readily analyzed analytically within nna , see the previous sections [ sec3 ] and [ sec5 ] , the situation with fully - interacting arrays ( fia ) presents a challenge for an analytical treatment .
let us briefly outline general analytical and numerical results obtained so far .
* a fia locsiton band is not symmetric with respect to the atomic resonance , @xmath120 ; it is shorter by the factor @xmath218 on the anti - lorentz side , see ( [ eq4.5 ] ) , as opposite to the nna .
* respectively , fia resonances are grouped tighter on the anti - lorentz side of the band , albeit their number is the same as for nna . near the lorentz side of the band ,
the nna - predicted resonances coincide more closely with those obtained by fia numerical calculations .
the major source of these effects is the first factor , i.e.a strongly asymmetric ( with respect to the detuning frequency , @xmath146 ) shape of the dispersion relation , fig .
a more detailed mathematical consideration of this problem , including a very good _ analytical _ fit for the dispersion relation is found in appendix [ app ] below .
less significant , although interesting as far as the eigenmodes of 1d arrays are concerned , is the fact that simple nna eigen - wavenumbers ( [ eq5.3 ] ) , @xmath219 , obtained based on the nna boundary conditions @xmath193 [ eq . ( [ eq3.3 ] ) ]
are not exact anymore .
one has to use now more extended , `` beyond - the - boundary '' conditions ( [ eq3.1 ] ) , which are the signature of fia , whereby @xmath220 instead of just two end - points in nna . to explore the problem ,
let us simplify it first by considering only non - dissipating atoms , @xmath221 , and rewrite the full - interaction dispersion relation ( [ eq4.1 ] ) for this case as : @xmath222 assuming @xmath128 real .
the equation for the field is written then as : @xmath223 with the boundary conditions ( [ eq6.1 ] ) .
the plot @xmath128 for real @xmath126 s due to the dispersion relation ( [ eq6.2 ] ) is shown in fig .
[ fig3 ] with the solid curve .
the new qualitative difference now between fia , ( [ eq6.2 ] ) , and the nna case , ( [ eq4.2 ] ) without dissipation , i.e. @xmath224 , is as follows . with @xmath225 ,
the nna equation has _ only real _
solutions for @xmath126 , whereas eq .
( [ eq6.2 ] ) , even within the locsiton band , @xmath226 , aside from real solution for @xmath227 , has , as one can show , an _
infinite _ number of _ complex _ solutions , @xmath126 , for _ each _ single real solution @xmath228 ( i.e. for the same @xmath229 ) .
all of them , in addition to fast spatial oscillation terms , have a rapidly rising / falling exponential factor .
these exponential modes are negligibly small almost over the entire array length , if @xmath100 , and they need to be accounted for only very near the end - points of the array , where they are instrumental in zeroing out the field and excitation at the points @xmath230 and @xmath231 .
let us illustrate the formation of those exponential ( or _ evanescent _ ) modes and their role in boundary conditions for the _ two_-near - neighbors approximation , _ 2-nna _ , whereby the field eq .
( [ eq6.3 ] ) becomes : @xmath232 = 1 , \label{eq6.4}\ ] ] with the boundary conditions for _ two _ pairs of end - points : @xmath233 the dispersion relations approximating ( [ eq6.2 ] ) will read now as : @xmath234 = d(q ) , \label{eq6.6}\ ] ] see fig .
[ fig3 ] , long - dashed curve , and the locsiton band is determined by @xmath235 the real solutions @xmath228 of ( [ eq6.6 ] ) are those for which @xmath236 ; they give rise to strata modes of eq .
( [ eq6.4 ] ) , @xmath237 however , having in mind that @xmath238 , one can readily see that ( [ eq6.6 ] ) has also solutions with @xmath183 , i.e. those that correspond to exponential modes , with complex @xmath239
. introducing for those modes @xmath240 with real @xmath188 , we obtain from ( [ eq6.6 ] ) that for each given real @xmath228 the exponent @xmath188 is determined by : @xmath241 and the respective exponential mode is @xmath242 thus , these 2-nna modes are antiferromagneticlike strata , modulated by fast exponents . indeed , since @xmath243 , we have @xmath244 . since the exponential , or evanescent , modes have so short `` tails ''
, they produce relatively small correction for the respective eigen - wavelengths , @xmath245 , and for @xmath246 , compared to the nna oscillatory modes , ( [ eq5.2 ] ) , ( [ eq5.3 ] ) , so that we can look for the @xmath245 corrected for 2-nna at the points of resonances as : @xmath247 with the correction @xmath248 , similarly to , e.g. , oscillations in a violin string with a `` soft '' suspension at its ends .
to find @xmath249 of the @xmath201-th resonance , we seek a solution for @xmath105 as a sum of two modes : oscillatory and exponential ones .
then , for symmetric modes , i.e. with odd @xmath201 , we have a full solution written as @xmath250 ( -1)^n , \nonumber \\ & & \mbox{with}\quad \bar{n } = n - \frac{n + 1}{2 } , \label{eq6.13}\end{aligned}\ ] ] where @xmath251 is a constant . for anti - symmetric modes , with even @xmath201 , we have : @xmath252 ( -1)^n
. \label{eq6.14}\ ] ] using now conditions ( [ eq6.5 ] ) , we can write for the points @xmath253 and @xmath254 , respectively , the following equations for the symmetric modes ( [ eq6.13 ] ) : @xmath255 for the anti - symmetric modes ( [ eq6.14 ] ) , one has to replace @xmath256 and @xmath257 functions in ( [ eq6.15 ] ) , ( [ eq6.16 ] ) with @xmath258 and @xmath259 functions , respectively . from these two equations we can compute @xmath249 and @xmath251 . indeed , approximating @xmath260 in ( [ eq6.15 ] ) and ( [ eq6.16 ] ) or in the respective equations for anti - symmetric modes , since @xmath261 , we obtain two equations for @xmath249 and @xmath251 , which are readily solved for @xmath249 and @xmath251 for _ both _ symmetric and anti - symmetric modes as : @xmath262 , \label{eq6.18 } \\ & & \mbox{with}\quad q_\mathrm{nna } = \frac{\pi k}{n + 1 } \nonumber \\ & & \mbox{and}\quad c = q_\mathrm{nna } \delta_k ( -1)^{\bar{k}/2 } e^{-\chi(n+1 ) } , \label{eq6.19}\end{aligned}\ ] ] where from ( [ eq6.10 ] ) , @xmath263 $ ] , while @xmath264 for symmetric modes , and @xmath265 for the anti - symmetric ones .
it is worth noting that the boundary conditions ( [ eq6.5 ] ) at @xmath266 are now satisfied automatically because of our choice of the coordinate @xmath267 in ( [ eq6.13 ] ) and ( [ eq6.14 ] ) . for lw resonances , @xmath268 , ( [ eq6.18 ] ) reduces to @xmath269 which is the maximum magnitude of @xmath249 ,
whereas for sw resonances , @xmath270 , ( [ eq6.18 ] ) reduces to @xmath271 which is substantially smaller than the lw correction ( [ eq6.20 ] ) .
one can readily see that @xmath249 is a monotonically decreasing function of @xmath201 . in the middle of the locsiton band , @xmath272
, we have @xmath273 interestingly , a fairly good fit to ( [ eq6.18 ] ) is provided by a much simpler formula : @xmath274 . \label{eq6.23}\ ] ] now , once the correction @xmath249 is found , one can substitute @xmath275 , ( [ eq6.12 ] ) , into the dispersion relation ( [ eq6.6 ] ) and calculate the respective frequency detuning for the @xmath201-th resonance , @xmath276 , with @xmath277 being the closest to the lorentz
lorenz resonance , i.e. a lw mode , and @xmath278 closest to the anti - lorentz edge of the locsiton band , a sw mode .
a way to extend the 2-nna approximation to a full - array interaction is to apply the result ( [ eq6.18 ] ) for the correction of the eigen - wavenumber , @xmath279 , but use it now in the full - blown dispersion relation ( [ eq6.2 ] ) , instead of the 2-nna relation ( [ eq6.6 ] ) , to calculate the resonance detuning @xmath197 .
the other avenue , of course , is to seek for higher order approximations .
for example , one can take into account two edge sets of _ 3 points _ each , i.e. similarly to ( [ eq6.5 ] ) , stipulate 3-nna : @xmath280 following the same route as for 2-nna case , we obtain now , similarly to ( [ eq6.10 ] ) , a _ second order _
equation for the _ exponential eigenmode
_ complex wavenumber , @xmath281 , for any given oscillation wavenumber , @xmath228 , as @xmath282 [ we leave it untransformed to @xmath283 , unlike ( [ eq6.9 ] ) , ( [ eq6.10 ] ) , since in 3-nna case the solution becomes more complicated than ( [ eq6.11 ] ) , and that transformation does not simplify the problem . ] the higher - order equation for the _ exponential eigenmodes _ ensue the choice of a higher number of end - points in the boundary conditions .
it has to be noted , however , that for specifying parameters of oscillating eigenmodes , the increase of the precision by accounting for higher - order exponential modes is of very limited , if purely academic , significance . essentially , for large arrays , @xmath100 , the combination of the nna for predicting the wavenumbers @xmath228 , ( [ eq6.12 ] ) or ( [ eq5.3 ] ) , with these @xmath228 s used then in ( [ eq6.2 ] ) to calculate the resonance eigen - frequencies @xmath146 ,
does already a good job .
a further step in increasing the precision provided by the 2-nna corrections to eigen - wavenumbers , is to again use the corrected @xmath228 s in ( [ eq6.2 ] ) for the same purpose , and it is more than sufficient .
a special case is a small array , e.g. @xmath284 , whereby it is actually preferable to simply solve the general 1d - array equations analytically , similarly to @xcite in the case of @xmath108 .
a fundamental effect of self - induced cancellation of local - field suppression emerges near the atomic resonance , @xmath120 , at certain `` magic '' numbers @xmath81 .
if @xmath143 , the _ uniform _ ( lorentz lorenz ) lf ( [ eq3.4 ] ) at @xmath285 is very small , @xmath286 [ eq . ( [ eq3.5 ] ) ] .
however , in the near - neighbor approximation , if @xmath287 where @xmath288 is a `` magic '' number within nna , locsitons at @xmath120 show canceled lf suppression at some atoms . the highest cancellation is attained at @xmath120 and @xmath289 , with the atomic dipoles lining up as in ( [ eq4.12 ] ) , whereby the lf amplitude at odd - numbered atoms is at maximum , @xmath290 , and the enhancement @xmath291 could be a few orders of magnitude ; the lf at the two other atoms almost zeros out .
the self - induced cancellation effect is produced by a standing wave with the nodes at atoms with even numbers .
this results in a `` virtual '' size - related resonance at @xmath292 ( i.e. at the exact atomic resonance ) , which manifests itself in the enhancement ( the resonant peak transpires in @xmath293 _ vs _ @xmath146 ) .
thus , _ the nature of magic numbers is the coincidence of the atomic resonance with one of the size - related locsiton resonances_. the effect holds also for the interaction of each atom with _ all _ other atoms ( [ eq3.1 ] ) ; the magic number in ( [ eq7.1 ] ) takes on a `` devilish '' likeness here : @xmath294 .
it is due to the fact that for @xmath120 and @xmath295 , the first root @xmath296 of eq .
( [ eq4.1 ] ) with zero r.h.s .
, @xmath297 has its property of @xmath298 almost coinciding with a rational number , @xmath299 @xmath300 ( @xmath301 ) , see appendix [ app ] below .
the locsiton wavelength is @xmath302 , and the lowest integer of @xmath303 to exactly match an integer of @xmath7 is @xmath304 , which requires 14 atoms .
we have now @xmath305 , with enhancement @xmath306 .
as we have shown in @xcite , 2d lattices can also exhibit `` _ _ magic shapes _ _ '' with similar properties ; the simplest one within nna is a six - point star with an atom at its center , thus making the total number of atoms again @xmath206 .
more details on magic shapes for 2d lattices will be discussed by us elsewhere .
if the locsitons are waves , can they be excited outside the driving field area ? and
thus travel away from the lorentz lorenz uniform local field area ?
how far can they travel before being extinguished ?
how fast do they propagate ?
having in mind their dissipation , what is the size of a finite array to have well pronounced standing waves and size - related resonances ?
of course , the locsitons can propagate in an array even in the areas unaccessible for the external ( laser ) field .
if the spatial profile gradient of the driving wave is large enough , a lf excitation can be found beyond the driving field area ; the terminological irony here is that the _ local_-field phenomenon is due to a _
nonlocal _ interaction , and the locsitons can propagate away from their origination point .
this may happen when the external laser field is non - uniform and has a large gradient , e.g. when one entirely screens out a part of the array by imposing a `` sharp knife '' over the array .
even simpler and more transparent case is when only an end - point of a semi - infinite 1d array [ say , with @xmath307 in ( [ eq3.1 ] ) ] is illuminated by a laser field _ via _ a pin - hole .
( [ eq3.1 ] ) has then non - zero ( unity ) r.h.p.for @xmath307 only , and the r.h.p .
zeroes out at all the other points . in this case
, _ no _ lorentz lorenz local field exists for any atom at @xmath308 , and the only field and atomic excitation passed along the array of atoms will be locsitons .
this may be the best way to excite and observe `` pure '' locsitons , with them not being masked by any external , averaged , mean , etc .
with such an arrangement , the 1d array ( or a sufficiently thin atomic `` cylinder '' or carbon nanotube ) becomes a true and effective waveguide for locsitons , capable of transmitting non - diffracting radiation and atomic excitation from one location ( e.g. in opto - electronic circuits ) to another .
the main issue here is how far the locsiton can propagate .
one can investigate it by studying the dispersion relation ( [ eq4.1 ] ) or ( [ eq4.2 ] ) , which predicts not only the locsiton wavenumbers , @xmath309 , for any given frequency detuning @xmath146 , but also their dissipation depth or distance ( in terms of numbers of atoms ) for each wavenumber , as @xmath310 where @xmath311 . using only the nna dispersion relation ( [ eq4.2 ] ) , amply sufficient here , since the calculation of the dissipation distance does nt require the same precision as for the real wavenumbers , we find out that the exact solutions for @xmath296 and @xmath312 are determined by : @xmath313 and @xmath314 from ( [ eq8.1 ] ) , for very small dissipation , i.e. when @xmath129 , we have , as expected , @xmath315 the dissipation length in terms of the numbers of atoms , @xmath310 , is readily found from ( [ eq8.2 ] ) . at the exact atomic resonance , @xmath285 , if @xmath143 , we have : @xmath316 in general , in the most part of the locsiton band ( except for the edge areas , @xmath317 ) , we find that @xmath318 finally , at the locsiton band edges , defined as @xmath319 , ( [ eq8.2 ] ) results in @xmath320 and it remains roughly the same as @xmath321 reaches @xmath322 .
thus , in the most of the locsiton band , the dissipation length , in terms of the numbers of atoms , is @xmath323 , which also determines the maximum size of array to still enable the size - related resonances , in agreement with section [ sec5 ] .
let us address now the characteristic velocities of the locsitons .
neglecting decay in ( [ eq4.2 ] ) ( @xmath129 ) , the group velocity of locsitons , @xmath324 , is found as : @xmath325 hence @xmath326 where @xmath327 is the rabi frequency of the self - interacting array ( [ eq4.6 ] ) , and @xmath328 is a characteristic rabi speed , @xmath329 which could be even slower than a typical speed of sound in condensed matter .
this effect can be used e.g. for developing nano - size delay lines in molecular computers and in optical gyroscopes .
the lw - locsiton phase velocity is @xmath330
so far we studied linear ( in field ) excitations of atomic 1d arrays .
the nonlinear interactions open door to a huge new landscape of effects .
the simplest and very generic nonlinearity in a two - level system is the saturation of its absorption ( [ eq2.4 ] ) , ( [ eq2.5 ] ) , which translates into a nonlinear response of the atom polarization to the local field ( [ eq2.6 ] ) ; this , in turn , nonlinearly affects the strength of interaction between atoms ( [ eq2.9 ] ) .
this represents a rare case whereby the nonlinear change ( decrease ) of absorption directly affects the eigen - frequencies of the system ( 1d array ) , by directly reducing the interaction .
many nonlinear effects are brought up in short pulse modes , e.g. discrete solitons , to be considered by us elsewhere .
however , spectacular effects , such as hysteresis and optical bistability , emerge even in _ cw _ mode . to write a set of nonlinear equations for an infinite 1d array , we first scale all fields to the characteristic saturation field , @xmath331 ( [ eq2.5 ] ) , instead of scaling to the incident field , @xmath332 , so that the dimensionless local fields at the @xmath204-th atom , @xmath333 , and the incident field , @xmath334 , are @xmath335 and the nonlinear counterpart of ( [ eq3.1 ] ) and ( [ eq3.2 ] ) for the array is written now as : @xmath336 where @xmath113 is defined by ( [ eq3.4 ] ) , so it covers either incident polarization .
the nonlinear counterpart of the nna equation ( [ eq3.3 ] ) is now @xmath337 & & \mathcal{e}_0 = \mathcal{e}_{n+1 } = 0 .
\nonumber\end{aligned}\ ] ] none of this is an easy object even for numerical solution , let alone analytical one .
we have , however , derived in @xcite a closed analytical solution for the nonlinear mode in the most fundamental system an array of just two atoms and found bistability and hysteresises in such a mode .
in this paper , more as a matter of illustration , we find an analytical multistable solution and hysteresis for the simplest case of lorentz lorenz uniform field ; however , our computer simulations showed that multistability and hysteresises exist in the vicinity of each size - related resonance in the system .
for the sake of simplicity , we consider here the case of lorentz lorenz uniform mode . in the near - neighbor approximation ( [ eq9.3 ] )
[ the fia results will essentially differ only by a factor of @xmath109 ] , we have @xmath339 , and thus the nonlinear equation for the uniform local field , @xmath340 , is @xmath341 = x , \label{eq9.4}\ ] ] or for the field intensity , @xmath342 , @xmath343 ^ 2 + \delta_\mathrm{ll}^2}{(1 + \delta^2 + y ) ^2 } = x^2 . \label{eq9.5}\ ] ] it can be readily seen that the strongest nonlinear effect emerges near the lorentz lorenz resonance , @xmath344 . assuming small losses , @xmath345
, we can analyze the threshold of the multistability and hysteresis mode by stipulating that @xmath346 , @xmath347 , and thus ( [ eq9.5 ] ) can be further simplified as @xmath348 / \delta_\mathrm{ll}^4 \approx x^2 . \label{eq9.6}\ ] ] the threshold for the multistable solution @xmath349 of this equation is determined by the condition @xmath350 , which results in the critical requirement that @xmath351 amazingly , as one recalls that @xmath352 and @xmath295 , the critical ( threshold ) driving intensity @xmath353 to initiate multistability and hysteresis could be orders of magnitude lower than the saturation one , @xmath45 , which is mostly due to resonant nature of the effect that emerge in the vicinity of the lorentz
lorenz resonance .
thus , the required saturation nonlinearity is indeed just a slight perturbation to the resonant linear mode .
this should be of no big surprise , since the nature of the effect is the same as in many other so called _ vibrational hysteresises _
@xcite in resonant nonlinear systems , from pendulum @xcite to electronic circuits @xcite , to optical bistability in a fabri
perot resonator @xcite , to a cyclotron resonance of a slightly relativistic electron @xcite .
in all these nonlinear resonances , it is enough for a narrow resonant curve to be tilted beyond its resonance width to reach multistability by becoming a `` pisa - like tower '' .
the formation of bistability and hysteresis is depicted in fig .
[ fig5 ] , where the resonant curve @xmath354 is shown for different driving @xmath334 for the full equation ( [ eq9.5 ] ) . _
vs _ detuning @xmath146 for different normalized driving amplitudes @xmath352 for the case of @xmath355.,width=325 ] the formation of a tri - stable solution results in the so called bistability , whereby only the states with the maximum and minimum intensities are stable , whereas the middle branch of the solution is absolutely unstable . to an extent ,
it is similar to one of the hysteretic patterns for the local field found for the case of the local field @xcite and scattered light @xcite of just two atoms ; the hysteretic resonance here is produced by a lorentz - like , or ferromagneticlike , excitation of atoms with lw locsitons .
this effect is also reminiscent of optical intrinsic bistability for uniform lf in dense materials predicted earlier in @xcite and observed experimentally in @xcite .
the above result for hysteresis and bistability was found for the uniform , lorentz , field , i.e. near the lorentz
lorenz resonance , using a simple field distribution .
however , it is clear from the above argument with a `` pisa - like tower '' , a similar hysteretic effect can be expected in the close vicinity of _ every _ size - related locsiton resonance within the locsiton band .
some of the most interesting hysteresises , though , are excited for the sw locsitons , when the nearest atoms counter - oscillate in an antiferromagneticlike fashion , as shown in @xcite for the anti - lorentz band edge for two atoms .
this hysteresis results in a `` split - fork '' bistability and will be discussed by us in application to long arrays elsewhere .
it is only natural to expect the effects studied here to manifest themselves in other finite systems of discrete resonant oscillators that are externally driven and interact with each other . on the one hand , this may help to demonstrate and verify some major results reported here in much simpler and easily handleable settings ; on the other hand , this may bring up new features in these other systems that escaped attention in seemingly well developed and researched fields .
we consider here a few such systems . perhaps , the most classical example would be a mechanical finite array of identical physical pendulums with the same individual resonant frequencies , weakly coupled to each other ( say , by using weak strings between neighboring pendulums ) , with each pendulum driven independently by an external feed ( say , _ via _ em solenoids ) with the same phase for all of them . by tuning the frequency of the driving feed around the resonant frequency of the pendulums
, one may expect to observe stratified excitation of the pendulums , from long - wave , ferromagneticlike strata , to short - wave , antiferromagneticlike ones .
one can also expect to observe `` magic '' numbers and related effect of non - moving pendulums surrounded by strongly excited ones .
( it is worth noting that we are talking here about a _ cw _
motion , in contrast to the well known effect in two coupled non - driven pendulums , whereby the pendulums periodically alternate periods of zero - excitation in one and strong excitation in the other one , with the frequency of the alternation being inversely proportional to the coupling . ) the major effect here would be again due to locsitons , and the major critical condition for their excitation would be similar to the condition on the strength of interaction , @xmath58 , ( [ eq2.10 ] ) , to exceed a critical one .
the driven pendulum array would be a great classroom - demonstration tool .
however , although purely mechanical , it might take some effort to implement , due to the need of an independent feed to each pendulum . from that point of view ,
a much simpler , easily implementable , controllable , recordable , and versatile system could be an electronic array of individual resonant circuits , electronically coupled to each other .
a long while ago , similar systems have been used as transmission lines , whereas a finite set of circuits has also been of interest for such applications as bandpass filters .
however , the detailed picture of behavior of individual circuits inside such systems have apparently not attracted too much attention , with a few reasons for that .
the main difference between these systems and the ones proposed by us , is that we need an independent feed with the same phase for each individual circuit , which may be arranged , by e.g. using individual cable for each one of them .
the coupling between the individual circuits can be engineered in such a way that one can arrange strictly near - neighbor interaction , two - near - neighbors interaction , etc . besides , the contribution of each neighboring circuit can be independently controlled , so the term @xmath356 can now be changed to any desirable function of the neighbor spacing .
a close example , which might have important implications for large radio - frequency antenna arrays , for e.g. radio - astronomy applications , as well as in multi - dish radar systems , is the interaction of radiators in those arrays , if the strength of this interaction exceeds some critical value .
since in the case of electronic circuits the polarization of the feed and the spatial anisotropy as in ( [ eq2.2 ] ) is not a factor anymore , the dipole - like interaction is simplified , as its spatial polarization form - factor @xmath96 ( as in section [ sec3 ] ) can be now dropped , and the equations of motion ( [ eq3.1])([eq3.4 ] ) can be simplified . furthermore , since now the direction of a `` dipole '' with respect to the `` incident polarization '' is not a factor , one can arrange a _ loop _ or _ ring _ of those circuits , instead of a linear 1d array , which allows for periodic boundary condition instead of zero boundary conditions as e.g. in ( [ eq3.1 ] ) .
this would greatly simplify the theory and comparison with the experiment , on the one hand , and allow for new interesting effects in ring arrays ( such as e.g. greatly enhanced sagnac effect and related gyroscope applications ) , to be discussed by us elsewhere , on the other hand . an interesting and exotic opportunity with 1d arrays , and especially ring arrays ,
is a possibility to develop a toy model of `` discrete space quantum mechanics '' , whereby a wave function is replaced by a set of oscillating fields at discrete spatial points , instead of a regular wave distributed in space .
this discrete - space quantum mechanics ( qm ) could be of interest for the theory of certain systems , as well as from fundamental point of view ; for example , the ring arrays then would allow to build a theory of a bohr - like `` discrete - qm h atom '' , with finite number of primary quantum numbers , unlike an infinite numbers of quantum levels as in a regular - qm h atom .
a very interesting analogy , especially from the application point of view , is _ magnetic spin resonance _ , in particular _ nuclear magnetic resonance _
( nmr ) @xcite , as well as magnetic resonances in finite spin systems , in particular so called molecular magnets @xcite .
they have some common points with the interacting systems considered here , in particular , two - level nature of resonances in both systems . in optical domain , an observation of the effect discussed here
can be done in a few ways .
nanostrata and locsitons can be observed either _ via _ size - related resonances in scattering of laser radiation , or _ via _ x - ray or electron energy loss spectroscopy of the strata .
the effect has promising potential for molecular computers @xcite and nanodevices .
the major advantage of locsitons _ vs _ electrons in semiconductors is that they are not based on electric current or charge transfer .
this may allow for a drastic reduction of the size limit for computer logic elements currently based on metal - oxide - semiconductor technology , which may suffer from many irreparable problems on a scale below 10 nm .
as such , locsiton - based devices could be an interesting entry into the field , as complimentary or alternative to emerging technologies like plasmonics @xcite or spintronics @xcite .
they can offer both passive ( e.g. transmission lines and delays ) , and active elements , e.g. for switching and logics .
a ring - array may be used as a basis for a sagnac - locsiton - based gyroscope ; low locsiton velocity may allow for a high sensitivity in a small ring .
another promising application of locsitons could be biosensing devices , where target - specific receptor molecules either form a locsiton - supporting lattice or are attached to its sites ; a localized locsiton occurs whenever a target biomolecule attaches to a receptor . finally , exciting opportunities exist in atomic arrays and lattices with inverse population created by an appropriate ( e.g. optical ) pumping , which may lead to a laser - like locsiton stimulated emitter ( `` locster '' ) , to be discussed elsewhere .
in conclusion , in this study of strong stratification of local field and dipole polarization in finite groups of atoms predicated by us earlier @xcite , we developed a detailed theory of the phenomenon in one - dimensional arrays of atom or resonant particles .
in strong departure from lorentz lorenz theory , the spatial period of those strata may become much shorter than the incident wavelength . by exploring nanoscale elementary excitations , locsitons , and resulting size - related resonances and large field enhancement in finite arrays of atoms
, we showed that their spatial spectrum has both long waves , reminiscent of ferromagnetic domains , and super - short waves corresponding to the counter - oscillating neighboring polarizations , reminiscent of antiferromagnetic spins .
the system also exhibits `` hybrid '' modes of excitation that have no counterpart in magnetic ordering , and are more representative of the effect .
our theory that goes beyond ising - like near - neighbor approximation and describes the excitation whereby each atom interacts with all the other atoms in the array , reveals the existence of infinite spectrum of `` exponential '' , or `` evanescent '' eigenmodes in the such arrays .
we explored the phenomenon of `` magic '' numbers of atoms in an array , whereby resonant local - field suppression can be canceled for certain atoms in an array .
we also demonstrated the existence of nonlinearly induced optical bistability and hysteresis in the system .
we discussed a stratification effect similar to that in atomic arrays , which may exist in broad variety of self - interacting systems , from mechanical ( pendulums ) to electronic circuits , to radar arrays , and to the nuclear magnetic resonance .
we pointed out a few potential applications of the atomic and similar arrays in such diverse field as low - losses nano - elements for optical computers , small - size sagnac - effect - based gyroscopes , and bio - censors .
the authors appreciate helpful discussions with v. atsarkin and r. w. boyd . this work is supported by afosr .
in this appendix we consider mathematical properties of the fourier series in the dispersion relation ( [ eq4.1 ] ) , i.e. the function @xmath357 which is a generalized form of ( [ eq4.1 ] ) [ in ( [ eq4.1 ] ) , @xmath140 due to the chosen geometry of the problem 1d array of point dipoles ] , but we will restrict ourselves to natural numbers @xmath358 .
note , for example , that in the case of infinite `` dipole strings '' , parallel to each other and equidistantly arranged in a 2d plane , @xmath359 , whereas for `` dipole planes '' parallel to each other and equidistantly arranged in the 3d space , @xmath360 .
our main purpose here is to find a closed finite analytical form of the function @xmath361 that originated the fourier series ( [ eqa.1 ] ) , and when it is impossible in terms of more or less regular analytical functions , at least find a closed finite derivative @xmath362 with minimum derivative order @xmath49 , which will help us to analyze in great detail the behavior of the function @xmath361 in its physically interesting points , e.g. at @xmath363 , the ratio of its min / max values @xmath364 , etc .
those derivatives are actually the main tool in our search . indeed , differentiating @xmath361 in ( [ eqa.1 ] ) @xmath365 times ,
one obtains @xmath366 and @xmath367 recalling that @xmath368 and @xmath369 , we recognize the sums @xmath370 in ( [ eqa.2 ] ) and ( [ eqa.3 ] ) as taylor expansions of @xmath371 . in the case of even numbers @xmath358 ,
the sum ( [ eqa.2 ] ) in the interval @xmath372 ( which is of the most interest ) yields : @xmath373 it is discontinuous at @xmath374 .
even numbers @xmath358 are thus a `` lucky '' case ; the result ( [ eqa.4 ] ) is easily integrated to restore function @xmath361 . in a physically interesting case , @xmath359 , by integrating ( [ eqa.4 ] ) once we have @xmath375 , \nonumber\end{aligned}\ ] ] which is a continuous ( but not smooth ) function with @xmath376 and @xmath377 , so that @xmath378 . as to the closed integrability ,
one is not as lucky with odd numbers @xmath358 ; the summation of ( [ eqa.3 ] ) yields : @xmath379 , \label{eqa.6 } \\ & & \quad\mbox{i.\,e.}\quad \sigma_1 ( q ) \to
\infty \quad\mbox{at}\quad q = 2m\pi , \nonumber\end{aligned}\ ] ] which can not be integrated in simple known analytical functions , but it gives a good analytical tool to analyze the behavior of @xmath361 . in the case of most interest to us , @xmath140 , we have that near the maximum of @xmath380 at small wavenumbers @xmath381 where @xmath382 , whereas near the minimum of @xmath380 , i.e.near @xmath170 , we have @xmath383 notice that here @xmath384 , and @xmath385 at @xmath386 . for the magic numbers in the case of full interaction of individual atoms with all the other atoms in the array , it is important to know zeros of the function @xmath387 . as it has been mentioned before
[ see ( [ eq7.2 ] ) and related text ] , it turns out that the value of the ratio @xmath388 for the first positive root of the equation @xmath389 is very close to a small rational number : @xmath390 for all practical purposes , a good approximation for @xmath387 is provided by : @xmath391 } { \ln(s + \delta s ) } \cdot \frac{\sin^2 ( q/2)}{4 } \right\ } , \label{eqa.10}\end{aligned}\ ] ] where @xmath392 .
it coincides with the results provided by numerical summation of ( [ eqa.1 ] ) with @xmath140 with the precision better than @xmath393 of @xmath394 , and their zeros coincide with each other and with ( [ eqa.9 ] ) with precision better than @xmath395 . m. born and e. wolf , _ principles of optics _ , ( cambridge univ . press , cambridge , 1997 ) , sections 2.3 and 2.4 ; j. van kranendonk and j. e. sipe , in _ progress in optics _ , vol .
15 , ed . e. wolf , ( n. holland , amsterdam , 1977 ) , p. 245 .
a. yariv , _ quantum electronics _ , ( wiley , new york , 1989 ) ; v. s. butylkin , a. e. kaplan , yu . g. khronopulo , and e. i. yakubovich , _ resonant nonlinear interactions of light with matter _ , ( springer - verlag , new york , 1989 ) ; v. s. butylkin , a. e. kaplan , and yu . g. khronopulo , sov .
jetp * 32 * , 501 ( 1971 ) ; c. m. bowden and j. p. dowling , phys . rev .
a * 47 * , 1247 ( 1993 ) .
a. k. sarychev and v. m. shalaev , _ electrodynamics of metamaterials _ , ( world scientific , hong kong , 2008 ) ; w. a. murray and w. l. barnes , adv . mater .
* 19 * , 3771 ( 2007 ) ; _ surface plasmon nanophotonics _ , eds .
m. l. brongersma and p. g. kik ( springer , new york , 2007 ) ; y. fainman , k. tetz , r. rokitski , and l. pang , opt . photonics news , * 17 * , no . 7 , 24 ( 2006 ) . | the major assumption of the lorentz lorenz theory about uniformity of local fields and atomic polarization in dense material does not hold in finite groups of atoms , as we reported earlier [ a. e. kaplan and s. n. volkov , phys .
rev .
lett .
* 101 * , 133902 ( 2008 ) ] .
the uniformity is broken at sub - wavelength scale , where the system may exhibit strong stratification of local field and dipole polarization , with the strata period being much shorter than the incident wavelength . in this paper , we further develop and advance that theory for the most fundamental case of one - dimensional arrays , and study nanoscale excitation of so called `` locsitons '' and their standing waves ( strata ) that result in size - related resonances and related large field enhancement in finite arrays of atoms . the locsitons may have a whole spectrum of spatial frequencies , ranging from long waves , to an extent reminiscent of ferromagnetic domains , to super - short waves , with neighboring atoms alternating their polarizations , which are reminiscent of antiferromagnetic spin patterns . of great interest is the new kind of `` hybrid '' modes of excitation , greatly departing from any magnetic analogies .
we also study differences between ising - like near - neighbor approximation and the case where each atom interacts with all other atoms in the array .
we find an infinite number of `` exponential eigenmodes '' in the lossless system in the latter case . at certain `` magic '' numbers of atoms in the array ,
the system may exhibit self - induced ( but linear in the field ) cancellation of resonant local - field suppression .
we also studied nonlinear modes of locsitons and found optical bistability and hysteresis in an infinite array for the simplest modes .
this paper is now published , with very minor changes , in phys .
rev .
a * 79 * , 053834 ( 2009 ) . |
transition metal oxides show remarkable physical properties , such as colossal magnetoresistance in manganites@xcite , or superconductivity in cuprates@xcite , which after decades of researches , remain only partly understood .
this reflects our understanding of the structural , electronic and magnetic phenomena , which is well established only in the limit where the systems show localized or itinerant electron behavior@xcite .
doping with electrons or holes is the canonical method to explore the intermediate regime .
for example , in r@xmath8d@xmath9mno@xmath0 manganese perovskites ( r : trivalent cation , d : divalent cation ) , the mn valence is formally intermediate 3+x , affecting the electrical conductivity and leading to a complex phase diagram
. however , even undoped compounds display an intriguing evolution of their magnetic properties as the geometry and coordination are changed .
for example , in the rmno@xmath0 and dmno@xmath0 end members of the r@xmath8d@xmath9mno@xmath0 series , mn has undoubtedly well localized electrons , making these compounds prototype antiferromagnetic ( af ) mott insulators .
their magnetic properties fit a heisenberg picture , in which superexchange ( se ) interactions couple high spin mn@xmath10@xmath11 and mn@xmath12@xmath13 ions at @xmath14 and @xmath15 respectively@xcite .
this can be contrasted to the very unusual magnetic properties of some hexagonal manganese halides containing single valent mn ions .
csmnbr@xmath0@xcite and csmni@xmath0@xcite compounds containing s=5/2 high spin mn@xmath6 ions arranged in infinite strings of face - sharing mno@xmath5 octahedra parallel to the c axis , from which exchange frustration is expected between the direct and the super - exchange interactions of different sign within the chains .
exchange frustration in the basal plane also bring about chirality of the magnetic state and a spin dynamic corresponding to new universality classes@xcite .
one may be tempted to conclude that these unusual magnetic properties arise from the quasi one - dimensional nature of the exchange , combined with easy axis anisotropy .
however , isostructural bamno@xmath0 ( s=3/2 ) does not show such effects , but it has instead a simple antiferromagnetic structure and rather classical magnetic behavior with the ordered moment value saturating at low temperatures@xcite . in order to separate the influence upon the magnetic properties of the lattice geometry from more conventional chemical effects , the study of polymorphic compounds , which display different stable or metastable structures with the same chemical composition is particularly valuable . as discussed by nagas and roth@xcite ,
the stability of the structures adopted by @xmath16 compounds depends on the tolerance factor @xmath17 , where @xmath18 is the average ionic radius of the a - site cations , while @xmath19and @xmath20 are those of the b - site metal and of the anion ( oxygen , halogen or chalcogen ) .
most manganites have @xmath21 , and crystallize in orthorhombic or rhombohedral perovskite structures showing cooperative tilting of the corner sharing mno@xmath5 array , with respect to the ideal cubic perovskite . for sufficiently large @xmath18 ,
@xmath22 materials adopt a different hexagonal symmetry , as in the case of bamno@xmath0 and the aforementioned halides . for @xmath23 ,
polytypes are obtained depending on the synthesis procedure@xcite .
these polytypes are characterized by the stacking sequence of the ao@xmath0 layers , which controls how the mno@xmath5 octahedra arrange .
octahedra are sharing corners across the central ao@xmath0 layer of _ .. abc .. _ sequences , while they share faces across _ .. aba .. _ type sequences .
for example , synthesized at ambient pressure in air , srmno@xmath0(@xmath24 ) has a four layer ( 4l ) _ abac _ stacking sequence of sro@xmath0 layers as shown in fig.[magtopo ] , whereas modified synthesis routines are required to obtain the cubic perovskite@xcite .
4l is a rather rare polytype among abx@xmath0 compounds , and represents an intermediate case between the hexagonal ( 2l ) polytype adopted by bamno@xmath0 , and the familiar perovskite structure of camno@xmath0 containing only corner sharing octahedra .
it is therefore interesting to study the magnetism in srmno@xmath0 in comparison with that of the perovkite and 2l hexagonal materials .
this could also help to shed light on the striking differences of magnetic behavior in isostructural csmnbr@xmath0 and bamno@xmath0 . in srmno@xmath0 ,
a complex magnetic behavior is expected from the exchange topology , which results from the arrangement of pairs of face sharing mno@xmath5 linked connected by the octahedra corners . in this structure , direct mn - mn exchange interactions between mn@xmath12ions in face sharing octahedra ( @xmath25 ) compete with the @xmath26 af - se linkages through the face shared by mno@xmath5 octahedra ( @xmath27 ) as in bamno@xmath0 , and @xmath28 af - se linkages across the common oxygen of corner sharing mno@xmath5 octahedra ( @xmath29 ) , like in camno@xmath0 . as in several hexagonal transition metal oxides or sulfides ,
@xmath25 is probably very strong , since the metal - metal distance is much lower than that observed in the corresponding intermetallic compounds containing the same transition metal element@xcite . as first noticed by battle _
_ et al.__@xcite , srmno@xmath0shows unusual magnetic properties : the magnetization data , which we have also measured ( see fig.[ppms ] ) show a transition above room temperature ( @xmath30 ) , but neutron powder diffraction ( npd ) demonstrates that this does not correspond to the af ordering .
af order sets in only at lower temperatures , where zero field cooled and field cooled @xmath31 curves deviate ( @xmath32@xmath33)@xcite . recently , raman spectroscopy was used to evidence a structural transformation in srmno@xmath0 below @xmath34@xcite . to explain the subtle difference observed by raman spectroscopy , the authors proposed a 4l to 6l transition , allowing the hexagonal symmetry to be retained . in this article
, we present a complete study of the magnetic and structural properties of srmno@xmath0 , using in combination electron microscopy , neutron powder diffraction ( npd ) and high - resolution synchtrotron x - ray powder diffraction ( hr - sxpd ) .
polycristalline samples of srmno@xmath0 were synthesized by solid - state reaction in air of a stoichiometric mixture ( 1:0.5 ) of srco@xmath0 and mn@xmath35o@xmath0 .
the powder was heated at 950@xmath36c during two weeks with intermediate grindings , then pelletized and sintered at 1000@xmath36c for 12 hrs .
this low - temperature synthesis was determined from the phase diagrams @xcite as the best route to obtain pure hexagonal 4l - structure without oxygen deficiency .
samples quality and stoichiometry were determined by preliminary neutron diffraction measurements . at room temperature ,
the reconstruction of the reciprocal space was carried out with a jeol 200 cx electron microscope and high resolution electron microscopy ( hrem ) images at room temperature were recorded using a topcon 002b microscope ( cs = 0.4mmm ) .
the electron diffraction patterns versus temperature were collected with a jeol 2010 electron microscope operating at 200kv and equipped with a double - tilt liquid n2 sample holder ( tilt @xmath37 , @xmath38 , and @xmath39 ) .
all the patterns were recorded under the same experimental conditions , i.e. the same exposure time , the same electron beam alignment and the same beam intensity and increasing the temperature from @xmath40 to @xmath41 , in steps of @xmath42 degrees , waiting for temperature stabilization before recording .
the three microscopes are equipped with energy dispersive analysis spectrometers ( eds ) , which allow determining the sr / mn ratio .
magnetic susceptibility was collected under a magnetic field of 1000 oe under zero - field cooled ( zfc ) and field cooled ( fc ) processes .
data were collected on warming between 2k and 400k at a sweep rate of 1k / min , using a vibrating sample magnetometer ( quantum design , ppms ) .
high resolution synchrotron and time - of - flight neutron powder diffraction data were collected at 100 and 350k , using the id31 beam line at the esrf , grenoble with a wavelength @xmath43 , and hrpd at the isis facility ( uk ) , respectively .
subsequent medium - resolution neutron diffraction data were collected on the general materials ( gem ) diffractometer of the isis facility to study the temperature dependence of the crystal structure and the magnetic structure .
data sets at temperature between @xmath44 and @xmath41 with @xmath45 steps were collected on warming using a standard he cryostat .
the diffraction patterns were used to refine the srmno@xmath0 structure by the rietveld method@xcite using the fullprof suite ( ref . ) .
symmetry analysis for the structural and magnetic phase transitions was performed using the basireps program , part of the fullprof suite as well as programs available on the bilbao crystallographic server @xcite .
the dc magnetic susceptibility of srmno@xmath0 between 2 and 400k is reported in fig.2 .
plot of the inverse suceptibility , not shown , indicates that a curie - weiss behavior is not obeyed in this temperature range , in agreement with earlier results @xcite .
however , our results below 400k are qualitatively different to what was previously reported .
below 380k , there is a clear upturn in the susceptibility followed by a relatively shallow decrease below 300k .
the upturn could be associated with the structural phase transition reported herein , and discussed in the following sections , as evidenced by x - ray and neutron diffraction .
however , it was impossible to track precisely the structural details as a function of temperature due to the extremely weak intensities of the superlattice reflections and therefore confirm unambiguously the correlation between structural and magnetic behaviors .
also , this upturn could be associated to short - range magnetic correlations within the face - sharing mno@xmath5 octahedra , as suggested by battle et al.@xcite . + at lower temperatures ( t@xmath46300k ) where @xmath47 decreases , the behavior is clearly reminiscent of antiferromagnets .
this is in agreement with previous work , in particular with the largely negative weiss temperature of -1210k derived by susceptibility measurements at high temperature@xcite .
the determination of the nel temperature from susceptibility alone is not straightforward due to the broad maximum observed in the variation of @xmath47 .
battle and co - workers identified a nel temperature of 278k from mossbauer spectroscopy on fe - doped samples@xcite .
our measurements suggest the same nel transition temperature since the susceptibility curves in field cooled ( fc ) and zero - field cooled ( zfc ) conditions slightly deviates at around 280k . moreover , this accident coincides with the onset of a long - range ordered antiferromagnetic structure evidenced by our neutron diffraction experiment , as discussed below . in this section ,
we report a detailed crystallographic characterization of srmno@xmath0 from electron microscopy and high - resolution x - ray and neutron powder diffraction experiments .
a large number of crystallites have been analyzed by eds and electron diffraction , confirming an homogeneous sr / mn ratio = 1 in the limit of the accuracy of the technique .
the reconstruction of the reciprocal space was carried out by tilting the sample around the crystallographic axes .
it evidences an hexagonal cell with @xmath48 ( @xmath49 is the parameter of the ideal cubic perovskite cell ) , @xmath50 and the extinction condition _
hh2@xmath51l = 2n_.
these results are consistent with the 4l polytype proposed for srmno@xmath0 as well as the space group @xmath1 previously reported@xcite .
the @xmath52 $ ] ed pattern at room temperature is given in fig.[tem1]a .
the quality of the crystallites was checked using hrem , in order to detect the presence of any intergrowth defects ( variations in the layer stacking along @xmath53 , i.e. the presence of different polytypes ) .
the lattice images , not shown , confirm that the stacking of the 4l structure is quite perfect .
the density of defects is very low ( of the order a few events for a nanometer - sized crystallite ) , which is rather rare in the hexagonal polytypes , allowing one to conclude that these defect do not significantly affect the symmetry .
+ the reciprocal space was reconstructed at @xmath40 , tilting especially around the @xmath54 axis .
one clearly observes two important points : first , the lattice parameters remains unchanged , especially @xmath55 , attesting that the 4l stacking mode is retained and second , the appearance of another system of weak reflections , _
l = no condition _ , which clearly violate the mirror @xmath56 symmetry .
this is illustrated comparing fig.[tem1]a and [ tem1]b with the evolution of the @xmath52 $ ] ed pattern between rt and @xmath57 . at this step
, a careful rotation around @xmath54 , selecting the @xmath58 $ ] ed patterns with large @xmath59 and @xmath60 values , shows that , in our conditions of reflection , the bragg peaks _ 00l : l=2n+1 _ are scarcely visible but not null .
increasing @xmath61 , step by step , up to room temperature shows that the intensity of the extra reflections remain unchanged between @xmath40 and @xmath34 , it decreases from @xmath34 to @xmath62 and ,
lastly , disappears for @xmath63 when keeping our working conditions constant .
this transition temperature is probably an underestimate , due to the heating of the sample under electron beam irradiation . in agreement with the tem results , high resolution x - ray and neutron powder diffraction
indicate that a number of ( hkl ) bragg reflections with odd l index appear at low temperature . in the synchrotron x - ray diffraction pattern at 100k ,
the strongest of these additional peaks is ( 221 ) whereas in the neutron data , two reflections indexed ( 221 ) and ( 223 ) in the hexagonal setting are the most clearly visible .
the fact that these additional reflections appear at high value of the scattering vector ( q ) and are present in the x - ray data , indicate their nuclear origin . at 350k ,
the reflections are almost extinct but a close inspection of the diffraction pattern reveal a weak contribution even at this temperature .
nevertheless , their intensities at 350k are too weak to determine the change of structure . in this respect , the model of battle et al . , with hexagonal symmetry
p6@xmath0/mmc , remains a valid approximation at high temperature .
the extracted parameters from the combined refinement of x - ray and neutron data at 350k , assuming p6@xmath0/mmc symmetry , are reported in table [ tablecombi ] .
+ at low temperature , this model is obviously invalid due to the presence of new reflections .
also , our preliminary study with medium resolution neutron diffraction as a function of temperature , indicates that the thermal parameter of some atoms , in particular , o(1 ) , increases when the temperature is lowered , suggesting that p6@xmath0/mmc is a pseudo - symmetry . in agreement with our tem results , npd and hr - sxpd data
are consistent with the absence of the _ c _ glide - plane symmetry operation .
none of the maximal isotropy subgroups of order 2 , as listed in the international tables of crystallography@xcite , is able to fit the data .
the solution was found among isotropy subgroups of higher index , which are compatible with the landau and lifshitz conditions of the theory of second order phase transitions , as listed by hatch and stokes@xcite or equivalently , by the program isotropy@xcite .
the only solution fitting the data corresponds to the orthorhombic space group @xmath2 with a super - cell related to the hexagonal cell in the following way : @xmath66 , @xmath67 , @xmath68 .
this phase transition is a perfect example of pseudo - symmetry problems since there are no direct indication of the orthorhombic symmetry from splitting of equivalent reflections or anisotropic broadening .
the system remains metrically hexagonal within the very high resolution of our x - ray and neutron experiments ; therefore in - plane lattice parameters were constrained accordingly in the refinements .
this also indicate that the atomic displacements involved must be small enough to remain decoupled from the lattice strain .
+ in the low - temperature structure , the atomic positions for all atoms , with the exception of the mn site , split into two orbits . in our model , all refinable atomic positions were allowed to vary , while thermal parameters of atoms belonging to the two related orbits were constrained to be equal .
the combined x - ray and neutron rietveld refinement , presented in fig.[combi ] , is in excellent agreement with the data .
reliability factors of the refinement and refined structural parameters are listed in table [ tablecombi ] .
it is important to note that the combined refinement is of crucial importance here because the difference in contrast in the x - ray and neutron experiments ( o is a strong neutron scatterer , mn neutron scattering length is negative ) is extremely selective .
+ the nature of the phase transition , involving all ions in the crystal , is difficult to comprehend without decomposing the respective atomic displacements .
we have used group theory , to explain in details the symmetry descent across the phase transition .
the atomic displacements , shown in fig . [ modes ] , are decomposed in symmetry - adapted modes of the irreducible representation @xmath69 corresponding to the observed symmetry descent . more details about the treatment with representation theory are given in appendix [ sa ] , and the basis vectors of these modes are reported in tables [ bsf ] . figs.[modes]a and [ modes]b represent individual modes while fig.[modes]c shows the resulting displacements consistent with the final refinement of atomic positions .
the transition involves a slight tilting of the corner - sharing mn@xmath3o@xmath4 units composed of 2 face - sharing mno@xmath5 octahedra .
this is coupled to a shear motion of sr@xmath6 cations in the _ ab_-plane .
it is clear that the largest displacements are associated with the o(1 ) ions which explain our preliminary results of large thermal displacements on this site in the hexagonal setting .
the sequential refinements of gem data ( @xmath70 ) do not show any obvious discontinuity for most of the geometrical parameters of the structure : excepted for volume contraction , the mn - o bond distances and angles within the mn@xmath3o@xmath4 units are temperature independent .
the structural transition essentially manifests in the temperature dependance of some of the mn - o - mn angles across corner sharing octahedra , which very slightly reduce by @xmath71 in the low temperature phase .
the raman data@xcite must therefore be reinterpreted in the light of the rather subtle transition we have established . *
6c + & sr@xmath72 & sr@xmath73 & mn@xmath74 & o@xmath75 & o@xmath76 + position & ( 0,0,0 ) & ( 1/3,2/3,1/4 ) & ( 1/3,2/3,z ) & ( 1/2,0,0 ) & ( -x , x,3/4 ) + & & & z=0.61264 ( 20 ) & & x=0.81858 ( 25 ) + b@xmath77 & 0.67 ( 8) & 0.40(8 ) & 0.26(7 ) & 0.69(8 ) & 0.40(7 ) + + * 6c + atom & wyck & x & y & z & b@xmath77 + sr1 & @xmath78 & 0.0096(8 ) & 0 & 0 & 0.44(3 ) + sr2 & @xmath79 & 0 & 1/3 & 1/4 & 0.44(3 ) + mn & @xmath80 & -0.0123(8 ) & 1/3 & 0.6131(2 ) & 0.33(4 ) + o11 & @xmath78 & 0.5212(6 ) & 0 & 0 & 0.64(4 ) + o12 & @xmath80 & 0.7712(6 ) & 1/4 & 0.0085(5 ) & 0.64(4 ) + o21 & @xmath79 & 0 & -0.1798(2 ) & 1/4 & 0.52(4 ) + o22 & @xmath80 & 0.2696(3 ) & 0.0899(1 ) & 0.2412(6 ) & 0.52(4 ) + below room temperature , additional bragg reflections are observed in the powder neutron diffraction patterns , which can all be indexed with a propagation vector * * k**=0 with respect to the c - centered orthorhombic crystallographic unit cell .
these reflections are clearly magnetic in origin since their intensities decrease rapidly as a function of the scattering vector q , following the magnetic form factor .
we note that magnetic scattering and nuclear scattering attributable to the structural phase transition give rise to different contributions in the diffraction pattern , allowing to treat both problems separately .
symmetry analysis of the magnetic structure has been performed using representation analysis and is detailed in the appendix [ sa ] .
the magnetic representation @xmath81 decomposed into the direct sum of irreducible representations : @xmath82 , where all @xmath83 ( i=1,4 ) are one - dimensional representations of the little group g@xmath84 , which in this case ( * * k**=0 ) is simply the crystallographic space group . only two modes , @xmath85 or @xmath86 , belonging to different representations @xmath87 and @xmath88 , fit the experimental data .
these modes give the same arrangement of spins with components along the _ a_- and _ b_-directions respectively and are in fact equivalent due to the hexagonal pseudo - symmetry .
in fact , these modes belong to a single irreducible representation when the analysis is performed in the hexagonal setting .
for the same reason , it is impossible to determine the direction of the spins in the _ ab_-plane , and we have arbitrarily chosen to direct them along the _ a_-axis .
the refinement , shown in fig.[gemmag ] , is of very good quality .
the global user - weigthed @xmath89 , and the bragg r - factors for each pattern are @xmath90 , @xmath91 , @xmath92 , @xmath93 , @xmath94 and @xmath95 for the patterns collected on the @xmath96 , @xmath97 , and @xmath98 detectors banks , respectively . the magnetic structure is reported in fig.[mag ] .
it can be described as a simple antiferromagnetic structure since the magnetic moments of all first neighbor mn ions are aligned antiparallel .
the most surprising finding is the value of the manganese magnetic moment extracted from our rietveld refinements as a function of temperature , as shown in fig.[gemmag ] . for a quenched orbital contribution ,
the expected ordered moment for high spin mn@xmath12 ions is 3 @xmath99 per mn ion but the observed value , saturating at 2.27(1 ) at low temperature , is significantly reduced .
the nature of the structural phase transition and in particular its connection with the magnetic properties , are not immediately clear . the transition does not manifest as an abrupt decrease of the interatomic mn - mn distance , so that it can not be obviously related to some magnetic exchange - striction , which could strengthen direct antiferromagnetic interactions over the superexchange interactions . in the measured temperature range ,
the mn - mn distance decreases smoothly ( @xmath100=2.485(2 ) at 100k and 2.497(3 ) at 350k ) following essentially the expected thermal expansion .
data collected on a finer temperature scale would be needed to exclude completely the presence of exchange striction in this material . even though it has not been precisely determined here , the structural phase transition temperature is certainly above 350k and it probably coincides with the changes observed in magnetization at @xmath101380k .
therefore , it has clearly no connection with the onset of magnetic order at @xmath102286k .
in fact , preliminary neutron measurements show that the structural transition temperature increases when ca is substituted for sr in the system , highlighting the critical role of ionic radius on the a - site . from this
, we conclude that the transition is primarily due to a steric effect .
this probably influences indirectly the magnetic properties as follows .
the distortion induces a small bending of the mn - o - mn angle between some corner sharing octahedra , which could reduce the electron transfer of the 3d electron between mn pairs .
this can accordingly enhance correlations associated to the direct exchange interactions inside each pair below @xmath103 , before the magnetic ordering is established below @xmath32=286k .
let s come back first to the exchange topology of srmno@xmath0 , as introduced in fig.[magtopo ] .
it is clear that for the superexchange path @xmath29 , the nearly 180@xmath36 mn - o - mn bond angle between corner - sharing mno@xmath5 octahedra favors antiferromagnetic arrangement according to the goodenough - kanamori - anderson ( gka ) rules@xcite .
however , gka rules predict a ferromagnetic superexchange interaction for the path @xmath27 between mn spins of face - sharing octahedra , since here , the mn - o - mn angle is nearly 90@xmath36 .
the observed antiferromagnetic arrangement between these mn ions indicates that the exchange @xmath25 through direct overlap of d orbitals , dominates over super - exchange interactions .
the sign of the direct exchange interaction depends on the interatomic distance between mn ions , with shorter bonds favoring an antiferromagnetic arrangement .
the effect of interatomic distances has been systematically studied in mn intermetallic compounds . in r(mn ,
fe)@xmath5a ( r : rare earth , a : sn or ge ) , a crossover from ferromagnetic to antiferromagnetic interactions was clearly observed for distances below 2.61 @xcite , whereas in various solid solutions derived from mnsb , that crossover is found for average mn - mn distances below 2.83@xcite .
a strong antiferromagnetic direct interaction arising from the small value of @xmath104 reported here for the mn - mn distance in srmno@xmath0 is perfectly consistent with the trend seen in intermetallic compounds .
let us compare now the properties of the present pure , defect free and stoichiometric srmno@xmath0 , with that of bamno@xmath0 and camno@xmath0 keeping in mind that most camno@xmath0 compounds are oxygen deficient .
bamno@xmath0@xcite and camno@xmath0@xcite have also simple magnetic structures with af arrangements of nearest neighbor mn .
the ordered moment reported for bamno@xmath0 and that extrapolated for stoichiometric camno@xmath0 , fits the expected gs=3@xmath99 value for high spin mn@xmath12 ions .
this is not surprising considering camno@xmath0 as a classical example of heisenberg magnet with se interactions . in the case of bamno@xmath0 ,
the magnetic ordering shows a very classical behavior , which probably results from the fact that the mn - mn distance is shorter ( @xmath105 ) than in srmno@xmath0 .
the enhancement of direct antiferromagnetic exchange interaction in such case can explain why the moment saturates in bamno@xmath0 and not in srmno@xmath0 , despite the quasi - one dimensional effects expected from the infinite strings of face sharing octahedra in bamno@xmath0 . from the presented data and the comparison with other materials
, we therefore suggest that the heisenberg / se picture does not fully hold in srmno@xmath0 , and that this results from an enhanced competition between se and direct exchange interactions between neighboring mn atoms in face sharing octahedra . instead of isolated high spin mn@xmath12 ions ,
the proper magnetic unit should rather be considered as mn@xmath35o@xmath106 pairs .
the loss of the mn@xmath12 ion identity is reminiscent to the loss of v@xmath10 ion identity which has been discussed in the case of the prototype v@xmath35o@xmath0 system@xcite . like srmno@xmath0
, v@xmath35o@xmath0 shows pairs of face sharing vo@xmath5 octahedra interconnected by the edges in the corrundum structure . here
also , it has been concluded that the localized - electron unit in the antiferromagnetic insulating ground state may be the nearest - neighbor pairs , with the electrons delocalized within a pair , which has been later thoroughly confirmed by theoretical investigation of effective spin / orbital hamiltonians@xcite .
the reduced moment in srmno@xmath0 at low temperatures indicates that something similar happens within the mn - mn dimers of srmno@xmath0 .
in summary , we have shown that a structural transition does take place in srmno@xmath0 , but that this occurs at much higher temperature than previously reported from raman scattering@xcite , and is probably correlated with the transition observed in magnetization at @xmath107 .
the structural transformation does not correspond to a change of polytype , but to a subtle distortion within the 4l structure lowering the hexagonal @xmath1 symmetry to a @xmath2 orthorhombic pseudo - symmetry , which preserves the hexagonal metrics .
this structural transition seems to be primarily due to steric effects , which only indirectly affects magnetic exchange .
npd reveals that srmno@xmath0 has a reduced moment at low temperatures .
we interpret this might be due to the presence of delocalized electrons within mn@xmath35o@xmath106 pairs , a finding that deserves confirmation from theoretical studies of the electronic structure .
the symmetry properties of the low temperature structure of srmno@xmath0have been determined by representation analysis .
the atomic positions are written @xmath108 for atoms in crystallographic cells indexed by @xmath109 that were at average positions @xmath110 in the high temperature phase , which are now displaced by vectors @xmath111 .
these displacement vectors are decomposed with the fourier sum @xmath112 where @xmath113 is the wave vector characterizing how the translational symmetry is broken at the transition .
the symmetry relations between the vectors @xmath114 for atoms of the same crystallographic orbit are obtained for each irreducible representation ( irrep ) @xmath115 using the projection operator : @xmath116 the sum is over the symmetry elements @xmath117 of the little group transforming @xmath113 into an equivalent wave vector .
@xmath118 are elements of the matrix representation of @xmath117 for the irrep @xmath119 .
this operator projects an arbitrary set of displacement components on atoms of a crystallographic orbit of multiplicity @xmath120 , onto symmetrized basis functions ( sbf ) .
we conveniently write these sbf as @xmath121 , where @xmath122 represents the displacement vectors components on an atom @xmath123 in the direction @xmath124 .
an arbitrary set of linearly independent sbf are chosen from all the constructed ones . working out with sbf belonging to single irreps
is equivalent to assume a distortion compatible with a second order transition . in our case ,
the propagation vector observed experimentally is * * k**=(0,0,0 ) , a special case where the complex fourier component @xmath125 identifies with the real displacement @xmath111 . as always with * * k**=(0 0 0 ) , all the symmetry elements of @xmath1 leave * k * invariant , hence , @xmath126 coincides with the space group @xmath1 . irreps and basis functions were obtained using the program basireps . in our case
however , the program outputs 8 real one - dimensional representations , and 4 two - dimensional complex representations .
the sbf being expected to be real , it is desirable to transform the complex representations into real using an appropriate unitary similarity transformation matrix @xmath127 , for which @xmath128 before projection . using + @xmath129 real matrices @xmath130 for the irreps @xmath69,@xmath131,@xmath132,@xmath133 have been thus obtained from the complex matrices @xmath134 , which are output by basireps ( see tables ii and iii ) .
the projection of the sbf is illustrated for the two - dimensional representation labeled @xmath69 , which correspond to the irrep actually chosen at the phase transition in srmno@xmath0 . with the special choice of unitary transform
, we could straightforwardly double check that the symmetry modes associated to the transition from @xmath1 to @xmath2 and given by the program symmodes of the bilbao crystallographic server@xcite indeed correspond to a subsets of the sbf calculated for the irrep @xmath69 by basireps .
representation analysis is more general , and also applies to other type of phase transitions where atoms acquire a new scalar , vectorial , or tensorial property in the low - symmetry phase . instead of dealing with atomic displacements ( polar vectors ) in the case of structural transitions
, we can indeed deal with the appearance of a magnetic moment to treat the case of magnetic transitions : @xmath135 the only change in the orthogonalization procedure and the construction of the sbf summarizing this time the symmetry relation of the fourier components @xmath136 , is the action of the space group operation @xmath117 in eq.[proj ] , which is different for axial and polar vectors .
we similarly obtained the symmetry analysis of the possible magnetic structure of srmno@xmath0 , which is summarized in table [ magsymm ] . *
7c ireps & + it symbol & 1 & 3 + 0,0,z & 2 ( 0,0,1/2 ) 0,0,z & 2 x , x,0 & 3- 0,0,z & 6- ( 0,0,1/2 ) 0,0,z + & @xmath137 & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 + it symbol & 2 0,y,0 & 2 x ,-
x,1/4 & 6 + ( 0,0,1/2 ) 0,0,z & 2 x,0,0 & 2 2x , x,1/4 & 2 x,2x,1/4 + & @xmath143 & @xmath144 & @xmath145 & @xmath146 & @xmath147 & @xmath148 + it symbol & -1 0,0,0 & -3 + 0,0,z ; 0,0,0 & m x , y,1/4 & m x ,- x , z & -3- 0,0,z ; 0,0,0 & -6- 0,0,z ; 0,0,1/4 + & @xmath149 & @xmath150 & @xmath151 & @xmath152 & @xmath153 & @xmath154 + it symbol & m 2x , x,2z & c x , x , z & -6 + 0,0,z ; 0,0,1/4 & m x,2x , z & c 0,y , z & c x,0,z + & @xmath155 & @xmath156 & @xmath157 & @xmath158 & @xmath159 & @xmath160 + @xmath161 & 1 & 1 & 1 & 1 & 1 & 1 + & 1 & 1 & 1 & 1 & 1 & 1 + & 1 & 1 & 1 & 1 & 1 & 1 + & 1 & 1 & 1 & 1 & 1 & 1 + @xmath162 & 1 & 1 & 1 & 1 & 1 & 1 + & 1 & 1 & 1 & 1 & 1 & 1 + & -1 & -1 & -1 & -1 & -1 & -1 + & -1 & -1 & -1 & -1 & -1 & -1 + @xmath87 & 1 & 1 & 1 & 1 & 1 & 1 + & -1 & -1 & -1 & -1 & -1 & -1 + & 1 & 1 & 1 & 1 & 1 & 1 + & -1 & -1 & -1 & -1 & -1 & -1 + @xmath88 & 1 & 1 & 1 & 1 & 1 & 1 + & -1 & -1 & -1 & -1 & -1 & -1 + & -1 & -1 & -1 & -1 & -1 & -1 + & 1 & 1 & 1 & 1 & 1 & 1 + @xmath69 & @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & @xmath168 + & @xmath169 & @xmath170 & @xmath171 & @xmath172 & @xmath173 & @xmath174 + & @xmath175 & @xmath176 & @xmath177 & @xmath178 & @xmath179 & @xmath180 + & @xmath181 & @xmath182 & @xmath183 & @xmath184 & @xmath185 & @xmath186 + @xmath131 & @xmath187 & @xmath188 & @xmath189 & @xmath190 & @xmath191 & @xmath192 + & @xmath193 & @xmath194 & @xmath195 & @xmath196 & @xmath197 & @xmath198 + & @xmath199 & @xmath200 & @xmath189 & @xmath190 & @xmath191 & @xmath192 + & @xmath193 & @xmath194 & @xmath195 & @xmath196 & @xmath197 & @xmath198 + * 7c ireps & + it symbol & 1 & 3 + 0,0,z & 2 ( 0,0,1/2 ) 0,0,z & 2 x , x,0 & 3- 0,0,z & 6- ( 0,0,1/2 ) 0,0,z + & @xmath137 & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 + it symbol & 2 0,y,0 & 2 x ,-
x,1/4 & 6 + ( 0,0,1/2 ) 0,0,z & 2 x,0,0 & 2 2x , x,1/4 & 2 x,2x,1/4 + & @xmath143 & @xmath144 & @xmath145 & @xmath146 & @xmath147 & @xmath148 + it symbol & -1 0,0,0 & -3 + 0,0,z ; 0,0,0 & m x , y,1/4 & m x ,- x , z & -3- 0,0,z ; 0,0,0 & -6- 0,0,z ; 0,0,1/4 + & @xmath149 & @xmath150 & @xmath151 & @xmath152 & @xmath153 & @xmath154 + it symbol & m 2x , x,2z & c x , x , z & -6 + 0,0,z ; 0,0,1/4 & m x,2x , z & c 0,y , z & c x,0,z + & @xmath155 & @xmath156 & @xmath157 & @xmath158 & @xmath159 & @xmath160 + @xmath201 & 1 & -1 & 1 & -1 & 1 & -1 + & 1 & -1 & 1 & -1 & 1 & -1 + & 1 & -1 & 1 & -1 & 1 & -1 + & 1 & -1 & 1 & -1 & 1 & -1 + @xmath202 & 1 & -1 & 1 & -1 & 1 & -1 + & 1 & -1 & 1 & -1 & 1 & -1 + & -1 & 1 & -1 & 1 & -1 & 1 + & -1 & 1 & -1 & 1 & -1 & 1 + @xmath203 & 1 & -1 & 1 & -1 & 1 & -1 + & -1 & 1 & -1 & 1 & -1 & 1 + & 1 & -1 & 1 & -1 & 1 & -1 + & -1 & 1 & -1 & 1 & -1 & 1 + @xmath204 & 1 & -1 & 1 & -1 & 1 & -1 + & -1 & 1 & -1 & 1 & -1 & 1 + & -1 & 1 & -1 & 1 & -1 & 1 + & 1 & -1 & 1 & -1 & 1 & -1 + @xmath132 & @xmath205 & @xmath206 & @xmath207 & @xmath208 & @xmath209 & @xmath210 + & @xmath211 & @xmath212 & @xmath213 & @xmath214 & @xmath215 & @xmath216 + & @xmath217 & @xmath218 & @xmath219 & @xmath220 & @xmath221 & @xmath222 + & @xmath212 & @xmath211 & @xmath214 & @xmath213 & @xmath216 & @xmath215 + @xmath133 & @xmath223 & @xmath224 & @xmath207 & @xmath208 & @xmath209 & @xmath210 + & @xmath225 & @xmath226 & @xmath227 & @xmath228 & @xmath229 & @xmath230 + & @xmath231 & @xmath232 & @xmath233 & @xmath234 & @xmath235 & @xmath210 + & @xmath225 & @xmath226 & @xmath227 & @xmath228 & @xmath229 & @xmath230 + * 5c + symm & x , y , z & -x ,- y , z+1/2 & y , x ,- z & -y ,- x ,- z+1/2 + atoms : & mn@xmath236 & mn@xmath35 & mn@xmath0 & mn@xmath239 + @xmath240 & ( 1 0 0 ) & ( -1 0 0 ) & ( 1 0 0 ) & ( -1 0 0 ) + @xmath241 & ( -1 -2 0 ) & ( 1 2 0 ) & ( -1 -2 0 ) & ( 1 2 0 ) + * 7c + symm & x , y , z & -x ,- y , z+1/2 & -y , x - y , z & y ,-
x+y , z+1/2 & -x+y ,- x , z & x - y , x , z+1/2 + atoms : & o1@xmath236 & o1@xmath35 & o1@xmath0 & o1@xmath239 & o1@xmath242 & o1@xmath5 + @xmath243 & ( 2 0 0 ) & ( -2 0 0 ) & ( 0 -1 0 ) & ( 0 1 0 ) & ( 1 1 0 ) & ( -1 -1 0 ) + @xmath244 & ( 0 0 0 ) & ( 0 0 0 ) & ( -2 -1 0 ) & ( 2 1 0 ) & ( -1 1 0 ) & ( 1 -1 0 ) + @xmath245 & ( 0 0 0 ) & ( 0 0 0 ) & ( 0 0 1 ) & ( 0 0 1 ) & ( 0 0 -1 ) & ( 0 0 -1 ) + @xmath246 & ( 0 0 0 ) & ( 0 0 0 ) & ( 0 -1 0 ) & ( 0 1 0 ) & ( -1 -1 0 ) & ( 1 1 0 ) + @xmath247 & ( 0 0 0 ) & ( 0 0 0 ) & ( 0 1 0 ) & ( 0 -1 0 ) & ( 1 1 0 ) & ( -1 -1 0 ) + @xmath248 & ( 0 0 2 ) & ( 0 0 2 ) & ( 0 0 -1 ) & ( 0 0 -1 ) & ( 0 0 -1 ) & ( 0 0 -1 ) + * 7c + symm & x , y , z & -x ,- y , z+1/2 & -y , x - y , z & y ,-
x+y , z+1/2 & -x+y ,- x , z & x - y , x , z+1/2 + atoms : & o2@xmath236 & o2@xmath35 & o2@xmath0 & o2@xmath239 & o2@xmath242 & o2@xmath5 + @xmath249 & ( 0 0 1 ) & ( 0 0 1 ) & ( 0 0 0 ) & ( 0 0 0 ) & ( 0 0 -1 ) & ( 0 0 -1 ) + @xmath250 & ( 0 0 1 ) & ( 0 0 1 ) & ( 0 0 -2 ) & ( 0 0 -2 ) & ( 0 0 1 ) & ( 0 0 1 ) + * 6c + & 1 & 2 ( 0,0,1/2 ) 0,0,z & 2 0,y,1/4 & 2 x,0,0 + & @xmath251 & @xmath252 & @xmath253 & @xmath254 + & 1 & 1 & 1 & 1 + & 1 & 1 & -1 & -1 + & 1 & -1 & 1 & -1 + & 1 & -1 & -1 & 1 + + & mn@xmath236 & mn@xmath35 & mn@xmath0 & mn@xmath239 + & @xmath251 & @xmath252 & @xmath253 & @xmath254 + & @xmath255 & @xmath256 & @xmath257 & @xmath258 & @xmath259 + @xmath161 & @xmath260 & @xmath261 & @xmath262 & @xmath263 & @xmath264 + & @xmath265 & @xmath266 & @xmath267 & @xmath268 & @xmath269 + & @xmath265 & @xmath256 & @xmath257 & @xmath270 & @xmath271 + @xmath162 & @xmath272 & @xmath261 & @xmath262 & @xmath273 & @xmath274 + & @xmath275 & @xmath266 & @xmath267 & @xmath276 & @xmath277 + & @xmath278 & @xmath256 & @xmath279 & @xmath258 & @xmath271 + @xmath87 & @xmath85 & @xmath261 & @xmath280 & @xmath263 & @xmath274 + & @xmath281 & @xmath266 & @xmath282 & @xmath268 & @xmath277 + & @xmath283 & @xmath256 & @xmath279 & @xmath270 & @xmath259 + @xmath88 & @xmath284 & @xmath261 & @xmath280 & @xmath273 & @xmath264
+ & @xmath285 & @xmath266 & @xmath282 & @xmath276 & @xmath269 + viewed along the [ 110 ] direction . the sr , mn and o atoms are shown as large grey spheres , small grey spheres and black spheres respectively .
face - sharing mno@xmath5 octahedra are also represented as transparent grey units ; right panel : magnetic exchange topology of the mn sublattice .
black bonds represent magnetic exchange between mn ions of the mn@xmath35o@xmath106 units , composed of antiferromagnetic direct exchange ( j@xmath286 ) and ferromagnetic super - exchange ( j@xmath236 ) .
grey bonds represent super - exchange interactions ( j@xmath35 ) between mn ions of the hexagonal units .
see text for details.,title="fig : " ] + $ ] ed pattern of srmno@xmath0 at rt ( a ) and @xmath57 ( b ) .
the low temperature data shows the appearance of another system of weak reflections , _
l = no condition _ , which clearly violate the mirror @xmath56 .
a careful rotation around @xmath54 , selecting the @xmath58 $ ] ed patterns with large @xmath59 and @xmath60 values , shows that , in our conditions of reflection , the conditions _ 00l : l=2n+1 _ are scarcely visible but not null and hence , are not due to double diffraction.,title="fig : " ] + structure at 100k with ( a ) npd hrpd and ( b ) hr - sxpd id-31 data collected on srmno@xmath0 at @xmath65 . for both patterns , the continuous line correspond to the model fitting to the data ,
the lowest continuous line , the difference curve , and ticks are indicating the position of the bragg reflections .
for the npd ( a ) , the upper , middle and lower ticks represents the contribution from the nuclear scattering , the magnetic scattering , and the contribution from the vanadium can ( asterisk ) .
arrows indicate the position of the few reflections violating the extinction conditions of the high temperature space group @xmath1.,title="fig : " ] + .
the arrows and the atom labels gives a graphical representation of the basis vectors of irrep .
@xmath69 involved in the transition , which are given table iv , appendix a ( a ) displacement modes for mn ( medium sized grey spheres ) , sr@xmath236 ( large black spheres ) and o@xmath35 ( small grey spheres ) ( b ) displacement modes for o@xmath236 ( c ) resultant distortion from the linear combination @xmath287 of all basis functions : the refinement leads to @xmath288 , @xmath289 and @xmath290 for sr@xmath236 , mn , and o@xmath35 respectively for o@xmath236 , the mixing coefficients can be constrained to have no component along b , as @xmath291,title="fig : " ] + at @xmath292 with gem data collected .
data from a detector bank situated at 35@xmath293 2@xmath294 are shown .
the experimental data points are shown as grey crosses while the calculated pattern is represented by a grey solid line .
contribution from magnetic scattering is plotted separately as a thick black solid line .
the lowest continuous line shows the difference curve between data and refinement , while tick marks indicate the positions of bragg reflections.,title="fig : " ] + | the structural and magnetic properties of the hexagonal four - layer form of srmno@xmath0 have been investigated by combining magnetization measurements , electron diffraction and high - resolution synchrotron x - ray and neutron powder diffraction .
below 350k , there is subtle structural phase transition from hexagonal symmetry ( space group @xmath1 ) to orthorhombic symmetry ( space group @xmath2 ) where the hexagonal metric is preserved .
the second - order phase transition involves a slight tilting of the corner - sharing mn@xmath3o@xmath4 units composed of 2 face - sharing mno@xmath5 octahedra and the associated displacement of sr@xmath6 cations .
the phase transition is described in terms of symmetry - adapted displacement modes of the high symmetry phase . upon further cooling , long range magnetic order with propagation vector @xmath7 sets in below 300k .
the antiferromagnetic structure , analyzed using representation theory , shows a considerably reduced magnetic moment indicating the crucial role played by direct exchange between mn centers of the mn@xmath3o@xmath4 units . |
low - mass stars ( i.e. , @xmath8 ) represent the largest fraction of the stellar population in the galaxy .
the availability of a reliable theoretical mass radius relationship is of considerable interest both _
per se _ and in view of the application to the study of exoplanet - hosting stars .
moreover , below about @xmath9 , the depth of the subsurface convection zone extends enough to reach the centre and stars become fully convective ( hereafter fc ) .
this new regime has peculiar characteristics and presents additional challenges . in constructing theoretical models of fc stars ,
special care is required in handling all the main input physics ingredients , such as the equation of state , the opacities , and the atmospheric boundary conditions , due to the lower temperature and higher density regime ( see the reviews by @xcite and @xcite ) .
although convection becomes very efficient at quite shallow depths ( i.e. @xmath10 at @xmath11 , see e.g. section 7.3.3 of @xcite ) , it is still necessary to rely on the mixing length theory ( mlt , @xcite ) for the convective transport in the outermost layers and in the atmosphere , which can have a significant impact on the global structure parameters @xcite .
double - lined eclipsing binary systems ( hereafter ebs ) provide very accurate determination of the fundamental parameters of stars , allowing to test theoretical model predictions .
there is mounting evidence that low - mass stars in ebs have radii about @xmath12 per cent larger and effective temperatures about @xmath2 per cent cooler than theoretical models ( see @xcite and references therein ) .
interestingly , the discrepancies are in the right proportions to produce roughly the same luminosity .
to explain these results , @xcite proposed that close ebs may not be representative of the whole low - mass stars population because of the high rotation regimes produced by spin
orbit synchronization via tidal interaction .
this explanation is corroborated by the significant correlation found by @xcite between the difference in the radii and the level of magnetic activity .
indeed , both rotation and magnetic fields can reduce the efficiency of convection , which is measured by the parameter @xmath0 in the standard mlt framework ( e.g. * ? ? ?
a faster rotation , moreover , can enhance the magnetic activity through the dynamo effect , leading to a substantial increase in the star - spots coverage of the surface .
both effects were phenomenologically taken into account by @xcite , using a lower @xmath0 with respect to the solar - calibrated value and reducing the radiative flux at the surface proportionally to the star - spots filling factor @xmath13 , respectively .
their results show that the parameters @xmath0 and @xmath13 have a degenerate effect on the stellar radius .
they also claim that fc stars are almost insensitive to a variation of @xmath0 and very low values ( e.g. , @xmath14 ) are required to reconcile the models with the observations . finally , according to @xcite
, the discrepancy may be partially accounted for by a systematic error in the light curve - derived radii , which would occur if a value of @xmath13 of about @xmath15 per cent and a concentration of the spots near the stellar poles are postulated .
as @xcite have pointed out , however , theoretical models calculated with a quite restricted set of parameters ( e.g. , solar metallicity ) , have been compared with observations of stars of poorly determined or even unknown metallicity , assuming a general validity that is not rigorously justified .
even a moderate increase in metallicity can result in theoretical radii increases which are significant in comparison with the discrepancies discussed so far .
a combination of all these effects is thus the most likely solution to the low - mass stellar radii problem . in this paper , we attempt to model the components of the two ebs known to date to host fc stars .
our main focus is to determine the sensitivity of the best - fitting models to various possible choices of the parameters , with as much generality as possible .
the paper is organized as follows : in section [ sec : obs ] we discuss the available data and compare the physical properties of the two systems ; in section [ sec : models ] we present the code and input physics used in our calculations and in section [ sec : ttau ] we show the impact of the atmospheric boundary conditions on our models . in sections [ sec :
k126 ] and [ sec : cmdra ] we discuss our best - fitting solutions for the systems koi-126 and cm draconis ( hereafter cm dra ) , respectively .
we compare our results for koi-126 with those obtained by @xcite in section [ sec : comp ] .
finally , in section [ sec : disc ] , we review results and conclusions of this work .
the systems koi-126 ( kic number : @xmath16 ) and cm dra host the only ebs known to date in the fc mass range .
the former was recently discovered by the kepler mission @xcite ; it is a hierarchical triple system , with a low - mass binary subsystem ( orbital period : @xmath17 d ) orbiting the primary star with a period of @xmath18 d. the second system ( e.g. @xcite , @xcite , and references therein ) is composed of two fc stars ( orbital period : @xmath19 d ) which have common proper motion with a white dwarf . for the stars in these systems , masses and
radii are determined with very high precision from the modelling of the light curve eclipses .
other parameters however , such as metallicity and effective temperatures , are much less well known .
table [ tab : parms ] summarises the known physical parameters of the two systems .
spectroscopic observations of koi-126 only show features associated with the primary star ( * ? ? ?
* see also the online supporting material ) .
however , the light curve analysis provides a determination of the fluxes of the two low - mass stars relative to the primary , which can be used to derive a rough estimate of their effective temperatures .
we proceeded as follows : using synthetic spectra from the nextgen ( ng ) grid ( constructed with the phoenix 1d code , * ? ? ?
* ) , with solar metallicity and appropriate @xmath20 and @xmath21 , we calculated the theoretical integrated flux for each star of the system , taking into account the response function of kepler @xcite .
we estimate that the effective temperatures fall in the range @xmath22@xmath23 k ; since the radii are known , the luminosities can be derived as well ( see table [ tab : parms ] ) .
although considerable effort has been devoted to the study of cm dra , both spectroscopically and photometrically ( @xcite , @xcite ) , the direct determination of its metallicity is still controversial @xcite ; the system is considered a population ii member based on its measured proper motion .
[ tab : parms ] the masses of the fc components of both systems fall in the range @xmath24 and have remarkably similar ratios .
interestingly , koi-126 c and cm dra 2 have the same mass within the error , but the radius of the latter is significantly larger ( of about @xmath1 per cent ) .
similarly , koi-126 b is about @xmath25 per cent more massive than cm dra 1 , but their measured radii are the same within the error .
the mass radius relationship followed by cm dra 1 and 2 is indeed very different from that of the fc components of koi 126 , as fig .
[ fig : mrrel ] shows
. correspondingly , the effective temperatures of cm dra are significantly lower than the range estimated for koi-126 b and c. from the raw observational data , taken at face value , we can expect the best - fitting models for the fc stars in the two systems to have quite different properties . in modelling the two systems
, we used different strategies to cope with their different characteristics and available observational constraints ( e.g. , spectroscopy of koi-126 a , @xmath3 of cm dra 1 and 2 , etc . ) .
for koi-126 , we looked for a self - consistent solution ( in age and composition ) for all the three components , without making any _ a priori _ assumption on the age , as is explained in detail in section [ sec : k126 ] .
for cm dra , on the other hand , we adopted the age estimate from the literature , since the age sensitivity of fc models is too weak to provide a meaningful self - consistency check ( e.g. , the radius variation is of the order of @xmath26@xmath1 per cent over the whole main sequence lifetime , @xmath27@xmath12 gyr , and about @xmath27 per cent between @xmath27 and @xmath2 gyr ) .
gyr , calculated with different metallicities and @xmath28.,scaledwidth=50.0% ]
all the models were calculated using the yale rotating stellar evolution code ( yrec , see @xcite ) in its non - rotating configuration .
we used the @xcite opacities at low temperatures , the opal rosseland opacities at high temperatures @xcite , and the opal 2005 equation of state @xcite .
the energy generation rates are calculated according to the prescription of @xcite and the diffusion coefficients for helium and heavy elements are those given by @xcite .
convection is described using the mlt @xcite .
we adopt the @xcite value of the solar metallicity , @xmath29 .
the integration of atmospheric layers in yrec requires that the temperature optical depth ( @xmath30@xmath31 hereafter ) relation is specified .
while the @xcite ( ks in the following ) or eddington grey approximations give reasonably accurate results for the sun or more massive stars , respectively , this is not true for the fc models ( see section 2.5 of @xcite , and also section [ sec : ttau ] below ) .
for this reason , in all the modelling discussed here , we used the @xmath30-@xmath31 relations taken from the ng model atmospheres of appropriate @xmath32 .
as a reference , with the input physics described so far , a standard solar model calculated with yrec has an initial helium content of @xmath33 and a value of the mlt parameter of @xmath34 ( @xmath35 ) when using the eddington ( ks ) @xmath30@xmath31 relation .
-@xmath30 atmospheric profile for a @xmath36 , @xmath25 gyr - old star , calculated with @xmath37 , @xmath38=0.0 $ ] , and @xmath28 .
different @xmath30@xmath31 relations are used in the atmosphere .
lower panel : the same as in the upper panel , but with @xmath38=-0.5$].,title="fig:",scaledwidth=45.0% ] -@xmath30 atmospheric profile for a @xmath36 , @xmath25 gyr - old star , calculated with @xmath37 , @xmath38=0.0 $ ] , and @xmath28 .
different @xmath30@xmath31 relations are used in the atmosphere .
lower panel : the same as in the upper panel , but with @xmath38=-0.5$].,title="fig:",scaledwidth=45.0% ] the global parameters of a fc star ( i.e. @xmath39 , @xmath40 , @xmath3 ) are highly sensitive to the treatment of the outermost layers . in particular , the model atmosphere used to determine the surface boundary conditions plays a crucial role . as was pointed out by @xcite ,
a @xmath30@xmath31 relation based on a grey atmosphere and/or radiative equilibrium is incorrect for @xmath41 k , because the recombination of molecular hydrogen in the envelope reduces the entropy and the adiabatic gradient , thus favouring the penetration of convection in optically thin layers .
moreover , a grey treatment produces cooler atmospheric profiles below the photosphere , thus leading to overestimated @xmath3 @xcite .
we investigated the impact of these effects on our calculations by comparing models of fc stars constructed using the eddington , ks and ng @xmath30@xmath31 relations .
[ fig : ptplane ] shows the pressure temperature profile in the outer layers of a star of @xmath36 with solar and moderately metal - poor composition ( cf . fig .
5a - b of @xcite ) .
the models calculated with the grey @xmath30@xmath31 relations have a discontinuity in the temperature gradient near the photosphere ( i.e. around @xmath42 ) .
this is due to the abrupt truncation of convection at the transition from the envelope to the atmosphere , where radiative equilibrium is artificially enforced throughout .
the models using the ng temperature profile , on the other hand , have a much shallower temperature gradient and a smooth envelope - atmosphere transition .
the net effect is that , using eddington ( ks ) grey atmosphere models , we obtain effective temperatures which are overestimated of about @xmath43 k ( @xmath44 k ) with respect to the ng models .
convection is treated according to the standard mlt in the phoenix 1d code .
since the models in the ng grid were all constructed with @xmath45 , we were unable to estimate directly the impact of the value of @xmath0 in the atmosphere on the global stellar quantities .
moreover , a value of @xmath45 in the atmosphere is larger than the one needed to achieve a satisfactory fit of spectral line profiles ( @xcite , see also section 2 of @xcite ) .
in fact , inconsistent values of @xmath0 between the atmosphere and the interior of the same stellar model are often required to match satisfactorily all the available observational constraints .
numerical simulations of convection also show that assuming the same value of the mixing length throughout all the outer stellar layers might be an oversimplification ( see , e.g. * ? ? ? * ) . in our modelling ,
however , this inconsistency does not seem to affect significantly the temperature profile , as is apparent from fig .
[ fig : ptplane ] .
for the system koi-126 , we tried to achieve a self - consistency in the best - fitting models of the three components , as follows . given its mass and the preliminary estimate of the age of the system reported by @xcite of @xmath46 gyr , koi-126 a is probably in the early post - main sequence phase , on the way to the red giant branch .
helium and heavy element diffusion and core overshooting were taken into account when modelling this star . with fixed values of the core overshooting parameter @xmath47 and metal content @xmath38 $ ] , we adjusted the helium content and the mlt parameter in order to produce the model of koi-126 a with the closest agreement with the observed @xmath48 and @xmath49 ( we choose to use the @xmath3-@xmath39 plane to take advantage of the high precision in the radius determination ) .
the resulting best - fitting model of the primary star gives the most stringent constraint on the age of the system , since the low - mass components undergo very small changes once they are on the main sequence .
we then tried to fit the radii of koi-126 b , c for each composition , determining an age range that can be used to check for compatibility with the model of the primary . in modelling koi-126 a , which is massive enough to have a convective core , we tried different values of the core overshooting parameter , within the range @xmath50@xmath51 .
this range is typically adopted for early post - main sequence subgiant stars ( see , e.g. , @xcite , @xcite , @xcite ) .
interestingly , for this star , we found a very close agreement between the evolutionary tracks calculated with ng ( @xmath52 k , @xmath53 ) and eddington ( grey ) @xmath30@xmath31 relations .
table [ tab : k126a ] reports the parameters for the best - fitting models . with the values of @xmath47 ,
@xmath38 $ ] , and @xmath54 held fixed , the mlt parameter @xmath0 determines the evolutionary path of the star in the @xmath55 plane .
the range of @xmath0 shown corresponds to evolutionary tracks that cross the observational box in the @xmath55 plane , i.e. , it corresponds to the models that passed within @xmath56 or less from both @xmath49 and @xmath48 .
since the radius determination is much more precise compared with that of the @xmath3 , our age estimate is that of the model which has @xmath57 .
these intervals of @xmath0 and age do not have the meaning of formal error bars , but are reported here to give a quantitative idea of the sensitivity of the models to the various parameters . in general , @xmath47 influences the best - fitting value of @xmath0 and therefore has an indirect impact on the age ( lower @xmath0 values result in younger age estimates ) . as for the composition , higher helium content produces younger models , while a higher metallicity requires larger values of @xmath0 and leads to an increase in the age of the models .
[ tab : k126a ] we calculated models of the fc components of koi-126 with each possible choice of ( @xmath54 , @xmath38 $ ] ) shown in table [ tab : k126a ] , keeping the mlt parameter fixed with the value @xmath58 .
we did not make any attempt to account for the @xmath3 constraint , since only a very rough estimate is available for these stars .
table [ tab : k126bc ] shows the age interval compatible with the observed radius for each composition .
consistent solutions ( marked in bold in tables [ tab : k126a ] and [ tab : k126bc ] ) exist for the following compositions : @xmath38=+0.15 $ ] , @xmath37 and @xmath59 ; @xmath38=0.23 $ ] , @xmath60 ; the ages are @xmath61 gyr , @xmath62 gyr and @xmath63 gyr , respectively . in conclusion , using the central determination of @xmath38 $ ] and a value of @xmath37 , close to the solar calibration , we obtain a consistent solution for the system , with an estimated age range of @xmath64@xmath65 gyr .
.age intervals compatible with the observational constraints on the radius for koi-126 b and c. all the models were calculated with @xmath58 . [
cols="^,^,^,^,^,^,^ " , ] [ tab : k126bc ] -@xmath39 relations at @xmath25 gyr calculated with different compositions and @xmath28 .
lower panel : as above , but keeping the composition fixed at solar values ( i.e. @xmath37 , @xmath66 ) and using different values of @xmath0.,title="fig:",scaledwidth=45.0% ] -@xmath39 relations at @xmath25 gyr calculated with different compositions and @xmath28 .
lower panel : as above , but keeping the composition fixed at solar values ( i.e. @xmath37 , @xmath66 ) and using different values of @xmath0.,title="fig:",scaledwidth=45.0% ] for cm dra , we assumed the age quoted in the literature , of about 4 gyr , and we allowed for a variation of the chemical composition in a very broad range , to take into account the high level of uncertainty in its determination . as was already clear from fig .
[ fig : mrrel ] , it is impossible to reconcile the models of cm dra with the observed values of the radii when using a metallicity in the range quoted in the literature ( @xmath67<-0.6 $ ] , see also the upper panel of fig . [
fig : cmdra ] ) .
in fact , even allowing all the parameters to freely vary within reasonable limits , we always find that our theoretical models underestimate the radii ( and overestimate the @xmath3 ) .
the upper panel of fig .
[ fig : cmdra ] also shows that , to bring the theoretical models almost within @xmath56 of agreement with the observations , a rather extreme metal - rich composition must be used ( @xmath38=+0.5 $ ] , or @xmath68 ) .
a moderate effect of the helium content is also apparent from the upper panel of fig .
[ fig : cmdra ] , with higher values of @xmath54 resulting in slightly higher effective temperatures and larger radii .
such a high metallicity is unlikely for a system of population ii , but it should be emphasized that this membership attribution is based only on the circumstantial evidence provided by the proper motion and that the direct estimate of the metallicity is very controversial ( see the discussion in section 5 of * ? ? ?
the possibility of non - standard effects in the structure of cm dra components has been invoked in the past to explain the radius anomaly .
both stars are usually assumed to have synchronised their spin with the orbital period , which makes of them quite fast rotators .
they are also known to have spotted surfaces ( see , e.g. @xcite ) , which is an indicator of the presence of subsurface magnetic fields .
a fast rotation and/or highly magnetised regime can reduce the efficiency of the convective transport @xcite .
this effect was taken into account by @xcite , as well as a simplified parametrization of the star - spots .
they found that both effects can be phenomenologically represented by a reduction of the mlt parameter @xmath0 .
we thus tried to construct models of cm dra with reduced @xmath0 , which suffices to the purpose of investigating such activity - induced effects within the standard theoretical framework , keeping the number of free parameters to a minimum .
as the lower panel of fig .
[ fig : cmdra ] shows , with @xmath37 and solar metallicity , the disagreement can be brought within about @xmath69 if a value of @xmath0 as low as @xmath51 is used . however , even reducing @xmath0 down to @xmath70 , we were unable to construct models with stellar radii compatible with the observations within the @xmath71 uncertainty . for @xmath72 ,
the mlt treatment implemented in yrec gives rise to numerical instabilities . in this sense , tuning @xmath0 as a free parameter is insufficient , by itself , to reconcile the models of cm dra 1 and 2 with the observations .
as the moderate sensitivity of fc stars to @xmath0 requires such a drastic , possibly unphysical reduction , we also tried to model cm dra using a combination of the composition ( @xmath54 and @xmath73 ) variation and @xmath0 reduction effects .
with @xmath38=+0.3 $ ] ( i.e. @xmath74 ) and @xmath37@xmath75 , for example , we find that the discrepancy is lower than @xmath6 for @xmath45 .
clearly , a better determination of the composition of the system , based on more robust evidences than kinematic arguments , is required to conclusively settle this issue . since this information is currently missing , we refrain from attempting a precise modelling of cm dra , as was done in section [ sec : k126 ] for koi-126 .
we believe , nonetheless , that our results suggest a revision of the metallicity of this system towards at least solar - like ( if not higher ) values .
dedicated models for the system koi-126 were constructed , independently of the present work , by @xcite , who used the dartmouth stellar evolution program ( dsep , see @xcite and references therein ) .
they were able to match very well the observational constraints and to reproduce satisfactorily the mass radius relationship for the fc components , using a metallicity in the quoted range and a solar - calibrated value of the mlt parameter .
their estimate for the age of the system is @xmath76 gyr .
the codes dsep and yrec have a distant common origin ; they differ mainly in the treatment of some pieces of input physics , namely , the equation of state , the atmospheric boundary conditions , and the presence of turbulent diffusion . as was shown in the previous section ,
both codes are quite successful in modelling the system koi-126 ; the differences between the two codes manifest themselves in the estimated age . to ease the comparison , fig .
[ fig : comp ] shows the evolutionary tracks for the radius of koi-126 a , constructed with yrec using the solar - calibrated mlt parameter and helium content ( @xmath33 , @xmath77 , eddington @xmath30@xmath31 relation , @xmath78 ) , and the same choice of mass and metallicity as in fig . 1 of @xcite .
in this way , we arrive at the age range @xmath79 gyr , with a slightly larger uncertainty , but compatible with the results of @xcite .
interestingly , our tracks in fig .
[ fig : comp ] intersect the observed radius when the star is already out of the main sequence phase , while this is not the case for the models in fig . 1 of @xcite . ,
@xmath80 and @xmath81 gyr.,scaledwidth=45.0% ]
we modelled the systems koi-126 and cm dra with the yrec code , using up - to - date input physics .
both systems have very precise determination of stellar masses and radii .
this allows to test in detail the theoretical models of the fc components , which is the main goal of this work . as fig .
[ fig : mrrel ] shows , for koi-126 b and c the theoretical mass radius tracks at @xmath25 gyr in the @xmath82@xmath83 mass range , with a metallicity between solar and @xmath38=+0.30 $ ] ( @xmath37 and @xmath28 ) , match quite well the observational data , consistently with the detailed modelling discussed in section [ sec : k126 ] . for the system
cm dra the situation is quite different .
we find significant discrepancies in radii and @xmath3 when using @xmath84 and a metal - poor , solar or even moderately metal - rich composition ( see the upper panel of fig .
[ fig : cmdra ] ) .
dedicated modelling of cm dra was attempted in the past .
early studies used polytropic models ( e.g. @xcite ) ; @xcite , using a complete stellar modelling approach , found a good agreement between their models ( with @xmath38=-0.5 $ ] and @xmath60 ) and the observations of @xcite .
the more precise data available today , however , can not be reconciled with the models of @xcite , the discrepancy being worse for a metal - poor composition with respect to a solar composition @xcite .
modelling non - standard input physics , such as stellar activity and/or magnetic fields , has been attempted as well @xcite .
these effects , however , need not be the only ( nor the dominant ) explanation .
the composition , in fact , which is very poorly constrained , can have a significant impact on @xmath39 and @xmath3 @xcite .
in conclusion , our models match satisfactorily the observational data for all the components of koi-126 . with @xmath38=+0.15 $ ] and @xmath37
, we obtain an age of about @xmath85 gyr ( based mainly on the modelling of koi-126 a ) , which is compatible with the initial estimates of @xcite and with the results of @xcite within the errors .
we are unable , on the other hand , to reconcile our models with the observed radii and @xmath3 of cm dra if we assume the currently estimated metal - poor composition .
we find that the mlt parameter has a modest effect on the structure of fc stars , which is nevertheless significant with respect to the very precise measurements available for these systems for @xmath86 .
for example , the relative change in radius for a star of @xmath87 is up to @xmath88 per cent when @xmath0 goes from @xmath89 to @xmath90 .
the agreement of cm dra can therefore be improved using both higher metallicity and a reduced @xmath0 ( e.g. it can be brought within @xmath6 for @xmath91 and @xmath92 or @xmath93 and @xmath94 ) .
the authors are grateful to s. sofia for reading the manuscript and providing valuable comments and to b. chaboyer for private communication on his work in advance of publication .
fs acknowledges support from the yale institute for biospheric studies through the yibs postdoctoral fellowship . | we present models of the components of the systems koi-126 and cm draconis , the two eclipsing binary systems known to date to contain stars with masses low enough to have fully convective interiors .
we are able to model satisfactorily the system koi-126 , finding consistent solutions for the radii and surface temperatures of all three components , using a solar - like value of the mixing - length parameter @xmath0 in the convection zone , and phoenix nextgen 1d model atmospheres for the surface boundary conditions .
depending on the chemical composition , we estimate the age of the system to be in the range @xmath1@xmath2 gyr .
for cm draconis , on the other hand , we can not reconcile our models with the observed radii and @xmath3 using the current metal - poor composition estimate based on kinematics .
higher metallicities lessen but do not remove the discrepancy .
we then explore the effect of varying the mixing length parameter @xmath0 .
as previously noted in the literature , a reduced @xmath0 can be used as a simple measure of the lower convective efficiency due to rotation and induced magnetic fields .
our models show a sensitivity to @xmath0 ( for @xmath4 ) sufficient to partially account for the radius discrepancies .
it is , however , impossible to reconcile the models with the observations on the basis of the effect of the reduced @xmath0 alone .
we therefore suggest that the combined effects of high metallicity and @xmath0 reduction could explain the observations of cm draconis .
for example , increasing the metallicity of the system towards super - solar values ( i.e. @xmath5 ) yields an agreement within @xmath6 with @xmath7 .
[ firstpage ] binaries : eclipsing convection methods : numerical stars : interiors stars : low - mass |
the past several years have witnessed dramatic progress in our understanding of gluon scattering amplitudes , especially in the maximally supersymmetric @xmath2 super - yang - mills theory ( sym ) .
these advances have provided a pleasing mix of theoretical insights , shedding light on the mathematical structure of amplitudes and their role in gauge / string duality , and more practical results , including impressive new technology for carrying out previously impossible calculations at tree level and beyond .
it has recently been pointed out @xcite that there are reasons to suspect @xmath3 supergravity ( sugra ) to have even richer structure and to be ultimately even simpler than sym . despite great progress
@xcite however , our understanding of sugra amplitudes is still poor compared to sym , suggesting that we are still missing some key insights into this problem .
nowhere is the disparity between our understanding of sym and sugra more transparent than in the expressions for what should be their simplest nontrivial scattering amplitudes , those describing the interaction of 2 particles of one helicity with @xmath4 particles of the opposite helicity . in sym
these maximally helicity violating ( mhv ) amplitudes are encapsulated in the stunningly simple formula conjectured by parke and taylor @xcite and proven by berends and giele @xcite , which we express here ( as throughout this paper ) in on - shell @xmath2 superspace @xmath5 in contrast , all known explicit formulas for @xmath6-graviton mhv amplitudes are noticeably more complicated .
the first such formula was conjectured 20 years ago @xcite and a handful of alternative expressions of more or less the same degree of complexity have appeared more recently @xcite . beyond mhv
amplitudes the situation is even less satisfactory , though the kawai - lewellen - tye ( klt ) relations @xcite may be used in principle to express any desired amplitude as a complicated sum of various permuted squares of gauge theory amplitudes and other factors .
these relations are a consequence of the relation between open and closed string amplitudes , but they remain completely obscure at the level of the einstein - hilbert lagrangian @xcite . in this paper
we present an algorithm for writing down an arbitrary tree - level sugra amplitude .
our result was largely made possible by combining and extending the results of two recent papers . in @xcite an explicit formula for all tree amplitudes in sym
was found by solving the supersymmetric version @xcite of the on - shell recursion relation @xcite , greatly extending an earlier solution @xcite for split - helicity amplitudes only .
we will review all appropriate details in a moment , but for now it suffices to write their formula for the color - ordered sym amplitude @xmath7 very schematically as @xmath8 where the sum runs over a collection of dual superconformal @xcite invariants @xmath9 .
the set @xmath10 is dictated by whether @xmath11 is mhv ( in which case there is obviously only a single term , 1 , in the sum ) , next - to - mhv ( nmhv ) , next - to - next - to - mhv ( nnmhv ) , etc .
our second inspiration is an intriguing formula for the @xmath6-graviton mhv amplitude obtained by elvang and freedman @xcite which has the feature of expressing the amplitude in terms of sums of squares of gluon amplitudes , in spirit similar to though in detail very different from the klt relations .
their formula reads @xmath12 ^ 2 \ , g^{\rm mhv}(1,\ldots , n)\,,\ ] ] where the sum runs over all permutations of the labels @xmath13 through @xmath14 and @xmath15 is a particular ` gravity factor ' reviewed below .
our result involves a natural merger of ( [ eqone ] ) and ( [ eqtwo ] ) , expressing an arbitrary @xmath6-graviton super - amplitude in the form @xmath16 ^ 2 \sum _ { \{\alpha \ } } [ r_\alpha(\lambda_i , \widetilde{\lambda}_i , \eta_i ) ] ^2 \ , g_\alpha(\lambda_i , \widetilde{\lambda}_i)\,.\ ] ] two important features worth pointing out are that the sum runs over precisely the same set @xmath10 that appears in the sym case ( [ eqone ] ) , rather than some kind of double sum as one might have guessed , and that the ` gravity dressing factors ' @xmath17 do not depend on the fermionic coordinates @xmath18 of the on - shell @xmath3 superspace .
all of the ` super ' structure of the amplitudes is completely encoded in the same @xmath19-factors that appear already in the sym amplitudes .
we begin in the next section by reviewing some of the necessary tools for carrying out our calculation . in section
iii we provide detailed derivations of explicit formulas for mhv , nmhv , and nnmhv amplitudes . finally in section
iv we discuss the structure of the gravity dressing factors @xmath17 for more general graviton amplitudes .
we will use the supersymmetric version @xcite of the on - shell recursion relation @xcite @xmath20 where we follow the conventions of @xcite in choosing the supersymmetry preserving shift @xmath21 so that the sum in ( [ sbcf ] ) runs over all factorization channels of @xmath22 which separate particle @xmath23 and particle @xmath6 ( into @xmath24 and @xmath25 , respectively ) .
the value of the shift parameter @xmath26 is chosen so that the shifted intermediate momentum @xmath27 goes on - shell at @xmath28 .
the recursion relation ( [ sbcf ] ) can be seeded with the fundamental 3-particle amplitudes @xcite @xmath29 ) between a physical gravity amplitude @xmath22 and the sum over its ordered subamplitudes @xmath30 .
we draw an arrow indicating the cyclic order of the indices between the special legs @xmath6 and @xmath23.,title="fig : " ] -0.5
cm color - ordered amplitudes in sym have a cyclic structure such that only those factorizations preserving the cyclic labeling of the external particles appear in the analogous recursion ( [ sbcf ] ) .
in contrast , gravity amplitudes must be completely symmetric under the exchange of any particle labels , so vastly more factorizations contribute to ( [ sbcf ] )
. we can deal with this complication once and for all by introducing the notion of an ordered ` gravity subamplitude ' @xmath30 .
these non - physical but mathematically useful objects are related to the complete , physical amplitudes @xmath22 via the relation @xmath31 depicted graphically in .
this decomposition only makes a subgroup of the full permutation symmetry manifest .
however it is the largest subgroup that the recursion ( [ sbcf ] ) allows us to preserve since two external lines are singled out for special treatment .
the relation ( [ main ] ) does not uniquely determine the subamplitudes for a given @xmath22 , since one could add to @xmath30 any quantity which vanishes after summing over permutations .
we choose to define the subamplitudes @xmath32 recursively via ( [ sbcf ] ) restricted to factorizations which preserve the cyclic ordering of the indices , just like in sym theory : @xmath33 this recursion is also seeded with the three - point amplitudes ( [ threepoint ] ) since there is no distinction between @xmath34 and @xmath35 .
note however that unlike the color - ordered sym amplitudes @xmath7 , the gravity subamplitude @xmath30 is _ not _ in general invariant under cyclic permutations of its arguments .
it remains to prove the consistency of this definition .
that is , we need to check that the subamplitudes defined in ( [ mdef ] ) , when substituted into ( [ main ] ) , do in fact give correct expressions for the physical gravity amplitude @xmath22 .
this straightforward combinatorics exercise proceeds by induction , beginning with the @xmath36 case which is trivial and then assuming that ( [ main ] ) is correct up to and including @xmath14 gravitons . for @xmath6 gravitons
we then have @xmath37 the first line is the superrecursion for the physical amplitude , including a sum over all partitions of @xmath38 into two subsets @xmath11 and @xmath39 , not just those which preserve a cyclic ordering . in the second line
we have thrown in a spurious sum over all permutations of @xmath38 at the cost of dividing by @xmath40 to compensate for the overcounting .
this is allowed since we know that @xmath22 is completely symmetric under the exchange of any of its arguments .
inside the sum over permutations we are then free to choose @xmath41 and @xmath42 as indicated on the third line , including the factor @xmath43 to count the number of times this particular term appears . on the fourth line
our prior assumption that ( [ main ] ) holds up to @xmath14 particles allows us to replace @xmath44 inside the sum over permutations .
the last line invokes the definition ( [ mdef ] ) and completes the proof that the physical @xmath6-graviton amplitude may be recovered from the ordered subamplitudes via ( [ main ] ) and the definition ( [ mdef ] ) .
the astute reader may have objected already to ( [ eqtwo ] ) in the introduction .
the sym mhv amplitude ( [ amhv ] ) involves the delta function @xmath45 expressing conservation of the total supermomentum @xmath46 since the square of a fermionic delta function is zero , it would seem that it makes no sense for the quantity @xmath47 ^ 2 $ ] to appear in ( [ eqtwo ] ) . throughout this paper
it will prove extremely convenient to adopt the convention that the square of an @xmath1 superspace expression refers to an @xmath0 superspace expression in the most natural way .
for example , it should always be understood that @xmath48 ^ 2 = \delta^{(16)}(q)\,,\ ] ] where the @xmath49 on the right - hand side is given by the same expression ( [ qdef ] ) but with @xmath50 .
this notation will prove especially useful for lifting results of grassmann integration from @xmath2 to @xmath3 superspace .
this trick works because we can break the @xmath51 symmetry of a @xmath52 integration into @xmath53 by taking @xmath54 for @xmath55 and @xmath56 for @xmath57 .
then every @xmath58 integral can be rewritten as a product of two sym integrals and the @xmath51 symmetry of the answer is restored simply by adopting the convention ( [ convention ] ) . for a specific example
consider the basic sym integral @xmath59 which expresses the superrecursion for the case of mhv amplitudes . by ` squaring ' this formula we immediately obtain the answer for a similar @xmath3 grassmann integral , @xmath60 ^ 2 [ a^{\rm mhv}(-\widehat{p},3,\ldots,\overline{n})]^2 = p^2 \frac{\delta^{(16)}(p ) } { ( \ket{1}{2 } \ket{2}{3 } \cdots \ket{n}{1})^2}\,.\ ] ] note the extra factor of @xmath61 which appears on the right - hand side because we have , for obvious reasons , chosen not to square the propagator @xmath62 on the left .
given the above considerations it should come as no surprise that we will be able to import much of the structure of sym amplitudes directly into our sugra results . therefore we now review the results of @xcite for tree amplitudes in sym . here and in all that follows
we use the standard dual superconformal @xcite notation @xmath63 where all subscripts are understood mod @xmath6 .
we will base our expression for the sugra amplitudes on an expression for the sym amplitudes which is equivalent to , but not exactly the same as the one presented in @xcite .
the reason is that the cyclic symmetry of the yang - mills amplitudes implies certain identities for the invariants @xmath9 appearing in ( [ eqone ] ) .
this symmetry was used in @xcite when solving the recursion relations .
instead it is helpful to have a different expression which is more suitable to the gravity case where the subamplitudes @xmath32 do not have cyclic symmetry . to be precise we need to return to the construction of @xcite and make sure that when considering the right - hand side of the bcf recursion relation we always insert the lower point amplitudes so that leg 1 of the left amplitude factor corresponds to the shifted leg @xmath64 .
we also need to have the leg @xmath6 of the right amplitude factor corresponding to the shifted leg @xmath65 , but this was already the choice made in @xcite .
the expression for all @xmath66 sym amplitudes is given in terms of paths in a particular rooted tree diagram .
here we will be using a different ( but equivalent ) diagram , shown in .
each vertex in the diagram , say with labels @xmath67 , corresponds to a particular dual conformal invariant .
these invariants take the general form @xcite @xmath68 where the chiral spinor @xmath69 is given by @xmath70 as in @xcite this expression needs to be slightly modified when any @xmath71 index attains the lower limit of its range .
we indicate by means of a superscript on @xmath19 the nature of the appropriate modification .
specifically , @xmath72 indicates the same quantity ( [ generalr ] ) but with the understanding that when @xmath73 reaches its lower limit , we need to replace @xmath74 we now have all of the ingredients necessary to begin assembling the complete amplitude , which is given by the formula @xmath75 where @xmath76 is given by the sum over vertical paths in beginning at the root node . to each such path
we associate a nested sum of the product of the associated @xmath19-invariants in the vertices visited by the path .
the last pair of labels in a given @xmath19 are those which are summed first , these are denoted by @xmath77 in row @xmath78 of the diagram .
we always take the convention that @xmath79 and @xmath80 are separated by at least two ( @xmath81 ) which is necessary for the @xmath19-invariants to be well - defined .
the lower and upper limits for the summation variables @xmath82 are indicated by the two numbers appearing adjacent to the line above each vertex .
the differences between the new diagram and the one of @xcite are : 1 .
all pairs of labels in the vertices appear alphabetically in the form @xmath83 .
the edges on the extreme left of the diagram are labeled by @xmath71 rather than @xmath84 , and the summation variables must be greater than or equal to these lower limits @xmath71 .
the edges on the extreme right of the diagram are labelled by @xmath6 rather than @xmath14 , and the summation variables must be strictly less than this upper limit @xmath6 .
all superscripts on @xmath19-invariants which detail boundary replacements are left superscripts ( i.e. for lower boundaries only ) . in a given cluster ,
e.g. the cluster shown in , the superscript associated to the left - most vertex is obtained from the sequence written in the vertex by deleting the final pair of labels and reversing the order of the last two labels which remain .
thus the sequence ends @xmath85 for some @xmath86 .
then proceeding to the right in the cluster , the next vertex has the same superscript , but with alphabetical order of the final pair , i.e. it ends @xmath83 .
going further to the right in the cluster one obtains the relevant superscripts by sequentially deleting pairs of labels from the right .
[ cc][cc]@xmath23 [ cc][cc]@xmath87 [ cc][cc]@xmath88 [ cc][cc]@xmath89 [ cc][cc]@xmath90 [ cc][cc]@xmath91 [ cc][cc]@xmath92 [ cc][cc]@xmath93 [ cc][cc]@xmath13 [ cc][cc]@xmath6 [ cc][cc]@xmath94 [ cc][cc]@xmath95 [ cc][cc]@xmath96 [ cc][cc]@xmath97 [ cc][cc]@xmath98 [ cc][cc]@xmath99 [ cc][cc]@xmath100 [ cc][cc]@xmath101 [ cc][cc]@xmath102 [ cc][cc]@xmath103 [ cc][cc]@xmath104 [ cc][cc]@xmath105 [ cc][cc]@xmath6 [ cc][cc]@xmath106 given the complexity of this prescription it behooves us to illustrate a few cases explicitly .
there is one path of length zero , whose value is simply 1 and this corresponds to the mhv amplitudes , @xmath107 then there is one path of length one which gives the nmhv amplitudes .
we get @xmath108 , summed over the region @xmath109 , as always with the convention that @xmath110
. there are no boundary replacements so we have @xmath111 the two paths of length two give the nnmhv amplitudes .
this time we get superscripts on the @xmath19-invariants as dictated by the rules in point 4 above , @xmath112 continuing to n@xmath113mhv amplitudes we find five paths of length three , giving the following nested sums , @xmath114 \ , .\end{aligned}\ ] ] these examples hopefully serve to illustrate how to write a general sym amplitude , though a more thorough discussion may be found in @xcite .
elvang and freedman have shown that the @xmath6-graviton mhv amplitude may be written in the form and have expressed the amplitude in @xmath3 superspace . ]
@xmath115 ^ 2 g^{\rm mhv}(1,\ldots , n)\ ] ] in terms of @xmath116 the formula ( [ efmhv ] ) is valid for @xmath117 ; @xmath36 will always be treated as a special case with @xmath118 .
comparison of ( [ efmhv ] ) with ( [ main ] ) suggests that we should identify the mhv ordered subamplitude as @xmath119 ^ 2 g^{\rm mhv}(1,\ldots , n)\,.\ ] ] let us now check that our definition ( [ mdef ] ) yields precisely the same expression for the subamplitude ( they may have differed by terms which cancel out when one sums over all permutations in ( [ main ] ) ) . -0.5
cm we will again proceed by induction , assuming that ( [ submhv ] ) satisfies ( [ mdef ] ) for @xmath14 and fewer gravitons . to calculate @xmath120 for @xmath6 gravitons from the definition ( [ mdef ] )
we first note that only the single term @xmath121 contributes , giving @xmath122 as shown in .
the calculation is rendered essentially trivial by plugging in the relations @xmath123
^ 2\ , , \cr m^{\rm mhv}(-\widehat{p},3,\ldots,\overline{n } ) & = [ a^{\rm mhv}(-\widehat{p},3,\ldots,\overline{n})]^2 \,g^{\rm mhv}(-\widehat{p},3,\ldots,\overline{n})\end{aligned}\ ] ] between ordered graviton and yang - mills amplitudes .
the @xmath124 factor in ( [ mhvbcf ] ) comes along for the ride as we perform the @xmath58 integral using the square of the analogous yang - mills calculation as explained above ( [ integralexample ] )
. therefore with no effort we find that ( [ mhvbcf ] ) gives @xmath125 ^ 2 \ , p^2 g^{\rm mhv}(-\widehat{p},3,\ldots,\overline{n})\,.\ ] ] a simple calculation using the shift ( [ shift ] ) now reveals that @xmath126 this completes the inductive proof that the formula ( [ submhv ] ) obtained by elvang and freedman is precisely the mhv case of the ordered subamplitudes that we have defined in ( [ mdef ] ) .
-0.5 cm next we turn our attention to the nmhv amplitude .
the two kinds of diagrams which contribute to the recursion are shown in .
let us begin with @xmath127 , in which case the first diagram is absent and only the term @xmath128 appears in the sum . according to the definition ( [ mdef ] ) we then have @xmath129 ^ 2 \,p^2 \
, g^{\rm mhv}(\widehat{1},2,3,\widehat{p})\cr & \equiv [ a^{\rm nmhv}(1,\ldots,5)]^2 \ , g^{\rm nmhv}(1,\ldots,5)\,.\end{aligned}\ ] ] here , following the example set in the previous subsection , evaluating the grassmann integral leads to the square of the analogous sym result , times the gravity factor @xmath130}{\bra{4}{1}}\,.\ ] ] one can check that this result it is consistent with the known answer ( for example , from the klt relation ) .
let us now turn to the general nmhv case . in the previous section we recalled the sym result obtained in @xcite ,
@xmath131 it was shown in @xcite that the @xmath132 term in ( [ dhnmhv ] ) corresponds to the sum over @xmath133 diagrams in , while the @xmath134 terms arise iteratively from the @xmath135 diagram .
now we claim that the nmhv gravity subamplitude is given by @xmath136 ^ 2 \sum_{i = 2}^{n-3 } \sum_{j = i+2}^{n - 1 } r_{n;ij}^2 g^{\rm nmhv}_{n;ij}\,\ ] ] where @xmath19 is the same dual superconformal invariant ( [ generalr ] ) as in sym and the nmhv gravity factor can be split for future convenience into three parts as follows , @xmath137 to express the gravity factor we introduce the notation @xmath138 which is overkill at the moment but will be fully utilized below when we move beyond the nmhv level . in the numerators only dual conformal chains of @xmath139-matrices appear , while in the denominators the chains are not dual conformal due to the break in the way the labels are arranged .
the break is denoted by the semi - colon in the subscript of @xmath140 while in the denominator of @xmath141 it is immediately after the left - most spinor @xmath142 .
then the first factor in ( [ gnmhv ] ) is given by @xmath143 while the remaining two are @xmath144 to check that the formula ( [ nmhvclaim ] ) is correct it is useful to first have a general formula for @xmath145 , where the shift is defined so that @xmath146 .
this tells us that the shift parameter is given by ( [ zp ] ) , i.e @xmath147}\,.\ ] ] then we have @xmath148 \\ & = \frac{x_{1v}^2 \langle n | x_{1i } | 1 ] - x_{1i}^2 \langle n | x_{1v } | 1 ] } { \langle n | x_{1i } | 1 ] } \\ & = \frac{\langle n | x_{1v } ( x_{1v } - x_{1i } ) x_{1i}|1 ] } { \langle n | x_{1i } | 1 ] } \label{firstchoice } \\ & = \frac{\langle n | x_{1v } x_{iv } x_{1i } | 1]}{\langle n | x_{1i } | 1 ] } \\ & = -\frac { \langle n | x_{nv } x_{vi } x_{i2 } x_{2n } | n \rangle } { \langle n | x_{i2 } x_{2n } |
n \rangle } \equiv - z^{n , v , i,2,n}_{n;i,2,n}\,.\end{aligned}\ ] ] note that instead of writing ( [ firstchoice ] ) we could have alternatively written it as @xmath149 } { \langle n | x_{1i } | 1 ] } \\ & = \frac{\langle n | x_{1i } x_{iv } x_{1v}|1]}{\langle n | x_{1i } | 1 ] } \\ & = \frac{\langle n | x_{ni } x_{iv } x_{v2 } x_{2n } | n \rangle } { \langle n | x_{i2 } x_{2n } | n \rangle } \equiv z^{n , i , v,2,n}_{n;i,2,n}\,.\end{aligned}\ ] ] the freedom to write this factor in these two various forms is useful because in certain cases either one or the other form simplifies by cancelling factors from the numerator and denominator . finally we are set up to check our claim ( [ gnmhv ] ) for the nmhv @xmath124-factor .
we first check the case @xmath150 which comes entirely from mhv @xmath151 mhv diagrams . from these diagrams
we obtain @xmath152 from which we find @xmath153 which is in agreement with equations ( [ gnmhv ] ) to ( [ gr ] ) for the case @xmath150 . for the case @xmath154 we must consider diagrams of the form @xmath155 .
from these diagrams we obtain @xmath156 the sum splits into two contributions , @xmath157 and @xmath158 .
the first gives @xmath159 in agreement with equations ( [ gnmhv ] ) to ( [ gr ] ) for the case @xmath157 . to go from ( [ a=3first ] ) to ( [ a=3second ] )
we have used the fact that @xmath160 where the simplification of the @xmath140-factor is due to a cancellation between its numerator and denominator . for the contributions to ( [ twoterms ] ) where @xmath158 we find @xmath161 which is again in agreement with equations ( [ gnmhv ] ) to ( [ gr ] ) .
the factor @xmath162 completes the factor @xmath163 to @xmath164 just as in the mhv case .
this completes the verification of the formula ( [ nmhvclaim ] ) for nmhv graviton amplitudes .
appendix b contains some notes on extracting nmhv graviton amplitudes from the super - amplitude ( [ nmhvclaim ] ) . in this section
we consider the nnmhv case as an exercise towards finding the general algorithm for all tree - level gravity amplitudes .
the structure of the result is just like in yang - mills and similar to the nmhv case ( [ nmhvclaim ] ) except that we now have two more subscripts on both the yang - mills @xmath19-factors and the gravity factors , @xmath165 ^ 2}= \sum_{2\leq a , b \leq n-1 } r_{n;a b}^2\bigl[\sum_{a \leq c , d < b } ( r^{ba}_{n;ab;cd})^2 h^{(1)}_{n;ab;cd } + \sum_{b \leq c , d < n } ( r_{n;cd}^{ab})^2 h^{(2)}_{n;ab;cd } \bigr]\ , .
\label{mnnmhv}\ ] ] the factors @xmath166 and @xmath167 can be written in the form @xmath168 in this formula @xmath169 , @xmath170 and @xmath171 are defined as before in the case of the nmhv amplitude ( see formulae ( [ f2 ] ) , ( [ gl ] ) and ( [ gr ] ) ) .
the factor @xmath172 in @xmath166 is given by @xmath173 and the factor @xmath174 in the second term in the parentheses is given by @xmath175 finally the new @xmath124-factors are given by @xmath176 let us now check the claim ( [ mnnmhv ] ) . as before we begin with the case @xmath150 which comes purely from nmhv @xmath151 mhv diagrams and mhv @xmath151 nmhv diagrams .
we start by calculating the former kind . from these diagrams
we obtain @xmath177 the sum over @xmath178 splits into two pieces , @xmath179 and @xmath180 . for the terms where @xmath179 we have @xmath181 \bigl [ x_{\widehat{1},i+1}^2 p^{i , n-3}_n\bigr]\ , .
\label{hnnmhva = c=2}\end{aligned}\ ] ] here as in the previous subsection we have used the fact that certain @xmath140-factors simplify . for example , reading the @xmath140-factor from the formula ( [ gl ] ) and taking into account the fact that the spinor @xmath182 can be replaced in both the numerator and denominator of @xmath140 by @xmath183 , we would obtain @xmath184 .
the sequence of indices @xmath185 implies however that one can factor out @xmath186 .
since the sequence is present in both the numerator and the denominator , it can simply be replaced by @xmath13 .
thus we arrive at the form of the @xmath140-factor in the first set of parentheses in ( [ hnnmhva = c=2 ] ) .
to verify that equation ( [ hnnmhva = c=2 ] ) is consistent with ( [ h1 ] ) it remains to substitute the @xmath140-factors appropriate to the factors @xmath187 and @xmath188 .
doing so we obtain @xmath189 \bigl [ -z^{n , i+1,i,2,n}_{n;i,2,n } p^{i , n-3}_n\bigr]\,.\ ] ] the factor @xmath190 gives the required contribution @xmath191 , while the factor in the second factor in square brackets is @xmath192 .
the remaining factor in the first set of square brackets is the contribution from @xmath193 and the other @xmath140 and @xmath141 factors in ( [ h1 ] ) .
now let us look at the terms where @xmath180 .
we have @xmath194 \notag\\ & \bigl [ x_{\widehat{1},i+1}^2 p^{i , n-3}_n\bigr]\,.\end{aligned}\ ] ] again , substituting for @xmath195 and @xmath188 we find agreement with ( [ h1 ] ) .
now let us turn our attention to the latter kind of diagrams , namely the mhv @xmath151 nmhv diagrams . from these diagrams we find @xmath196 as before the sum over @xmath178 splits into two pieces . for @xmath197
we find @xmath198 \bigl[x_{\widehat{1 } d}^2 \bigl(-z^{n , i+1,d , i , n}_{n;d , i , n } p^{i , d-3}_{d , i , n } \bigr ) \bigl(-z^{n , d+1,d , i , n}_{n;d , i , n } p^{d , n-3}_{n } \bigr)\bigr ] \ , , \label{hnnmhv2a = c=2}\end{aligned}\ ] ] while for @xmath199 we find @xmath200 \bigl[x_{\widehat{1},i+1}^2 \bigl(-z^{n , d , c-1}_{n;c-1 } p^{i , c-2}_{n } \bigr ) \bigl(-z^{n , c+1,d , c , n}_{n;d , c , n } p^{c , d-3}_{d , c,,n } \bigr ) \bigl(-z^{n , d+1,d , c , n}_{n;d , c , n } p^{d , n-3}_{n } \bigr)\bigr ] \,.\end{aligned}\ ] ] making the usual substitutions for the factors of the form @xmath201 we find agreement with ( [ h2 ] ) in both cases . to check the terms for @xmath154 we need to consider @xmath202 diagrams .
these diagrams give us @xmath203 \notag \\ & = \sum_{3\leq a , b < n } \!\!\!\ ! r^{2}_{n;ab } \bigl [ \sum_{a\leq c , d < b } \!\!\!\ !
( r^{ba}_{n;ab;cd})^2 p^2 h^{(1)}(\widehat{p},\ldots,\overline{n } ) + \sum_{b\leq c , d < n } \!\!\!\ ! ( r^{ab}_{n;cd})^2 h^{(2)}(\widehat{p},\ldots,\overline{n } ) \bigr]\,.\end{aligned}\ ] ] as in the nmhv case , the sum over @xmath73 splits into a part where @xmath157 and a part where @xmath158 .
the calculation is essentially the same as in the nmhv case , with the factor of @xmath204 providing the necessary piece of @xmath169 in both cases .
this completes the verification of the formula ( [ mnnmhv ] ) for nnmhv amplitudes .
because of the association between vertices in the rooted tree diagram with individual terms appearing in the iterative solution of the recursion relation ( [ sbcf ] ) , it is clear that the procedure applied in the previous section can be generalized to express an arbitrary n@xmath205mhv @xmath6-graviton super - amplitude in the form @xmath206 ^ 2 \sum _ { \{\alpha \ } } [ r_\alpha(\lambda_i , \widetilde{\lambda}_i , \eta_i ) ] ^2 \ , g_\alpha(\lambda_i , \widetilde{\lambda}_i)\,,\ ] ] where @xmath9 are precisely the same dual superconformal invariants ( [ generalr ] ) that appear in sym and @xmath17 are some additional
, non - dual conformally invariant , dressing factors .
explicit formulas for the mhv , nmhv , and nnmhv gravity factors are given respectively in ( [ mhvg ] ) , ( [ gnmhv ] ) , and ( [ h1])([h2 ] ) .
the gravity factors @xmath17 for a general amplitude can be worked out on a case - by - case basis .
they always have the form @xmath207 where @xmath106 is some combination of @xmath208 , @xmath209 and @xmath210 factors .
the iterative construction of any desired amplitude is no more difficult than the examples we have already studied in detail .
actually one only needs to take care of the factor @xmath211 , because the other parts just go from lower points to higher points automatically under the usual rules @xmath212 and @xmath213 as , for example , in going from the nmhv formula ( [ gnmhv ] ) to the nnmhv formula ( [ h1 ] ) and ( [ h2 ] ) .
the @xmath208 factors arise at each level for the simple reason that an extra propagator @xmath61 appears in on - shell recursion for gravity as compared to the ` square ' of the corresponding yang - mills result , a fact which we noted already back in ( [ integralexample ] ) as we already explained carefully in previous section for the nmhv case , the factor @xmath211 is needed to satisfy the recursion relation .
although it is simple to describe the algorithm for a general amplitude in words and by appealing to the examples detailed above , we have not identified a pattern which would allow us to write down a general explicit formula , as was done for sym in @xcite . as noted above each @xmath9 invariant comes with its own @xmath208-type factor , and each path in which ends on a vertex with indices @xmath214 leads to an associated factor of the form @xmath215 where the general @xmath208 , @xmath209 and @xmath210 are suitably defined following the examples in the previous section .
specifically we have @xmath216 the @xmath208 factors can be of two types , @xmath172 and @xmath174 . the first type
are defined as follows , @xmath217 the second type are given by @xmath218 in addition to the factors ( [ endfactors ] ) , other @xmath209 and @xmath210 factors also appear .
if we try the simplest guess which is that we should be able to associate these factors to the vertices in such that every path ending on a given vertex picks up the factors associated to that vertex , then we find that : 1 . for each cluster ( see )
, the leftmost descendant vertex picks up the same factors as the parent vertex , and in addition a @xmath209 factor with the indices of the parent , 2 .
the next descendant vertex to the right is exactly the same , except that the additional factor is a @xmath210 instead of @xmath209 , and 3 .
going further to the right along the descendant vertices , there is a more complicated structure @xmath209 and @xmath210 factors whose indices are modified from those of the parent .
we emphasize that we have attempted here only to illustrate some features of the general structure ; in order to determine precisely the factors which appear for a given path it seems necessary to work out recursively which kinds of n@xmath219mhv@xmath151n@xmath220mhv factorizations that particular path corresponds to .
to stress that the algorithm can be simply exploited to generate higher and higher n@xmath221mhv amplitudes , we give here the formula for n@xmath113mhv amplitudes : @xmath222 ^ 2 \sum_{2 \leq a_1 , b_1 < n } \!\!\!\!\!\ !
r^2_{n;a_1b_1 } \bigl [ \notag \\ & \sum_{a_1 \leq
a_2,b_2 < b_1 } \!\!\!\!\!\ !
( r_{n;a_1b_1;a_2b_2}^{b_1a_1})^2 \bigl(\sum_{a_2 \leq a_3,b_3 < b_2 } \!\!\!\!\!\ ! ( r_{n;a_1b_1;a_2b_2;a_3b_3}^{a_1b_1b_2a_2})^2 g^{(1)}_{n;a_1b_1;a_2b_2;a_3b_3 } \nonumber \\ & \qquad\qquad + \sum_{b_2 \leq a_3,b_3 < b_1 } \!\!\!\!\!\ ! ( r_{n;a_1b_1;a_3b_3}^{a_1b_1a_2b_2})^2 g^{(2)}_{n;a_1b_1;a_2b_2;a_3b_3}+ \sum_{b_1 \leq a_3,b_3 < n } \!\!\!\!\!\ ! ( r_{n;a_3,b_3}^{a_1b_1})^2 g^{(3)}_{n;a_1b_1;a_2b_2;a_3b_3 } \bigr ) \nonumber \\ & + \sum_{b_1 \leq a_2,b_2 < n } \!\!\!\!\!\ ! ( r_{n;a_2,b_2}^{a_1b_1})^2 \bigl(\sum_{a_2 \leq a_3,b_3 < b_2 } \!\!\!\!\!\ ! ( r_{n;a_2b_2;a_3b_3}^{b_2a_2})^2 g^{(4)}_{n;a_1b_1;a_2b_2;a_3b_3 } + \sum_{b_2 \leq a_3,b_3 < n } \!\!\!\!\!\ ! ( r_{n;a_3,b_3}^{a_2b_2})^2 g^{(5)}_{n;a_1b_1;a_2b_2;a_3b_3}\bigr ) \bigr]\,.\end{aligned}\ ] ] the five different @xmath124-factors are in correspondence with the five different vertical paths from the root node to the vertices on the lowest row explicitly shown in .
explicitly they are given by @xmath223 where @xmath124 is shorthand for @xmath224 ( with the same subscripts on both ) .
the expressions we have found can certainly be used in the calculation of loop amplitudes in supergravity .
it is straightforward to apply the generalized unitarity technique in a manifestly supersymmetric way @xcite ; the basic ingredients in this procedure are the tree - level super - amplitudes .
it would of course be extremely interesting to unlock the general pattern of @xmath124-factors to allow one to write down a general explicit formula
. it would also be interesting to see if the sugra ` bonus relations ' @xcite could be usefully exploited beyond mhv amplitudes .
there is no doubt that much additional structure remains to be found .
hopefully , much simpler and more beautiful formulas await than the ones obtained here .
certainly this should be the case if the notion that sugra amplitudes are even simpler than those of sym is to come to full fruition .
we are grateful to nima arkani - hamed , zvi bern , lance dixon , henriette elvang , dan freedman , johannes henn , chrysostomos kalousios , steve naculich and cristian vergu for stimulating discussions and helpful correspondence .
this work was supported in part by the french agence nationale de la recherche under grant anr-06-blan-0142 ( jmd ) , the us department of energy under contract de - fg02 - 91er40688 ( ms ( oji ) and av ) , and the us national science foundation under grants phy-0638520 ( ms ) and phy-0643150 career pecase ( av ) .
here we give some formulae to establish the conventions we are using for the two - component spinors .
we have @xmath225 @xmath226 for the commuting spinors , the following notation has been used , @xmath227 where @xmath228 and @xmath229 are antisymmetric tensors . for the contractions of these spinor variables we write for example @xmath230
@xmath231 = \lambda_i^{\alpha } x_{\alpha \dot\alpha } \widetilde{\lambda}_j^{\dot \alpha } , \hspace{30pt } \langle i | x_1 x_2 \ldots x_{2 m } | j\rangle = \lambda_i^{\alpha } x_{1 \alpha \dot \alpha } x_2^{\dot \alpha \beta } \ldots x_{2 m}^{\dot \gamma \gamma } \lambda_{j \gamma}\,.\ ] ] for the dual coordinates we use @xmath232 similarly for the grassmann odd dual coordinates we have @xmath233
here we provide some additional details regarding the formula for general nmhv super - amplitudes proven in section iii.b , @xmath234 ^ 2 \sum_{s=2}^{n-3 } \sum_{t = s+2}^{n-1 } r_{n;st}^2 g^{\rm nmhv}_{n;st}\,,\ ] ] where the @xmath124-factor is given in ( [ gnmhv ] ) and the dual superconformal invariant is @xmath235 in terms of @xcite @xmath236\,.\ ] ] in order to extract the @xmath6-particle nmhv graviton amplitude from this superspace expression we should perform the integral over @xmath237 for the three negative helicity gravitons @xmath86 .
it is convenient to choose particles 1 and @xmath6 to be two of these three since @xmath238 does not depend on @xmath239 or @xmath240 .
these two variables appear only inside the supermomentum conserving delta function @xmath241 which may be put into the form @xcite @xmath242 the @xmath243 integrals are then trivial , leading to @xmath244 ^ 2 g_{n;st}^{\rm nmhv}\,.\ ] ] here we have chosen , without loss of generality , particle 2 to be the third negative helicity graviton .
the analogous nmhv gluon amplitude simplifies further due to the fact that @xmath245 only depends on @xmath246 when @xmath247 ; thus performing the integral eliminates the sum over @xmath248 @xcite . here in the case of gravity
it is unfortunately cumbersome to proceed analytically because the sum over permutations in ( [ nmhvgraviton ] ) generates many terms , even for the simplest nontrivial case @xmath249 where the sum over @xmath248 and @xmath250 produces just three terms and the corresponding @xmath124-factors simplify considerably , @xmath251 \ , \langle 4|3 + 2|1 ] } { \langle 6|5 + 4|1 ] \ , \langle 6|3 + 2|1]}\ , \\ g^{\rm nmhv}_{6;25 } & = + ( p_5 + p_6)^2 \ket{2}{3 } \bra{3}{4 } \frac { \langle 6|1 + 2|3 + 4|5 + 6| 1 ] } { \ket{5}{6 } \bra{1}{5 } } \frac { \langle 4|5 + 6|1 ] } { \langle 2|5 + 6| 1]}\ , , \\
g^{\rm nmhv}_{6;35 } & = - ( p_1+p_2)^2 \ket{3}{4 } \bra{4}{5 } \ket{5}{6}^2 \frac { \langle 2|3 + 4|5 ] \ , \langle 6|1 + 2|3 ] } { \ket{2}{6}\
, \langle 6|1 + 2|3 + 4|6 \rangle}\,.\end{aligned}\ ] ] therefore we do not provide explicit analytic formulas for graviton amplitudes , which instead may be evaluated numerically as needed .
we have checked numerically that our expression agrees with other representations for the @xmath249 particle nmhv graviton amplitude in the literature ( see for example @xcite ) .
n. arkani - hamed , f. cachazo and j. kaplan , arxiv:0808.1446 [ hep - th ] .
z. bern , d. c. dunbar and t. shimada , phys .
b * 312 * , 277 ( 1993 ) [ arxiv : hep - th/9307001 ] .
z. bern , l. j. dixon , d. c. dunbar , m. perelstein and j. s. rozowsky , nucl .
b * 530 * , 401 ( 1998 ) [ arxiv : hep - th/9802162 ] .
z. bern , living rev .
* 5 * , 5 ( 2002 ) [ arxiv : gr - qc/0206071 ] .
s. giombi , r. ricci , d. robles - llana and d. trancanelli , jhep * 0407 * , 059 ( 2004 ) [ arxiv : hep - th/0405086 ] .
v. p. nair , phys .
d * 71 * , 121701 ( 2005 ) [ arxiv : hep - th/0501143 ] .
f. cachazo and p. svrcek , arxiv : hep - th/0502160 . n. e. j. bjerrum - bohr , d. c. dunbar , h. ita , w. b. perkins and k. risager , jhep * 0601 * , 009 ( 2006 ) [ arxiv : hep - th/0509016 ] .
n. e. j. bjerrum - bohr , d. c. dunbar and h. ita , nucl .
* 160 * , 215 ( 2006 ) [ arxiv : hep - th/0606268 ] .
z. bern , l. j. dixon and r. roiban , phys .
b * 644 * , 265 ( 2007 ) [ arxiv : hep - th/0611086 ] .
p. benincasa , c. boucher - veronneau and f. cachazo , jhep * 0711 * , 057 ( 2007 ) [ arxiv : hep - th/0702032 ] .
z. bern , j. j. carrasco , l. j. dixon , h. johansson , d. a. kosower and r. roiban , phys .
* 98 * , 161303 ( 2007 ) [ arxiv : hep - th/0702112 ] . n. arkani - hamed and j. kaplan , jhep * 0804 * , 076 ( 2008 ) [ arxiv:0801.2385 [ hep - th ] ] .
r. kallosh and m. soroush , nucl .
b * 801 * , 25 ( 2008 ) [ arxiv:0802.4106 [ hep - th ] ] .
s. g. naculich , h. nastase and h. j. schnitzer , jhep * 0811 * , 018 ( 2008 ) [ arxiv:0809.0376 [ hep - th ] ]
. m. bianchi , h. elvang and d. z. freedman , jhep * 0809 * , 063 ( 2008 ) [ arxiv:0805.0757 [ hep - th ] ] .
s. g. naculich , h. nastase and h. j. schnitzer , nucl .
b * 805 * , 40 ( 2008 ) [ arxiv:0805.2347 [ hep - th ] ] .
n. e. j. bjerrum - bohr and p. vanhove , jhep * 0810 * , 006 ( 2008 ) [ arxiv:0805.3682 [ hep - th ] ] .
z. bern , j. j. m. carrasco and h. johansson , phys .
d * 78 * , 085011 ( 2008 ) [ arxiv:0805.3993 [ hep - ph ] ] .
r. kallosh , arxiv:0808.2310 [ hep - th ] .
z. bern , j. j. m. carrasco , l. j. dixon , h. johansson and r. roiban , arxiv:0808.4112 [ hep - th ] .
l. mason , d. skinner , arxiv:0808.3907 [ hep - th ] .
s. badger , n. e. j. bjerrum - bohr and p. vanhove , arxiv:0811.3405 [ hep - th ] .
r. kallosh and t. kugo , arxiv:0811.3414 [ hep - th ] .
r. kallosh , c. h. lee and t. rube , arxiv:0811.3417 [ hep - th ] .
s. j. parke and t. r. taylor , phys .
. lett . * 56 * , 2459 ( 1986 ) .
f. a. berends and w. t. giele , nucl .
b * 306 * , 759 ( 1988 ) .
f. a. berends , w. t. giele and h. kuijf , phys .
b * 211 * , 91 ( 1988 ) .
j. bedford , a. brandhuber , b. j. spence and g. travaglini , nucl .
b * 721 * , 98 ( 2005 ) [ arxiv : hep - th/0502146 ] .
z. bern , j. j. carrasco , d. forde , h. ita and h. johansson , phys .
d * 77 * , 025010 ( 2008 ) [ arxiv:0707.1035 [ hep - th ] ] .
h. elvang and d. z. freedman , jhep * 0805 * , 096 ( 2008 ) [ arxiv:0710.1270 [ hep - th ] ]
. m. spradlin , a. volovich and c. wen , arxiv:0812.4767 [ hep - th ] .
h. kawai , d. c. lewellen and s. h. h. tye , nucl .
b * 269 * , 1 ( 1986 ) .
z. bern and a. k. grant , phys .
b * 457 * , 23 ( 1999 ) [ arxiv : hep - th/9904026 ] .
s. ananth and s. theisen , phys .
b * 652 * , 128 ( 2007 ) [ arxiv:0706.1778 [ hep - th ] ] .
j. m. drummond and j. m. henn , arxiv:0808.2475 [ hep - th ] .
j. m. drummond , j. henn , v. a. smirnov and e. sokatchev , jhep * 0701 * , 064 ( 2007 ) [ arxiv : hep - th/0607160 ] .
j. m. drummond , g. p. korchemsky and e. sokatchev , nucl .
b * 795 * , 385 ( 2008 ) [ arxiv:0707.0243 [ hep - th ] ] .
j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , arxiv:0807.1095 [ hep - th ] .
a. brandhuber , p. heslop and g. travaglini , arxiv:0807.4097 [ hep - th ] .
r. britto , f. cachazo and b. feng , nucl .
b * 715 * , 499 ( 2005 ) [ arxiv : hep - th/0412308 ] .
r. britto , f. cachazo , b. feng and e. witten , phys .
lett . * 94 * , 181602 ( 2005 ) [ arxiv : hep - th/0501052 ] .
r. britto , b. feng , r. roiban , m. spradlin and a. volovich , phys .
d * 71 * , 105017 ( 2005 ) [ arxiv : hep - th/0503198 ] .
j. m. drummond , j. henn , g. p. korchemsky and e. sokatchev , arxiv:0808.0491 [ hep - th ] . | we present an algorithm for writing down explicit formulas for all tree amplitudes in @xmath0 supergravity , obtained from solving the supersymmetric on - shell recursion relations .
the formula is patterned after one recently obtained for all tree amplitudes in @xmath1 super yang - mills which involves nested sums of dual superconformal invariants .
we find that all graviton amplitudes can be written in terms of exactly the same structure of nested sums with two modifications : the dual superconformal invariants are promoted from @xmath2 to @xmath3 superspace in the simplest manner possible by squaring them and certain additional non - dual conformal gravity dressing factors ( independent of the superspace coordinates ) are inserted into the nested sums . to illustrate the procedure we give explicit closed - form formulas for all nmhv , nnmhv and nnnmv gravity super - amplitudes .
brown - het-1572 lapth-1291/08 1 cm |
magnetic reconnection is the process that enables to convert magnetic energy into heat and kinetic energy of accelerated particles @xcite . in
many astrophysical and laboratory plasma situations plasma beta ( the ratio of thermal and magnetic pressures ) is much smaller than unity .
thus , magnetic reconnection has attracted a considerable attention as the plasma heating and charged particle acceleration mechanism .
it is believed that magnetic reconnection is responsible for solar @xcite and stellar flares .
it has been also observed in the earth geomagnetic tail @xcite .
this work has several motivations : \(i ) the longstanding problems with resistive mhd ( magnetohydrodynamic ) , i.e. _ collisional _ description of the magnetic reconnection has been its too slow rate if a consistent , the - first - principles approach is used @xcite ; or a fast rate @xcite , but a lack of a proper motivation ( @xcite , p. 79 ) ( one can still get the fast reconnection if the resistivity is enhanced in the diffusion region , as one expects in physical applications ) . in the _ collisionless _ regime there is no known simple , analytical reconnection rate ( e.g. analogous to sweet - parker , or petschek rates ) .
what does exist , however , is a bulk of mostly numerical simulation work @xcite .
it should be mentioned that a significant progress ( including analytical ) has been made in study of the details of collisionless reconnection such as the diffusion region structure , outflows , reconnection rates @xcite .
however these were either specific to a narrow class of collisionless reconnection models , namely tearing unstable harris current sheet ; or too slow reconnection rates were obtained @xcite .
some numerical simulation models @xcite can produce as fast reconnection rates as @xmath7 where @xmath8 is the inflow alfvnic mach number .
the lack of a simple , analytical model of collisionless reconnection ( _ that is not specific to any initial configuration _ ) has been somewhat hampering the progress in understanding .
\(ii ) quite often , particularly in space plasmas , there is a good reason to go beyond resistive , single - fluid mhd .
a simple line of reasoning can be outlined on an example of e.g. solar corona .
fixing coronal temperature at @xmath9 k , coulomb logarithm at 18.1 , the lundquist number ( using spitzer resistivity ) is @xmath10 . here
@xmath11 m was used .
one of the arguments for going beyond resistive mhd is comparing typical width of a sweet - parker current sheet @xmath12 m to the ion inertial length .
typical scale associated with the hall term in the generalised ohm s law at which deviation from electron - ion coupled dynamics is observed is , @xmath13 m. here particle density of @xmath14 m@xmath15 is used .
hence , the fact that @xmath16 warrants going _ beyond single fluid resistive mhd approximation _ ( a similar conclusion is reached by @xcite in their figure 12 based on laboratory plasma experiment known as mrx ( magnetic reconnection experiment ) ) .
\(iii ) previous results on collisionless reconnection both in tearing unstable harris current sheet @xcite and stressed x - point collapse @xcite has shown that magnetic field is frozen into electron fluid and the term in the generalised ohm s law that is responsible for breaking the frozen - in condition is electron pressure tensor off - diagonal ( non - gyrotropic ) components . thus ,
a need for inclusion of the _ electron pressure tensor non - gyrotropic components _ in a model of collisionless reconnection has become clear .
\(iv ) there is a growing amount of work @xcite that suggests that in the collisionless regime , on the scales less than @xmath17 magnetic field is frozen into the electron fluid rather than bulk of plasma .
one can write in general @xmath18 .
this relation clearly shows that in collisional regime ( when the number density @xmath19 is large ) , the difference between electron and ions speeds diminishes @xmath20 .
however , as one enters collisionless regime ( when the number density @xmath19 is small ) the deviation between electron and ion speeds starts to show .
in @xcite we proposed a possible explanation why the reconnection is fast when the hall term is included .
inclusion of the latter means that in the reconnection inflow magnetic field is frozen into _ electron _ fluid .
as it was previously shown in @xcite ( see their figs.(7 ) and ( 11 ) ) speed of electrons , during the reconnection peak time , is at least 4 - 5 times greater than that of ions .
this means that electrons can bring in / take out the magnetic field attached to them into / away from the diffusion region much faster than in the case of single fluid mhd which does not distinguish between electron - ion dynamics .
thus , a need for inclusion _ magnetic field transport by electrons _ in a model of collisionless reconnection has become clear .
\(v ) a pioneering work of @xcite has demonstrated a clear transition from collisional to collisionless reconnection regimes by varying the plasma density , and established that in the collisionless regime the reconnection rate is much greater than the sweet - parker rate . despite a well motivated explanation of their experimental results ,
one was surprised to see only experimental data points and numerical simulation results on their figure 12 .
thus , we set out with a task of formulating a simple model of magnetic reconnection in a collisionless regime . in what follows we shall be using well - motivated arguments of sweet - parker model , but shall be applying it to the collisionless case
it is instructive to re - iterate key points of the sweet - parker model .
the first step towards the derivation of the reconnection rate is that the plasma outflow speed from the diffusion region is of the order of the alfvn speed , @xmath21 .
this follows from the following consideration : taking the fluid to be incompressible and assuming a steady state condition , one obtains @xcite , p. 285
@xmath22 where @xmath23 and @xmath24 are thermal pressures inside and outside the diffusion region .
this pressure difference can be set to the magnetic pressure @xmath25 .
from which the above result @xmath26 readily follows .
the second step is applying the continuity equation @xmath27 where @xmath3 and @xmath28 are the diffusion region length and width respectively .
the third step is using the generalised ohm s law ( e.g. @xcite p. 108 ) @xmath29 where @xmath30 and @xmath31 are electric and magnetic fields , @xmath32 is plasma velocity , @xmath33 is pressure tensor ( @xmath34 matrix ) , @xmath19 is plasma number density , @xmath35 is mass and @xmath36 is electric charge .
subscript @xmath36 stands for an electron . in eq.(3 ) we balance two terms advection ( @xmath37 ) and resistive diffusion ( @xmath38 ) : @xmath39 where @xmath40 is used , with @xmath41 . note that in the ( collisional ) resistive mhd regime all other terms in the generalised ohm s law are insignificant and @xmath20 .
finally , substituting @xmath42 from eq.(2 ) and after some simple algebra one arrives at @xmath43 where @xmath44 , the lundquist number has been introduced .
eq.(5 ) constitutes the classical sweet - parker reconnection rate . in order to obtain collisionless reconnection rate in an analogy with the above derivation of the classical sweet - parker rate
, we now balance the advection ( @xmath37 ) with electron pressure tensor ( @xmath45 ) , because now spitzer resistivity in the collisionless regime is not large enough and also @xmath46 .
previous results of the reconnection in the collisionless regime , both in tearing unstable harris current sheet @xcite and stressed x - point collapse @xcite , have shown that : ( i ) the term in the generalised ohm s law that is responsible for breaking the frozen - in condition is electron pressure tensor non - gyrotropic components and ( ii ) the magnetic field is frozen into electron fluid .
@xcite formulate two dimensional model , which shows that in a steady state , at the magnetic null ( x - point ) where the electron flow velocities and magnetic field are zero , the only terms in the generalized ohm s law that can support out - of - plane electric field ( and hence the reconnection ) are electron pressure tensor non - gyrotropic components ( see their eqs.(4)-(6 ) ) : @xmath47 here the two @xmath48- and @xmath49-coordinates are along ( outflows ) and across ( inflows ) the current sheet respectively .
further simplification can be made observing that @xmath50 but @xmath51 because width of the current sheet is much smaller than its length : @xmath52 where @xmath53 is the width of the current sheet in the collisionless regime ( @xmath54 ) .
thus in an analogy with eq.(4 ) we now have @xmath55 note that the latter equation is similar to eq.(11 ) from @xcite .
in addition to balancing the different terms , note also that the two crucial differences from eq.(4 ) are that ( i ) we use electron speed @xmath56 because magnetic field advection on scales less than @xmath17 is done by electrons and ( ii ) width of the diffusion region ( current sheet ) is @xmath57 which we shall specify below .
let us now specify electron pressure tensor non - gyrotropic component @xmath58 . based in earlier works , which use second moment of vlasov equation ( i.e. multiplying the vlasov equation by @xmath59 and integrating over the velocity space ) , @xcite give explicit expression for the electron pressure tensor components .
intermediate expression for @xmath58 in the static case , which is generic and yet is not specific to tearing unstable harris current sheet configuration , is given by eq.(14 ) in @xcite : @xmath60 where @xmath61 takes into account linear variation of magnetic field with distance in the vicinity of the magnetic null .
the electron pressure tensor non - gyrotropic ( off - diagonal ) components are generally much smaller than gyrotropic ( diagonal ) ones .
the deviations from gyrotropic pressure are possible due to electron meandering motion in the regions of weak magnetic field close to the x - point @xcite .
away from the x - point the particles are magnetized and pressure is isotropic .
thus , it would be reasonable to assume that electron meandering motion will be effective up to @xmath62 . here
@xmath63 is the electron larmor radius .
@xmath64 is electron thermal speed and @xmath65 is electron cyclotron frequency .
further away from the magnetic null , when @xmath66 then pressure becomes isotropic .
thus in the above expression for @xmath67 we set @xmath62 , i.e. @xmath68 . in eq.(9 ) we can also assume the gyrotropic pressure component @xmath69 is of the order of the magnetic pressure @xmath25 , set @xmath70 as the electron inflow speed @xmath71 , also estimate @xmath72 .
thus for the @xmath58 we have @xmath73 naturally , the next step is to specify width of the diffusion region in the collisionless regime , which we do by using conservation of mass , but again taking into account that now magnetic field is advected by the electrons .
thus , in lieu of eq.(2 ) we have @xmath74 where @xmath75 is the _ electron _ alfvn speed . thus substituting eq.(10 ) into eq.(8 ) , using eq.(11 ) and defining the collisionless inflow alfvnic mach number , @xmath76 as @xmath77 , after simple algebra ( also noting that @xmath78 and @xmath79 ) we obtain @xmath80 note that factor @xmath81 has been omitted because this rate is a crude estimate .
eq.(12 ) can be regarded as the main result of this work , as it provides the collisionless reconnection rate independent of an initial configuration .
simple analysis shows that collisionless reconnection rate @xmath82 e.g. for solar coronal parameters ( @xmath83 m@xmath15 , @xmath84 k , @xmath11 m , @xmath85 t ( 100 gauss and hence @xmath86 m s@xmath87 ) , @xmath88 ) is two orders of magnitude faster than the classical sweet - parker rate @xmath89 . in the context of solar flares
this means that the collisionless model , presented here , effectively alleviates longstanding problem of the interpretation of solar flares by means of magnetic reconnection .
the reconnection rate is also interpreted as the ratio of alfvn time ( @xmath90 s ) and resistive ( or reconnection ) times .
this means in the sweet - parker model resistive ( or reconnection ) time is @xmath91 s @xmath92 days .
while our model provides reconnection time of @xmath93 s @xmath94 minutes , which is commensurate of solar flare observations .
it should be mentioned that petschek model also can provide an appropriately fast reconnection rate , although its justification from the fundamental point of view is not without a debate @xcite .
we now shall compare the master equation eq.(12 ) to the mrx experimental data of @xcite .
their figure 12 presents how the empirical reconnection rate scales with the parameter @xmath95 , which is essentially the ratio of ion inertial length and the width of the sweet - parker current sheet . @xcite
have clearly established that in the collisionless regime , i.e. when @xmath96 the normalised reconnection rate ( after starting from the sweet - parker rate ) attains values much larger than unity . in order to compare our scaling law , eq.(12 ) to the data from ref.@xcite , we need to normalise the collisionless reconnection rate by the sweet - parker rate @xmath97 .
also , in order for the results to be applicable in the collisional regime too , we combine collisional and collisionless rates as follows : @xmath98 . noting that @xmath99 we obtain , @xmath100 note that despite a factor @xmath101 appearing in eq.(13 ) , @xmath102 does not depend on ion mass as @xmath103 .
also the same is evident from the fact that @xmath76 ( according to eq.(12 ) ) only contains physical quantities pertaining to an electron , and @xmath104 is independent of ion mass , so should be their sum .
our collisionless reconnection rate seems to fit the experimental data @xcite and the two - fluid simulation results @xcite well in the range @xmath96 .
it should be noted that each experimental data point was obtained for different plasma parameters .
hence each data point has its own @xmath105 associated with it .
@xmath105 changes from roughly @xmath106 to @xmath107 as we move in the lowest measured values of @xmath108 to @xmath109 in the horizontal axis ( m. yamada , private communication ) . in order to provide theoretical _ curves
_ rather than sets of the theoretical points we argue that in eq.(13 ) the dependence on @xmath105 is weak ( via @xmath110 )
. therefore , solid curve in fig.1 is produced using a fixed value of @xmath111 .
other parameters used in fig .
1 were @xmath112 ev , @xmath113 m and @xmath114 tesla .
some deviation in the region of @xmath115 can be explained by , perhaps , several factors that were not taken into account in our simple model .
let us mention the obvious two : ( i ) strictly speaking our reconnection rate eqs.(12 ) or ( 13 ) was obtained assuming a steady state , while as @xcite mention , the pull magnetic reconnection lasts for about 40 @xmath116s .
although this is much longer than the typical alfvn time @xmath117s , yet one can argue that _ nearly _ steady state was achieved in the experiment .
thus the comparison between the theory and experiment should be taken with caution .
( ii ) our theoretical reconnection rate , naturally , does not include contributions from turbulence . despite these shortcomings
the match between the theory and experiment seems good . based on eqs.(11 ) and
( 12 ) it is also straightforward to derive the width of the diffusion region ( current sheet ) in the collisionless regime : @xmath118 which is independent of global reconnection scale @xmath3 and is only prescribed by micro - physics ( electron inertial length , @xmath119 , and electron larmor radius , @xmath5 ) .
@xcite have provided mrx laboratory experimental data and 2d particle - in - cell simulation results of scaling of the electron diffusion region width , @xmath120 , with the electron inertial length , @xmath121 .
it was found that the experimental data can be fitted with a straight line @xmath122 , while pic simulation data consistently produced a less steep scaling of @xmath123 .
naturally , we tried to apply the diffusion region ( current sheet ) width formula from our model ( eq.(14 ) ) to @xcite .
useful sketch to aid visualisation can be found in @xcite , p.91 , fig.3.1 .
when plasma inflows in the diffusion region ions decouple from electrons at ion inertial length scale of @xmath124 , while electrons deflect from the diffusion region ( and hence forming it ) at electron inertial length scale of @xmath125 .
one of the convincing findings of @xcite was that the electron diffusion region width , @xmath120 , is independent of which ion species are used in the experiment ( h , d@xmath126 or he ) . thus one can conclude that they indeed were observing electron diffusion region .
this corroborates the fact that at the electron inertial length scales the magnetic field transport is done my electrons , as well as the currents are carried by electrons .
our width of the diffusion region according to eq.(14 ) is shown as solid curve in fig.2 .
a reasonably good fit is obtained , supporting the idea that eq.(14 ) indeed provides a good theoretical expression for the electron diffusion region width .
let us estimate the coronal heating flux which is produced by our collisionless reconnection model .
one can obtain the heating flux ( w m@xmath127 ) by taking magnetic energy density @xmath128 ( j m@xmath15 ) , multiplying it by a typical scale @xmath129 m and dividing by the reconnection times which we estimated above , @xmath130 s ( 2 min ) and @xmath131 s ( 0.1 days ) assuming the reconnection time is a good proxy for the energy release time in the solar corona . fixing @xmath85 t , in the case of collisionless reconnection , the heating flux
is obtained @xmath132 w m@xmath127 , which is more than enough to meet the coronal heating requirement ; while as expected sweet - parker model produces an equivalent of 450 w m@xmath127 .
@xcite , p. 359 quotes solar active region typical heating requirement value of 2000 w m@xmath127 .
the obtained collisionless reconnection heating flux is an order of magnitude larger than that .
naturally , neither all of the collisionless reconnection heating flux will go into dissipation ( by definition collisionless process is dissipationless ) , nor all 100% of the magnetic energy will be released by the reconnection process . even if 1/10th of it will ultimately end up as the heat , then the coronal heating requirement can be met .
thus , the importance of the present model for the coronal heating seems evident .
next , let us comment on the observational consequences of the collisionless reconnection model in solar coronal context . the reconnection time ( the flare energy release time ) ,
@xmath133 ( because , @xmath134 , and @xmath135 ) .
thus , if we assume that magnetic field and plasma density remains the same as we go from small to large flares ( this assumption implies that for simplicity of the argument @xmath136 in @xmath137 and @xmath121 in eq.(12 ) stay the same ) , the flare time should increase ( quadratically ) with flare size .
there is some ambiguity how to define actual spatial size of a flare ( dr . e. kontar of university of glasgow , private communication ) : at high energies one sees loop footpoints only ; hence is the geometric size a distance between the footpoints or total ( unknown ) area ?
in soft x - rays ( 6 - 10 kev ) only the looptop source is observed .
x - ray telescope ( xrt ) on board of japanese space mission hinode often sees the entire flaring loop . however , this mission is quite recent and no good flare statistics exists as yet .
@xcite claim that the flare duration increases with increasing the flare importance class , hence with the flare area , as reported by a number of papers ( see their table 7 ) . based on a statistical analysis of almost 50 000 soft x - ray flares observed during the period 1976 - 2000 @xcite
show that the scatter plots of the flare duration , rise time and decay time as function of the flare fluence ( j m@xmath127 ) show strong correlations . whether the latter is a good proxy of the flare geometrical size is not entirely certain .
@xcite report a statistical study of flares observed with the soft x - ray telescope ( sxt ) on board of yohkoh in the year 2000 .
they measured physical parameters of 77 flares .
their fig .
( 3 ) plots the spatial scale @xmath3 against the flare timescale and shows that the spatial scale @xmath3 tends to be larger with the increasing timescale .
thus , in summary , our scaling seems to be in agreement with the solar flare observations .
one should be aware of the fact that in the solar flare observations we see only post - reconnection dynamics and the geometric size of `` particle accelerator '' , the location of actual energy deposition site is never resolved , due to poor spatial resolution of the instruments .
some source of a concern is also the aspect ratio of the diffusion region ( i.e. the current layer , where particles are accelerated ) . from eq.(14 ) it is clear for @xmath138 the aspect ratio is about 7500 : 1 .
the question is whether such fine scale structure can survive in the turbulent corona .
however , since current observations do not have enough resolution to prove or disprove existence if such elongated current layers , the jury is still out . on the laboratory plasma side , comparison of our model scaling with the mrx data ( see fig.(2 ) )
needs a clarification as to why pic simulation results ( open squares ) do not fit the experimental data while our simple analytical model does . at first sight
one might expect the opposite to be true because pic simulation includes all relevant physics , i.e. all relevant terms in the generalized ohm s law , time dependence , etc . in author
s opinion the source of the discrepancy are the boundary conditions and the mass ratio mismatch .
@xcite ( see their fig .
3 ) either use boundary conditions commensurate to mrx ( conducting boundary conditions for electromagnetic fields and elastic reflection of particles at wall ) for unrealistic mass ratio of 400 , or open boundary conditions for a realistic mass ratio of 1836 ( for hydrogen only ) .
it is no surprise that the outcomes of numerical simulations depend on correct ( appropriate ) boundary conditions used .
there are two issues here : first , when mrx boundary conditions are used , it may well be that a better fit could have been achieved , if the correct mass ratio were used ( note that experimental data points lie on the line ( @xmath139 ) that is factor of @xmath140 5 above pic simulation line @xmath141 .
accidentally such is also the mass ratio mismatch @xmath142 .
note , however , that the theoretically derived width according to eq.(14 ) scales as @xmath143 ) .
hence if our analytical treatment is correct , eq.(14 ) still does not alleviate the problem in full , as @xmath144 can not account for the factor of 5 difference between the mrx data and pic simulations .
but the trend is in the right direction ; secondly , when the correct mass ratio is used ( in ref.@xcite , fig.3 ) open boundary conditions are not a good representation of the experiment . hence the results based on open boundary conditions should be discarded .
thus , we conjecture that when simultaneously ( a ) the correct mass ratio and ( b ) boundary conditions commensurate to mrx are used , than pic simulation results will possibly follow our theoretical result ( eq.(14 ) ) . until such simulation is performed , however , the jury in this issue is still out .
some clarification is necessary concerning the use of notation for @xmath3 . in eq.(11 ) it plays a role of the _ electron _ diffusion region length ; i.e. the region of space where the electron dynamics is decoupled from that of ions . since in the solar flare observations the geometric size of `` particle accelerator '' , the location of actual energy deposition site is never resolved , due to poor spatial resolution of the instruments , we use @xmath11 m , a typical length scale in the corona . in context of mrx , we use the length of the magnetic scale @xmath113 m. however , on the mhd scale ( to which our numerical estimates apply ) one also expects that at some distance downstream from the x - point , electron and ion dynamics becomes coupled again .
this implies that the outflowing electrons will decelerate from the electron alfvn speed to ion alfvn seed .
recent pic simulations @xcite suggest that the electron diffusion region length can extend for much longer distances downstream than previously thought , and hence alleviating , in part , the above problem .
a new model of collisionless reconnection is formulated that is based on simple conservation laws . the obtained collisionless reconnection rate , @xmath0 ,
naturally does not depend on lundquist number , @xmath105 , ( because it is collisionless ) and is much faster than the sweet - parker rate , but yet somewhat slower than the petschek rate ( at least for solar coronal plasma parameters ) .
in particular , for the same set of parameters e.g. for solar coronal plasma the rates are @xmath145 , @xmath138 , @xmath146 .
the main implication for solar flares is that if this collisionless rate is used , flare time can be as short as a few minutes , that is commensurate with the observations .
note that the formulated collisionless reconnection model is general and is not specific to any initial configuration e.g. harris type tearing unstable current sheet , x - point collapse or otherwise . in differ to previous results
it relates the reconnection rate to simple , generic spatial scales such as electron inertial length , larmor radius and global reconnection length .
therefore it is easily applicable to different space or laboratory plasma situations .
author would like to thank prof .
m. yamada and his team at mrx princeton plasma physics laboratory for kindly providing their experimental data set and for useful replies to queries .
useful discussions with dr .
g. vekstein and other participants of fifths international cambridge workshop on magnetic reconnection 2008 , bad honnef , germany are acknowledged .
author acknowledges useful discussion of solar flare observational aspects with dr .
e. kontar .
author also would like to thank an anonymous referee who helped to improve the paper significantly .
this research was supported by the united kingdom s science and technology facilities council ( stfc ) . | based on the first principles ( i.e. ( i ) by balancing the magnetic field advection with the term containing electron pressure tensor non - gyrotropic components in the generalised ohm s law ; ( ii ) using the conservation of mass ; and ( iii ) assuming that the weak magnetic field region width , where electron meandering motion supports electron pressure tensor off - diagonal ( non - gyrotropic ) components , is of the order of electron larmor radius ; ) a simple model of magnetic reconnection in a collisionless regime is formulated .
the model is general , resembling its collisional sweet - parker analogue in that it is not specific to any initial configuration e.g. harris type tearing unstable current sheet , x - point collapse or otherwise .
in addition to its importance from the fundamental point of view , the collisionless reconnection model offers a much faster reconnection rate ( @xmath0 ) than sweet - parker s classical one ( @xmath1 ) .
the width of the diffusion region ( current sheet ) in the collisionless regime is found to be @xmath2 , which is independent of global reconnection scale @xmath3 and is only prescribed by micro - physics ( electron inertial length , @xmath4 , and electron larmor radius , @xmath5 ) . amongst other issues , the fastness of the reconnection rate alleviates e.g. the problem of interpretation of solar flares by means of reconnection , as for the typical solar coronal parameters the obtained collisionless reconnection time can be a few minutes , as opposed to sweet - parker s equivalent value of @xmath6 a day . the new theoretical reconnection rate is compared to the mrx ( magnetic reconnection experiment ) device experimental data by @xcite and a good agreement is obtained . |
there has been renewed interest in directional bayesian analysis in view of its fundamental applications to molecular biology @xcite . due to chemical constraints on the bonds of biomolecules , the geometry of these molecules can be described by a set of angles .
other applications include locating and tracking an electric signal @xcite and the analysis of forensic fingerprint evidence @xcite .
all these applications involve circular data which is naturally modelled by the von mises distribution .
the probability density function of the von mises distribution with mean @xmath0 on the unit circle and concentration parameter @xmath1 is given by @xcite @xmath2 where @xmath3 is the modified bessel function of the first kind and order @xmath4 .
the circular variance can be described by @xmath5 where @xmath6 .
let @xmath7 be a vector of observations from a von mises distribution .
when a conjugate prior is used , the posterior distribution of the mean @xmath8 is itself von mises , and can be easily sampled via @xcite .
let @xmath9 be the conjugate prior for the concentration .
the posterior is @xmath10 where @xmath11 and @xmath12 are observed constants ; in this case @xmath13 and @xmath14 . for the case @xmath15 and @xmath16 the normalization constant is @xmath17 , however in general the normalization constant is intractable @xcite . in this paper we shall call the _ bessel exponential distribution_. existing algorithms to sample tend to generate from approximate distributions @xcite or have a large overhead of sampled auxiliary variables @xcite .
we present a new , extremely fast algorithm to sample from the bessel exponential distribution . for large @xmath18 ,
@xmath19 is approximately @xmath20 ( * ? ? ?
9.7.1 ) . plugging this approximation into yields a gamma density with shape @xmath21 and rate @xmath22 .
this insight motivates us to use a gamma - based acceptance - rejection sampler for @xmath18 .
however , the above approximation for @xmath19 breaks down for small @xmath18 , and thus great care is needed to ensure our rejection sampler is efficient for all values of @xmath18 .
we derive the optimal gamma - based proposal distribution and show that the resulting sampler has an acceptance probability of at least 0.7 for all @xmath23 and @xmath24 .
the minimum acceptance probability of @xmath25 occurs when the distribution is concentrated around @xmath26 .
the algorithm is described in section [ sect : alg ] and derived in section [ sect : derivation ] .
enhancements are considered in section [ sect : speedup ] and the algorithm s efficiency is explored in section [ sect : efficiency ] .
as discussed above , we can approximate the bessel exponential distribution with a gamma distribution .
however , because the ratio of these densities diverges as @xmath27 , we can not directly use a gamma proposal for our rejection sampler .
instead we propose values @xmath28 where @xmath29 and @xmath30 has a gamma distribution .
this is an application of marsaglia s exact approximation procedure , see @xcite for more details . using a shifted gamma proposal with shape @xmath31 and
scale @xmath32 leads to the envelope function @xmath33 for @xmath34 , where the amplitude @xmath35 is chosen to ensure the ratio @xmath36 is bounded below one .
we can generate a sample from the bessel exponential distribution by generating a sample @xmath18 from @xmath37 and accepting it with probability @xmath38 where @xmath39 and @xmath40 .
the samples generated via this procedure have the correct bessel exponential distribution for any choice of the proposal parameters @xmath41 , though the values of these parameters will affect the algorithm s efficiency .
in section [ sect : derivation ] we show that the approximate optimal choices for the proposal parameters are @xmath42 where the terms @xmath43 and @xmath44 are @xmath45 and @xmath46 is the principal branch of the lambert w function defined as @xmath47 for @xmath48 . the acceptance - rejection algorithm to generate a sample from the bessel exponential distribution proceeds as follows : 1 .
find efficient proposal parameters @xmath41 , which will depend on @xmath49 .
2 . draw @xmath30 from a gamma distribution with shape @xmath50 and rate @xmath51 .
3 . draw @xmath52 from a uniform distribution on @xmath53 $ ] .
4 . accept @xmath28 if @xmath54 , else go to 2 .
the detailed procedure is described in algorithm [ alg : kappa ] . when implementing the algorithm , both the bessel functions and the lambert w function
can be computed using software such as the general scientific library @xcite or its r wrapper , the cran package * gsl*. in practice it is often possible to avoid computing these functions , as we show in section [ sect : speedup ] .
@xmath55 [ algline : startsetup ] @xmath56 @xmath57 @xmath58[algline : kappastar ]
@xmath59 @xmath60 @xmath61 @xmath62 @xmath63 [ algline : betadefn ] @xmath64 @xmath65[algline : lambert ] @xmath66 @xmath67 [ algline : endsetup ] [ algline : startloop ] @xmath68 sample from a @xmath69 left - truncated at @xmath70 [ algline : xsampling ] @xmath71 @xmath72 sample from a @xmath73 [ algline : acceptreject ] + [ algline : returnline ]
we will now derive the optimal parameters @xmath41 for the proposal distribution of @xmath74 which follows a gamma distribution with shape @xmath50 and rate @xmath51 .
we do so by maximizing the expected probability of acceptance @xmath75\nonumber \\&=\frac{(\eta \beta)^{\eta \alpha+1}}{\gamma(\eta \alpha+1)}\exp\{-\eta \beta\varepsilon-\eta g(\kappa_0;\alpha,\beta,\varepsilon)\}\int_{0}^\infty
\frac{\exp(-\eta \beta_0\kappa)}{i_0(\kappa)^\eta } { \mathrm{d}}\kappa \label{eq : acceptprob}\end{aligned}\ ] ] over @xmath76 subject to the constraint @xmath40 . in order for the maximum to be finite as @xmath77
we require @xmath78 . by taking logs we see that maximizing with respect to @xmath18 is equivalent to maximizing @xmath79 the constraint @xmath40 implies either @xmath80 or @xmath81 .
the lagrangians for constrained optimization corresponding to these conditions are @xmath82 and @xmath83 , neither of which have interior critical points over @xmath84 because the @xmath85 and @xmath70 derivatives have no common root .
thus the optimal parameters must lie on the boundary of the parameter space .
an examination of the boundaries show that the maximum satisfies @xmath81 and @xmath86 .
intuitively , this says that for any @xmath43 we should pick @xmath70 as small as possible while still having @xmath43 be the maximizer of @xmath87 .
thus our lagrangian is @xmath88 our optimal parameters are either a critical point of @xmath89 or lie on one or more of the boundaries @xmath90 or @xmath91 .
if either @xmath92 or @xmath93 then direct differentiation shows the maximum occurs when @xmath94 and @xmath43 is the unique positive root of @xmath95 .
we shall see this is a limiting case of the critical point solution .
the only other boundary is @xmath91 ; we shall see this is the solution when @xmath24 is close to @xmath96 . to find the critical points of @xmath89 ,
we start by setting the derivatives with respect to @xmath97 and @xmath98 to zero and rearranging yields the optimal parameters as functions of @xmath43 and @xmath99 . this yields @xmath100 where @xmath101 is the digamma function .
we must have @xmath102 so that @xmath85 and @xmath70 are positive .
notice that the limit @xmath103 corresponds to the boundary case @xmath104 discussed above .
the above equations give all optimal parameters in terms of @xmath99 and @xmath43 .
note that since constraint @xmath105 at @xmath106 is satisfied whenever @xmath107 , we are free to choose alternative , sub - optimal values for the other parameters if the true optimal values are too difficult to compute
. we shall explore this in section [ sect : speedup ] when we use an approximation for the lambert w function . finally , we find the optimal @xmath99 as follows
. this value must either lie on the boundary @xmath91 or else satisfy @xmath108 unfortunately plugging into and solving for @xmath99 as a function of @xmath43 alone is analytically intractable .
however , one can check that @xmath109 decreases from positive infinity at @xmath110 to negative infinity as @xmath111 . since all admissible @xmath99 lie in the finite interval @xmath112 we can easily find the optimal @xmath99 through any standard one - dimensional root - finding algorithm . if the root lies to the right of @xmath113 , the optimal value is @xmath91 .
we plug the optimal @xmath99 into to find all of our optimal parameters in terms of @xmath43 .
doing this for each @xmath43 and plugging the resulting parameters @xmath114 into @xmath115 yields a function which we numerically maximize over @xmath43 .
let @xmath116 be the optimal value of @xmath43 and let @xmath117 be the optimal parameters corresponding to @xmath116 .
these are the desired parameters that maximize the expected acceptance probability .
the above numeric maximizations for @xmath99 and @xmath43 may be acceptable when @xmath23 and @xmath24 are known a priori .
however , they are computationally prohibitive in the standard monte carlo case where we wish to generate many samples from the bessel exponential distribution with different values of @xmath23 and @xmath24 for each sample .
thus our next task is to approximate the optimal parameters with easily computable functions of @xmath23 and @xmath24 . for all @xmath23 and @xmath24 , @xmath116
is well approximated by @xmath118 , the positive root of @xmath119 ; indeed @xmath118 is the exact optimum in the boundary case @xmath104 . to approximate @xmath118 we use the bounds (
11 ) , @xmath120 rearranging these bounds shows that @xmath121 where @xmath122 these bounds are relatively tight , we found that the convex combination @xmath123 with @xmath124 provides a good approximation to @xmath118 and hence to @xmath116 .
the parameter @xmath125 is exactly equal to @xmath113 when @xmath24 is close to its lower bound of negative one . for @xmath24 sufficiently large , @xmath125 drops from its upper limit @xmath113 towards its lower limit @xmath126 .
the transition between the two limits is very rapid for @xmath127 .
we achieve good accuracy with the approximation @xmath128 where @xmath129 .
this approximation is very good when @xmath130 is large or when @xmath131 .
a slower , more precise approximation for @xmath125 may lead to parameters which provide better efficiency for small @xmath23 and @xmath132 ; we address this in section [ sect : efficiency ] . given these approximations of @xmath116 and @xmath125 , the parameters @xmath133 and @xmath134
are given by .
the truncated gamma on line [ algline : xsampling ] can be sampled using dagpunar s algorithm @xcite . alternatively , one can use a standard gamma sampling algorithm such as marsaglia
tsang @xcite and reject when @xmath135 .
indeed , the marsaglia tsang algorithm is itself a rejection sampler with a gaussian proposal , and its rejection step can be combined with the rejection step on line [ algline : acceptreject ] for an additional speed - up .
the function @xmath136 on line [ algline : lambert ] can be approximated by ( winitzki , @xcite ) @xmath137 with no noticeable drop in the expected probability of acceptance .
finally , we can implement the simple squeezes @xmath138 to avoid computing the bessel function within the rejection loop on line [ algline : acceptreject ] .
specifically , the loop on lines [ algline : startloop ] to [ algline : returnline ] can be replaced with algorithm [ alg : squeezed ] .
@xmath139 @xmath68 sample from a @xmath140 left - truncated at @xmath70 @xmath71 @xmath141 @xmath72 sample from a @xmath73 @xmath142 * if * @xmath143 * or * @xmath144 * then * + * if * @xmath145 * or * @xmath146 * then * +
we now analyze the efficiency of algorithm [ alg : kappa ] . when using the winitzki approximation , the initial setup ( lines [ algline : startsetup][algline : endsetup ] ) involve arithmetic operations , four square roots , two bessel function evaluations , one logarithm and one exponentiation . in total
this setup requires approximately 70 microseconds on a 2.4ghz intel i5 computer when using the r package * gsl*. implementation in a lower - level language would increase the speed significantly .
each iteration of the rejection loop requires a gamma sample , a uniform sample , between three and five logarithms , and in the worst case a bessel function evaluation .
the squeeze in algorithm [ alg : squeezed ] does a good job of avoiding the bessel computation most of the time , and each iteration of the loop requires approximately @xmath147 microseconds .
most of these iterations are accepted , and the algorithm , implemented in r , yields approximately 80,000 von mises samples per second when @xmath148 and @xmath24 is drawn uniformly over @xmath149 .
when using a compiled language such as c++ , the algorithm yields over one million samples per second . in figure
[ fig : efficiency ] we plot the expected probabilities of acceptance as functions of @xmath23 and @xmath24 .
the figures were generated by numerically integrating the expected probability of acceptance for each @xmath150 and for each of 2000 equally spaced values of @xmath151 .
there is a noticeable dip in efficiency near @xmath152 . recalling that @xmath153
, we see that this region corresponds to diffused @xmath154 , i.e. the true @xmath18 is near zero .
this is precisely the region where our bessel function approximation fails , so this drop is to be expected .
fortunately the drop in efficiency is not severe and our efficiency remains above @xmath155 for all @xmath23 and @xmath24 . from figure
[ fig : efficiency ] we see that our algorithm with the approximate optimal parameters does noticeably worse than the numerically computed true optimal parameters when @xmath132 .
this corresponds to the transition region where the optimal @xmath99 rapidly drops from its upper limit of @xmath113 to its lower limit of @xmath156 .
our approximation of the optimal @xmath99 is inaccurate in this transition region .
a more sophisticated approximation of the optimal @xmath99 would increase the algorithm s efficiency , however , the region @xmath157 is not usually an area of primary interest and we prefer to use the faster approximation .
table[x index=0,y index=1 ] eff1.csv ; table[x index=0,y index=2 ] eff1.csv ; table[x index=0,y index=1 ] eff10.csv ; table[x index=0,y index=2 ] eff10.csv ; table[x index=0,y index=1 ] eff100.csv ; table[x index=0,y index=2 ] eff100.csv ;
we have described a highly efficient algorithm to sample from the bessel exponential distribution .
it is suitable for any application where one wishes generate samples from the posterior distribution for the concentration parameter of the von mises distribution . | motivated by molecular biology , there has been an upsurge of research activities in directional statistics in general and its bayesian aspect in particular .
the central distribution for the circular case is von mises distribution which has two parameters ( mean and concentration ) akin to the univariate normal distribution .
however , there has been a challenge to sample efficiently from the posterior distribution of the concentration parameter .
we describe a novel , highly efficient algorithm to sample from the posterior distribution and fill this long - standing gap . |
the successful detection of co emission in many local ( @xmath15 ) iras galaxies revealed large reservoirs of molecular gas mass that fuel nuclear starbursts and active nuclei ( e.g. tinney et al .
1988 ; sanders , scoville & soifer 1991 ; solomon 1997 ) . in the most ir - luminous galaxies ( @xmath16 ) gas - rich mergers and interactions
are thought to play a crucial role in causing the rapid accumulation of molecular gas in the center and thus initiating the onset of spectacular starbursts .
subsequent high resolution imaging of the co emission in ultra luminous ir galaxies ( hereafter ulirgs ) ( e.g. scoville , yun & bryant 1997 ; downes & solomon 1998 ) confirmed this picture by demonstrating the presence of molecular gas with high mass surface densities ( @xmath17 ) and large velocity widths ( @xmath18 km s@xmath5 ) confined within a few hundred parsec in the nuclear regions . besides being responsible for some of the most spectacular starbursts in the local universe
the merging process is thought to play an important role in galaxy formation at high redshift , especially for spheroidal systems .
the favorite formation model for these systems is that they grow from the hierarchical clustering of gas - rich `` fragments '' where the oldest stars ( @xmath19 gyr ) have already formed ( e.g. baron & white 1987 ; white 1996 ) .
the subsequent intense star formation rapidly consumes the gas mass and finally gives rise to a present - day giant - elliptical galaxy with its red colors , evolved stellar population and large mass , but devoid of substantial amounts of gas and dust .
powerful high - z radio galaxies ( hereafter hzrgs ) may be the progenitors of the massive present - day ellipticals hosting a radio - loud agn . since such galaxies seemed to have already settled into ellipticals ( best , longair , & rttgering 1997 , 1998 and 1999 ) by @xmath20 , it is natural to assume that hzrgs at @xmath21 with their frequently irregular morphologies ( e.g. pentericci et al .
1999 ) is where merger - induced large scale starbursts may have occured . the first detection of co in iras 10214 + 4724 at z=2.286 ( brown & vanden bout 1991 , 1992 ) and the subsequent detection of its mm / sub - mm continuum from dust ( see downes et al . 1992 and references therein ) initiated ongoing efforts to detect co and mm / sub - mm emission from the copious amounts of gas and dust expected in the high - z counterparts of ulirgs and in galaxies undergoing their formative starbursts .
the large negative k - corrections expected for the thermal dust spectrum and the high - j co lines ( e.g. hughes 1996 ; van der werf & israel 1996 ) of high - z galaxies as well as gravitational lensing help to make such objects detectable to the current mm / sub - mm instruments .
lensing amplifies the expected emission but usually complicates the gas and dust mass estimates .
the galaxy iras 10214 + 4724 and many high - z co / sub - mm luminous objects like the qso h1413 + 117 ( barvainis et al . 1994 ) , the recent sub - mm selected galaxies ( frayer et al 1998 ; frayer et al .
1999 ) and the bal quasar apm 08279 + 5255 ( downes et al .
1999 ) are lensed .
recently there have been detections of large amounts of dust in hzrgs ( e.g. hughes 1996 ; rttgering et al .
1998 ; hughes & dunlop 1998 ; best et al . 1998 ) , and a systematic survey is now been conducted to detect sub - mm emission from dust for all @xmath22 radio galaxies ( rttgering et al .
1999 ) in order to understand the large range of fir luminosities of these objects .
the most notable examples are the extremely fir - luminous radio galaxies 8c 1435 + 635 ( ivison et al .
1998 ) and 4c 41.17 ( dunlop et al . 1994 ; chini & krgel 1994 ) whose properties do not appear to be influenced by gravitational lensing .
in both cases the inferred gas masses are large enough ( @xmath23 ) to suggest that the large fir luminosities of these objects ( @xmath24 ) are due to the formative starbursts . however , despite systematic efforts ( evans et al . 1996 ; van ojik 1997 ) no co emission has been detected from these two objects or any other powerful hzrgs . in this paper
we present the detection of dust continuum and co j=43 line emission in two powerful high - z radio galaxies and discuss the latter in the context of the earlier unsuccessful attempts to detect co in this class of objects . throughout this work
we assume @xmath25 and @xmath26 .
we used the sub - mm common user bolometer array ( scuba ) at the 15-m james clerk maxwell telescope ( jcmt ) in the photometry mode to observe 4c 60.07 on 1998 april 15 and 6c 1909 + 722 on 1997 october 4 and 5 as part of an ongoing program to observe all hzrgs at @xmath22 ( rttgering et al .
scuba is a dual camera system cooled to @xmath27 k allowing sensitive observations with two arrays simultaneously .
the short - wavelength array at 450 @xmath0 m contains 91 pixels and the long - wavelength array at 850 @xmath0 m 37 pixels .
the resolution is diffraction - limited with @xmath28 ( 850 @xmath0 m ) , and @xmath29 ( 450 @xmath30 m ) . for a full description of the instrument see holland et al .
( 1998 ) .
we employed the recommended rapid beam switching at a frequency of 8 hz and a beam throw of @xmath31 in azimuth .
the pointing and focus of the telescope were monitored frequently using crl 618 and qso 0836 + 710 , and the typical rms pointing error was @xmath32 .
frequent skydips were used to deduce the atmospheric extinction .
typical opacities at 850 @xmath0 m were @xmath33 for our 1998 run and @xmath34 for our 1997 run .
beam maps of mars , crl 618 and uranus were used to derive the gain @xmath35 .
the data were flatfielded , corrected for extinction and sky noise was removed before they were co - added following standard procedures outlined in stevens et al .
the estimated rms of our measurements is consistent with what is expected from the total integration times and the nefds at the sky conditions of our runs . the systematic uncertainty in the flux density scale at 850 @xmath0 m is of the order of @xmath36 ( holland , private communication ) .
we used the iram plateau de bure interferometer ( pdbi ) between 1998 april 20 and 1998 may 15 in the d configuration to observe co j=43 ( @xmath37 ghz ) in 6c 1909 + 722 at z=3.534 ( van breugel private communication ) in four tracks ranging from 4 - 14 hours each .
the same configuration was also used on 1998 november 8 and 29 to obtain two 8-hr tracks for 4c 60.07 at z=3.788 ( chambers et al .
the correlator setup used for the 3 mm receivers involved 4 x 160 mhz modules covering a total bandwidth of 560 mhz tunned in single sideband .
the band center was positioned at 96.290 ghz for 4c 60.07 and 101.770 ghz for 6c 1909 + 722 .
the 1 mm receivers were used in double sideband mode with the two remaining correlator modules as backends to simultaneously observe the continuum at 240 ghz .
typical ssb system temperatures were @xmath38120 k ( 96.290 ghz ) and @xmath39150 k ( 101.770 ghz ) , and at 240 ghz they were @xmath40 k ( dsb ) .
bandpass calibration was obtained using 3c 454.3 and amplitude and phase calibration were obtained using iap@xmath410444 + 634 and nrao 150 . for the 3 mm receivers the residual phase noise was @xmath42 and for the amplitude @xmath43 .
for the 1 mm receivers these figures are @xmath44 and @xmath45 .
the flux density scale is accurate to within @xmath36 .
after calibration and editing the data were processed with the standard nrao aips software package . at 3 mm no continuum emission was detected in the line - free channels and no continuum subtraction was performed .
the maps were produced using the task mx and in cases of low s / n , clean was not applied .
the rms noise of the maps is consistent with the thermal noise expected for the total observing time and average @xmath46 in the runs .
both radio galaxies are detected in 850 @xmath0 m and co j=43 and are the brightest sub - mm objects found in the ongoing scuba survey of hzrgs at @xmath22 so far , this being the main reason for selecting them for the follow - up sensitive co observations . in 4c 60.07
the co(43 ) emission is clearly resolved ( figure 1 ) . a remarkable characteristic of the co emission in this galaxy is its large velocity range ( @xmath47 ) and its two distinct components with line centers separated by @xmath48 km s@xmath5 .
one component has @xmath49 ( extending beyond the observed band ) , while the other is narrower with @xmath50 .
the feature with the narrow linewidth coincides with the position of the suspected radio core and thus the agn , but a significant part of the broad linewidth component is clearly offset from it .
this aspect of the co j=43 emission is reminiscent of the co j=54 emission detected in the radio - quiet quasar br1202 - 0725 at z=4.69 ( omont et al .
1996 ) where a broad and a narrow linewidth component are also detected . in that object the narrow linewidth component is centered on the agn position while the broad one is offset by @xmath51 .
the velocity - integrated map in figure 2 shows extended co emission with two peaks that are @xmath52 apart ( @xmath53 kpc at @xmath54 ) . towards the same region 1.25 mm emission from dust is also detected , shown in figure 3 . both the co and 1.25 mm continuum maps
are overlaid on a map of the non - thermal radio emission at 6 cm ( carilli et al .
1997 ) .
the radio galaxy 4c 60.07 is a powerful radio galaxy ( @xmath55 w hz@xmath8 ) with a fanaroff - riley ii ( fr ii ) edge - brightened double lobe morphology ( fanaroff & riley 1974 ) , hence there is the possibility of a significant contribution of the non - thermal emission to the observed 1.25 mm and 850 @xmath56 m continuum . extrapolating the non - thermal flux density of @xmath57 mjy of its brightest component with its associated spectral index of @xmath58 ( de breuck , private communication ) yields a negligible contribution ( @xmath59 ) in both 1.25 mm and 850 @xmath0 m bands .
further confirmation of the thermal origin of the 1.25 mm continuum is offered by the high resolution map in figure 3 which shows that most of this emission is not associated with the observed non - thermal radio continuum and the peak 1.25 mm brightness is @xmath51 ( 17.6 kpc ) offset from the radio core where the agn probably resides ( de breuck private communication ) .
deep k@xmath60-band images obtained with keck ( van breugel et al .
1998 ) reveal faint emission towards the weak eastern radio component but none from the regions with the brightest 1.25 mm continuum .
since at @xmath61 the k@xmath60-band is rest - frame b - band , this is consistent with the brightest 1.25 mm emission marking the place of a massive gas / dust reservoir with large amounts of extinction . in order to obtain the highest s / n maps for the two distinct co j=43 features we averaged all the appropriate channels and the two resulting maps are shown in figure 4 overlaid with the 1.25 mm continuum . from these maps
it becomes obvious that the component with the largest dynamical and molecular gas mass is closely associated with the peak of the dust emission as expected . in the case of the radio galaxy 6c 1909 + 722 a map of the averaged co j=43 emission and its spectrum
are shown in figure 5 . there
it can be seen that its linewidth is also rather large .
extrapolation of the non - thermal flux @xmath62 mjy with the observed power law @xmath63 ( de breuck private communication ) yields a non - thermal contribution of @xmath64 at 850 @xmath0 m .
all the observed parameters for the two hzrgs are tabulated in table 1 .
editor : place table 1 here
the two galaxies have extended radio emission and are unlikely candidates for gravitationally lensed objects . in the case of 4c 60.07 , high - resolution radio images show a classical fr ii source .
sensitive 6 cm ( carilli et al .
1997 ) , k - band ( chambers et al .
1996 ) and deep k@xmath65-band images ( van breugel et al . 1998 ) do not reveal any obviously lensed features . for 6c 1909 + 722
, hst images at 7000@xmath66 do not reveal any lensed features either ( pentericci , private communication ) .
henceforth we assume that the two sources are unlensed and organize the discussion as follows : \1 .
estimate the molecular gas mass implied by the co j=43 luminosity assuming global gas excitation conditions similar to the ones found in local starburst galaxies .
briefly explore the influence of various galactic environments on the molecular gas excitation conditions over large scales .
special emphasis is given on the effects of the various environments on the high - j co lines and their detectability at high redshift .
use the mm / sub - mm data to find the dust masses and fir luminosities and estimate the @xmath67 ratios , which we then compare to the values in the local universe .
discuss the evolutionary status of these objects in terms of their gas mass relative to their dynamic mass , and their star formation rates and efficiencies .
we focus particularly on 4c 60.07 where the two distinct molecular gas components strongly suggest a large scale starburst event unlike the usually nuclear starbursts observed in local ulirgs .
the estimate of molecular gas mass from the co j=10 luminosity involves the so - called standard conversion factor @xmath68 ( e.g. young & scoville 1982 ; bloemen 1985 ; dickman et al .
1986 ; young & scoville 1991 ) , which is `` calibrated '' using molecular clouds in the milky way . from the co(43 ) luminosity @xmath69 is then given by @xmath70 \int _ { \delta v } s_{\nu _ { \rm obs } } d v,\ ] ] where @xmath71 is the luminosity distance for @xmath72 , @xmath73 is the rest - frame frequency of the co j=43 transition , @xmath74 is the ( 43)/(10 ) line ratio of the area / velocity - integrated brightness temperatures and @xmath75 is the observed flux density .
substituting astrophysical units yields @xmath76\ m_{\odot } .\ ] ] the dependence of @xmath77 on the ambient conditions of the @xmath78 gas has been extensively explored ( e.g. bryant & scoville 1996 ; sakamoto 1996 ) .
an important recent result is that in the intense starburst environments of ulirgs the warm , diffuse gas phase dominating the @xmath1co emission does not consist of virialized clouds , hence leading to an overestimate of @xmath69 when the standard @xmath79 is used ( solomon et al .
1997 ; downes & solomon 1998 ) .
a more appropriate value for such ism environments and hence their high - z counterparts is @xmath80 ( downes & solomon 1998 ) .
a plausible value for @xmath81 can be found assuming an ism environment similar to that of a local `` average '' starburst .
the average line ratios measured towards such systems are @xmath82 ( e.g. braine & combes 1992 ; aalto et al .
1995 ) , @xmath83 ( devereux et al . 1994 ) , while the @xmath84co/@xmath85co j=10 , 21 ratios are @xmath86 ( aalto et al .
a large velocity gradient ( lvg ) code with the aforementioned ratios as inputs was employed to model the physical conditions of the gas ( e.g. richardson 1985 ) .
the conditions corresponding to the best fit are @xmath87 k , @xmath88 @xmath89 and @xmath90/dv / dr = 3\times 10^{-5}\ ( km \ s^{-1 } \ pc^{-1})^{-1}$ ] , similar to the ones deduced for the high - z starburst iras 10214 + 4724 ( solomon , downes , radford 1992 ) . for these conditions
it is @xmath91 , which for @xmath92 , is slightly enhanced to @xmath93 .
this ratio and @xmath94 yield @xmath95 for 4c 60.07 and @xmath96 for 6c 1909 + 722 .
these are most likely lower limits since even in ulirgs the ratio @xmath74 can be significantly smaller but not much larger . indeed the assumption of an `` average '' starburst excitation environment , while convenient , hides the fact that there is a wide range of physical conditions for the gas in ir - luminous galaxies , and in much of that range co(j+1@xmath97j ) , @xmath98 can be significantly weaker than co(10 ) .
as we shall see , the potential faintness of the high - j co transitions may partly explain why attempts to detect high - z co have been less successful than the ones trying to detect mm / sub - mm continuum from dust .
especially for hzrgs two large systematic searches ( evans et al .
1996 ; van ojik et al .
1997 ) gave null results .
on the other hand observations of the sub - mm continuum fare better with 7 such objects detected at 850 @xmath0 m ( see van der werf 1999 and references therein ) .
a similar situation applies also in other types of high - z objects .
while the uncertainties of the gas - to - dust ratio and a rising @xmath99 ratio with fir luminosity ( van der werf 1999 ) can still adequately account for this , the wide range of gas excitation observed in ir - luminous galaxies can not but have an effect on the luminosity of co lines especially at high j levels . in the attempts to detect co in high - z objects various transitions
are observed depending on the particular redshift and the receiver used . for @xmath100 mostly
co(j+1@xmath13j ) with @xmath101 is observed and in the systematic searches in hzrgs transitions with j+1=4 - 9 were routinely observed ( evans et al .
1996 ; van ojik 1997 ) .
it has been argued that observing higher j transitions can offset the dimming due to the distance in high - z objects in a manner similar to the negative k - corrections of the thermal spectrum from dust ( van der werf & israel 1996 ) . in this picture the warm and dense gas in an `` average '' starburst environment thermalizes the co transitions up to j+1=6 .
however the luminosity of the high - j co transitions may be much smaller for two basic reasons , namely \1 ) the presence of a warm ( @xmath102 k ) but diffuse ( @xmath103 @xmath104 ) and subthermally excited ( for @xmath105 ) gas phase that dominates the @xmath106co emission in ulirgs ( aalto et al .
1995 ; downes & solomon 1998 ) .
the most conspicuous such galaxy is arp 220 , frequently used as the standard ulirg for comparison with high - z fir - luminous sources . in this source a low @xmath107 and high @xmath108 , @xmath109 ratios are observed ( aalto et al .
1995 ; papadopoulos & seaquist 1998 ) , in contrast to the `` average '' starburst ratios mentioned previously . another important characteristic of this gas phase is the moderate optical depths ( @xmath110 ) of the @xmath111co ( 10 ) transition ( aalto et al .
1995 ) .
\2 ) a large reservoir of cold and/or subthermally excited gas extending beyond the warm starbursting nuclear region of an ulirg .
such excitation gradients are observed when large beams that sample the extended emission are used ( papadopoulos & seaquist 1998 ) .
the associated cool dust in the ulirg vv 114 was imaged recently with scuba ( frayer et al .
1999 ) and extremely cold gas ( @xmath112 k ) is inferred over large scales for starbursts like ic 5135 and ngc 7469 ( papadopoulos & seaquist 1998 ) .
this gas phase , if present , can easily dominate the _ global _
co excitation especially in high - z systems where a beam of @xmath113 at @xmath114 corresponds to @xmath115 kpc . here
we must stress that the effectiveness of a cold gas component in suppressing the observed global gas excitation is not altered by the higher cmb temperature at high z. indeed , while the higher cmb temperature enhances the populations of the high j levels of co , it also corresponds to a higher background against which the respective lines must be detected .
moreover the effective temperature of cold gas at high z is not simply the sum of the temperature of this gas phase at z=0 and @xmath116 k as one might naively assume .
this can be demonstrated with some simple arguments .
obviously the excitation temperature @xmath117 of any collisionally excited line at any redshift is bounded as @xmath118 where @xmath119 k is the present epoch cmb temperature and @xmath120 is the gas kinetic temperature .
if we further assume that on large scales in the ism of a galaxy thermodynamic equilibrium exists between gas and dust , i.e. @xmath121 , then for a dust emissivity of @xmath122 the energy balance within a typical giant molecular cloud is described by @xmath123,\ ] ] where @xmath124 is the energy density of the interstellar radiation field ( o , b , a stars ) that heats the grains directly and @xmath125 is the `` mechanical '' heating energy density deposited to the molecular clouds ( e.g. supernovae shocks , turbulent cascade , cloud - cloud collisions ) that first heats the gas and then the dust .
these are the two major heating mechanisms of an average molecular cloud and since , a ) more cooling processes are at play than only dust radiation ( e.g. c@xmath126 , ci , co lines ) , b ) usually @xmath127 ( except in the surfaces of uv - illuminated clouds , hollenbach & tielens 1997 ) , it follows that the dust temperature yielded by the last equation will be an upper limit to the gas temperature .
the aforementioned physical processes responsible for the heating of the ism do not depend on the particular value of the cmb temperature at a given z. thus the last equation , being redshift - invariant , allows us to find the high - z equivalent temperature of any given ism component from its temperature in the local universe ( see also combes , maoli & omont 1999 ) .
it is easy to see that at high z , the low - temperature component of the the local ism has @xmath128 which also means ( equation 3 ) that lines become thermalized at the expense of becoming invisible against the enhanced cmb radiation field . choosing a nominal redshift of @xmath129 as representative of the redshifts of the two hzrgs we find that for a warm ism component of @xmath130 k the @xmath131 is essentially identical .
however for a cold component of @xmath132 k it is @xmath133 k , i.e. just @xmath134 k above the @xmath135 k temperature of the cmb at that redshift . in table 2
we display the expected line ratios at @xmath136 for three sets of conditions , namely a ) an `` average '' starburst , b ) diffuse and warm gas of moderate @xmath137co j=10 optical depth , and c ) cold gas , representative of the phase that may exist in ulirgs beyond the central starburst region .
editor : place table 2 here this table shows that a wide range is expected for co(j+1@xmath138j ) , j@xmath142 line luminosities relative to co j=10 .
several published @xmath139 line ratios ( lisenfeld et al .
1996 ) that sample the _ global _ co emission in ulirgs , and a recent survey of this ratio in many nearby galaxies ( mauersberger et al .
1999 ) reveal a large range of values , namely @xmath140 , thus a similar or larger range is to be expected for the higher j transitions .
moreover since the flux density ratio @xmath141 determines whether a high - j transition is easier to detect than j=10 ( assuming equal sensitivities ) , it can be seen ( table 2 ) that co j=32 is the highest transition for which this ratio is @xmath142 for all the expected conditions . hence many non - detections of co(j+1@xmath138j ) , j@xmath142 in high - z systems can be due to the globally sub - thermal excitation of these lines rather than a true molecular gas mass deficiency . at the same time this makes the upper limits for @xmath143 from such non - detections
much less stringent .
for example if the co j=54 line was observed , and for @xmath144 , the upper limit in @xmath69 can be @xmath145 times larger that what is usually reported for this line ( e.g. van ojik 1998 ) . finally for gravitationally lensed objects with underlying steep gas excitation gradients ,
differential amplification can render line ratios useless in deducing the average gas excitation unless an accurate description of the lensing potential is available .
such gradients are expected in nuclear starbursts where gas warm and dense enough to excite high - j co lines is more centrally confined in the immediate area of the starburst .
recent mapping of co j=43 in m 51 and ngc 6946 ( nieten et al .
1999 ) showed that even in more quiescent galaxies the highly excited gas is strongly concentrated in the center . under such circumstances
differential amplification can enlarge the effective size of the high - j co emitting region with respect to the low - j one and thus alter the observed _ global _ line ratios towards the ones expected for warm and dense gas .
the frustrating aspect of this effect is that it can become progressively severe for lines that are widely separated in j , i.e. exactly the ones whose ratios are most sensitive to the gas excitation conditions under normal circumstances
. clearly more observations of high - j co lines in local ulirgs are needed in order to reveal the range of their _ global _ luminosities and relative brightness distributions and thus allow a better understanding of similar systems at high - z .
there are many uncertainties associated with the estimate of the dust mass from mm / sub - mm measurements ( e.g. gordon 1995 ) , with the uncertainty of the dust temperature being one of the most important ones ( hughes 1996 ) . in the case of 4c 60.07 the detection of the dust continuum both at 1.25 mm and 850 @xmath0
m allows us to place some broad constraints on the dust temperature as long as the emissivity law is @xmath147 .
indeed , for an optically thin isothermal reservoir of dust mass it is @xmath148 where @xmath149 is the dust temperature and @xmath150 is the exponent of the emissivity law . for 4c 60.07
we find @xmath151 .
we adopt an emissivity law of @xmath122 , which was deduced in other high - z objects , e.g. iras 10214 + 4724 ( downes et al .
1992 ) and 8c 1435 + 635 ( ivison et al . 1998 ) over a similar rest - frame spectral range . for this emissivity law
the @xmath152 range of the observed @xmath153 yields a range of @xmath154 k. given the starburst nature of these two objects and their luminous co ( 43 ) line ( the j=4 level is @xmath155 k above the ground state ) we assume a dust temperature of @xmath156 k. then the dust mass can be estimated from @xmath157^{-1},\ ] ] where @xmath158 is the emitted frequency , @xmath159 is the planck function , @xmath160 is the cmb temperature at redshift z , and @xmath161 m@xmath162 kgr@xmath5 is the adopted dust emissivity ( e.g. krgel , steppe , & chini 1990 ) . for the assumed @xmath149
the cmb term can be omitted with @xmath163 error .
here it is worth noting that for cold dust this term can be significant .
for example @xmath164 k , hence a dust component with @xmath165 k _ at that redshift _ is a very cold component , just @xmath166 k above @xmath167 .
an observed wavelength of 1.25 mm corresponds to @xmath168 k at @xmath169 and for this wavelength the cmb term is @xmath170 of the dust term .
assuming similar parameters for the dust emission in 6c 1909 + 722 the 850@xmath0 m flux density yields @xmath171 for both objects .
this gives warm gas - to - dust ratios of @xmath172 ( 4c 60.07 ) and @xmath173 ( 6c 1909 + 722 ) , i.e. within the range of the values found for local iras galaxies ( e.g. young et al .
1986 ; stark et al . 1986 ; young et al . 1989 ) and ulirgs ( sanders et al .
this suggests that in a universe at @xmath36 of its current age these two hzrgs already have heavy - element abundance comparable to the galaxies in the contemporary universe .
it is also worth noting that , while many uncertain factors enter into the estimate of the gas - to - dust ratio , the values chosen for them here are the ones considered plausible from the study of the ism in local ulirgs .
the fir luminosities of the two hzrgs were estimated by assuming the underlying spectrum of an optically thin , isothermal reservoir of dust mass , hence @xmath174 after using equation 6 to substitute @xmath175 this finally yields @xmath176 where @xmath177 and @xmath178 a numerical constant that depends on the emissivity law . in astrophysical units
this becomes @xmath179 for the assumed @xmath180 k the last expression will yield a lower limit to the true @xmath146 since in some ulirgs the optical depth may become significant even at fir frequencies ( e.g. solomon et al . 1997 ) , and in a starburst environment dust can be warmer still .
we find @xmath181 for both hzrgs , a luminosity comparable to 8c 1435 + 635 and 4c 41.17 ( ivison et al .
1998 and references therein ) the other two high - z radio galaxies detected in the sub - mm ( rest - frame fir ) spectral range whose properties do not appear to be influenced by gravitational lensing .
the usual conclusion drawn when such large fir luminosities are deduced for a galaxy is that we are witnessing it during a starburst phase .
however many of the ir - luminous galaxies in the local universe and certainly the two particular hzrgs also harbor an agn , thus part of the fir luminosity maybe due to agn - heated dust .
nevertheless , recent studies of ulirgs ( genzel et al . 1998 ; downes & solomon 1998 ) reveal that most of them are powered by recently formed massive stars . also , even when an agn does make a significant contribution to @xmath182 , it is usually @xmath183 ( genzel et al .
1998 ; downes & solomon 1998 ) . therefore the qualitative arguments behind converting fir luminosities to star formation rates are not likely to change even if an agn is present . for 4c 60.07
the case for the starburst origin of its fir luminosity is made stronger by the fact that most of its 1.25 mm dust emission is not emanating in the vicinity of the agn .
the fir luminosity provides a measure of the current star formation rate ( sfr ) according to @xmath184 where @xmath185 , depending on the assumed imf ( telesco 1988 and references therein ) .
adopting the most conservative value @xmath186 , the fir luminosity of the two hzrgs implies @xmath187 .
such a high rate of star formation can , if sustained , produce @xmath188 of stars in @xmath189 gyr and it is comparable to the sfrs found for 8c 1435 + 635 and 4c 41.17 ( ivison et al .
the efficiency with which such a burst converts molecular gas into stars is given by @xmath190 or equivalently by @xmath191 ( assuming the same @xmath192 for all the galaxies ) . on many occasions the quantity @xmath193 is used instead , but since the @xmath77 conversion factor is @xmath194 times smaller in ulirg systems the latter ratio can be misleading when a single value for @xmath68 is used for galaxies spanning a fir range from moderately ir - luminous galaxies ( @xmath195 ) to ulirgs ( @xmath196 ) .
the star formation efficiencies estimated are @xmath197 ( 4c 60.07 ) and @xmath198 ( 6c 1909 + 722 ) , comparable to the ones found for ulirgs ( solomon et al . 1997 ) once @xmath69 has been estimated using the same @xmath68 factor .
the implied star formation rates and efficiencies of these galaxies point towards spectacular starbursts occurring at high redshift .
in addition the large linewidths observed , which in the case of 4c 60.07 exceed @xmath199 km s@xmath200 , are routinely observed towards ulirgs ( e.g. solomon et al .
1997 ) and are a kinematic signature of the ongoing mergers / interactions that trigger their enormous starbursts ( sanders , scoville , & soifer 1991 ) .
the tantalizing question raised when high - z starbursts are found is to what extent the observed star formation episode is forming the bulk of their eventual stellar mass . the main difficulty in answering this important question stems from the uncertainties involved in deducing the total gas mass from co measurements .
if we assume an atomic - to - molecular gas ratio of @xmath201 found for iras galaxies ( andreani , casoli , & gerin 1995 ) and a large sample of spirals ( casoli et al .
1998 ) , we obtain a total gas content of @xmath202 ( 4c 60.07 ) and @xmath203 ( 6c 1909 + 722 ) . for the `` average '' starburst value @xmath204 this gives gas masses of the order of @xmath205 .
these are an order of magnitude higher than typical gas masses in local ulirgs ( e.g. downes and solomon 1998 ) and constitute @xmath206 of the total stellar mass of a typical elliptical associated with the 3cr radio galaxies at @xmath20 ( best , longair & rttgering 1998 ) . thus it is clear that these enormous starbursts still have a vast reservoir of gas to be eventually turned into stars . the best evidence yet that at least in the case of 4c 60.07 we are witnessing an extraordinary starburst and not a scaled - up version of an ulirg comes from the fact that in this galaxy the co j=43 emission from gas is distributed over scales of @xmath53 kpc .
this is in contrast with the local ulirgs where most of the molecular gas and starburst activity is confined within the central @xmath207 kpc . since starburst activity is usually co - extensive with the molecular gas reservoir , and this will be particularly true for the high excitation co j=43 line , it seems that the starburst in 4c 60.07 occurs on galaxy - wide scales .
equally intriguing is the fact that the co emission consists of two distinct components widely separated in velocity ( figures 1 , 4 ) .
their velocity - integrated co flux densities of @xmath208 jy km s@xmath209 ( wide - linewidth component ) and @xmath210 jy km s@xmath8 ( narrow - linewidth component ) yield masses of @xmath211 and @xmath212 respectively ( equation 2 , @xmath213 ) .
we derived size estimates for these two components by fitting both the image and the visibility plane of the co emission shown in figure 4 with an underlying gaussian brightness distribution .
the narrow - linewidth component appears unresolved with an upper limit of @xmath214 while the wide - linewidth one is marginally resolved with the largest size being @xmath215 . in the absence of adequate spatial / kinematic information allowing a distinction between the possible geometrical arrangements of the co - emitting gas we assume that the largest estimated size @xmath216 corresponds to the diameter of a disk ,
hence its mass is given by @xmath217 where i is the inclination of the disk and @xmath218 ( bryant & scoville 1996 ) .
for the wide - linewidth component the largest estimated size corresponds to @xmath219 kpc , which gives @xmath220 , comparable to the mass of a present - day giant elliptical .
the ratio of the inferred molecular to dynamic gas mass then is @xmath221 . for the narrow - velocity component
the upper limit corresponds to @xmath222 kpc yielding a dynamic mass of @xmath223 .
this is @xmath43 of the mass of a typical giant elliptical and a comparison with its molecular gas mass gives @xmath224 .
thus , geometrical factors aside , this component is significantly richer in molecular gas than the more massive one and it maybe the one where star formation has yet to form the bulk of its eventual stellar mass .
it is easy to contrive a combination of different excitation properties , @xmath77 values , and inclinations so that @xmath225 in one or both molecular gas components is altered significantly .
still the fact that remains unaltered is that in 4c 60.07 , unlike in a typical ulirg , the intense star formation occurs over large scales in two spatially and kinematically distinct molecular gas reservoirs .
this brings in mind the scenario for the formation of a giant elliptical at high redshift where several star - forming clumps merge to eventually form the galaxy . in this picture a gas - rich low mass clump like the one seen in 4c 60.07 is still in the process of merging and vigorous star formation while the higher mass object has already formed most of its stars .
of course the possibility that the wide - linewidth component itself consists of several virialized gas - rich clumps that do not necessarily constitute a bound system can not be discarded in the light of the present data . for 6c 1909 + 722
the co emission is unresolved with a size @xmath226 ( @xmath227 kpc ) , which for the observed linewidth ( table 1 ) corresponds to @xmath228 and @xmath229 .
thus , while there is little doubt that this galaxy is a starburst , it can still be a high - z counterpart of an ulirg where most of its eventual stellar population has already formed and the intense star formation is confined in its central region .
finally , the case of 4c 60.07 makes it clear that when searching for high - z co lines a large velocity coverage is necessary not only because of their often uncertain redshift , but in order to be able to detect various components that may be far apart in velocity .
higher resolution observations are necessary in order to reveal more structure in the molecular gas reservoir of this galaxy and allow better constraints of its dynamical mass
. however the faintness of the co emission observed here may hinder such efforts and a systematic study of such objects may have to await the advent of the next generation mm interferometers with significantly larger collecting areas .
in this paper the detection of mm / sub - mm emission from dust and the first detection of co j=43 in two powerful high - z radio galaxies has been presented .
the analysis of the data leads us to the following conclusions : \1 .
using the most conservative co / h@xmath230 conversion factor ( 1/5 of the galactic value ) the co j=43 emission implies molecular gas masses of the order of @xmath231 in the two high - z powerful radio galaxies .
their sub - mm continuum corresponds to large fir luminosities ( @xmath232 ) which imply that we are witnessing intense starburst phenomena at @xmath233 converting the aforementioned gas mass into stars .
the wide range of the molecular gas excitation expected in ultra luminous ir galaxies over large scales is briefly explored and we conclude that observing objects with similar ism conditions at high z using high - j co transitions may in some cases hinder their detection .
this can partly explain explain why , despite systematic efforts , co detections of unlensed objects are fewer than detections of their mm / sub - mm continuum from dust .
therefore it is still too early to draw any general conclusions about the abundance of molecular gas in high - z systems from the data currently available in the literature .
the estimated molecular gas - to - dust ratios found in these two objects are in the range found for the local iras galaxies , revealing that in both of them most of the heavy elements have already been produced at present - day abundances . \4 . in 4c 60.07
the co emission and presumably the starburst activity implied by its large fir luminosity is distributed over scales of @xmath53 kpc and consists of two distinct components spanning a total velocity range of @xmath7 km s@xmath5 .
thus this galaxy does not seem to be just a scaled - up high - z version of a local ultra luminous ir galaxy where most of the starburst activity and the accompanying bright co / dust emission are confined to a more compact region in the center .
a plausible scenario is that we are witnessing the ongoing formation event of a giant elliptical galaxy , the future host of the residing radio - loud agn .
we acknowledge the iram staff from the plateau de bure and from grenoble for carrying the observations and help provided during the data reduction .
p. papadopoulos would like to thank especially dieter nrnberger and anne dutrey for their patient help during the data reduction in grenoble and jessica arlett for providing us the lvg code .
we thank the referee david sanders as well as simon radford for valuable comments , chris carilli for making available the 6 cm maps of 4c 60.07 and carlos de breuck for help with the radio images and stimulating discussions .
the work of w. v. b. at igpp / llnl was performed under the auspices of the us department of energy under contract w-7405-eng-48 .
p. is supported by the `` surveys with the infrared space observatory '' network set up by the european commission under contract erb fmrx - ct96 - 0086 of its tmr programme .
best , p. , longair , m. , & rttgering , h. j. a. 1999 in `` the most distant radio galaxies '' , page 59 - 69 , proceedings of the colloquium , amsterdam , 15 - 17 october 1997 , edited by h.j.a rttgering , p. n. best , & m. d. lehnert ( royal netherlands academy of arts and sciences ) holland , w. s. , robson , e. i. , gear , w. k. , cunningham , c. r. , lightfoot , j. f. , jenness , t. , ivison , r. j. , stevens , j. a. , ade , p. a. r. , griffin , m. j. , duncan w. d. , murphy j. a. , & naylor , d. a. 1998 ( in press ) , astro - ph/9809122 lisenfeld , u. , hills , r. e. , radford , s. j. e. , & solomon , p. m. 1996 , in `` cold gas at high redshift '' , pg .
55 - 59 , m. n. bremer , p. p.
van der werf , h. j. a. rttgering , & c. l. carilli ( eds ) , kluwer academic publishers .
rttgering , h. j. a. , best , p. , pentericci , l. , de breuck , c. , & van breugel , w. 1998 , the most distant radio galaxies and the formation of massive galaxies . , in s. dodorico , a. fontana , e. giallongo ( eds ) , the young universe in monte porzio , p. 49
scoville , n. z. , yun , m. s. , & bryant , p. m. 1997 in `` co : twenty - five years of millimeter - wave spectroscopy '' , pg 305 - 313 , iau symposium no 170 edited by william b. latter , simon j. e. radford , philip r. jewell , jeffrey g. mangum and john bally .
( kluwer academic publishers ) .
solomon , p. m. 1997 in `` co : twenty - five years of millimeter - wave spectroscopy '' , pg 289 - 298 , iau symposium no 170 edited by william b. latter , simon j. e. radford , philip r. jewell , jeffrey g. mangum and john bally .
( kluwer academic publishers ) .
van der werf , p. p. , & israel , f. p. 1996 in `` science with large millimeter arrays '' ,
pg 51 , proceedings of the eso - iram - nfra - onsala workshop , held at garching , germany december 1995 , peter a. shaver ( ed . )
van der werf , p. in `` the most distant radio galaxies '' , pg 161 - 172 , proceedings of the colloquium , amsterdam , 15 - 17 october 1997 , edited by h.j.a rttgering , p. n. best , & m. d. lehnert ( royal netherlands academy of arts and sciences ) | we report the detection of sub - mm emission from dust at 850 @xmath0 m and of the @xmath1co j=43 line in the two distant powerful radio galaxies 4c 60.07 ( z=3.79 ) and 6c 1909 + 722 ( z=3.53 ) . in the case of 4c 60.07
the dust emission is also detected at 1.25 mm .
the estimated molecular gas masses are large , of the order of @xmath2 .
the large fir luminosities ( @xmath3 ) suggest that we are witnessing two major starburst phenomena , while the observed large velocity widths ( @xmath4 km s@xmath5 ) are characteristic of mergers . in the case of 4c 60.07
the co emission extends over @xmath6 kpc and spans a velocity range of @xmath7 km s@xmath8 .
it consists of two distinct features with fwhm of @xmath9 km s@xmath5 and @xmath10 km s@xmath8 and line centers separated by @xmath11 km s@xmath5 .
the least massive of these components is probably very gas - rich with potentially @xmath12 of its dynamical mass in the form of molecular gas .
the extraordinary morphology of the co emission in this object suggests that it is not just a scaled - up version of a local ultra luminous infrared galaxy , and it may be a formative stage of the elliptical host of the residing radio - loud agn . finally we briefly explore the effects of the wide range of gas excitation conditions expected for starburst environments on the luminosity of high - j co lines .
we conclude that in unlensed objects , co ( j+1@xmath13 j ) , j+1@xmath143 lines can be significantly weak with respect to co j=10 and this can hinder their detection even in the presence of substantial molecular gas masses . |
over the last twenty years , much effort has been invested in searching for a `` third parameter '' to reduce the scatter in the two - parameter tully - fisher relation ( tfr ) between the stellar luminosity and the rotation speed of spiral galaxies ( strauss & willick 1996 , courteau 1997 ; hereafter c97 , and references therein ) . for late - type spiral galaxies ,
no other structural parameter ( such as the size , light concentration , or mean surface brightness of the galaxy ) appears to lead to a significant reduction in scatter . in short
, the stellar luminosity alone seems to determine the rotation speed ( which we will define more precisely below ) of a spiral galaxy .
it is common to model the rotation curves of individual galaxies with multi - component mass models , usually consisting of bulge , disk and dark halo components ( broeils 1992 , hereafter b92 ; van der kruit 1995 , hereafter vdk95 ) .
the mass components representing the bulge and the disk are usually derived from the deprojected light - distribution , assuming a constant mass - to - light ratio , to be determined by the data .
the assumed constancy of @xmath16 across a given galaxy is often justified on the basis of modest radial color gradients ( de jong 1995 ) and the small changes of vertical scale heights as a function of radius in late - type spirals ( van der kruit 1988 , de grijs & peletier 1997 ) . especially for data taken in the redder bands
this approximation should be valid . with only a single value of @xmath16
, analytical galaxy formation models can explain simultaneously the properties of hsb and lsb galaxies , like the slope and zero - point of the tf relation , very well ( dalcanton 1997 , mo , mao , & white 1998 ) .
this provides further support for the constancy of @xmath16 among disk galaxies .
dark matter halos in rotation curve decompositions have been modeled traditionally as isothermal spheres with a homogeneous core .
because these multi - component fits to the rotation curve data are not unique , it appeared sensible to add a constraint , such as the maximum disk " or minimum dark matter " solution ( sancisi & van albada 1985 , carignan & freeman 1985 ) , to obtain a unique solution .
these fits imply the highest possible value for @xmath16 , typically ranging from 3 to 5 in the @xmath17 band ( broeils & courteau 1997 ; hereafter bc97 ) and , in turn , lead to larger core radii for the halo .
`` maximum disk '' fits have been very successful , matching almost all and rotation curve data ( b92 , buchhorn 1992 , bc97 ) , reproducing the small scale features of optical rotation curves well ( freeman 1993 ) , and satisfying constraints on spirals arms for non - interacting galaxies ( athanassoula 1987 ) .
maximum disk models imply that for bright hsb sb - sc spirals , the stellar disk and bulge are the dominant mass components inside two exponential scale lengths .
these galaxies are typically hi - deficient in their inner parts and the mass of the neutral gas is negligible ( broeils 1992 ) .
an operational definition for the maximum disk hypothesis is that the stellar disk provides @xmath18 of the total rotational support of the galaxy at @xmath19 ( sackett 1997 ) . the lower boundary accounts for bars and massive bulges which reduce the contribution of the disk to the overall rotation curve .
the upper boundary of this ratio is kept at 95% in order to prevent hollow cores in the dark halo .
still , the validity of the maximum disk hypothesis has yet to be demonstrated unequivocally in external galaxies because of our inability to constrain @xmath16 , independent of mass models . to illustrate this point ,
we have computed mass decompositions for 10 late - type spirals with extended rotation curves previously modeled with an isothermal halo and maximum disk by broeils ( 1992 , p. 246 ) .
we have repeated broeils modeling , but with a cosmologically motivated halo ( navarro , frenk , & white 1996 ; hereafter nfw ) and two values of @xmath16 corresponding to a maximal disk and lighter disk with @xmath20 .
the data - model residuals are shown in figs . 1 & 2 .
details regarding the fitting procedure will be given elsewhere .
clearly , one can not discriminate in favor of any one mass model based on the quality of these multi - component fits ; the observed rotation curves are reproduced equally well by each fit .
a similar result was also found by bc97 who modeled 290 rotation curves with @xmath17-band surface brightness profiles and found equally good fits using either a pure disk or a pure halo .
this latter argument is however trivial given the limitations to constrain fitting parameters from shallow rotation curves .
note also that the particular shape of the nfw halo does not play a role in constraining @xmath21 ; this is because of the strong covariance that exists between the concentration parameter of the nfw halo and @xmath16 .
this is true of any other halo model with a core parameter .
an example of such fitting is found in navarro ( 1997 , p. 408 ) for ngc 3198 .
a related argument in support of maximum disks is that the luminous matter alone can reproduce the overall shape and small scale features of optical rotation curves ( buchhorn 1992 , freeman 1993 ) .
the small - scale features are almost certainly the result of spiral wave streaming and offer weak support for maximum disks .
this is confirmed by matching the bumpy features of the rotation curve with similar features in major - axis luminosity cuts .
the match to the overall shape of the rotation curve is met just as well by pure halo or pure disks models ( bc97 ) . in principle
, there exists a technique to constrain @xmath16 uniquely , but it is difficult to implement in practice , and it still requires statistical arguments . for pure exponential disks ,
the vertical ( azimuthal ) velocity dispersion and scale height of the stellar disk are linked to its mass density , independent of the halo mass ( van der kruit & searle 1981 , bahcall & casertano 1985 ) .
if the disk scale length is also known , one can directly infer the mass of the disk .
the measurement of velocity dispersions in stellar disks away from the central bulge and disk inhomogeneities is however difficult for large samples .
the rapid surface brightness decline in stellar disks requires that absorption profiles beyond 1 - 2 disk scale lengths be measured with 8 - 10 m class telescopes .
this technique also requires that galaxy properties be matched statistically since disk scale heights and scale lengths can not be measured in face - on and edge - on galaxies simultaneously .
one attempt at this technique has been published so far ( bottema 1993 ) which suggests , based on a sample of 12 bulge - less systems and some assumptions about their stellar populations , that late - type spirals are sub - maximal with @xmath21 equal to @xmath22 at @xmath23 locally , most investigations of the galactic mass density near the sun support the notion of a significantly sub - maximal milky way disk with @xmath24 at @xmath25 ( kuijken 1995 and references therein ) .
sackett ( 1997 ) however argues that new or revised structural data for the milky way would favor a maximal galactic disk .
a crucial element in these models is the precise contribution of the massive central bar ( zhao , spergel , & rich 1995 ) or elongated bulge ( kuijken 1995 ) to the dynamical support in the inner parts of the galaxy , which is quite uncertain at present .
the uncertainty in the disk scale length is also an issue ; a 25% difference above or below the nominal value of 3.5 kpc is sufficient to lend a sub - maximal result or not
. published values for the disk scale length of the milky way currently differ by more than 30% ( sackett 1997 )
. numerical and analytical models of disk formation in a dissipationless dark matter halo also predict , for realistic total fractions of baryonic to dark matter , that spiral disks should be far from maximal ( blumenthal 1986 , barnes 1987 , flores 1993 , moore 1994 , nfw , dalcanton 1997 , mo , mao , & white 1998 , navarro 1998 ) . in this paper
we explore whether the tully - fisher relation , and its seemingly sole dependence on stellar luminosity , is consistent with `` maximal '' disks .
the basic idea of our test is best described by a _
gedankenexperiment _ : consider a pure exponential disk galaxy of a given luminosity , mass - to - light ratio , and disk scale length . in the absence of dark matter or bulge , its rotation curve is given by ( freeman 1970 ) , @xmath26,\ ] ] where , @xmath27 is the gravitational constant , @xmath28 the central surface brightness , @xmath4 the disk exponential scale length , @xmath29 , and @xmath30 and @xmath31 are the modified bessel functions of the first and second kind .
the disk rotation curve will peak at @xmath32 at a value of @xmath2 .
for disks of finite thickness ( say , @xmath33 ) the rotation curve will have a very similar shape but a @xmath34% lower peak ( casertano 1983 ) ; this will slightly affect the shape " term in square brackets in eq .
( 1 ) , but leave the subsequent scalings untouched .
imagine compressing " the same disk to a somewhat smaller scale length .
the more compacted disk will peak at a higher rotation velocity , and dimensional analysis of eq .
1 yields @xmath35 at fixed @xmath36 . thus , if dark matter is negligible inside 2.2@xmath37 , a 20% change in disk scale length at a given disk mass should translate into a 10% change of @xmath2 , hence into a @xmath38% offset in luminosity from the mean tfr .
such an offset should be easily detectable in a statistical sense , given the size and quality of tfr samples currently available .
of course , this effect will be altered by the presence of a dark matter halo and of a bulge , and we will devise a sequence of models that predicts @xmath39 for a wide range of disk - to - halo mass ratios . even though in practice such a test requires to substitute @xmath40 for @xmath36 , our test does not rely on any assumption that @xmath41const ; it merely requires that @xmath16 be independent of surface brightness and that any gradient in @xmath42 for comparable disks be self - similar . in 2 of this paper
, we determine the mean value of @xmath39 for a set of well defined samples of luminous ( @xmath43 ) , non - barred hsb galaxies . in 3 , we compare this observed value to the predictions from a number of disk@xmath44halo models and examine the effects due to color gradients .
this enables us to estimate what fraction of the rotation speed , @xmath14 at @xmath45 arises from the stellar disk .
we discuss our results in 4 and offer a brief conclusion in 5 .
to date , two large optical tf data sets have been published , from which we can determine @xmath2 for many hundreds of non - barred hsb galaxies .
all galaxies are selected from the field , where dark matter halos have likely not overlapped or recently merged .
we first use the collection of 306 sb - sc northern galaxies from courteau and faber ( courteau 1992 ; hereafter referred to as cf ) .
the cf galaxies were selected from the ugc catalog using the following criteria : @xmath46 , blue major axis diameter @xmath47 , inclination between 55 and 75 , no bar , and no apparent interaction .
the velocities @xmath2 were modeled via parametric fitting of the h@xmath48 rotation curves ( c97 ) and the ccd magnitudes and disk scale lenghts were measured in the gunn @xmath17 band ( courteau 1996 ) .
specific details about galaxy selection can be found in courteau ( 1992 , 1996 ) .
we use a sub - sample of the cf catalog in a region of quiet hubble flow ( qhf ) to minimise the effects of streaming and local infall motions on distance estimates ( courteau 1993 ) . for this analysis , the cf qhf sub - sample
is further restricted to rotation curves that extend beyond @xmath49 ( so that @xmath2 is defined ) and to @xmath50 . only the brighter galaxies ( @xmath43 ) are retained where the dynamical contributions from the stellar disk are likely to be largest .
all other cases , expect perhaps for bar - dominated galaxies , presumably yield a lower fraction of visible - to - total matter at @xmath45 .
the @xmath51 color is also computed from the rc3 b - band and cf @xmath17-band magnitudes are corrected for galactic and internal extinction . after all culling ,
124 cf / qhf galaxies are kept for final analysis .
we also use the collection of 1355 southern late - type spirals from mathewson , ford , & buchhorn ( 1992 ; hereafter mat ) .
peculiar motions are not as conspicuous in this sample and no spatial culling is applied . excluding or retaining galaxies in the vicinity of the great attractor does not change the final result . as above for the cf sample , we restrict the total mat sample to @xmath52 and @xmath53 , and require galaxies to have a measured @xmath54 color and i - band scale length .
this yields a final sample of 406 mat galaxies .
the @xmath55-band magnitudes and @xmath54 colors were corrected for internal and galactic extinction as described in willick ( 1997 ) .
the fitting of @xmath2 for the mat galaxies is described in c97 .
the disk scale lengths , from byun ( 1995 ) , were extracted from 2d bulge / disk decompositions of the galaxy images .
five galaxies overlap between the cf and mat samples . with those
, we can verify that our cf scalelengths and those measured by byun agree to within 10% .
the typical uncertainty in published scalelengths for a given galaxy is about 23% ( knapen & van der kruit 1991 ) .
the matching rotational amplitudes @xmath2 also agree very well , within 5% ( c97 ) .
it is of interest to verify how adequately the surface density of disk galaxies is described by a pure exponential profile . using this information
, we can show that our final result does not depend on any assumption about the shape of the luminosity profile of the disk .
we compute the fraction of `` exponential light '' at a given radius as the ratio of the integrated light , based on the disk exponential fit , to the total observed light inside that radius .
we measure two separate ratios at @xmath45 and at `` infinity . ''
the total magnitude measured at infinity includes all the light from the galaxy plus a small extrapolation of the disk profile ( typically @xmath56 ) from its observed edge to infinity .
a disk is deemed `` purely exponential '' if the ratio of `` exponential '' to observed light at a given radius does not deviate by more than 20% from unity . in fig . 3
, we show exponential light ratios computed at @xmath45 and infinity for all 306 cf galaxies .
more than three - quarters of the light profiles deviate from the idealized exponential disk model , some by more than 50% .
galaxies left of the peak of the histogram have typically a larger bulge or can be described by a freeman type i luminosity profile ( freeman 1970 ) , whereas those on the right side correspond to a freeman type ii profile or will show a truncated disk .
based on the criterion above , 85 cf galaxies , or 22% of the total cf sample , qualify as pure exponential disks .
this ratio is closer to 18% for all mat galaxies .
the tf distribution of all cf galaxies is shown in fig . 4 . the two distributions with circles and squares in each panel use absolute magnitudes _ measured _ at @xmath45 and infinity respectively .
the solid line is the qhf tf fit from c97 and the dashed line is offset from that fit by 05 .
this magnitude difference is that expected between the light measured at @xmath45 and at infinity for a pure exponential disk .
the tf slopes are statistically equivalent and the zero - points differ by 062 @xmath57 005 , a little over the exponential disk value .
the bottom panel shows the same distributions for pure exponential disks only ; here , the tf zero - point offset is exactly 05 , as expected by construction .
the tf scatter for the qhf galaxies increases from 034 when total extrapolated magnitudes are used to 040 if the light interior to @xmath45 is used . as expected , there is no dependence of the tf scatter on the choice of magnitudes for the pure exponential galaxies ( @xmath58 in either case ) . in all cases , whether one considers the full cf sample or only the distribution of pure exponential disks , the tf slopes are equivalent within their statistical error ( independent of the choice of magnitude ) , meaning that dynamically , all late - type spirals behave as exponential disks .
( c97 ) . ] .
the data used in this study are : @xmath2 , @xmath4 , @xmath3 or @xmath59 , a @xmath51 or @xmath54 color term , distance of the galaxy in mpc ( see c97 ) , and the fraction of exponential light in a galaxy .
distance - dependent quantities assume @xmath60 as in c97 .
a galaxy s rotation speed , size , and color correlate with its absolute luminosity ; these correlations are known as the tully - fisher relation ( tully & fisher 1977 ) , freeman s law ( freeman 1970 ) , and the color - magnitude relation ( tully 1982 ) , and are shown in fig . 5 for our two samples .
at present , we want to ask how various galaxy properties correlate _ at a given absolute luminosity_. hence we need to fit and remove the trends of @xmath2 , @xmath4 , and color ( @xmath61 or @xmath62 ) with absolute magnitude ( @xmath63 or @xmath64 for each sample ) .
these three relations were fit by the functional form @xmath65 , where @xmath66 stands for @xmath67 , @xmath68 , and @xmath61 , respectively , @xmath63 is the absolute magnitude of the cf galaxies , and the upper bar denotes a median value .
the same notation applies for the mat sample with the appropriate terms replaced . to achieve a robust fit to the data @xmath69 $ ]
, we varied @xmath70 and @xmath71 to minimize the sign of the data@xmath44model deviation ( press 1992 , 15.7 ) .
these fits are shown in table 1 .
standard least - squares fits give essentially the same solutions .
l c c c + [ 2pt ] & & & + [ -12pt ] & cf / qhf galaxies ( n=124 ) & & cf pure exp . disks ( n=28 ) + [ 1pt ] & & & + [ -13pt ] @xmath72 & @xmath73 & & @xmath74 + [ 1.2pt ] @xmath75 & @xmath76 & & @xmath77 + [ 1.2pt ] @xmath78(\mr)$ ] & @xmath79 & & @xmath80 + [ 3pt ] & & & + [ -12pt ] & mat galaxies ( n=406 ) & & mat pure exp . disks ( n=73 ) + [ 1pt ] & & & + [ -13pt ] @xmath81 & @xmath82 & & @xmath83 + [ 1.2pt ] @xmath84 & @xmath85 & & @xmath86 + [ 1.2pt ] @xmath87(\mi)$ ] & @xmath88 & & @xmath89 + [ 3pt ] the fits were applied to the sample of exponential disks only and are shown as solid lines in fig .
open and closed symbols show all the galaxies and the restricted sample of pure exponential disks respectively .
we can now define the residuals for each object @xmath90 as @xmath91 $ ] .
6 shows the residuals for @xmath67 , @xmath68 , and color plotted against each other .
> from inspection of the plots , it is not obvious which range of correlation slopes @xmath92 is statistically acceptable .
we devise a robust non - parametric test by rotating the set of residuals @xmath93 by various angles @xmath94 to get @xmath95 and then applying a spearman rank test ( press 1992 , 14.6 ) for correlation between the quantities @xmath96 .
the acceptable range of @xmath97 can be calculated from the range of angles @xmath94 for which the @xmath98 are _ not _ significantly correlated .
table 2 shows correlation slopes for the full and restricted cf / qhf and mat samples .
the error quoted for each slope gives the 95% confidence level .
plus 1pt l c c c cc + & & & + [ -14pt ] & & & + & & & + [ -13pt ] & all & exp . only & & all & exp . only + & & & + [ -13pt ] @xmath39 & @xmath99&@xmath100&&@xmath101&@xmath102 + [ 1pt ] @xmath103&@xmath104&@xmath105&&@xmath106&@xmath107 + [ 1pt ] @xmath108&@xmath109&@xmath110&&@xmath111 & @xmath112 + the first line of table 2 shows that there is tentative evidence that @xmath113 , indicating that the disk mass does play a role in setting @xmath114 .
however , the case of @xmath9 , expected for the disk - only case , is clearly ruled out .
further , the residual analysis shows that there is a slight ( but not statistically significant ) tendency that redder disks ( at a given luminosity ) are slightly more compact , and that redder disks have higher @xmath114 ( see table 2 ) .
these trends are expected in a variety of evolutionary models for disk galaxies ( firmani & avila - reese 1998 ) . before proceeding with a dynamical analysis ( 3 ) one
should explore the potential impact of stellar population variations among disks of the same luminosity .
qualitatively , one would expect redder disks to have higher mass - to - light ratios , whether their colors are affected by dust , greater age , or higher metallicity . since
, more compact disks are observed to be marginally redder ( table 2 ) , this would imply that the derived @xmath39 is lower than its true " value , and consequently the observed mean value of @xmath115 is an _
upper _ limit to the actual dynamical contribution from the disk stars . to be more quantitative about m / l variations _ among _ different disks
, we have used population models ( vazdekis 1996 , their table 2 ) to determine how the stellar @xmath17-band @xmath116 varies with color . as a full exploration of this question
is beyond the scope of this paper , we follow the common practice of using single age population models to match colors that arise from a more complex star - formation history , but which has resulted in the same luminosity weighted stellar age .
we use population models with a mean age of @xmath117 gyrs and near - solar metallicity to find that @xmath118 \sim 1 $ ] if the color differences are attributable to age differences , @xmath118 \sim 0.4 $ ] if they are attributable to metallicity differences , and @xmath118 \approx r_{b - r}/2.5 \sim 1 $ ] for dust reddening , where @xmath119 is the ratio of total @xmath17-band extinction to the differential extinction @xmath120 .
this last quantity will depend on the details of the dust distribution .
for the benchmark case of a simple absorption screen with a mean galactic extinction curve , @xmath121 ( savage and mathis , 1979 ) . in the context of these models , it appears the reddening due to stellar population age differences , and dust reddening imply the largest @xmath122 differences , with @xmath123 corresponding to a 25% increase in @xmath122 .
another potential bias on the final value of @xmath39 is the wavelength dependence on scalelengths .
the stellar disk mass is dominated by low - mass stars which emit most of their light in the redder bandpasses .
therefore , k - band scalelengths may be most representative of the surface density of the disk population . scale lengths at red band passes are typically @xmath124 shorter than in the blue ( de jong 1995 , peletier & balcells 1997 ) .
the ratio between the @xmath17 and @xmath55 band scalelengths and k - band scalelengths is about half that value but we will test for a more extreme difference of 10% .
that is , we now measure @xmath125 at @xmath126 and use the shorter scalelength as our new fiducial . for both cf and mat galaxies ,
this yields an 8% increase of @xmath127 .
this effect is negligible and does not impact our conclusions . in the following
we develop a dynamical analysis under the assumption that _ within our sample _ the stellar populations of disks at the same luminosity are similar .
if the stellar disk were the only relevant mass component , we would expect @xmath9 , in the absence of color gradients . if , on the other hand , the stars are only test particles , then a difference in @xmath37 simply shifts the physical radius at which we determine @xmath2 . therefore , in this limiting case @xmath39 just reflects @xmath129 , the logarithmic gradient of the halo rotation curve at @xmath0 disk scale lengths .
note that for halos with rising rotation curves ( isothermal with a core , or nfw ) one finds @xmath130 . as a consequence
, one would expect a surface brightness , or disk size , dependence in the @xmath131 relation , both for the case of dominant and of negligible disk mass , yet with _
opposite sign_. consequently , there must be an intermediate case , where the stellar disk is somewhat important and the surface brightness disappears .
the actual situation will be bracketed by these two extremes .
below , we explore how @xmath39 depends on the fraction of the circular velocity at @xmath45 that is attributable to the stellar disk , @xmath21 .
we first examine a simple exponential disk embedded in dark matter halo . for the _ initial _ dark matter halo structure ,
we use a density profile @xmath132^{-2}$ ] found in the collisionless cdm simulations of halo formation by nfw .
these simulations indicate that this initial structure only depends on the total halo mass , which translates into the total stellar mass , if the baryonic fraction is constant among halos of identical mass , and if most cooled baryons have been converted into stars .
if the central accumulation of the baryons occurs on a longer time scale than the local dynamical time , the dissipationless dark matter halo will contract obeying adiabatic invariance ( blumenthal 1986 ) .
calculating this adjustment is easy in spherical geometry ( flores 1993 , nfw ) , but requires replacement of the disk configuration by the spherical density profile that has the same enclosed mass @xmath133 ( see binney & tremaine 1987 , eq . 2 - 170 ) .
it is well known that this procedure leads to a somewhat incorrect rotation curve , but we are only interested in derivatives .
the final disk rotation curve is still computed using eq . 1 , and the final dark matter distribution is given by @xmath134 where @xmath135 is the total mass within radius @xmath136 before disk formation ( nfw ) . using a scdm halo profile from nfw with @xmath137 and assuming a baryon fraction of @xmath138 , we calculated the combined disk@xmath44halo rotation curves for various @xmath16 and for @xmath139 kpc , corresponding to a range of @xmath21 at @xmath45 .
then we changed @xmath4 by 5% retaining the total disk mass and the _ initial _
( uncontracted ) halo profile , and repeated the calculation to estimate @xmath39 for each value of @xmath21 .
our choice of @xmath140 and @xmath4 is representative of the milky way and the @xmath141 galaxies considered in this study
. we will show below that our result is rather insensitive to reasonable variations in our disk and halo parameters .
experiments with @xmath142 showed that the finite change has no significant impact on the estimate of the gradient . in cases
where a central bulge is included ( 3.2 ) , the spheroid remains unchanged when considering the small changes in @xmath4 for the gradient calculation .
an example of such a contraction is shown in fig .
the estimated gradients are shown as the short - dashed line connecting the open circles in fig .
note also that for low mass disks , @xmath143 , the gradient @xmath39 becomes positive , because the halo rotation curve is rising , and an increase in @xmath4 means that @xmath2 is measured at greater metric radii .
these results are insensitive to the adopted disk scale length or rotation curve amplitude provided representative values are used . if we choose @xmath144 kpc ( 4 kpc ) in the calculations , the value of @xmath39 at a given @xmath21 increases ( decreases ) in the mean by only 0.03 ( see fig .
9 ) from our nominal case .
disk scale lengths for hsb galaxies vary typically between 2kpc and 6kpc ( courteau 1996 ) .
similar modest excursions for @xmath39 are found if we use different rotation curve amplitudes .
the value of @xmath140 is a slow function of the optical velocity , or @xmath2 , and is typically smaller than @xmath2 for hsb galaxies ( the opposite is true for lsb galaxies ) .
we have experimented with a range of halo velocities from 160 to 200 , or a variation in @xmath2 of 150 to 300 ( depending on the cosmology ) .
the value of @xmath39 increases in the mean by less than 0.03 per 10 increments in @xmath140 for a given disk mass and mass - to - light ratio .
this is because the shape of the logarithmic nfw rotation curve changes little over such amplitude increments at @xmath2 .
this is also depicted in fig .
these tests were performed with and without a bulge component ( see below ) .
thus our results are only weakly dependent on the exact value of @xmath140 and @xmath37 provided we use reasonable estimates for @xmath141 galaxies .
our nominal halo velocity , @xmath145 corresponds to @xmath146 for an @xmath147 cdm universe ( navarro 1998 ) .
our choice of baryon fraction is consistent with the latest d / h abundances ( burles & tytler 1998 , levshakov 1998 ) which give @xmath148 .
the specific choice of @xmath149 is inconsequential for this analysis ; results are essentially unchanged if we adopt @xmath150 ( = 70 ) . even considering the late - type spirals of our sample
, the bulge component may not be dynamically negligible .
therefore , we repeated the above calculation including in the rotation curve calculation a spheroid with 20% of the disk mass and a scale radius of 500 pc .
such a bulge mass fraction is an upper - limit for these sb@xmath44sc galaxies ( bc97 ) .
we repeated the same analysis as above with this new component in the halo contraction and rotation curve calculation .
the resulting @xmath39 are shown as the solid line connecting the triangles in fig 9 .
it is evident that even a large bulge for these galaxies makes little difference for this calculation at @xmath45 ( see also mo 1997 ) .
finally , we explore how sensitive are these results to the particular functional form of the assumed dark matter profile . to this end
, we repeated our calculations starting from an isothermal dark matter halo with an initial core radius @xmath151 .
the resulting dependence of @xmath39 on @xmath21 is shown by the long - dashed line connecting the open squares in fig .
the result is again quite similar to the previous two cases .
choosing a larger core radius makes little difference for heavy disks ( @xmath152 ) , but leads to rotation curve shapes for very light disks that are linearly rising within @xmath153 , grossly inconsistent with the observations for these luminous , hsb disks ( c97 ) .
the derivative @xmath154 is a function of ( at least ) four variables : the initial values of @xmath140 ( assumed to be directly coupled to the nfw concentration parameter ) and @xmath37 , the disk size , mass and disk mass - to - light ratio . if the initial halo structure and size of the disk were coupled , _
i.e. _ if @xmath140 varied systematically with @xmath37 , it would change the interpretation of @xmath154 . to impact our conclusions , the denser , or more concentrated , halos of a given mass scale ( nfw predict a 25 % scatter in their concentration parameter ) would need to harbor larger than average disks .
the dark matter could aid the lower surface brightness stellar disks to yield the same rotation speed at @xmath2 as that of their higher surface brightness cousins and remove any surface brightness trend in the tf relation , even in the presence of a near - maximal disk .
there seems to be broad agreement that ( at least for systems with small bulges , little merging ) the disk size is regulated by the initial angular momentum ( _ e.g. _ mo , mao and white , 1998 ) . for the purpose of our analysis
, it suffices that there exists a monotonic mapping between total initial angular momentum and final disk size ; this would mean that larger disks ( at a given total baryonic mass ) come from baryon / dark matter systems with higher initial spin parameters @xmath155 ( see also 3.5.2 in mo , mao & white 1998 ) . when considering halos of the same mass scale but with different spin - parameters @xmath155 one might suspect that the halos , too , become more extended ( or less concentrated ) at higher @xmath155 , analogous to the disks . in this case
the surface mass density of both disk and halo would drop at higher @xmath155 , and one would necessarily expect an sb dependence of the tfr .
the halo concentration is found to be independent of the spin parameter @xmath155 ( see also 3.5.2 in mo , mao & white 1998 ) .
hence , at least in the context of hierarchical formation simulations , one would not expect any correlation between the ( initial ) halo structure and size of the disk .
this is implicit in our derivation .
we calculate @xmath39 using halo parameters that depend only on the total mass .
hence our conclusion is somewhat model dependent , but one which follows from current hierarchical scenarios .
note that the initial halo choice matters least for a dominant disk , where the final halo structure in the inner parts is predominantly determined by the ( adiabatic ) contraction process .
we have not investigated the ( random ) scatter in the initial halo concentrations but this would merely contribute to weakening our model - data comparison and the power of our statistical tests and we believe our conclusions to be conservative in this respect .
the mean value of @xmath39 in both cf and mat samples of exponential disks is @xmath156 . by taking the range of @xmath21 for which our models predict @xmath39 in this range ( fig . 8)
, we find a mean value of @xmath157 .
this range can be widened if we adopt limits determined from each sample separately ; in this case , we would have @xmath158 .
therefore , our mean relation is inconsistent with @xmath159 , the mimimum boundary for a `` maximal disk '' ( sackett 1997 ) , at the 95% confidence level .
this statement holds for any reasonable combination of @xmath140 , @xmath37 , the mass of the disk , and @xmath16 .
that is , we have tested for any combination of our model input parameters and can not produce @xmath159 for @xmath160 , the absolute maximum allowed by our global data exceeds 0.75 . ] .
the need to consider the actual distribution of the light in any dynamical measurements ( mass decomposition ) , instead of assuming _ a priori _ an average luminosity profile , can not be overemphasized ( see kalnajs 1983 ) .
we did indeed restrict our analysis to galaxies with a nearly pure exponential light profile
. however , this distinction turns out to matter little for our statistical analysis .
inclusion of all types of galaxy profiles would bias the value @xmath39 towards larger fractions of dark - to - total matter at @xmath45 by no more than 15% of the actual value for pure exponential disks , with @xmath161 and a 95% upper limit of 0.66 ( courteau & rix 1998 ) .
inspection of eqs .
1 and 2 shows that our analysis is insensitive to the assumption of exponential disks : as long as the rotation curve shape factor in eq .
1 , @xmath162 $ ] , does not depend strongly on @xmath4 at a fixed luminosity , eq .
2 holds exactly . for the remainder of this discussion
, we adopt the ratio @xmath163 as our nominal result .
our main argument is based on possible correlations between @xmath2 and @xmath4 which can be ruled out statistically in the presence of the observed scatter .
as long as the sources of the scatter are not systematically linked to the correlation of interest , their nature , cause or origin is inconsequential .
for example , inclination errors and hubble - flow distance errors are not systematically linked to either @xmath2 and @xmath4 and we did not consider them explicitly . color gradients ( dust , stellar populations ) are relevant sources of scatter that might vary systematically with disk scale length . however , as we saw in 2.3 , correction for color differences can only increase the ratio of dark - to - visible matter at @xmath45 .
this is because the mean inferred value of @xmath21 is an upper limit to the contribution of the disk mass to the rotation speed at @xmath45 .
the dynamical interpretation of this gradient is only slightly altered if we consider the dependence of stellar population on disk mass - to - light ratios ( bottema 1997 , hereafter b97 ) .
b97 examined population differences between lsb and hsb galaxies using crude population synthesis arguments , and showed that dust or metallicity gradients in spiral galaxies can account for @xmath124 of @xmath164 our own estimates in 2.3 using the population models of vazdekis 1996 , suggest that the impact of age and dust on @xmath21 can be even larger with an increase of 25% in @xmath122 , or a 50% decrease of @xmath21 .
this would lower @xmath21 to @xmath165 for color offsets @xmath166 from the mean .
this value is in agreement with the data if the maximum observed color trends seen in fig .
8 for @xmath4 and @xmath2 are removed from the distribution of @xmath39 . however , because colors depend strongly on the star formation history of the galaxy , more detailed chemical and spectral evolutionary models based on different galaxy formation scenarios ( merging , extended collapse ) are needed to address the interplay between colors and the dynamical properties of galaxies on firmer grounds ( firmani & avila - reese 1998 , sommerville & primack 1998 )
. arguments in favor of sub - maximal disks have been given by van der kruit ( 1995 ) and a few were highlighted in 1 .
abiding by the definition that a disk is `` maximal '' if @xmath167 ( sackett 1997 ) , we have effectively shown that non - barred late - type spirals are , _ on average _ , sub - maximal at @xmath45 , and that the halo contributes @xmath168 , within that radius .
sub - maximal disks with @xmath169 imply @xmath170 , a factor of two smaller than maximal disk values ( bottema 1993 , bc97 ) .
this corresponds to twice more dark matter in the inner parts of late - type spirals than previously accounted for by maximal disk fits .
evolutionary synthesis models of spiral galaxies suggest mass - to - light ratios in the range 1 - 2 ratios depends significantly on the choice of initial mass function . here , we assume a scalo imf truncated at @xmath171 and @xmath172 in the @xmath17-band , with possible extremes up to 5 for very red systems ( bruzual & charlot 1993 ) .
thus , there is good agreement between these models and our result .
the same is also true with the work of bottema ( 1993 ) who combined measurements of velocity dispersion , scale length , and scale height of exponential disks to infer that old disks contribute at most @xmath173 of the observed maximum rotation at a given luminosity .
another supportive result is that of quillen & sarajedini ( 1998 ) who find that low values of the disk @xmath174 in the range 1 - 2 are required to explain stable disks at @xmath175 . in order to avoid unrealistically thick disks
, one must invoke a substantial amount of dark matter in the inner parts of these distant disks , in agreement with our study of low redshift galaxies .
swing amplification theory ( athanassoula 1987 ) puts a constraint on the density of galactic disks in order to preserve m=2,3 modes ( and inhibit m=1 modes ) .
the low values of @xmath16 observed here are in complete agreement with the observed spiral structure of late - type disks .
while our statistical argument favors sub - maximal disks in the majority of hsb non - barred disk galaxies , it does not completely exclude the possibility that some galactic disks could be maximal .
such disks are more likely to be found in lighter or less centrally concentrated halos or equivalently , systems of very high surface brightness or low angular momentum .
ostriker and peebles ( 1973 ) have shown that a massive halo will suppress bar instabilities in the disk .
thus , it is not surprising that bar - dominated galaxies may exhibit a higher fraction of luminous - to - dark matter at their center , to the point of possibly being maximal ( quillen & frogel 1998 , bottema & gerritsen 1998 , debattista & sellwood 1998 , weiner 1998 ) .
the current analysis together with other recent studies , suggest a continuum of luminous - to - dark matter content in spiral galaxies directly proportional to their mean surface brightness ( or initial angular momentum ) .
lsb and dwarf galaxies appear to be fully dm dominated at all radii ( broeils 1992 , zwaan 1995 , bottema 1997 , ct 1997 , dalcanton 1997 , de blok & mcgaugh 1997 ) , non - barred hsbs are sub - maximal in their inner parts ( bottema 1993 , this work ) , and bar - dominated galaxies which exhibit , on average , higher surface brightnesses than typical tully - fisher galaxies may possibly be maximal . following this argument
, one may infer that our milky way , if it does harbor a strong bar at its center , could indeed be maximal ( sackett 1997 ) .
more explorations of the central regions of our galaxy and the structure of extragalactic disks and bulges are desperately needed .
we have shown that the tf relation , when cast as @xmath3 _ vs. _ @xmath2 , is independent of surface brightness for hsb galaxies . we are lead to interpret this empirical result as evidence for large quantities of non - luminous matter in the inner parts ( @xmath176 ) of late - type galaxies
. we have employed simple models for the contraction of the dark matter halo by adiabatic infall of baryons ( the luminous component " ) . for halos with initially rising rotation curves , this contraction leads to approximately flat , or featureless , rotation curves , _ and _ to a tf relation that is independent of surface - brightness , if @xmath177 in the mean .
this analysis assumes that the initial halo structure is uncorrelated with the size of the resulting disk , as suggested by current cosmological simulations .
combined with the empirical tf evidence , this model invalidates the notion of maximal disks for the majority of hsb late - type spirals . our result is independent of the shape of the luminosity profiles and large variations in our choice of model parameters .
the authors would like to thank roelof bottema , alice quillen and julio navarro for useful comments , the referee jerry sellwood for a critical and insightful review , and adrick broeils for use of his software to produce figs . 1 & 2 . athanassoula , e. , bosma , a. , & papaioannou , s. 1987 , , 179 , 23 bahcall , j. , & casertano , s. 1985 , , 293 , 7 barnes , j. 1987 , in _
nearly normal galaxies _ , ed .
faber , ( new york : springer - verlag ) , 154 binney , j. & tremaine , s. 1987 , _ galactic dynamics _ ( princeton univ . press ) blumenthal , g.r . ,
faber , s.m . , flores , r. , & primack , j.p .
1986 , , 301 , 27 bottema , r. 1993 , , 275 , 16 bottema , r. 1997 , , 328 , 517 bottema , r. & gerritsen , 1998 , mnras , 290 , 585 broeils , a.h .
1992 , ph .
d. thesis , univ . of groningen , the netherlands broeils , a.h . & courteau , s. 1997 , in _ dark and visible matter in galaxies and cosmological implications _ , eds . m. persic & p. salucci ( san francisco : asp , vol .
117 ) , 74 [ bc97 ] bruzual , a.g . & charlot , s. 1993 , , 405 , 538 buchhorn , m. 1992 , ph.d .
thesis , australian national univ .
burles , s. & tytler , d. 1998 , in _ proceedings of the second oak ridge symposium on atomic and nuclear astrophysics _ , ed .
a. mezzacappa .
burstein , d. & rubin , v. 1985 , , 297 , 423 byun , y.i .
1995 , , 448 , 563 carignan , c. & freeman , k.c . 1985 , , 294 , 494 casertano , s. 1983 , , 203 , 735 casertano , s. & van gorkom , j.h .
1991 , , 101 , 1231 ct , s. 1997 , in _ dark and visible matter in galaxies and cosmological implications _ , eds . m. persic & p. salucci ( san francisco : asp , vol .
117 ) , 52 courteau , s. 1992 , ph.d .
thesis , univ .
california , santa cruz courteau , s. 1996 , , 103 , 363 courteau , s. 1997 , , 114 , 2402 [ c97 ] courteau , s. , faber , s.m . , dressler , a. , & willick , j.a .
1993 , , 412 , l51 courteau , s. , & rix , h.w .
1998 , in _ galactic halos _ , ed .
dennis zaritsky ( san francisco : asp , vol .
136 ) , 196 dalcanton , j.j .
, spergel , d.n . , & summers , f.j .
1997 , , 482 , 659 debattista , v.p . &
sellwood , j.a .
1998 , , 493 , l5 de blok , w.j.g . & mcgaugh , s.s . , 1997 , , 290 , 533 de grijs , r. , peletier , r.f . , &
van der kruit , p.c .
1997 , , 327 , 966 de jong , r.s .
1995 , ph.d .
thesis , univ . of groningen , the netherlands flores , r. , primack , j.p . ,
blumenthal , g.r . , &
faber , s.m .
1993 , , 412 , 443 flynn & fuchs 1994 , , 270 , 471 freeman , k.c .
1970 , , 160 , 811 ( see also binney & tremaine 1987 , eq . 2 - 169 )
freeman , k.c .
1993 , in _ physics of nearby galaxies : nature or nurture ? _ , eds .
thuan , c. balkowski , j.thanh van ( paris : frontires ) , 201 kalnajs , a.j .
1983 , in _ internal kinematics of galaxies _ , iau symp .
e. athanassoula ( dordrecht : reidel ) , 87 knapen , j. , & van der kruit , p. c. 1991 , , 248 , 57 kuijken , k. 1995 , in _ stellar populations _ ,
iau 164 ( dordrecht : kluwer ) , 195 kuijken , k. & gilmore , g. 1991 , , 367 , 9 levshakov , s.a . ,
kegel , w.h .
& takahara , f. 1998 , , 499 , 1 martin , p. & roy , j.r .
1994 , , 424 , 599 mathewson , d. s. , ford , v. l. , & buchhorn , m. 1992 , , 81 , 413 mckeith , c.d .
, castles , j. , greve , a. , & downes , d. 1993 , , 272 , 98 mo , h.j . , mao , s. , & white , s.d.m .
1998 , mnras , 295 , 319 moore , b. 1994 , nature , 370 , 629 navarro , j.f .
1998 , in _ galactic halos
dennis zaritsky ( san francisco : asp , vol .
136 ) , 409 navarro , j.f . ,
frenk , c.s .
, & white , s.d.m .
1996 , , 462 , 563 [ nfw ] ostriker , j.p . & peebles , p.j.e . 1973 , ,
186 , 467 peletier , r.f . & balcells , m. 1997 , new astronomy , 1 , 349 prada , f. , beckman , j.e . ,
mckeith , c.d . ,
castles , j. , & greve , a. 1994 , , 423 , 35 press , w.h . , teukolsky , s.a . , vettering , w.t .
, & flannering , b.p .
1992 , _ numerical recipes _ ( cambridge univ . press ) ryder , s.d .
1995 , , 444 , 610 quillen , a.c . &
sarajedini , v.l . , 1998 , aj , 115 , 1412 quillen , a.c .
& frogel , j. 1997 , , 487 , 603 sackett , p.d .
1997 , , 483 , 103 savage , b. and mathis , j. , 1979 , ara&a , 17 , 73 .
sommerville , r. s. , & primack , j.r .
1998 , astro - ph/9802268 strauss , m. a. & willick , j. a. 1995 , phys . rep .
, 261 , 271 tully , r.b . &
fisher , j.r .
1977 , , 54 , 661 tully , r.b . , mould , j.r . , & aaronson , m. 1982 , , 257 , 527 van albada , t.s . & sancisi , r. 1986 , phil .
lond . , 320 , 447 van der kruit , p.c .
1988 , , 192 , 117 van der kruit , p.c .
1995 , in _ stellar populations _ , iau 164 ( dordrecht : kluwer ) , 205 [ vdk95 ] vazdekis , a. , casuso , e. , peletier , r. f. , & beckman , j.e .
1996 , apjs , 106 , 307 weiner , b.j .
1998 , phd thesis , rutgers univ .
willick , j.a . ,
courteau , s. , faber , s.m . , burstein , d. , dekel , a. & strauss , m.a .
1997 , , 109 , 333 zaritsky , d. , kennicutt , r.c . , &
huchra , j.p .
1994 , , 420 , 87 zhao , h .- s . ,
spergel , d.n . , & rich , r.m .
1995 , , 440 , l13 zwaan , m.a .
, van der hulst , j.m . , de blok , w.j.g . , & mcgaugh , s.s .
1995 , mnras , 273 , l35 | we show that for luminous , non - barred , high - surface brightness ( hsb ) spirals the tully - fisher ( tf ) relation residuals can be used to estimate the relative mass contributions of the stellar disk and the dark halo at the peak of the disk rotation , near @xmath0 exponential scale lengths . for maximal disks " ,
a large fraction ( @xmath1 ) of the total rotational support , @xmath2 , at such radii should arise from their stellar mass .
therefore the disk size , or surface - brightness , should be a significant additional parameter in the tf relation . at a given absolute luminosity , @xmath3 , more compact disks ( as measured by the disk scale length @xmath4 ) should have higher rotation speeds , @xmath2 . using a well - defined sample of late - type spirals , deviations , @xmath5 and @xmath6 , from the mean relations , @xmath7 and @xmath8 ,
are not significantly correlated .
the case of @xmath9 expected for a maximal disk is ruled out for the majority of these hsb galaxies .
we model adiabatic infall of varying amounts of luminous matter into dark matter halos to explore the range of possible values for @xmath10 from this , we find that the tf relation residuals require a mean value of @xmath11 , fairly insensitive to the details of the initial dark matter halo and to the presence of a bulge .
this translates to @xmath12 within @xmath13 , or roughly twice more dark matter in the inner parts of late - type spirals than previously accounted for by maximum disk fits .
we show that any stellar population differences between disks of different scalelengths lead to lower values of @xmath14 .
our result is independent of the shape of the luminosity profile and relies only on the assumption of adiabatic contraction and that the dark matter halo rotation rises in the central parts of the galaxy .
sub - maximal disks establish a natural continuity between hsb and low - surface brightness ( lsb ) galaxies which appear to be completely dark matter dominated even in their inner regions .
= cmsy10 scaled 1 22v_2.2 22r_disk # 10.5truecm_#1 _ u # 1#2#3#4 # 1 , _ # 2 _ , * # 3 * , # 4 .
# 1#2#1#2 22 # 1to 0pt#1 # 1#2#1@xmath15#2 |
the notion of subgroup distortion was first formulated by gromov in @xcite . for a group @xmath7 with finite generating set @xmath8 and a subgroup @xmath9 of @xmath7 finitely generated by @xmath10 , the distortion function of @xmath9 in @xmath7
is @xmath11 where @xmath12 represents the word length with respect to the given generating set @xmath10 , and similarly for @xmath13 .
this function measures the difference in the word metrics on @xmath7 and on @xmath9 . as usual
, we only study distortion up to a natural equivalence relation . for non - decreasing functions @xmath14 and @xmath15 on @xmath16
, we say that @xmath17 if there exists an integer @xmath18 such that @xmath19 for all @xmath20 .
we say two functions are equivalent , written @xmath21 , if @xmath17 and @xmath22 .
when considered up to this equivalence , the distortion function becomes independent of the choice of finite generating sets .
if the subgroup @xmath9 is infinite then the growth of the distortion function is at least linear , and therefore one does not extend the equivalence classes using the equivalence defined by the inequality @xmath23 a subgroup @xmath9 of @xmath7 is said to be undistorted if @xmath24 .
if a subgroup @xmath9 is not undistorted , then it is said to be distorted , and its distortion refers to the equivalence class of @xmath25 .
[ poa ] suppose there exists a subsequence of @xmath16 given by @xmath26 where @xmath27 for @xmath28 .
if there exists @xmath29 such that @xmath30 , for all @xmath28 , and @xmath31 , then @xmath32 .
here we study the effects of distortion in various subgroups of the wreath products @xmath33 , for @xmath34 , and more generally , in @xmath0 where @xmath1 is finitely generated abelian .
note that wreath products @xmath35 where @xmath1 is abelian play a very important role in group theory for many reasons .
given any semidirect product @xmath36 with abelian normal subgroup @xmath37 then any two homomorphisms from @xmath38 and @xmath39 ( uniquely ) extend to a homomorphism from @xmath35 to @xmath40 also , if @xmath41 is presented as a factor - group @xmath42 of a @xmath43-generated free group @xmath44 then the maximal extension @xmath45 $ ] of @xmath41 with abelian kernel is canonically embedded in @xmath46 ( see @xcite . ) wreath products of abelian groups give an inexhaustible source of examples and counter - examples in group theory .
for instance , the group @xmath47 is the simplest example of a finitely generated ( though not finitely presented ) group containing a free abelian group of infinite rank . in @xcite the group @xmath47
is studied in connection with diagram groups and in particular with thompson s group . in the same paper
, it is shown that for @xmath48 , where the group @xmath49 appears @xmath50 times , there is a subgroup @xmath51 having distortion function @xmath52 .
in contrast to the study of these iterated wreath products , here we obtain polynomial distortion of arbitrary degree in the group @xmath47 itself . in @xcite the distortion of @xmath47 in baumslag s metabelian group
is shown to be at least exponential , and an undistorted embedding of @xmath47 in thompson s group is constructed . in this note , rather than embedding the group @xmath47 into larger groups , or studying multiple wreath products , we will study distorted and undistorted subgroups in the wreath products @xmath0 with @xmath1 finitely generated abelian .
the main results are as follows .
[ x ] let @xmath1 be a finitely generated abelian group . 1
. for any finitely generated infinite subgroup @xmath53 there exists @xmath54 such that the distortion of @xmath9 in @xmath0 is @xmath55 2 .
if @xmath1 is finite , then @xmath56 ; that is , all subgroups are undistorted .
if @xmath1 is infinite , then for every @xmath54 , there is a @xmath3-generated subnormal subgroup @xmath9 of @xmath0 having distortion function @xmath55 the following will be explained in subsection [ fs ] . [ ot ] for every @xmath54 , there is a @xmath3-generated subgroup @xmath9 of the free @xmath57-generated metabelian group @xmath58 having distortion function @xmath59 [ xx ] if we let the standard generating set for @xmath47 be @xmath60 , then the subgroup @xmath61,b ] , \cdots , b ] \rangle$ ] , where the commutator is @xmath62-fold , is @xmath63 subnormal , isomorphic to the whole group @xmath47 , with distortion @xmath64 . in particular the normal subgroup @xmath65 \rangle$ ]
has quadratic distortion . corollary [ xx ] is proved at the end of this paper .
because the subgroup @xmath66 , b \rangle$ ] of @xmath47 is normal , it follows by induction that the distorted subgroup @xmath67 is subnormal .
[ mmm ] there are distorted embeddings from the group @xmath47 into itself as a normal subgroup .
for example , the map defined on generators by @xmath68 $ ] extends to an embedding , and the image is a quadratically distorted subgroup by corollary [ xx ] . by lemma [ t2 ]
, @xmath47 is the smallest example of a metabelian group embeddable to itself as a normal subgroup with distortion .
[ sc ] there is a distorted embedding of @xmath47 into thompson s group @xmath69 . under the embedding of remark [ mmm ]
, @xmath47 embeds into itself as a distorted subgroup .
it is proved in @xcite that @xmath47 embeds to @xmath69 .
therefore , corollary [ sc ] is true .
it is interesting to contrast theorem [ x ] part @xmath70 with the following , which will be discussed in section [ ppq ] . throughout this paper ,
we use the convention that @xmath71 represents the finite group @xmath72 .
[ pqp ] the group @xmath73 for @xmath74 , has a finitely generated subgroup @xmath9 with distortion at least @xmath75 .
some of the techniques to be introduced in this paper include some computations with polynomials .
we will use the theory of modules over principal ideal domains in section [ mt ] to reduce the problem of subgroup distortion in @xmath33 to the consideration of certain @xmath3-generated subgroups in @xmath47 .
every such subgroup is associated with a polynomial , and therefore we need to define and compute the distortion of arbitrary polynomial , as in theorem [ tgs ] .
all of these techniques are used in conjunction with theorem [ t6 ] , which provides a formula for computing the word length in arbitrary wreath product and makes computing subgroup distortion more tangible in the examples we consider .
here we provide some examples of distortion as well as some basic facts to be used later on . \1 .
consider the three - dimensional heisenberg group @xmath76 , [ a , c]=[b , c ] = 1 \rangle.$ ] it has cyclic subgroup @xmath77 with quadratic distortion , which follows from the equation @xmath78 $ ] .
the baumslag - solitar group @xmath79 has cyclic subgroup @xmath80 with at least exponential distortion , because @xmath81 however , there are no similar mechanisms distorting subgroups in @xmath47
. therefore , a natural conjecture would be that free metabelian groups or the group @xmath47 do not contain distorted subgroups .
this conjecture was brought to the attention of the authors by denis osin .
the result of theorem [ x ] shows that the conjecture is not true .
the following facts are well - known and easily verified .
when we discuss distortion functions , it is assumed that the groups under consideration are finitely generated .
[ wkf ] \1 .
if @xmath82 and @xmath83 < \infty$ ] then @xmath84 .
\2 . if @xmath85 then @xmath86 \3 . if @xmath85 then @xmath87 \4 . if @xmath9 is a retract of @xmath7 then @xmath84
\5 . if @xmath7 is a finitely generated abelian group , and @xmath82 , then @xmath84 .
we consider the wreath products @xmath35 of finitely generated groups @xmath88 and @xmath89 .
we introduce the notation that @xmath35 is the semidirect product @xmath90 , where @xmath91 is the direct product @xmath92 of isomorphic copies @xmath93 of the group @xmath94 we view elements of @xmath91 as functions from @xmath41 to @xmath1 with finite support , where for any @xmath95 , the support of @xmath14 is @xmath96 .
the ( left ) action @xmath97 of @xmath41 on @xmath91 by automorphisms is given by the following formula : for any @xmath98 and @xmath99 we have that @xmath100 .
any element of the group @xmath35 may be written uniquely as @xmath101 where @xmath102 .
the formula for multiplication in the group @xmath35 is given as follows .
for @xmath103 we have that @xmath104 in particular , @xmath41 acts by conjugation on @xmath91 in the wreath product : @xmath105 therefore the wreath product is generated by the subgroups @xmath41 and @xmath106 where non - trivial functions from @xmath107 have support @xmath108 in what follows , the subgroup @xmath107 is identified with @xmath109 and so @xmath110 and @xmath111 is a finite set of generators in @xmath112 in particular , @xmath47 is generated by @xmath113 and @xmath114 where @xmath113 generates the left ( passive ) infinite cyclic group and @xmath114 generates the right ( active ) one . here we observe that a finitely generated abelian subgroup of @xmath115 with finitely generated abelian @xmath1 and @xmath41 is undistorted .
it should be remarked that the authors are aware that the proof of the fact that abelian subgroups of @xmath33 are undistorted is available in @xcite . in that paper
it is shown that @xmath33 is a subgroup of the thompson group @xmath69 , and that every finitely generated abelian subgroup of @xmath69 is undistorted .
however , our observation is elementary and so we include it .
[ abel ] let @xmath1 and @xmath41 be finitely generated abelian groups .
then every finitely generated abelian subgroup @xmath9 of @xmath116 is undistorted .
it follows from the classification of finitely generated abelian groups @xmath7 that every subgroup @xmath10 is a retract of a subgroup of finite index in @xmath7 , and so we are done if @xmath9 is a subgroup of @xmath1 or @xmath41 , or if @xmath117 , by lemma [ wkf ] .
therefore we assume that @xmath118 since @xmath9 is abelian , this implies that the the factor - group @xmath119 is finite .
then it suffices to prove the lemma for @xmath120 since @xmath121\le \infty.$ ] because @xmath122 is finitely generated , it is contained in a finite product of conjugate copies of @xmath1 .
that is to say , @xmath123 for a wreath product @xmath124 where @xmath125 has finite index in @xmath41 .
we are now reduced to our earlier argument , thus completing the proof .
in fact , under the assumptions of lemma [ abel ] , @xmath9 is a retract of a subgroup having finite index in @xmath112 we now return to one of the motivating ideas of this paper , and complete the explanation of remark [ mmm ] .
[ t2 ] the group @xmath47 is the smallest metabelian group which embeds to itself as a normal distorted subgroup in the following sense . for any metabelian group @xmath7 ,
if there is an embedding @xmath126 such that @xmath127 and @xmath128 is a distorted subgroup in @xmath7 , then there exists some subgroup @xmath9 of @xmath7 for which @xmath129 by lemma [ wkf ] , we have that the group @xmath130 is infinite , else @xmath128 would be undistorted .
being a finitely generated solvable group , @xmath130 must have a subnormal factor isomorphic to @xmath49 . because @xmath131 , one may repeat this argument to obtain a subnormal series in @xmath7 with arbitrarily many infinite cyclic factors . therefore
, the derived subgroup @xmath132 has infinite ( rational ) rank .
since the group @xmath133 is finitely presented , the action of @xmath41 by conjugation makes @xmath132 a finitely generated left @xmath41 module .
hence , @xmath134 for some finitely generated @xmath135 . because it is a finitely generated abelian group ,
@xmath136 is a product of cyclic groups .
therefore for some @xmath137 we have a subgroup @xmath138 of finite rank in @xmath132 but @xmath139 has infinite rank .
then @xmath1 has an element @xmath113 such that the @xmath140-submodule generated by @xmath113 has infinite rank , and so it is a free @xmath140-module .
it follows that @xmath113 and @xmath114 , where @xmath141 , generate a subgroup of the form @xmath47 . in @xcite
, magnus shows that if @xmath142 is an absolutely free group of rank @xmath43 with normal subgroup @xmath143 , then the group @xmath45 $ ] embeds into @xmath144 .
this wreath product is a semidirect product @xmath145 where the action of @xmath7 by conjugation turns @xmath91 into a free left @xmath146$]-module with @xmath43 generators . for more information in an easy to read exposition , refer to @xcite .
the monomorphism @xmath147 \rightarrow { { \mathbb z}}^k { \textrm { wr } } g$ ] is called the magnus embedding .
we let @xmath148 denote the @xmath43-generated derived length @xmath149 free solvable group .
[ y ] if @xmath150 , then the group @xmath148 contains a subgroup isomorphic to @xmath47 .
it is well known ( and follows from the magnus embedding ) that any nontrivial @xmath151 and @xmath152 generate @xmath47 .
it should be noted that by results of @xcite , the group @xmath153 can not be embedded into any free metabelian or free solvable groups .
subgroup distortion has connections with the membership problem .
it was observed in @xcite and proved in @xcite that for a finitely generated subgroup @xmath9 of a finitely generated group @xmath7 with solvable word problem , the membership problem is solvable in @xmath9 if and only if the distortion function @xmath154 is bounded by a recursive function . by theorem 2 of @xcite , the membership problem for free solvable groups of length greater than two is undecidable .
therefore , because of the connections between subgroup distortion and the membership problem just mentioned , we restrict our primary attention to the case of free metabelian groups .
it is worthwhile to note that the membership problem for free metabelian groups is solvable ( see @xcite ) .
lemma [ y ] motivates us to study distortion in @xmath47 in order to better understand distortion in free metabelian groups .
distortion in free metabelian groups is similar to distortion in wreath products of free abelian groups , by lemma [ y ] and the magnus embedding . in particular , if @xmath155 then @xmath156 thus by lemma [ wkf ] , given @xmath157 we have @xmath158 this explains corollary [ ot ] .
on the other hand , given @xmath159 then we have @xmath160 based on this discussion , we ask the following .
an answer would be helpful in order to more fully understand subgroup distortion in free metabelian groups .
what effects of subgroup distortion are possible in @xmath161 for @xmath162 ?
here we aim to further understand how the length of an element of a wreath product @xmath163 depends on the canonical form of this element .
let us start with @xmath164 where @xmath165 because the subgroup @xmath91 of @xmath166 is abelian , we also use additive notation to represent elements of @xmath91 .
[ yyy ] in the case of @xmath167 , we use module language to write any element as @xmath168 is a laurent polynomial in @xmath169 , and the sums are finite .
similarly , if @xmath1 is a finitely generated abelian group , then by the definition of @xmath170 arbitrary element in @xmath0 is of the form @xmath171 where @xmath172 are laurent polynomials and @xmath173-s generate @xmath1 .
this form is unique if @xmath174 is a free abelian with basis @xmath175 .
we will use the notation that @xmath176 equals the conjugate @xmath177 for @xmath178 and @xmath179 the normal form described in remark [ yyy ] for elements of @xmath180 is necessary to obtain a general formula for computing the word length . [ yy ] arbitrary element of @xmath0 may be written in a normal form , following @xcite , as @xmath181 where @xmath182 , and @xmath183 are elements in @xmath184 the following formula for the word length in @xmath0 is given in @xcite .
[ t5 ] given an element in @xmath0 having normal form as in remark [ yy ] , its length is given by the formula @xmath185 where @xmath186 is the length in the group @xmath94 the formula from lemma [ t5 ] becomes more intelligible if one extends it to wreath products @xmath35 of arbitrary finitely generated groups .
we want to obtain such a generalization in this section since we consider non - cyclic active groups in section [ ppq ] .
we fix the notation that , with respect to the symmetric generating set @xmath187 , the cayley graph @xmath188 is defined as follows .
the set of vertices is all elements of @xmath7 . for any @xmath189 , @xmath15 and @xmath190
are joined by an edge pointing from @xmath15 to @xmath190 whose label is @xmath191 .
any @xmath192 can be expressed as follows : @xmath193 where @xmath194 and for @xmath195 we have @xmath196 .
the expression @xmath197 is unique , up to a rearrangement of the ( commuting ) factors @xmath198 . for any @xmath199 with canonical form as in equation @xmath197 we consider the set @xmath200 of paths in the cayley graph @xmath188 which start at @xmath201 , go through every vertex @xmath202 and end at @xmath15 .
we introduce the notation that @xmath203 @xmath204 we also define the norm of any such representative @xmath205 of @xmath91 by @xmath206 we have the following formula for word length , which generalizes that given for the case where @xmath207 in the paper @xcite .
( * caution * : the _ right_-action definition of wreath product would be incompatible with the standard definition of cayley graph in the proof of theorem [ t6 ] . ) [ t6 ] for any element @xmath199 , we have that @xmath208 where @xmath209 is the canonical form of equation @xmath197 .
we will use the following pseudo - canonical ( non - unique ) form in the proof .
this is just the expression of equation @xmath197 but without the assumption that all @xmath210 are distinct or @xmath211-s are non - trivial . for any element @xmath192 which is expressed in pseudo - canonical form
we may define a quantity depending on the given factorization by @xmath212 first we show that for @xmath213 in canonical form @xmath214 it holds that @xmath215 . by the choice of generating set @xmath216 of @xmath35 , we have that any element @xmath192 may be written as @xmath217 where @xmath218 and @xmath219 can be trivial , but all other factors are non - trivial
. we may choose the expression so that @xmath220 observe that we may use the expression from equation @xmath221 to write @xmath222 where @xmath223 and @xmath224 , for @xmath225 .
then we have by definition that for the pseudo - canonical form @xmath226 , @xmath227 it is possible that in the form of equation @xmath226 , some @xmath228 for @xmath229 . when taking @xmath213 to the canonical form @xmath230 of equation @xmath197 , we claim that @xmath231 obtaining the canonical form requires a finite number of steps of the following nature .
we take an expression such as @xmath232 and replace it with @xmath233 the assertion of equation @xmath234 follows because @xmath235 equation @xmath236 is true because @xmath237 which implies that @xmath238 @xmath239 finally , we have that @xmath240 because the right hand side is the length of a particular path in @xmath200 : the path which travels from @xmath201 to @xmath241 to @xmath242 to @xmath243 to @xmath15 .
it follows that the length of this path is at least as large as the length of @xmath244 .
thus for a canonical form @xmath209 we see by equations @xmath245 , @xmath234 and @xmath236 that @xmath246 to obtain the reverse inequality , take @xmath247 in @xmath35 in canonical form . by the definition
, @xmath244 will be a path that starts at @xmath201 , goes in some order directly through all of @xmath202 , and ends at @xmath248 we may rephrase this to say that for some @xmath249 there is a path @xmath250 in @xmath200 such that @xmath251 .
in other words , @xmath252 moreover , in the wreath product we have that @xmath253 @xmath254 this implies that @xmath255 @xmath256
we begin with the following result , the proof of which exploits the formula of theorem [ t6 ] .
[ t3 ] the group @xmath5 contains distorted subgroups . this is interesting in contrast to the case of @xmath258 which has no effects of subgroup distortion .
the essence in the difference comes from the fact that the cayley graph of @xmath49 is one - dimensional , and that of @xmath259 is asymptotically two - dimensional , which gives us more room to create distortion using theorem [ t6 ] .
we will use the following notation in the case of @xmath260 : @xmath113 generates the passive group of order @xmath3 while @xmath114 and @xmath261 generate the active group @xmath262 the canonical form of equation @xmath197 will be denoted by @xmath263 for @xmath264 distinct elements of @xmath259 and @xmath265 .
we may do this because any nontrivial element of @xmath266 is just equal to the generator @xmath113 .
[ ss ] let @xmath82 be generated by a nontrivial element @xmath267 as well as the generators @xmath268 of @xmath259
. then @xmath269 . we know that @xmath270 is a free module over @xmath271 $ ]
therefore , we may think of @xmath91 as being the laurent polynomial ring in two variables , say , @xmath169 for @xmath114 and @xmath272 for @xmath261 .
we can use the module language to express any element as @xmath273 , where for @xmath274 we have that @xmath275 .
this corresponds to the canonical form @xmath276 where @xmath277 for @xmath278 .
we now have all the required facts to prove proposition [ t3 ] . of proposition [ t3 ]
: let @xmath279 and @xmath280 where @xmath281=(1+x)a$ ] .
by lemma [ ss ] we have that @xmath269 .
let @xmath282 the element @xmath283 is in canonical form , when written in the additive group notation as @xmath284 by theorem [ t6 ] , we have that its length in @xmath9 is at least @xmath285 since the support of it has cardinality @xmath286 , and the length of arbitrary loop going through @xmath286 different vertices is at least @xmath286 .
+ now we compute the length of @xmath287 in @xmath7 .
we have that @xmath288a.\ ] ] theorem [ t6 ] shows that @xmath289 .
this is because the shortest path in @xmath290 starting at @xmath201 , passing through @xmath291 and @xmath292 and ending at @xmath201 is given by traversing the perimeter of the rectangle , and so gives the length of @xmath293
. therefore the subgroup @xmath9 is at least quadratically distorted .
the subgroup @xmath9 is not normal in @xmath7 because the element @xmath294 is not in @xmath9 .
the proof of proposition [ t3 ] can be generalized as follows .
consider the group @xmath295 for @xmath296 and @xmath162 .
then the subgroup @xmath297 where @xmath298,b_2], ... b_{k-1}]$ ] has distortion at least @xmath75 .
this is a restatement of proposition [ pqp ] . by ( the analogue of ) lemma [ ss ] we have that @xmath269 and
so we can compute lengths using theorem [ t6 ] .
consider the element @xmath299 in @xmath9 .
then it has length in @xmath9 at least equal to @xmath300 because the path in @xmath301 arising from theorem [ t6 ] would need to pass through at least @xmath2 vertices : @xmath302 for @xmath303 . in the group @xmath7 , @xmath304 this has linear length , which follows because the vertices of the support are placed along the edges of a @xmath43-dimensional parallelotope , such that the length of any edge of the parallelotope is at most @xmath149 .
although the notion of equivalence has only been defined for functions from @xmath16 to @xmath16 , we would like to define a notion of equivalence for functions on a finitely generated group .
we say that two functions @xmath305 are equivalent if there exists @xmath18 such that for any @xmath306 we have @xmath307 if there is a function @xmath308 such that @xmath309 , then for any subgroup @xmath9 of @xmath7 , @xmath310 .
we need to establish a looser way of estimating lengths in @xmath47 , than the formula introduced in lemma [ t5 ] .
recall that this group has standard generators @xmath311 ( passive ) and @xmath114 ( active ) . here
we call _ exemplary _ any sugroup @xmath312 where @xmath313 we have @xmath314 where @xmath315 and @xmath316 this follows without loss of generality by conjugating of @xmath205 by a power of @xmath114
. thus we associate a polynomial @xmath317 $ ] with any exemplary subgroup @xmath318 [ xxx ] the mapping @xmath319 extends to a monomorphism of the wreath product @xmath47 onto the exemplary subgroup this follows because in this case @xmath91 is a free module with one generator @xmath113 over the domain @xmath320,$ ] @xmath314 and the mapping @xmath321 ( @xmath322 ) is an injective module homomorphism . then for any element @xmath323 , we may write @xmath324 where @xmath325 is a laurent polynomial
. denote by @xmath326 the sum @xmath327 for this element , consider the norms @xmath328 letting @xmath329 @xmath330 we define @xmath331 consider the function @xmath332 the following lemma shows that we may simplify computations of word length in exemplary subgroups .
[ r2 ] let @xmath333 be an exemplary subgroup .
then we have that @xmath334 recall that by lemma [ t5 ] , we have the following formulas .
for @xmath323 with the notation established above , we have that : @xmath335 and @xmath336 the following inequality follows from the definitions : @xmath337 similarly , we have that @xmath338 we have that @xmath339 thus , @xmath340 @xmath341 the first inequality follows from equation ( [ bb ] ) , the second from equation ( [ bbb ] ) . on the other hand , we have that @xmath342 @xmath343 the first inequality follows from equation ( [ bbb ] ) , the second from equation ( [ c ] ) , and the third from equation ( [ cc ] ) . thus @xmath344 and @xmath345 are equivalent .
in order to understand distortion in exemplary subgroups of @xmath47 , we will introduce the notion of distortion of a polynomial .
[ pd ] let @xmath346 be a subring of a field with a real valuation , and consider the polynomial ring @xmath347 $ ] .
we will define the norm function @xmath348 \rightarrow \mathbb{r}^+$ ] which takes any @xmath349 $ ] to @xmath350 for any @xmath351 $ ] and @xmath29 , we define the distortion of the polynomial @xmath352 from @xmath16 to @xmath16 by @xmath353 taking into account the inequality @xmath354 , one can easily find some explicit upper boundes @xmath355 for the modules of the coefficient at @xmath356 of @xmath357 in formula , starting with the lowest coefficients .
therefore the supremum in equation is finite .
furthermore , if @xmath358 or @xmath359 then the supremum is taken over a compact set of polynomials of bounded degree with bounded coefficients , and since @xmath10 is a continuous function , one may replace @xmath360 by @xmath361 in definition [ pd ] .
note that the distortion does not depend on the constant @xmath261 , up to equivalence , and so we will consider @xmath362 .
the following fact makes concrete our motivation for studying distortion of polynomials .
[ hpu ] let @xmath9 be an exemplary subgroup @xmath363 and @xmath364 for @xmath365 $ ] .
then @xmath366 by lemma [ r2 ] , we have that @xmath367 let @xmath368 be so that @xmath369 .
there exists @xmath370 so that @xmath371 and @xmath372 where @xmath373 is a regular polynomial .
it is easy to check that @xmath374 and @xmath375 now observe that @xmath376 and @xmath377 .
therefore , @xmath378 on the other hand , let us choose any polynomials @xmath379 such that @xmath380 @xmath381 and @xmath382 then by lemma [ t5 ] , @xmath383 while @xmath384 it follows that @xmath385 and the lemma is proved .
given any polynomial @xmath386 , d_0,d_t \ne 0 $ ] with complex , real or integer coefficients , we are able to compute the equivalence class of its distortion function .
[ hnc ] the distortion @xmath362 of @xmath352 with respect to the ring of polynomials over @xmath387 or @xmath359 is bounded from below by @xmath388 , up to equivalence , where @xmath261 is a complex root of @xmath352 of multiplicity @xmath389 and modulus one .
let @xmath261 be a complex root of @xmath352 of multiplicity @xmath389 and modulus @xmath201 .
that is , @xmath390 over @xmath359 .
let @xmath391 then the product @xmath392 satisfies @xmath393 , because @xmath394 . on the other hand
, because @xmath395 , we have that @xmath396 this implies that if @xmath397 ; i.e. @xmath398 , then @xmath399 , where the distortion is considered over @xmath359 , @xmath400 or over @xmath401 . we will show that a similar computation holds over @xmath400 and over @xmath49 even in the case when @xmath402 .
let @xmath403 be the complex conjugate of @xmath261 . by hypothesis that @xmath404 we know that @xmath405 then @xmath406 is a root of @xmath407 of multiplicity @xmath389 as well , and @xmath408 where @xmath409 has real coefficients .
consider the product @xmath410 where @xmath411 a simple calculation shows that each of the coefficients of this product is a sum of the form @xmath412 this is a geometric progression with common ratio @xmath413 .
therefore , the modulus of every such coefficient is at most @xmath414 and so @xmath415 is @xmath416 .
this computation implies that the products @xmath417 have the sum of the modules of their coefficients which are @xmath416 .
the polynomial @xmath418 has real coefficients .
there is a polynomial @xmath419 with integer coefficients such that each coefficient of @xmath420 has modulus at most one .
thus @xmath421 is also @xmath416 .
we will show that the sums of modules of coefficients of @xmath419 grow at least as @xmath388 on a subsequence from remark [ poa ] .
it suffices to obtain the same property for @xmath422 since @xmath395 , we have that the sum of the modules of the coefficients of @xmath423 is at least @xmath424 we will show that there exists a subsequence @xmath425 so that on this sequence , @xmath426 we have that @xmath427 because @xmath406 .
therefore @xmath428 and similarly , @xmath429 one of these two numbers must be at least one half because @xmath430 .
thus either @xmath149 or @xmath431 can be included in the sequence @xmath425 , and all required properties are shown .
in order to obtain upper bounds on distortion of polynomials we require some facts from linear algebra .
fix an integer @xmath432 and let @xmath433 be arbitrary .
[ ob ] let @xmath434 be @xmath435 column vectors .
suppose that the sum of the modules of all coordinates of @xmath436 is bounded by some constant @xmath261 , and that the modulus of each coordinate of @xmath437 and @xmath438 is also bounded by @xmath261 .
suppose further we have the relationship @xmath439 where @xmath1 is a @xmath440 matrix , in jordan normal form , having only one jordan block .
then the modulus of each coordinate of arbitrary @xmath441 is bounded by @xmath442 where @xmath50 depends on @xmath1 only . in the case where the eigenvalue of @xmath1 does not have modulus one ,
the modulus of each coordinate of arbitrary @xmath441 is bounded by @xmath443 , where @xmath50 depends on @xmath1 only .
all matrix entries are assumed to be complex .
let @xmath444 be the eigenvalue of @xmath1 , so that @xmath445 we will consider cases . * first suppose that @xmath446 . from formula
we derive : @xmath447 the following formula for @xmath448 is well - known because @xmath1 is assumed to be a jordan block ; it may also be checked easily using induction .
we have that @xmath449 with the understanding that if @xmath450 , any terms of the form @xmath451 where @xmath452 are @xmath453 .
arbitrary nonzero element of the matrix @xmath448 is of the form @xmath451 for some @xmath454 .
let @xmath455 denote the @xmath456 entry of @xmath448 .
then @xmath455 is either zero or of the form @xmath451 for some @xmath457 depending on @xmath458 and @xmath191 .
then @xmath459 which is a constant depending on @xmath1 and not on @xmath137 , because the series on the right is convergent when @xmath460 .
let @xmath461 let @xmath462 be the @xmath440 matrix whose @xmath456 entry is @xmath463 , and the column @xmath464 be obtained by placing in the @xmath465 row the sum of the modules of the entries of the @xmath465 row of all @xmath466 and @xmath437 .
then every entry of @xmath464 is bounded by @xmath467 .
observe that the modulus of every entry in the right side of is bounded by an entry of @xmath468 , which is in turn bounded by @xmath469 , which does not include any power of @xmath57 at all .
* let @xmath470 . because @xmath471 is an eigenvalue of @xmath472 , there exists a decomposition @xmath473 where @xmath474 letting @xmath475 and @xmath476 we see by equation that @xmath477 for @xmath478 .
observe that the sum of modules of coordinates of @xmath479 is less than or equal to @xmath480 , where @xmath458 depends on @xmath10 ( and hence on @xmath1 ) only .
similarly , the sum of all modules of all coordinates of @xmath481 is bounded above by @xmath480 .
this case now follows just as the previous one to obtain constant upper bounds on the modules of the entries in @xmath482 . finally , the modulus of any coordinate of @xmath483 is bounded by @xmath484 times the modulus of a coordinate of @xmath479 .
* let @xmath485 . in this case , we have that @xmath486 @xmath487 it follows from equation @xmath488 that every entry of @xmath489 is bounded above by @xmath490 .
[ yx ] let @xmath434 be @xmath435 column vectors .
suppose that the sum of the modules of all coordinates of @xmath436 is bounded by some constant @xmath261 , and that the modulus of each coordinate of @xmath437 and @xmath438 is also bounded by @xmath261 .
suppose further we have the relationship @xmath491where @xmath1 is a @xmath440 matrix .
then the modulus of each coordinate of arbitrary @xmath441 is bounded by @xmath492 where @xmath50 depends on @xmath1 only , and @xmath493 is the maximal size of any jordan block of the jordan form of @xmath1 having eigenvalue with modulus one .
there exists a jordan decomposition , @xmath494 .
let @xmath495 and let @xmath496 .
then for @xmath476 and @xmath475 we have that @xmath497 by hypothesis , the sum of the modules of all coordinates of @xmath481 is bounded by @xmath498 and the coordinates of @xmath499 and @xmath500 are bounded by @xmath501 as well .
as we will explain , our problem can be reduced to the similar problem for @xmath502 in ( [ s ] ) .
suppose that the modules of coordinates of every @xmath502 are bounded by @xmath503 where @xmath50 depends on @xmath1 only .
then , letting @xmath504 and @xmath505 we have by definition of @xmath502 that arbitrary element of @xmath489 has modulus bounded above by @xmath506 where @xmath507 only depends on @xmath508 , as required .
lemma [ ob ] says that if @xmath508 has only one jordan block , then the bound is constant if the eigenvalue does not have modulus one .
otherwise , we have in this case that @xmath509 and the claim is true . if there is more than one jordan block present in @xmath508 , the problem is decomposed into at most @xmath43 subproblems , each with only one jordan block of size smaller than @xmath43 . again
, we are done by lemma [ ob ] .
we will use lemma [ yx ] to prove the following fact , which requires establishing some notation prior to being introduced .
let @xmath510 where @xmath511 .
let the @xmath512 matrix @xmath513 have @xmath465 column , for @xmath514 , given by @xmath515^{t}$ ] , where @xmath516 first appears as the @xmath465 entry in this @xmath465 column .
given the matrix @xmath513 , we may also construct the matrix @xmath517 where @xmath518 , for @xmath519 .
let @xmath389 be the maximal size of a jordan block of the jordan form of @xmath1 having eigenvalue with modulus one .
[ r9 ] suppose that @xmath520^t$ ] is a solution to the system of equations @xmath521 , where @xmath522^t$ ] .
then it is possible to bound the modules of all coordinates @xmath523 of the vector @xmath524 such that @xmath525 where @xmath526 for @xmath527 and the constant @xmath261 depends upon @xmath528 only . prior to proving lemma [ r9 ]
we prove an easier special case .
[ a ] it is possible to bound the coordinates @xmath529 of the vector @xmath524 from lemma [ r9 ] from above by @xmath530 where @xmath526 and @xmath531 . by cramer
s rule , we have the explicit formula that @xmath532 where @xmath533 is the @xmath440 upper left submatrix of @xmath513 corresponding to the first @xmath43 equations , and @xmath534 is obtained by replacing column @xmath137 in @xmath533 with @xmath535^t$ ] . because @xmath536 , it suffices to show that the desired bounds exist for @xmath537 that is , we must show that there exists a constant @xmath538 depending on @xmath539 only such that @xmath540 for @xmath541 . by expanding along the @xmath542 column in @xmath534 , we find that @xmath543 where for each @xmath541 , @xmath544 is a function of @xmath539 only obtained as the determinant of a submatrix containing none of @xmath545 .
the proof is complete by the triangle inequality .
note that the @xmath546 for @xmath547 are similarly bounded by @xmath548 for the same @xmath114 and some @xmath549 as in lemma [ a ] .
it is clear according to lemma [ a ] that we may assume that @xmath550 for the same @xmath551 for all @xmath552 .
we proceed with the proof of lemma [ r9 ] .
it suffices to obtain upper bounds for @xmath553 when @xmath554 . for such indices , we have that @xmath555 in other words , @xmath556 where @xmath557 and @xmath518 .
let @xmath558^{t}$ ] and let @xmath559^t$ ] .
then for the matrix @xmath1 of equation we have the recursive relationship @xmath560 for @xmath561 .
observe that the matrix @xmath1 depends on @xmath528 only , and that the sum of modules of the entries in all @xmath562 are bounded by @xmath563 .
we see by lemma [ a ] that lemma [ yx ] applies to our situation .
therefore , the modules of coordinates of arbitrary @xmath564 , @xmath565 are bounded by @xmath566 , where @xmath50 depends only on @xmath528 , @xmath567 .
[ cz ] let @xmath568 , where @xmath569 .
then the distortion of @xmath352 is at most @xmath388 where @xmath389 is the maximal size of a jordan block of the jordan form of @xmath570 of equation with eigenvalue having modulus one .
consider any @xmath571 as in definition [ pd ] .
then consider @xmath572 .
the coefficients @xmath573 are given by the matrix equation @xmath574 where @xmath575^t , y=[y_{s } , \dots , y_{s+p+t}]^t$ ] and @xmath576 is an @xmath577 matrix .
by lemma [ r9 ] we have that for each @xmath578 that @xmath579 where @xmath580 therefore , @xmath581 the following theorem shows that the upper and lower bounds are the same , and so we can compute exactly the distortion of a polynomial .
[ tgs ] let @xmath568 be a polynomial in @xmath582 $ ] .
then the distortion of @xmath352 is equivalent to a polynomial .
further , the degree of this polynomial is exactly one plus the maximal multiplicity of a ( complex ) root of @xmath407 having modulus one . on the one hand
, lemma [ hnc ] shows that the distortion is bounded from below by the polynomial of degree one plus the maximal multiplicity @xmath389 of a root of @xmath407 having modulus one . on the other hand ,
the characteristic polynomial @xmath583 of the matrix @xmath1 in lemma [ cz ] equals @xmath584 .
and so the real polynomials @xmath583 and @xmath407 have the same roots with modulus @xmath201 ( and with the same multiplicities )
. since the size of a jordan block does not exceed the multiplicity of the root of the characteristic polynomial , we have @xmath585 by lemma [ cz ] .
the theorem is proved .
theorem [ tgs ] will be used here for polynomials with integer coefficients , but it is valid ( with the same proof ) for polynomials with complex or real coefficients .
theorem [ tgs ] and lemma [ cz ] imply the following .
[ exempl ] the distortion of any exemplary subgroup @xmath9 of @xmath47 is equivalent to a polynomial .
the degree of this polynomial is exactly one plus the maximal multiplicity of a ( complex ) root having modulus one of the polynomial @xmath407 associated with @xmath9 .
for every @xmath586 the wreath product @xmath47 has subgroups @xmath587 isomorphic to @xmath588 and so we are forced to study distortion in the groups @xmath33 even we are interested in @xmath47 only .
let @xmath589 be canonical generators of @xmath590 if a subgroup @xmath9 of @xmath591 is generated by @xmath592 where every @xmath593 belongs to the normal closure @xmath594 of @xmath173 ( @xmath594 = the submodule @xmath320a_i$ ] of @xmath91 ) then we say that @xmath9 is a _ tame _
subgroup of @xmath590 if @xmath595 then the subgroup @xmath596 is an exemplary subgroup of the wreath product @xmath597 [ lcc ] for the tame subgroup @xmath9 , we have that @xmath598 observe that @xmath599 is an undistorted embedding , due to that fact that @xmath596 is a retract of @xmath9 ( and similarly for @xmath600 ) .
therefore , by lemma [ wkf ] we have that @xmath601 for every @xmath137 , and therefore @xmath602 to prove the other inequality , we consider an element @xmath603 with @xmath604 then there is a unique decomposition @xmath605 where @xmath606 and for @xmath607 we have @xmath608 since @xmath9 is tame .
then we have @xmath609 since @xmath610 is a retract of @xmath40 therefore the required inequality will follow from the inequality @xmath611 this inequality is true indeed by theorem [ t6 ] because @xmath612 since @xmath613 and @xmath614 since @xmath9 is a tame subgroup of @xmath40 [ tame ] every tame subgroup of @xmath615 has a polynomial distortion .
the statement follows from corollary [ exempl ] and lemma [ lcc ] .
to get rid of the word ` tame ' in the formulation of lemma [ tame ] , we will need few remarks about modules .
the following is well known ( see also @xcite ) .
[ r1 ] the ring @xmath616 $ ] is a principal ideal domain if @xmath69 is a field .
[ ccc ] suppose that @xmath617 is a submodule of a free module @xmath618 of rank @xmath43 over a ( commutative ) principal ideal ring @xmath619 then @xmath620 is a free module of rank @xmath621 , and modules @xmath618 and @xmath617 have bases @xmath622 and @xmath623 respectively such that for some @xmath624 @xmath625 at first we apply this statement to the following special case of theorem [ x ] part @xmath70 .
[ ta ] if @xmath626 is a prime , then any finitely generated subgroup @xmath9 of @xmath627 containing the generator @xmath114 is undistorted . by lemma [ wkf ]
, it suffices to show that @xmath9 has finite index in a retract @xmath628 of @xmath40 since @xmath626 is a prime , @xmath629 is a field .
this implies by lemma [ r1 ] , that the ring @xmath630 $ ] is a principal ideal ring .
let @xmath631 then @xmath632 is a free @xmath346-module by lemma [ ccc ] , and we have that @xmath632 and @xmath91 have bases @xmath633 and @xmath634 respectively , for @xmath635 such that @xmath636 for some polynomials @xmath637 .
thus we can choose the generators for @xmath7 and @xmath9 to be @xmath638 and @xmath639 , respectively , and @xmath9 is a subgroup of the retract @xmath628 of @xmath7 , where @xmath628 is isomorphic to @xmath640 and is generated by @xmath641 now @xmath632 is a submodule of the @xmath642$]-module @xmath643 generated by @xmath644 and the factor - module @xmath645 is a direct sum of cyclic modules @xmath646 hence @xmath645 is finite since it is easy to see that each @xmath647 has finite order at most @xmath648 since the subgroup @xmath9 contains @xmath649 the index of @xmath9 in @xmath628 is also finite .
we return to our discussion of module theory .
let @xmath650 be generated by @xmath114 , as well as any elements @xmath651 .
let @xmath632 be the normal closure of @xmath652 in @xmath653 i.e. , the @xmath320$]-submodule of @xmath91 generated by @xmath654 let @xmath655 and @xmath656 . observe @xmath617 and @xmath618 are free modules over @xmath657}$ ] of respective ranks @xmath43 and @xmath621 .
[ cb ] it follows from lemma [ ccc ] that there exist @xmath658 with @xmath659 where @xmath660}$ ] . moreover , the modules generated by @xmath661 and @xmath662 are free .
[ ca ] there is a bijective correspondence between the set of finitely generated @xmath663}$ ] submodules @xmath143 of @xmath663}^k$ ] and the set of subgroups @xmath664 of @xmath33 such that the finite set of generators of @xmath628 is of the form @xmath665 , @xmath666 .
[ bfz ] let @xmath667 and @xmath668 be generated as submodules over @xmath663}$ ] by the elements from remark [ cb ] : @xmath669 and @xmath634 respectively .
let @xmath122 and @xmath670 be subgroups of @xmath671 generated by @xmath672 and @xmath673 respectively .
it follows by remark [ cb ] that that @xmath674 and @xmath675 .
[ bc ] observe that under the correspondence of remark [ ca ] each generator @xmath676 of the group @xmath122 is in the normal closure of only one generator @xmath544 of @xmath670 , i.e. , @xmath122 is a tame subgroup of @xmath677 [ ba ] there exists @xmath678 so that @xmath679 and @xmath680 by remark [ ca ] we have that @xmath632 is a finitely generated @xmath663}$ ] module with generators @xmath652 . for each @xmath593 , we have that the element @xmath681 . therefore , by lemma [ ccc ] , there are @xmath682}$ ] so that @xmath683 first observe that @xmath684 because @xmath685 .
next , there exists @xmath686 so that @xmath687 where @xmath688}$ ]
. let @xmath689 then for any @xmath690 , we have that @xmath691 where @xmath692},$ ] and therefore , @xmath693 as required .
a similar argument works for obtaining @xmath694 .
[ ab ] let @xmath695 and let @xmath696 be the normal closure of elements @xmath666 .
suppose that there exists @xmath697 and a finitely generated subgroup @xmath698 so that @xmath699 then @xmath700 we will use the notation that @xmath701 observe that the mapping @xmath702 is an injective homomorphism which restricts to an isomorphism @xmath703 .
an easy computation which uses lemma [ t5 ] and the definition of @xmath704 shows that for any @xmath705 , we have that latexmath:[\[\label{e1 } computed in @xmath7 with respect to the usual generating set @xmath707 . observe that under the map @xmath704 we have that @xmath708 where the lengths in @xmath709 are computed with respect to the images under @xmath704 of a fixed generating set of @xmath710 . by their definitions
, we have the embeddings @xmath711 by equation ( [ ac ] ) there exists @xmath712 depending only on the chosen generating sets of @xmath710 and @xmath713 so that @xmath714 it also follows by equation ( [ ac ] ) that there exists a constant @xmath715 depending only on the chosen generating sets of @xmath709 and @xmath713 so that @xmath716 first we show that @xmath717 let @xmath718 be such that @xmath719 and @xmath720
. then there exists @xmath721 such that @xmath722 .
therefore , it follows that @xmath723 the first and second equalities follow by definition , the third by equation ( [ e2 ] ) , and the inequality is true because by equation ( [ e1 ] ) we have that @xmath724 we claim that @xmath725 let @xmath705 be such that @xmath726 .
then @xmath727 which follows from equations ( [ e1 ] ) , ( [ e3 ] ) and by definition . on the other hand
, we will show that @xmath728 let @xmath729 be such that @xmath730 .
then @xmath731 which follows from equations ( [ e1 ] ) , ( [ xa ] ) and by definition .
therefore , we have that @xmath732 we say that @xmath9 is a _ subgroup with @xmath114 _ in a wreath product @xmath733 if @xmath734 where @xmath735 . [ addm ]
let @xmath9 be a subgroup with @xmath114 in @xmath591 .
then the distortion of @xmath9 in @xmath33 is equivalent to the distortion of a tame subgroup @xmath122 of a wreath product @xmath736 @xmath737 this follows from the results of section [ mt ] .
recall that the tame subgroup @xmath122 of the group @xmath670 was defined in lemma [ bfz ] , and these groups were associated to the given @xmath82 .
it follows from lemmas [ ba ] and [ ab ] that @xmath738 and therefore @xmath739 [ withb ] the distortion of every subgroup with @xmath114 in @xmath33 is polynomial .
this follows from corollary [ tame ] and lemma [ addm ] .
in this section , we will reduce distortion in subgroups of @xmath0 where @xmath1 is finitely generated abelian to that in subgroups of @xmath33 only .
[ grga ] let @xmath1 be a finitely generated abelian group and consider @xmath740 assume that @xmath43 is the torsion - free rank of @xmath741 if @xmath9 is a subgroup with @xmath114 in @xmath7 then the distortion of @xmath9 in @xmath7 is equivalent to that of a subgroup with @xmath114 in @xmath33 .
there exists a series of subgroups @xmath742 for @xmath743 where @xmath744 has prime order for @xmath745 .
we induct on @xmath746 . if @xmath747 , then @xmath748 and the claim holds .
now let @xmath749 .
observe that @xmath107 is a finitely generated abelian group with a series @xmath750 of length @xmath63 .
therefore , by induction , any subgroup with @xmath114 in @xmath751 has distortion equivalent to that of a subgroup with @xmath114 in @xmath33 , for some @xmath43 . by lemma [ ta ] ,
all subgroups with @xmath114 of @xmath752 are undistorted .
denote the natural homomorphism by @xmath753 let @xmath754 observe that @xmath755 the product is semidirect because @xmath756 is a normal subgroup which meets @xmath757 trivially , and it is isomorphic to the wreath product by definition : the action of @xmath114 on the module @xmath758 is the same .
let @xmath759 $ ] .
observe that @xmath346 is a noetherian ring .
this follows from basic algebra because @xmath49 is a commutative noetherian ring .
therefore , @xmath91 is a finitely generated module over the noetherian ring @xmath346 , hence is noetherian itself .
thus , the @xmath346-submodule @xmath760 is finitely generated .
let @xmath761 generate @xmath760 as a @xmath346-module .
let @xmath762 be a set of generators of @xmath9 modulo @xmath756 ; that is , the canonical images of these elements generate the subgroup @xmath763 of @xmath670 .
then the set @xmath764 generates @xmath9 .
furthermore , the collection @xmath765 generates the subgroup @xmath766 of @xmath767 .
let @xmath323 have @xmath768 .
then the image @xmath769 in @xmath670 belongs to @xmath122 , because @xmath323 , and has length @xmath770 it follows by lemma [ ta ] that @xmath122 is undistorted in @xmath677 therefore , there exists a linear function @xmath771 ( which does not depend on the choice of @xmath15 ) such that @xmath772 that is to say , there exists a product @xmath200 of at most @xmath773 of the chosen generators @xmath762 of @xmath122 such that @xmath774 in @xmath122 .
taking preimages , we obtain that @xmath775 . because @xmath9 is a subgroup of @xmath7 , there exists a constant @xmath261 depending only on the choice of finite generating set of @xmath9 such that for any @xmath776 we have that @xmath777 ) that @xmath778 this follows because @xmath775 by construction , and @xmath323 by choice .
further , @xmath779 because it is a product of some of the generators of @xmath9 . since @xmath780
we see that @xmath781 . using the fact that @xmath7 and @xmath767 are wreath products together with the length formula in lemma [ t5 ]
, we have that for any @xmath782 @xmath783 by induction , the finitely generated subgroup @xmath784 of @xmath767 has distortion function @xmath785 equivalent to that of a subgroup @xmath786 with @xmath114 in @xmath33 for some @xmath43 .
that is , @xmath787 .
in particular , for any @xmath788 , @xmath789 the first inequality follows from equation ( [ ax ] ) , the second from equation ( [ cx ] ) , and the last from equation ( [ xc ] ) . because @xmath790 there is a constant @xmath43 such that for any @xmath791 combining all previous estimates , we compute that @xmath792 thus , at this point we have shown that @xmath793 since @xmath14 is linear . on the other hand , @xmath794 by lemma [ ab ] . by lemma [ wkf ]
we have that @xmath795 and so @xmath796 .
[ awrz ] for any finitely generated abelian group @xmath1 , the distortion of every subgroup @xmath9 with @xmath114 in @xmath0 is polynomial .
@xmath9 is undistorted if @xmath1 is finite .
this follows from lemma [ grga ] and corollary [ withb ] .
[ r4 ] let @xmath7 be a group having normal subgroup @xmath91 and cyclic @xmath797 .
then any finitely generated subgroup @xmath9 of @xmath7 may be generated by elements of the form @xmath798 where @xmath666 .
[ r5 ] let @xmath9 be a finitely generated subgroup of @xmath0 not contained in @xmath91 , where @xmath1 is finitely generated abelian .
then the distortion of @xmath9 in @xmath0 is equivalent to the distortion of a subgroup @xmath812 with @xmath114 in @xmath813 where @xmath814 is also finitely generated abelian . by lemma [ r4 ] the generators of @xmath9
may be chosen to have the form @xmath815 where @xmath666 .
therefore , for this value of @xmath191 we have that @xmath9 is a subgroup of @xmath802 . using the isomorphisms of lemmas
[ zz ] and [ zzz ] we have that @xmath9 is a subgroup of @xmath816 generated by the image of @xmath817 under the two isomorphisms : elements @xmath818 . finally , because @xmath819<\infty$ ] we have by lemma [ wkf ] that the distortion of @xmath9 in @xmath0 is equivalent to the distortion of its image in @xmath820 .
theorem [ x ] parts @xmath821 and @xmath70 follow from lemma [ abel ] if the subgroup @xmath9 is abelian .
otherwise they follow from corollary [ awrz ] and lemma [ r5 ] .
now we complete the proof of theorem [ x ] , part @xmath822 .
let @xmath1 be a finitely generated abelian group of rank @xmath43 .
consider the @xmath3-generated subgroup @xmath157 constructed as follows .
let @xmath54 .
consider @xmath823 .
then the distortion of the polynomial @xmath352 is seen to be equivalent to @xmath64 , by lemma [ tgs ] . by lemma [ hpu ] , this means that the @xmath3-generated subgroup @xmath824 has distortion @xmath825 .
the subgroup @xmath47 is a retract of @xmath0 if @xmath1 is infinite .
therefore , the distortion of @xmath67 in @xmath47 and in @xmath0 are equivalent by lemma [ wkf ] .
if we adopt the notation that the commutator @xmath826=aba^{-1}b^{-1}$ ] , then we see that in @xmath47 , the element of @xmath91 corresponding to the polynomial @xmath827 is @xmath828,b ] , \cdots , b]$ ] where the commutator is @xmath62-fold .
this explains corollary [ xx ] . | we study the effects of subgroup distortion in the wreath products @xmath0 , where @xmath1 is finitely generated abelian .
we show that every finitely generated subgroup of @xmath0 has distortion function equivalent to some polynomial .
moreover , for @xmath1 infinite , and for any polynomial @xmath2 , there is a @xmath3-generated subgroup of @xmath0 having distortion function equivalent to the given polynomial .
also a formula for the length of elements in arbitrary wreath product @xmath4 easily shows that the group @xmath5 has distorted subgroups , while the lamplighter group @xmath6 has no distorted ( finitely generated ) subgroups . |
the early 1970s witnessed the start of the intensive study of radiation emanated by superluminal sources @xcite , which received a new impetus due to possibility of launching propagation - invariant localized waves @xcite , whose fundamental representations and classification schemes are discussed in @xcite . among different types of the localized wave structures , the x - shaped waves are one of the most well - known @xcite .
recami et al .
@xcite proposed a toy model of localized x - shaped field generation by an axisymmetric superluminal four - current @xmath0 , where @xmath1 is the speed of light , @xmath2 the electric charge density , and @xmath3 the only non - zero component of the current density vector @xmath4 in the cylindrical coordinates @xmath5 .
considering the inhomogeneous electromagnetic problem for the four - potential in the form @xmath6 , @xmath7 , @xmath8 , where @xmath9 is a @xmath10-cone variable , introduced for constant superluminal velocity @xmath11 , and @xmath12 the time variable , the investigation was inevitably constrained to the cases that ( i ) describe a steady - state process that lasts all times from @xmath13 to @xmath14 , ( ii ) require similar four - current dependence @xmath15 , and ( iii ) bound the nonzero components of the four - vectors by relations @xmath16 and @xmath17 due to , respectively , the continuity equation and the lorentz gauge .
the problem was solved for the case of the rectilinear motion of a point - like charge using the spectral method ( fourier - bessel and fourier transforms with respect to @xmath18 and @xmath19 ) , yielding a localized `` true x wave '' solution @xmath20 being symmetric with respect to time reversal and involving equally the advanced and retarded components , such a model of x wave generation is acausal .
later attempts to remedy this drawback introducing the unilateral conical - wave source @xcite and applying the unidirectional decomposition @xcite resulted in more realistic models .
all these studies , however , do not take into account the temporal constraint of the source motion even in its weak form that assumes the existence of an initial moment , prior to which neither the source pulse nor the emanated wave can exist .
the present work considers wave generation by superluminal charges within the framework of the riemann - volterra approach , introducing a direct space - time domain method that , in contrast to constructing the wavefunctions via the bateman transform @xcite or green s method @xcite , deals with the initial value problem that includes , in addition to the inhomogeneous wave equation @xmath21\psi \left ( \tau , \rho , z \right ) = s \left ( \tau , \rho , z \right ) , \ ] ] the time asymmetric initial condition @xmath22 which defines the _ arrow of time _ and , as will be shown , represents a sufficient condition for existence of a unique causal solution . for
the electromagnetic waves , we assume that @xmath23 and @xmath24 represent , respectively , the @xmath25-components of the four - potential @xmath26 and four - current @xmath27 ( @xmath28 stands for the magnetic constant ) . furthermore , let us get over one singularity , considering , instead of the line @xmath29-pulse , a line source of arbitrary shape @xmath30 propagating with a constant superluminal speed in the positive @xmath25-direction .
in view of ( [ inhomogenwaveeq ] ) , ( [ timeasyminicond ] ) , such a source can be expressed in the form @xmath31 where @xmath32 is the elementary charge while @xmath33 and @xmath34 stand for the dirac delta and heaviside step functions .
one can easily notice that due to ( [ timeasyminicond ] ) in no case the problem in question can yield a pure localized solution meeting the propagation - invariant condition stated by lu ( ref .
@xcite , p. 98 ) as reducibility to the form @xmath35 . instead , it provides us with bounded - support , finite - energy wavefunctions ( signals ) that can serve as real - world , physically admissible approximations of the ideal temporally unbounded localized waves . borrowing the terminology from the numerical analysis
, we may say that the description of a localized propagation - invariant wave by such a signal is convergent if the signal asymptotically approaches the localized wave structure as @xmath36 .
the continuity equation @xmath37 and representation ( [ sourcecurrent ] ) immediately implies that the charge density is uniquely defined by equation @xmath38 = \frac{1}{c\beta } j+ \left ( -e \right)\frac{\delta \left ( \rho \right)}{2\pi \rho } h\left ( \tau \right)\delta \left ( z \right ) \int_0^{\beta \tau}{d \xi f \left ( \xi \right ) } , \end{split}\ ] ] and the other non - zero component of the four - potential @xmath39 automatically satisfying @xmath40 can be obtained as a consequence of ( [ inhomogenwaveeq ] ) , ( [ timeasyminicond ] ) , and ( [ formulaforelcharge ] ) from the lorenz gauge @xmath41 by a simple integration @xmath42 here the second term describes accumulation of the opposite charge at the pulse - generation point @xmath43 due to current outflow . for models concerning the wave emanation by hypothetical superluminal particles , tachyons
, such an accumulation can be nulled if we suppose that the generation of a particle is accompanied by generation of an antiparticle ( an opposite point - like charge ) moving in different direction . for other models involving more traditional source structures the excessive charge accumulation
can be avoided in the long term by considering ( bipolar ) current pulses for which @xmath44 is vanishing or , at least , limited ( demonstrating , for example , oscillating behavior ) .
one of such embodiments discussed by ziolkovski et al .
@xcite and later by saari @xcite envolves superluminal sink - source - type fictitious currents @xcite . as discussed in @xcite with reference to fig . 1
therein , the simplest and most straightforward implementation of the superluminal wave motion can be achieved with a type of `` scissor effect '' , creating an interference pattern of two ( luminal - speed ) plane waves whose fronts are inclined with respect to each other .
similar patterns can be created , e.g. , at oblique incidence of wavefronts onto locally plane targets ( screens ) . on the cosmic scale , this phenomena is well discussed by astrophysicists ( see , e.g. , @xcite ) .
remarkably , recent advances in ultraintense laser interaction science enables such huge - scale phenomena to be reproduced in the micrometer scale in the laboratory conditions .
excitation of a source current pulse by an ultra - intense , ultra - short pulsed laser beam , incident at an angle to a thin plasma filament induced in a medium by another laser beam is an example ( here significant current density can be obtained due the effects akin to the wake - field acceleration while the desired orientation of the macroscopic current along the filament can be asserted by imposing a strong magnetic field ; description of pertinent phenomena can be found in @xcite ) .
separating @xmath18 by the fourier - bessel transform @xmath45 one gets from ( [ inhomogenwaveeq ] ) , ( [ timeasyminicond ] ) and ( [ sourcecurrent ] ) a simpler time asymmetric initial value problem for the klein - gordon equation @xmath46 with the initial condition @xmath47 one can easily check by direct enumeration that condition ( [ conditionforkge ] ) uniquely defines the transition to the ( characteristic ) variables @xmath48 that : ( i ) transforms the klein - gordon equation to the first canonical form and ( ii ) results in the geometry of the solution support , the area of @xmath49 , corresponding to the well - established layout of the riemann - volterra method ( see , e.g. , fig . 30 in @xcite or fig . 1 in @xcite ) .
ad hoc _ riemann - volterra procedure involving known riemann function @xmath50 ( see @xcite for details ) readily provides a unique causal solution @xcite , which after the inverse transition to the independent variables @xmath51 takes the form @xmath52 using the inverse fourier - bessel transform , the closure equation ( see , e.g. , @xcite , p. 691 ) @xmath53 and the representation of the delta function with simple zeros @xmath54 ( @xcite , p. 87 ) @xmath55 ( among the two simple zeros , only one lies within the integration limits ) , we get the solution to the initial space - time domain problem ( [ inhomogenwaveeq ] ) , ( [ timeasyminicond ] ) in the form @xmath56 as illustrated in fig .
[ fig_integration_segment ] , @xmath23 is defined by integration over the curve segment @xmath57 that is the intersection of : the initial triangle integration area @xmath58 ; the half - plane @xmath59 , the support of the step function @xmath60 ; the half - plane @xmath61 , the support of the step function @xmath62 ; and the hyperbola @xmath63 , the support of the delta function @xmath64 .
instantiation of general formula ( [ dblintegralsolution ] ) with the help of the @xmath65 plane diagrams ( fig .
[ fig_zp_taup_sequence ] ) reveals the explicit structure of the solution , if @xmath66 , @xmath67 otherwise ( @xmath68 ) @xmath69 that depends on the following parameters : @xmath70 $ ] , @xmath71 , @xmath72 , @xmath73 , @xmath74 , and @xmath75 .
the result illustrates capability of the method that does not resort to _ a priory _ causality conditions like retarded nature of the argument , wave propagation in a fixed direction , etc . to yield solutions whose space - time structure admits easy _ posterior _ interpretation in terms of causal propagation of information : for an observation point @xmath76 there is no intersection between @xmath77 and @xmath78 ( no wave registered ) until the moment @xmath79 , in which the hyperbola touches the boundary of @xmath77 in the ( tangency ) point @xmath80 , @xmath81 , as depicted in fig .
[ fig_zp_taup_sequence](a ) .
for @xmath82 the value of @xmath83 represents the minimum time necessary for the electromagnetic energy , originated in the space - time point @xmath84 , to reach the observation point @xmath85 , @xmath86 first propagating during time @xmath87 along the @xmath88 axis at the superluminal speed @xmath89 with the source - pulse front and then during time @xmath90 out of this axis toward @xmath91 at the luminal speed with the front of the emanated electromagnetic wave ( see the upper diagram of fig .
[ fig_zp_taup_sequence](a ) ) . for cases corresponding to figs .
[ fig_zp_taup_sequence](a , b ) , eq .
( [ psiextraordinarycases ] ) characterizes quantitatively a source pulse that appears to the observer `` suddenly growing out of a point '' @xmath92 at @xmath79 and expanding in the opposite directions , just as was predicted by gron @xcite on the basis of purely geometrical consideration ( see also fig .
15 of @xcite or fig . 1 of @xcite ) .
notably , apart from @xcite , other pioneering papers showed how a superluminal source , after having suddenly revealed in an `` optical - boom '' phase , may subsequently appear as a couple of objects receding one from the other , in particular , in astrophysical observations ( so - called `` superluminal expansions '' ) .
for a detailed discussion of these models , both `` orthodox '' in which a superluminal pattern is created by a coordinated motion of the subluminally moving constituents and `` true superliminal '' in which actual superluminal motions are treated within the framework of the extended special relativity theory , the interested reader is directed to recami et al . @xcite and references therein
. causal solution ( [ psiextraordinarycases ] ) demonstrates other phenomena discussed in the literature from the standpoint of the extended special relativity : as expound in @xcite , the superluminal ( tachyonic ) motions are always observed being forward in time , but can appear reversed in direction ; as well , the duality between the source and detector makes possible the emission to be observed as absorption . in the case in question ,
the `` post - boom '' evolution of the source pulse is observed as forward and backward expansion .
the latter stops as the pulse reaches its origin @xmath93 ( figs .
[ fig_zp_taup_sequence](c , d ) ) and at this point the ( apparently reversed ) generation process manifests as the pulse absorption . in the limiting case of a @xmath29-pulse
the half - plane @xmath77 degenerates into a line and the segment of integration @xmath94 into points @xmath95 and @xmath96 , representing two images of the same tachyon .
source @xmath95 is eventually absorbed at @xmath93 , abruptly diminishing the wave amplitude ( as will be shown later , by half ) .
the @xmath97 plane map of the three areas differing in the solution representation is shown in fig .
[ fig_case_areas_z_tau ] .
.,scaledwidth=50.0% ] aiming at construction of the propagation - invariant waves , it is worthwhile to express the wavefunction in terms of the propagation variable @xmath98 . in contrast to the pure steady - state solutions , representation of wavefront - limited causal waves does not exclude the time dependence , and passing from @xmath99 to @xmath100 _ transforms _ : \(i ) the non - zero wave condition @xmath101 _ into _ @xmath102 , a causal analog of the condition @xmath103 , reported in ( @xcite , eq .
( 8) ) for the steady - state electromagnetic field of a charged tachyon in agreement with predictions expound by recami et al . on the basis of the extended theory of special relativity ( see @xcite , especially fig . 4 , as well as earlier works on superluminal lorentz transformations @xcite ) ; \(ii ) the case - limiting condition @xmath104 , _ into _ @xmath105 , where @xmath106 are two roots of the equation @xmath107 ( iii ) case formulas ( [ psiordinarycases ] ) , ( [ psiextraordinarycases ] ) _ into _ + @xmath108 @xmath109 @xmath110 @xmath111 @xmath112 \right ) } { r \left(\tau ' , \rho , \xi \right)}},\ ] ] @xmath113 + @xmath114 \right ) } { r \left(\tau ' , \rho , \xi \right)}};\ ] ] ( iv ) the case map of fig .
[ fig_case_areas_z_tau ] ( @xmath51 plane ) _ into _ the case map represented in fig . [ fig_case_areas_xi_tau ] ( @xmath115 plane ) .
.,scaledwidth=50.0% ] contour plot , representing the wave propagation via a characteristic space - time scale @xmath116 and the dimensionless quantities @xmath117 , @xmath118 , @xmath119 , @xmath120 , is given in fig .
[ fig_wfunction_supp](a ) ( @xmath121 , as in numerical illustrations of ref .
@xcite ) . for each fixed moment of time
@xmath122 the solid isoline @xmath123 defines the boundary of the wavefunction support ( area @xmath124 corresponding to @xmath125 ) while the dashed isoline @xmath126 traces the boundary between cases ( [ psi_caseb ] ) and ( [ psi_casec ] ) .
the emanated wave has an expanding droplet - shaped support , defined exclusively by the parameters @xmath127 , and @xmath128 . fig .
[ fig_wfunction_supp](b ) represents the structure of this droplet - shaped wave for @xmath129 ; the wavefunction is represented in dimensionless units via normalization @xmath130 . ) and the boundary between cases ( [ psi_caseb ] ) and ( [ psi_casec ] ) ( dashed isolines @xmath131 ) ; ( b ) a snapshot of @xmath132 , clipped by @xmath133 in the vicinity of its singularities , taken at @xmath134.,scaledwidth=80.0% ]
using the above approximation of an infinitely short source current pulse @xmath129 , which in the case discussed in @xcite led to the model of x wave generation by a charged tachyon , one readily arrives at the solution that in the area @xmath135 and @xmath136 ( @xmath125 ) reads @xmath137 the equation for the delta - function roots coincides with ( [ eq_for_tau12 ] ) , so these roots , @xmath138 , are defined by formula ( [ tau12 ] ) . using ( [ deltafuncwithsimple0 ] ) and passing to the dimensionless parameters finally reduce ( [ psicasesfordeltapulse ] ) to the formula @xmath139 one can check by direct calculation that the second case of ( [ psifordeltadimless ] ) corresponds to solution of the inhomogeneous wave equation of ref .
@xcite , describing the steady - state x wave produced by a charged tachyon ( delta - pulse source ) .
introduction of the initial moment of particle generation results in launching of the same propagation - invariant waveform , which however is restricted by the droplet - shaped support illustrated in fig .
[ fig_wfunction_supp](a ) .
as the time increases , this waveform expands in both @xmath140 and @xmath141 directions , tending to the charged - tachyon x - shaped field , as illustrated in fig .
[ fig_drop_wave_dynam ] .
diverges , are omitted).,scaledwidth=100.0% ] for @xmath142 the singularity corresponding to @xmath143 ( in fig .
[ fig_zp_taup_sequence ] , to @xmath144 ) resides outside the integration segment @xmath94 , diminishing @xmath23 by half .
corresponding electromagnetic field vectors @xmath145 and @xmath146 ( of which only @xmath147 , @xmath148 , and @xmath149 components are nonzero ) , can readily be found using eqs .
( [ formulaforelcharge ] ) and ( [ formulaforao ] ) .
in particular , for the observation times @xmath150 ( a steady - state wave zone located outside the singularities arising from the potential discontinuities on the case delimiting boundaries of fig . [ fig_case_areas_xi_tau ] ) the magnetic field ( normalized by @xmath151 ) is characterized by @xmath152^ { - 3/2}$ ] . as in the case described in @xcite , sec .
iii , it remains the only component that does not vanish when @xmath153 , revealing a `` magnetic monopole '' behavior . as put forward in @xcite , the reference frame in which the particle velocity tends to infinity plays for tachyons the same role as the rest frame for ordinary particles ( bradyons ) , and there exists a duality between subliminal electric charges and superluminal magnetic monopoles .
so , for @xmath153 one might expect the magnetic field to have a structure akin to that of the electric field of a charged particle .
notably , earlier applications of the proposed technique to the waves emanated by subluminal sources ( see , for instance , @xcite ) result in peak- or ball - like shapes akin to the subliminal wave bullets obtained in secs .
ii - iv of ref . @xcite .
while electromagnetic fields of different line currents propagating at luminal speed present singularity at one `` point of accumulation '' , @xmath154 ( see , e.g. , fig . 1 of @xcite ) , for the superluminal delta current the area of singularity spreads along the conical surface @xmath155 , representing , according to @xcite an initially point - like structure ( perceived by superluminal observers ) highly distorted by the superluminal lorentz transformation . in toto ,
the results obtained support the general idea about the shape of superluminal particles and distribution of the field associated with superluminal charges : `` while the simplest subluminal object is obviously a sphere or , in the limit , a space point , the simplest superluminal object is on the contrary an x - shaped pulse '' @xcite .
the present work introduces a new type of the propagation - invariant , localized droplet - shaped waves that , being solutions of the inhomogeneous wave equation and satisfying zero initial conditions , admit causal generation by real pulsed superluminal sources widely discussed in the literature ( see in particular the monograph @xcite and more recent review @xcite by recami ) .
although the presented analysis is limited to the most illustrative case of @xmath29-pulse current , the general integral solution ( [ psi_casea])-([psi_casec ] ) enables the propagation - invariant localized droplet - shaped waves emanated by various line sources to be investigated both analytically and numerically using the @xmath65 diagrams similar to one of fig .
[ fig_zp_taup_sequence ] .
various possible anstze are discussed in @xcite for free space and @xcite for waveguides .
the author is grateful to michel zamboni - rached and erasmo recami for several useful discussions that significantly improved the introductory part of the article as well as for their kind cooperation that stimulated publication of this study .
the research was partially based on work supported by the international science foundation under grant m3h000 _ generation and propagation of localized waves_. e. recami , m. zamboni - rached , k. z. nobrega , c. a. dartora , and h. e hernandez f. , `` on the localized superluminal solutions to the maxwell equations , '' ieee journal of selected topics in quantum electronics , * 9 * , 59 - 73 ( 2003 ) .
i. m. besieris , m. abdel - rahman , a. m. shaarawi , and a. chatzipetros , `` two fundamental representations of localized pulse solutions to the scalar wave equation , '' prog .
19 * , 148 ( 1998 ) .
p. saari , `` superluminal localized waves of electromagnetic field in vacuo , '' in d. mugnai , a. ranfagni , and l. s. shulman , eds . , _ time arrows , quantum measurement and superluminal behaviour _ ( cnr , 2001 ) , pp .
3748 ( arxiv : physics/0103054v1 ) .
p. caldirola , g. d. maccarrone , and e. recami , `` second contribution on solving the < < imaginary quantities > > problem in superluminal lorentz transformations , '' lett .
nuovo cimento series 2 * 29 * , 241250 ( 1980 )
. a. b. utkin , `` riemann - volterra time - domain technique for waveguides : a case study for elliptic geometry , '' wave motion , to be published , doi:10.1016/j.wavemoti.2011.12.001 , + http://www.sciencedirect.com/science/article/pii/s0165212511001430 | a model of steady - state x - shaped wave generation by a superluminal ( supersonic ) point - like source infinitely moving along a straight line is extended to a more realistic causal scenario of a source pulse launched at time zero and propagating rectilinearly at constant superluminal speed . in the case of infinitely short ( delta )
pulse , the new model yields an analytical solution , corresponding to the propagation - invariant x - shaped wave clipped by a droplet - shaped support , which perpetually expands along the propagation and transversal directions , thus tending the droplet - shaped wave to the x - shaped one
. _ ocis codes _ : 350.5720 , 320.5550 , 350.7420 , 070.7345 , 350.5500 , 070.0070 _ pacs numbers _ : 43.20.bi , 03.50.de , 41.20.jb |
analysis of the lyapunov exponent plays an important role in the problems of propagation of waves .
much attention in recent years has been given to periodic and disordered one - dimensional structures ( see , e.g. , @xcite and references therein ) .
application of such structures include photonic crystals , coupled - resonator optical waveguides , slow - light devices , metamaterials , semiconductor superlattices with given transmission properties and more .
characteristics of real materials , however , may be considerably different from predicted ones due to inevitable disorder of geometric and material parameters . structural perturbation and manufacturing inaccuracies may lead to a substantial reduction of performance @xcite , @xcite and even to complete suppression of wave propagation @xcite , @xcite . in the strongest form the disorder may manifest itself in the anderson type localization , which is a general wave phenomenon for classical and quantum waves in disordered systems . due to multiple scattering , destructive interference occurs , so that the wave amplitude decreases exponentially in the medium . the rate of decay of the solution is defined by the lyapunov exponent , @xmath0 ( the inverse of the localization length ) .
the lyapunov exponent is of major importance in the problems of slowing down light by orders of magnitude , a phenomenon which has been extensively discussed in the literature @xcite .
such a possibility can be useful in a variety of optical and microwave applications .
choosing wave frequency near stationary points of the electromagnetic dispersion relation allows one to reduce the group velocity by two orders of magnitude @xcite-@xcite .
another model for slowing down light proposed in @xcite is based on the periodic necklace - type waveguide .
an essential common feature of such devices is that the frequency of propagating wave must be located near the spectral band edge or degenerate band edge of the corresponding medium .
one of the major factor , however , limiting performance of optical waveguides is their structural imperfection , and investigation of the influence of different kinds of disorder on the lyapunov exponent becomes of primary importance in manufacturing of optical devices .
papers @xcite-@xcite performed sensitivity analysis of photonic crystals near degenerate and regular spectral band edge under perturbations of geometrical and material parameters , and the impact of these perturbations was compared .
numerical simulations were used to investigate the sensitivity of field enhancement effects at fabry - perot resonances against the bandwidth of the excitation . the gigantic amplitude increase produced by such resonances
was observed , which dramatically deteriorates under small perturbations of the thickness of layers . in particular , the peak field intensity of a 16 layer stack regular band edge photonic crystal decreases by an order of magnitude under @xmath3 perturbation of the layer thicknesses .
the central idea of chaotic behaviour of dynamical systems is widely seen as strong dependence on initial conditions @xcite . in a chaotic system ,
a very small change in the initial conditions can cause a big change in the trajectory .
quantitative criterion for sensitivity is characterized by the lyapunov exponent ( also called the maximal lyapunov exponent ) which measures exponential growth rate of the solution . in all the applications ,
the influence of the strength of perturbation @xmath1 is essentially different depending on how close the frequency @xmath2 is to the band edge . while @xmath4 in the bulk of the band , it has order @xmath5 near the band edges . therefore , it is important to construct uniform two - parameter asymptotic expansion of the lyapunov exponent for all frequencies of wave propagation .
let @xmath6 be the lyapunov exponent of the optical equation @xmath7 where @xmath8 denotes the strength of perturbation of either the value of the periodic refraction index @xmath9 or the length of its period , and @xmath10 is located in an @xmath11-neighbourhood of a non - degenerate band edge .
let also @xmath12 be the lyapunov exponent of the schrdinger equation @xmath13 with the white noise potential @xmath14 and the frequency @xmath2 located near the spectral edge @xmath15 .
the frequency @xmath2 is defined by @xmath10 using the relation @xmath16 here @xmath17 is the transfer matrix of ( [ oe ] ) .
then in @xcite it was shown that @xmath18 and @xmath0 estimate each other uniformly .
that means @xmath19 where positive constants @xmath20 and @xmath21 depend only on @xmath22 and @xmath11 but not on the values of @xmath1 and @xmath10 .
thus , a uniform two - parameter asymptotic expansion of the lyapunov exponent of the schrdinger equation ( [ sche ] ) provides a uniform estimate of the lyapunov exponent of the wave equation ( [ oe ] ) .
because of this , we consider in details the schrdinger equation with different relations between the intensity of noise @xmath1 and the frequency @xmath2 .
we will obtain a uniform asymptotic expansion of @xmath0 followed by numerical evaluation of @xmath0 to determine the range of parameters for which the asymptotic formulas can be used .
in this paper we study the asymptotic behaviour of the lyapunov exponent @xmath23 of the random one - dimensional schrdinger operators @xmath24 \psi(0 ) = \cos { \vartheta}_0 , \quad { \displaystyle}\frac{\psi^\prime ( 0)}{\omega } = \sin { \vartheta}_0 , \end{array } \label{h}\ ] ] as a function of the energy @xmath25 and the coupling constant ( intensity ) @xmath1 of the white noise potential . here
@xmath26 ( equations with any real @xmath25 will be considered ) .
the white noise @xmath14 can be understood as the limit of a stationary gaussian process @xmath27 with independent increments on intervals of length @xmath28 ( cf .
@xcite ) : @xmath29 an equivalent alternative approach to define the operator with a white noise potential ( which is used in the present paper ) is based on the reduction of ( [ h ] ) to a system of it stochastic differential equations of the first order for the phase @xmath30 and amplitude @xmath31 of @xmath32 . in what follows
we will denote @xmath33 . then the it and stratonovich representations of ( [ h ] ) coincide in coordinates @xmath34 and both give the same result ( see @xcite ) .
after the standard substitution ( in the prffer form ) @xmath35 equation ( [ h ] ) takes the form @xmath36 usually the reduction of the schrdinger equation to a system of two first order equations is accompanied by the phase - amplitude formalism , i.e. , by using the polar coordinates @xmath37 { \displaystyle}v=\frac{\psi^\prime}{{\omega } } = r ( x ) \cos { \vartheta}(x ) , \end{array } \label{pol}\ ] ] where @xmath38 . the lyapunov exponent @xmath39
is defined by @xmath40 this limit exists _
on the probability space @xmath41 defined by the white noise @xmath42 in ( [ h ] ) ( see @xcite , @xcite ) . we will discuss the limit behaviour of @xmath39 in the different regimes when @xmath43 or @xmath44 ( small and large disorder ) and @xmath45 or @xmath46 ( long and short waves ) .
we also introduce some combinations of @xmath47 and @xmath48 to describe the intermediate asymptotics .
asymptotic behaviour of @xmath0 for @xmath49 and fixed @xmath50 was established in papers @xcite and @xcite where it was shown that @xmath51 paper @xcite contains the formula @xmath52 for fixed @xmath53 and @xmath54 .
we show that in fact @xmath55 , where @xmath56 and @xmath57 is a function of @xmath58 .
thus , ( [ gam0 ] ) is also valid for arbitrary @xmath1 and @xmath2 such that @xmath59 in this paper we suggest a simple approach to obtaining the asymptotic expansion for @xmath0 as @xmath60 as well as @xmath61 . therefore , different combinations of large and small perturbations and frequencies are covered .
we show that @xmath62 .
in particular , it provides low frequency asymptotics of @xmath0 when @xmath1 is fixed and @xmath63 .
the case of @xmath64 is also considered .
the next section contains an exact formula for @xmath0 followed by a section with derivation of its asymptotics expansion .
we will use equations ( [ sfo ] ) to obtain the integral representations for @xmath0 .
the following formulas for the lyapunov exponent hold p - a.s . 1 .
if @xmath65 then @xmath66 where @xmath67 with a normalizing constant @xmath68 given by @xmath69 2 . if @xmath70 then @xmath71 where @xmath72 and @xmath73 let @xmath74 we rewrite ( [ sfo ] ) in polar coordinates ( [ pol ] ) but use @xmath75 instead of @xmath30 .
the resulting equations will be much simpler in the coordinates @xmath34 than in @xmath76 .
one only needs to keep in mind that the process @xmath77 has boundary conditions at infinity : if @xmath78 when @xmath79 approaches from the left to some finite value @xmath80 , then @xmath81 .
this jump corresponds to the transition of the angle variable @xmath82 from @xmath83 to zero .
we will use the it formula in order to write equations for @xmath34 .
the it formula says that for any smooth function @xmath84 , @xmath85 where @xmath86 are given by ( [ sfo ] ) and @xmath87 .
there is only one quadratic in @xmath86 term on the right , since equation ( [ sfo ] ) for @xmath88 does not contain @xmath89 .
if we put @xmath86 from ( [ sfo ] ) into the it formula we arrive at @xmath90 & = \left(\frac{{\partial}f}{{\partial}u}\omega v-\frac{{\partial}f}{{\partial}v } \omega u + \frac{1}{2}\frac{{\partial}^2 f}{{\partial}v^2 } \frac{\sigma^2 u^2}{\omega^2 } \right)dx + \frac{\sigma}{\omega } \frac{{\partial}f}{{\partial}v } udw .
\label{genf}\end{aligned}\ ] ] since @xmath91 ( see ( [ pol ] ) ) , we apply ( [ genf ] ) to function @xmath92 . for this function
@xmath93 we have @xmath94 and therefore , @xmath95 we will need function @xmath96 , not @xmath31 .
thus from ( [ genf ] ) with @xmath97 we obtain @xmath98 the right - hand side of the latter equation does not depend on @xmath31 .
thus the solution of the original schrdinger equation ( [ h ] ) is reduced to a solution of the stochastic equation ( [ eqz ] ) followed by a direct integration of ( [ eqln ] ) .
consider the generator of the diffusion process @xmath99 : @xmath100 the probability density @xmath101 ( with respect to @xmath102 with initial point @xmath103 ) of the diffusion process @xmath99 satisfies the equation @xmath104 and the limiting probability density @xmath105 is given by @xmath106 or @xmath107 the solution of the last equation has the form ( [ p ] ) where the normalizing constant is defined by the condition @xmath108 that is , @xmath109 we make the substitution @xmath110 here and evaluate the integral in @xmath111 .
this leads to ( [ cp ] ) .
after the limiting probability density @xmath112 is found , formula ( [ gamma1 ] ) is an immediate consequence of ( [ gamma ] ) and ( [ eqln ] ) .
one needs only to note that the boundedness of the integrand in the second term in ( [ eqln ] ) implies that @xmath113-a.s .
@xmath114 if @xmath115 then we denote @xmath116 in ( [ h ] ) , and the system ( [ sfo ] ) becomes @xmath117 the stochastic differential equation for @xmath118 analogous to ( [ eqz ] ) has the form @xmath119 the stochastic equation for @xmath120 similar to ( [ eqln ] ) is @xmath121 dx -\frac{\sigma}{\omega}\frac{z}{1+z^2}\,dw.\ ] ] determining the limiting probability density @xmath105 as in ( [ dep ] ) we obtain that it satisfies the equation @xmath122 the solution of this equation is given by ( [ p- ] ) with a normalizing constant determined from ( [ c- ] ) .
therefore , the expression of the lyapunov exponent is given by ( [ gamma- ] ) .
the exact expression of the lyapunov exponent @xmath0 is given by ( [ gamma1 ] ) if @xmath123 and by ( [ gamma- ] ) if @xmath115 , and can be calculated numerically .
although equation ( [ h ] ) depends on two parameters @xmath25 and @xmath1 , formula ( [ gamma1 ] ) implies that @xmath0 depends on the product of @xmath124 and a function of @xmath125 .
the next theorem provides the asymptotic behaviour of the lyapunov exponent for @xmath61 and @xmath60 .
lyapunov exponent of the schrdinger operator ( [ h ] ) has the following asymptotic behaviour : * if @xmath126 and @xmath61 ( long wave asymptotics ) then @xmath127 where exact value of @xmath128 is given by ( [ c ] ) . *
if @xmath129 and @xmath130 ( short wave asymptotics ) then @xmath131 * if @xmath132 and @xmath130 then @xmath133 * let us find the asymptotic expansion of @xmath134 in ( [ cp ] ) using the substitution @xmath135 . if @xmath50 then @xmath136 & = \sum_{n=0}^\infty ( -1)^n a_n\ , \nu^{\frac{2(n-1)}{3 } } , \quad a_0 = \pi^{\frac{1}{2}}\,2^{\frac{1}{3}}\ , 3^{-\frac{5}{6}}\ , \gamma \left(\frac{1}{6 } \right ) , \quad \nu \to 0 , \label{a}\end{aligned}\ ] ] where @xmath137 and @xmath138 is the gamma function .
hence , for the constant @xmath68 we obtain @xmath139 + to find asymptotics of @xmath0 we use two substitutions @xmath140 and @xmath141 in ( [ gamma1 ] ) and ( [ p ] ) , respectively
. then @xmath142 & = \omega \nu^{-1 } c\int_{-\infty}^{\infty } \frac{\nu^\frac{2}{3}-s^2}{(\nu^\frac{2}{3}+s^2)^2}\ , q(s,\nu)\,ds , \label{gamma11}\end{aligned}\ ] ] where @xmath143}\ , dt .
\label{q1}\end{aligned}\ ] ] denote by @xmath144 a contour in the complex @xmath145-plane which consists of the real axis with the interval @xmath146 replaced by the semicircle @xmath147 , @xmath148 .
then for @xmath149 @xmath150 one can show that @xmath151 and each of its derivatives with respect to @xmath152 are uniformly bounded in @xmath153 , @xmath154 .
thus , the first term in ( [ cont ] ) is infinitely smooth function of @xmath155 , @xmath156 , and @xmath157 the main term in ( [ main ] ) can also be written in the form @xmath158 & = { \displaystyle}{-\hspace{-12pt}\int}_{-\infty}^\infty \frac{q(0,0 ) - q(s,0)}{s^2}\,ds - \pi i,\end{aligned}\ ] ] where @xmath159 stands for the cauchy principle value .
the second term in ( [ cont ] ) is analytic in @xmath160 and hence @xmath161 & = -\pi i \left[a + b\nu^{\frac{1}{3 } } + o(\nu^\frac{2}{3 } ) \right ] , \label{}\end{aligned}\ ] ] where @xmath162 and @xmath163 .
thus , we obtain two formulas for computation of @xmath0 : through the principal value and as a contour integral @xmath164 & = { \displaystyle}\omega \nu^{-1}c \left\{-\int_{l}\frac{q(s,0)}{s^2}\,ds + \pi
i + o(\nu^\frac{2}{3})\right\ } { \nonumber}\\[2 mm ] & = c\ , \sigma ^{\frac{2}{3}}\left(1 + o(\nu^{\frac{2}{3}})\right ) , \quad \nu \to 0 , \label{gamma_a}\end{aligned}\ ] ] where @xmath165 + if @xmath64 then formula ( [ c- ] ) for @xmath134 differs from ( [ cp ] ) only by a sign in the exponent
. therefore expansion ( [ a ] ) with @xmath166 replaced by @xmath167 holds : @xmath168 thus , asymptotics ( [ c1 ] ) is valid in this case .
+ the lyapunov exponent is determined from ( [ p- ] ) and ( [ gamma- ] ) : @xmath169 & { \displaystyle}-2\omega \,c\int_{-\infty}^{\infty } \frac{x e^{\nu(\frac{x^3}{3}-x)}}{1+x^2}\ , dx\int_x^{\infty } e^{-\nu(\frac{t^3}{3}-t)}\ , dt { \nonumber}\\[2 mm ] & { \displaystyle}=\omega \nu^{-1 } c\int_{-\infty}^{\infty } \frac{\nu^\frac{2}{3}-x^2}{(\nu^{\frac{2}{3}}+x^2)^2}\ , r(x,\nu ) \ , dx -2\omega \nu^{-\frac{1}{3}}c \int_{-\infty}^{\infty } \frac{xr(x,\nu)\,dx}{\nu^\frac{2}{3}+x^2 } , \label{gamma31}\end{aligned}\ ] ] where @xmath170 denote the terms on the right hand side of ( [ gamma31 ] ) by @xmath171 and @xmath172 , respectively .
then @xmath171 is analogous to that in ( [ gamma11 ] ) and gives the main contribution to @xmath0 . to determine asymptotics of @xmath171
, we use the same contour @xmath144 in the complex @xmath145-plane as in ( [ cont ] ) : @xmath173 { \raggedright } & = \omega \nu^{-1}c\left\{-\int_{l } \frac{r(s,0)}{s^2}\ , ds - \pi i r^\prime_x ( i\nu^{\frac{1}{3}},\nu ) + o(\nu^\frac{2}{3 } ) \right\ } { \nonumber}\\[2 mm ] { \raggedright } & = \omega \nu^{-1}c\left\{-\int_{l } \frac{q(s,0)}{s^2}\ , ds + \pi i + o(\nu^\frac{2}{3 } ) \right\ } { \nonumber}\\[2 mm ] { \raggedright } & = \omega \nu^{-1}c\left\{{-\hspace{-12pt}\int}_{-\infty}^{\infty } \frac{q(0,0)-q(s,0)\,ds}{s^2 } + o(\nu^\frac{2}{3})\right\ } = c\ , \sigma ^{\frac{2}{3}}\left(1 + o(\nu^{\frac{2}{3}})\right ) , \quad \nu \to 0,\end{aligned}\ ] ] where @xmath128 is defined by ( [ c ] ) .
the second term @xmath172 in ( [ gamma31 ] ) has smaller order in @xmath58 : @xmath174 { \raggedright } & = -2\omega \nu^{-\frac{1}{3}}c\left\{\int_{l } \frac{r(s,0)}{s}\ , ds + \pi i r ( i\nu^{\frac{1}{3}},\nu ) + o(\nu^\frac{2}{3 } ) \right\ } { \nonumber}\\[2 mm ] { \raggedright } & = -2\omega \nu^{-\frac{1}{3}}c\biggl\{{-\hspace{-12pt}\int}_{-\infty}^\infty \frac{r(s,0)}{s}\ , ds - \pi i\res_{s = i\nu^\frac{1}{3 } } \frac{r(s,0)}{s } + \pi i r
( i\nu^{\frac{1}{3}},\nu)+ o(\nu^\frac{2}{3 } ) \biggr\ } { \nonumber}\\[2 mm ] { \raggedright}&= -2\omega \nu^{-\frac{1}{3}}c\left\{{-\hspace{-12pt}\int}_{-\infty}^{\infty } \frac{q(s,0)}{s}\,ds + o(\nu^\frac{2}{3})\right\}=\sigma^\frac{2}{3 } o(\nu^\frac{2}{3 } ) , \quad \nu \to 0 .
\end{aligned}\ ] ] thus , asymptotics ( [ gamma_a ] ) is valid for @xmath64 as well .
* using the substitution @xmath175 we compute asymptotics of @xmath134 in ( [ cp ] ) @xmath176 b_n & = \frac{(-1)^{n}}{n ! } \sqrt{\pi}\ , \frac{\gamma \left ( 3n+\frac{1}{2}\right ) } { 12^{n } } , \quad \nu \to \infty.\end{aligned}\ ] ] therefore , for asymptotics of @xmath68 we have @xmath177 next , we find asymptotics of ( [ p ] ) using repeated integration by parts : @xmath178 & = \frac{e^{-\nu \left ( x+ \frac{x^3}{3}\right)}}{\nu ( 1+x^{2})}\biggl ( 1-\frac{2x}{\nu ( 1+x^{2})^{2 } } + \frac{2(5x^{2}-1)}{\nu ^{2}(1+x^{2})^{4}}+o(\nu ^{-3})\biggr ) , \quad \nu \to \infty . \label{int_asy_inf}\end{aligned}\ ] ] this asymptotics is uniform in @xmath79 .
hence , from ( [ gamma1 ] ) , ( [ p ] ) , ( [ c_asy_inf ] ) , and ( [ int_asy_inf ] ) we obtain @xmath179 { \raggedright}&=\frac{\sigma^2}{8\omega^2 } \left ( 1-\frac{15}{16}\,\nu ^{-2}+o(\nu ^{-4})\right ) , \quad \nu \to \infty.\end{aligned}\ ] ] * the exponent in the integrand ( [ c- ] ) has a maximum at @xmath180 . therefore , laplace s method @xcite immediately gives @xmath181 we rewrite the lyapunov exponent as follows : @xmath182 & -2\omega \,c\int_{-\infty}^{\infty } \frac{x e^{\nu(\frac{x^3}{3}-x)}}{1+x^2}\ , dx\int_x^{\infty } e^{-\nu(\frac{t^3}{3}-t)}\ , dt .
\label{gamma41}\end{aligned}\ ] ] + observe that the maximum of the exponent in the second integral is attained at @xmath183 .
therefore , laplace s method applied to the double integral ( [ gamma41 ] ) ( see @xcite ) together with ( [ c- ] ) gives @xmath184 & = 2\omega \,c \frac{\pi}{2\nu}\,e^{\frac{4\nu}{3}}\ , ( 1+o(\nu^{-1 } ) ) = \omega ( 1+o(\nu^{-1})),\quad \nu \to \infty.\end{aligned}\ ] ] since the integrand of the first term in ( [ gamma41 ] ) vanishes at @xmath185 , the same approach leads to the asymptotics @xmath186 hence , in this case we have @xmath187
since we do not estimate the remainders in asymptotic formulas ( [ gamma0])-([gamma_neg ] ) , we compare the asymptotic formulas numerically with the exact ones given by ( [ gamma1])-([gamma- ] ) in order to determine the range of their validity .
figures [ pdf1 ] and [ pdf2 ] show the graphs of the normalized lyapunov exponent @xmath188 ( [ gamma1 ] ) as a function of @xmath189 for @xmath50 . as @xmath61 in figure [ pdf1 ]
the relative error of the asymptotic approximation of @xmath0 becomes less than @xmath190 when @xmath191 .
the relative error of asymptotic approximation ( [ gamma_h ] ) for large values of @xmath58 is less than @xmath192 if @xmath193 in figure [ pdf2 ] .
the graphs of the normalized lyapunov exponent @xmath188 ( [ gamma- ] ) as a function of @xmath189 for @xmath64 are shown in figures [ gamma0_neg ] and [ gamma1_neg ] . for small values of @xmath58 ( figure [ gamma0_neg ] )
the relative error of the asymptotic formula ( [ gamma0 ] ) becomes less than @xmath197 if @xmath191 .
the relative error of ( [ gamma_neg ] ) for large values of @xmath58 ( figure [ gamma1_neg ] ) is less than @xmath190 as soon as @xmath198 .
m. notomi , k. yamada , a. shinya , j. takahashi , c. takahashi , and i. yokohama , 2001 _ extremely large group - velocity dispersion of line - defect waveguides in photonic crystal slabs , _ phys .
87 ( 2001 ) 253902 . | we study the influence of disorder on propagation of waves in one - dimensional structures .
transmission properties of the schrdinger equation with the white noise potential can be expressed through the lyapunov exponent @xmath0 which we determine explicitly as a function of the noise intensity @xmath1 and the frequency @xmath2 .
we find uniform two - parameter asymptotic expressions for @xmath0 which allow us to evaluate @xmath0 for different relations between @xmath1 and @xmath2 . |
this work is supported in part by the national science council of r.o.c . under grant
no : nsc-97 - 2112-m-006 - 001-my3 .
v. barger _
et al . _ ,
b*580 * , 186 ( 2004 ) [ arxiv : hep - ph/0310073 ] ; v. barger _
et al . _ ,
b*596 * , 229 ( 2004 ) [ arxiv : hep - ph/0405108 ] ; v. barger _
et al . _ ,
b*598 * , 218 ( 2004 ) [ arxiv : hep - ph/0406126 ] ; k. cheung _ et al .
b*652 * , 285 ( 2007 ) [ arxiv : hep - ph/0604223 ] ; v. barger _
et al . _ ,
rev . d*80 * , 055008 ( 2009 ) [ arxiv:0902.4507 [ hep - ph ] ] ; arxiv:0906.3745 . c. h. chen and h. hatanaka , phys .
rev . d*73 * , 075003 ( 2006 ) [ arxiv : hep - ph/0602140 ] ; q. chang , x. q. li and y. d. yang , jhep * 0905 * , 056 ( 2009 ) [ arxiv:0903.0275 [ hep - ph ] ] ; x.
g. he and g. valencia , phys .
b*680 * , 72 ( 2009 ) [ arxiv:0907.4034 [ hep - ph ] ] ; c. w. chiang _ et al .
_ , arxiv:0910.2929 [ hep - ph ] ; c. w. chiang , r. h. li and c. d. lu , arxiv:0911.2399 [ hep - ph ] . | the evidence of a large cp phase has been shown by cdf and d collaborations in the time - dependent cp asymmetry ( cpa ) of @xmath0 decay , where the nonvanished cpa clearly implies the existence of a non - kobayashi - maskawa phase in @xmath1 transition .
we study the new phases originated by the extra family dependent @xmath2 model and examine their impact on @xmath3 or @xmath4 violating observables in @xmath5 decays with @xmath6 . adopting the constraints from the current data of @xmath7 and the branching ratios for @xmath8 and @xmath9 , we find that @xmath10 and the t - odd observables of above @xmath11 in the decay chain @xmath12 ) \ell^+ \ell^-$ ] can be achieved .
in addition , we demonstrate that the longitudinal and transverse polarizations of @xmath13 and the up - down asymmetry of @xmath14 in the same decay chain are also sensitive to the @xmath15-mediated effects .
although three generations of the standard model ( sm ) provide a unique cp violating phase in the cabibbo - kobayashi - maskawa ( ckm ) matrix @xcite to interpret the observed cp violating phenomena in @xmath14 and @xmath16 systems , due to the failure of the km phase in the explanation of matter - antimatter asymmetry , some new source of cp violation ( cpv ) is expected .
therefore , it is an important issue to probe and find the new cp violating effects . by the @xmath17 production in tevatron run
ii , cdf and d have observed the @xmath17 oscillation of @xmath18 ps@xmath19 @xcite and @xmath20 ps@xmath19 @xcite , respectively , and the large direct cp asymmetry ( cpa ) of @xmath21 for @xmath22 @xcite .
moreover , surprisingly both detectors have also shown an unexpected large cp phase in the mixing - induced cpa for @xmath23 with the average value given by @xcite ^j/_s=-2_s + ^np_s= -2[0.10 , 1.42][eq : phis ] at the @xmath24 confidence level ( cl ) , where @xmath25 @xcite is the cp phase of the sm while @xmath26 is a phase from new physics . if the measurement of eq .
( [ eq : phis ] ) indicates the existence of a new cp phase in @xmath1 transition @xcite , it is interesting to expose the effects in other @xmath1 related phenomena , such as the puzzles of cpas in @xmath27 and of polarizations in @xmath28 . nevertheless , as known that owing to the inevitable large uncertainty of nonperturbative qcd effects , it is difficult to definitely uncover the new cp phase from the cpas of the nonleptonic exclusive decays .
therefore , the best environment to look for new phase is that qcd effects are less involved and sm contributions are highly suppressed .
therefore , we propose that the large cp violating phase in @xmath1 transition can be directly probed by measuring t - odd observables in @xmath29 decays with @xmath30 . to understand the property of @xmath3 violating approach for probing the new cp phase , we briefly discuss the characters of cp - odd and t - odd observables . in a decay process
, the direct cpa or cp - odd observable is defined by @xmath31 , where @xmath32 ( @xmath33 ) is the partial decay rate of the ( cp - conjugate ) process . due to @xmath34
, it always needs weak cp phase @xmath35 and strong phase @xmath36 simultaneously to get the nonvanished cpa .
thus , the prediction efficiency on such cpa is mainly dictated by the uncertain calculations of strong phase .
another way to probe cp phase is through the spin - momentum triple correlation of @xmath37 @xcite , where @xmath38 is the spin carried by one of outgoing particles and @xmath39 and @xmath40 denote any two independent momentum vectors .
the triple momentum correlation is t - odd observable since it changes sign under the time reversal ( @xmath41 ) transformation of @xmath42 .
we note that @xmath43 is used to distinguish from real time - reversal transformation which also contains the interchange of initial and final states . by the cp@xmath44 invariant theorem , @xmath43 violation ( tv ) implies cpv .
therefore , study of the t - odd observable can also help us to understand the origin of cpv .
intriguingly , the triple correlation is proportional to @xmath45 .
this leads that the strong phase in t - odd observable is not necessary . to satisfy the criteria of a small qcd uncertainty with the high suppression in the sm for the @xmath1 transition mentioned earlier
, we find that tv in @xmath46 with polarized vector meson could provide a good chance to directly observe the new phase revealed by cdf and d . to search for the new cp phase in the extension of the sm , a model independent analysis is not suitable for the current study .
in this paper we investigate the minimal extension in the gauge sector by including one family dependent @xmath2 to the sm gauge symmetry , i.e. the considered gauge symmetry is @xmath47 .
although lots of extensions of the sm predict additional @xmath2 gauge symmetry , such as string theory @xcite , grand unified theories @xcite , little higgs models @xcite , light u - boson model @xcite , etc , a well - motivated @xmath15 model for low energy systems is the so - called nonuniversal @xmath15 .
besides the @xmath2 charges carried by the fermions are family dependent @xcite , the model also leads to flavor changing neutral currents ( fcncs ) at tree level and provides the solution to the puzzles in @xmath16 physics @xcite .
the recent progress of the nonuniversal @xmath15 on the phenomenological applications could be referred to refs .
@xcite .
we start to set up the relevant interactions with the new @xmath15 gauge particle . following the convention in refs .
@xcite , we write the couplings of the @xmath15-boson to fermions as ^z&= & -g_2 j^f_z z _ , with j^f_z&= & |f ^ f , [ eq : jf ] where @xmath48 and the indexes @xmath49 ( i=1,2,3 ) denote the same type of quarks ( qs ) or charged leptons ( @xmath50 ) , @xmath51 are diagonal @xmath52 matrices in flavor space with @xmath53 and are represented by @xmath54 .
note that @xmath55 could be different for different generations in the nonuniversal @xmath15 model .
hereafter , @xmath55 are taken as free parameters .
although the third component of @xmath56 and @xmath57 in the sm also provide two neutral gauge bosons , we will not show them explicitly since they involve only flavor conserved processes and are well known . to obtain the interactions in physical states
, we introduce @xmath58 to diagonalize the associated yukawa matrix of fermion . in terms of mass eigenstates , eq . ( [ eq : jf ] ) could be written as j^f_z&= & |f ^ f , [ eq : intzp ] + l^f & = & v^l_f v^l_f , r^f = v^r_f v^r_f .
due to the unequal elements in @xmath59 , obviously , flavor misaligned effects in @xmath60 and @xmath61 will lead to fcncs at tree level . in order to display the new physics effects with more compact form in the processes , we use vector and axial - vector currents instead chiral currents . thus , eq . ( [ eq : intzp ] ) is rewritten by j^f_z&= & |f ^ f , + & = & + , = - .
consequently , the @xmath15-mediated interactions for @xmath1 are given by ^z_bs & = & -g_2 |s ^ ( c^q_vsb + c^q_asb _ 5 ) b z _ + h.c .
[ eq : int_bs ] for semileptonic decays , the couplings to the leptons are taken as ^z _ & = & -g_2 |^ ( c^_v + c^_a _ 5 ) z_. [ eq : int_ll ] since we will concentrate on the @xmath62-pair production , to avoid the problem on the lepton flavor violation and nonuniversality and simplify our analysis , we assume that @xmath63 are the same for different families , i.e. @xmath64 . in the following analysis
, we will show that the impacts of @xmath15-mediated fcncs on the processes of @xmath17 mixing , @xmath65 and @xmath5 .
note that since we do nt have any information on @xmath66 except the ckm matrix defined by @xmath67 , the elements in @xmath68 should be regarded as free parameters . therefore , the parameters associated with the processes for @xmath69 and @xmath70 transitions could be independent of those associated with the @xmath1 transition .
based on the new effects , we now formulate the related phenomena for @xmath71 and @xmath72 processes . in terms of eq .
( [ eq : int_bs ] ) , the tree induced effective hamiltonian for @xmath71 is read by ^z_b=2= ^2 , where @xmath73 , @xmath74 is the gauge coupling of @xmath56 and @xmath75 . including the wick contracting factor and using the matrix elements parametrized by |b | |s _ p _ b
|s ^ p_b | b & = & m_b f^2_b _ b , + & = & m_b _ b f^2_b . combining the result with the sm contribution , the transition matrix element for @xmath76 is given by m^s_12&= & a^sm_12 e^-2i_s + a^z_12 e^2i(^z_s-_s ) , [ eq : ms12 ] where @xmath77 and the cp phase of @xmath78 is parametrized by @xmath79 @xcite . as a result
, the @xmath17 mixing is obtained as m_s = 2|m^s_12|=m^sm_s ( 1 + 2r2^z_s + |r|^2)^1/2 with @xmath80 . due to @xmath81 in the b - system
, the time - dependent cpa could be simplified to be -s_j/ & & i m ( ) = ( 2_s - ^z_s ) , + ^z_s&= & ( ) . for @xmath72 processes , we focus on the important decay mode of @xmath82 . from eqs .
( [ eq : int_bs ] ) and ( [ eq : int_ll ] ) , the corresponding effective interactions can be found to be ^z_bs & = & |s _ ( c^q_vsb + c^q_asb_5 )
b |^ ( c^_v + c^_a_5 ) .
[ eq : bsll_int ] the direct application of eq .
( [ eq : bsll_int ] ) is the chiral suppressed decay of @xmath83 .
using @xmath84 and including the sm contributions , the branching ratio ( br ) for @xmath85 is expressed by ( |b_s^+ ^-)=_b_s m_b_sf^2_b_sm^2 _
( 1-)^1/2|a _ |^2 with a _ & = & y(x_t ) + c^q_asb c^_a , + & = & ( ) ^2 , where @xmath86 , @xmath87 is the weinberg angle , @xmath88 and @xmath89 with @xmath90 @xcite . to study the @xmath3 violating effects in @xmath91 decays , besides the new interactions of eq .
( [ eq : bsll_int ] ) , we need the hadronic effects for the @xmath92 transition . as usual ,
the form factors associated with various current operators are parametrized as @xmath93 with @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 and @xmath99 . here
, we have used @xmath16 ( @xmath100 ) instead of @xmath101 ( @xmath102 ) and @xmath103 .
hereafter , we will adopt the notation for shorthand .
hence , the transition amplitude for @xmath104 ( @xmath105 ) with the polarized @xmath100 can be arranged by @xcite @xmath106 where @xmath107 , @xmath108 , subscript @xmath109 $ ] while @xmath110 @xmath111 $ ] ( @xmath112 , @xmath113 , @xmath114 ) and their explicit expressions are h_1 & = & + c_7t_1 , + h_2 & = & ( m_b+m_v)a_1 pq c_7 t_2 , + h_3 & = & + c_7(t_2+t_3 ) , + g_1&= & , + g_2 & = & ( m_b+m_v)a_1 , + g_3 & = & .
[ eq : ampk1 ] here , we have suppressed the @xmath115-dependence in the form factors , @xmath116 are the wilson coefficients of the sm and their specific results could be referred to ref .
@xcite and @xmath117 . to obtain
the @xmath3 violating obervables , which are connected to the polarizations of the vector meson in @xmath118 decays , we have to consider the decay chain @xmath119 in which @xmath120 is @xmath121 as @xmath122 . for calculations
, we set the positron lepton momentum as @xmath123 in the @xmath115 rest frame and the k momentum @xmath124 in the @xmath100 rest frame . with some tedious algebra ,
the differential decay rate with polarized v - meson can be formulated by @xcite & = & ( k^*[]k ) + & & \{16 ^ 2_k_a=1,2| m_a^0| ^2 + 8 ^ 2_k_a=1,2 ( | m_a^+|^2+| m_a^-|^2 ) + & & -8 ^ 2_k + & & + 32_k } [ eq : difangle ]
where @xmath125 , @xmath126 is the polar angle of k - meson while the @xmath127 is the relative angle of decaying planes of @xmath128 and @xmath129 , @xmath130 and @xmath131 denote the longitudinal and transverse polarizations of @xmath100 respectively and they are given as follows : @xmath132 with @xmath133 and @xmath134 .
since we have integrated out the angles associated with the lepton , some important effect such as the forward - backward asymmetry ( fba ) of the lepton does not show up in eq .
( [ eq : difangle ] ) .
however , we find an up - down ( ud ) asymmetry of the @xmath14-meson , which is associated with @xmath135 and similar to the lepton fba , could be generated . according to eq .
( [ eq : difangle ] ) , it is clear that besides the real pieces , the imaginary parts that represent the tv ( or cpv ) also appear in the angular distribution .
if we can find the proper @xmath136-odd operators denoted by @xmath137 , in terms of the concept of quantum mechanics , the average of the @xmath136-odd operator can be written as _ t_i = d_k o_t_i sign(_i ) where sign(@xmath138 )
is chosen to get a nonvanished asymmetric observables when integrating the angles of phase space .
then , the @xmath136-odd observable could be defined by _
t_i&= & .
accordingly , we find that the two @xmath3 violating terms in eq .
( [ eq : difangle ] ) can be extracted by using @xmath136-odd momentum correlations , defined by _
t_1&= & 4 | _ b| , + & = & 22_k , + o_t_2&= & = _ k where the momenta of particles are taken in the @xmath100-meson reset frame , @xmath139 and @xmath140 . as a result
, we get _ t_1&= & - , [ eq : at1 ] + a_t_2&= & .
[ eq : at2 ] although our main purpose is to explore the existence of new cp phase , the @xmath15-mediated effects also have significant contributions to cp - conserved quantities .
therefore , besides the physical observables defined in eqs .
( [ eq : at1 ] ) and ( [ eq : at2 ] ) , we could also invent interesting measurable quantities , such as longitudinal and transverse polarizations of the @xmath100-meson and the angular asymmetry of the @xmath14-meson , etc @xcite . according to the definition in ref .
@xcite , the longitudinal and transverse polarizations of @xmath100-meson appearing in eq .
( [ eq : difangle ] ) can be found as _
l(q^2)&= & , [ eq : pl ] + p_t(q^2)&= & .
[ eq : pt ] in addition , due to the appearance of @xmath141 in eq .
( [ eq : difangle ] ) , the ud asymmetry of the @xmath14-meson can be obtained as @xcite ^k_ud&= & .
[ eq : uda ] so far we have introduced seven independent free parameters of @xmath142 , @xmath143 and @xmath144 and @xmath145 and @xmath146 in our analysis . in order to further reduce the number of free parameters to simplify the phenomenological analysis , we set @xmath147 and @xmath148tev and force @xmath149
. although the restricted condition of the latter is unnecessary , it could correspond to a special scenario of yukawa matrices , e.g. hermitian yukawa matrices which result in @xmath150 @xcite . for numerical estimations , we need the data to constrain the remaining five parameters .
therefore , we adopt the values of @xmath151 ps@xmath19 @xcite , @xmath1520}\mu^{+ } \mu^{-})= ( 0.45^{+0.12}_{-0.10})[1.05^{+0.15}_{-0.13}]\times 10^{-6}$ ] @xcite and @xmath153 @xcite as the inputs whereas the sm predictions are taken as @xmath154 ps@xmath19 @xcite , @xmath1520}\mu^{+ } \mu^{-})_{\rm sm}= 0.58[1.3]\times 10^{-6}$ ] @xcite and @xmath155 @xcite .
since @xmath156 is very similar to @xmath157 , we use the latter as the delegate .
we now present the numerical results . in order to realize the strong correlation among @xmath158 , @xmath7 , @xmath159 and @xmath1600}\ell^{+ } \ell^{-})$ ] influenced by the same parameters , we show the observables versus @xmath7 in fig .
[ fig : phis ] . here ,
for the @xmath161 transition form factors , we employ the results calculated by light - cone sum rules ( lcsrs ) @xcite . by the figure
, we see clearly that besides the br of @xmath65 can be enhanced to be close to @xmath162 , the large @xmath163 generated by @xmath15-mediated effects can be consistent with the current measurements of d and cdf . , @xmath164 , @xmath165 , @xmath166 ) and @xmath7 by the influence of nonuniversal @xmath15.,width=4 ] with the values of parameters which are constrained by the current data of @xmath7 , @xmath167 and @xmath1680}\ell^{+ } \ell^{-})$ ] , the predicted t - odd observables of @xmath169
are displayed in fig .
[ fig : tv ] .
it is clear that we can obtain @xmath170 of @xmath11 and @xmath171 of @xmath172 at large and small @xmath115 , respectively . in @xmath173 as a function of @xmath115 in nonuniversal @xmath15 model.,width=4 ] moreover , with the same set of values of parameters , the predicted longitudinal [ transverse ] polarization and the ud asymmetry defined in eq .
( [ eq : pl ] ) [ eq .
( [ eq : pt ] ) ] and eq .
( [ eq : uda ] ) are given in fig .
[ fig : ud](a ) and ( b ) , respectively , where the solid and dashed lines denote the sm predictions .
we find that the deviation of 10% from the sm value in @xmath174}$ ] can be achieved in the @xmath15 models .
additionally , the crossing of @xmath175 and @xmath176 at large @xmath115 is also sensitive to the new physics effects . besides a large deviation from the sm result in the spectrum of @xmath177
could be accomplished , the zero point of @xmath177 is also sensitive to the @xmath15-mediated effects .
polarization @xmath178}$ ] of @xmath13 and ( b ) up - down asymmetry of k in @xmath179 , where solid and dashed lines stand for the sm predictions . ,
width=4 ] in summary , we have studied the @xmath15-mediated effects on @xmath180 , @xmath158 , @xmath181 and @xmath182 in the family dependent @xmath2 model . with the constraints of the current data , we have found that in the nonuniversal @xmath15 model , the time - dependent cp violating phase of the @xmath17 oscillation could be @xmath10 and the t - odd observables @xmath170 and @xmath171 could be @xmath11 and @xmath172 at the large and small @xmath115 regions , respectively .
in addition , we have also shown that the longitudinal and transverse polarizations of @xmath13 and the ud asymmetry of @xmath14 in the decay chain @xmath183 could be significantly influenced by the @xmath15 effects . |
globular clusters have long been noted to deviate , sometimes substantially , from perfect spherical symmetry @xcite . what is the reason why many globular clusters are non - spherical ?
a conclusive answer to such an important question in the study of globular cluster dynamics , with potentially important ramifications in the study of the evolution of globular cluster stars @xcite , has so far not been conclusively established ( see * ? ? ? * for an extensive review and references ) .
several possible causes of deviations of globular cluster shapes from sphericity have been suggested , including tidal stresses due to the presence of strong galactic tidal fields @xcite , gravothermal shocks during the passage of a cluster through the disk @xcite , velocity anisotropies , and internal rotation ( which may also be acquired by interaction with a massive galaxy ; lee et al .
2004 and references therein ) .
in particular , the former two phenomena are now well documented , with a particularly impressive example being given by palomar 5 @xcite .
the possibility that rotation may be the main driver of cluster ellipticities @xcite is supported by the @xcite and @xcite studies of @xmath3 centauri ( ngc 5139 ) , which are based upon a detailed analysis of its structural and dynamical profiles .
yet , it still remains unclear , both from an empirical and from a theoretical perspective , whether rotation can be the main cause of flattening of globular star clusters . on the observational side
, we currently lack detailed velocity profiles for most globular clusters ; and even though we do have detailed rotational information for a few clusters , such as 47 tucanae ( ngc 104 ) , @xmath3 cen , m4 ( ngc 6121 ) and m15 ( ngc 7078 ) @xcite , the lack of a wider sample , as well as of reliable measurements of the variation in flattening as a function of radius for clusters encompassing the full range of observed ellipticities , has hindered further progress in establishing the general correspondence , or lack thereof , between flattening and cluster rotation .
indeed , the latest catalogs of globular cluster ellipticities , published by @xcite and by @xcite , are entirely based on photographic data , consider exclusively the outermost cluster regions , and , relying on visual data , can be severely affected by differential reddening @xcite .
our group is currently building a new , detailed catalog of globular cluster ellipticities , based on 2mass ( near - infrared ) data , which will greatly improve on the current situation @xcite .
theoretically , too , the case for rotation as one of the main drivers of globular cluster ellipticity is far from being settled . in particular , @xcite has recently argued that globular clusters can rotate to very fast rates without deviating substantially from sphericity .
this result is in contrast with those from @xcite , who instead were unable to produce models of fast - rotating globular cluster - like structures _ without _ an accompanying deviation from sphericity , in the sense that their modeled structures became lengthened along one or two axes , depending on the initial anisotropy of the velocity distributions .
an ideal test , from an empirical viewpoint , of the impact of rotation on the shapes of globular star clusters , is to measure the rotation velocity of a highly flattened , isolated extragalactic globular cluster , where there is little possibility that tidal effects and/or gravothermal shocks may be responsible for its flattening .
the closest possible approximation to such an isolated , flattened system is represented by the single , old globular cluster in the wolf - lundmark - melotte ( wlm ) dwarf irregular galaxy ( ddo 221 ) .
while distant ( 885 kpc ; * ? ? ?
* ) , the mass of the wlm galaxy is much smaller than even the lmc or smc , with @xmath4 of @xmath5 , @xmath6 , and @xmath7 , respectively ( van den bergh 2000 ) , thus effectively ruling out the possibility that the non - spherical shape of the wlm cluster ( hereafter wlm-1 ) can be due to tidal effects or gravothermal shocks .
note , in this sense , that this is not necessarily the case even in the magellanic clouds @xcite , where non - spherical globular clusters of all ages can be found ( * ? ? ? * and references therein ) . in this paper
we begin by reanalyzing archival hst wfpc2 data in section [ sec : hstobservations ] , where we measure the ellipticity and surface brightness profiles of wlm-1 .
we then discuss our new vlt uves observations and their reduction in section [ sec : vltobservations ] , and attempt to measure the rotation ( section [ sec : rotation ] ) and velocity dispersion ( section [ sec : veldisp ] ) of wlm-1 .
we conclude with a discussion of the possibility of velocity anisotropy in wlm-1 as a possible explanation for its observed morphology , and give some final remarks in section [ sec : conclusions ] .
in order to measure accurate ellipticity and surface brightness profiles of wlm-1 , we have downloaded calibrated hst wfpc2 f814w and f555w images from the hst archive ( program go 6813 , pi p. w. hodge ) . in these images
wlm-1 is centered on the pc chip .
the only processing required was to combine the two images in each filter to reject cosmic rays .
we also combined the images from different filters to reject the remaining few cosmic rays .
this final image , rotated to put north at the top , is shown in figure [ fig : wlm-1 ] .
the ellipticity and position angle as a function of semi - major axis distance are measured with the image reduction and analysis facility ( iraf ) routine ellipse , where @xmath8 , and @xmath9 and @xmath10 are the semi - major and semi - minor axis lengths respectively .
the position angle is the angle ( @xmath11 ) made by the major axis measured degrees east of north .
the results are illustrated in figure [ fig : ellipticity ] .
the mean ellipticities and position angles for our spectral extraction regions are listed in table [ tab : ellipticity ] .
lcc 0.1 0.5 & @xmath12 & @xmath13 + 0.5 1.0 & @xmath14 & @xmath15 + 1.0 1.5 & @xmath16 & @xmath17 + 1.5 5.0 & @xmath18 & @xmath19 [ tab : ellipticity ] note that since there are many resolved stars in the hst image , there are many irregularities in the measurement of the ellipticity and position angle , especially at larger radii where these stars contribute a significant fraction of the total flux ( see discussion in navarro et al .
2005 ) . also , when the ellipticity gets close to zero ( nearly circular ) , the position angle becomes poorly constrained and shows large fluctuations .
the mean ellipticity of wlm-1 in the region @xmath20 is @xmath21 .
this ranks among the highest values among galactic globulars according to the lists tabulated by @xcite and @xcite , and only @xmath22 galactic globular clusters have ellipticities greater than this according to the recent analysis of 2mass data by navarro et al .
it has previously been noted that lmc and smc globular clusters of all ages also have significant ellipticities , and in fact tend to be more elliptical than either galactic or m31 clusters ( * ? ? ?
* ; * ? ? ?
* and references therein ) , suggesting an inverse relationship between galaxy mass and globular cluster ellipticity which would appear to be broadly consistent with the case of wlm-1 .
however , @xcite have recently shown that the giant elliptical galaxy ngc 5128 , with @xmath23 mag @xcite , also contains a sizable population of flattened clusters .
it is also interesting to note that @xcite have suggested that , while galactic and m31 globular clusters shapes are broadly consistent with oblate spheroids , lmc and smc clusters are instead more consistent with triaxial ellipsoids .
as well known , triaxiality is often accompanied by isophote twisting @xcite . in the case of wlm-1 ,
we do find some interesting evidence that isophote twisting may indeed be present , but this is clearly restricted to the innermost @xmath24 of the cluster ( fig .
[ fig : ellipticity ] ) .
we have independently measured the surface brightness profile on hst wfpc2 f555w ( @xmath25 ) and f814w ( @xmath26 ) images obtained from the hst archive .
using the iraf ellipse task , with the ellipticity and position angle fixed at the mean values of @xmath27 and @xmath28 respectively , we measured the average flux in each elliptical annulus , which were then converted to surface brightness using the photometric zero points ( zp ) from the hst manual [ @xmath29 , @xmath30 .
the data from the f555w filter are shown as points in figure [ fig : sbprofile ] .
finally we fit the surface brightness measurements with @xcite profiles using a chi - squared minimization algorithm and using the rms scatter in each annulus as an estimate of the measurement errors ( although this is actually a measure of the true range of flux values at any particular semi - major axis distance rather than a measure of how accurately one can measure the surface brightness ) .
pixels within @xmath31 of the center of the cluster are saturated , and were therefore excluded from the fit .
these excluded points are indicated by open circles in figure [ fig : sbprofile ] .
the resulting fits for each filter were then averaged together to give the final values .
we find a best - fit core radius of @xmath32 , and a tidal radius of @xmath33 , where the magnitude of the errors is driven by the large scatter of fluxes in each annulus ( due to the resolved nature of the cluster population ) .
the fitting also takes into account the background surface brightness ; the best fit values are @xmath34 mag / arcsec@xmath35 in f555w and @xmath36 mag / arcsec@xmath35 in the f814w filter .
note that we have _ not _ taken into account the finite size of the hst psf , which has a nominal value of less than @xmath37 in both filters .
assuming @xmath38 @xcite , one arcsecond corresponds to 4.28 parsecs at the distance of wlm-1 , and our measurements of the core and tidal radii convert to @xmath39 pc and @xmath40 pc .
both of these values fall squarely in the range of parameters exhibited by galactic globulars : @xmath41 and @xmath42 with median values of @xmath43 pc and @xmath44 pc @xcite .
this implies a very compact globular cluster , with @xmath45 . we can not discard the possibility of a collapsed core
note that our measured half - light radius for the wlm-1 cluster , along with its @xmath46 ( sec . [ sec : veldispest ] ) , are not inconsistent with the trend between @xmath47 and @xmath4 for globular clusters in dwarf galaxies and in the outer galactic halo suggested by ( * ? ? ?
* their fig .
the structural parameters of wlm-1 have been previously derived using starcounts by @xcite who found a core radius of @xmath48 ( @xmath49 pc ) and a tidal radius of @xmath50 ( @xmath51 pc ) .
a comparison of the background - subtracted star count profile and background - subtracted surface brightness profile ( fig .
[ fig : sbprofile ] ) yields a very good match down to @xmath52 , where the star counts start to become incomplete . at @xmath53
the two profiles diverge ; the star counts turn over while the surface brightness profile continues to rise down to @xmath54 , at which point the ccd becomes saturated .
we speculate that this incompleteness in the star counts is the source of hodge s large parameter values .
other quantities which will be of use later in this paper include the half - light radius @xmath47 , the central @xmath55-band surface brightness @xmath56 , and the mean surface brightness inside the half - light radius @xmath57 .
we estimate these parameters from the best - fit king model to the f555w ( @xmath55 ) band hst data .
we find : @xmath58 arcseconds ( @xmath59 pc ) , @xmath60 mag / arcsec@xmath35 , and @xmath61 mag / arcsec@xmath35 .
the new data analyzed in this paper were obtained with the uv - visual echelle spectrograph ( uves ) on the 8.2 meter vlt ut2 ( kueyen ) telescope .
uves is a two - arm cross - dispersed echelle spectrograph with a dichroic splitting the red and blue components of the light .
the blue light is recorded on a single 2k @xmath62 4k eev ccd , and the red light on a mosaic of an eev ccd and an mit / ll ccd for the reddest orders . in order to obtain the highest spatial resolution we used a @xmath24 wide slit , which allows us to obtain velocity information on a similar scale . while slightly undersampling the spectral information , we chose to bin the ccds @xmath63 to improve the signal - to - noise ratio in the faint outer parts of the cluster .
this binning yields a plate scale of @xmath64/pixel in the blue and @xmath65/pixel in the red .
the blue slit has a length of @xmath66 ( @xmath67 pixels ) and the red slit @xmath68 ( @xmath69 pixels ) . the wavelength coverage and resolution of this configuration
is listed in table [ tab : wavelengths ] .
wlm-1 ( 00:01:49.5 , @xmath70:27:30.7 , j2000 ) was observed for 3060 seconds in each of two slit positions , one aligned with the major axis ( position angle @xmath71 ) , and another aligned along the minor axis ( @xmath72 ) .
an hst image of the cluster @xcite with the slit orientations indicated is shown in figure [ fig : wlm-1 ] .
ccccc blue & 3275 & 4509 & 0.034 & 56,700 +
red - l & 4610 & 5607 & 0.038 & 67,220 + red - u & 5682 & 6650 & 0.046 & 66,450 a / pixel b2-pixel resolution [ tab : wavelengths ]
the data reduction was carried out with iraf .
each of the three uves ccds ( blue , red - l , red - u ; see table [ tab : wavelengths ] for wavelength coverages ) were reduced independently , beginning with overscan , bias subtraction , and removal of the scattered inter - order light . before dividing by the flat field ,
any variations with scales larger than a few pixels were removed from the illuminated regions by fitting and dividing by a high - order spline3 fit using the apflatten task . the resulting `` flattened '' flat field was then normalized to unity and divided into the science frames .
lastly we located and interpolated over cosmic rays and bad pixels in the science frames using the iraf median filtering task crmedian .
the center of the cluster was traced in all orders , and then extracted using the apertures discussed in section [ sec : extractions ] .
the calibration lamp spectrum and flat field were extracted using identical traces and extraction apertures as the cluster to insure the most accurate wavelength calibration .
the extracted science spectra were then divided by the normalized extracted flat field to remove the blaze function and any wiggles due to absorption or emission features in the flat field .
finally we removed any large - scale problems from the extracted spectra , such as bad columns or cosmic ray strikes , using linear interpolation from 100 pixels on either side of the blemish .
the wavelength solution was derived from a thorium - argon calibration lamp spectrum .
the solutions have an rms scatter of @xmath73 , and use @xmath74 , @xmath75 , and @xmath76 lines in each of the blue , red - l , and red - u wavelength regions respectively .
note that we rederived the wavelength solution for each extraction .
we performed one last cleaning of any remaining cosmic rays in the spectra by fitting by a cubic spline of high enough order to fit the absorption features , and iteratively replacing @xmath77 and @xmath78 outliers with the fit .
finally we trimmed the noise - dominated ends of each order , and combined the orders into a single spectrum the key to measuring a rotation velocity in data such as these , where a highly concentrated object is barely resolved , is the careful analysis of the light coming from the _ edges _ of the cluster . while the center of the cluster contains most of the light , and hence has the highest signal - to - noise ratio , the slit of the spectrograph is @xmath24 wide , thus we can obtain no velocity information in the central @xmath24 .
the seeing , which ranged from @xmath79 to @xmath80 , and the coarseness of our binned pixels ( @xmath64/pixel in the blue and @xmath65/pixel in the red ) also blur the spatial resolution .
we must therefore concentrate on the outer regions of the cluster .
however , the cluster is very compact , with a core radius of @xmath81 ( sec . [
sec : sbprofile ] ) , and the light profile plummets as one moves away from the center ; the cluster light fades into the background only @xmath82 from the center .
we have extracted five different regions of various sizes and radii in hopes of obtaining the best measure of the cluster rotation profile .
since the cluster is fairly elliptical ( @xmath83 in this region ) we scale the minor - axis extraction apertures to insure that we consistently measure the same physical radius of the cluster .
the extraction regions are listed in table [ tab : extractions ] .
the first region includes the entire half of the cluster , and is meant to give a luminosity - weighted rotation estimate .
however , as mentioned before , due to the width of the spectrograph slit , the innermost regions will include an equal mix of approaching and receding stars , so the second extraction region excludes the inner @xmath24 .
the final three extraction regions are very narrow , and will be used to try to estimate the rotation profile of the cluster .
ccccc @xmath84 & 0.003.05 & 0.004.12 & 0.002.56 & 0.003.46 + @xmath85 & 1.023.05 & 1.374.12 & 0.852.56 & 1.153.46 + @xmath86 & 0.001.02 & 0.001.37 & 0.000.85 & 0.001.15 + @xmath87 & 1.022.03 & 1.372.75 & 0.851.71 & 1.152.31 + @xmath88 & 2.033.05 & 2.754.12 & 1.712.56 & 2.313.46 + background & none & 7.2013.19 & none & 6.0411.10 1extraction regions in pixels .
[ tab : extractions ] when possible the extractions use background subtraction , where the background is calculated as the median in two regions on either side of the cluster .
these background regions are scaled according to the ellipticity for the minor axes extractions .
note that the slit is too narrow for background subtraction on the blue chip ; however , as there are no significant sky lines in the blue , this should not affect the results .
one should note that the wlm-1 cluster is relatively large .
the tidal radius is @xmath89 ( sec . [ sec : sbprofile ] ) , while the slits used in this paper to obtain the spectra are only @xmath66 and @xmath68 long .
thus the _ entire _ slit has contributions from cluster stars , and any attempt at background subtraction will also subtract cluster light . we plot the three narrow spectral extractions along the major axis in figure [ fig : spectra ] .
this very small piece of the total spectrum illustrates the variation in signal - to - noise ratio with each extraction and the abundance of absorption features available for measuring velocities .
terrestrial atmospheric absorption lines are noticeable in our red - u spectra .
two bands are especially strong : h@xmath90o at 58806020 , and o@xmath90 at 62006360 .
we did not observe atmospheric standards , and therefore we have no way to accurately correct for this absorption .
however , if not removed these absorption lines will bias our red - u derived rotation velocity toward zero since they are at exactly the same wavelengths on both sides of the cluster . to remove the telluric absorption lines we simply set any such absorption to the continuum value .
we start with a high - resolution atmospheric transmission atlas , smooth it to our resolution , and create a `` telluric mask '' for our spectra .
this mask has regions of high transmission set to one , and absorption regions , where the transmission is less than 95% , are set to a very large number . before cross - correlating the spectra , but after continuum subtraction , we divide by this telluric mask , and effectively set all significant regions of atmospheric absorption to the continuum level .
we thus lose any velocity information about wlm-1 which may have been contained in these masked regions , but the cross - correlation will no longer be biased by these strong atmospheric features .
a region of telluric absorption is shown in figure [ fig : telluric ] before and after masking .
the goal of these observations is to unambiguously measure the rotation velocity of wlm-1
. one method of getting an estimate of the magnitude of this rotation would be to follow the original theoretical calculations of @xcite for the ellipticity of a rotating cluster , which is equivalent to assuming that the cluster is a rotating fluid body . assuming a distance modulus of @xmath38 @xcite , which corresponds to 4.28 pc / arcsec , we find that the rotation velocity of wlm-1 should be of the order @xmath91 where @xmath92 is the effective density calculated for non - homogeneous clusters @xcite
. thus if this cluster is indeed elliptical as a result of rotation , we might expect to measure a difference of @xmath93 km s@xmath2 along the major axis , and zero along the minor axis ( see sec .
[ sec : veldisp ] for a more detailed estimate of the expected rotation ) .
while our spectra are not of high signal - to - noise ratio , especially in the very narrow extractions , we do have a reasonably large wavelength coverage .
thus while measuring the position of a single line yields a relatively large error , combining the results for the hundreds of lines present across the spectra increases the precision greatly . one way to simultaneously compare the positions of all features in two spectra is cross - correlation .
we use the fourier cross - correlation technique of @xcite , which has been implemented in the iraf fxcor task .
this routine computes the height , width , and location of the cross - correlation peak of two spectra . for each axis
we correlate the north / south ( major ) and east / west ( minor ) extractions which lie at equal radii .
the results of these correlations for the narrow extractions along both the major and minor axes are shown in figure [ fig : xcor ] .
the center of the peak is determined by fitting the top 5 points with gaussian functions .
this yields a direct measure of the velocity shift between the two , which is equal to twice the cluster rotation velocity at that distance .
we start with the reduced , extracted spectra from each chip . in preparation for the cross - correlation
we subtract the continuum by fitting with a cubic spline .
the telluric absorption is removed from the red - u spectra following the procedure outlined in section [ sec : telluric ] .
we then cross - correlate the spectra from each ( blue , red - l , red - u ) chip .
since the reductions , extractions , and wavelength calibrations are completely separate for each chip , this should give three independent , although weaker , measures of the cross - correlation velocity for each extraction .
the results of this piecewise cross - correlation are listed in table [ tab : piecewisexcor ] for all symmetric extractions , and allow one to compare the quality of the correlation for each chip .
we then combine the spectra from all the chips , and cross - correlate the result , excluding the @xmath94 gaps between the chips .
the results of correlating the combined spectra are listed in table [ tab : xcor ] , and shown in figure [ fig : rotation ] .
rather than only make one cross - correlation measurement to obtain the velocity at each radius , we can also cross - correlate all possible permutations , and find the velocities which are most consistent with all correlations . in other words ,
cross - correlating the spectrum extracted from @xmath95 and @xmath96 will give a measure of @xmath97 .
however , one can also correlate spectra from different radii , @xmath95 and @xmath98 , to obtain the relative velocity difference @xmath99 . using all six narrow extractions yields 15 correlation permutations .
thus we have 15 measurements which depend on only three quantities , assuming that we should find equal but opposite velocities at the same radius on either side of the cluster .
we solve for the most probable velocities by minimizing the sum of the squares of the deviation of each correlation .
thus instead of basing a rotation velocity on only one measurement of the cross - correlation of two spectra at equal distances from the cluster center , we measure 15 correlations between all 6 spectral extractions and find the three velocities which are in best agreement with all measurements .
these `` best fit '' rotation velocities are given as a function of semi - major axis distance in table [ tab : vrot ] .
we find no statistically significant rotation along the line of sight at any distance along either axis of wlm-1 .
cccc major - axis & & & + blue & @xmath100 & @xmath101 & @xmath102 + red - l & @xmath103 & @xmath104 & @xmath105 + red - u & @xmath106 & @xmath107 & @xmath108 + combined & @xmath109 & @xmath110 & @xmath111 + minor - axis & & & + blue & @xmath112 & @xmath113 & @xmath114 + red - l & @xmath115 & @xmath116 & @xmath117 + red - u & @xmath118 & @xmath119 & @xmath120 + combined & @xmath121 & @xmath122 & @xmath123 avelocities in km s@xmath2 . [
tab : piecewisexcor ] ccccc @xmath84 & 0.68 & @xmath124 & 0.67 & @xmath125 + @xmath85 & 0.26 & @xmath126 & 0.28 & @xmath127 + @xmath86 & 0.70 & @xmath109 & 0.69 & @xmath121 + @xmath87 & 0.23 & @xmath110 & 0.27 & @xmath122 + @xmath88 & 0.16 & @xmath111 & 0.16 & @xmath123 avelocities in km s@xmath2 .
[ tab : xcor ] ccc @xmath86 & @xmath128 & @xmath129 + @xmath87 & @xmath130 & @xmath131 + @xmath88 & @xmath132 & @xmath133 avelocities in km s@xmath2 .
[ tab : vrot ] in order to estimate the robustness of our results we have performed various cross - correlation simulations .
all of these tests start out with a base spectrum which is the average of the major- and minor - axis integrated spectra of wlm-1 . to test the dependence of the cross - correlation velocity on the signal - to - noise ratio of the spectra we create pairs of spectra at various signal levels by adding random noise to two copies of the base spectrum , treating them as opposite sides of the cluster .
we then remove the telluric absorption and shift one of the spectra by 5 km s@xmath2 .
the pair of spectra are then cross - correlated as described in section [ sec : rotation ] to obtain their relative velocity .
the results of these tests are illustrated in figure [ fig : sims ] , along with the height and fwhm of the cross - correlation peak .
this figure shows no systematic trends with signal - to - noise ratio , and verifies that the velocity errors returned by iraf are indeed valid . from this analysis
we estimate that the errors associated with the measurement of the cross - correlation fwhm are approximately three times the reported velocity errors , and have plotted them as such in figure [ fig : sims ] . to estimate the errors of the @xmath134-minimization technique used to determine the `` best '' rotation velocity we ran 50 simulations consisting of six copies of the integrated spectrum of wlm-1 to mimic the six @xmath24 extractions .
each of the six spectra is shifted such that the simulated cluster has a monotonically increasing rotation velocity with radius , and equal but opposite rotation velocities on opposite sides of the cluster .
specifically , the spectra were shifted by @xmath135 , @xmath136 , @xmath137 , + 1 , + 2 , and + 3 km s@xmath2 each .
noise was added to each spectrum so that the cross - correlation height measured in the simulation ( fig .
[ fig : sims ] ) matched that measured in the real data ( columns 2 and 4 in table [ tab : xcor ] ) .
the velocity error distributions ( recovered input ) resulting from the 50 simulations have widths of 0.02 , 0.05 , and 0.06 km s@xmath2 for the central , middle , and outer velocities respectively . for each simulation
we also record the average error of each measurement which results from making all cross - correlation measurements consistent ( the square of this quantity is actually what is minimized when solving for the best set of rotation velocities ) .
the mean of this error over all simulations is 0.031 km s@xmath2 .
however , in the real data , the error is 0.056 km s@xmath2 along the major axis and 0.074 km s@xmath2 along the minor axis .
this discrepancy indicates that the simulation was too idealistic , and we have therefore scaled the estimated velocity errors by 1.8 for the major axis , and 2.4 for the minor axis , yielding our formal error estimates of 0.04 , 0.09 , and 0.11 km s@xmath2 for the central , middle , and outer extractions along the major axis , and 0.05 , 0.12 , and 0.14 km s@xmath2 along the minor axis .
the velocity dispersion can be derived from the width of the cross - correlation peak , which is the sum of the line widths of the two spectra being correlated added in quadrature . in the case of the globular cluster wlm-1 ,
the line width is a combination of the intrinsic stellar line width , the instrumental profile , and the velocity dispersion of the cluster .
unfortunately , no template star was measured at the time of our observations . in order to estimate the velocity dispersion of wlm-1 , we used two high - resolution stellar spectra obtained from the uves paranal observatory project @xcite as templates .
we chose hd 30562 and hd 45067 ( both f8v stars ) because of their similarities in line strengths to the wlm-1 spectrum .
it is important to note that the template spectra were obtained using the @xmath24 slit and no binning , while the wlm-1 spectra used @xmath63 binning .
we therefore rebinned the template stars to the same resolution as our observations . in order to increase the signal - to - noise ratio of the wlm-1 spectra
, we averaged together the spectra extracted from opposite sides of the cluster ( since we only want to compare the velocity dispersions along the major and minor axes ) .
we then analyzed the spectra measured on each chip separately to allow optimal fourier filtering and rejection of undesirable lines .
for example , we rejected the blue end of the blue chip which was contaminated by broad hydrogen lines and caii h&k lines , the region around h@xmath138 on the red - l chip , and h@xmath139 on the red - u chip .
we then cross - correlated each wlm-1 spectrum with each of the two template spectra .
this gives a cross - correlation width for each template , which are then averaged .
the resulting cross - correlation widths derived for the integrated cluster spectra are 28.7 and 29.0 km s@xmath2for the major and minor axes respectively .
this width is the quadratic sum of the wlm-1 and template line widths . in order to estimate the line widths of the templates , we cross -
correlate the two templates with one another , and averaging the values from each chip , we find a width of 16.4 km s@xmath2 .
if we make the ( generous ) assumption that the wlm-1 and template spectra have the same instrumental width and intrinsic stellar line widths , it is now possible to subtract these components to find the stellar velocity dispersion .
subtracting in quadrature yields velocity dispersions of 23.5 and 23.9 km s@xmath2 along the major and minor axes respectively .
following the same procedure for each of the various spectral extractions of wlm-1 yields the velocity dispersion profile shown in figure [ fig : veldisp ] , where the errors are three times the velocity errors , as estimated in section [ sec : rotationerrors ] . here the extraction semi - major axis distances are shown using flux - weighted centers .
for comparison we have overplotted the velocity dispersion profiles predicted by the isotropic maxwellian distribution of the king model .
these curves show the velocity dispersions computed according to eq .
( 31 ) in @xcite with a concentration @xmath140 2.00 , 2.25 , and 2.50 , scaled to match the observed central velocity dispersion ( but see sec .
[ sec : veldispest ] ) , where the @xmath141 profile is the flattest and , of these three curves , the best fit to the data .
the discrepancy between the model and the measured data points ( given that the light profile indicates @xmath142 ) suggests that either the mass distribution is more extended than predicted by a king profile or that there is a systematic problem in our measurement of the velocity dispersion . in order to verify that any observed differences ( or lack thereof ) in line widths are inherent in the source spectra and not due to changes in the instrumental / atmospheric contribution ,
we have also compared the widths of unresolved telluric absorption lines in each using the atmospheric transmission spectrum described in section [ sec : telluric ] as the template .
we first excise a region of each wlm-1 spectrum in the range 62706320 which is dominated by absorption lines due to atmospheric o@xmath90 .
we remove the continuum and broad features by fitting with a high - order cubic spline , and estimate the instrumental width of each via cross - correlation with the atmospheric spectrum .
the widths of the telluric lines present in the major- and minor - axis spectra differ by less than 0.05 km s@xmath2 , confirming the stability of the conditions during the observations .
our _ absolute _ measurement of the velocity dispersion profiles is uncertain due to the low signal - to - noise of our spectra , and the imperfect match with the templates which were obtained at a different time using a different instrument configuration .
however , the _ relative _ velocity dispersion between the major and minor axes should be much more reliable .
the bottom of figure [ fig : veldisp ] shows the difference between the velocity dispersions measured along the major and minor axes .
all spectral extraction regions are consistent with zero difference , and comparing the integrated cluster spectra measured along the major and minor axes yields a difference of only @xmath143 km s@xmath2 ( major minor ) .
we can obtain an estimate of the central velocity dispersion of wlm-1 as a reality check of our measurement of the velocity dispersion profile . the simplest way to obtain a rough estimate of the velocity dispersion
is to utilize the correlation between the central velocity dispersion and the cluster absolute magnitude which was found by @xcite to be followed by globular clusters in several different galaxies .
on the basis of figure 3 ( top panel ) in djorgovski et al . , we find @xmath144 where @xmath145 is in km s@xmath2 .
we use the apparent magnitude of the cluster from table i in @xcite , @xmath146 mag , and the extinction and distance moduli estimated by @xcite , @xmath147 mag and @xmath38 mag , respectively , to find @xmath148 mag for wlm-1 .
the above relation then gives a central velocity dispersion @xmath149 km s@xmath2 .
a perhaps more accurate estimate for the central velocity dispersion of the cluster can be obtained from the `` fundamental plane '' correlation for globular clusters , as provided by @xcite . in particular
, djorgovski has shown that ( galactic ) globulars follow tight relations between the central velocity dispersion @xmath145 , the central surface brightness @xmath56 , and the core radius @xmath150 , on the one hand ; and between @xmath145 , the half - light radius @xmath47 , and mean surface brightness within @xmath47 , @xmath151 . based on his equations ( 3 ) and ( 6 ) , we find : @xmath152 and @xmath153 where @xmath150 and @xmath47 are in parsecs , and @xmath145 in km s@xmath2 .
using equation ( [ eqn : sigrc ] ) with the structural parameters derived in section [ sec : sbprofile ] , core radius @xmath154 pc and central @xmath55-band surface brightness @xmath155 mag / arcsec@xmath35 , we estimate the velocity dispersion to be @xmath156 km s@xmath2 .
equation ( [ eqn : sigrh ] ) with the half - light radius @xmath157 pc and the mean surface brightness inside the half - light radius @xmath158 mag / arcsec@xmath35 gives a central velocity dispersion of @xmath159 km s@xmath2 .
these estimates of the central velocity dispersion are lower than our measurements in figure [ fig : veldisp ] , but the uncertainties do not exclude our value .
the larger velocity dispersion implies that the mass of wlm-1 lies on the high side of the relation derived for galactic clusters .
extracting an integrated spectrum from the entire cluster using a @xmath160 ( @xmath161 ) aperture yields a relatively high signal - to - noise spectrum relative to the partial extractions used to measure cluster rotation .
these integrated spectra have many strong lines from which we can measure the net radial velocity of the cluster . in order to make this measurement , we first calculated and corrected for terrestrial motion , based on the observation times of 2003/10/03 at 3:23 and 4:23 ut for the minor and major - axis respectively .
we obtained an atomic line list from @xcite , and used the iraf task rvidlines to calculate the radial velocity .
centroiding @xmath162 lines in each of the major and minor axis spectra yields heliocentric radial velocities of @xmath163 and @xmath164 km s@xmath2 respectively . averaging the results we estimate that the heliocentric radial velocity of wlm-1 is @xmath165 km s@xmath2 .
the radial velocity of the wlm galaxy is @xmath166 km s@xmath2 @xcite , thus the cluster has a velocity of @xmath167 km s@xmath2 with respect to its host .
it appears that the ellipticity of wlm-1 can not be due to rotation , so perhaps wlm-1 is ( in a sense ) more like elliptical galaxies , where anisotropies in the velocity dispersion play an important role in the observed morphology . to address this hypothesis
we employ the tensor virial theorem , as described in @xcite , to obtain information about the relation between the ratio of rotation velocity and velocity dispersion @xmath168 , on the one hand , and ellipticity , on the other .
we start by assuming that a globular cluster is an oblate spheroid , rotating around the @xmath169 axis , and is seen edge - on .
we assume , moreover , that the system s isodensity surfaces are concentric ellipsoids . defining @xmath170 as the mass - weighted mean - square rotation speed , and @xmath171 as the mass - weighted mean - square random velocity along the line of sight , we arrive at the following relation : @xmath172 - 2,\ ] ] where @xmath173 is a parameter that measures the anisotropy of the velocity dispersion tensor and @xmath174 is the _ eccentricity _ , which is related to the ellipticity @xmath175 as follows : @xmath176 using these equations
, we can build a family of curves giving @xmath177 as a function of ellipticity for different values of @xmath178 .
the result is shown in figure [ fig : velanisotropy ] .
binney & tremaine ( 1987 ) argue that the representative points in the @xmath179 plane of systems that are flattened by rotation lie close to the curve @xmath180 , _ independent _ of the inclination angle @xmath181 .
the mean ellipticity of wlm-1 , as measured in section [ sec : ellipticity ] , is @xmath182 , which is indicated in figure [ fig : velanisotropy ] by the vertical dashed line .
as the low mass of the wlm galaxy effectively rules out the possibility that wlm-1 owes its ellipticity to tidal effects or gravothermal shocks , this figure very nicely illustrates the kinematical combinations available to explain the ellipticity of wlm-1 . at one extreme , the ellipticity of wlm-1 could be produced entirely by rotation . in this scenario
there is no velocity anisotropy ( @xmath183 ) , and reading from the plot , we see that @xmath184 .
taking a conservative estimate of 10 km s@xmath2 for the central velocity dispersion of the cluster ( sec .
[ sec : veldisp ] ) , the required rotation would be 5.7 km s@xmath2 .
however , as we find no measurable rotation ( @xmath185 km s@xmath2 ) , we exclude this possibility . at the other extreme ,
the ellipticity of wlm-1 is due entirely to velocity anisotropy .
the quantity @xmath186 , defined as the ratio between the diagonal terms of the velocity dispersion tensor along the rotation ( `` @xmath169 '' ) axis and along the elongation ( `` @xmath187 '' ) axis , should scale as @xmath188 from figure [ fig : velanisotropy ] , one sees that if there is zero rotation ( @xmath189 ) , it is possible to produce the observed ellipticity with a velocity dispersion parameter of @xmath190 .
therefore , wlm-1 s flattening could be accounted for if the velocity dispersion along the rotation axis were smaller than along the elongation axis by @xmath191 again taking the nominal velocity dispersion to be @xmath192 km s@xmath2 , this velocity anisotropy would only produce a difference of 0.7 km s@xmath2 in velocity dispersion between the major and minor axes .
such a difference would be too small to be detected on the basis of our data .
therefore , while we can not prove that velocity anisotropy is responsible for the observed shape of wlm-1 , it certainly appears to be the only option that is consistent with our observations .
we have analyzed vlt uves long - slit spectra and hst wfpc2 images of the extragalactic globular cluster wlm-1 in an attempt to understand the cause of its elliptical morphology .
unlike the situation for galactic globular clusters whose detailed shapes may be influenced by their massive host galaxy , wlm is a low - mass dwarf irregular galaxy , eliminating the possibilities that the non - sphericity of wlm-1 could be caused by either tidal stresses or gravothermal shocks .
this leaves two explanations for the flattening of this cluster : internal rotation and/or velocity dispersion anisotropy .
cluster rotation seemed to be a prime candidate for the origin of wlm-1 s ellipticity since two - body relaxation , which erases any primordial anisotropy , is relatively efficient in star clusters .
in addition , rotation has been measured for several galactic globular clusters using both proper motion and radial velocity techniques [ the review by @xcite lists the rotation velocities of 11 clusters ] .
the most notable case is that of @xmath3 cen , which exhibits correlated rotation and flattening profiles , strongly suggesting that its ellipticity is due to rotation .
we therefore obtained high - resolution long - slit echelle spectra along the major and minor axes of the cluster .
we used cross - correlation to look for velocity shifts on opposite sides of the cluster , dividing the cluster into annuli in an attempt to obtain a rotation velocity profile .
however , we found no evidence for rotation ( @xmath193 km s@xmath2 ) along either the major or minor axes of the cluster .
as the cluster would require a rotation of @xmath194 km s@xmath2 to sustain its shape , we conclude that rotation is definitely not the main driver of wlm-1 s ellipticity .
the only remaining plausible source of flattening in the wlm cluster is velocity dispersion anisotropy , the main reason elliptical galaxies are elliptical .
we attempted to measure the stellar velocity dispersion in wlm-1 based on the width of the cross - correlation peak ; however , due to our large errors , we could see no statistically significant difference between the velocity dispersions measured along the major and minor axes .
calculations reveal that the magnitude of the velocity anisotropy required to produce the observed ellipticity is relatively small : a 7% difference in velocity dispersion , or less than 1 km s@xmath2 , which is smaller than our velocity dispersion measurement errors .
support for this work was provided by a princeton - catlica prize fellowship , proyecto fondecyt regular no . 1030976 , and a gemini science fellowship ( aws ) , and by proyecto fondecyt regular
no . 1030954 ( mc ) . based on observations made with eso telescopes at the paranal observatories under programme i d 072.d-0355 ; and with the nasa / esa _ hubble space telescope _
, obtained at the space science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 , retrieved from the eso st - ecf archive .
djorgovski , s. g. , ct , p. , meylan , g. , et al .
2003 , in new horizons in globular cluster astronomy , asp conf .
g. piotto , g. meylan , s. g. djorgovski , & m. riello ( san francisco : asp ) , 479 | globular clusters have long been known for presenting ( at times ) significant deviations from spherical symmetry . while rotation has been the main proposed explanation , other complicating factors such as their constant interaction with the strong gravitational potential of their host galaxy
have made it difficult for a consensus to be reached . to address this question
we have obtained high - resolution spectra of wlm-1 , the lone globular cluster associated with the isolated , low - mass dwarf irregular galaxy wlm . using archival hst wfpc2 data
, we measure the radial ellipticity profile of wlm-1 , finding it to be highly elliptical , with a mean value of 0.17 in the region @xmath0which is comparable to what is found in our galaxy for the most elliptical globular clusters .
there is no evidence of isophote twisting , except for the innermost regions of the cluster ( @xmath1 ) . to investigate whether the observed flattening can be ascribed to rotation , we have obtained long - slit high - resolution vlt / uves spectra of this cluster along and perpendicular to the axis of flattening . using cross - correlation
we find that the velocity profile of the cluster is consistent with zero rotation along either axis .
thus neither cluster rotation nor galactic tides can be responsible for the flattened morphology of wlm-1 .
we argue that the required velocity dispersion anisotropy between the semi - major and semi - minor axes that would be required to account for the observed flattening is relatively small , of order 1 km s@xmath2 . even though our errors preclude us from conclusively establishing that such a difference indeed exists
, velocity anisotropy remains at present the most plausible explanation for the shape of this cluster . |
kolmogorov complexity ( also known as kolmogorov - chaitin or program - size complexity ) is recognized as a fundamental concept , but it is also often thought of as having little or no applicability because it is not possible to provide stable numerical approximations for finite particularly short strings by using the traditional approach , namely lossless compression algorithms .
we advance a method that can overcome this limitation , and which , though itself limited in ways both theoretical and numerical , nonetheless offers a means of providing sensible values for the complexity of short strings , complementing the traditional lossless compression method that works well for long strings .
this is done at the cost of massive numerical calculations and through the application of the coding theorem from algorithmic probability theory that relates the frequency of production of a string to its kolmogorov complexity .
bennett s logical depth , on the other hand , is a measure of the complexity of strings that , unlike kolmogorov complexity , measures the _ organized _ information content of a string . in
@xcite an application inspired by the notion of logical depth was reported in the context of the problem of image classification .
however , the results in this paper represent the first attempt to provide direct numerical approximations of logical depth . the independence of the two measures kolmogorov complexity and logical depth which has been established theoretically , is also numerically tested and confirmed in this paper .
our work is in agreement with what the theory predicts , even for short strings despite the limitations of our approach .
our attempt to apply these concepts to practical problems ( detailed in a series of articles ( see e.g. @xcite ) ) is novel , and they are indeed proving to have interesting applications where evaluations of the complexity of finite short strings are needed @xcite . in sections
[ kolmochap ] , [ ld ] , [ codingchap ] and [ formal ] , we introduce the measures , tools and formalism used for the method described in section [ dist ] . in section [ comparison ] ,
we report the numerical results of the evaluation and analysis of the comparisons among the various measures , particularly the connection between number of instructions , integer valued program - size complexity , kolmogorov complexity approximated by means of the coding theorem method , and logical depth .
when researchers have chosen to apply the theory of algorithmic information ( ait ) , it has proven to be of great value despite initial reservations @xcite .
it has been successfully applied , for example , to dna false positive repeat sequence detection in genetic sequence analysis @xcite , in distance measures and classification methods @xcite , and in numerous other applications @xcite .
this effort has , however , been hamstrung by the limitations of compression algorithms currently the only method used to approximate the kolmogorov complexity of a string given that this measure is not computable .
central to ait is the basic definition of plain algorithmic ( kolmogorov - chaitin or program - size ) complexity @xcite : @xmath3 where @xmath4 is a universal turing machine and @xmath5 the program that , running on @xmath4 , produces @xmath1 . traditionally , the way to approach the algorithmic complexity of a string has been by using lossless compression algorithms .
the result of a lossless compression algorithm applied to @xmath1 is an upper bound of the kolmogorov complexity of @xmath1 .
short strings , however , are difficult to compress in practice , and the theory does not provide a satisfactory solution to the problem of the instability of the measure for short strings . the invariance theorem , however , guarantees that complexity values will only diverge by a constant @xmath6 ( e.g. the length of a compiler or a translation program ) . +
* invariance theorem * ( @xcite ) : if @xmath7 and @xmath8 are two universal turing machines , and @xmath9 and @xmath10 the algorithmic complexity of @xmath1 for @xmath7 and @xmath8 respectively , there exists a constant @xmath6 such that : latexmath:[\[\label{invariance } hence the longer the string , the less important @xmath6 is ( i.e. the choice of programming language or universal turing machine ) .
however , in practice @xmath6 can be arbitrarily large , thus having a very great impact on the stability of kolmogorov complexity approximations for short strings .
a measure of the structural complexity ( i.e. richness of structure and organization ) of a string can be arrived at by combining the notions of algorithmic information and time complexity . according to the concept of logical depth @xcite , the complexity of a string
is best defined by the time that an unfolding process takes to reproduce the string from its shortest description .
while kolmogorov complexity is related to compression length , bennett s logical depth is related to decompression time . a typical example that illustrates the concept of logical depth , underscoring its potential as a measure of complexity , is a sequence of fair coin tosses .
such a sequence would have a high information content ( kolmogorov complexity ) because the outcomes are random , but it would have no structure because it is easily generated . the string 1111 1111 would be equally shallow as measured by logical depth . its compressed version , while very small , requires little time to decompress into the original string . in contrast , the binary expansion of the mathematical constant @xmath12 is not shallow , because though highly compressible and hence having a low kolmogorov complexity , it requires non - negligible computational time to produce arbitrary numbers of digits from its shortest program ( or indeed from any short program computing the digits of @xmath12 ) .
a detailed explanation pointing out the convenience of the concept of logical depth as a measure of organized complexity as compared to plain algorithmic complexity , which is what is usually used , is provided in @xcite . for finite strings , one of bennett s formal approaches to the logical depth of a string
is defined as follows : + let @xmath1 be a string and @xmath13 a significance parameter . a string s
depth at significance @xmath13 is given by @xmath14 with @xmath15 the length of the shortest program for @xmath1 , ( therefore @xmath16 ) .
in other words , @xmath17 is the least time @xmath18 required to compute @xmath1 from a @xmath13-incompressible program @xmath5 on a universal turing machine @xmath4 .
each of the three linked definitions of logical depth provided in @xcite comes closer to a definition in which near - shortest programs are taken into consideration . in this experimental approach
we make no such distinction among significance parameters , so we will denote the logical depth of a string @xmath1 simply by @xmath2 . like @xmath16 ,
@xmath2 as a function of @xmath1 is uncomputable .
a novel feature of this research is that we provide exact numerical approximations for both measures @xmath16 and @xmath2 for specific short @xmath1 , allowing a direct comparison .
this was achieved by running a large set of random turing machines and finding the smallest and fastest machines generating each output string .
hence these approximations are deeply related to another important measure of algorithmic information theory .
the algorithmic probability ( also known as levin s semi - measure ) of a string @xmath1 , is a measure that describes the expected probability of a random program @xmath5 running on a universal ( prefix - free . for details see @xcite ) . ] ) turing machine @xmath4 producing @xmath1 .
formally @xcite , @xmath19 i.e. the sum over all the programs for which @xmath4 with @xmath5 outputs @xmath1 and halts .
levin s semi - measure @xmath20 defines a distribution known as the _ universal distribution _ @xcite .
it is important to notice that the value of @xmath20 is dominated by the length of the smallest program @xmath5 ( when the denominator of @xmath21 reaches its largest value ) .
the length of the smallest program @xmath5 that produces the string @xmath1 is @xmath16 .
the semi - measure @xmath20 is therefore also uncomputable , because for every @xmath1 , @xmath20 requires the calculation of @xmath22 , involving @xmath23 , which is itself uncomputable . an extension of @xmath20 to non - binary alphabets
is natural .
more formally , @xmath24 can be associated with the original definition for binary strings .
however , one may want to extend @xmath24 to @xmath25 , in which case for every @xmath26 , the function @xmath27 is semi - computable ( for the same reason that @xmath28 is uncomputable ) .
an alternative @xcite to the traditional use of compression algorithms to approximate @xmath23 can be derived from a fundamental theorem that establishes the exact connection between @xmath20 and @xmath16 . +
* coding theorem * ( levin @xcite ) : @xmath29 where we will use @xmath0 to indicate that @xmath23 has been approximated by means of @xmath21 through the coding theorem .
an important property of @xmath21 as a semi - measure is that it dominates any other effective semi - measure @xmath30 , because there is a constant @xmath31 such that for all @xmath1 , @xmath32 .
for this reason @xmath20 is often called a _ universal distribution _ @xcite .
the ability of a universal turing machine to simulate any algorithmic process has motivated and justified the use of universal turing machines as the language framework within which definitions and properties of mathematical objects are given and studied . however , it is important to describe the formalism of a turing machine , because exact values of algorithmic probability for short strings will be approximated under this model , both for @xmath16 through @xmath20 ( denoted by @xmath0 ) , and for @xmath16in terms of the number of instructions used by the smallest turing machine producing @xmath1 .
+ consider a turing machine @xmath33 with alphabet @xmath34 symbols , @xmath35 states and an additional halting state denoted by @xmath36 ( as defined by rado in his original busy beaver paper @xcite ) . at the outset
the turing machine is in its initial state @xmath37 .
the machine runs on a @xmath38-way unbounded tape .
its behavior is determined by the transition function @xmath39 .
so , at each step : the machine s current `` state '' ( instruction ) @xmath1 ; and the tape symbol the machine s head is scanning @xmath26 define the transition @xmath40 with a unique symbol @xmath41 to write ( the machine can overwrite a @xmath37 on a @xmath36 , a @xmath36 on a @xmath37 , a @xmath37 on a @xmath37 , and a @xmath36 on a @xmath36 ) ; a direction @xmath13 to move in : @xmath42 ( left ) , @xmath37 ( right ) or @xmath36 ( none , when halting ) ; and a state @xmath43 to transition into ( which may be the same as the one it was in ) .
the machine halts if and when it reaches the special halt state @xmath36
. there are @xmath44 turing machines with @xmath45 states and @xmath38 symbols according to the formalism described above , as there are @xmath46 entries in the transition table and any of them may have @xmath47 possible instructions : there are @xmath38 halting instructions ( writing ` 0 ' and ` 1 ' ) and @xmath48 non - halting instructions ( @xmath38 movements , @xmath38 possible symbols to write and @xmath45 states ) .
the output string is taken from the number of contiguous cells on the tape the head of the halting @xmath45-state machine has gone through .
a turing machine is considered to produce an output string only if it halts .
the output is what the machine has written on the tape .
in order to arrive at an approximation of @xmath20 , a method based on the coding theorem was advanced in @xcite .
it is captured in the following function .
let @xmath33 be a turing machine in @xmath49 with empty input . then : @xmath50 where @xmath51 denotes the number of elements of @xmath52 .
let @xmath21 be fixed .
it has been proved @xcite that the function @xmath53 is non - computable ( due to the denominator ) .
however , @xmath54 for fixed and small values @xmath45 and @xmath21 is computable for values of the busy beaver problem @xcite that are known . for @xmath55 , for example , the busy beaver function @xmath56 tells us that @xmath57 @xcite , so given a turing machine with 4 states running on a blank tape that hasnt halted after 107 steps , we know it will never stop . more generally , for every string @xmath1 ( with alphabet @xmath58 ) one can compute a sequence @xmath59 which converges to @xmath60 when @xmath61 . for @xmath62
we compute for @xmath18 steps all @xmath21-turing machines with @xmath45 states ( there is a finite number of them ) and compute the quotient for @xmath60 for machines that halted before @xmath18 steps . since @xmath62 converges for every @xmath1 to @xmath63 (
@xmath21 fixed , @xmath45 fixed , @xmath1 fixed ) , the value of @xmath64 converges for fixed @xmath21 and @xmath45 . in this specific sense @xmath65
is approachable , even if @xmath62 as a function of increasing time @xmath18 may increase when a machine produces @xmath1 , or decrease when a machine halts without producing @xmath1 . by the invariance theorem ( eq . [ invariance ] ) and the coding theorem ( eq . [ coding ] ) , @xmath63 is guaranteed to converge to @xmath20 .
exact values of @xmath62 were previously calculated @xcite for @xmath66 symbols and @xmath67 states for which the busy beaver values are known .
that is , a total of 36 , 10000 , 7529536 and 11019960576 turing machines respectively .
the distributions were very stable and are proving to be capable of delivering applications that are also in agreement with results from lossless compression algorithms @xcite for boundary cases ( where both methods can be applied ) , hence validating the utility of both methods ( compression being largely validated by its plethora of applications and by the fact that it achieves an approximation of @xmath23 , see e.g. @xcite ) .
the chief advantage of the coding theorem method , however , is that it is capable of dealing with short entities ( unlike compression algorithms , which are designed for large entities ) .
there are 26559922791424 turing machines with 5 states and 2 symbols , and the values of busy beaver functions for these machines are unknown . in what follows
we describe how we proceeded .
calculating @xmath68 is an improvement on our previous numerical evaluations and provides a larger data set to work with , allowing us to draw more significant statistical conclusions vis - - vis the relation between these calculations and strict integer value program - size complexity , as well as to make a direct comparison to bennett s logical depth .
we did not run all the turing machines with 5 states to produce @xmath69 , because one can take advantage of symmetries and anticipate some of the behavior of the turing machines directly from their transition tables without actually running them ( this is impossible generally due to the halting problem , but some reductions are possible ) .
if @xmath70 is the set of turing machines with @xmath45 states and @xmath21 symbols , as defined above , we reduce it to : @xmath71 where @xmath39 is the transition function of @xmath33 .
so @xmath72 is a subset of @xmath70 , with machines with the transition corresponding to initial state @xmath37 and symbol @xmath36 ( this is the initial transition in a ` 0'-filled blank tape ) moving to the right and changing to a state different from the initial and halting ones . for machines with two symbols , @xmath73 as there are @xmath74 different initial transitions ( the machine can write ` 0 ' or ` 1 ' and move to one of the @xmath75 states in @xmath76 ) , and for the other @xmath77 transitions there are @xmath47 possibilities , as in @xmath70 .
after running @xmath78 on a ` 0'-filled tape , the procedure for completing the output strings so that they reach the frequency they have in @xmath70 is : * for every @xmath79 , * * if @xmath33 halts and produces the output string @xmath1 , add one occurrence of @xmath80 , the reverse of @xmath1 .
* * if @xmath33 does not halt , count another non - halting machine .
+ these two completions add the output ( or number of non - halting machines ) of @xmath81 new machines , one for each machine in @xmath78 .
these new machines are left - right symmetric to the machines in @xmath78 .
formally , this is the set @xmath82 when the original machine halts , its symmetric counterpart halts too and produces the reversed strings , and if the original machine does not halt , neither does the symmetric machine . this way we consider the output of all machines with the initial transition moving to the left and to a state not in @xmath83 . *
include @xmath84 occurrences of string `` 1 '' .
this corresponds to the machines writing ` 1 ' and halting at the initial transition .
there is just one possible initial transition for these machines ( move to the halting state , write ` 1 ' and remain in the initial cell ) .
the other @xmath77 transitions can have any of the @xmath47 possible instructions .
* include @xmath84 occurrences of string `` 0 '' .
this is justified as above , for machines writing ` 0 ' and halting at the initial transition . *
include @xmath85 additional non - halting machines , corresponding to machines remaining in the initial state in the initial transition ( these machines never halt , as they remain forever in the initial state ) .
there are @xmath86 initial transitions of this kind , as the machine can write @xmath38 different symbols and move in @xmath38 possible directions .
if we sum @xmath87 and the machines considered above , having completed them in the manner described , we get the output corresponding to the @xmath88 machines in @xmath89 .
moreover , we need the output of those machines starting with a ` 0'-filled tape and with a ` 1'-filled tape .
but we do not run any machine twice , as for every machine @xmath90 producing the binary string @xmath1 starting with a ` 1'-filled tape , there is also a 0 - 1 symmetric machine @xmath91 ( where the role of 1 ( of 0 ) in the transition table of @xmath92 is the role of 0 ( of 1 ) in the transition table of @xmath33 ) that when starting with a ` 0'-filled tape produces the complement to one of @xmath1 , that is , the result of replacing all 0s in s with 1s and all 1s with 0s .
so we add the complement to every one of the strings found and count the non - halting machines twice to obtain the output of all machines in @xmath89 starting both with a ` 0'-filled tape and with a ` 1'-filled tape .
to construct @xmath69 , we ran the @xmath93 machines in @xmath94 , which is @xmath95 of @xmath96 .
the output strings found in @xmath94 , together with their frequencies , were completed prior to constructing @xmath69 , following the procedure explained above .
it is useful to avoid running machines that we can easily determine will not stop .
these machines will consume the runtime without yielding an output . as we have shown above
, we can avoid generating many non - halting machines .
in other cases , we can detect them at runtime , by setting appropriate filters .
the theoretical limit of the filters is the halting problem , which means that they can not be exhaustive . but a practical limit is imposed by the difficulty of checking some filters , which takes up more time than the runtime that is saved .
we have employed some filters that have proven useful .
briefly , these are : * * machines without transitions to the halting state*. while the transition table is being filled , the simulator checks to ascertain whether there is some transition to the halting state .
if not , it avoids running it .
* * escapees*. these are machines that at some stage begin running forever in the same direction . as they are always reading new blank symbols , as soon as the number @xmath6 of non - previously visited positions is greater than the number @xmath45 of states , we know that they will not stop , because the machines have necessarily entered an infinite loop . given that @xmath97 , while visiting the last @xmath6 new cells , some of the @xmath45 states have been repeated , and will repeat forever , as the machine s behavior is deterministic .
* * cycles of period two*. these cycles are easy to detect .
they are produced when in steps @xmath18 and @xmath98 the tape is identical and the machine is in the same state and the same position .
when this is the case , the cycle will be repeated infinitely .
these filters were implemented in our c++ simulator , which also uses the reduced enumeration of section [ sec : some - reductions ] . to test them we calculated @xmath99 with the simulator and compared the output to the list that was computed in @xcite , arriving at exactly the same results , and thereby validating our reduction techniques .
running @xmath100 without reducing the enumeration or detecting non - halting machines took 952 minutes . running the reduced enumeration with non - halting detectors took 226 minutes .
the busy beaver for turing machines with 4 states is known to be 107 steps @xcite , that is , any turing machine with 2 symbols and 4 states running longer than 107 steps will never halt .
however , the exact number is not known for turing machines with 2 symbols and 5 states , although it is believed to be 47176870 , as there is a candidate machine that runs for this length of time and halts and no machine with a greater runtime has yet been found .
so we decided to let the machines with 5 states run for 4.6 times the busy beaver value for 4-state turing machines ( for 107 steps ) , knowing that this would constitute a sample significant enough to capture the behavior of turing machines with 5 states .
the chosen runtime was rounded to 500 steps , which was used to construct the output frequency distribution for @xmath69 .
not all 5-state turing machines have been used to build @xmath69 , since only the output of machines that halted at or before 500 steps was taken into consideration . as an experiment to ascertain how many machines we were leaving out , we ran @xmath101 random turing machines for up to 5000 steps . among these ,
only 50 machines halted after 500 steps and before 5000 ( that is , a fraction less than @xmath102 , because in the reduced enumeration we do nt include those machines that halt in one step or that we know wo nt halt before we generate them , so it s a smaller fraction ) , with the remaining 1496491379 machines not halting at 5000 steps . as far as these are concerned and given that the busy beaver values for 5 states are unknown we do not know after how many steps they would eventually halt , if they ever do . according to the following analysis , our election of a runtime of 500 steps therefore provides a good estimation of @xmath69 .
the frequency of runtimes of ( halting ) turing machines has theoretically been proven to drop exponentially @xcite , and our experiments are closer to the theoretically predicted behavior . to estimate the fraction of halting machines that were missed because turing machines with 5 states were stopped after 500 steps , we hypothesize that the number of steps @xmath56 a random halting machine needs before halting is an exponential random variable , defined by @xmath103 we do not have direct access to an evaluation of @xmath104 , since we only have data for those machines for which @xmath105 .
but we may compute an approximation of @xmath106 , @xmath107 , which is proportional to the desired distribution . a non - linear regression using ordinary least - squares
gives the approximation @xmath108 with @xmath109 and @xmath110 .
the residual sum - of - squares is @xmath111 ; the number of iterations with starting values @xmath112 and @xmath113 is nine .
the model s @xmath114 is the same @xmath114 appearing in the general law @xmath104 , and may be used to estimate the number of machines we lose by using a 500 step cut - off point for running time : @xmath115 .
this estimate is far below the point where it could seriously impair our results : the less probable ( non - impossible ) string according to @xmath69 has an observed probability of @xmath116 .
although this is only an estimate , it suggests that missed machines are few enough to be considered negligible .
we now study the relation of @xmath0 to the minimal number of instructions used by a turing machine producing a given string , and to bennett s concept of logical depth .
as expected , @xmath0 shows a correlation with the number of instructions used but not with logical depth .
first , we are interested in the relation of @xmath117 to the minimal number of instructions that a turing machine producing a string @xmath1 uses .
machines in @xmath69 have a transition table with 10 entries , corresponding to the different pairs @xmath49 , with @xmath1 one of the five states and @xmath21 either `` 0 '' or `` 1 '' .
these are the 10 instructions that the machine can use .
but for a fixed input not all instructions are necessarily used .
then , for a blank tape , not all machines that halt use the same number of instructions .
the simplest cases are machines halting in just one step , that is , machines whose transition for @xmath118 goes to the halting state , producing a string `` 0 '' or `` 1 '' .
so the simplest strings produced in @xmath68 are computed by machines using just one instruction .
we expected a correlation between the @xmath0-complexity of the strings and the number of instructions used . as we show
, the following experiment confirmed this .
we used a sample of @xmath119 random machines in the reduced enumeration for @xmath68 , that is , @xmath120 the total number of machines .
the output of the sample returns the strings produced by halting machines together with the number of instructions used , the runtime and the instructions for the turing machine ( see fig . [
fig : distribins ] ) . in order to save space
, we only saved the smallest number of instructions found for each string produced , and the smallest runtime corresponding to that particular number of instructions .
values according to the minimum number of instructions required . each
drop - like " distribution is the set of strings that are minimally produced with the same number of instructions ( horizontal axis ) .
the more instructions needed to produce the strings , the more complex they are ( vertical axis in @xmath121 units).,width=377 ] after doing the appropriate symmetry completions we have 99584 different strings , which is to say almost all the 99608 strings found in @xmath68 .
the number of instructions used goes from 1 to 10 .
when 1 instruction is used only `` 0 '' and `` 1 '' are generated , with a @xmath0 value of @xmath122 . with 2 instructions ,
all 2-bit strings are generated , with a @xmath0 value of @xmath123 . for 3 or more instructions , fig .
[ fig : distribins ] shows the distribution of values of @xmath0 .
table [ tab : meankl ] shows the mean @xmath0 values for the different numbers of instructions used .
.mean @xmath0 and string length for different numbers of instructions used .
[ cols=">,>,>",options="header " , ] we now provide some examples of the discordance between @xmath0 and @xmath124 .
`` 0011110001011 '' is a string with high @xmath0 and low @xmath124 .
[ fig : tablelowtime ] shows the transition table of the smallest machine found producing this string .
the runtime is low just 29 steps ( of the 99584 different strings found in our sample , only 3360 are produced in fewer steps ) , but it uses 10 instructions and produces a string with complexity @xmath125 .
it is the greatest complexity we have calculated for @xmath0 .
[ fig : execlowtime ] shows the execution of the machine . on the other hand , `` @xmath126 ''
is a string with high @xmath124 but a low @xmath0 value .
[ tmtablehightime ] shows the transition table of the machine found producing this string , and fig .
[ tmrunhightime ] depicts the execution .
the machine uses 9 instructions and runs for 441 steps ( only 710 strings out of the 99584 strings in our sample require more time ) but its @xmath0 value is @xmath127 .
this is a low complexity if we consider that in @xmath0 there are 99608 strings and that 90842 are more complex than this one . ' ' .,width=415 ] '' . with a high runtime ,
it produces a string with low complexity.,width=453 ] we may rate the overall strength of the relation between @xmath128 and @xmath124 by the correlation @xmath129 , corresponding to a medium positive link . as we previously mentioned , however , the fact that the length @xmath130 of the strings is linked with both variables may bias our interpretation .
a more relevant measure is thus @xmath131 , a slight negative but with no significant value between @xmath128 and @xmath124 once @xmath130 is controlled .
the results in this paper are important because these measures can be better studied and understood under a specific but widely known general formalism .
what we have found is very interesting because it is what one would wish in the best case scenario , stable and reasonable distributions rather than chaotic and unstable ones .
the results also suggest that these measures can be applied even if numerically approximated using a specific model of computation .
for example , as we expected , the kolmogorov - chaitin complexity evaluated by means of levin s coding theorem from the output distribution of small turing machines correlates with the number of instructions used but not with logical depth .
logical depth also yields a measure that is different from the measure obtained by considering algorithmic complexity ( @xmath23 ) alone , and this investigation proves that all these three measures ( kolmogorov - chaitin complexity , solomonoff - levin algorithmic probability and bennett s logic depth ) are consistent with theoretical expectations .
@xmath23 as a measure of program - size complexity is traditionally expected to be an integer ( the length of a program in bits ) , but when evaluated through algorithmic probability using the coding theorem it retrieves non - integer values ( still bits ) .
these results confirm the utility of non - integer values in the approximation of the algorithmic complexity of short strings , as they provide finer values with which one can tell apart small differences among short strings which also means one can avoid the longer calculations that would be necessary in order to tell apart the complexity of very small objects if only integer values were allowed .
thus it also constitutes a complementary and alternative method to compression algorithms . an _ on - line algorithmic complexity calculator _ ( or oacc ) is now available at http://www.complexitycalculator.com .
it represents a long - term project to develop an encompassing universal tool implementing some of the measures and techniques described in this paper .
it is expected to be expanded in the future as it currently only implements numerical approximations of kolmogorov complexity and levin s semi - measure for short binary strings .
more measures , more data and better approximations will be gradually incorporated in the future , covering a wider range of objects , such as longer binary strings , non - binary strings and multidimensional arrays ( such as images ) .
bennett , logical depth and physical complexity in rolf herken ( ed ) _ the universal turing machine a half - century survey , _ oxford university press 227257 , 1988 .
bennett , how to define complexity in physics and why . in _ complexity , entropy and the physics of information .
_ zurek , w. h. , addison - wesley , eds .
sfi studies in the sciences of complexity , p 137 - 148 , 1990 .
brady , the determination of the value of rado s noncomputable function @xmath132 for four - state turing machines , _ mathematics of computation 40 _ ( 162 ) : 647665 , 1983 .
calude , _ information and randomness _ , springer , 2002 .
calude and m.a .
stay , most programs stop quickly or never halt , _ advances in applied mathematics _
, 40 , 295 - 308 , 2008 .
chaitin , on the length of programs for computing finite binary sequences : statistical considerations , _ journal of the acm _ , 16(1):145159 , 1969 .
_ from philosophy to program size , _ 8th .
estonian winter school in computer science , institute of cybernetics , tallinn , 2003 .
r. cilibrasi , p. vitanyi , clustering by compression , _ ieee transactions on information theory , _ 51 , 4 , 15231545 , 2005 .
t.m . cover and j.a .
thomas , _ information theory , _
j. wiley and sons , 2006 .
delahaye , _ complexit alatoire et complexit organise , _ editions quae , 2009 .
delahaye , h. zenil , towards a stable definition of kolmogorov - chaitin complexity , arxiv:0804.3459 , 2007 .
delahaye and h. zenil , on the kolmogorov - chaitin complexity for short sequences . in c.
calude ( ed . ) , _ randomness and complexity : from leibniz to chaitin _ , world scientific , 2007 .
delahaye & h. zenil , numerical evaluation of the complexity of short strings : a glance into the innermost structure of algorithmic randomness , _ applied math . and comp .
w. kircher , m. li , and p. vitanyi , the miraculous universal distribution , _ the mathematical intelligencer , _ 19:4 , 715 , 1997 .
kolmogorov , three approaches to the quantitative definition of information , _ problems of information and transmission _ , 1(1):17 , 1965 . l. levin , laws of information conservation ( non - growth ) and aspects of the foundation of probability theory .
, _ problems in form . transmission _ 10 . 206210 , 1974 . m. li , p. vitnyi , _ an introduction to kolmogorov complexity and its applications , _ springer , 2008 .
. rivals , m. dauchet , j .-
delahaye , o. delgrange , compression and genetic sequence analysis . , _ biochimie _ , 78 , pp 315 - 322 , 1996 . t. rad , on non - computable functions , _ bell system technical journal , _ vol .
3 , pp . 877884 , 1962 .
solomonoff , a formal theory of inductive inference : parts 1 and 2 . _ information and control _ , 7:122 and 224254 , 1964 .
h. zenil , une approche exprimentale la thorie algorithmique de la complexit , dissertation in fulfilment of the degree of doctor in computer science ( jury members : j .- p .
delahaye and c.s .
calude , g. chaitin , s. grigorieff , p. mathieu and h. zwirn ) , universit de lille 1 , 2011 .
h. zenil , f. soler - toscano , j .-
delahaye and n. gauvrit , two - dimensional kolmogorov complexity and validation of the coding theorem method by compressibility , arxiv:1212.6745 [ cs.cc ] .
h. zenil , j .-
delahaye and c. gaucherel , image information content characterization and classification by physical complexity , _ complexity _ , vol .
173 , pages 2642 , 2012 . | we show that real - value approximations of kolmogorov - chaitin ( @xmath0 ) using the algorithmic coding theorem as calculated from the output frequency of a large set of small deterministic turing machines with up to 5 states ( and 2 symbols ) , is in agreement with the number of instructions used by the turing machines producing @xmath1 , which is consistent with strict integer - value program - size complexity .
nevertheless , @xmath0 proves to be a finer - grained measure and a potential alternative approach to lossless compression algorithms for small entities , where compression fails .
we also show that neither @xmath0 nor the number of instructions used shows any correlation with bennett s logical depth @xmath2 other than what s predicted by the theory .
the agreement between theory and numerical calculations shows that despite the undecidability of these theoretical measures , approximations are stable and meaningful , even for small programs and for short strings .
we also announce a first beta version of an online algorithmic complexity calculator ( oacc ) , based on a combination of theoretical concepts , as a numerical implementation of the _ coding theorem method_. + * keywords : * coding theorem method ; kolmogorov complexity ; solomonoff - levin algorithmic probability ; program - size complexity ; bennett s logical depth ; small turing machines . |
levy and jokipii [ 1976 ] showed that the dominant force on small dust grains in the solar system is the lorentz - force caused by the solar wind magnetic field sweeping by the dust grains , which are charged to a positive surface potential , mainly by electron emission due to solar uv photo - effect .
it was concluded that due to the coupling to the radially outward flowing solar wind , small interstellar grains do not exist inside the heliosphere which defines the domain of the solar wind . for big grains ,
lorentz - force is less important , because they have smaller charge - to - mass ratios , but from the determination of the size range and size distribution of interstellar grains in our galaxy by fitting the wavelength dependence of extinction of starlight on its way through the interstellar medium [ e.g. _ mathis et al .
_ , 1977 ] , it was believed that typical interstellar grains are not larger than @xmath3 in diameter .
so the conclusion was drawn that there are no big interstellar dust grains and the small ones are swept out of the solar system by the solar wind magnetic field .
gustafson and misconi [ 1979 ] on the other hand , argued that the electromagnetic interaction enhances the spatial density of small ( radii between @xmath4 and @xmath5 ) interstellar grains upstream of the sun .
the detection of interstellar grains in the solar system by the dust detector on - board ulysses [ ] and its confirmation by data from the galileo dust detector [ ] proved the existence of grains larger than @xmath3 and that smaller grains can penetrate into the solar system , although they are depleted in number [ ] .
it was concluded [ ; ; ] that the lorentz - force on interstellar grains not only repels but also focuses them , depending on the configuration of the solar wind magnetic field , and thus on the phase of the solar cycle .
the objective of this work is to quantitatively determine the spatial distribution of interstellar dust grains in the solar system as a function of the solar cycle . to meet this objective ,
the motion of interstellar grains in the solar system is simulated by numerically integrating the equation of motion for a big ( @xmath6 to @xmath7 ) set of interstellar grains . in section [ sec_modeldescription ] the assumptions and approximations used in the model are described .
section [ sec_results ] presents the resulting three - dimensional and time - dependent spatial distributions of grains in the solar system .
the results are discussed and compared with the in - situ measurements of the ulysses and galileo spacecraft in section [ sec_discussion ] .
as initial conditions , a homogeneous , mono - directional stream of grains is assumed arriving from an upstream direction of @xmath8 ( heliocentric ecliptic longitude ) and @xmath9 ( heliocentric ecliptic latitude ) .
this direction is equal to the upstream direction of interstellar dust grains in the solar system as it was determined by analyzing the directional information of the ulysses in - situ measurements [ ; ] .
this direction is also compatible with the upstream direction of interstellar neutral helium atoms , determined by the measurements of the ulysses / gas experiment [ ] .
it was shown [ ] , that the impact velocity of interstellar grains measured by ulysses after jupiter fly - by was compatible with the velocity of the interstellar helium atoms .
because the impact velocity of an individual grain is determined with an uncertainty of factor @xmath10 by the in - situ measurement [ ] , the helium velocity of @xmath11 was adopted for the model . in the simulation , a `` wall '' of grains is placed at a heliocentric distance of @xmath12 every time interval @xmath13 .
the wall covers an area of @xmath14 for big and @xmath15 for small grains , and is @xmath16 thick .
initially the grains are distributed uniformly inside the wall .
the position @xmath17 and velocity @xmath18 of each grain as a function of time is determined by integrating the equation of motion given by : @xmath19 where @xmath20 is the strength of radiation pressure expressed as the ratio of the magnitude of radiation pressure force to the magnitude of gravity , @xmath21 is the gravitational coupling constant , @xmath22 the mass of the sun , @xmath23 and @xmath24 the charge and mass of the grain , @xmath25 the solar wind velocity , and @xmath26 the solar wind magnetic field . in the case
@xmath27 , i.e. if the lorentz - force on the grains is negligible , the spatial distribution of grains can be determined analytically [ ] and the solution is rotationally symmetric about the axis parallel to the upstream direction . if the lorentz - force can not be neglected , the solution to equation ( [ dgl ] ) has no continuous symmetry and the dust grain trajectories are essentially three - dimensional .
equation ( [ dgl ] ) contains three basic parameters : the strength of radiation pressure @xmath20 , the charge - to - mass ratio @xmath28 , and the strength of the radial @xmath29 and azimuthal @xmath30 components of the solar wind magnetic field . to reduce the number of free parameters in the model , spherical
grains with a constant density of @xmath31 are assumed . in this case ,
@xmath20 and @xmath28 can be expressed as a function of grain radius @xmath32 . in the model
the calculation of @xmath20 as a function of grain radius by gustafson [ 1994 ] is used ( see also table [ tab_cycle ] below ) . in this mie - theory calculation , the optical constants of `` astronomical silicates '' introduced by draine and lee [ 1984 ] are assumed .
the parameter @xmath20 reaches its maximum for @xmath33 and is greater than unity for @xmath34 . in this regime , interstellar grains are effectively repelled by the sun . for @xmath35
, @xmath20 reaches a constant value of @xmath36 , and for @xmath37 , @xmath20 depends on @xmath32 like @xmath38 , which is valid in the limit of geometric scattering . for the charge on a dust grain in interplanetary space
it is assumed that the charging process due to photo - emission of electrons induced by solar uv photons partially balanced by the interplanetary plasma environment produces a positive surface potential of @xmath39 on the grain surface [ ] .
the charge of the grain is then given by @xmath40 where @xmath41 is the permeability of the vacuum .
thus , @xmath42 , which means that electromagnetic interaction with the solar wind magnetic field is only important for small grains .
in the model , parker s [ 1958 ] model is used , which has been found to be a good large scale approximation of the observed magnetic field in interplanetary space [ ; ; ] .
the radial , azimuthal , and normal components ( @xmath43 , @xmath44 , @xmath45 ) of the parker - field can be expressed as a function of location in heliographic coordinates ( @xmath46 , @xmath47 , @xmath48 ) : @xmath49 where @xmath50 are the radial and azimuthal field strength at @xmath51 , respectively [ markcite_gustafson , _ 1994 ] .
the lorentz - force @xmath52 on a dust grain in interplanetary space caused by the solar wind magnetic field is dominated by the effect of the field sweeping by the grain with the solar wind speed , because the grain velocity is much smaller than the solar wind velocity .
therefore , the lorentz - force is governed by the term @xmath53 ( see equation ( [ dgl ] ) ) . although it would seem that the lorentz - force depends on the solar wind speed , this is not the case , because at higher solar wind speeds , the parker spiral becomes steeper , and consequently the azimuthal component @xmath44 is reduced [ ] , effectively keeping the lorentz - force constant
. the observed sector structure of the polarity of the solar wind magnetic field has been modeled by alfven [ 1977 ] using a `` ballerina model '' .
this model describes a warped current sheet that separates opposite polarities of the magnetic field in interplanetary space . in the ecliptic
, opposite polarities are observed in separate sectors , the boundaries of which are given by the intersections of the current sheet with the ecliptic plane . for dust grains in the considered size - range between @xmath54 and @xmath55 ,
the instantaneous polarity can be replaced by an average polarity @xmath56 which is a function of the heliographic latitude @xmath48 .
this averaging process is a good approximation for grains for which the larmor - frequency is orders of magnitude smaller than the solar rotation frequency @xmath57 , therefore , @xmath58 where @xmath59 is the interplanetary magnetic field strength at @xmath60 and @xmath61 is the solar rotation frequency ( rotational period of 27 days ) .
the effect of the sector structure on very small ( @xmath62 to @xmath63 ) dust grains which have been ejected by the jovian system , has been observed by ulysses [ ] .
the average polarity @xmath56 seen by a dust grain at heliographic latitude @xmath48 , is given by @xmath64 where @xmath65 is the tilt angle of the ( equivalent ) flat current sheet , that rotates with the @xmath66-year period of the solar cycle [ ] . from equation ( [ eqn_mean_polarity ] ) it is clear that the _ average _ field strength @xmath67 is small at low heliographic latitudes , whereas the _ instantaneous _ field strength is high at low heliographic latitudes ( see equation ( [ eqn_parkerfield ] ) ) .
this is because a particle at low latitudes experiences a negative polarity nearly as long as a positive polarity .
it was shown [ ] that the stochastic perturbation of the trajectories of small dust grains by the sectored solar wind magnetic field produces an outward directed diffusion of these grains .
this is not taken into account by the model , because the sector structure was removed by using an average polarity .
the average polarity @xmath56 defines the sign of the azimuthal component of the average magnetic field and thus the sign of the lorentz - force , and because the average polarities in the northern and southern hemispheres are always opposite , the lorentz - force accelerates dust grains away from the solar equatorial plane for @xmath68 and towards this plane for @xmath69 .
the sequence of this focusing / defocusing cycle is shown in table 1 . to simulate the motion of interstellar grains in the heliosphere , equation ( [ dgl ] )
was integrated by using a simple runge - kutta integrator of fourth order [ ] .
after the first grains have completely traversed the heliosphere , the spatial distribution of dust grains has been determined at each time step by counting grains inside each cell of an orthogonal , equidistant grid .
the resulting spatial distributions of interstellar grains in the heliosphere are discussed in the following sections .
the model described above uses some non - trivial approximations . for the determination of the basic parameters @xmath20 and @xmath28 ,
a spherical grain shape has been assumed . on the other hand , the observation of the polarization of starlight ( for summary see li and greenberg [ 1997 ] ) has shown that interstellar dust grains ( at least the part of the population that causes the extinction ) must have elongated shapes .
a deviation from a spherical shape would increase @xmath28 for the same surface potential @xmath70 , because the charge carriers on the grains are separated by larger distances .
the @xmath20-value would also increase for non - spherical particles , because it is proportional to the cross - section - to - volume ratio which is minimal for spheres .
it can therefore be concluded that a non - spherical particle of the same size as a spherical one has larger values for the basic parameters @xmath28 and @xmath20 , or , a non - spherical particle with the same @xmath28 and @xmath20 values used in the simulation is larger than indicated in table [ tab_dynpara ] .
still , @xmath28 and @xmath20 are quantities that depend on material properties of the grains and have not been measured directly .
therefore , they have to be treated as free parameters .
another assumption made in the model is the constancy of @xmath23 . due to short - term variations in the solar wind , the surface potential and thus @xmath23
is expected to fluctuate .
these charge fluctuations have the same effect on the grain s motion as short - term variations of the ambient magnetic field . as shown above
, it is a good approximation to use an average strength of the lorentz - force , with the averaging time - scale being the larmor - period of the grain .
furthermore it was shown by zook et al . [ 1996 ] , who modeled the motion of very small jupiter stream particles through interplanetary space , that the effect of charge fluctuations on grain dynamics can be neglected for the grain sizes under consideration .
finally it can be argued that the solar wind magnetic field is not well represented by a parker spiral during the solar maximum .
the physical situation during solar maximum is a highly disordered field with the polarity distribution showing no clear separation of polarities between the northern and southern heliographic hemisphere [ ] .
as shown above , the motion of grains in the size range of @xmath71 depends only on the _ average _ polarity @xmath56 . in the highly disordered configuration of the solar maximum , regions of positive and negative polarity act on the grains in a stochastic way , independent of the heliographic latitude of the grain s position .
therefore @xmath72 and thus vanishing average lorentz - force for all heliographic latitudes is a good approximation of the physical situation during the solar maximum , independent of the actual form of the field lines .
the spatial distribution of interstellar grains inside the heliosphere shows how the concentration of grains of different sizes is reduced or enhanced by solar gravity , radiation pressure , and the solar wind magnetic field . to display the distributions that result from the model described in section [ sec_modeldescription ] ,
a coordinate - system is chosen such that the @xmath73-axis is anti - parallel to the initial velocity vector .
the @xmath74-axis is then chosen to be perpendicular to the @xmath73-axis and to lie in the plane of the initial velocity vector @xmath75 and the solar rotation axis vector @xmath76 ( see figure [ fig_coorsys ] ) .
the @xmath77-axis is set to be perpendicular to the @xmath73- and @xmath74-axes such that a right - handed coordinate - system is formed .
the angle @xmath78 between the downstream - direction and the solar rotation axis ( as shown in figure [ fig_coorsys ] ) , has been determined by fitting the directional information of the ulysses in - situ measurements .
the best fit was achieved at @xmath79 [ ] .
the result of the simulation is shown for four different grain radii : @xmath80 , @xmath3 , @xmath81 , @xmath82 , and @xmath71 .
for the dynamical parameters @xmath20 and @xmath28 of these grains see table 2 .
the spatial distribution of the @xmath80 and @xmath83 grains is shown in figure [ fig_dd0705 ] . because of the small charge - to - mass ratio of these grains , the coupling to the time - variable solar wind magnetic field is weak , and therefore their spatial distribution is nearly constant and rotationally symmetric about the @xmath73-axis ( downstream direction ) .
the density enhancement downstream of the sun is produced by gravitational bending of the grain s trajectories towards the sun . due to their smaller @xmath20-value ,
the downstream concentration increase of the @xmath3-grains is less pronounced .
the comparison with the result of the analytical calculation [ ] , which assumes @xmath84 , shows that electromagnetic interactions can indeed be neglected for these particles .
the spatial distribution of grains with radii of @xmath81 is already slightly affected by their interaction with the solar wind magnetic field . for even smaller grains
the interaction with the solar wind magnetic field becomes important , as shown in figure [ fig_dd02_xy ] , in which the time - sequence of the density distribution of interstellar @xmath85-grains in the @xmath73-@xmath77-plane ( approximately in the ecliptic plane ) is displayed . as shown in table 1 ,
the solar maximum 1991 was the end of a focusing cycle .
the density enhancement in the @xmath73-@xmath77-plane is the result of the focusing of grains to this plane . at the maximum of solar activity
the average magnetic field is weak , because the current sheet extends to high heliographic latitudes . in this configuration sectors of positive and negative polarity
cancel each other efficiently as they pass by the dust grains , creating a weak average lorentz - force .
therefore , the region of enhanced grain density moves downstream without much perturbation until in 1996 the defocusing effect becomes important .
the void region downstream of the sun is expanded by the defocusing , and the overall density in the @xmath73-@xmath77-plane is reduced until the end of the defocusing cycle in 2002 . after that the new focusing cycle starts to increase the density in the @xmath73-@xmath77-plane until the distribution in 2012 is similar to the one in 1991 when the defocusing cycle started . as a consequence of the large @xmath20-value of @xmath86 ,
the concentration of @xmath85-grains downstream of the sun is very low . in the @xmath73-@xmath74-cut ( plane approximately perpendicular to the ecliptic that also contains the stream vector ) of the density distribution shown in figure [ fig_dd02_xz ]
one can see that the density distribution is no longer symmetric about the @xmath73-axis for @xmath82-grains .
the first panel in figure [ fig_dd02_xz ] also shows that because the upstream direction lies close to but not in the plane of the solar equator , the distribution is asymmetric about the @xmath73-@xmath77-plane . at the end of the focusing cycle in 1991
the grains are concentrated along the plane of the solar equator and the density at higher latitudes is reduced .
as the defocusing cycle proceeds , the density at higher latitudes increases as more grains are deflected into this region . in 2002 , at the end of the defocusing cycle the regions of enhanced density move downstream and the new focusing cycle starts again to concentrate grains around the plane of the solar equator until in 2012 the distribution of 1991 is restored .
the spatial distribution of grains with radii of @xmath71 is dominated by the interaction of these grains with the solar wind magnetic field due to their high charge - to - mass ratio of @xmath87 . at the end of the focusing solar cycle ,
the grains are concentrated in a sheet about the plane of the solar equator as can be seen in the first panel in figure [ fig_dd01_xz ] . because the solar wind magnetic field moves radially outward with the solar wind , the lorentz - force on the grains is not perfectly perpendicular to the solar equator , but also contains a small outward directed component .
this force decelerates small grains and causes a region of enhanced density upstream of the sun , which is visible as the arc - shaped feature in the first panel of figure [ fig_dd01_xy ] . as the average field strength is low during the solar maximum , this feature can propagate towards the sun and the concentration around the plane of the solar equator is split into a northern and southern part by the new defocusing field ( 1993 to 2001 in figure [ fig_dd01_xz ] ) .
as can be seen in the panel for 1996 in figure [ fig_dd01_xy ] , the region of high grain density is diverted before it reaches the solar vicinity .
therefore , the flux of @xmath71 grains measured in the solar vicinity ( roughly inside jupiter s orbit ) is expected to be reduced during the full solar cycle due to the interaction with the solar wind magnetic field .
this deficiency of small grains has been observed in the in - situ data [ ; ] .
after a new focusing cycle starts in 2002 , grains are again concentrated around the solar equator as can be seen in the panels from 2006 to 2012 in figure [ fig_dd01_xz ] until the configuration of 1991 is restored .
the simulation shows that the flux of small interstellar grains in the solar vicinity depends on the phase of the solar cycle . since in the interstellar medium , small grains ( @xmath88 )
dominate the dust population by number as indicated by extinction measurements [ ; ] , the total flux of interstellar grains in the solar system should be dominated by smallest grains that are not filtered by the solar wind magnetic field .
therefore , the total flux of interstellar dust depends on the phase of the solar cycle . in the next section the flux of interstellar grains measured in - situ with the galileo and ulysses dust detectors
is compared with the temporal variation predicted by the model .
the magnitude @xmath89 of flux of interstellar dust measured by the dust detectors on - board ulysses and galileo in the time - interval @xmath90 $ ] is determined by @xmath91 , where @xmath92 is the number of interstellar grains detected in the given time - interval and @xmath93 is the average sensitive detector - area , accumulated between @xmath94 and @xmath95 .
@xmath93 is given by : @xmath96 where @xmath97 is the spacecraft rotation angle ( ulysses and galileo are spin - stabilized spacecraft , and therefore the sensitive area for a dust stream from one direction has to be averaged over the rotation angle ) , @xmath98 is the sensitive detector - area of the dust detector which is a function of time and rotation angle , @xmath99 is the dust velocity at large distances from the sun , and @xmath100 is the relative velocity of the spacecraft and the dust as a function of time . by introducing the factor @xmath101 , the effect of the flux measured in the spacecraft frame being enhanced if the spacecraft moves upwind and reduced during downwind motion is taken into account .
therefore , @xmath102 is the flux in the inertial frame at the location of the spacecraft at time @xmath103 . in the following @xmath104 predicted by the model
is compared with the flux measured with the ulysses and galileo spacecraft . for the comparison it is assumed , that the population of interstellar dust grains is dominated by grains out of a narrow size - interval which can be represented by one of the simulated grain - sizes ( and thus dynamical parameters @xmath20 and @xmath28 ) . in figure [ fig_modelfluxes ] the model prediction for the flux of interstellar grains at the location of the ulysses and galileo spacecraft
is shown as a function of time .
for the galileo measurements the simulation predicts a constant flux of interstellar grains on the detector for all but the smallest grain sizes .
this is because of the short period of time covered by the galileo data .
interstellar grains could be identified clearly in the galileo data only after it left the inner solar system in mid-1993 and before it reached the jovian system at the end of 1995 .
this time interval is short compared to the duration of the solar cycle , therefore the change of the solar cycle did not affect the flux of @xmath85- and @xmath81-grains .
the predicted increase in the flux of @xmath71-grains is due to to galileo s upstream directed motion towards the region of enhanced spatial density as can be seen in the first three panels of figure [ fig_dd01_xy ] .
the prediction for the long - term measurements of ulysses from february 1992 ( jupiter flyby ) to end of 1997 ( latest data ) shows that for @xmath71- , @xmath85 , and @xmath81-grains the effects of the solar cycle should modulate the measured flux . in the ulysses data ,
the flux of @xmath71-grains is predicted to be strongly reduced .
after 1996 the simulation predicts a decrease in the flux of the @xmath85- and @xmath105-grains .
this is due to the defocusing effect of the solar wind magnetic field on the trajectories of these grains .
the flux of larger grains is nearly constant and close to its value at large distances , because they couple weakly to the solar wind magnetic field .
figures [ fig_gal_flux_comp ] and [ fig_uls_flux_comp ] show the comparison of the temporal variation of the flux predicted by the simulation with the galileo and ulysses in - situ measurements , respectively .
since the spatial density of grains represented by one of the simulated grain sizes at large heliocentric distances is not known , the normalization of the spatial distribution is a free parameter that can be scaled by an overall factor such that the square of the difference of prediction and measurement is minimized .
if grains with dynamical parameters similar to the simulated @xmath71- , @xmath85- , or @xmath81-grains dominate the interstellar dust population measured by galileo and ulysses , the predicted temporal variation of the flux of the corresponding grain size should match the measurements .
galileo has measured an increasing flux of interstellar grains on its way to jupiter . according to the simulation
this is best explained with grains that interact strongly with the solar wind magnetic field and are thus deficient in the inner solar system .
because galileo was moving upwind to larger heliocentric distances , it detected an increasing flux .
the same deficiency of grains in the inner solar system can be created by the effect radiation pressure , but to explain an increase of flux outside @xmath106 observed by galileo , the grains must have very high @xmath20 values of @xmath107 and larger , which is not supported by the mie - calculations .
compared to the duration of the solar cycle , galileo spent a short time in the interplanetary space of the outer solar system .
therefore , the comparison of the galileo measurements with the results of the simulation is rather ambiguous and does not give any information about the effect of the changing solar cycle on the distribution of interstellar dust grains in the solar system .
since the ulysses measurements cover nearly @xmath108 years , the effect of the changing solar cycle can be observed in the ulysses data .
the measured flux stayed constant after ulysses left the ecliptic plane in february 1992 until mid-1996 .
after that the flux of interstellar grains decreased until it reached its minimal value of @xmath109 at the end of 1997 .
for @xmath71 grains the simulation does not predict a constant flux at the ulysses location between @xmath110 and @xmath111 .
furthermore , the flux of the simulated @xmath71-grains decreases by an order of magnitude within one year after mid-1996 .
this is also not observed .
therefore , the in - situ measurements as well as the simulation indicates that grains with charge - to - mass ratios of @xmath112 or larger do not contribute significantly to the flux of interstellar grains in the solar system within jupiter s orbit . for @xmath85- and @xmath81-grains ( charge - to - mass ratios between @xmath113 and @xmath2 ) the simulation predicts a flux variation that is in good agreement with the observation , though long - term measurements are necessary to confirm the agreement .
interstellar grains with lower charge - to - mass ratios do not show a decrease in flux ( see right panel of figure [ fig_modelfluxes ] ) .
this explains why bigger grains dominate the mass distribution of interstellar grains measured by ulysses after mid-1996 compared to the mass distribution measured between 1992 and mid-1996 [ ] .
the model described in section [ sec_modeldescription ] explains the observed time variability of the flux and grain mass distribution of interstellar dust in the solar system as due to electromagnetic interaction of the grains with the changing solar wind magnetic field .
it also shows that very small grains with large charge - to - mass ratios are repelled from the sun by the average electromagnetic effects . since the grain size distribution in the lic
is believed to be very steep , the most abundant interstellar grains have charge - to - mass ratios just small enough ( or , equivalently , sizes large enough ) for their inertia to overcome the electromagnetic repulsion . in this marginal case
the electromagnetic perturbation is still strong and defines the spatial distribution of the grains .
the knowledge of the spatial distribution of interstellar dust grains can be used to predict in - situ measurements by spacecraft ( e.g. cassini and stardust ) [ ] depending on their location in the solar system .
the simulation has furthermore shown that the spatial distribution of small interstellar dust grains deviates strongly from the distribution of bound , interplanetary grains .
this asymmetry as well as the time variability of the spatial distribution can be used to observe the infrared emission of interstellar dust in the solar system by telescopic means and distinguish it from the radiation emitted by interplanetary grains .
because of the low average spatial density ( about @xmath114 to @xmath115 ) of interstellar grains , an instrument of very high sensitivity is needed for such an observation .
further long - term in - situ measurements are needed to confirm the time variation of the interstellar dust flux discovered in the ulysses data .
the simulation has shown that even small interstellar dust grains reach heliocentric distances of @xmath60 .
therefore , an earth orbiting satellite can be used to measure or even collect dust grains from outside the solar system .
this work was performed while ml held a national research council - nasa / jsc research associateship and is based on ml s phd thesis `` modellierung der dynamik und interpretation der in - situ - messung interstellaren staubs in der lokalen umgebung des sonnensystems '' , ruprecht karls universitt heidelberg , 1998 .
m. baguhl , d.p .
hamilton , e. grn , s.f .
dermott , h. fechting , m.s .
hanner , j. kissel , b .- a .
lindblad , d. linkert , g. linkert , i. mann , j.a.m .
mcdonnell , g.e .
morfill , c. polanskey , r. riemann , g. schwehm , p. staubach , and h.a .
dust measurements at high ecliptic latitudes .
, 268:10161019 , 1995 .
p. c. frisch , j. dorschner , j. gei , j. m. greenberg , e. grn , m. landgraf , p. hoppe , a. p. jones , w. krtschmer , t. j. linde , g. e. morfill , w. reach , j. slavin , j. svestka , a. witt , and g. p. zank .
dust in the local interstellar wind . submitted to apj , 1999 .
e. grn , h.a .
zook , m. baguhl , a. balogh , s.j .
bame , h. fechtig , r.forsyth , m.s .
hanner , m. horanyi , j. kissel , b .- a .
lindblad , d. linkert , g. linkert , i. mann , j.a.m .
mcdonnell , g.e .
morfill , j.l .
phillips , c. polanskey , g. schwehm , n. siddique , p. staubach , j. svestka , and a. taylor .
discovery of jovian dust streams and interstellar grains by the ulysses spacecraft .
, 362:428430 , 1993 .
b..s . gustafson and s.m .
interstellar grain flow through the solar wind cavity around 1992 . in b..s .
gustafson and m.s .
hanner , editors , _ physics , chemistry and dynamics of interplanetary dust _ ,
volume 104 of _ astronomical society of the pacific conference series _ ,
pages 3539 , 1996 .
the relationship of the large - scale solar field to the interplanetary magnetic field : what will ulysses find ? in r.g .
marsden , editor , _ the sun and the heliosphere in three dimensions _ , pages 241254 .
astrophysics and space science library , 1985 .
m. witte , m. banaszkiewicz , and h. rosenbauer .
recent results on the parameters of interstellar helium from the ulysses / gas experiment . in r. von steiger , r. lallement , and m.a .
lee , editors , _ the heliosphere in the local interstellar medium _ , volume 78 of _ space science reviews _ , pages 289296 .
kluwer academic publishers , october 1996 . * [email protected] & & & mass @xmath24 @xmath116 $ ] & charge @xmath23 @xmath117 $ ] + @xmath118&@xmath119 & @xmath118&@xmath120 & @xmath118&@xmath121 & @xmath122 & @xmath123 + @xmath118&@xmath124 & @xmath118&@xmath125 & @xmath118&@xmath126 & @xmath127 & @xmath128 + @xmath118&@xmath129 & @xmath130&@xmath130 & @xmath118&@xmath131 & @xmath132 & @xmath133 + @xmath118&@xmath10 & @xmath130&@xmath134 & @xmath130&@xmath135 & @xmath136 & @xmath137 + @xmath118&@xmath130 & @xmath130&@xmath10 & @xmath124&@xmath138 & @xmath139 & @xmath140 + | the interaction of dust grains originating from the local interstellar cloud with the environment inside the heliosphere is investigated . as a consequence of this interaction the spatial distribution of interstellar dust grains
changes with time .
since dust grains are charged in the interplanetary plasma and radiation environment , the interaction of small grains with the heliosphere is dominated by their coupling to the solar wind magnetic field .
the change of the field polarity with the solar cycle imposes a temporal variation of the spatial distribution and the flux of small ( radius smaller than @xmath0 ) interstellar dust grains in the solar system , whereas the flux of large grains is constant because of their negligible coupling to the solar wind magnetic field .
the flux variation observed by in - situ measurements of the galileo and ulysses spacecraft are reproduced by simulating the interaction of interstellar grains with charge - to - mass ratios between @xmath1 and @xmath2 with the interplanetary environment . |
searches for galactic dark matter by detecting flux variations of source stars caused by gravitational microlensing have been and are being carried out by several groups ( macho : alcock et al .
1993 ; eros : aubourg et al . 1993 ;
ogle : udalski et al .
1993 ; duo : alard & guibert 1997 ) .
to increase the event rate , these searches are being conducted towards very dense star fields such as the galactic bulge and the magellanic clouds .
while searches towards these dense star fields result in an increased event rate , it also implies that the observed light curves are affected by the unwanted flux from unresolved nearby stars : blending effect .
the light curve of a microlensing event with an isolated source star is represented by @xmath4 where @xmath5 is the unlensed flux of the source star ( baseline flux ) .
the gravitational amplification is related to the lens - source separation @xmath6 normalized by the angular einstein ring radius by @xmath7^{1/2 } , \eqno(2)\ ] ] where the lensing parameters @xmath8 , @xmath9 , and @xmath10 represent the lens - source impact parameter , the time of maximum amplification , and the einstein ring radius crossing time scale ( einstein time scale ) , respectively .
these lensing parameters are determined by fitting theoretical light curves to the observed one .
one can obtain information about the lens mass @xmath11 because the einstein time scale is proportional to the square root of the lens mass , i.e. @xmath12 .
when an event is affected by blended light , on the other hand , its light curve differs from that of the unblended event by @xmath13 where @xmath14 represents the flux from blended stars .
then , to fit the observed light curve of a blended event , one should include @xmath14 as an additional parameter in addition to the three fitting parameters ( @xmath8 , @xmath15 , and @xmath10 ) of an unblended event . as a result ,
the uncertainties in the determined einstein time scale and the corresponding lens mass for a blended event are significantly larger compared to those for an unblended event ( di stefano & esin 1995 ; woniak & paczyski 1997 ; han 1997 ; alard 1997 ) . to resolve the blending problem , a newly developed technique to detect and measure light variations caused by gravitational microlensing was proposed by tomaney & crotts ( 1996 ) , alard & lupton ( 1998 ) , and alard ( 1999 ) .
this so - called difference image analysis ( dia ) method measures the variation of source star flux by subtracting observed images from a normalized reference image , i.e. @xmath16 where @xmath17 and @xmath18 represent the source star fluxes measured from the image obtained during the progress of the event and from the reference image , respectively . since not only the baseline flux of the lensed source star but also the flux from blended stars are subtracted by the dia method , the light variation measured from the subtracted image is free from the effect of blending .
since photometric precision is improved by removing the blended light , the dia method was adopted by the macho group and actually applied to microlensing searches ( alcock et al . 1999a , 1999b ) .
however , even with the dia method dramatic reduction of the uncertainties in the determined einstein time scales of gravitational microlensing events will be difficult .
this is because the dia method , by its nature , has difficulties in measuring the baseline flux @xmath5 of a source star .
unless the blended light fraction of the source star flux measured in the reference image , and thus the baseline flux @xmath19 , is determined by some other means , one still has to include @xmath14 as an additional fitting parameter .
of high amplification events , one can determine @xmath5 with uncertainties less than 50% .
] therefore , detecting the blending effect and estimating the blended light fraction in the observed source star flux is still an important issue to be resolved ( see more discussion in 2 ) .
there have been several methods proposed for the detection of the blending effect . for a high amplification event
, one can determine the unblended baseline flux of the source star from the shape of the light curve obtained by using the dia method itself ( han 2000 ) .
in addition , if the color difference between the lensed and blended stars is large , the effect of blending can be detected by measuring the color changes during the event ( buchalter , kamionkowski , & rich 1996 ) .
one can also identify the lensed source among blended stars by using high resolution images obtained from the _ hubble space telescope _
( hst ) observations ( han 1997 ) .
in addition , han & kim ( 1999 ) showed that the effect of blending can be detected from the astrometric observations of an event by using high resolution interferometers such as the _ space interferometry mission _ ( sim ) .
however , these methods either have limited applicability only for several special cases of microlensing events or impractical due to the requirement of using highly - demanding instrument for space observations . a much more practical method for the detection of the blending effect that can be applicable for general microlensing events
is provided by measuring the linear shift of the source star image centroid towards the lensed source star ( hereafter centroid shift ) during gravitational amplification ( alard , mao , & guibert 1995 ; alard 1996 ; goldberg 1998 , see more detail in 3 ) .
goldberg & woniak ( 1997 ) actually applied this method to the ogle-1 database and demonstrated the efficiency of this method by detecting centroid shifts greater than @xmath20 for nearly half of the total tested events ( seven out of 15 events ) .
however , even with this method the blending effect for an important fraction of blended events , especially for low amplification events , can not be detected due to their small centroid shifts ( han , jeong , & kim 1998 ) . in this paper , we show that if the blending effect is investigated by detecting the shift of a source star image centroid , the dia method will allow one to detect the blending effect with a significantly enhanced efficiency compared the efficiency of the current method based on psf photometry ( psf method ) .
this is because for a given event the centroid shift measurable by using the dia method , @xmath0 , is _ always _ larger than the centroid shift measurable by using the psf method , @xmath1 .
we find that the ratio @xmath2 rapidly increases with increasing fraction of blended light .
in addition , for events affected by the same fraction of blended light , the ratio @xmath3 is larger for the event with a lower amplification .
therefore , centroid shift measurements by using the dia method will be an efficient method to detect the blending effect especially of highly blended events , for which the uncertainties in the determined einstein time scale are large , as well as of low amplification events , for which the current method is highly inefficient .
even with the blending - free flux variations of a gravitational microlensing event measured by using the dia method , it will be difficult to know whether the event is affected by the blending effect or not .
this because the best - fit light curve obtained under the wrong assumption that the event is not affected by the blending effect matches well with the observed light curve .
the relations between the best - fit lensing parameters of a microlensing event resulting from the wrong determination of its source star baseline flux and their corresponding true values are provided by the analytic equations derived by han ( 1999 ) .
if a blended event is misunderstood as an unblended event , the baseline flux of the source star is overestimated by @xmath21 , causing mis - normalization of the amplification curve .
then the best - fit impact parameter @xmath22 determined from the mis - normalized amplification curve differs from the true value @xmath8 by @xmath23^{1/2};\qquad a_{\rm p } = { a_{\rm p,0}+\eta\over 1+\eta } , \eqno(5)\ ] ] where @xmath24 and @xmath25 represent the peak amplifications of the true and the mis - normalized amplification curves and @xmath26 is the fractional deviation of the mis - determined baseline flux .
mis - normalization of the amplification curve makes the best - fit einstein time scale also differ from the true value by @xmath27 here @xmath28 and @xmath29 represent the threshold impact parameter and the corresponding threshold amplification for event detection , respectively .
however , due to the mis - normalization of the amplification curve , the actually applied threshold amplification and the corresponding threshold impact parameter differ from @xmath30 and @xmath31 by @xmath32 and @xmath33^{1/2}. \eqno(8)\ ] ] figure 1 shows the degeneracy problem in the light curve measured by using the dia method . in the figure
, the solid curve presents the light variation curve when an example event with lensing parameters @xmath34 and @xmath35 is observed by using the dia method .
the event has baseline flux of @xmath36 and it is affected by the blended flux of @xmath37 .
the dotted curve represents the best - fit light curve obtained by assuming that the event is not affected by the blending effect .
the best - fit lensing parameters of the mis - normalized light curve determined by using the relations in equations ( 5)(8 ) are @xmath38 and @xmath39 , respectively .
one finds that the two light curve matches very well , implying that it will be not easy to detect the blending effect only from the light variation curve obtained by using the dia method .
in previous section , we show that detection of the blending effect for general microlensing events will be difficult even with the blending - free light variation curve obtained by using the dia method .
however , we show in this section that if the blending effect of a microlensing event is investigated by detecting the centroid shift of a source star image , the dia method will allow one to detect the blending effect with a significantly increased efficiency than the current method based on psf photometry does .
this is because for a given event the centroid shift measurable by using the dia method is _ always _ larger than the shift measurable by the psf method . for
the comparison of the centroid shifts of an event measurable by the dia and the psf methods , let us consider a blended source star image within which multiple stars with individual positions and fluxes @xmath40 and @xmath41 are included .
gravitational amplification occurs only for one of these stars .
arcsec . therefore , simultaneous lensing for multiple source stars hardly happens .
] since the position of the lensed star usually differs from the centroid of the blended image , the centroid of the source star image is shifted toward the lensed source during the event . if the lensed source star with a baseline flux @xmath42 is located at a position @xmath43 , the amount of this centroid shift , which is measurable by using the current psf method , is @xmath44 where @xmath45 represents the fractional flux of the lensed source out of the total flux of the blended stars ( including @xmath42 ) within the integrated seeing disk and @xmath46 is the position of blended image centroid before the gravitational amplification , i.e. the position of the blended source image centroid on the reference frame ( goldberg 1998 ) . on the other hand ,
_ the position of the centroid measured on the subtracted image by using the dia method is the true position of the lensed source star _ , i.e. @xmath43 .
therefore , the centroid shift measureable by using the dia method is @xmath47 then the ratio between the centroid shifts measurable by the two methods is @xmath48 .
since @xmath49 , and thus @xmath50 , for a given event the centroid shift measurable by the dia method is always larger than the shift measurable by the psf method . in figure 2 , we illustrate the centroid shifts measurable by the dia and the psf methods for visualization . on the left side , we present the contours ( measured at an arbitrary flux level ) of two source stars with identical baseline fluxes ( the inner two dotted circles with their centers marked by ` + ' ) and their integrated images ( the outer solid curve centered at ` x ' ) before ( upper part ) and during ( lower part ) gravitational amplification . among the two stars within the blended image ,
the right one is lensed . between the two left contours of integrated image
, we marked the centroid shifts measurable by the psf ( @xmath51 ) and the dia ( @xmath52 ) methods .
to better show the centroids shifts , the region enclosed by a dot - dashed line is expanded and presented on the right side .
one finds that @xmath53 .
then , how much larger is the centroid shift measurable by the dia method than the shift measurable by the dia method ?
the ratio @xmath54 depends not only on the blended light fraction @xmath55 , but also on the amplification @xmath56 of an event . to see these dependencies
, we present in figure 3 the relations between the ratio @xmath57 and the fraction of blended light for events with various impact parameters . from the figure ,
one finds the two following imporant trends .
first , the ratio @xmath58 increases rapidly with the increasing fraction of blended light .
this implies that compared to the psf method the dia method will be able to better detect the blending effect of highly blended events for which the uncertainties in the determined einstein time scales are large .
second , if events are affected by the same fraction of blended light , the ratio @xmath59 is bigger for the event with lower amplification , implying that the blending effect of low amplification events can be better detected by using the dia method . for events with low amplifications ,
the expected centroid shifts measurable by the psf method are very small , making it difficult to detect the blending effect by this method .
therefore , by increasing the detection efficiency for centroid shifts for low amplification events , the dia method will allow one to detect the blending effect for a significant fraction of galactic microlensing events .
we analyze the centroid shifts of the source star image centroid of gravitational microlensing events caused by the blending effect .
the findings from the comparison of the centroid shifts measurable by the current method based on psf photometry and the newly developed dia method are summarized as follows . 1 . for a given event
the centroid shift measurable by using the dia method is always larger than the shift measurable by using the psf method , allowing one to better detect the blending effect of microlenisng events . 2 .
the ratio between the centroid shifts measurable by using the dia and the psf methods , @xmath2 , rapidly increases as the fraction of the blended light increases .
therefore , detection of the centroid shifts by using the dia method is an efficient method for the detection of the blending effect especially for highly blended events which cause large uncertainties in the determined einstein time scales .
3 . if the blended light fraction is the same , the ratio @xmath60 is larger for the events with a lower amplification
. therefore , the dia method enables one to better detect the source star image centroid shifts of low amplification events , for which detection of their blending effect by using the psf is very difficult due to their small expected shifts .
alard , c. 1996 , in iau symp .
173 , astrophysical application of gravitational microlensing , eds . c. s. kochanek & j. n. hewitt ( dordrecht : kluwer ) , 215 alard , c. 1997 , , 321 , 424 alard ,
c. 1999 , , 343 , 10 alard , c. , & lupton , r. h. 1998 , , 503 , 325 alard , c. , mao , s. , & guibert , j. 1995 , , 300 , l17 alcock , c. , et al .
1993 , nature , 365 , 621 alcock , c. , et al . 1999a , , 521 , 602 alcock , c. , et al .
1999b , , 124 , 171 aubourg , e. , et al .
1993 , nature , 365 , 623 buchalter , a. , kamionkowski , m. , & rich , m. r. 1996 , , 469 , 676 di stefano , r. , & esin , a. a. 1995 , , 448 , l1 goldberg , d. m. 1998 , , 498 , 156 goldberg , d. m. , & woniak , p. r. 1998 , acta astron .
, 48 , 19 han , c. 1997 , , 490 , 51 han , c. 1999 , , 309 , 373 han , c. , jeong , t. , & kim , h .- i .
1998 , , 507 , 102 han , c. , & kim , t .- w . 1999 , , 305 , 795 tomaney , a. b. , & crotts , a. p. s. 1996 , , 112 , 2872 udalski , a. , et al .
1993 , acta astron . ,
43 , 289 woniak , p. , & paczynski , b. 1997 , , 487 , 55 the degeneracy problem in the light variation curve of a gravitational microlensing event observed by the dia method .
the solid curve represents the light variation curve which is expected when an example event with @xmath34 , @xmath35 , and @xmath36 is observed by using the dia method .
the dot - dashed curve represents the best - fit curve obtained under the assumption that the flux of the source star is not affected by the blending effect , despite that the source star flux in the reference image is influenced by the blended light with an amount @xmath37 .
illustration of the source star image centroid shifts measurable by usng the psf ( @xmath1 ) and the dia ( @xmath0 ) methods . on the left side ,
the contours ( measured at an arbitrary flux level ) of two source stars with identical baseline fluxes ( the inner two dotted circles with their centers marked by ` + ' ) and their integrated images ( the outer solid curve centered at ` x ' ) before ( upper part ) and during ( lower part ) gravitational amplification are presented . among the two stars within the blended image ,
the right one is lensed . between the two left contours of integrated image , the centroid shifts measurable by the psf ( @xmath61 ) and the dia ( @xmath52 ) methods
are marked .
to better show the centroids shifts , the region enclosed by a dot - dashed line is expanded and presented on the right side .
the relations between the centroid shift ratio @xmath54 and the fraction of blended light @xmath55 for events with various impact parameters . to better show the relations with @xmath62 ,
the region is enlarged and presented in a separate box . | due to the choice of very dense star fields for a higher event rate , the current microlensing searches suffer from large uncertainties caused by blending effect . to measure light variations of microlensing events free from the effect of blending , a newly developed method of differential image analysis ( dia )
was proposed for microlensing searches . however , even with the light variation curve obtained by using the dia method , dramatic reduction of the uncertainty in the determined einstein time scale is not expected due to the difficulty in determining the baseline flux of a source star .
however , we show in this paper that if the blending effect is investigated by detecting the shift of a source star image centroid , the dia method will allow one to detect the blending effect with a significantly enhanced efficiency compared the efficiency of the current method based on psf photometry ( psf method ) .
this is because for a given event the centroid shift measurable by using the dia method , @xmath0 , is _ always _ larger than the centroid shift measurable by using the psf method , @xmath1 .
we find that the ratio @xmath2 rapidly increases with increasing fraction of blended light .
in addition , for events affected by the same fraction of blended light , the ratio @xmath3 is larger for the event with a lower amplification .
therefore , centroid shift measurements by using the dia method will be an efficient method to detect the blending effect especially of highly blended events , for which the uncertainties in the determined einstein time scale are large , as well as of low amplification events , for which the current method is highly inefficient .
6.5 in submitted to _ the astrophysical journal _ : nov 10 , 1999 preprint : cnu - a&ss-11/99 |
hadronic b decays attract a lot of attention because of its role in determining the cabibbo - kobayashi - maskawa ( ckm ) matrix elements , extracting cp - violating angles and even revealing physics beyond the standard model ( sm ) .
however in most cases , a deep understanding on the strong dynamics in hadronic b decays is prerequisite for the above purposes .
phenomenologically the naive factorization ansatz ( nf ) @xcite , supported by color transparency argument @xcite , is widely used in hadronic two - body b decays .
however the unphysical dependence of the decay amplitude on renormalization scale indicates a prominent role of qcd corrections to nf . in this respect ,
@xmath11 decays are of special interest as these channels vanish in the approximation of nf , due to the spin - parity and vector current conservation .
therefore they provide a good opportunity to study the qcd corrections to nf .
it was generally believed that the branching ratios of these channels should be quite small as the qcd corrections are either suppressed by strong coupling @xmath12 or @xmath13 .
but babar @xcite and belle @xcite have found a surprisingly large branching ratio of @xmath14 decay , @xmath15 actually this large branching ratio is even comparable , for example , to that of @xmath16 decay which is not forbidden in nf .
another surprising observation is that , the upper limit of @xmath17 decay is roughly an order of magnitude smaller than the observed branching ratio of @xmath18 decay @xcite , @xmath19 while naively the branching ratios of @xmath20 decays are expected to be at the same order . in the following
we shall discuss these decay channels using the qcd factorization ( qcdf ) approach @xcite . in this framework ,
the final state light meson is described by the light - cone distribution amplitude(lcda ) , while for the @xmath21-wave charmonium @xmath22 , we shall adopt the covariant projection method of non - relativistic qcd @xcite .
it is well known that , for the inclusive decay and production of @xmath21-wave charmonia , the color - octet mechanism must be introduced to guarantee the infrared safety .
however it is still unclear how to incorporate this mechanism in a model - independent way into exclusive processes .
thus the decay amplitudes @xmath23 would be inevitably infrared divergent when only the color - singlet picture is adopted for @xmath24 , which is shown explicitly in @xcite .
thus strictly speaking , the qcdf approach is not applicable for @xmath20 decays due to the breakdown of factorization . in this paper , to get a model estimation , we will introduce the binding energy @xmath1 @xcite as an effective cut - off to regularize the infrared divergence appearing in the diagrams of vertex corrections ( see fig .
1 ) . in fact , the logarithmic divergence @xmath25 term in the limit @xmath26 for the vertex corrections in @xmath20 decays is similar to the @xmath25 term found in ref .
@xcite for the production of @xmath21-wave charmonium in @xmath27 collisions . as for the spectator scattering contributions
, there appears logarithmic divergence at twist-2 level and linear divergence at twist-3 level .
phenomenologically we shall parameterize these divergence as @xmath28 $ ] and @xmath29 respectively , where the non perturbative parameter @xmath30 again acts as an effective cut - off to regularize the endpoint divergence @xcite . according to the qcdf approach ,
all other contributions are power suppressed by @xmath13 .
we find that , with the above method , the branching ratio of @xmath14 decay is about @xmath31 , which is several times smaller than the experimental measurements . at the same time , we also get the branching ratio of @xmath32 decay at about @xmath33 , which is significantly larger than the upper limit @xmath34 observed by babar @xcite . but the above estimation is very crude in that the strong phases effects are completely ignored . notice further that for the spectator contributions , there contains only logarithmic divergence at twist-2 level , while linear divergence appears at the twist-3 level , the strong phases of the twist-2 and twist-3 spectator terms could be quite different .
we then briefly discuss the potential strong phases effects and argue that very different strong phases between twist-2 and twist-3 spectator terms together with a slightly larger @xmath8 seems to be able to reproduce the experimental hierarchy @xmath35 .
( 150,90 ) ( 0,0 ) contributions to @xmath36 decay . fig .
( a)-(d ) and fig .
( e)-(f ) are called vertex corrections and spectator scattering diagrams , respectively.,title="fig : " ]
in the qcdf approach , @xmath37 meson is described by the following light - cone projection operator in momentum space@xcite @xmath38 where @xmath39 is the momentum of @xmath37 meson and @xmath40(@xmath41 ) is the momentum of quark ( antiquark ) in k meson .
@xmath42 and @xmath43 are leading twist and twist-3 distribution amplitudes of @xmath37 meson , respectively .
it is understood that only after the factor @xmath44 in the denominator is canceled , may we take the collinear limit @xmath45 , @xmath46 .
notice that in principle we could also start directly from the original light - cone projector of k meson in coordinate space @xcite , and the physical results should be the same .
but in this case care must be taken that , only with a proper regularization , can one do the relevant convolution integrals correctly .
the readers may refer to the appendix of @xcite for further details .
since @xmath21-wave charmonium @xmath47 is involved , we shall use covariant projection method @xcite to calculate the decay amplitude @xmath48|_{q=0}~.\ ] ] here @xmath49 is the standard qcd amplitude for @xmath50 production , amputated of the heavy quark spinors , @xmath51 is the color singlet projector . while @xmath52 is the @xmath53 heavy quark spinor projector @xmath54 where @xmath21 is the momentum of charmonium and @xmath55 is the relative momentum between the @xmath56 pair in @xmath47 .
@xmath57 is the polarization tensor of @xmath22 which satisfies the following sum over polarization relation @xcite @xmath58 with @xmath59 here m is the mass of @xmath47 . for charmonium
b decays , we shall start with the effective hamiltonian @xcite @xmath60 where @xmath61 are wilson coefficients which are perturbatively calculable and @xmath62(@xmath63 ) are the effective tree ( qcd penguin ) operators . notice that we have dropped the electroweak penguin contributions here which are numerically negligible .
the four - quark effective operators are defined as @xmath64 here @xmath65 denotes all the active quarks at the scale @xmath66 , i.e. q = u , d , s , c , b. while @xmath67 and @xmath68 are color indices .
it is then straightforward to get the decay amplitude of @xmath69 decay by considering the vertex and spectator corrections drawn in fig . 1 , @xmath70 where @xmath71 is the derivative of the @xmath47 wave function at the origin and @xmath72 the form factor of @xmath73 .
the function @xmath74 represents the contributions from vertex corrections while @xmath75(@xmath76 ) arising from the twist-2 ( twist-3 ) spectator contributions .
the vertex function @xmath74 is actually infrared divergent and therefore depends on the binding energy @xmath1 . in the following
we shall keep @xmath77 term and drop the terms suppressed by @xmath78 .
the explicit expressions of @xmath79 are as follows @xmath80})}{(1 - 4a)^2\sqrt{3a } } \ln{\left [ \frac{-b}{m } \right ] } + f^i_{0\,\rm fin}+{\cal o}(b / m)~ , \nonumber \\ f^i_2&=&\frac{32 { \cal e}^{(2)*}_{\alpha \beta } p_b^\alpha p_b^\beta \sqrt{a } ( ( 1 + 12a)(1 - 4a)+16a \ln{[4a]})}{m_b(1 - 4a)^3 } \ln{\left [ \frac{-b}{m } \right ] } + f^i_{2\,\rm fin}+{\cal o}(b / m)~,\end{aligned}\ ] ] where @xmath81 , and @xmath82 is the finite part of the function @xmath74 in the limit @xmath83 .
the explicit expressions of @xmath82 are as follows , @xmath84 } + 9\ln{a } + 2 ( a(-3 + 4a)(13 - 46a + 56 a^2 ) - \nonumber \\ & & 4(1 - 2a)^3 \ln{[64a]})\ln{a } + 16(1 - 2a)^3\ln{2 } \ln{[-1 + 4a ] } ) - \nonumber \\ & & \left .
64a(1 - 2a)^3 \left ( \mbox{li2}\left [ \frac{2 - 4a}{1 - 4a}\right ] + \mbox{li2}[1 - 4a ] - \mbox{li2}\left [ \frac{1 - 2a}{1 - 4a } \right ] \right ) \right \}~ , \nonumber \\ f^i_{2\,\rm fin}&=&\frac{-{\cal e}^{(2)*}_{\alpha \beta } p_b^\alpha p_b^\beta}{4m_b \sqrt{a } } \left \ { \frac{32a}{(1 - 2a)^3(1 - 4a ) } \left ( \frac { } { } 4\ln{2}(1 - 2a)^2+(1 - 4 a ) ( 4 a^2 ( 1 + 2\ln{2 } ) - 1 ) \right
. \nonumber \\ & & \hspace*{2 cm } \left .
- 8 a ( 3 a - 1 ) ( \ln{a } - \ln{[4 a - 1]})\frac{}{}\right ) + v_{ab}[a ] \right \}~,\end{aligned}\ ] ] where the function @xmath85 $ ] denotes the finite part of vertex corrections from fig .
. the analytical form of @xmath85 $ ] is too complicated to be shown here , but numerically it has a very mild dependence on the parameter a , for example , @xmath86=11.3~ , \hspace*{1.5 cm } v_{ab}[0.15]=11.9~.\ ] ] as for the spectator functions , we have @xmath87 here @xmath88 is the momentum fraction of the light spectator quark in the @xmath89 meson , and @xmath90 the light - cone momentum fraction of the quark in the @xmath37 meson which is from the spectator quark of @xmath89 meson .
notice that our expressions for twist-2 spectator function @xmath91 are consistent with those of @xcite .
to get a numerical estimation on the branching ratios of @xmath69 decays , several parameters appearing in eqs .
( [ am ] @xmath92 [ fii ] ) should be first decided on .
the derivative of @xmath47 wave function at the origin @xmath93 may be either estimated by qcd - motivated potential models @xcite , or extracted from @xmath47 decays @xcite .
@xmath94 varies from @xmath95 to @xmath96 in different potential models @xcite while using @xmath47 decays , for instance @xcite , @xmath97 it is easy to get @xmath98 if we take @xmath99 .
this result is a little bit lower than , but still consistent with the potential model calculations , especially considering that it is very sensitive to the choice of charm quark mass . in this paper
, we shall take @xmath100 as input . for the binding energy ,
if we take @xmath101 , the ratio @xmath78 is about @xmath102 for @xmath103(@xmath104 ) , while @xmath105
. the qcd scale @xmath106 should be order of @xmath107 , as in charmless @xmath89 decays , which is about @xmath108 . in the following
we shall fix the scale @xmath109 with @xmath110 .
notice also that the wilson coefficients should be evaluated at leading order , to be consistent with the leading order formula of eq .
( [ am ] ) , @xmath111 the relevant ckm parameters are chosen to be @xmath112 and @xmath113 .
as for the spectator contributions , we adopt the following lcdas for the final @xmath37 meson , @xmath114 where @xmath115 are gegenbauer polynomials . the parameters @xmath116 are set to be @xcite : @xmath117 then logarithmic and linear divergences appear in eq.([fii ] ) , which may be phenomenologically parameterized as @xcite @xmath118 with @xmath30 .
notice that the above parametrization of linear divergence would violate the power counting of qcdf , but we do not have better way yet to deal with it .
this is clearly a very rough estimation , for example , we do not consider here the strong phase effect .
we also know little about @xmath89 wave function , but fortunately only the following integral is involved which may be parameterized as @xmath119 and we shall simply fix @xmath120 in our calculation .
the chirally enhanced ratio @xmath121 is chosen to be @xmath122 , which corresponds to taking @xmath123 and @xmath124 .
the form factor @xmath125 may be read from @xcite , in which as stated , the uncertainty of form factor at @xmath126 is likely to be smaller than that of @xmath127 , which is about @xmath128 .
therefore we will cite @xmath129 as our input .
the decay constants are set as @xmath130 and @xmath131 . with the above input , we get @xmath132 we also show separately the contributions from vertex corrections and hard spectator scattering diagrams in table i , with all the input parameters taken at their central values .
.the numerical estimations of vertex corrections and hard spectator scattering contributions , with all the parameters taken at their central values . the constant @xmath133 . [ cols="^,^,^,^",options="header " , ] for the case of @xmath2 channel , our results are approximately four times smaller than the average of babar and belle measurements , @xmath134 , while our prediction on @xmath135 decay is obviously too large compared with the experimental upper limit , @xmath136 .
this is a little bit surprising , because for charmonium b decays , the theoretical results are normally a few times _ smaller _ than the experimental measurements .
the careful reader may have noticed that in the above analysis we did not consider the uncertainty related to the parameter @xmath81 .
in fact a larger @xmath137 could enhance the branching ratio of @xmath138 decay significantly , but unfortunately it would also enhance that of @xmath17 decay with similar magnitude .
notice that @xmath20 share many common inputs , the ratio of branching ratios of these two channels should have mild dependence on the input parameters , for example it is independent on the parameter @xmath93 .
our numerical analysis shows that this is indeed the case , with @xmath139 : @xmath140 where the uncertainties arise from the parameters @xmath137 , @xmath141 , @xmath142 and @xmath143 , respectively .
the above ratio is clearly in strong contradiction with the experimental hierarchy @xmath144 .
notice that the chirally enhanced power corrections , namely twist-3 spectator contributions in this case , have been included in the above estimation . for the rest part of power corrections , there is no systematic way to estimate them yet .
but since the power corrections are suppressed by @xmath13 , intuitively they might lead to an uncertainty of about @xmath145 to the decay amplitude , which is unlikely to be able to change our estimation eq.([ratio ] ) dramatically . in our model , the parameters @xmath8 and @xmath146 will introduce additional uncertainties to @xmath20 decays .
it is very unlikely that we could reproduce the experimental observations by fine tuning @xmath146 , because although a larger @xmath146 would lead to a smaller branching ratio for @xmath147 decay , it would also make the already too small branching ratio of @xmath148 decay even smaller .
however a larger @xmath8 does help to close the gap between our predictions and the experimental data , due to the fact that a larger @xmath8 will lead to a significantly smaller branching ratio for @xmath149 decay while @xmath148 decay does not change much . of course we can not choose a too large @xmath8 , say larger than @xmath150 , because it is anyway a non perturbative parameter . as an illustration
, we take @xmath151 and get @xmath152 although it seems to be on the right way , this effort alone is still not enough to accommodate the experimental data .
let us take a closer look at the decay amplitudes . from table
i , it is clear that the spectator hard scattering mechanism is dominant in @xmath17 decay and also very important for @xmath2 channel .
furthermore there is significant destructive ( constructive ) interference between the twist-2 spectator term and the twist-3 one for @xmath148 ( @xmath149 ) mode .
this is probably the reason that we get too small @xmath2 decay as well as too large @xmath3 decay in our model .
notice that there are logarithmic and linear divergences appear in the spectator contributions , which are parameterized by eq .
( [ endpoint ] ) .
it is obviously a very rough model estimation and for example , strong phases effects are completely ignored .
it is also reasonable to assume that the strong phase of twist-2 spectator term could be different from that of twist-3 part . as an illustration
, the endpoint divergences could be parameterized as @xcite : @xmath153 with @xmath154 and the phase @xmath155 completely free . in the above equations
, @xmath156 denotes the parameters for twist-2 spectator term and @xmath157 for twist-3 one . in this case , the interference effects and therefore the predictions of the branching ratios , could be changed dramatically .
for example , if we take a somewhat extreme case @xmath158 with @xmath159 while keep all other input parameters fixed at their central values , we will get @xmath160 which are in good agreement with the experimental observations .
certainly , due to the non perturbative nature of the above strong phases , there is strong model dependence of our predictions .
therefore it is not so meaningful to fine tune the parameters to get the best fit of the experimental data .
the key point here is that , different strong phases between twist-2 and twist-3 spectator terms might be able to account for the experimental hierarchy that @xmath161 is at least an order of magnitude larger than @xmath162 .
the authors of ref .
@xcite also studied the @xmath163 decay with the same qcdf method .
but they used the gluon mass and gluon momentum cutoff , instead of binding energy adopted in this paper , to regularize the infrared divergences of the vertex corrections .
another difference is that they calculated the spectator contributions directly from the original light - cone projector of k meson in coordinate space and got a different result from this paper .
they claimed that the difference was due to the light - cone projector adopted in this paper which is inappropriate for @xmath164 channels : to get the projector eq.([projector ] ) from the original one in coordinate space @xcite , the integration by parts has been used and the boundary terms were dropped .
however because of the linear singularities appeared in the above calculations , the boundary terms seem to be divergent and thus the justification of using the integration by parts is in doubt in this case .
but beneke has elaborated on this subtle point in the appendix of ref . @xcite and it is shown there that the boundary terms are indeed zero provided the propagators are regularized carefully when they go close to the mass - shell
. therefore the integration by parts can be used here and the light - cone projector adopted in this paper is justified .
certainly , with a proper regularization , the calculation starting directly from the coordinate space projector should give the same results as this paper .
most recently , the @xmath165 decay was discussed by using the pqcd method @xcite .
notice that the vertex corrections were not included in their calculations , and the spectator contributions alone are enough in their paper to account for the experimental data .
it would be very interesting to see whether they could also reproduce the very small branching ratio of @xmath149 channel observed by babar , which has not been done yet .
the @xmath165 decay was also analyzed with light - cone sum rules @xcite . although there are some discrepancies in their papers , they agreed on the point that their results were too small to accommodate the experimental data .
a large charmed meson rescattering effects @xmath166 could account for the surprisingly large @xmath165 decay@xcite , but generally it will also lead to a large branching ratio for @xmath149 mode . in summary ,
we discuss in this paper the vertex corrections and spectator hard scattering contributions to @xmath20 decays .
since there is no model independent way yet to estimate the color - octet contribution to exclusive processes , it is no wonder that the vertex corrections here are infrared divergent .
the non - zero binding energy @xmath167 makes the charm quark slightly off - shell inside @xmath47 , and effectively acts as a cut - off to regularize the vertex part .
there are also less serious logarithmic and linear endpoint divergences which appears in the spectator contributions and are parameterized in a model - dependent way as usually done in charmless b decays .
this means that the spectator diagrams are actually dominated by soft gluon exchange , which in a sense could be viewed as a model estimation of color - octet contributions .
then our numerical analysis predicts the branching ratio of @xmath5 decay to be about @xmath168 , about four times smaller than the experimental observations , while for @xmath169 decay , we get @xmath170 , which is about five times larger than the experimental upper limit . but concerning the large theoretical uncertainties , it is more interesting to consider the ratio @xmath171 , in which a large part of the theoretical uncertainty can be eliminated
. numerically we find the ratio to be @xmath172 , in sharp contrast to the experimental observation that this ratio should be about @xmath7 or even smaller , if the babar analysis of the upper limit of @xmath173 decay will be confirmed by farther measurements .
we then have a closer look at the decay amplitudes .
one observation is that , @xmath148 channel is not very sensitive to the spectator infrared cutoff parameter @xmath8 while a larger @xmath8 could reduce the branching ratio of @xmath149 decay significantly .
another observation is that , in our model there is large destructive ( constructive ) interference between the twist-2 and twist-3 spectator terms for @xmath148 ( @xmath149 ) mode . but notice that the twist-2 spectator contributions contain only logarithmic endpoint divergence while twist-3 ones contain more severe linear endpoint divergence , it should be reasonable to assume that their strong phases could be quite different .
since the interference effects are very sensitive to the strong phases difference , this might change our model predictions dramatically .
as an illustration , we then show in an explicit case that , with a slightly larger @xmath8 and large strong phases difference between twist-2 and twist-3 spectator terms , our predictions are in good agreement with the experimental data . in conclusion , what we have shown in this paper
, is that by adjusting the parameters for the spectator hard scattering contributions , as with the annihilation terms for charmless @xmath89 decays , qcdf is able to produce appreciable non factorizable contributions to @xmath174 decays close to experiments , in contrast with the @xmath175 decay which needs a large factorizable contribution in addition to the small non factorizable one obtained in qcdf @xcite .
g.z acknowledges the support from alexander - von - humboldt stiftung . | in this paper , we investigate the vertex corrections and spectator hard scattering contributions to @xmath0 decays , which has no leading contribution from naive factorization scheme .
a non - zero binding energy @xmath1 is introduced to regularize the infrared divergence of the vertex part .
the spectator diagrams also contain logarithmic and linear infrared divergences , for which we adopt a model dependent parametrization .
if we neglect possible strong phases in the hard spectator contributions , we obtain a too small branching ratio for @xmath2 while too large one for @xmath3 , as can be seen from the ratio of the branching ratio of @xmath4 to that of @xmath5 , which is predicted to be @xmath6 in our model , while experimentally it should be about @xmath7 or even smaller . but
a closer examination shows that , assuming large strong phases difference between the twist-2 and twist-3 spectator terms , together with a slightly larger spectator infrared cutoff parameter @xmath8 , it is possible to accommodate the experimental data .
this shows that , for @xmath9 decays with no factorizable contributions , qcdf seems capable of producing decay rates close to experiments , in contrast to the @xmath10 decay which is dominated by the factorizable contributions . |
this paper deals with the justification of the two - scale asymptotic expansions method applied to a thermo - diffusion problem arising in the context of transport of densities of hot colloids in media made of periodically - distributed microstructures .
following @xcite , we study a system of two coupled semi - linear parabolic equations , where the diffusivity for the concentration @xmath1 is of order @xmath2 and for the temperature @xmath3 it is of order @xmath4 . here
@xmath5 denotes the characteristic length scale of the underlying microstructure .
we rigorously justify the expansions @xmath6 and @xmath7 and prove an estimate of the type @xmath8 where @xmath9 denotes the macroscopic domain and @xmath10 is the perforated reference cell .
estimate basically gives a quantitative indication of the speed of the ( two - scale ) convergence between the unknowns of our problem and their limits , which is detailed in the forthcoming sections .
this work follows up previous successful attempts of deriving quantitative corrector estimates using periodic unfolding ; see e.g. @xcite .
the unfolding technique allows for homogenization results under minimal regularity assumptions on the data and on the choice of allowed microstructures .
the novelty we bring in here is the combination of three aspects : ( i ) the asymptotic procedure refers to a suitably perforated domain , ( ii ) presence of a cross coupling in gradient terms , and ( iii ) lack of compactness for @xmath3 .
our working techniques combines @xmath0-independent a priori estimates for the solutions and periodic unfolding - based estimates such as the _ periodicity defect _ in @xcite and the _ folding mismatch _ in @xcite .
estimate improves existing convergence rates for semi - linear parabolic equations with possibly non - linear boundary conditions in @xcite or small diffusivity in @xcite from @xmath11 to @xmath12 .
this improvement is obtained by studying all equations in the two - scale space @xmath13 and by suitably rearranging and controlling occurring error terms @xmath14 .
it is worth noting that the availability of corrector estimates for the thermo - diffusion system allows in principle the construction of rigorously convergent multiscale numerical methods ( for instance based on msfem like in @xcite ) to capture thermo - diffusion effects in porous media .
interestingly , for the thermo - diffusion system posed in perforated domains such convergent multiscale numerical methods are yet unavailable .
the paper is structured as follows : in section [ sec : model ] , we introduce the thermo - diffusion model and prove existence as well as a priori estimates for the solutions of the microscopic problem respective the two - scale limit problem . the periodic unfolding method and auxiliary corrector estimates are presented in section [ subsec : two - scale ] and [ subsec : auxiliary ] , respectively .
finally , the corrector estimates in are proved in section [ subsec : proof ] .
we conclude our paper with a discussion in section [ sec : discuss ] .
we investigate a system of reaction - diffusion equations which includes mollified cross - diffusion terms and different diffusion length scales .
the cross - diffusion terms are motivated by the incorporation of soret and dufour effects as outlined in @xcite . for more information on phenomenological descriptions of thermo - diffusion ,
we refer the reader to @xcite .
the concentrations of the transported species through the perforated domain @xmath15 are denoted by @xmath1 , while @xmath3 is the temperature . the overall interplay between transport and reaction is modeled here by the following system of partial differential equations : @xmath16 supplemented with the neumann boundary conditions @xmath17 and the initial conditions @xmath18 first of all , it is important to note that the @xmath0-scaling for some of the terms in the system is variable with @xmath19 .
we refer to the suitably scaled heat conduction - diffusion interaction terms @xmath20 and @xmath21 as `` weak thermal couplings '' , while the `` high - contrast '' is thought here particularly in terms of the heat conduction properties of the composite material that can be seen in @xmath22 . in this context , @xmath23 denotes the normal outer unit vector of @xmath15 . the matrix @xmath24 is the diffusivity associated to the concentration of the ( diffusive ) species @xmath1 ,
@xmath25 is the heat conductivity , while @xmath26 and @xmath27 are the soret and dufour coefficients .
note that @xmath24 , @xmath25 , @xmath28 , and @xmath29 are either positive definite matrices , or they are positive real numbers .
furthermore , the reaction term @xmath30 models the smoluchovski interaction production . in the original model from @xcite ,
the function @xmath31 is an additional unknown modeling the mass of deposited species on the pore surface @xmath32 , and it is shown to possess the regularity @xmath33 .
here we assume @xmath31 as given data .
we point out that the linear boundary terms are relevant for the regularity of solutions , but that they are not required to prove the convergence rate of order of @xmath34 in . to deal with perforated domains we employ the method of periodic unfolding as presented in @xcite .
let @xmath35 denote the standard unit - cell .
we fix here and for all the following assumptions on the domain and the microstructure .
[ assump : domain ] our geometry is designed as follows : the domain @xmath36 is a @xmath37-polytope with length @xmath38 for all @xmath39 .
the reference hole @xmath40 is an open lipschitz domain and the perforated cell @xmath41 satisfies @xmath42 .
moreover @xmath43 is a connected lipschitz domain and @xmath44 is identical on all faces of @xmath45 .
the set of all nodal points is given via @xmath46 .
with this we define the pore part @xmath47 and the perforated domain @xmath15 , which is connected , via @xmath48 where @xmath49 denotes the interior of the set @xmath50 .
both sets are open and form together the original domain @xmath51 .
[ fig : cell ] the assumptions on the domain guarantee the existence of suitable extensions from @xmath15 to @xmath52 ( cf .
theorem [ thm : extension ] ) . also traces exist and are well - defined on the boundaries @xmath53 and @xmath54 . with this ,
perforated domains with isolated holes as well as the prominent `` pipe - model '' for porous media are included in our considerations , see figure [ fig : cell ] .
the boundary of the perforated domain @xmath15 is given by @xmath55 .
indeed , intersected pore structures at the boundary @xmath56 as in figure [ fig : cell](ii ) are not excluded .
[ rem : boundary ] in the following we denote by @xmath0 a sequence @xmath57 of numbers satisfying @xmath58 .
this implies that all microscopic cells @xmath59 , for @xmath60 , are contained in @xmath15 and no intersected cells occur at the boundary @xmath61 .
this assumption ( tremendously ) simplifies the presentation in this paper , however , we believe that the same results can be obtained for lipschitz domains @xmath52 by considering a bigger @xmath37-polytope @xmath62 with @xmath63 . then , all relevant coefficients , functions , and solutions are suitably extended from @xmath15 to @xmath62 .
[ assump : data ] we impose the following restrictions on the data : the diffusion matrices @xmath24 and @xmath25 are given via @xmath64 where @xmath65 are symmetric and uniformly elliptic , i.e. @xmath66 the constants @xmath67 are non - negative .
the reaction term @xmath68 is globally lipschitz continuous , i.e. @xmath69 for all @xmath70 .
the sink / source term @xmath31 is given via @xmath71 for any data @xmath72 ; ( \mathrm{w}^{1,\infty}({\omega } ; { \mathrm{l}}^2(\partial t)))$ ] .
here we denote with @xmath73 the scalar product of vectors in @xmath74 and set @xmath75 for technical reasons , we introduce the mollified gradient @xmath76 which is given as follows : for @xmath77 , we introduce the mollifier @xmath78 where the constant @xmath79 is selected such that @xmath80 . using @xmath81 we define for @xmath82 the mollified gradient @xmath83,\end{aligned}\ ] ] where @xmath84 denotes the ball centered at @xmath85 with radius @xmath86 . according to (
c.4 ) there holds @xmath87 and @xmath88 we assume throughout this text that @xmath0 and @xmath86 are chosen such that @xmath89 holds .
this assumption arises in lemma [ lemma : prop - t-2 ] .
now , let us consider the case @xmath90 .
this subsection and the next one are devoted the existence of weak solutions to our target problem .
[ thm : exist - eps ] let the assumptions [ assump : domain ] and [ assump : data ] hold and let the initial condition @xmath91 satisfy @xmath92 then there exists for every @xmath93 a unique solution @xmath94 of with @xmath95 moreover the solution is non - negative , i.e. @xmath96 almost everywhere in @xmath97\times { \omega}_{\varepsilon}$ ] , and uniformly bounded @xmath98 where the constants @xmath99 are independent of @xmath0 .
the existence of solutions , non - negativity , and uniform boundedness follow from the lemmata 3.2 3.6 and theorem 3.8 in @xcite by replacing @xmath25 and @xmath100 with @xmath101 and @xmath102 , respectively .
note that the proof can be generalized from diffusion coefficients @xmath103 to symmetric matrices as in assumption [ assump : data](i ) . in equation ( 35 )
respective ( 57 ) in @xcite it holds for @xmath104 and @xmath105 : @xmath106 this argumentation also requires linear boundary terms .
otherwise one has to argue as in ( * ? ? ?
3.2 ) or ( * ? ? ?
* prop . 1 ) and
differentiate the whole equation with respect to time and then use a second grnwall argument .
since our solutions are uniformly bounded in @xmath107 , we may consider reaction terms with arbitrary growth as in @xcite .
also note that estimate remains valid for all @xmath108 and @xmath109 . for the parameters @xmath90
, we obtain in the limit @xmath110 the following two - scale system @xmath111 supplemented with the boundary conditions @xmath112 and the initial conditions @xmath113 here @xmath23 and @xmath114 denote the normal outer unit vector of @xmath52 and @xmath43 , respectively . to capture the oscillations in the limit
we define the space of @xmath45-periodic functions @xmath115 via @xmath116 where @xmath117 and @xmath118 are opposite faces of the unit cube @xmath45 with @xmath119 . with this
the effective coefficients are given via the standard unit - cell problem @xmath120 \cdot [ { \nabla_{\!y}}\phi { + } \xi ] { \,\mathrm{d}}y .\end{aligned}\ ] ] note that the integral is taken over @xmath121 and not the average @xmath122 .
in full , formula reads @xmath123 with @xmath124 here .
for the boundary data @xmath31 , we obtain in the limit @xmath110 the usual average @xmath125\times{\omega}:\quad v_0(t , x ) = \dashint_{\partial t } \mathbb{v}(t , x , y ) { \,\mathrm{d}}y .\end{aligned}\ ] ] finally , we state the existence and uniqueness of solutions for the limit system .
[ thm : exist - limit ] let the assumptions [ assump : domain ] and [ assump : data ] hold and let the initial value @xmath126 satisfy @xmath127 and @xmath128 . there exists a unique solution @xmath129 of with @xmath130 the existence and boundedness of unique solutions @xmath129 follows by galerkin approximation as in @xcite . in particular , the higher @xmath131-regularity of @xmath132 follows by ( * ? ? ?
* thm . 5 ) .
by slightly modifying the proof of ( * ? ? ?
4 ) after equations ( 40)(42 ) , the assumptions on the initial values can be relaxed from @xmath133 and @xmath134 to @xmath135 and @xmath136 . to prove the @xmath137-estimates for the gradients and the @xmath138-estimates for the time derivative we can argue as in @xcite by exploiting the symmetry of @xmath139 and @xmath140 as in as well as the fact that the boundary terms are linear .
the usual two - scale decomposition is given via the mappings @xmath141 : { \mathbb{r}}^d \to \mathbb{z}^d$ ] and @xmath142 .
for @xmath143 , @xmath144 $ ] denotes the component - wise application of the standard gauss bracket and @xmath145 $ ] is the remainder . with this , the _ periodic unfolding operator _
@xmath146 , for @xmath147 , is defined by ( ( * ? ? ? * def . 2.3 ) ) @xmath148 + { \varepsilon}y \right ) .\end{aligned}\ ] ] note that we do not need to extend @xmath149 by @xmath150 outside @xmath15 since there occur no intersected cells at the boundary @xmath151 , cf .
also remark [ rem : boundary ] .
we have indeed @xmath152 + y ) \in { \omega}_{\varepsilon}$ ] for all @xmath153 such that @xmath154 is well - defined in . in the same manner
we define the _ boundary unfolding operator _ @xmath155 by ( ( * ? ? ?
* def . 5.1 ) ) @xmath156 + { \varepsilon}y \right ) .\end{aligned}\ ] ] following @xcite we define the _ folding ( averaging ) operator _ @xmath157 via @xmath158 + y _ * \right ) } u\left ( z , \lbrace\tfrac{x}{{\varepsilon}}\rbrace \right ) { \,\mathrm{d}}z \;\bigg\vert_{{\omega}_{\varepsilon}},\end{aligned}\ ] ] where @xmath159 denotes the usual average and @xmath160 is the restriction of @xmath149 to @xmath15 . to derive quantitative estimates for the differences @xmath161 and @xmath162
, we need to test the weak formulation of the original system with @xmath163-functions which are one - scale pendants of the limiting solution @xmath129 .
there are two options to naively fold a two - scale function @xmath164 , namely @xmath165
however @xmath1 is only well - defined in @xmath163 , if at least @xmath166 belongs to @xmath167 , and our limit @xmath129 ( respective the corrector @xmath168 for @xmath169 ) does not satisfy strong differentiability in general .
the second option @xmath170 is neither a suitable test function , since it is not @xmath163-regular . to overcome this regularity issue
, we define the _ gradient folding operator _ following @xcite and adapt its definition to perforated domains .
the gradient folding operator @xmath171 is defined as follows : for every @xmath172 , the function @xmath173 is given in @xmath163 as the solution of the elliptic problem @xmath174 note that @xmath175 is uniquely determined by the lax
milgram lemma implying the well - definedness of @xmath176 . for simplicity
, we define for @xmath177 the norm @xmath178 where the second identity follows from lemma [ lemma : prop - t ] .
both folding operators , @xmath179 and @xmath176 , are linear and bounded operators satisfying @xmath180 where the first estimate is due to jensen s inequality , while the second one is due to hlder s inequality .
we are now collecting several results which are essential ingredients in the proof of our error estimates .
note that @xmath105 also belongs to the space @xmath163 since @xmath181 and we can apply the unfolding operator via @xmath182 , where @xmath183 denotes the characteristic function of the set @xmath15 .
for the sake of brevity @xmath183 is omitted in the following .
[ lemma : unfold - err ] for all @xmath184 and @xmath105 we have @xmath185 respectively , where @xmath79 only depends on the domains @xmath52 and @xmath43 .
the proof for the first estimate is based on the application of the poincar
wirtinger inequality on each cell @xmath59 , see ( * ? ? ?
3.1 ) or ( * ? ? ?
2.3.4 ) with @xmath186 .
the second estimate follows from the first one with @xmath187 cf .
also ( * ? ? ?
* eq . ( 3.4 ) ) .
note that @xmath188 is indeed well - defined for one - scale functions . to control the mollified gradient we prove : [ lemma : prop - t-2 ] for @xmath89 and all @xmath189 and @xmath153 we have @xmath190 where @xmath191 depends on the mollifier @xmath81 and @xmath43 .
according to ( * ? ? ?
* thm . 6 ) we obtain for every @xmath153 @xmath192(x , y ) & = \left[\operatorname{\mathcal{t}_{\varepsilon}}\left ( \int_{b(x,\delta ) } \nabla j_\delta(x - \xi)u(\xi ) { \,\mathrm{d}}\xi \right ) \right ] ( x , y ) \\ & = \int_{b \left ( { \varepsilon}\left[\tfrac{x}{{\varepsilon } } \right ] + { \varepsilon}y,\delta \right ) } \nabla j_\delta({\varepsilon}[\tfrac{x}{{\varepsilon } } ] { + } { \varepsilon}y - \xi)u(\xi ) { \,\mathrm{d}}\xi . \end{aligned}\ ] ] for @xmath89 , we define the following @xmath37-dimensional annulus @xmath193 of thickness @xmath194 and with volume @xmath195 .
we arrive at @xmath196 + { \varepsilon}y,\delta \right ) } \nabla j_\delta({\varepsilon}[\tfrac{x}{{\varepsilon } } ] { + } { \varepsilon}y - \xi)u(\xi ) { \,\mathrm{d}}\xi - \int_{b(x,\delta ) } \nabla j_\delta(x - \xi)u(\xi ) { \,\mathrm{d}}\xi \right\vert \\ & \leq \left\vert \int_{b_\mathrm{diff } } \nabla j_\delta(x - \xi)u(\xi ) { \,\mathrm{d}}\xi \right\vert \leq \vert \nabla j_\delta \vert_{{\mathrm{l}}^2(b_\mathrm{diff } ) } \vert u\vert_{{\mathrm{l}}^2(b_\mathrm{diff } ) } \leq \sqrt{{\varepsilon } } c \vert j_\delta
\vert_{\mathrm{c}^\infty({\mathbb{r}}^d ) } \vert u\vert_{{\mathrm{l}}^2({\omega } ) } , \end{aligned}\ ] ] which proves the assertion .
having defined two folding operators , @xmath179 being dual to @xmath154 and @xmath176 assuring @xmath197-regularity , we call their difference _ folding mismatch _ and control it as follows . [ thm : fold - mismatch ] for @xmath79 only depending on @xmath52 and @xmath43 it holds @xmath198 the proof is based on ( * ? ? ?
3.2 ) and adapted to perforated domains in appendix [ app : appendixb ] .
since unfolded sobolev functions @xmath199 are in general not @xmath45-periodic , we need to control the so - called _ periodicity defect _ ,
cf . @xcite . in the case of slow diffusion
it reads : [ thm : per - defect ] for every @xmath200 , there exists a @xmath45-periodic function @xmath201 such that @xmath202 where the constant @xmath79 only depends in the domains @xmath52 and @xmath43 .
the proof relies on ( * ? ? ?
* thm.2.2 ) which we can apply after suitably extending @xmath203 from the perforated domain @xmath15 to the whole domain @xmath52 .
let @xmath204 denote the extension of @xmath203 as in theorem [ thm : extension ] . according to ( * ? ? ?
* thm.2.2 ) , there exists a two - scale function @xmath205 satisfying @xmath206 with @xmath79 only depending on @xmath52 , @xmath45 , and @xmath207 defined on the whole unit - cell as in , cf .
also @xcite .
note that it holds @xmath208 .
recalling the definition of @xmath209 in with @xmath210 , we define @xmath201 via @xmath211 , which gives @xmath212 and the proof is finished .
for the case of classical diffusion , we consider @xmath177 instead of @xmath163 .
this is related to the fact that , in the limit system , the @xmath149-equation is given in the macroscopic domain @xmath52 , whereas the @xmath132-equation as posed in the two - scale space @xmath13 , and hence , it can not be reduced to @xmath52 only . [ thm : per - defect-2 ] for every @xmath177 , there exists a @xmath45-periodic function @xmath201 such that @xmath213 where the constant @xmath79 only depends in the domains @xmath52 and @xmath43 .
for @xmath177 the desired estimates hold with @xmath205 according to ( * ? ? ? * thm.2.3 ) . choosing @xmath214 as in the proof of theorem [ thm :
per - defect ] yields the assertion .
having collected all preliminaries , we can now state and prove the corrector estimates for our thermo - diffusion model .
[ thm : main ] let @xmath94 and @xmath129 denote the unique solution of @xmath215 and @xmath216 , respectively , according to theorem [ thm : exist - eps ] and theorem [ thm : exist - limit ] . if the initial values satisfy @xmath217 then we have @xmath218 where the constant @xmath79 depends on the given data and the norms in and .
note that the domain @xmath52 is convex , bounded , and has a lipschitz boundary .
since @xmath219 and @xmath220 belong to the space @xmath221 , we can apply ( * ? ? ?
3.2.1.3 ) and obtain that the limit @xmath222 belongs to the better space @xmath223 . if not stated otherwise , the following notion of weak formulation is to be understood pointwise in @xmath97 $ ] . *
part a : slow diffusion .
* note that for @xmath224 and @xmath225 the following two norms are equivalent up to an error of order @xmath226 , i.e.@xmath227 which is due to @xmath228 by lemma [ lemma : unfold - err ] .
_ step 1 : reformulation of @xmath3-equation . _
the weak formulation of the @xmath3-equation reads @xmath229 for all admissible test functions @xmath230 .
applying the periodic unfolding operators @xmath154 and @xmath231 , with @xmath232 , and exploiting their properties in lemma [ lemma : prop - t ] and [ lemma : prop - t-2 ] gives @xmath233 we choose @xmath234 in , which is by construction of the gradient folding operator @xmath176 an admissible test function in @xmath163 so that @xmath235 + \mu { \nabla_{\!y}}(\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon } ) \cdot \operatorname{\mathcal{t}_{\varepsilon}}({\nabla^\delta}u_{\varepsilon } ) \operatorname{\mathcal{t}_{\varepsilon}}(\theta_{\varepsilon}{- } \operatorname{\mathcal{g}_{\varepsilon}}\theta ) { \,\mathrm{d}}x
{ \,\mathrm{d}}y \nonumber \\ & \quad + \int_{{\omega}\times\partial t } g \operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}\theta_{\varepsilon}\operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}(\theta_{\varepsilon}{- } \operatorname{\mathcal{g}_{\varepsilon}}\theta ) { \,\mathrm{d}}x{\,\mathrm{d}}\sigma(y ) .\end{aligned}\ ] ] adding @xmath236 respective @xmath237 gives @xmath238 where the folding mismatch @xmath239 reads @xmath240 to treat the boundary term , we exploit the continuous embedding @xmath241 , i.e. @xmath242 using the @xmath76-estimates in as well as the boundedness of the solution @xmath94 in , in particular , the improved time - regularity @xmath243 and @xmath244 by , gives @xmath245 inserting @xmath246 respective @xmath247 , applying the triangle inequality , and using the norm preservation of @xmath154 gives @xmath248 using the higher @xmath131-regularity @xmath249 , applying proposition [ thm : fold - mismatch ] for the folding mismatch , and lemma [ lemma : unfold - err ] for the unfolding error gives @xmath250 _ step 3 : reformulation of @xmath132-equation .
_ the weak formulation of the @xmath132-equation reads @xmath251 for all admissible test functions @xmath252 .
we choose @xmath253 according to proposition [ thm : per - defect ] such that we can control the periodicity defect of @xmath254 for arbitrary functions @xmath230 , namely @xmath255 where the periodicity defect @xmath256 is given via @xmath257 applying hlder s inequality and the embedding yield @xmath258 according to the higher @xmath131-regularity of @xmath132 in it is @xmath259 with @xmath260 such that we can apply theorem [ thm : per - defect ] and obtain
@xmath261 choosing @xmath234 in yields @xmath262 adding @xmath236 , and respectively @xmath263 , as in step 1 gives @xmath264 where the folding mismatch @xmath265 is determined by @xmath266 the estimation of @xmath265 follows along the lines of @xmath239 in step 1 using the boundedness of the limit @xmath129 , in particular , the boundedness of @xmath267 . finally , we insert the test function @xmath234 into the @xmath268-estimate in and apply young s inequality with @xmath269 @xmath270 _ step 4 : derivation of grnwall - type estimates . _
subtracting equation from and using @xmath271 gives @xmath272 \cdot [ { \nabla_{\!y}}(\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta ) ] \nonumber \\ & \hspace{50pt } + \mu [ { \nabla_{\!y}}(\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon } ) \cdot \operatorname{\mathcal{t}_{\varepsilon}}({\nabla^\delta}u_{\varepsilon } ) - { \nabla_{\!y}}\theta\cdot{\nabla^\delta}u](\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta ) \big\rbrace { \,\mathrm{d}}x{\,\mathrm{d}}y \nonumber \\ & \quad + \int_{{\omega}\times\partial t } g |\operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}\theta_{\varepsilon}{- } \theta |^2 { \,\mathrm{d}}x{\,\mathrm{d}}\sigma(y ) \nonumber \\ & \quad + \delta^{\theta_{\varepsilon}}_\mathrm{fold } - \delta^{\theta}_\mathrm{per , fold } .\end{aligned}\ ] ] we continue by estimating each term on the right - hand side in separately .
exploiting the interpolation inequality ( cf .
e.g. @xcite ) @xmath273 and then young s inequality with @xmath274 lead to @xmath275 reformulating the @xmath29-term gives @xmath276(\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta ) { \,\mathrm{d}}x{\,\mathrm{d}}y \nonumber \\ \label{eq : slow - gronwall - mu-1 } & = \int_{{\omega}\times y _ * } \mu { \nabla_{\!y}}(\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon } ) \cdot [ \operatorname{\mathcal{t}_{\varepsilon}}({\nabla^\delta}u_{\varepsilon } ) { - } { \nabla^\delta}u](\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta ) { \,\mathrm{d}}x{\,\mathrm{d}}y \\ \label{eq : slow - gronwall - mu-2 } & \quad + \int_{{\omega}\times y _ * } \mu { \nabla_{\!y}}(\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta ) \cdot{\nabla^\delta}u(\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta ) { \,\mathrm{d}}x{\,\mathrm{d}}y .\end{aligned}\ ] ] using in that @xmath277 is bounded as well as hlder s and young s inequality with @xmath278 gives @xmath279 in a similar manner we obtain for by adding @xmath280 and using estimate @xmath281(\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta ) { \,\mathrm{d}}x{\,\mathrm{d}}y \right\vert \nonumber\\ & \leq \mu \vert { \nabla_{\!y}}(\operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon } ) \vert_{{\mathrm{l}}^2({\omega}\times y _ * ) } \vert \operatorname{\mathcal{t}_{\varepsilon}}({\nabla^\delta}u_{\varepsilon } ) { - } { \nabla^\delta}u \vert_{{\mathrm{l}}^\infty({\omega}\times y _ * ) } \vert \operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta \vert_{{\mathrm{l}}^2({\omega}\times y _ * ) } \nonumber\\ & \leq c \vert \operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta \vert_{{\mathrm{l}}^2({\omega}\times y _ * ) } \left\lbrace \vert \operatorname{\mathcal{t}_{\varepsilon}}({\nabla^\delta}u_{\varepsilon } ) { - } \operatorname{\mathcal{t}_{\varepsilon}}({\nabla^\delta}u ) \vert_{{\mathrm{l}}^\infty({\omega}\times y _ * ) } + \vert \operatorname{\mathcal{t}_{\varepsilon}}({\nabla^\delta}u ) { - } { \nabla^\delta}u \vert_{{\mathrm{l}}^\infty({\omega}\times y _ * ) } \right\rbrace \nonumber\\ & \leq c_\delta \vert \operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta \vert_{{\mathrm{l}}^2({\omega}\times y _ * ) } \left\lbrace \vert u_{\varepsilon}{- } u \vert_{{\mathrm{l}}^2({\omega}_{\varepsilon } ) } + \sqrt{{\varepsilon } } \vert u \vert_{{\mathrm{l}}^2({\omega } ) } \right\rbrace \nonumber\\ & \leq c \left\lbrace \vert \operatorname{\mathcal{t}_{\varepsilon}}\theta_{\varepsilon}{- } \theta \vert^2_{{\mathrm{l}}^2({\omega}\times y _ * ) } + \vert u_{\varepsilon}{- } u \vert^2_{{\mathrm{l}}^2({\omega}_{\varepsilon } ) } + { \varepsilon}\vert u \vert^2_{{\mathrm{l}}^2({\omega } ) } \right\rbrace .\end{aligned}\ ] ] overall we can estimate equation with the uniform ellipticity of @xmath140 and such that @xmath282 choosing @xmath283 , integrating over @xmath284 $ ] with @xmath285 , as well as recalling and yields @xmath286 * part b : classical diffusion .
* we point out that the higher regularity of the limit @xmath287 implies the higher @xmath131-regularity of the corrector @xmath288 which is the unique minimizer of the unit - cell problem with @xmath289 .
_ step 1 : reformulation of @xmath1-equation . _
the weak formulation of the @xmath1-equation is given via @xmath290 for all test functions @xmath200 .
first of all note that the cross - diffusion term @xmath291 is of order @xmath226 thanks to hlder s inequality and the boundedness in and @xmath292 applying the unfolding operators @xmath154 and @xmath231 , in particular , rewriting @xmath293 and @xmath294{+}y),y)$ ] , and using the properties in lemma [ lemma : prop - t ] gives @xmath295 wherein we replaced the boudnary term @xmath296 with @xmath297 and created the approximation error @xmath298 using that @xmath299 { + } y)| \leq { \varepsilon}\mathrm{diam}(y)$ ] holds for all @xmath300 , we obtain the pointwise estimate @xmath301 thanks to the lipschitz continuity of @xmath302 .
together with embedding we obtain for the approximation error @xmath303 we choose the test function @xmath304 in such that @xmath305 + r(\operatorname{\mathcal{t}_{\varepsilon}}u_{\varepsilon } ) ( \operatorname{\mathcal{t}_{\varepsilon}}u_{\varepsilon}{- } u ) { \,\mathrm{d}}x{\,\mathrm{d}}y \nonumber\\ & \quad + \int_{{\omega}\times \partial t } \left ( a \operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}u_{\varepsilon}+ b \mathbb{v } \right ) ( \operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}u_{\varepsilon}{- } u ) { \,\mathrm{d}}x{\,\mathrm{d}}\sigma ( y ) + \delta^{u_{\varepsilon}}_\mathrm{cross , app , fold } , \end{aligned}\ ] ] where we added @xmath306 and @xmath307 $ ] , and created the folding mismatch @xmath308 ( u { - } \operatorname{\mathcal{t}_{\varepsilon}}u + { \varepsilon}\operatorname{\mathcal{t}_{\varepsilon}}\operatorname{\mathcal{g}_{\varepsilon}}u ) \nonumber \\ & \hspace{50pt } { - } \mathbb{d } \operatorname{\mathcal{t}_{\varepsilon}}(\nabla u_{\varepsilon } ) [ \nabla u { + } { \nabla_{\!y}}u - \operatorname{\mathcal{t}_{\varepsilon}}(\nabla ( u { + } { \varepsilon}\operatorname{\mathcal{g}_{\varepsilon}}u ) ) \big\rbrace { \,\mathrm{d}}x{\,\mathrm{d}}y \nonumber\\ & \quad + \int_{{\omega}\times \partial t } \left ( a \operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}u_{\varepsilon}+ b \mathbb{v } \right ) ( u { - } \operatorname{\mathcal{t}_{\varepsilon}}u + { \varepsilon}\operatorname{\mathcal{t}_{\varepsilon}}\operatorname{\mathcal{g}_{\varepsilon}}u ) { \,\mathrm{d}}x{\,\mathrm{d}}\sigma ( y )
. \end{aligned}\ ] ] exploiting the higher regularity @xmath309 we obtain with lemma [ lemma : unfold - err ] and theorem [ thm : fold - mismatch ] @xmath310 { - } \nabla_y u \vert_{{\mathrm{l}}^2({\omega}\times y _ * ) } \leq { \varepsilon}c \vert u \vert_{{\mathrm{h}}^1({\omega } ; { \mathrm{h}}^1(y _ * ) ) } .\end{aligned}\ ] ] the boundary term in is controlled via @xmath311 while noting that @xmath312 . with this , @xmath313 by , hlder s inequality , and the boundedness of @xmath1 in , we obtain @xmath314 _ step 2 : reformulation of @xmath149-equation .
_ the weak formulation reads @xmath315 for all @xmath177 which is equivalent to @xmath316 \cdot [ \nabla \varphi { + } \nabla_y \phi ] + r(u)\varphi { \,\mathrm{d}}x { \,\mathrm{d}}y \\ & \quad + \int_{{\omega}\times\partial t } ( a u + b \mathbb{v } ) \varphi { \,\mathrm{d}}x{\,\mathrm{d}}\sigma ( y ) \end{aligned}\ ] ] for all test functions @xmath203 and @xmath317 .
we choose @xmath201 such that we can control the periodicity defect of @xmath318 as in theorem [ thm : per - defect-2 ] @xmath319 \cdot \operatorname{\mathcal{t}_{\varepsilon}}(\nabla \varphi ) + r(u)\operatorname{\mathcal{t}_{\varepsilon}}\varphi { \,\mathrm{d}}x { \,\mathrm{d}}y \nonumber \\ & \quad + \int_{{\omega}\times\partial t } ( a u + b \mathbb{v } ) \operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}\varphi { \,\mathrm{d}}x{\,\mathrm{d}}\sigma ( y ) + \delta^u_\mathrm{per}.\end{aligned}\ ] ] the periodicity defect is given via @xmath320 ( \varphi { - } \operatorname{\mathcal{t}_{\varepsilon}}\varphi ) { - } \mathbb{d } [ \nabla u { + } { \nabla_{\!y}}u ] \cdot [ \nabla \varphi { + } { \nabla_{\!y}}\phi_{\varepsilon}- \operatorname{\mathcal{t}_{\varepsilon}}(\nabla \varphi ) ] { \,\mathrm{d}}x { \,\mathrm{d}}y \nonumber \\ & \quad + \int_{{\omega}\times\partial t } ( a u + b \mathbb{v } ) ( \varphi { - } \operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}\varphi ) { \,\mathrm{d}}x{\,\mathrm{d}}\sigma ( y ) \end{aligned}\ ] ] and it is controlled by applying lemma [ lemma : unfold - err ] , theorem [ thm : per - defect-2 ] with @xmath321 \in { \mathrm{h}}^1({\omega } ; { \mathrm{l}}^2(y_*))$ ] , arguing as in for the boundary term , and using the boundedness of @xmath149 in via @xmath322 now , we choose @xmath323 , where @xmath324 denotes the extension from @xmath163 to @xmath325 according to theorem [ thm : extension ] .
note that the test function belongs to the space @xmath325 which differs from step 1 wherein it belonged to @xmath163 .
indeed it holds @xmath326 almost everywhere in @xmath13 .
inserting @xmath203 into and rearranging gives @xmath327 \cdot [ \operatorname{\mathcal{t}_{\varepsilon}}(\nabla u_{\varepsilon } ) -
( \nabla u { + } { \nabla_{\!y}}u ) ] \nonumber \\ & \hspace{140pt } + r(u)(\operatorname{\mathcal{t}_{\varepsilon}}u_{\varepsilon}{- } u ) \big\rbrace { \,\mathrm{d}}x { \,\mathrm{d}}y \nonumber \\ & \quad + \int_{{\omega}\times\partial t } ( a u + b \mathbb{v } ) ( \operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}u_{\varepsilon}{- }
u ) { \,\mathrm{d}}x{\,\mathrm{d}}\sigma ( y ) + \delta^u_\mathrm{per , fold } \end{aligned}\ ] ] and another folding mismatch @xmath328 ( u { - } \operatorname{\mathcal{t}_{\varepsilon}}u + { \varepsilon}\operatorname{\mathcal{g}_{\varepsilon}}u ) \nonumber \\ & \hspace{60pt } { - } \mathbb{d } [ \nabla u { + } { \nabla_{\!y}}u ] \cdot [ \nabla u { + } { \nabla_{\!y}}u - \operatorname{\mathcal{t}_{\varepsilon}}(\nabla u { + } { \varepsilon}\nabla \operatorname{\mathcal{g}_{\varepsilon}}u ) ] \big\rbrace { \,\mathrm{d}}x{\,\mathrm{d}}y \nonumber \\ & \quad + \int_{{\omega}\times\partial t } ( a u + b \mathbb{v } ) ( u { - } \operatorname{\mathcal{t}_{\varepsilon}}u + { \varepsilon}\operatorname{\mathcal{g}_{\varepsilon}}u ) { \,\mathrm{d}}x{\,\mathrm{d}}\sigma ( y ) .\end{aligned}\ ] ] the folding mismatch @xmath329 has the same form as @xmath330 in when replacing @xmath149 with @xmath331 .
finally , we control the norm @xmath332 by using once more lemma [ lemma : unfold - err ] and theorem [ thm : fold - mismatch ] . applying young s inequality with @xmath333 in yields @xmath334 _ step 3 : derivation of grnwall - type estimates . _ subtracting equation from yields @xmath335 \cdot [ \operatorname{\mathcal{t}_{\varepsilon}}(\nabla u_{\varepsilon } ) - ( \nabla u { + } { \nabla_{\!y}}u ) ] \\ & \hspace{50pt } + [ r(\operatorname{\mathcal{t}_{\varepsilon}}u_{\varepsilon } ) - r(u ) ] ( \operatorname{\mathcal{t}_{\varepsilon}}u_{\varepsilon}{- } u ) \big\rbrace { \,\mathrm{d}}x{\,\mathrm{d}}y \\ & \quad + \int_{{\omega}\times\partial t } a |\operatorname{\mathcal{t}_{\varepsilon}^\mathrm{b}}u_{\varepsilon}{- } u
|^2 { \,\mathrm{d}}x{\,\mathrm{d}}\sigma ( y ) + \delta^{u_{\varepsilon}}_\mathrm{cross , app , fold } - \delta^u_\mathrm{per , fold}.\end{aligned}\ ] ] using the uniform ellipticity of @xmath336 , the lipschitz continuity of @xmath337 , and the estimations of the periodicity defect in gives @xmath338 choosing @xmath339 and integrating over @xmath340 with @xmath285 we get @xmath341 * final step .
* we add and and finally obtain @xmath342 the application of grnwall s lemma and the convergence of the initial values in complete the proof of .
our corrector estimates generalize the qualitative homogenization result obtained in @xcite in two ways : on the one hand we prove quantitative estimates . on the other hand ,
we consider slow thermal diffusion as well as different scalings @xmath343 and @xmath344 of the cross - diffusion terms . under slightly more general assumptions on the data with respect to the @xmath131-dependence
, our estimates imply in particular the rigorous but qualitative homogenization limit for this system . * what is the limit for arbitrary @xmath19 ?
* for all @xmath345 the limiting @xmath149-equation remains as it is and the cross - diffusion @xmath346 disappears in the limit @xmath110 .
for @xmath347 we have a priori that @xmath348 weakly in @xmath349 and we expect the additional term @xmath350 in the limit .
the choice @xmath351 is not meaningful , since the cross - diffusion term is unbounded with @xmath352 . for @xmath353
the cross - diffusion term @xmath354 vanishes in the limiting @xmath132-equation and for @xmath355 it diverges with @xmath356 .
indeed only the choice @xmath357 is meaningful , since it corresponds to the scaling of @xmath358 .
* possible generalizations concerning the data . *
our analysis allows for not - exactly periodic coefficients such as @xmath359 with @xmath360 as in @xcite .
the coefficients @xmath100 and @xmath361 as well as the reaction term @xmath362 may also be not - exactly periodic in the same manner .
moreover all coefficients may additionally depend lipschitz continuously on time .
the sink / source term @xmath31 may be less regular by choosing @xmath363(x)$ ] to capture possible spatial discontinuities in @xmath72 ; { \mathrm{h}}^1({\omega } ; { \mathrm{l}}^2(\partial t)))$ ] . on the boundary @xmath364 we may consider globally lipschitz continuous reaction terms @xmath365 . in this case ,
the boundary term in is controlled by @xmath366 , where @xmath367 denotes the global lipschitz constant .
non - linear boundary terms may require better initial values to derive the @xmath368-regularity of the time derivatives as in @xcite , however the error estimates hold as they are .
* on the choice of the initial values .
* for given @xmath369 the obvious choice is @xmath370 such that the assumption @xmath371 is satisfied .
perturbations of the form @xmath372 , which preserve non - negativity , are possible as well . in the case of slow diffusion
such a direct choice is not possible mainly because @xmath373 and @xmath374 live in spaces of dimension @xmath37 and @xmath375 , respectively .
let @xmath376 be given .
one possible choice is @xmath377 , however we are not able to prove @xmath378 in this case , since @xmath379 is not a hilbert space . hence , we assume strong differentiability , such as @xmath380 or @xmath381 , so that @xmath382 is well - defined in @xmath383 .
we recall elementary properties for the periodic unfolding operator @xmath154 and the boundary unfolding operator @xmath231 as well as extensions operators .
[ lemma : prop - t ] let @xmath384 with @xmath385 .
the operators @xmath154 and @xmath231 are linear and bounded .
the product rule @xmath386 holds for all @xmath387 , @xmath388 and @xmath389 , @xmath390 , respectively .
the norms are preserved via @xmath391 for all @xmath392 and @xmath393 , respectively .
one has the integration formulas @xmath394 for all @xmath395 and @xmath396 , respectively . if @xmath397 , then it is @xmath398 with @xmath399 . for all @xmath392 it holds @xmath400 = r(\operatorname{\mathcal{t}_{\varepsilon}}u)$ ] where @xmath68 is an arbitrary function .
assertions 1.-5 .
follow from ( * ? ? ?
2.5 & 5.2 ) and 6 . from @xmath401 ( x , y ) = r[u({\varepsilon}[x/{\varepsilon } ] { + } { \varepsilon}y ) ] = [ t(\operatorname{\mathcal{t}_{\varepsilon}}u)](x , y)$ ] for all @xmath153 .
[ thm : extension ] under the assumptions [ assump : domain ] on the domain there exists a family of linear operators @xmath402 such that for every @xmath397 it holds @xmath403 where @xmath79 only depends on the domains @xmath52 , @xmath45 , and @xmath404 .
the proof of proposition [ thm : fold - mismatch ] for the folding mismatch follows @xcite and is adapted to perforated domains .
we define the scale - splitting operator @xmath405 by @xmath406-lagrangrian interpolants , as customary in finite element methods ( fem ) , following ( * ? ? ?
* sec . 3 ) .
by the sobolev extension theorem , there exists for every @xmath407 and @xmath288 a function @xmath408 and @xmath409 , respectively , such that it holds @xmath410 where @xmath79 only depends on the domain @xmath52 .
then @xmath411 is given via : for every node @xmath412 we define @xmath413 ( note that this definition is slightly different than in @xcite .
therein the average is taken over balls @xmath414 centered at @xmath415 and not touching the pores @xmath47 .
the present definition has the advantage that the equality @xmath416 holds for all nodes . )
we define @xmath417 on the whole @xmath74 by interpolating the nodal values @xmath418 with @xmath406-lagrangian interpolants yielding polynomials of degree @xmath37 , for more details see ( * ? ? ?
4.1 ) or ( * ? ? ?
2.3.6 ) . for given two - scale functions @xmath420 of product form , we can now construct approximating sequences in @xmath163 via @xmath421 and require only the minimal regularity @xmath407 and @xmath422 . according to (
3.1 ) the macroscopic interpolants satisfy @xmath423 where @xmath79 only depends on @xmath52 and @xmath43 .
furthermore , we can control the difference between @xmath179 and @xmath405 by : the proof follows along the lines of ( * ? ? ?
3.4 ) by replacing @xmath45 with @xmath43 . in a first step estimate
is derived for @xmath426 using the `` folded function '' @xmath427 and the estimates . in a second step this result is generalized to arbitrary two - scale functions @xmath164 by exploiting the tensor product structure of the space @xmath428 and expressing @xmath429 in terms of an orthonormal basis @xmath430 with @xmath431 . | the present work deals with the derivation of corrector estimates for the two - scale homogenization of a thermo - diffusion model with weak thermal coupling posed in a heterogeneous medium endowed with periodically arranged high - contrast microstructures .
the terminology `` weak thermal coupling '' refers here to the variable scaling in terms of the small homogenization parameter @xmath0 of the heat conduction - diffusion interaction terms , while the `` high - contrast '' is thought particularly in terms of the heat conduction properties of the composite material . as main target
, we justify the first - order terms of the multiscale asymptotic expansions in the presence of coupled fluxes , induced by the joint contribution of sorret and dufour - like effects . the contrasting heat conduction combined with cross coupling lead to the main mathematical difficulty in the system .
our approach relies on the method of periodic unfolding combined with @xmath0-independent estimates for the thermal and concentration fields and for their coupled fluxes . *
msc 2010 : * 35b27 , 35q79 , 74a15 , 78a48 .
* keywords : * homogenization , corrector estimates , periodic unfolding , gradient folding operator , perforated domain , thermo - diffusion , composite media . |
this note addresses some questions raised by p. hegarty in @xcite . in that paper
he studied results about graphs inspired by the cauchy - davenport theorem .
all graphs in this paper are simple and loopless . for two vertices @xmath2 , denote the length of the shortest path between them by @xmath3 . for @xmath4 , define its _ ith neighborhood _ as @xmath5 .
the @xmath1th power of a graph @xmath0 , denoted @xmath6 , is constructed from @xmath0 by adding an edge between two vertices @xmath7 and @xmath8 when they are within distance @xmath1 in @xmath0 .
define the diameter of @xmath0 , @xmath9 , as the minimal @xmath1 such that @xmath6 is complete ( alternatively , the maximal distance between two vertices ) .
denote the number of edges of @xmath0 by @xmath10 . for @xmath4 and a set of vertices @xmath11 ,
define @xmath12 .
the cayley graph of a subset @xmath13 is constructed on the vertex set @xmath14 . for two distinct vertices
@xmath15 , we define @xmath16 to be an edge whenever @xmath17 or @xmath18 .
the following is a consequence of the cauchy - davenport theorem ( usually stated in the language of additive number theory @xcite ) .
[ thm : cauchydavenport ] let @xmath19 be a prime , @xmath20 a subset of @xmath14 , and @xmath0 the cayley graph of @xmath20 .
then for any integer @xmath21 : @xmath22 if we take @xmath20 to be the arithmetic progression @xmath23 , then equality holds in this theorem for all @xmath24 .
we might look for analogues of theorem [ thm : cauchydavenport ] for more general graphs @xmath0 .
in particular since these cayley graphs are always regular and ( when @xmath19 is prime ) connected , we might focus on regular , connected @xmath0 . in @xcite hegarty proved the following theorem : [ thm : hegarty ] suppose @xmath0 is a regular , connected graph with @xmath25 .
then we have @xmath26 with @xmath27 in other words , the cube of @xmath0 retains the original edges of @xmath0 and gains a positive proportion of new ones . in section 3
we prove this theorem with an improved constant of @xmath28 .
the requirement of regularity can not be easily dropped , as shown in @xcite .
theorem [ thm : hegarty ] leads to the question of how the growth behaves for other powers of the @xmath0 . note that theorem [ thm : hegarty ]
can not be used recursively to obtain such a result since the cube of a regular graph is not necessarily regular . in @xcite
it was shown that no equivalent of theorem [ thm : hegarty ] exists with @xmath29 replaced by @xmath30 , and it was asked what happens for higher powers . in this note
we address that question .
we prove the following theorem : [ thm : higherpowers ] suppose @xmath0 is a regular , connected graph , and @xmath31 .
then we have : @xmath32 let the degree of each vertex be @xmath33 .
fix some @xmath34 with @xmath35 nonempty .
consider any vertex @xmath36 .
then for any @xmath37 satisfying @xmath38 , there is a @xmath39 such that @xmath40 .
for such a @xmath41 , all vertices @xmath42 have @xmath43 .
all such @xmath7 are contained in @xmath44 , hence @xmath45 note that each @xmath46 satisfies @xmath47 . summing the bound ( [ eq : bound ] ) over all these @xmath37 , noting that any edge is counted at most once , we obtain @xmath48
now we sum this over all @xmath49 .
note that since the edges counted above go from some @xmath50 to @xmath51 with @xmath52 , each edge is counted at most once .
also we havent yet counted any of the original edges of @xmath0 , so we might as well add them .
hence @xmath53 obviously there was nothing particularly special about @xmath34 .
we can get a similar expresssion using @xmath54 , namely @xmath55 averaging ( [ eq : vsum ] ) and ( [ eq : vsum2 ] ) we get @xmath56 note that for any @xmath57 we have @xmath58 this is because @xmath59 . putting the bound ( [ eq : sumbound ] ) into the sum ( [ eq : vsumaverage ] ) we obtain @xmath60 thus the theorem is proven .
note that for @xmath61 the bounds in theorem [ thm : higherpowers ] are trivial .
in particular it says nothing about the increase in the number of edges of the cube of a regular , connected graph .
such an increase was already demonstrated by hegarty in theorem [ thm : hegarty ] .
here we give an alternative proof of that theorem , yielding a slightly better constant .
[ thm : improvedhegarty ] suppose @xmath0 is a regular , connected graph with @xmath62
. then we have @xmath63 let the degree of each vertex be @xmath33 .
note that as @xmath0 is regular , and not complete , every @xmath4 will have a non - neighbour in @xmath0 .
together with connectedness this implies that each @xmath4 has at least one new neighbour in @xmath30 .
this implies the theorem for @xmath64 . for the remainder of the proof , we assume that @xmath65 .
the proof rests on the following colouring of the edges of @xmath0 : for an edge @xmath66 in @xmath0 , colour @xmath67 notice that if @xmath66 is a blue edge , then there are at least @xmath68 neighbours of @xmath69 in @xmath30 .
this is because @xmath69 will be connected to everything in @xmath70 except itself , and @xmath71 for @xmath66 blue . if , in addition , we have some @xmath7 connected to @xmath69 by an edge ( of any colour ) , then @xmath7 will be at distance at most @xmath72 from everything in @xmath73 . hence @xmath7 will have at least @xmath68 neighbours in @xmath29 .
+ partition the vertices of @xmath0 as follows : @xmath74 @xmath75 @xmath76 + by the above argument , if @xmath34 is in @xmath77 , then @xmath78 . recall that each @xmath79 will have at least one new neighbour in @xmath30 , giving @xmath80 .
summing these two bounds over all vertices in @xmath0 , noting that any edge is counted twice , gives @xmath81 recall that we are considering the case when @xmath65 .
thus to prove that @xmath82 , it suffices to show that @xmath83 . to this end
we shall demonstrate that @xmath84 .
first however we need a proposition helping us to find blue edges in @xmath0 .
[ cliqueproposition ] for any @xmath4 there is some @xmath85 such that @xmath86 .
. then there are vertices @xmath7 and @xmath8 such that @xmath88 forms a path between @xmath69 and @xmath34 .
we will show that one of the edges @xmath89 , @xmath16 or @xmath90 is blue .
this will prove the proposition assuming that there are _ any _ blue edges to begin with .
however , it also shows the existence of blue edges because @xmath25 .
so , suppose that the edges @xmath89 and @xmath91 are red . then we have @xmath92 , and @xmath93 . using this and @xmath94 gives @xmath95 therefore @xmath96 . hence @xmath16 is blue , proving the proposition
now we will show that @xmath84 .
suppose @xmath97 . by the definition of @xmath98 , there is a @xmath99 such that @xmath100 is an edge .
this edge is neccesarily red as @xmath101 . using @xmath102,we have @xmath103 .
hence @xmath104 .
proposition [ cliqueproposition ] implies that there is some @xmath97 such that @xmath105 is an edge .
since @xmath105 is red , we have @xmath106 . using this
, the fact that @xmath107 , and ( [ eq : upperbound ] ) , gives @xmath108 double - counting the edges between @xmath11 and @xmath98 using the bounds ( [ eq : upperbound ] ) and ( [ eq : lowerbound ] ) gives a contradiction unless @xmath109 .
therefore @xmath110 as required .
theorem [ thm : higherpowers ] answers the question of giving a lower bound on the number of edges that are gained by taking higher powers of a graph .
we obtain growth that is linear with @xmath1 just as in theorem [ thm : cauchydavenport ] .
* the constant @xmath111 in theorem [ thm : higherpowers ] can not be improved to something of the form @xmath112 with @xmath113 . to see and consider the following sequence of graphs @xmath114
as @xmath33 tends to infinity : + take disjoint sets of vertices @xmath115 , with @xmath116 if @xmath117 and @xmath118 otherwise .
add all the edges within each set and also between neighboring ones .
so if @xmath119 , @xmath120 , then @xmath66 is an edge whenever @xmath121 ( see figure [ fig : img1 ] ) .
+ .,width=377 ] + the number of edges in @xmath114 is at least the number of edges in the larger classes which is @xmath122 .
+ the @xmath1th power @xmath123 has less than @xmath124 edges which is less than @xmath125
. therefore , @xmath126 + the graphs @xmath114 are not regular , but if @xmath127 , it is possible to remove a small ( less than @xmath128 ) number of edges from the graphs and make them @xmath33-regular without losing connectedness ( any cycle passing through all the vertices in @xmath129 would work ) . call these new graphs @xmath130 . by the same argument as before we have @xmath131 +
if @xmath132 , a similar trick can be performed , but we d need to start with @xmath116 if @xmath133 and @xmath118 otherwise .
+ so the factor of @xmath134 can not be improved for regular graphs .
all these examples are inspired by one given in @xcite to show that for any @xmath135 there are regular graphs @xmath0 with @xmath136 .
* despite the above example , there is certainly room for further improvement in theorems [ thm : higherpowers ] and [ thm : improvedhegarty ]
. in particular , theorem [ thm : improvedhegarty ] does nt seem tight in any way .
the graphs @xmath130 seem to give essentially the slowest possible growth for _ all _ powers of regular graphs . considering the graphs @xmath137 leads to the conjecture of @xmath138 for @xmath0 regular , connected , and @xmath25 . +
a shortcoming of theorem [ thm : higherpowers ] is that it only gives a good bound if the diameter of @xmath0 is close to @xmath1 .
when this is not the case , the number of edges in @xmath6 seems to grow faster . it would be interesting to obtain a good lower bound on @xmath139 involving both @xmath1 and @xmath9 .
* all the questions from this paper and @xcite could be asked for directed graphs .
in particular one can define directed cayley graphs for a set @xmath13 by letting @xmath16 be a directed edge whenever @xmath140 .
then the cauchy - davenport theorem implies an identical version of theorem [ thm : cauchydavenport ] for directed cayley graphs . in this setting
it is easy to show that there is growth even for the square of an out - regular oriented graph @xmath141 ( a directed graph where for a pair of vertices @xmath69 and @xmath34 , @xmath66 and @xmath142 are not both edges ) . in particular , we have @xmath143 this occurs because every vertex @xmath34 has @xmath144 in an out - regular oriented graph
. it s easy to see that this is best possible for such graphs .
one can construct out - regular oriented graphs with an arbitrarily large proportion of vertices @xmath34 satisfying @xmath145 .
+ however if we insist on _ both _ in and out - degrees to be constant , ( 8) no longer seems tight .
such graphs are always eulerian . in @xcite
there is a conjecture attributed to jackson and seymour that if an oriented graph @xmath141 is eulerian , then @xmath146 holds .
if this conjecture were proved , it would be an actual generalization of the directed version of theorem [ thm : cauchydavenport ] , as opposed to the mere analogues proved above .
i would like to thank my supervisors jan van den heuvel , and jozef skokan for much helpful advice and remarks . | for a graph @xmath0 , its @xmath1th power is constructed by placing an edge between two vertices if they are within distance @xmath1 of each other . in this note
we study the amount of edges added to a graph by taking its @xmath1th power . in particular we obtain that either the @xmath1th power is complete or
many " new edges are added .
this is an extension of a result obtained by p. hegarty for cubes of graphs . |
in our concerned problem , the photons with the electromagnetic fields @xmath1 are co - propagating or counter - propagating in two ensembles , having their kinetic hamiltonian ( here we give the form of co - propagation ) @xmath65 together with the control field of the rabi frequency @xmath0 , they will induce the polarization fields @xmath66 and the rydberg spinwave fields @xmath67 in the medium , as described by the many - body version of the atomic - level - scheme hamiltonian @xmath68 where @xmath9 is atom - field coupling constant , @xmath8 the atomic density of the ensembles , and @xmath44 the detuning of the photons .
the hamiltonian for the interaction between the rydberg spinwave fields takes the form @xmath69 meanwhile , as the result of the interaction , the population of the intermediate levels spontaneously decaying at the rate @xmath18 gives rise to the photon dissipation , and we describe it by the stochastic hamiltonian @xmath70 in terms of the coupling between the polarization fields and quantum noise fields , where the random - variable noise operators satisfy @xmath71=\delta_{lm}\delta({\bf x}-{\bf x}^{\prime})\delta(t - t^{\prime})$ ] .
an advantage of this stochastic hamiltonian approach is that the physical process can be modeled by the formal unitary evolution operator @xmath72 with @xmath73 , which naturally leads to the heisenberg - langevin equations of the relevant quantum fields as shown below . in deriving the dynamical equations ,
we write @xmath74 , which satisfy the ito s rules ( a generalization from those in @xcite ) : @xmath75 then an increment of the polarization fields , for example , will be found as @xmath76\nonumber\\ & + & \gamma \int d{\bf x}'\big(2\hat{p}^\dagger ( { \bf x}',t)\hat{p}({\bf x},t)\hat{p}({\bf x}',t)-\hat{p}({\bf x},t)\hat{p}^\dagger({\bf x}',t)\hat{p}({\bf x}',t)-\hat{p}^\dagger({\bf x}',t)\hat{p}({\bf x}',t)\hat{p}({\bf x},t)\big),\end{aligned}\ ] ] where the above ito s rules have been applied to the expansion up to the second order of @xmath77 . in this way the exact dynamical equations for the three types of quantum fields are obtained as follows ( @xmath24 ) : @xmath78 @xmath79 @xmath80 where @xmath81 for @xmath2 and @xmath3 . as in other photon storage scenarios
@xcite , we will replace the field operators in the above dynamical equations by their mean values to see how their profiles evolve under given conditions .
the dynamical equations for the field profiles in eqs .
( 1)-(3 ) of the main text are thus obtained by averaging out the quantum noise operators , @xmath82 .
[ [ b.more-examples-of-pulse-evolution-under-realistic-interaction ] ] b. more examples of pulse evolution under realistic interaction ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the interaction potential a pulse experiences in the concerned processes is determined by the spatial extension of the other rydberg spinwave packet , as indicated by eq .
( [ inter ] ) or eq .
( 4 ) of the main text , which comes from the full dynamics discussed in the above .
this leads to a significant difference from the interaction with a point rydberg atom .
we illustrate the fact with the comparison in fig .
the finally obtained realistic potential profiles in the illustrated situations correspond to those of the standstill rydberg spinwaves . though the extending spinwave packets lower the potential magnitudes , the potential values distribute less abruptly over the ensembles .
such potentials due to wave packets instead of point particles exist for any type of interaction .
for instance , the dipole - dipole interaction potentials between the stored wave packets should also be evaluated in this way , and are found to be more inhomogeneous than those of vdw interaction ; see fig .
the dipole - dipole interaction is applied in another photonic gate proposal based on stored photon pulses @xcite .
more realistic considerations should be therefore included in quantum information processing by the similar methods .
one of the meaningful consequences of such realistic interaction potentials is that they lead to different dissipation rates over the pulses . in ref .
@xcite about another topic related to rydberg eit , the interaction between pulses is also modeled as the one between extending wave packets , but their dissipation is assumed to be with a global rate independent of the pulse distributions .
the exact details of pulse dissipation should be clarified by approaching the processes with the complete dynamics of the quantum field profiles . here
we present two more examples to show how such space - time dependent dissipation manifest in photon storage .
the first one in fig .
s-3 is about the even closer ensembles in fig .
3(a1 ) of the main text .
the interaction between the pulses in these setups becomes stronger .
since the counter - propagating pulses have spatial extensions , the interaction between their fronts is intense before they are stopped in the medium .
an extreme result is that half of a pulse will be absorbed .
negligible decay of the rydberg levels after the control field is turned off maintains stable asymmetric spinwave profiles during the storage period .
the other example in fig .
s-4 illustrates the dynamical evolutions of two co - propagating photon pulses and their associated spinwaves before they are stopped together .
here we use a narrower pulse bandwidth ( corresponding to longer pulse durations @xmath39 ) so that the pulse dissipation is almost completely due to their interaction .
this example clearly demonstrates the significant effects of pulse interaction . due to a longer interaction time
, only a small portion of the initially induced spinwaves will remain in the end .
we now take a further look at the effects from a fast - changing rabi frequency @xmath0 of control field . to neglect the photon losses from these effects
, the systems should evolve along adiabatic passages during storage and retrieval of photons .
the condition for realizing such adiabatic passages can be formulated in terms of a change rate of the mixing angle @xmath83 as @xmath84 , where @xmath85 is the energy difference between bright - state and dark - state polaritons ( see , e.g. @xcite ) .
the change rate , which takes the form @xmath86 scales inversely with the square root of the atomic density @xmath8 in our concerned regime realizing slow light ( @xmath87 ) . in ensembles of high density @xmath8 ,
insignificant non - adiabatic effects are therefore expected .
this is seen from fig .
s-5(a ) about the processes without the interaction between pulses , in which the storage efficiency goes up with the density and the difference between fast and slow switching of a control field is not so large . meanwhile , in the presence of the pulse interaction as shown in fig .
s-5(b ) , the tendency of storage efficiency with atomic density will be totally reversed .
the interaction between spinwaves becomes stronger in an ensemble of higher density , because the spinwave profiles @xmath88 are magnified by the square root of @xmath8 .
accordingly more significant dissipation will exist in the ensemble .
these results demonstrate that the dissipation from non - adiabatic corrections is hardly comparable with that due to pulse interaction . stopping and regenerating photons in the setups
can be done with quickly switching control field too . c. w. gardiner and p. zoller , _ quantum noise _ ( springer - verlag , berlin , 2000 ) .
a. v. gorshkov , a. andre , m. d. lukin , and a. s. srensen , phys .
a 76 , 033805 ( 2007 ) .
d. paredes - barato and c. s. adams , phys .
112 , 040501 ( 2014 ) .
b. he , a. v. sharypov , j. sheng , c. simon , and m. xiao , 112 , 133606 ( 2014 ) .
d. petrosyan and k. mlmer , phys . rev .
a 87 , 033416 ( 2013 ) . | stopping and regenerating a pair of single - photon pulses at adjacent locations in coherently prepared rydberg atomic ensembles are significantly affected by their effective interaction mediated by rydberg excitations , and the similar processes can differ notably from the one exhibiting the common rydberg blockade as with the stationary propagation of multi - photon light beams in the same medium .
based on the complete dynamics , we reveal the detailed features in such processes by finding how the profiles of the involved quantum fields evolve in various situations .
the findings help to determine the proper regimes for implementing photon - photon gates and photon transistors .
in addition , we discuss the non - adiabatic corrections associated with quickly changing control fields , and illustrate a method that restores the photon pulses original amplitude during their retrieval unless they are heavily damped before storage . since its first experimental observations @xcite
, the phenomenon of electromagnetically induced transparency ( eit ) in ensembles of cold rydberg atoms has attracted extensive studies .
different from the ordinary eit @xcite , there exists van der waals ( vdw ) or dipole - dipole interaction between atoms , which modifies the dispersive and dissipative properties of light , as well as the photon correlations , in rydberg medium @xcite .
these interactions also indicate the possible formation of bound states @xcite and lattice structures @xcite of photons .
more recent experiments @xcite have demonstrated numerous interesting features related to rydberg - atom - mediated interactions .
the above mentioned long - range interactions make rydberg eit medium a promising candidate for quantum information processing with photons .
a fascinating application of this kind is implementing a deterministic two - qubit gate , the building block of photonic quantum computing .
apart from the scenarios working with slowly traveling photons @xcite , a gate operation could be performed with stopped photons that are converted to rydberg atomic excitations , so that the nonlocal interaction between the long - living rydberg spinwaves achieves much more significant nonlinear phases than those generated by the collisional interaction between atoms @xcite . however , because the vdw interaction also brings about unwanted effects , the previous schemes @xcite in this category require more sophisticated technical steps , while it was not clear whether the simplest way to store and regenerate two photons ( without extra procedures ) properly under vdw potential could realize a good gate operation .
on the other hand , other devices such as switches @xcite and transistors @xcite should perform a different function that scatters more photons with a stored one . stopping light in rydberg medium and/or re - converting rydberg excitations into light
have been demonstrated experimentally @xcite , and the vdw interaction potential in the used multi - photon beams is generally high and stable , leading to the well - known blockade effect .
yet the details of the corresponding processes for individual photon pulses under their varying interaction , which are important to the operations of the devices such as photon - photon gates and photon transistors , had not been fully understood .
the previously unknown features in these processes will be illustrated in this work .
in addition to the application concerns , the main purpose of our study is to provide a more realistic picture for the non - steady motions of interacting wave packets in rydberg medium . to reflect the spatial distributions of photons and rydberg excitations ,
we treat them as delocalized fields for their interactions and find their continuous evolutions in space and time .
this differs from most other related studies , which rely on a discontinuous blockade radius from point - point interaction potential .
a typical example of our concerned problem is the setup of two pencil - shaped ensembles in fig . 1 . here
two single - photon pulses either travel along the same direction ( co - propagation ) or respectively enter the opposite tips of the ensembles ( counter - propagation ) .
the control fields with the rabi frequency @xmath0 are being turned off with time , so that the pulses will be stopped and converted to rydberg excitations that are close to each other .
after the two rydberg spinwave packets interact for a certain time , they will be turned on again to retrieve the photon pulses in the phase matched directions .
this process can be modeled by the dynamical evolutions of the electromagnetic fields @xmath1 of the two photons ( @xmath2 and @xmath3 ) with the kinetic hamiltonian @xmath4 , the induced polarization fields @xmath5 and rydberg spinwave fields @xmath6 , where @xmath7 .
a high atomic ensemble density @xmath8 enhances the photon - atom coupling ( with the constant @xmath9 ) seen in the hamiltonian @xmath10 , which describes the atomic level scheme . a narrow band photon propagates with negligible absorption under the eit condition @xmath11 [ @xmath12 is defined in fig .
1(a ) ] .
however , under the spinwave interaction with the hamiltonian @xmath13 , where @xmath14 is the vdw potential , the condition will be violated by a shift @xmath15 of the levels @xmath16 in the relevant atoms . the consequent dissipation from populating the levels @xmath17 that decay at the rate @xmath18 can be depicted by a stochastic hamiltonian @xmath19 involving the quantum noise fields @xmath20 of the reservoirs @xcite .
it is sufficient to find the mean values @xmath21 of the quantum fields @xmath22 , and their evolutions under the total hamiltonian @xmath23 are determined by the following equations ( see appendix a and we set @xmath24 here ) : @xmath25 @xmath26 @xmath27 where @xmath28 for @xmath2 and @xmath3 . using eqs .
( [ 2 ] ) and ( [ 3 ] ) , one can expand the right - hand side of eq .
( [ 1 ] ) as @xmath29 then , in our concerned regime realizing slow light ( @xmath30 ) , the time derivative on the left - hand side of eq . ( [ 1 ] ) will be absorbed into the leading term in the above equation , reducing eq .
( 1 ) to the one only with a spatial derivative .
the discretization of this rearranged form of eq .
( 1 ) is used to get the spatial profile of @xmath31 at a specified moment , given the values of @xmath32 induced by the field @xmath31 at the preceding moment of iteration with the discretized eqs.([2 ] ) and ( [ 3 ] ) .
we apply a second order runge - kutta method in the iterative steps toward the quantities ( @xmath31 , @xmath32 , @xmath33 and @xmath34 ) all over the space - time grids . for simplicity ,
only symmetric propagations in two ensembles will be discussed below , but the generalization to different pulse motions is straightforward . as a time - dependent boundary condition for pulse evolution , the electromagnetic field of the input photon pulses
is supposed to have the profile @xmath35 at the entries of the ensembles , where @xmath36 is the maximum of the photons rabi frequency @xmath37 , and @xmath38 and @xmath39 are the time scales indicating the peak arrival and pulse duration , respectively , while @xmath40 is the bessel function of order zero with its first zero point @xmath41 .
the effects of a gradually increasing interaction can be best seen from stopping two counter - propagating pulses . in reality ,
storage of photons may fail as shown in fig .
2 , which illustrates the associated dynamical evolutions for the photons with the opposite - sign detuning values .
there a positively detuned photon will go through the medium , though it is damped .
if the photon s detuning changes the sign , it will not survive the travel .
such difference can be explained with the susceptibility @xmath42 , defined as @xmath43 , for a single - frequency wave under a constant interaction potential @xmath15 ; see the insets of fig .
2 . in the former situation
the attractive potential @xmath15 pulls " the central frequency component initially under the eit condition away from the absorption peak , but it pushes " the corresponding component in the latter toward the peak . then the positively detuned photons will virtually see two - level systems when they get closer , but the same medium becomes opaque to the negatively detuned ones .
these effects are more significant for a higher detuning @xmath44 , as the gap between the absorption peak and the two - level regime narrows down accordingly .
the scenario of negative detunings is most suitable for implementing a photon transistor . in the presence of many photons at the same time ,
a significant interaction potential can immediately push " the system across the absorption peak to the two - level regime on the other side , fitting the rydberg blockade model ( super - atom ) irrespective of the photons detuning signs @xcite which well describes the stationary propagation of light in rydberg medium . by contrast with the phenomenological models , our first - principle calculations provide the understandings of what happen to individual photons under their constantly varying interaction and a time - dependent @xmath0 simultaneously . in order to realize their storage , the photons , as well as the control field ,
should be better to be resonantly coupled to the energy levels ( @xmath45 ) so that the farther positions of absorption peaks leave more room for the frequency components not to enter the above mentioned regimes .
certainly one can take a stronger control field to widen the eit window , but the practice suppresses the induced spinwave , as seen from the relation @xmath46 for a single - frequency wave without interaction . accordingly
the photon pulses should have a sufficiently long duration @xmath39 ( corresponding to a narrower bandwidth @xmath47 ) to fit into the limited eit window .
such spatially extending photon pulses will be mapped to rydberg spinwaves distributing over the ensembles , after the control field is turned off .
when the process happens to two pulses counter - propagating respectively in close ensembles , the mutual interaction can diminish their fronts before they are stopped side by side ( the interaction between their back portions is still insignificant at this time ) , leaving a spinwave packet of asymmetric shape as in fig .
3(a1 ) .
the potential @xmath48 acting on other ensemble can therefore be extremely inhomogeneous as shown in fig .
3(b1 ) .
for an ideal photonic gate operation , each point on the stored wave packets should be under the same potential value @xmath15 , so that the wavefunction will gain a uniform phase .
thus one has to face a trade - off between the magnitude and the uniformity of the achieved potential @xmath48 .
unlike most other previous researches , we adopt the potential field in eq .
( [ v ] ) , which is generated by the real - time spinwave field @xmath49 , for the the mutual interaction between rydberg excitations .
the evolution an input pulse undergoes under this potential is very different from that under the interaction with a single - point rydberg atom .
the interaction potential value @xmath50 is simply the degree of deviating the eit condition @xmath11 at any space - time point @xmath51 , when the control field is on .
the consequent space - time dependent dissipation can be captured by our simulations directly based on eqs.([1])-([3 ] ) .
more examples of such pulse dissipation are given in appendix b. the calculations also specify the proper regime of larger ensemble separations , in which an approximately homogeneous potential can be realized for gate operation ; see fig .
3(b2 ) . in this regime beyond certain ensemble separation ,
co - propagation and counter - propagation actually make little difference .
the pulse motion in ensemble is largely influenced by the control field . given the rabi frequencies in fig .
1(b ) , the parameters @xmath52 and @xmath53 determine where a pulse will be stopped .
a fast switching with small @xmath53 can immediately stop the pulses at any moment .
one question is whether the non - adiabatic corrections @xcite , connected with terms containing the time derivatives of @xmath0 on the right - hand side of eq .
( [ ex ] ) , could lead to considerable pulse losses .
this can be clarified by our numerical calculations with arbitrary @xmath0 , which go beyond the previously studied adiabatic passages @xcite .
the results in figs .
3(c1 ) and 3(c2 ) indicate unremarkable difference between the disparate switching speeds .
further calculations manifest that , in the ensembles of high densities , the non - adiabatic effects are overshadowed by those from the pulse interaction ; see appendix c. under any circumstance the photon absorption is inevitable due to pulse interaction , limited eit window , non - adiabatic effects and others .
we here demonstrate a mechanism that compensates for the previous dissipations through photon regeneration . in fig .
4(a ) the control field in the read - out " is weaker than the control field for the write - in " .
the amplitude of the regenerated photons is less than that of the input photons , and they exit the medium at a lower group velocity ( flatter contour ) as well . the retrieval control field in fig .
4(b ) is , on the other hand , stronger than the one used for stopping the photons , so that the re - converted photon pulses become brighter and leave the ensembles at a higher group velocity ( steeper contour ) .
near the entries and exits of this counter - propagation setup ( almost no interaction exists there ) , the pulses can be regarded as dark - state polariton field @xmath54 , where @xmath55 .
the electromagnetic field @xmath56 of the pulses therefore takes the form @xmath57 at the beginning and @xmath58 in the end . to maintain their amplitudes ,
the dissipation that reduces the field @xmath59 to @xmath60 can be partly offset by the factor @xmath61 , though the retrieved pulses are deformed ( such deformation does not affect carrying quantum information ) .
the best enhancement rendering @xmath62 for the output photons demands an infinite @xmath63 , implying the impossibility of restoring the amplitudes of the severely damped pulses with a realistic @xmath64 [ see fig .
4(c ) ] .
because pulses dissipation is primarily decided by the duration and intensity of their interaction , a quick retrieval outdoes a slow one , contrary to what happens to the storage period ; compare fig .
4(d ) with fig .
3(c1 ) . a slow write - in " and
a fast read - out " are thus favored to our concerned processes . in summary , we have presented the first study of the details in stopping and regenerating individual photon pulses under their mutual interaction in rydberg eit medium . much more complex and varied than those of the stationary propagation of multi - photon beams , several important features in the processes originate from the ever - changing interactions which lead to space - time dependent dissipation .
our simulations provide more realistic pictures of the continuous photon pulse evolutions in space and time , dispensing with a boundary of discontinuity as the blockade radius in phenomenological models . apart from clarifying the regimes to implement photon transistors and gates , and ascertaining the insignificant role of the non - adiabatic corrections in high density ensembles ,
we show that the amplitudes of the weakened photon pulses can be enhanced by a stronger retrieval control field .
these understandings could be valuable guide for the relevant experimental researches .
l. y. is supported by the china scholarship council .
m.x . acknowledges funding support in part from nbrpc ( grant no .
2012cb921804 ) . j. h. w. is sponsored by nsfc ( grants no .
11174110 ) and nbrpc ( grants no .
2011cb921603 ) .
j. d. pritchard , d. maxwell , a. gauguet , k. j. weatherill , m. p. a. jones , and c. s. adams , phys .
rev .
lett .
105 , 193603 ( 2010 ) .
h. schempp , g. gter , c. s. hofmann , c. giese , s. d. saliba , b. d. depaola , t. amthor , m. weidemller , s. sevinli , and t. pohl , phys .
rev .
lett .
104 , 173602 ( 2010 ) .
m. fleischhauer , a. imamoglu , and j. p. marangos , rev .
mod .
phys .
77 , 633 ( 2005 ) . c. ates , s. sevinli , and t. pohl , phys .
rev .
a 83 , 041802(r ) ( 2011 ) .
s. sevinli , n. henkel , c. ates , and t. pohl , phys .
rev .
lett .
107 , 153001 ( 2011 ) .
d. petrosyan , j. otterbach , and m. fleischhauer , phys .
rev .
lett .
107 , 213601 ( 2011 ) .
j. d. pritchard , c. s. adams , and k. mlmer , phys .
rev .
lett .
108 , 043601 ( 2012 ) .
m. grttner and j. evers , phys .
rev .
a 88 , 033417 ( 2013 ) .
j. stanojevic , v. parigi , e. bimbard , a. ourjoumtsev , and p. grangier , phys .
rev .
a 88 , 053845 ( 2013 ) .
y .-
m .
liu , d. yan , x .- d .
tian , c .-
l .
cui , and j .- h .
wu , phys .
rev . a 89 , 033839 ( 2014 ) . w. li , d. viscor , s. hofferberth , and i. lesanovsky , phys . rev
. lett .
112 , 243601 ( 2014 ) .
h. wu , m .- m .
bian , l .- t .
shen , r .- x .
chen , z .- b .
yang , and s .- b .
zheng , phys .
rev .
a 90 , 045801 ( 2014 ) .
m. grttner , s. whitlock , d. w. schnleber , and j. evers , phys .
rev .
lett .
113 , 233002 ( 2014 ) .
d. viscor , w. li , and i. lesanovsky , new j. phys .
17 , 033007 ( 2015 ) .
y .-
m .
liu , x .- d .
tian , d. yan , y. zhang , c .-
l .
cui , and j .- h .
wu , phys .
rev . a 91 , 043802 ( 2015 ) .
p. bienias , s. choi , o. firstenberg , m. f. maghrebi , m. gullans , m. d. lukin , a. v. gorshkov , and h. p. bchler , phys .
rev .
a 90 , 053804 ( 2014 ) .
j. otterbach , m. moos , d. muth , and m. fleischhauer , phys .
rev .
lett .
111 , 113001 ( 2013 ) .
t. peyronel , o. firstenberg , q .- y .
liang , s. hofferberth , a. v. gorshkov , t. pohl , m. d. lukin , and v. vuleti , nature ( london ) 488 , 57 ( 2012 ) . y. o. dudin , f. bariani , and a. kuzmich , phys . rev .
lett . 109
( 2012 ) . c. s. hofmann , g. gnter , h. schempp , m. robert - de - saint - vincent , m. grttner , j. evers , s. whitlock , and m. weidemller , phys .
rev .
lett .
110 , 203601 ( 2013 ) .
o. firstenberg , t. peyronel , q .- y .
liang , a. v. gorshkov , m. d. lukin , and v. vuleti , nature ( london ) 502 , 71 ( 2013 ) .
d. maxwell , d. j. szwer , d. paredes - barato , h. busche , j. d. pritchard , a. gauguet , m. p. a. jones , and c. s. adams , phys .
rev . a 89 , 043827 ( 2014 ) .
i. friedler , d. petrosyan , m. fleischhauer , and g. kurizki , phys .
rev .
a 72 , 043803 ( 2005 ) .
b. he , a. macrae , y. han , a. lvovsky , and c. simon , phys .
rev .
a 83 , 022312 ( 2011 ) .
e. shahmoon , g. kurizki , m. fleischhauer , and d. petrosyan , phys .
rev .
a 83 , 033806 ( 2011 ) .
a. v. gorshkov , j. otterbach , m. fleischhauer , t. pohl , and m. d. lukin , phys .
rev .
lett .
107 , 133602 ( 2011 ) .
b. he , a. v. sharypov , j. sheng , c. simon , and m. xiao , 112 , 133606 ( 2014 ) .
a. rispe , b. he , and c. simon , 107 , 043601 ( 2011 ) . c. vo , s. riedl , s. baur , g. rempe , and s. drr , phys .
rev .
lett .
109 , 263602 ( 2012 ) .
d. paredes - barato and c. s. adams , phys .
rev .
lett .
112 , 040501 ( 2014 ) .
m. khazali , k. heshami , and c. simon , phys .
rev .
a 91 , 030301(r ) ( 2015 ) .
s. baur , d. tiarks , g. rempe , and s. drr , phys .
rev .
lett .
112 , 073901 ( 2014 ) .
d. tiarks , s. baur , k. schneider , s. drr , and g. rempe , phys .
rev .
lett .
113 , 053602 ( 2014 ) .
h. gorniaczyk , c. tresp , j. schmidt , h. fedder , and s. hofferberth , phys .
rev .
lett .
113 , 053601 ( 2014 ) .
y. o. dudin and a. kuzmich , science 336 , 887 ( 2012 ) .
d. maxwell , d. j. szwer , d. paredes - barato , h. busche , j. d. pritchard , a. gauguet , k. j. weatherill , m. p. a. jones , and c. s. adams , phys .
rev .
lett .
110 , 103001 ( 2013 ) .
m. fleischhauer and m. d. lukin , 65 , 022314 ( 2001 ) .
d. mller , l. b. madsen , and k. mlmer , phys .
rev .
lett .
100 , 170504 ( 2008 ) .
d. petrosyan and k. mlmer , phys .
rev .
a 87 , 033416 ( 2013 ) .
j. qian , j. zhai , l. zhang , and w. zhang , phys .
rev .
a 91 , 013411 ( 2015 ) . |
despite its simplicity , the london approach is a powerful tool in describing magnetic properties of the mixed state .
in fact , short of the full - blown microscopic theory , it is the only method available for low temperatures .
the approach is based on the london equation : @xmath5 where @xmath6 is the magnetic field , @xmath7 is the penetration depth ( a temperature dependent _ constant _ in uniform samples ) , @xmath8 is the flux quantum , and @xmath9 are vortex positions . for simplicity , the equation is written for isotropic materials .
this approach fails at distances of the order of the coherence length @xmath10 ; still , in materials with @xmath11 , there is a broad domain of intermediate fields @xmath12 where the complexity of the vortex core contributions to the total energy can be disregarded and the london approach suffices for the description of macroscopic magnetic properties . as far as the equilibrium properties of the flux - line lattice are concerned , the pivotal point is the expression for the free energy @xmath13 the right - hand side here is the interaction energy of vortices forming a periodic lattice , @xmath14 is the magnetic induction .
this expression is obtained by transforming the sum of pair - wise interactions of vortices to a sum over the reciprocal lattice , see e.g. , ref .. the sum ( or the integral over the reciprocal plane @xmath15 ) is logarithmically divergent so that a cutoff at @xmath16 is needed ( @xmath17 is the size of the vortex core ) .
this yields @xmath18 .
the parameter @xmath19 of the order unity is commonly introduced to account for uncertainty of the cutoff ( along with the uncertainty in the lower limit of the integral of the order of inverse intervortex spacing @xmath20 ) ; @xmath21 ... appears in eq.([f_lond ] ) for convenience of not having it in the expression for the magnetization .
again , the energy in the form of eq.([f_lond ] ) holds in intermediate fields @xmath22 , the domain existing only in materials with large ginzburg - landau parameter @xmath2 ( @xmath23 and @xmath24 are the lower and upper critical fields ) .
hence , although the length @xmath10 ( or the core size @xmath17 ) does not appear in the london equation ( [ london ] ) , it enters the energy expression ( [ f_lond ] ) through the cutoff and therefore affects , presumably weakly , macroscopic quantities such as the magnetization and other properties of the mixed state .
significant effort has recently been put in studies of the vortex core size @xmath17 ; see the review by sonier and references therein.@xcite notably , whatever definition of @xmath17 is adopted , the low temperature @xmath17 ( extracted from the @xmath4sr data ) decreases with increasing applied magnetic field in a number of materials such as nbse@xmath25 , v@xmath26si , luni@xmath25b@xmath25c , yba@xmath25cu@xmath26o@xmath27 , and ceru@xmath25 ; their physical characteristics have little to do with one other , except that all of them have a large gl parameter @xmath28 and exhibit large regions of reversible magnetic behavior .
one can add to this list a heavy fermion compound cecoin@xmath29 , for which the interpretation of small angle neutron scattering data ( sans ) requires a similar behavior of the coherence length.@xcite the dependencies @xmath30 for all tested materials are qualitatively similar : when the field increases toward @xmath24 , @xmath31 decreases roughly as @xmath32 . in other words , in large fields
@xmath17 is roughly proportional to the intervortex spacing .
a few qualitative reasons for the core shrinking with increasing field have been discussed in literature ; see the review .
perhaps the simplest defines the core boundary as a position where the divergent london current @xmath33 of an isolated vortex reaches the depairing value , i.e. , as @xmath34 . in the mixed state ,
neighboring vortices suppress the circulating current by contributing currents of the opposite direction .
hence , the depairing value is reached at a shorter distance from the vortex center , and consequently @xmath17 should decrease with increasing field .
the vortex core size , @xmath17 , is of the order of the coherence length @xmath10 and , in fact , it is often identified with @xmath35 . strictly speaking ,
the latter is defined only at the upper critical field : @xmath36 .
nevertheless the length @xmath10 is used to describe the mixed state at fields not necessarily close to @xmath24 .
the question of possible field dependence of @xmath10 has been considered by one of us for the isotropic case.@xcite it was shown that in the dirty limit one can use @xmath37 at any field within the mixed phase ; the same is true near the critical temperature @xmath38 for any scattering strength .
however , in general , when the field is reduced below @xmath24 , the value of @xmath0 increases , an effect that is profound in clean materials at low temperatures .
calculations of ref .
are in accord with the @xmath4sr results cited above . in the following ,
we denote as @xmath39 the value of @xmath0 at @xmath40 to stress that in general @xmath41 for @xmath42 .
another common theoretical definition of @xmath17 is based on the slope of the order parameter @xmath43 at the vortex axis @xmath44 , normalized either to its value @xmath45 far from the single vortex or to the value of @xmath46 half - way to the nearest neighbor in the mixed state : @xmath47 .
recent microscopic calculations of this quantity by miranovi _
et al . _ showed a variety of field dependent behaviors of @xmath17 at low temperatures depending on the scattering strength.@xcite in particular , this work suggests that @xmath30 may have a minimum which , however , has not been seen in @xmath4sr experiments .
there are many different ways to define @xmath17 customized for different experimental or theoretical needs ( see , e.g. , the discussion in ref . ) . for the purpose of this paper , these differences are irrelevant and we use the terms @xmath48 and @xmath49 as the same . in the following we provide experimental data for the field dependence of the reversible magnetization for a single crystal yni@xmath25b@xmath25c in a broad temperature region to demonstrate the known fact :
the data can not be described by the standard london model .
we then derive a closed form expression for the cutoff @xmath0 needed to represent correctly the data @xmath50 with the help of the london model .
we show that @xmath0 so chosen is qualitatively consistent with the field dependence of @xmath10 recorded by @xmath4sr and discussed theoretically . finally , we demonstrate that the new model generates a consistent description of the magnetization data for a number of unrelated materials with large @xmath2 .
the goal of this paper is to demonstrate that @xmath0 can be in principle extracted from the magnetization data , a less demanding experimental procedure as compared to @xmath4sr or sans . within our approach , _ the london penetration depth is field independent _ , whereas the field dependence of @xmath10 alone suffices to explain the data .
single crystals of yni@xmath25b@xmath25c , luni@xmath25b@xmath25c , and lu(ni@xmath51co@xmath52b@xmath25c were grown out of ni@xmath25b or ( ni@xmath51co@xmath52b flux in a manner similar to the growth of other borocarbide crystals . as discussed in ref .
, co doping serves as a convenient tool to move from clean to dirty limit .
nbse@xmath25 crystals were grown via iodine vapor transport technique ( ref ) and had @xmath53 k and the residual resistivity ratio @xmath54 .
mgb@xmath25 single crystals of submillimeter sizes were grown by a high pressure technique as described in ref .. a single crystal of v@xmath26si was grown using a bridgeman method , in which a floating zone was created by rf induction heating .
the samples for investigation were cut by a wire saw and oriented using laue method .
samples had typical dimension of @xmath55 mm .
the magnetization measurements were performed by using several quantum design mpms systems . in a typical experiment ,
a full @xmath56 loop was recorded and only its reversible part , above the irreversibility field , @xmath57 , was used for the analysis .
@xmath57 was determined as a field where ascending and descending branches coincided or were sufficiently close ( a weak hysteresis ) .
[ ptbh ] raw.eps the procedure is demonstrated in fig.[fraw ] .
the main frame shows raw data with clear range of reversible behavior above @xmath57 indicated by an arrow .
the left inset shows the expanded portion of the raw data in the vicinity of @xmath24 . a small paramagnetic background ( from the sample , the sample holder and perhaps residual flux on the crystal surface ) is clearly seen as a linear - in-@xmath58 contribution .
after this contribution is subtracted , we obtain the superconducting diamagnetic signal shown in the right inset .
the @xmath24 is indicated by an arrow .
superconducting transition temperature was measured in a small ( @xmath59oe ) applied field . in large fields of our interest ,
demagnetization effects are weak ; in the following we do not distinguish between the applied field @xmath58 and the induction @xmath14 .
as explained below , we do not need the sample volume in our analysis so that we can use on equal footing field dependencies of the magnetic moment ( in @xmath60 ) or of the magnetization ( in @xmath61 ) ; we use the notation @xmath62 for both quantities .
the standard london energy ( [ f_lond ] ) gives an equilibrium magnetization that is linear in @xmath63 in intermediate fields : @xmath64 hence , the standard london model requires a plot of @xmath62 versus @xmath63 to be a straight line .
fig.[fynbc ] shows _ reversible _ magnetization for a single crystal yni@xmath25b@xmath25c in fields parallel to the @xmath65 axis at @xmath66 , and @xmath67k . clearly , the deviations from the london prediction increase with decreasing @xmath68 ; at low temperatures @xmath3 is far from being linear .
[ tbh ] ynbcm.eps we note that in many materials with @xmath69 , it is difficult to distinguish between a narrow abrikosov domain near @xmath24 with @xmath70 and a broad london domain where the magnetization depends on the field in a slow , nearly logarithmic manner . for this reason the abrikosov part of @xmath50
is sometimes discarded altogether ; this amounts to setting @xmath71 in eq.([m_lond]).@xcite of course , this can not be done for materials with @xmath72 .
we will follow this simplification in our analysis and indicate the cases when this can not be done . to formally account for deviations of the data from the behavior prescribed by the standard london formula , we add to expression ( [ m_lond ] ) two additional terms : const@xmath73 ( to correct for the low field behavior ) and const@xmath74 ( to account for the high field curvature ) .
these are , perhaps , the simplest possible modifications one can think about.@xcite a restriction upon the constants is provided by a requirement that @xmath75 . hence @xmath76 ; \label{m_exp}\ ] ] where @xmath77 .
certainly , the form ( [ m_exp ] ) is not the only possiblity for representing the available data .
other forms were suggested in the literature@xcite based on different theories and assumptions .
we stress that the expression ( [ m_exp ] ) is just an empirical formula to represent the data .
we choose it because of its simplicity , and because - as is demonstrated below - it is sufficiently flexible to represent the magnetization data in a host of materials with very different physical properties .
the solid lines in fig.[fynbc ] are the data fits to eq.([m_exp ] ) .
the value of the upper critical field for each @xmath68 is read directly from the raw data as explained in fig.[fraw ] .
we are left with three fit parameters @xmath78 , and @xmath79 .
two of these are shown in the table of fig.[fynbc ] .
the inset in fig.[fynbcxi ] shows that the @xmath68-dependence of @xmath80 is qualitatively consistent with the behavior of the superfluid density @xmath81 .
the quality of the fits is good ; hence , the empirical form ( [ m_exp ] ) can be used to represent the data reasonably accurately .
examining possible modifications of the london model to account for the deviations of @xmath3 of fig.([fynbc ] ) from linear behavior , one should bear in mind the difference between the roles of two fundamental lengths , @xmath7 and @xmath10 , within the london theory .
the length @xmath7 enters eq.([london ] ) which is the basis of the whole approach . on the other hand ,
the length @xmath10 is absent in the london equation and enters the energy expression as an uncertain cutoff used to mend the inherent shortcoming of the london model .
therefore , considering possible modifications of this model , one still has some freedom - however limited - in working with @xmath10 , unlike the case of @xmath7 . comparing the data of fig([fynbc ] ) with predictions of the standard
london model one wonders why the model which describes correctly the field of vortices away of their cores , fails badly at low temperatures . from the point of view of a consistent london description , the only suspicious point in deriving the free energy ( [ f_lond ] ) and
the corresponding magnetization ( [ m_lond ] ) is the cutoff employed which generates the term @xmath82 .
hence , we write the free energy in the form : @xmath84 where @xmath0 is the _ field dependent cutoff _ ( the core size ) . clearly , @xmath85 is the standard coherence length associated with @xmath24 , so that @xmath86
. then , evaluating @xmath87 , we obtain : @xmath88 the idea of the following manipulation is to find a field @xmath89 that generates the form ( [ m_exp ] ) , or in other words , that represents the experimental data . after equating ( [ m1 ] ) and ( [ m_exp ] ) , one can solve a linear differential equation for @xmath90 with the boundary condition @xmath91 .
the result in terms of @xmath92 and @xmath93 reads : @xmath94 this corresponds to the normalized cutoff distance ( the core radius ) @xmath95 it is readily shown that the slope of @xmath96 at @xmath24 is determined by the parameter @xmath19 : @xmath97 hence , when the field decreases from @xmath24 , @xmath0 decreases for @xmath98 and increases for @xmath99 . using eq.([xi(b ) ] ) , we can calculate the normalized cutoff @xmath100 responsible for deviations of @xmath1 from the standard london behavior for yni@xmath25b@xmath25c shown in fig.[fynbc ] since we have @xmath19 and @xmath79 representing these data sets ( note that @xmath101 does not enter eq.([xi(b ) ] ) ) .
the curves @xmath100 for @xmath102 2 , 5 , 9 , and @xmath67k calculated with eq.([xi(b ) ] ) are shown in fig.[fynbcxi ] .
[ ptb ] ynibcxi.eps it is worth observing that @xmath0 so obtained is qualitatively similar to the @xmath14 dependence of the core size seen in @xmath4sr experiments.@xcite moreover , it is argued in ref . that the field dependence of @xmath10 should weaken with increasing temperature in accord with fig.[fynbcxi ] .
another point to make is only a moderate variation of @xmath10 which is needed to account for strong deviations from the linear @xmath3 .
for example , at 2k , @xmath10 changes only by a factor of 2 over most of the mixed state field domain .
since the cutoff enters the energy ( [ f1 ] ) under the log - sign , it might be surprising that such a difference suffices to cause a drastic deviation of the 2k curve in fig.[fynbc ] from a straight line .
the puzzle is resolved if one observes that the field dependence of @xmath10 translates to _ non - logarithmic _ corrections to the standard london magnetization , see eq.([m1 ] ) .
the same analysis has been applied to the magnetization data for a crystal of luni@xmath25b@xmath25c [ whose crystal structure and superconductive properties are similar to yni@xmath25b@xmath25c ] , yielding similar results as shown in fig.[lut ] . as mentioned , ref .
argues that the field dependence of the core size is weakened by increasing temperature and scattering . in order to study the scattering dependence of @xmath0 we turn to a series of crystals lu(ni@xmath51co@xmath52b@xmath25c in which the mean - free path is progressively reduced by increasing the co content.@xcite [ ptb ] lnbct.eps in the upper panel of fig.[lux ]
, @xmath50 is shown for @xmath103 , 3 , and 6% ( for which @xmath24 has been independently measured ) at the same temperature of 2k .
the fit parameters @xmath19 and @xmath79 are also shown and the calculated field dependent core sizes are given in the lower panel .
[ ptbh ] lucox.eps we note that for 6% co , the ratio of the zero-@xmath68 coherence length to the mean - free path has been estimated in ref . as exceeding 10 , which places this sample close to the dirty limit .
the core size for this crystal is seen to vary only by about 10% , which is in accord with the theoretical finding that field dependence of @xmath10 disappears in the dirty limit.@xcite the superconducting anisotropy of this material is stronger than in borocarbides discussed above . as is seen in fig.[fnbse2 ] ,
deviations of @xmath3 from the standard london linearity are profound along with the corresponding field dependence of @xmath10 .
[ ptb ] nbse2.eps [ ptbh ] mgb2.eps the upper panel of fig .
[ fmgb2 ] shows @xmath3 in fields parallel to the @xmath65 axis of a single crystal mgb@xmath25 .
one readily sees a qualitative difference from the preceding examples : the curvature of @xmath3 for @xmath104k being negative in low fields becomes positive in large fields .
still , we can fit well the data for all @xmath68 s to the form of eq.([m_exp ] ) with parameters @xmath19 and @xmath79 given in the table .
the most interesting feature is that the obtained values of @xmath19 exceed unity . according to eq.([xi(1 ) ] ) this means that starting from @xmath24 , @xmath10 decreases with decreasing field .
this is shown in the lower panel of fig.[fmgb2 ] .
we attribute this unusual behavior to the two - gap nature of this material : the small gap on the @xmath105-sheet of the fermi surface opens up in decreasing fields thus causing a decrease of @xmath10 .
if indeed the unusual behavior of @xmath0 for mgb@xmath25 is due to suppression of the small gap in fields of few kg for @xmath106 , and if the suppression field is _ isotropic _ , then by going to other field orientation away of the @xmath65 axis , we can push the effect of the small gap out of the high field domain of our interest . to check this hypothesis we acquired the data for the applied field at @xmath107 to the @xmath65 axis where @xmath108 is accessible with our equipment .
figure [ 45degr ] shows the result similar to that for y and lu - based borocarbides .
this suggests that , e.g. , for @xmath109k , in the field domain examined ( from @xmath110 t down to about 0.4 t or @xmath111 ) the small gap is not yet fully formed .
[ ptbh ] mgb245.eps of course , this interpretation is much too simple because for other than @xmath112 orientation the strong anisotropy of @xmath10 should be taken into account , the subject of our future work it should also be noted here that the macroscopic phenomenology of magnetic properties of mgb@xmath25 is still debated . despite the two - band nature of this material , within the london approach , we employ _ one _ penetration depth @xmath7 and _ one _ cutoff length @xmath10 when describing the vortex lattice in reciprocal space . judging by literature ,
this point of view is not universally shared by all mgb@xmath25 community . given the slope @xmath113t / k and @xmath114k , it is likely that the low temperature upper critical field of this material exceeds 20 t ; no direct measurements of @xmath24 were conducted in this field range .
one might treat @xmath24 as an extra fitting parameter to be extracted from the data on @xmath50 . however , the numerical procedure of extracting both @xmath24 and @xmath19 from the magnetization data is unstable because their product enters the formulas we use . for this reason we consider here only @xmath50 for @xmath115k for which @xmath24 was measured .
[ ptb ] v3sim.eps it is worth noting that eq .
( [ m_exp ] ) is good enough even for a quite unusual shape of @xmath3 in this material : the curvature of @xmath3 changes sign in all data we have examined . with the help of eq.([m_exp ] ) we readily find that the inflection point is at @xmath116 .
as is seen from the table of fig.[fv3si ] , for all curves given , @xmath117 which leads to @xmath118 hence the inflection points of @xmath119 are approximately in the same place , @xmath120 , for all temperatures.@xcite therefore , the curves @xmath3 are concave for @xmath121 , i.e. , at fields hardly in the domain of applicability of our high - field model . on the other side of the inflection point where the curves @xmath3 are convex , we may deal with the abrikosov domain where @xmath62 is linear in @xmath122 ; it is easy to see then that @xmath3 should be convex.@xcite application of the london approach in this domain can not be justified . therefore , there is no point in trying to extract @xmath0 from the data on v@xmath26si ( formally , since all @xmath19 s in the table of fig.[fv3si ] exceed unity , this extraction would have given @xmath0 decreasing with decreasing field which would have contradicted the @xmath4sr data of ref . ) .
the main point of this work is to argue that the field dependence of the core size is a generic low - temperature property of all sufficiently clean superconductors .
moreover , incorporating this dependence in the london approach broadens considerably its applicability for describing macroscopic reversible magnetic properties .
still , the empirical approach adopted here lacks microscopic justification .
if proven correct , the modified london scheme calls for revisiting many results obtained within the standard london model in which the field independent core size or the cutoff are involved .
field dependencies of the flux - flow resistivity , of the specific heat in the mixed state , or the core pinning are some examples
. one can foresee a number of difficulties in trying to develop such a justification .
the cutoff size we extract from @xmath50 data is not necessarily the same as the core size defined as being proportional to the slope of the order parameter at the vortex axis : approaching the core from the outside to determine the cutoff we may have a different result than when examining the core structure starting from the core center . having this in mind , it is not surprising that the microscopic calculations of @xmath123 in ref
. do not agree in detail with our empirical results ( to our knowledge , this is the only calculation of this sort that has been tried to date ) .
fig.[fmiranovich ] shows the output of this calculation normalized in the manner of this paper for a model isotropic material at @xmath124 and for a few values of the scattering parameter @xmath125 ( @xmath126 is the zero-@xmath68 bcs coherence length and @xmath127 is the mean - free path for the non - magnetic scattering ) .
we note that while the curves generated are mostly qualitatively similar to what we extract from the magnetization data , this calculation does not confirm our assertion about weakening of the field dependence of the core size with increasing scattering .
[ ptbh ] miranovich.eps a different approach to evaluation of the core size is chosen in ref . :
it is argued on physical grounds that since @xmath128 at the vortex center and the field is practically uniform within the core for @xmath69 , one can use the helfand - werthamer@xcite linearization technique employed for calculation of @xmath24 also for the core size in the high - field mixed state . within this approach ,
the order parameter near the vortex center satisfies a linear equation @xmath129 , where @xmath130 and @xmath10 is found solving the basic bcs self - consistency equation .
this produces @xmath131 in qualitative agreement with what we extract from the magnetization data in this work ( for all cases other than v@xmath26si and mgb@xmath25 in field along the @xmath65 axis ) ; in particular , the analytical field dependence of @xmath10 obtained in this way disappears when @xmath132 or @xmath133 . in our view , the question still remains which of these two theoretical approaches , ref . or ref .
, describes better various data on @xmath0 .
an important role in resolving the question belongs to experimental studies of how the field dependence of the core size affects other physical properties of the mixed state . as an example of such a cross - examination we took data on the flux - flow resistivity @xmath134 from measurements of the microwave surface impedance of yni@xmath25b@xmath25c.@xcite
the data show large deviations of the measured @xmath134 from the bardeen - stephen linear field dependence , @xmath135 .
this formula is obtained assuming a field independent core size @xmath136 .
clearly , if @xmath10 does depend on the field , one has : @xmath137 hence , for each data set @xmath138 , we can extract @xmath139 in other words , we can delegate deviations from @xmath140 to the field dependence of @xmath10 and see whether or not the obtained @xmath0 agrees with that extracted from magnetization data . utilising the flux - flow resistivity data of fig.3 of ref . and applying eq.([xi - rho ] ) , we obtain the result shown in the panel ( a ) of fig.[fflow ] . comparing it with our fig.[fynbcxi ] for the same material , we obtain reasonable agreement , notwithstanding the usage of different samples in these two experiments . as another example
, two selections of data for y(ni@xmath51pt@xmath52b@xmath25c from fig.3 of ref . were used to calculate @xmath141 , as shown fig.[fflow](b ) .
we see that in addition to the expected decrease with field , @xmath0 is suppressed by increasing scattering ( i.e. , increasing impurity content of pt ) again in accord with the examples discussed above .
panel ( c ) of fig.[fflow ] shows the result of the same exercise with a d - wave material ( an overdoped crystal of bi-2201).@xcite this example supports the idea that the field dependence of the london cutoff is a generic feature of type - ii superconductors with no direct relation to the order parameter symmetry .
the last panel of fig.[fflow ] presents the cutoff @xmath142 for two field orientations of mgb@xmath25 extracted from the flux - flow resistivity data of ref .
, again using eq.([xi - rho ] ) . for the field along @xmath143
, both @xmath50 and @xmath138 data yield qualitatively similar results .
this , however , is not the case for the field along @xmath65 as evident by comparing figs.[fmgb2 ] and [ fflow](d ) .
thus it appears that for the two - gap mgb@xmath25 with its greater complexity , the simple scheme of incorporating a field - dependent cutoff to the london model does not work for all field orientations .
[ ptbh ] fflow.eps deviations in @xmath3 from the standard london linear behavior have been thought to come from the effects of the nonlocal relation between current and the vector potential inherent for superconductors.@xcite the nonlocal corrections to london theory turned out to be an effective tool in describing evolution and transitions in vortex lattice structures.@xcite however , it is difficult at this stage to sort out what part of the deviations of @xmath62 from linear - in-@xmath63 behavior arises solely from the core - size field dependence and what part should be relegated to the nonlocality . in particular ,
the difficulty comes from analogous weakening of the two effects with increasing temperature and scattering strength .
looking back to the overall satisfactory data analysis of @xmath3 for lu(ni@xmath51co@xmath52 b@xmath25c of ref .
based on nonlocal effects , we note that the analysis produced an excessively rapid reduction of the `` nonlocality range '' with temperature for samples with elevated impurity content .
we have to conclude that more high precision experimental results are needed before a conclusive judgement is made on the relative importance of contributions to @xmath50 due to the nonlocality and to the field dependence of the core size .
as mentioned above , the basic london equation ( [ london ] ) along with the energy expression ( [ f_lond ] ) for intermediate fields ( or eq.([f1 ] ) with an unspecified cutoff ) imply that the london penetration depth @xmath7 is a temperature dependent _ constant _ in a homogeneous material . assuming a field dependent @xmath7 would have changed the london equation ( [ london ] ) _ per se _ : the quantity @xmath145 can not be taken out of differentiation operators . as a result ,
the fourier components of the solution @xmath146 for a single vortex would have been different from @xmath147 and the energy ( [ f_lond ] ) would have been different as well . therefore , unlike the case of the cutoff @xmath10 , relaxing the requirement of a constant @xmath7 causes basic changes in the london approach and therefore should not be taken lightly .
it also worth recalling that a constant @xmath7 is derived from the microscopic bsc theory ( as @xmath148 limit of the bcs kernel in the nonlocal connection between the persistent current and the vector potential ) .
to our knowledge , there is no microscopic justification for a field - dependent @xmath7 ( in non - magnetic superconductors ) . in other words ,
the london theory is rigid with respect to a constancy of @xmath7 , unlike the case of @xmath10 . yet , quite often , analyzing data with the help of the london model ( i.e. , starting with a constant @xmath7 ) it is concluded that @xmath7 should be field dependent .
numerous examples are found in the literature on @xmath4sr ( see , e.g. , ref . ) and in many recent publications on mgb@xmath25.@xcite an operational `` justification implied for this apparent contradiction usually goes like this : of course , the field distribution for a single vortex is described by eq.([london ] ) with a constant @xmath7 . however , in the mixed state the average order parameter is suppressed by overlapping vortex fields , and therefore , the macroscopic '' @xmath149 ( @xmath150 to @xmath7 calculated for @xmath151 ) should enter the free energy eq.([f_lond ] ) .
this macroscopic parameter may depend on the average magnetic field @xmath14 .
the inconsistency of such an argument is exposed if one considers the clean limit at zero temperature . in this case
the london @xmath7 does not depend on the order parameter ( in fact , it depends only on the total electron density ) , so that the mixed - state order parameter suppression can not be reffered to as a general justification for employing @xmath152 . the quantity @xmath153 is often taken as a measure of the superfluid density @xmath154 .
this assignment has unambiguous justification only for isotropic superconductors ( see , e.g. , ref . ) and when @xmath155 is defined as the penetration depth of a small magnetic field ( strictly speaking in the limit @xmath156 ) .
one of the attractive features of the standard london theory is that one can extract @xmath157 directly from the magnetization ( [ m_lond ] ) by measuring the constant slope @xmath158 for each temperature . this way of determining the superfluid density @xmath154 rests , therefore , upon whether or not the standard london model for @xmath62 is valid . as we have seen in a number of examples above , this is quite often not the case .
perhaps the best example of the futility of extracting @xmath154 from magnetization data is provided by the data for v@xmath26si . as is seen in fig.[fv3si ] , e.g. , for @xmath159k , the slope @xmath158 decreases with the field increasing up to @xmath160 t , but it grows with further field increase toward @xmath24 .
if one takes literally the proportionality between @xmath154 and the slope @xmath158 , one should conclude that the field suppresses the superfluid density " as long as @xmath14 is under @xmath161 t but for @xmath162 t , @xmath154 is enhanced by increasing field .
hence , extracting any quantitative information about the superfluid density or the penetration depth from the magnetization data with the help of the _ standard _ london model should be taken with a grain of salt " at best even for such simple " materials as v@xmath26si , not to speak about mgb@xmath25 in which the different field behavior of the two gaps further complicates the matter.@xcite to summarize , the field dependence of the core size @xmath0 has been incorporated in the london model to describe intermediate field reversible magnetization @xmath1 for materials with large @xmath2 . the dependence @xmath0 is directly related to deviations in @xmath3 from linear behavior prescribed by the standard london model . a method to extract @xmath0 from the magnetization data
is proposed .
for most materials examined , @xmath0 so obtained decreases with increasing field ; the dependence becomes weaker with increasing temperature or with strengthening the non - magnetic scattering , in qualitative agreement with theoretical predictions and with existing @xmath4sr , sans , and flux - flow resistivity data .
the method , however , fails when applied to mgb@xmath25 and - surprisingly - to v@xmath26si , the subject for a separate discussion .
the authors thank morten eskildsen for useful discussions and for sharing sans data on cecoin@xmath29 before publication .
ames laboratory is operated for us doe by the iowa state university under contract no .
research at ornl was sponsored by the division of materials sciences and engineering , office of basic energy sciences , u.s .
department of energy , under contract de - ac05 - 00or22725 with oak ridge national laboratory , managed and operated by ut - battelle , llc . r.p .
acknowledges a support from nsf grant number dmr-05 - 53285 and from the alfred p. sloan foundation .
j. karpinski , m. angst , j. jun , s.m .
kazakov , r. puzniak , a. wisniewski , j. roos , h. keller , a. perucchi , l. degiorgi , m.r .
eskildsen , p. bordet , l. vinnikov , and a. mironov , supercond .
* 16 * , 221 - 230 ( 2003 ) .
this suggest a possibility of a certain scaling procedure as a result of which all @xmath119 of fig.[fv3si ] nearly collapse to a single curve .
we will not dwell on this matter , since the physical meaning of such a scaling is still unclear .
one can calculate the linear slope @xmath165 using eq.([m_exp ] ) to obtain @xmath166 .
the abrikosov s slope is @xmath167 with @xmath168 for a triangular lattice at @xmath132 .
it is now easy to see that eq.([m_exp ] ) with parameters @xmath19 and @xmath79 from the table in fig.[fv3si ] for @xmath169k ( which is close to @xmath170k ) gives the slope of the same order as that calculated according to arikosov .
this suggests that in v@xmath26si the abrikosov domain might be very broad .
we note that if @xmath17 would behave as @xmath171 in a broad domain under @xmath24 , the intervortex distance would have been on the order of @xmath17 ; in other words , one would expect abrikosov s linear @xmath50 in this domain .
this speculation is worth of experimental examination .
m. angst , d. di castro , d.g .
eschkenko , r. khassanov , s. kohout , i.m .
savic , a. shengelaya , s.l .
budko , p.c .
canfield , j. jun , j. karpinski , s.m .
kazakov , r.a .
ribeiro , and h. keller , * 70 * , 224513 ( 2004 ) . | the field dependence of the vortex core size @xmath0 is incorporated in the london model , in order to describe reversible magnetization @xmath1 for a number of materials with large ginzburg - landau parameter @xmath2 . the dependence @xmath0 is directly related to deviations in @xmath3 from linear behavior prescribed by the standard london model . a simple method to extract @xmath0 from the magnetization data
is proposed . for most materials examined
, @xmath0 so obtained decreases with increasing field and is in qualitative agreement both with behavior extracted from @xmath4sr and small angle neutron scattering data and with that predicted theoretically . |
[ intro ] many attempts have been done to find a wide but not too inclusive class of functions that can be defined `` special '' , where `` special '' means something more that `` useful '' @xcite .
the actual main line of work for a possible unified theory of special functions is the askey scheme that is based on the analytical theory of linear differential equations @xcite .
however , a possible alternative point of view was established by employing considerations that belong to a field of mathematics seemingly quite far from them : the theory of representations of lie groups .
this way was introduced by wigner @xcite and talman @xcite and later developed mainly by miller @xcite and vilenkin and klimyk @xcite . in this line ,
previous papers by us shown a direct connection between some special functions and well defined lie groups .
the starting point of our work has been the paradigmatic example of hermite functions that are a basis on the hilbert space of the square integrable functions defined on the configuration space @xmath14 . as it is well known in the algebraic discussion of the harmonic oscillator , besides the configuration basis , @xmath15 , a discrete basis , @xmath16 related to the weyl - heisenberg group @xmath17 can be introduced such that hermite functions are the transition matrices from one basis to the other .
this scheme has been generalized to all the orthogonal polynomials we have , up to now , considered : legendre and laguerre polynomials @xcite , associated legendre polynomials and spherical harmonics @xcite .
we discuss here the lie group properties of jacobi polynomials and wigner @xmath2-matrices .
starting from the seminal work by truesdell @xcite , where a sub - class of special functions is defined by means of a set of formal properties , we proposed a possible definition of a fundamental sub - class of special functions , called algebraic special functions " ( asf ) .
these asf look to be strictly related to the hypergeometric functions but are constructed from the following algebraic assumptions : 1 .
a set of differential recurrence relations exists on these asf that can be associated to a set of ladder operators that span a lie algebra .
2 . these asf support an irreducible representation of this algebra .
3 . a hilbert space can be constructed on these asf where the ladder operators have the hermiticity properties appropriate for constructing a unitary irreducible representation ( uir ) of the associated lie group .
second order differential equation that define the asf can be reconstructed from all diagonal elements of the universal enveloping algebra ( uea ) and , in particular , from the second order casimirs of all subalgebras and of the whole algebra . from these assumptions ,
we have that : * applying the exponential map to asf different sets of functions can be constructed . if the transformation is unitary another algebraically equivalent basis of the hilbert space is obtained . when the transformations are not unitary , as in the case of coherent states , sets with different properties are found ( like overcomplete sets ) .
* the asf are also a basis of an appropriate set of @xmath9functions and of an appropriate hilbert space functions .
this , combined with the above properties , implies that the vector space of the operators operating on @xmath9 ( or hilbert ) space functions is homomorphic to the uea built on the algebra .
in @xcite it has been shown that hermite , laguerre and legendre polynomials are asf such that the hermite functions support a uir of the weyl - heisenberg group @xmath17 with casimir @xmath18 , while laguerre functions and legendre polynomials are both bases for the uir of @xmath19 with @xmath20 .
since hermite functions are a basis of square - integrable functions defined on the real line , as well as laguerre functions on the semi - line and legendre polynomials on the finite interval @xcite , all operators acting on such @xmath9functions can be written inside the universal enveloping algebra ( uea ) of @xmath21 or @xmath22 .
all these properties have been shown to not be restricted to the above mentioned 1-rank algebras ( and groups ) .
indeed in @xcite associated legendre polynomials and spherical harmonics are shown to share the same properties with underlying lie group @xmath23 that is of rank 2 like two , @xmath24 and @xmath25 , are the label parameters of these functions . here
we present a further confirmation of this scheme in terms of the jacobi polynomials and wigner @xmath2-matrices showing that they satisfy the required conditions to be considered asf and share the same properties .
indeed they can be associated to well defined `` algebraic jacobi functions '' ( ajf ) that support a uir of @xmath0 i.e. , a lie group of rank 3 like three are the parameters , @xmath26 , of the jacobi polynomials @xmath1 and three @xmath27 are also the ones of @xmath28 matrices . from an applied point of view for both , ajf and wigner @xmath2matrices ,
the relevant group chains are @xmath29 to consider together integer and half - integer spin @xmath8 and @xmath30 to describe them separately .
the paper is organized as follows .
section [ algebraicjacobifunctions ] is devoted to present the main properties of the ajf relevant for our discussion and their relations with the wigner @xmath2-matrices . in section
[ sua(2)otimessub(2 ) ] we study the symmetries of the ajf that keep invariant the principal parameter @xmath8 and change @xmath31 and/or @xmath32 .
we prove that these ladder operators determine a @xmath33 algebra , that allows us to build up a family of uir of the group @xmath34 , i.e. @xmath35 . in section [ su11groups ]
we construct , by means of four new sets of ladder operators that change the three parameters @xmath36 and @xmath32 in @xmath37 , each of them generating a @xmath22 algebra to which infinitely many uir s of @xmath5 supported by the ajf are associated . in sect .
[ su22section ] we show that the ladder operators , obtained in previous sections , span a @xmath4 lie algebra and the ajf and wigner @xmath2-matrices generate a uir of @xmath0 characterized by the eigenvalue of the quadratic casimir @xmath38 .
next section shows the ajf and wigner @xmath2matrices with labels all integers and half - integers can be classified in different and disconnected uir representations of @xmath5 or @xmath23 . in section [ operatorsl2 ] the homomorphism between the space of the operators on the @xmath9 space and the uea of @xmath4
is discussed .
finally some conclusions and comments are included .
a previous unpublished version of this work containing part of the results here presented can be found in @xcite .
[ algebraicjacobifunctions ] we consider the jacobi polynomials , @xmath1 , of degree @xmath39 as defined in terms of the hypergeometric functions @xmath40 @xcite by [ j ] j_n^(,)(x)=_2f_1 , where @xmath41 is the pochhammer symbol , or equivalently by @xcite [ jj ] j_n^(,)(x)=()^n_2f_1 . from an explicit polynomial expression
can be obtained @xcite [ jjj ] j_n^(,)(x)= _ s=0^n ( c n+ + s ) ( c n+ + n - s ) ( ) ^s()^n - s . where we have considered a generalized binomial coefficient @xmath42 being @xmath43 an arbitrary number and @xmath44 a positive integer .
however to obtain objects related to an algebraic structure like in @xcite , we define -alternatively to @xmath45- three other ( discrete ) variables and include an @xmath46-depending factor .
we first substitute @xmath47 with @xmath48 @xmath49 hence @xmath50 then , in a second and final step , we include a @xmath46-depending factor related with the integration measure of the jacobi polynomials .
so , the fundamental objects of this paper , we call `` algebraic jacobi functions '' ( ajf ) , have the final form [ ajf ] j_j^m , q(x):= ( ) ^ ( ) ^ j_j - m^(m+q , m - q)(x ) , where in order to obtain a group representation , as we will prove later , we have to impose the following restrictions for @xmath27 [ conditions ] j|m| , j|q| , 2j , j - m , j - q hence , @xmath27 are all together integers or half - integers ( see fig .
[ fig_00 ] and fig .
[ fig_0 ] where some afs for different integers and half - integers values of @xmath27 are displayed ) .
the conditions rewritten in terms of the original parameters @xmath47 are @xmath51 note that usually the jacobi polynomials @xmath1 are defined for @xmath52 and @xmath53 ( @xmath54 ) in such a way that a unique weight function @xmath55 allows their normalization .
however ( see also @xcite p. 49 ) we have to change such restrictions since we introduce the normalization inside the functions and the algebra requires eq .
( [ conditions ] ) . in principle
-because from the definition @xmath56 for @xmath57- the ajf could be extended to @xmath58 considering @xmath59 as @xmath60 0 & { j<|m| } \end{array}\right . .\ ] ] the ajf reveal additional symmetries hidden inside the jacobi polynomials .
indeed we have [ symmj ] lll j_j^m , q(x)=j_j^q , m(x ) , + j_j^m , q(x)=(-1)^l - m j_j^m ,- q(-x ) , + j_j^m , q(x)=(-1)^l - q j_j^-m , q(-x ) , + j_j^m , q(x)=(-1)^m+q j_j^-m
,- q(x ) .
the proof of these properties is straightforward .
the first one can be proved taking into account the following property of the jacobi polynomials for integer coefficients @xmath47 @xcite @xmath61 the second property can be derived from the well - known symmetry of the jacobi polynomials @xcite @xmath62 and the last two properties can be proved using the first two ones . these ajf for @xmath31 and @xmath32 fixed verify the orthogonality relation [ com ] _
-1 ^ 1j_j^m , q(x)(j+1/2)_j^m , q(x ) dx= _ j j as well as [ completitud ] _
j=_sup_(|m| , |q|)^_j^m , q(x ) ( j+1/2)_j^m , q(y ) = ( x - y ) .
both relations are similar to those of the legendre polynomials @xcite and the associated legendre polynomials @xcite : all are orthonormal only up to the factor @xmath63 .
the jacobi equation @xmath64\ , j_n^{(\a,\b)}(x)=0\ ] ] rewritten in terms of these new functions @xmath65 and of the new parameters @xmath27 becomes [ jacobiequation ] j_j^m , q(x)=0 , where the symmetry under the interchange @xmath66 is evident .
it is worthy noticing that the ajfs , with the substitution @xmath67 , are essentially the elements of the wigner @xmath2 rotation matrices @xcite @xmath68 that verify the conditions .
the explicit relation between them is [ relation ] d^j_q m()= j^m ,- q_j ( ) . eqs .
are equivalent to the well - known relations among the @xmath69 like , for instance , @xmath70 the paper could be rewritten in terms of the @xmath71 but the formal complexity should be larger and , perhaps , this is the reason why the lying algebras were not known up to now . coming back to the ajf , the starting point for finding
the algebra is now the construction of the rising / lowering differential operators that change the labels @xmath27 of the ajf by 1 or 1/2 .
the fundamental limitation of this analytical approach is that the indices are considered as parameters @xcite that , in iterated applications , must be introduced by hand .
this problem has been solved in @xcite where a consistent vector space framework was introduced to allow the iterated use of recurrence formulas by means of operators of which the parameters involved are eigenvalues . indeed in order to display the operator structure on the set @xmath72 we introduce , in consistency with the quantum theory approach , not only the operators @xmath73 and @xmath74 of the configuration space , such that @xmath75= -1\ , , \ ] ] but also three other operators @xmath76 , @xmath77 and @xmath78 such that [ jmq ] j_j^m , q(x ) = j_j^m , q(x ) m j_j^m , q(x ) = m_j^m , q(x ) q j_j^m , q(x ) = q_j^m , q(x ) , i.e. diagonal on the algebraic jacobi functions and , thus , belonging -in the whole algebraic scheme- to the cartan subalgebra .
[ sua(2)otimessub(2 ) ] we start from the differential - difference equations and the difference equations verified by the jacobi functions ( a complete list of which can be found in refs .
the procedure is laborious , so that , we only sketch the simplest case with @xmath79 , related with @xmath80 and well - known for the @xmath2 in terms of the angle @xcite .
let us start from the operators that change the values of @xmath31 only .
consider the equations ( 18.9.15 ) and ( 18.9.16 ) of ref .
@xcite @xmath81 \ds\frac{d}{dx}\left[(1-x)^\alpha ( 1+x)^\beta j_{n}^{(\alpha , \beta)}(x)\right ] & = & \ds -2(n+1 ) ( 1-x)^{\alpha-1 } ( 1+x)^{\beta-1}\,j_{n+1}^{(\alpha-1 , \beta-1)}(x ) , \end{array}\ ] ] which are far to be symmetric , but allow us to define the operators [ adiferential ] a _ : = d_x + ( x m+q ) , that act on the algebraic jacobi functions @xmath65 as [ actiona ] a__j^m , q(x ) = ) j_j^m1 , q(x ) .
the operators ( [ adiferential ] ) are a generalization for @xmath82 of the operators @xmath83 introduced in ref .
@xcite for the associated legendre functions related to the ajf with @xmath84
. moreover eqs .
( [ actiona ] ) , that are independent from @xmath32 , coincide with eqs .
( 2.11 ) and ( 2.12 ) of ref .
@xcite .
defining now @xmath85 and taking into account the action of the operators @xmath86 and @xmath87 on the jacobi functions , eqs . and , it is easy to check that @xmath86 and @xmath87 close a @xmath80 algebra , that commutes with @xmath76 and @xmath78 , denoted in the following by @xmath88 @xmath89=\pm a_\pm \qquad [ a_+,a_-]=2 a_3 .\ ] ] thus , the algebraic jacobi functions @xmath72 such that @xmath90 , @xmath91 and @xmath92 support the @xmath93-dimensional uir of the lie group @xmath94 independently from the value of @xmath32 . similarly to @xcite , starting from the differential realization of the @xmath86 operators we recover the jacobi differential equation from the casimir , @xmath95 , of @xmath88 [ eqcasimira ] _ j^m , q(x ) j_j^m , q(x ) = 0 . indeed , eq .
reproduces the operatorial form of , i.e. [ jacobiequationoperator ] _ j^m ,
q(x ) = 0 .
on the other hand we can make use of the factorization method @xcite , relating second order differential equations to product of first order ladder operators in such a way that the application of the first operator modifies the values of the parameters of the second one .
taking into account this fact , iterated application of gives the two equations l[jacobinodiferential1 ] _
j^m , q(x)=0 , + _ j^m
, q(x)=0 , that reproduce again the operator form of the jacobi equation . equations , are particular cases of a general rule : the defining jacobi equation can be recovered appliying to @xmath96 the casimir operator of any involved algebra and sub - algebra as well as any diagonal product of ladder operators .
now , using the symmetry @xmath66 of the @xmath65 , we construct the algebra that changes @xmath32 leaving @xmath8 and @xmath31 unchanged . from @xmath86 two new operators @xmath97 are thus defined [ bdiferential ] b _ : = d_x+ ( x q+m ) , and their action on the jacobi functions is [ bnodiferential ] b_j_j^m , n(x ) = ) j_l^m , q1(x ) .
obviously also the operators @xmath97 and @xmath98 close a @xmath80 algebra we denote @xmath99 @xmath100=\pm b_\pm \qquad [ b_+,b_-]=2 b_3 , \ ] ] and the jacobi functions @xmath72 with @xmath101 , @xmath102 and @xmath103 close the @xmath104-dimensional uir of the lie group @xmath105 independently from the value of @xmath31
. again we can recover the jacobi equation from the casimir , @xmath106 , of @xmath99 @xmath107 { \cal j}_j^{m , q}(x ) = \left [ b_3 ^ 2 + \frac12\{b_+,b_-\}-j(j+1)\right ] { \cal j}_j^{m , q}(x)=0\ ] ] as well as from the diagonal second order products @xmath108\ ; { \cal j}_j^{m , q}(x)=0\ ; , \\[0.3cm]\left[b_-\,b_+ -(j - q)\,(j+q+1)\right]\ ; { \cal j}_j^{m , q}(x)=0\;. \end{array}\ ] ] a more complex algebraic scheme appears in common applications of the operators @xmath86 and @xmath97 .
as the operators @xmath109 commute with @xmath110 , the algebraic structure is the direct sum of the two lie algebras @xmath111 a new symmetry of the ajfs emerges in the space of @xmath65 with @xmath8 fixed . for @xmath48 , all integer or half - integer , formulae , and are the well known expressions for the infinitesimal generators of the group @xmath112 .
the jacobi functions @xmath65 for fixed @xmath8 and @xmath113 , @xmath114 determine a uir of this group . from and , taking into account that always the operators @xmath77 and @xmath78 have been written at the right of @xmath73 and @xmath74 , it can be shown that @xmath115 and the representation is , as required , unitary . in fig .
[ fig_1 ] the action of the operators @xmath116 on the parameters @xmath27 that label the jacobi functions corresponds to the plane @xmath117 .
[ su11groups ] we mentioned before that many difference and differential - difference relations for jacobi polynomials are known @xcite .
starting from them a su(2,2 ) lie algebra can be constructed .
it has fifteen infinitesimal generators , where three of them are cartan generators ( for instance , @xmath118 and @xmath78 ) .
as the four generators that commute with @xmath76 ( @xmath86 and @xmath97 ) have been introduced in the preceeding paragraph , we have to construct eight non - diagonal operators more .
they are : [ differential8 ] l c_+:=+ d_x - ( x(j+1)-(j+1+m+q ) ) , + c_-:=- d_x - ( x j- ( j+m+q ) ) , + d_+:=- d_x + ( x ( j+1)+ ( j+1+m - q ) ) , + d_-:=+ d_x + ( xj + ( j+m - q ) ) , + e_+:=- d_x+ ( x ( j+1)+(j+1-m+q ) ) , + e_-:=+ d_x+ ( x j+(j - m+q ) ) , + f_+:=- d_x+ ( x ( j+1)-(j+1-m - q ) ) , + f_-:=+ d_x+ ( x j-(j - m - q ) ) .
all these differential operators act on the space @xmath119 for @xmath120 integer and half - integer such that @xmath121 .
the explicit form of their action is : [ nodifferential8 ] rcl c_+j_j^m , q(x)&=&_j+1/2^m+1/2,q+1/2(x ) , + c_-j_j^m , q(x)&=&_j-1/2^m-1/2,q-1/2(x ) , + d_+j_j^m , q(x)&=&_j+1/2^m+1/2,q-1/2(x ) + + d_-j_j^m , q(x)&=&_j-1/2^m-1/2,q+1/2(x , ) + e_+j_j^m , q(x)&= & _ j+1/2^m-1/2,q+1/2(x ) , + e_-j_j^m , q(x)&=&_j-1/2^m+1/2,q-1/2(x ) , + f_+j_j^m , q(x)&= & j_j+1/2^m-1/2,q-1/2(x ) , + f_-j_j^m , q(x)&=&_j-1/2^m+1/2,q+1/2(x ) . from ( or remembering the order in the operators ) we have @xmath122 i.e. all these rising / lowering operators have the hermiticity properties required by the representation to be unitary .
the operators change all parameters by @xmath37 .
we define all together @xmath123 the ones that change @xmath8 in @xmath124 , that in fig .
[ fig_1 ] correspond to the planes @xmath125 . from the eqs .
it is easily stated that llr[weylsym ] d_(x , d_x , m , q ) & = & c_(-x ,- d_x , m ,-
q ) , + e_(x , d_x , m , q ) & = & c_(-x ,- d_x ,-
m , q ) , + f_(x , d_x , m , q ) & = & - c_(x , d_x ,- m ,- q ) .
thus , because of the weyl symmetry of the roots , we limit ourselves to discuss the operators @xmath126 .
taking thus into account their action on the jacobi functions we get [ suc11 ] [ c_+,c_-]= -2 c_3 , = c _ where [ c3spectrum ] c_3:= j+ ( m+q)+12 .
hence @xmath127 close a @xmath22 algebra that is denoted @xmath128 . as in the cases of the operators @xmath86 and @xmath97 , we obtain the jacobi differential equation up to a non vanishing factor from the casimir @xmath129 of @xmath128 , written in terms of and , [ casimirsuc11 ] c_cj_j^m , q(x)_j^m , q(x)=_j^m , q(x ) .
indeed l[eqcasimirc ] _
j^m , q(x ) + j_j^m , q(x ) = 0 allows us to recover the jacobi eq . .
analogously the same result derives from eqs .
l[cjacobinodiferential1 ] _
j^m , q(x ) = 0 , + _
j^m , q(x)=0 , obtained by the factorization method . from we see that since @xmath130 the ir of su(1 , 1 ) with @xmath131 are obtained . hence , the set of ajf supports many infinite dimensional uir , of @xmath19 of the discrete series for @xmath132 @xcite .
similar results can be found for the other ladder operators @xmath133 with the substitutions in all eqs .
( [ suc11]-[cjacobinodiferential1 ] ) .
[ su22section ] if one represents the action of the twelve operators @xmath134 , that we have defined in previous sections , we obtain fig . [ fig_1 ] .
to obtain the root system of the simple lie algebra @xmath135 we have only simply to add three points in the origin corresponding to the elements @xmath118 and @xmath78 of the cartan subalgebra .
the lie commutators of the generators @xmath136 are @xmath137=0,\quad & [ j , m]=0,\quad & [ j , b_\pm]=0,\quad & [ j , q]=0 , \\[0.4 cm ] [ j , c_\pm]=\pm \frac12 \,c_\pm,\quad & [ j , d_\pm]=\pm \frac12 \,d_\pm,\quad & [ j , e_\pm]=\pm \frac12 \,e_\pm,\quad & [ j , f_\pm]=\pm \frac12 \,f_\pm , \\[0.4 cm ] [ m , b_\pm]=0,\quad & [ m , q]=0 , & & \\[0.4 cm ] [ m , c_\pm]=\pm\frac12\ , c_\pm,\quad & [ m , d_\pm]=\pm\frac12\ , d_\pm,\quad & [ m , e_\pm]=\mp\frac12\ , e_\pm,\quad & [ m , f_\pm]=\mp\frac12\ , f_\pm , \\[0.4 cm ] [ q , a_\pm]=0,\quad & & & \\[0.4 cm ] [ q , c_\pm]=\pm \frac12 \,c_\pm,\quad & [ q , d_\pm]=\mp \frac12 \,d_\pm,\quad & [ q , e_\pm]=\pm \frac12 \,e_\pm,\quad & [ q , f_\pm]=\mp \frac12 \,f_\pm , \\[0.4 cm ] [ a_+,a_-]=2 a_3,\quad & [ a_3,a_\pm]=\pm a_\pm , & ( a_3=m ) , & \\[0.4 cm ] [ b_+,b_-]=2 b_3,\quad & [ b_3,b_\pm]=\pm b_\pm , & ( b_3=q ) , & \\[0.4 cm ] [ c_+,c_-]=-2 c_3,\quad & [ c_3,c_\pm]=\pm c_\pm , & ( c_3= j+ \frac{1}{2}(m+q)+\frac12 ) , & \\[0.4 cm ] [ d_+,d_-]=-2 d_3,\quad & [ d_3,d_\pm]=\pm d_\pm , & ( d_3= j+ \frac{1}{2}(m - q)+\frac12 ) , & \\[0.4 cm ] [ e_+,e_-]=-2 e_3,\quad & [ e_3,e_\pm]=\pm e_\pm , & ( e_3= j+ \frac{1}{2}(-m+q)+\frac12 ) , & \\[0.4 cm ] [ f_+,f_-]=-2 f_3,\quad & [ f_3,f_\pm]=\pm f_\pm , & ( f_3= j-\frac{1}{2}(m+q)+\frac12 ) , & \\[0.4 cm ] [ a_\pm , b_\pm]=0,\quad & [ a_\pm , b_\mp]=0,\quad & & \\[0.4 cm ] [ a_\pm , c_\pm]=0,\quad & [ a_\pm , c_\mp]=\pm e_\mp,\quad & [ a_\pm , d_\pm]=0,\quad & [ a_\pm , d_\mp]=\mp f_\mp , \\[0.4 cm ] [ a_\pm , e_\pm]=\pm c_\pm,\quad & [ a_\pm , e_\mp]=0,\quad & [ a_\pm , f_\pm]=d_\pm , \quad & [ a_\pm , f_\mp]=0 , \\[0.4 cm ] [ b_\pm , c_\pm]=0,\quad & [ b_\pm , c_\mp]=\mp d_\mp,\quad & [ b_\pm , d_\pm]=\pm c_\pm , \quad & [ b_\pm , d_\mp]=0 , \\[0.4 cm ] [ b_\pm , e_\pm]=0,\quad & [ b_\pm , e_\mp]=\mp f_\mp , \quad & [ b_\pm , f_\pm]=\pm e_\pm , \quad & [ b_\pm , f_\mp]=0 , \\[0.4 cm ] [ c_\pm , d_\pm]=0 , \quad & [ c_\pm , d_\mp]=\mp b_\pm,\quad & [ c_\pm , e_\pm]=0 , \quad & [ c_\pm , e_\mp]=\mp a_\pm , \\[0.4 cm ] [ c_\pm , f_\pm]=0 , \quad & [ c_\pm , f_\mp]=0 , & & \end{array}\ ] ] @xmath138=0 , \quad & [ d_\pm , e_\mp]=0,\quad & [ d_\pm , f_\pm]=0 , \quad & [ d_\pm , f_\mp]=\mp a_\pm , \\[0.4 cm ] [ e_\pm , f_\pm]=0,\quad & [ e_\pm , f_\mp]=\mp b_\pm . & & \end{array}\ ] ] the quadratic casimir of @xmath4 has the form @xmath139 & \qquad\qquad\ds+\ , \frac{1}{2}\,\left ( a_3 ^ 2 + b_3 ^ 2 + c_3 ^ 2 + d_3 ^ 2 + e_3 ^ 2 + f_3 ^ 2 \right)\\[0.4 cm ] & = \ds \frac12\left(\{a_+,a_-\}+\{b_+,~b_-\}-\{c_+,c_-\}-\{d_+,d_-\ } -\{e_+,e_-\}-\{f_+,f_-\}\right)\\[0.3 cm ] & \qquad\qquad\ds+\,2 j(j+1)+m^2+q^2+\frac12\\[0.4 cm ] & \ds \equiv -\frac{3}{2 } .
\end{array}\ ] ] from it and taking into account the differential realization of the operators involved , , and , we recover again the jacobi equation .
hence , the ajf support a uir of the group @xmath0 with the value -3/2 of @xmath140 ( see fig . [ fig_3 ] ) . also , as we have seen along the previous sections , the jacobi equations is recovered form the casimir of any subalgebra of @xmath141 as well as from any diagonal product of ladder operators . in this uir of @xmath0
the integer and half - integer values of @xmath48 are putted all together .
the symmetries where integer and half - integer values belong to different uir will be discussed in next section .
[ symmetriessu11 ] in the preceding sections we have obtained a set of symmetries on the ajf @xmath65 where the operators @xmath86 ( @xmath97 ) change only the label @xmath31 ( @xmath32 ) of @xmath142 leaving invariant the remaining two labels and the other operators @xmath123 of the algebra change all the @xmath27 by a half - integer quantity .
integer and half - integer values of @xmath27 are indeed related to a unique uir of @xmath0 .
now , composing the action ot pairs of operators @xmath123 , we can construct two operators that change @xmath27 to @xmath143 .
indeed @xmath144 ( equivalently , @xmath145 , @xmath146 or @xmath147 ) are such operators
. moreover they , in general second order differential operators , reduce because of the jacobi equation ( [ jacobiequationoperator]) to a first order differential operators when applied to the @xmath96 . as discussed before the @xmath96 have @xmath148 and @xmath149 , but now we have to consider separately specific values of @xmath31 and @xmath32 .
to begin with let us start with states such that @xmath150 . from ( [ differential8 ] ) and ( [ jacobiequationoperator ] ) we get that the two conjugate hermitian operators [ kpm ] k_+:= f_+ c_+ , k_-:= f_- c_- can be written on the ajf as [ njmasmenosoperators ] k_+ & : = & ( -(1-x^2 ) d_x+ x ( j+1)+ ) + k_- & : = & ( ( 1-x^2 ) d_x+ x j+ ) [ k- ] always well defined because @xmath151 on these ajf . by inspection
, their action does not depend from the value of @xmath32 : lll [ actionk ] k_+_j^m , q(x ) & = & _ j+1^m , q(x ) , + k_-_j^m , q(x ) & = & _ j-1^m , q(x ) , and @xmath152 together with @xmath153 close a @xmath22 lie algebra @xmath154=-2 k_3,\qquad [ k_3,k_\pm]=\pm k_\pm\ ] ] such that the @xmath96 with @xmath31 and @xmath32 fix and @xmath155 are a basis of the uir of @xmath22 with casimir @xmath156 for the states where @xmath157 the procedure is analogous : we have only to interchange @xmath158 in @xmath152 .
the problem is more complex when @xmath159 as the action of @xmath160 is not well defined in eq.([k- ] ) for @xmath161 . to extend its definition to this case we have to consider not the eigenvalues of @xmath77 and @xmath78 but their limit i.e. @xmath162.\ ] ] in this way the action of @xmath160 does not change for @xmath163 while for @xmath161 , results from the product of a first factor that goes like @xmath164 an a second one that goes like @xmath165 and it is thus zero . in conclusion ,
all @xmath96 close , for fixed @xmath31 and @xmath32 , a uir of @xmath5 . if @xmath166 we have eqs .
( [ actionk ] ) with @xmath167 and casimir invariant @xmath168 while , for @xmath169 , we have to exchange everywhere @xmath32 and @xmath31 .
note that , unlike the uir of @xmath0 , each representation of @xmath5 contains only states with integer or half - integer values of its labels .
this is particularly relevant for the @xmath170 as usual symmetries in physics do not mix integer and half - integer spins i.e. bosons and fermions .
this @xmath5 can not in general be extended to larger groups as its operators can not be combined with other operators to construct a bigger algebra , but there are few exceptions .
indeed when @xmath84 , i.e. on the @xmath171 with @xmath172 and @xmath173 , we can define not only the @xmath174 but also the @xmath109 and thus the whole algebra @xmath175 described in @xcite .
the @xmath176 are indeed related to the associate legendre polynomials @xmath177 and thus a basis of a uir of @xmath23 . by inspection @xmath152
are exactly the ones reported in fomulas ( 2.13 - 14 ) of @xcite and @xmath86 are the @xmath83 of formulas ( 2.11 - 12 ) of the same paper . because of the symmetry @xmath66 we have a @xmath23 also for @xmath178 i.e. on the @xmath179 .
also `` fermions '' states @xmath180 and @xmath181 are related to the same algebra @xmath175 but , of course , to a representation of its covering group @xmath182 @xcite .
[ operatorsl2 ] as already discussed in details in @xcite and @xcite , special functions together with their lie group properties play a role in @xmath9functions and hilbert spaces
. formulas ( [ com ] ) and ( [ completitud ] ) where @xmath31 and @xmath32 are fixed allow to introduce @xmath8 and @xmath46 as conjugate variables on the same hilbert space .
let us thus begin discussing the case of section [ symmetriessu11 ] , where @xmath31 and @xmath32 are fixed parameters characterizing the uir of @xmath22 and its support hilbert space .
operators ( [ kpm ] ) written as ( [ actionk ] ) on the @xmath72 allow to define , in abstract form , for fixed @xmath31 and @xmath32 , the action on a hilbert space constructed in the space of eigenvectors @xmath183 of the operators @xmath76 as lll [ actionkh ] k_+ |j , ( m , q)&= & |j+1 , ( m , q ) , + k_- |j , ( m , q)&= & |j-1 , ( m , q ) . and
@xmath183 is a basis of the hilbert space i.e. [ comh ] j , ( m , q)| j , ( m , q ) = _ j j _ j=_sup_(|m| , |q|)^| j , ( m , q ) j , ( m , q)| = i .
this allows to define [ bb ] |x , ( m , q):= _
j=_sup_(|m| , |q|)^ @xmath184 are a basis of the configuration space @xmath185 i.e. @xmath186 this implies that the @xmath72 are the transition matrices between the two bases : @xmath187 and [ 123 ] |j,(m , q)= _
|x,(m , q ) j_j^m , q(x ) dx . as @xmath31 and @xmath32
are invariant of the representation , this situation is similar to the lagrange functions discussed in @xcite , where only two conjugate variables @xmath8 and @xmath46 ( one discrete and one continuous ) are present .
the difference is that in @xcite a basis was associated to each element @xmath188 of the group @xmath5 , while here we have a basis for each element of the triple @xmath189 .
the case of two discrete variables , related to the group @xmath23 , has been discussed in @xcite and it will be not reconsidered here . on the contrary we consider the new case of three variables case , related to @xmath0 where both @xmath31 and @xmath32 are modified by the algebra .
the hilbert space is now @xmath190 , where @xmath191 is again @xmath192 , @xmath193 is related to @xmath31 and @xmath194 to @xmath195 , as @xmath31 and @xmath32 are together integer or half - integer .
the space @xmath190 , with basis @xmath196 , is the direct sum of the hilbert spaces @xmath197 with @xmath31 and @xmath32 fixed , @xmath198 orthonormality and completeness are now , @xmath199 analogously to ( [ 123 ] ) we can now define inside the hilbert space a new basis @xmath200 with @xmath201 by @xmath202 so that @xmath203 the @xmath72 plays the role of transition matrices : @xmath204 like in ref .
@xcite the role of the algebraic jacobi functions @xmath72 as transition matrices reflects the fact that the generators can be seen as differential operators in @xmath190 , or algebraic operators in the spaces of labels @xmath205 , where @xmath206 is related to @xmath8 and the two @xmath207 to @xmath208 and @xmath209 .
this allows to make explicit the lie algebra structure in contrast with ref .
@xcite .
an arbitrary vector @xmath210 can be alternatively expressed as @xmath211 where @xmath212 f_j^{m , q } & : = & \langle j , m , q|f\rangle & = & \ds \int_{-1}^{+1}dx\ ; \sqrt{j+1/2}\;{\cal j}_j^{m , q}(x)\ , f^{m , q}(x)\ ; \end{array}\ ] ] and all the @xmath9functions defined on @xmath213 can be written as [ fl2 ] _ m , q=-^_j = sup(|m|,|q|)^ _ j^m , q(x ) f_j^m , q .
as the @xmath119 are a basis of a uir of @xmath0 and ( see eq.([fl2 ] ) ) , at the same time , a basis of the @xmath9functions defined on @xmath213 the @xmath214 belong to the same uir of @xmath0 .
this implies that every change of basis in the @xmath214 is related to an element @xmath188 of the group @xmath0 and that every operators that acts on the @xmath214 can be written inside the universal enveloping algebra of @xmath4 .
the algebraic jacobi functions are a particular case of the algebraic special functions recently introduced by us @xcite whose relevance seems to be related to the following points : 1 .
the role of intertwining between second order differential equations and lie algebras played by the algebraic special functions . 2 .
peculiar properties of the algebraic special functions , that perhaps could be assumed as their definition , look to be that taking into account the fundamental second order differential equation all diagonal elements of the uea can be found to be related to it and that all non - diagonal elements can be written as first order differential operators .
3 . the fact that asf ( here ajf ) are at the same time an irreducible representation of a lie algebra ( here @xmath4 ) and a basis of @xmath9functions ( here the ones defined on @xmath215 allows to establish a homorphism between the uea of the lie algebra and the vector space of the operators defined on the @xmath9functions . in this way the framework of quantum mechanics is reproduced with @xmath214 hilbert space and @xmath4-uea as space of operators acting on it .
4 . as the asf are a basis of a unitary irreducible representation of the corresponding lie group also , all sets obtained from asf applying an element @xmath188 of the lie group are bases in the space of the @xmath9functions .
moreover in the @xmath5 cases presented in sect .
[ symmetriessu11 ] each basis depends not only on the group element @xmath188 but also on the pair @xmath216 .
although the uir of @xmath0 put together integer and half - integer values of the labels @xmath27 , the analysis of sect . [ symmetriessu11 ] separates boson and fermions systems as usual in physical applications . in this case
@xmath217 is the hilbert space and @xmath22-uea is the space of operators acting on it .
the algebraic jacobi functions have the same symmetry of the elements of the @xmath2-matrices .
this enhances the interest of the algebraic special functions in physical applications since the wigner matrices play an important role in the description of quantum mechanics .
this work was partially supported by the ministerio de educacin y ciencia of spain ( projects fis2009 - 09002 with eu - feder support ) , by the universidad de valladolid and by infn - micinn ( italy - spain ) . | a symmetry @xmath0 group in terms of ladder operators is presented for the jacobi polynomials , @xmath1 , and the wigner @xmath2-matrices where the spins @xmath3 integer and half - integer are considered together .
a unitary irreducible representation of @xmath0 is constructed and subgroups of physical interest are discussed .
the universal enveloping algebra of @xmath4 also allows to construct group structures ( @xmath5 , @xmath6 , @xmath7 ) whose representations separate integers and half - integers values of the spin @xmath8 .
appropriate @xmath9functions spaces are realized inside the support spaces of all these representations .
operators acting on these @xmath9functions spaces belong thus to the corresponding universal enveloping algebra .
* lie groups of jacobi polynomials + and + wigner @xmath10-matrices * e. celeghini@xmath11 , m.a .
del olmo@xmath12 and m.a .
velasco@xmath12 @xmath13_dipartimento di fisica , universit di firenze and infn sezione di firenze + i50019 sesto fiorentino , firenze , italy _
+ @xmath12_departamento de fsica terica and imuva , universidad de valladolid , + e-47011 , valladolid , spain . _
+ e - mail : [email protected] , [email protected] keywords : special functions , jacobi polynomials , wigner @xmath2-matrices , lie algebras , square - integrable functions |
we studied the xxz model for an open chain of @xmath15 sites , which is divided into a left part with @xmath3 sites and a right part with @xmath16 sites .
both are connected by a bond of strength @xmath13 while for the rest of the chain we take @xmath17 .
the hamiltonian then has the form h=_i=1^l-1h_i+j_imph_l+_i = l+1 ^ 2l-2h_i [ chain ] where h_i= ( s_i^+s_i+1 ^ -+s_i^-s_i+1^+ ) + s_i^zs_i+1^z [ h ] and the anisotropy @xmath14 lies between @xmath18 and @xmath19 .
written in terms of spinless fermions , the @xmath20-term corresponds to a nearest - neighbour interaction which is repulsive ( attractive ) for @xmath12 ( @xmath11 ) .
the system was studied in its ground state which in the present case corresponds to total spin @xmath21 , and we took @xmath22 . in the fermionic formulation ,
the filling is @xmath23 and thus slightly above @xmath24 .
we chose the odd total length in order to avoid relatively large oscillations between even and odd @xmath3-values which are present in even - length chains .
these are connected with the different dimerization patterns in the spin correlation functions which are induced by the open ends .
although this is an interesting topic , we preferred to avoid it for the present purpose .
+ to determine the effect of the impurity bond on the entanglement entropy @xmath8 between left side and right side , we performed extensive dmrg calculations .
as usual in such calculations , we first determined the ground state wavefunction of ( [ chain ] ) and then traced out the spin degrees of freedom of the right side to obtain the reduced density matrix @xmath1 .
the defect was inserted only at the last step of the procedure .
we worked with @xmath25 states and used no sweeps .
the values of @xmath3 varied between 3 and 401 , corresponding to a total length of up to @xmath26 sites .
the computations were performed on the hpsc45 of icts , cas .
+ to judge the accuracy of these calculations , we used the xx case , @xmath27 , where the system becomes a simple free - fermion hopping model .
then one can determine @xmath1 from the one - particle correlation function @xmath28 by diagonalizing the matrix @xmath29 in the subsystem @xcite .
such calculations were done recently for segments in homogeneous chains with interface defects @xcite and in random chains @xcite . for the present geometry with open ends , this correlation function can be determined analytically for the pure system ( @xmath30 ) .
then one finds c_ij= [ corr ] with a defect , @xmath31 has to be calculated numerically using the single - particle eigenfunctions of @xmath32 and the filling .
the corresponding results were considered as benchmarks . in fig.[benchmark ] we show the difference between the dmrg results and the correlation function results for three defect strengths .
both @xmath33 and @xmath34 increase with the chain length , as one would expect .
from the upper panel one sees that the error in the entropy becomes about @xmath35 for @xmath36 .
the error in the effective central charge shown in the lower panel is larger and shows more noise .
this is due to the numerical derivative by which one calculates @xmath0 c(l)= 6 [ ceff ] and in which small differences enter .
thus @xmath34 reaches about @xmath37 at @xmath36 , which is still sufficient for our purposes .
we expect the same accuracy also for the interacting system with @xmath14 non - zero .
note that in ( [ ceff ] ) steps of 2 are used , because there still is a small even - odd oscillation in @xmath8 .
the two resulting values for @xmath0 approach each other for increasing @xmath3 . in the following
we always give results obtained for odd @xmath3 .
+ and @xmath38 .
upper panel : difference in the entanglement entropy @xmath8 .
lower panel : difference in the effective central charge @xmath0 .
, width=257 ]
before we present the results for the entanglement entropy we show some density - matrix spectra , since they determine the value of @xmath8 . in fig.[spec1 ] spectra of the pure chain for fixed length and five different values of @xmath14 are shown on the left .
plotted are the 100 largest eigenvalues @xmath39 of @xmath1 , ordered according to their magnitude .
one sees the typical initial exponential decrease which then flattens out .
a variation of @xmath14 leads only to relatively small changes in the curves and @xmath8 will also not be affected much .
note , however , that @xmath8 is mainly determined by the first few large eigenvalues . .
please note the logarithmic scale .
the insets show the first three eigenvalues.,title="fig:",width=257 ] .
please note the logarithmic scale .
the insets show the first three eigenvalues.,title="fig:",width=257 ] on the right , the analogous spectrum is plotted for a defect with strength @xmath40 . here
one can see a marked variation with @xmath14 .
the decay is fastest for @xmath41 and slowest for @xmath42 . because of the sum rule @xmath43 , a slower decay
is coupled with a smaller value for the first or first few eigenvalues @xcite .
this can be seen in the inset of the figure . +
the size dependence of the spectra is shown in fig.[spec2 ] for two values of @xmath14 . in both cases
the curves become flatter as @xmath3 increases , which is well - known from other dmrg calculations .
the behaviour of the first eigenvalues , however , is different . for @xmath41 one
has a region of relatively slow initial decay which becomes even slower as @xmath3 increases .
the first two eigenvalues decrease in accord with the remark made above . for @xmath8 one therefore expects a relatively large value , because several @xmath39 are of order 1 , and a marked size dependence . for @xmath42 ,
on the other hand , there is a gap between the first and the second eigenvalue and no visible change with @xmath3 . because of the rather fast decay of the complete spectrum , the sum rule is not effective here .
moreover , the largest eigenvalue @xmath44 is close to 1 and thus gives only a small contribution to @xmath8 .
thus one expects a small absolute value and a weak size dependence of @xmath8 here . and various sizes @xmath3 .
left : @xmath45 .
right : @xmath46 .
the insets show the first three eigenvalues.,title="fig:",width=257 ] and various sizes @xmath3 .
left : @xmath45 .
right : @xmath46 .
the insets show the first three eigenvalues.,title="fig:",width=257 ]
results for the entropy @xmath47 are shown in fig.[entro ] for @xmath48 and sizes up to @xmath49 .
one can see a monotonic increase with @xmath3 for all values of @xmath14 .
the absolute value of @xmath8 and the dependence on @xmath3 are large for negative @xmath14 and small for positive @xmath14 , confirming the expectations based upon the spectra .
the plot against @xmath50 on the right hand side shows relatively straight curves with similar slopes in the first case but rather bent curves which seem to become horizontal in the second one .
thus there are characteristic differences between repulsive and attractive interactions . and seven values of the anisotropy .
the data correspond to @xmath51 , from top to bottom.,title="fig:",width=257 ] and seven values of the anisotropy .
the data correspond to @xmath51 , from top to bottom.,title="fig:",width=257 ] a determination of the slope and the effective central charge via ( [ ceff ] ) leads to the results shown in fig .
[ ceff05 ] .
the quantity @xmath0 clearly tends to the value 1 of the pure system for large negative @xmath14 and to zero for large positive ones .
the data for @xmath52 also seem to fit into the pattern , while for @xmath27 the limit obviously differs from 1 .
this corresponds to the results in @xcite where a continuous variation of the effective central charge with the defect strength was found in this case . for a detailed comparison
, one has to take into account that the quantity @xmath9 used there referred to a subsystem coupled to the rest with e.g one modified and one unmodified bond .
the contributions of these two bonds add up and the relation to @xmath0 is therefore c_eff/3 = c/6 + 1/6 [ ceff1 ] in @xcite the asymptotic value of @xmath9 was determined by an extrapolation in @xmath53 .
the same procedure also works here and the results agree to 3 - 4 decimal places . in view of the rather different geometries in the two cases ,
this is a non - trivial consistency check of the calculations .
+ . left : as function of the system size for @xmath54 , from top to bottom .
right : as function of the anisotropy.,title="fig:",width=257 ] . left : as function of the system size for @xmath54 , from top to bottom .
right : as function of the anisotropy.,title="fig:",width=257 ] the right part of fig.[ceff05 ] gives @xmath0 as a function of @xmath14 and shows the different behaviour for positive and negative values very clearly . to show how these effects vary with the impurity strength , we present results for two other values of @xmath13 . in fig .
[ ceff02 ] this is done for a weaker bond , corresponding to a stronger perturbation .
then @xmath0 for the non - interacting case is only about 0.25 and even smaller for repulsion . for weak attraction ,
the values rise only slowly but for strong attraction they exceed 1 and approach it from above . , but for @xmath55 .
, title="fig:",width=257 ] , but for @xmath55 .
, title="fig:",width=257 ] in fig.[ceff09 ] , on the other hand , the results for an only slightly weakened bond , @xmath56 , are shown . in this case , not only the @xmath0 values for negative @xmath14 lie rather close to 1 , but also those for @xmath57 . only for the larger @xmath14 one
has a reduction below 1 , but the values are still rather high . from the left figure one sees that the curves for these @xmath14 bend downwards and presumably approach zero for longer chains .
however , the asymptotic region is beyond the range of our system sizes . , but for @xmath58 .
please note the different vertical scale in the left figure.,title="fig:",width=257 ] , but for @xmath58 .
please note the different vertical scale in the left figure.,title="fig:",width=257 ]
these results show a clear distinction between attractive and repulsive interactions and a crossover between the two which is determined by the strength of the perturbation .
this strength fixes the value of @xmath0 in the non - interacting system .
because this value depends only weakly on the size , it serves as a reference point through which the @xmath59 curves for different sizes go .
the curves , which we determined only at a number of points , will be continuous in all finite systems .
these findings are in line with the general predictions on the influence of an impurity in a luttinger liquid , namely that for attraction the perturbation heals with increasing system size , while for repulsion it becomes stronger . for the weakened bonds which we have considered
, the latter case means a scaling towards zero and a cutting of the chain .
the vanishing of @xmath0 is in agreement with this picture , but the cutting should not be taken too literally since it only refers to processes near the fermi energy . as fig.[entro ] shows , there remains a residual entanglement corresponding to the constant @xmath60 in ( [ log ] ) which depends on the interaction strength .
+ there is no general expression for @xmath8 in a finite system , not even for the xx case , but in @xcite a formula was given for a weak impurity in a luttiger liquid .
the contribution to @xmath8 is then s= -y^2 ^ 2 ( ) ^1-g ( ) - b(g ) [ levine ] where @xmath61 is the strength of the impurity potential , @xmath62 a small - distance cutoff , @xmath63 a positive constant and @xmath64 the luttinger liquid parameter measuring the interaction . for the xxz lattice model
, it can be expressed as = 2 ( 1- arccos ) [ luttinger ] for attraction ( @xmath11 ) one has @xmath65 and for repulsion @xmath66 . in the latter case , @xmath67 is always negative and diverges with @xmath3 .
the numerical results for @xmath67 on the left of fig.[delta - s ] actually show this behaviour .
one can also fit the data to equ.([levine ] ) as is shown on the right for @xmath68 .
the value of @xmath64 is then 0.75 and the fit gives @xmath69 and @xmath70 .
however , similar fits for @xmath71 and @xmath72 lead to very different results for @xmath62 and if one interprets it as an effective length determined by the scaling equations there are still inconsistencies . for attraction the @xmath3-dependent term in ( [ levine ] ) is positive and decreases with @xmath3 but the data for @xmath67 are more or less @xmath3-independent and can not be fitted with reasonable parameters .
thus we can not verify levine s formula in detail .
one should mention that he also predicts a @xmath73 term in @xmath67 for the non - interacting system which one does not see in the numerics
. it could be that he used a too simple geometry for the two - dimensional bosonic field theory in his approach .
+ to the entanglement entropy for @xmath74 . left : @xmath67 as function of @xmath3 for @xmath75 , from top to bottom .
right : fit of the data ( circles ) for @xmath68 according to eqn.([levine ] ) , see text.,title="fig:",width=257 ] to the entanglement entropy for @xmath74 . left : @xmath67 as function of @xmath3 for @xmath75 , from top to bottom .
right : fit of the data ( circles ) for @xmath68 according to eqn.([levine ] ) , see text.,title="fig:",width=257 ] summing up , we have shown by numerical calculations that the general effects of an impurity in a luttinger liquid also show up in the entanglement entropy .
its logarithmic behaviour remains for attraction but is suppressed for repulsion .
a more detailed analytical treatment is , however , still desirable .
+ _ acknowledgement : _
i.p . thanks the interdisciplinary center for theoretical studies , cas , beijing , for its hospitality .
he also acknowlegdes discussions with n. shannon and v. meden and correspondence with n. laflorencie .
holzhey , f. larsen and f. wilczek , nucl .
b * 424 * , 443 ( 1994 ) p. calabrese and j. cardy , journal of statistical mechanics jstat , p06002 ( 2004 ) j. cardy and i. peschel , nucl .
b * 300 * [ fs22 ] , 377 ( 1988 ) g. c. levine , phys . rev . lett . * 93 * , 266402 ( 2004 ) g. vidal , j. i. latorre , e. rico and a. kitaev , phys .
rev . lett . * 90 * , 227902 ( 2003 ) j. i. latorre , e. rico and g. vidal , quant .
inf . and comp . * 4 * , 48 ( 2004 ) , preprint : quant - ph/0304098 b .- q .
jin and v. e. korepin , j. stat . phys . * 116 * , 79 ( 2004 ) m. wolf , preprint quant - ph/0503219 d. gioev and i. klich , preprint : quant - ph/0504151 s. eggert and i. affleck , phys . rev .
b * 46 * , 10866 ( 1992 ) c. l. kane and m. p. fisher , phys . rev .
b * 46 * , 15233 ( 1992 ) s. qin , m. fabrizio and l. yu , phys . rev .
b * 54 * , 9643 ( 1996 ) s. qin , m. fabrizio , l. yu , m. oshikawa and i. affleck , phys .
b * 62 * , 9766 ( 1997 ) s. rommer and s. eggert , phys .
b * 56 * , 4370 ( 1997 ) v. meden , w. metzner , u. schollwck and k. schnhammer , phys .
b * 65 * , 045318 ( 2002 ) k. louis and c. gros , phys . rev .
b * 68 * , 184424 ( 2003 ) s. andergassen , t. enss , v. meden , w. metzner , u. schollwck and k. schnhammer , phys . rev .
b * 70 * , 075102 ( 2004 ) i. peschel , j. phys .
a : math . gen .
* 38 * , 4327 ( 2005 ) s.r .
white , phys .
b * 48 * , 10345 ( 1993 ) .
i. peschel , x. wang , m. kaulke and k. hallberg , _ density - matrix renormalization _ , + lecture notes in physics , vol .
* 528 * , ( springer - verlag , berlin 1999 ) .
i. peschel , j. phys .
a * 36 * , l205 ( 2003 ) n. laflorencie , preprint cond - mat/0504446 i. peschel and j. zhao , preprint quant - ph/0509002 | we study the entanglement properties of anisotropic open spin one - half heisenberg chains with a modified central bond .
the entanglement entropy between the two half - chains is calculated with the density - matrix renormalization method ( dmrg ) .
we find a logarithmic behaviour with an effective central charge @xmath0 varying with the length of the system .
it flows to one in the ferromagnetic region and to zero in the antiferromagnetic region of the model . in the xx case
it has a non - universal limit and we recover previous results .
the consideration of entanglement properties has brought a new element into the study of many - particle quantum states .
entanglement is related to a division of the system into two parts and can be quantified via the reduced density matrix @xmath1 and the entropy @xmath2 connected with it .
this quantity is particularly interesting for the ground state of critical one - dimensional systems .
for example , if one cuts an open chain into two halves of length @xmath3 , conformal invariance predicts the universal form @xcite @xmath4 where @xmath5 is the central charge in the conformal classification .
an analogous formula with @xmath6 replaced by @xmath7 holds for a segment of length @xmath3 in an infinite chain .
the logarithmic divergence is a particular signature of the criticality and can be related to analogous universal contributions to the free energy of critical two - dimensional systems with conical shape @xcite .
it has been verified numerically for a number of quantum chains @xcite and derived analytically for free fermions hopping on a chain @xcite . for this system
it can also be proven by putting proper bounds on @xmath8 @xcite .
since the entanglement entropy measures the mutual coupling of the two parts of a system in the wave function , a defect at their boundary should have a strong influence . in one dimension ,
defects show particularly interesting features in luttinger liquids , i.e. in strongly correlated fermionic systems .
their effective strength then depends on the sign of the interaction @xcite and goes to zero for attraction while it diverges for repulsion , as the system size increases .
this has been checked in various further studies , see e.g. @xcite .
one then wonders how the entanglement behaves in such a case and how the logarithmic law ( [ log ] ) is affected .
this question was first raised by levine @xcite who used bosonization and found results to lowest order in the impurity strength which are consistent with the general picture .
the simpler case of a free - fermion hopping model , corresponding to the xx spin chain , was treated numerically in ref .
@xcite . in this case
, the logarithmic behaviour was found to persist , but with a prefactor @xmath9 which depends continuously on the defect strength . in the present paper
we present a numerical study of this problem for the case of a planar xxz spin chain , which has @xmath10 and is the lattice version of a luttinger model .
we treat open chains with one modified bond in the middle and use density - matrix renormalization ( dmrg ) @xcite .
this method is ideally suited for such a study since the calculation of @xmath1 and its spectrum is an intrinsic part of the algorithm .
we find that the effective central charge , which will be called @xmath0 , tends asymptotically to one for @xmath11 ( ferromagnetic region , resp .
attraction ) and to zero for @xmath12 ( antiferromagnetic region , resp .
repulsion ) .
how fast this happens depends , however , on the defect strength , and for weak perturbations and interactions the asymptotic region is still far away , although we treated chains up to 800 sites .
we also compared to levine s formulae but could not obtain a quantitative agreement , although we verify his general picture . in the following
we first present the model and the method of calculation .
then we show some density - matrix spectra and their typical features .
after that we present the calculations of @xmath8 and the results for the quantity @xmath0 for a number of defect strengths @xmath13 and anisotropic parameters @xmath14 .
finally we compare with ref.@xcite and draw our conclusions . |
this supplemental material contains a proof that , under conditions specified in the manuscript and reiterated below , the liouvillians we consider possesses a finite dissipative gap . in particular
, the dissipative gap of the liouvillian @xmath19 is bounded below by that of the dissipator @xmath98 in the absence of a hamiltonian .
the first section briefly introduces relevant formalism , the second section discusses the structure of the liovillians considered in the manuscript in the language of this formalism , and the final section contains the proof .
an arbitrary density matrix of @xmath99 spin-@xmath3 s can be expanded as @xmath100 where the coefficients @xmath101 are real numbers and @xmath102 is the identity operator on site @xmath8 . for convenience
we also introduce a more compact notation in which the set of indices @xmath103 is replaced by a single collective index @xmath104 ( which runs over all possible assignments of values @xmath105 to all @xmath99 sites ) , the basis element @xmath106 is denoted simply by @xmath107 , and the density operator @xmath108 ( or any other operator ) is denoted with a modified ket notation as @xmath109 . in this notation , eq . becomes @xmath110 .
the @xmath111 must be real because the density matrix is hermitian , and the basis elements all are hermitian , but it is nevertheless useful to consider complex coefficients . in this notation
a general ( not - necessarily hermitian ) operator @xmath112 and its hermitian conjugate @xmath113 become , respectively , @xmath114 and @xmath115 .
the basis elements @xmath116 form a vector space ( liouville space ) endowed with the inner product @xmath117 , where we have used the fact that @xmath118 and @xmath119 .
we denote the action of a superoperator on an operator , @xmath120 , with the notation @xmath121 . the superoperator @xmath122 inherets a matrix representation from the equation @xmath121 by expanding @xmath123 and left multiplying both sides of the equation by @xmath124 , giving @xmath125 , where @xmath126 .
note that a physical liouvillian must be hermiticity preserving , which implies that its matrix elements in a basis of hermitian operators ( such as the operators @xmath107 defined above ) are always real . for our purposes below , an important property of each basis element
@xmath116 is the number of operators @xmath127 and @xmath128 that appear [ see eq . ] , which we denote by @xmath129 .
in literature on nuclear - magnetic resonance liouville space can be decomposed into a direct sum of subspaces with fixed @xmath129 , and we denote the basis elements of these subspaces by @xmath130 , with the understanding that the index @xmath131 enumerates only the basis elements in this subspace .
an arbitrary vector within this subspace is denoted by @xmath132 .
if we define @xmath133 as a superoperator that projects onto the subspace spanned by @xmath130 , then we can expand @xmath134 , which has a matrix representation @xmath135 .
the liouvillian superoperator can be decomposed as @xmath136 , where @xmath137 $ ] .
because @xmath98 and @xmath138 are each ( independently ) hermiticity preserving , they both have real - valued matrix representations in the chosen basis .
the dissipators we consider in the manuscript obey the condition @xmath139 ( eq . of the manuscript ) , which in the language used here implies that @xmath140 whenever @xmath141 .
in fact , because the dissipator is constructed of local terms , it is straightforward to see that @xmath142 vanishes unless either @xmath143 or @xmath144
. it is also readily verified that @xmath145 is block diagonal in the @xmath129-subspace decomposition , and that @xmath146 ( the latter condition following because @xmath1 commutes with all basis elements in the @xmath147 sector ) .
hence the complete liouvillian has the block - lower - triangular structure ( dropping subscripts ) @xmath148 { \mathcal{d}}^{10 } & { \mathcal{d}}^{11}+i{\mathcal{h}}^{11 } & 0 & \cdots & 0 & 0 \\[4pt ] 0 & { \mathcal{d}}^{21 } & { \mathcal{d}}^{22}+i{\mathcal{h}}^{22 } & \cdots & 0 & 0 \\[4pt ] \vdots & \vdots & \vdots & \ddots & \vdots & \vdots \\[4pt ] 0 & 0 & 0 & \cdots & { \mathcal{d}}^{n , n-1 } & { \mathcal{d}}^{nn}+i{\mathcal{h}}^{nn } \end{matrix}\right\rgroup . \ ] ] note that the block - lower - triangular structure of @xmath19 of ensures that the time evolution within the @xmath147 subspace , which completely determines the populations in the @xmath24-basis , is governed by a closed set of equations .
the dynamics within this sector is therefore equivalent to the dynamics of a classical master equation , a situation which sometimes arises in more general quantum systems whenever it is possible to eliminate the contribution of coherences from the equations of motion for populations @xcite .
the eigenvalues of a block - lower triangular matrix are given by the eigenvalues of the diagonal blocks , and therefore @xmath149 here it is to be understood that the eigenvalues of a projected operator are computed only with respect to vectors supported on that subspace , i.e. the eigenvalues of @xmath150 are given by the eigenvalues of the matrix @xmath151 .
the dissipators we consider are constructed as sums over single - site dissipators , @xmath152 , each of which independently obeys the constraints ( equivalent to eq . in the manuscript ) @xmath153={\rm
tr}[\hat{\mathds{1}}_j\mathcal{d}_j(\hat{\sigma}_j^y)]={\rm tr}[\hat{\sigma}^z_j\mathcal{d}_j(\hat{\sigma}_j^x)]={\rm tr}[\hat{\sigma}^z_j\mathcal{d}_j(\hat{\sigma}_j^y)]=0.\end{aligned}\ ] ] here it is understood that @xmath154 is an operator in the full liouville space , but because it only acts nontrivially on site @xmath8 we will ( in a slight abuse of notation ) temporarily use the symbol @xmath154 to represent an operator acting on just the liouville space associated with site @xmath8 , spanned by vectors @xmath155 ; regardless of the interpretation of which space @xmath154 acts on , its eigenvalues are the same up to degeneracies . with this interpretation in mind
, we can represent @xmath154 as a @xmath156 real matrix matrix @xmath157 .
using the shorthand @xmath158 ( note the temporary suppression of the index @xmath8 ) , ordering our basis as @xmath159 , and taking into account the constraints on @xmath154 imposed in eq . , we have @xmath160 0 & 0 & 0 & 0 \\ \mathcal{d}_{z1 } & \mathcal{d}_{zz } & 0 & 0 \\ \hline \mathcal{d}_{x1 } & \mathcal{d}_{xz } & \mathcal{d}_{xx } & \mathcal{d}_{xy } \\
\mathcal{d}_{y1 } & \mathcal{d}_{yz } & \mathcal{d}_{yx } & \mathcal{d}_{yy } \end{matrix}\right\rgroup.\end{aligned}\ ] ] in order to prove the existence of a dissipative gap for @xmath19 , we also make the following two assumptions : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ \(1 ) each local dissipator @xmath154 has a unique steady state and a dissipative gap @xmath161 , and @xmath162 .
since all eigenvalues of the dissipator must have non - positive real parts , this means that for any nonzero eigenvalue @xmath163 of @xmath154 , we must have @xmath164 .
\(2 ) in addition to eq . , the local dissipators obey the further constraints : @xmath165={\rm
tr}[\hat{\sigma}_j^x\mathcal{d}_j(\hat{\sigma}_j^y)],~~~~{\rm and}~~~~{\rm tr}[\hat{\sigma}_j^z\mathcal{d}_j(\hat{\mathds{1}}_j)]=0.\end{aligned}\ ] ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in the notation just introduced , we can restate these conditions as : ( 1 ) the matrix @xmath166 has a unique zero eigenvalue and a dissipative gap @xmath161 , and ( 2 ) the diagonal blocks of @xmath166 are symmetric ( @xmath167 and @xmath168 ) . note that the hamiltonian in eq
. of the manuscript , when rotated about the @xmath95 axis and at @xmath82 , becomes of the form in eq .
of the manuscript , while the jump operators become @xmath169 .
it is straightforward to verify that conditions ( 1 ) and ( 2 ) are satisfied for the corresponding dissipator , with @xmath170 .
as mentioned above , a block - lower - triangular matrix of this form shares its eigenvalues with the diagonal blocks .
let s denote the set of eigenvalues of the upper - left block by @xmath171 , and the set of eigenvalues of the lower - right block by @xmath172 ( it is straightforward to see that the upper - left block must have a zero eigenvalue ) . by conditions ( 1 ) and ( 2 ) , we know that the three non - zero eigenvalues are real , and that @xmath173 .
because each @xmath154 is supported on a single site , the eigenvalues of the full dissipator @xmath174 are simply sums of eigenvalues from each @xmath154 .
eigenvalues of the projected dissipator @xmath175 can be constructed as follows : defining @xmath176 to be a set containing @xmath129 sites , and also defining the notation @xmath177 , we have @xmath178 we can draw two immediate conclusions : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ \(a ) @xmath150 has precisely one zero eigenvalue , obtained when the zero eigenvalue in @xmath179 is chosen for all @xmath8 , and any other eigenvalue @xmath180 satisfies @xmath181 .
\(b ) any eigenvalue @xmath180 of @xmath175 ( with @xmath182 ) is real and satisfies @xmath183 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ now consider an arbitrary normalized eigenvector of @xmath184 in the @xmath185 subspace , @xmath186 , with eigenvalue @xmath187 , and write @xmath188 it is straightforward to show that the superoperator @xmath189 is hermitian ( for any hamiltonian ) , and thus @xmath190 . in general , there is not a simple relation between the real part of the expectation value of an operator and the real parts of its eigenvalues .
however , assumption ( 2 ) guarantees that @xmath175 is a real symmetric matrix .
therefore , we have @xmath191 where the inequality follows because the expectation value of a _ real symmetric matrix _ ( more generally a hermitian matrix ) is bounded between its smallest and largest eigenvalues .
returning to eq . , keeping in mind conclusion ( a ) , and using eq . , we find that @xmath19 has a single zero eigenvalue , and a dissipative gap of at least @xmath192 . | exactly solvable models have played an important role in establishing the sophisticated modern understanding of equilibrium many - body physics . and
conversely , the relative scarcity of solutions for _ non - equilibrium _ models greatly limits our understanding of systems away from thermal equilibrium .
we study a family of non - equilibrium models , some of which can be viewed as dissipative analogues of the transverse - field ising model , in that an effectively classical hamiltonian is frustrated by dissipative processes that drive the system toward states that do not commute with the hamiltonian .
surprisingly , a broad and experimentally relevant subset of these models can be solved efficiently in any number of spatial dimensions .
we leverage these solutions to prove a no - go theorem on steady - state phase transitions in a many - body model that can be realized naturally with rydberg atoms or trapped ions , and to compute the effects of decoherence on a canonical trapped - ion - based quantum computation architecture .
the understanding of equilibrium many - body physics in both classical and quantum systems has relied heavily on exact solutions of highly simplified models .
for example , solutions of the classical and quantum ( transverse - field ) ising models have elucidated the structure of classical and quantum phase transitions , respectively , and the deep connections between them @xcite . for quantum systems that are
_ not _ in thermal equilibrium such solutions are comparatively scarce , though important progress has been made in numerous specialized models that either : ( a ) are one dimensional and isolated @xcite , ( b ) impose nontrivial dissipation on otherwise free bosons or fermions @xcite , ( c ) impose highly fine - tuned dissipation @xcite , or ( d ) are coupled to an environment only at a boundary @xcite . on the other hand , interacting quantum spin systems that are dissipative in the bulk and driven so that they do not thermalize are expected to exhibit a variety of rich behaviors not found in their equilibrium counterparts , including unusual multi - critical points @xcite , critical exponents not found in or near equilibrium @xcite , and the existence of zero - entropy entangled steady states @xcite . and , naturally , such systems play an important role in the theory of quantum computation in the presence of decoherence . while recent experimental advances have enabled the controlled study of such physics in systems ranging from exciton - polariton fluids @xcite , to trapped ions @xcite , to rydberg gases @xcite , the minimal microscopic models expected to capture the essential qualitative physics
many - body quantum master equations @xcite continue to pose severe challenges to existing theoretical techniques . is frustrated , even at @xmath0 , because the ground state of the transverse field ( expressed as a density matrix ) does not commute with @xmath1 .
the models solved here generalize this scenario to the case where order is frustrated by the inclusion of a _ dissipative process _
whose _ steady state _ does not commute with @xmath1 . in the example shown ( right panel ) , markovian dissipation implemented by jump operators @xmath2 drives each spin towards the ground - state of a transverse field . ] in this manuscript we investigate a broad and experimentally relevant class of driven - dissipative many - body spin models , some of which can be viewed as dissipative analogues of the transverse - field ising model ( tfim ) , and show that they can be solved efficiently .
the tfim captures a characteristic feature of low - temperature quantum systems more generally : even at zero temperature , the ordering associated with a classical hamiltonian ( the ising model ) can be frustrated by the persistence of quantum fluctuations ( induced by the transverse field , left panel of fig.[fig : schematic ] ) . in particular
, the transverse field favors a zero - temperature density matrix that does not commute with the ising hamiltonian . as a result , energy minimization forces the system to develop quantum correlations even between arbitrarily distant spins , generically prohibiting exact calculations of its ground - state properties ( except in certain very well - known circumstances , e.g. nearest - neighbor interactions in 1d ) .
the models we consider realize a natural non - equilibrium analogue of this scenario , in which a classical hamiltonian is frustrated by the presence of _ dissipative _ fluctuations ( right panel , fig.[fig : schematic ] ) . strikingly , we find that the inclusion of a broad class of dissipative processes favoring a _ steady - state _ density matrix that does not commute with the hamiltonian is relatively benign , even in more than one spatial dimension .
in particular , correlations remain dynamically localized whenever the hamiltonian is of finite range , enabling the time - dependence of finite - order correlation functions to be efficiently obtained ( even for an infinite system ) by solving a finite - dimensional system of equations .
this structure exists , for example , even when the dissipation alone drives the system towards the ground state of a transverse field .
it also exists when the dissipation is explicitly derived from a temporally fluctuating transverse field ; evidently , even though the ising model is not ( in general ) expected to be solvable in the presence of a _
particular _
( static or time - varying ) transverse field , the dynamics of the transverse - field ising model averaged over _ all _ such fields can be computed exactly . to illustrate the utility and generality of these solutions , we use them to prove that an experimentally relevant subset of the models considered possess a finite dissipative gap , and therefore can not undergo dissipative phase transitions .
we then apply this result to a model of interacting rydberg atoms studied in refs .
@xcite , thereby confirming the structure of the phase diagram inferred from recently developed approximate techniques @xcite . from a more applied perspective
, we emphasize that ising - like hamiltonians , despite being `` classical '' , induce truly quantum dynamics and play a central role in the production of states sought for quantum information processing and quantum metrology @xcite .
their solvability in the presence of dissipation thus affords numerous exciting opportunities to investigate the effect of decoherence on the generation of useful entangled states . as an illustrative example
, we compute the effects of a fluctuating transverse field ( dephasing ) on ising dynamics generated by mlmer - srensen gates @xcite , which can impose important technical limitations in trapped - ion based approaches to quantum computation @xcite .
_ _ dissipative models.__the models we consider can be constructed in close analogy to the transverse - field ising model . there
, one starts with a many - body hamiltonian that is `` classical '' in the sense that it can be diagonalized by a choice of local basis .
thus its eigenstates have well defined local properties , and in this sense there are no quantum fluctuations @xcite . to be concrete
we restrict our focus to the spin-@xmath3 ising model , @xmath4 however , the scope of our results is significantly broader , and includes generic hamiltonians that can be diagonalized by a choice of local basis or , with some additional constraints , commuting hamiltonians that are not diagonalized by a choice of local basis ( e.g. the toric code @xcite ) .
extensions along these lines will be discussed at the end of the manuscript . in equilibrium ,
a natural question to ask is how one can modify @xmath1 such that , in the low temperature limit where thermal fluctuations vanish , quantum fluctuations remain .
one simple strategy is to add single - body terms to the hamiltonian that are not diagonalized by the eigenstates of @xmath1 ; the usual culprit is a transverse field , resulting in an additional term @xmath5 . even at zero temperature ,
i.e. in the quantum ground state , such a term induces fluctuations between the classical eigenstates of the interaction part of the hamiltonian . in a driven - dissipative setting , we are no longer interested in properties of the ground state , which control the low - temperature equilibrium physics , but rather the properties of the steady state , which control the long - time , non - equilibrium physics .
a natural dissipative generalization of the procedure used above to frustrate the zero - temperature ordering associated with @xmath1 is to introduce single - body dissipative processes that drive the system toward a steady state that does not commute with @xmath1 . assuming that dissipation can be treated in the born - markov approximation , the dynamics of an open quantum system with hamiltonian @xmath1 is governed by a markovian master equation of the form @xmath6+\mathcal{d(\hat{\rho})},\nonumber\\ \label{eq : dissipator } \mathcal{d}(\star)&\equiv\sum_{j,\alpha}\frac{\gamma_{j\alpha}}{2}\big(2\hat{j}^{\phantom\dagger}_{j\alpha}\star \hat{j}^{\dagger}_{j\alpha}-\{\hat{j}^{\dagger}_{j\alpha}\hat{j}^{\phantom\dagger}_{j\alpha},\star\}\big).\end{aligned}\ ] ] the dissipation is induced by jump operators @xmath7 , each of which we assume to be supported on a single site @xmath8 .
the index @xmath9 may take on multiple values in order to describe multiple dissipative channels on a given site , though in what follows we will generally consider only one jump operator on each site , dropping the index @xmath9 . consider the dissipative dynamics in the absence of the hamiltonian , described by @xmath10 .
the steady - state solution of the purely dissipative dynamics is determined implicitly by solving @xmath11 . if @xmath12 commutes with the hamiltonian , then it is automatically also a solution of @xmath13 and thus is a proper steady state of the complete dynamics including both coherent evolution and dissipation .
if it does not , then in close analogy to the ground state of the transverse - field ising model , we expect the steady state to possess fluctuations in the sense of admixing of classical states . surprisingly , a large class of dissipators that frustrate the classical hamiltonian , i.e. for which @xmath14\neq 0 $ ] , nevertheless admit exact solutions for the dynamics of observables .
in particular , suppose that @xcite @xmath15 for @xmath1 of the form in eq .
, we also have @xmath16)=0 $ ] , and therefore whenever the jump operators are chosen to satisfy eq .
, the complete liouvillian also obeys @xmath17 taken together with the equality @xmath18 , which follows trivially from the definition of @xmath19 ( and ensures the conservation of probability ) regardless of the form of the jump operators , eq .
can be understood colloquially as a statement that @xmath19 does not map coherences onto populations ( fig.[fig : lattice]a ) . in what follows
, we will show that finite - range hamiltonians of the form in eq .
, subjected to dissipation obeying eq . , produce correlations that are localized in space and can therefore be solved efficiently in the thermodynamic limit .
basis , @xmath19 does not map coherences to populations .
equivalently , in the heisenberg picture , @xmath20 does not cause diagonal operators to develop coherences .
( b ) when the conditions described in ( a ) are met , the dynamics of any observable supported on a subsystem @xmath21 can be computed by identifying the nearest neighbors of the set @xmath21 , denoted @xmath22 , tracing out the part of the density matrix supported on @xmath23 , and then solving the master equation projected into the remaining finite - dimensional system . ]
there are many natural jump operators satisfying eq .. for example , dephasing , spontaneous emission , and incoherent pumping along the @xmath24 axis all do , and the models studied here therefore subsume finite - range versions of the models studied in @xcite ( and realized experimentally in @xcite ) as special cases
. however , all the examples of dissipation just mentioned lead to steady states @xmath12 that commute with @xmath1 , and thus the steady states in the presence of @xmath25 remain trivial .
examples of jump operators satisfying eq . but producing steady states that _ do not _ commute with @xmath25 include dephasing along any direction in the @xmath26 plane ; @xmath27 , or spontaneous emission along any axis in the @xmath26 plane ; @xmath28 .
note that in the latter example , @xmath12 is the ground state of a transverse field .
_ localization of correlations_.the time - dependent expectation value of an arbitrary operator @xmath29 , initially supported on a set of sites @xmath21 , can be written as @xmath30 the heisenberg picture expression utilizes the adjoint liouvillian @xmath31+\mathcal{d}^{\ddagger}(\hat o)$ ] , with the adjoint dissipator given by @xmath32 .
the operator @xmath33 could be , for example , a product of two spin operators on different sites , in which case @xmath34 is an equal - time correlation function ( generalizations to unequal time correlation functions are also possible using the quantum regression formula ) .
equation can be recast in terms of the heisenberg - picture liouvillian as @xmath35=0 $ ] ; thus , @xmath36 does not map populations onto coherences ( fig.[fig : lattice]a ) .
this condition on @xmath36 can be restated in the following useful way : referring to an operator as `` diagonal on the set of sites @xmath37 '' if it commutes with all operators @xmath38 for @xmath39 , one can show that the set of operators that are diagonal on @xmath37 is closed under the action of @xmath36 . with this structure in mind
, we refer to @xmath36 as _ diagonality preserving _ ( in the @xmath24 basis ) , a property which will play a key role in the solution of eq .. to compute @xmath34 , it is useful to define the set of sites that are nearest neighbors of @xmath21 , denoted @xmath22 , as all sites outside of @xmath21 for which a term in the hamiltonian has simultaneous support on @xmath21 and @xmath22 .
the complement of @xmath40 , denoted @xmath23 , is therefore not directly connected to @xmath21 by terms in the hamiltonian ( see fig.[fig : lattice]b ) .
we then decompose the hamiltonian as @xmath41 , where @xmath42 contains all terms in @xmath1 that have support on @xmath21 , and @xmath43 contains all terms that do not ( note that , by the definition of @xmath22 , @xmath42 is supported on @xmath40 , while @xmath43 is supported on @xmath44 ) .
similarly , we decompose the heisenberg - picture dissipator as @xmath45 , where @xmath46 only contains jump operators supported on @xmath40 and @xmath47 only contains jump operators supported on @xmath23 .
writing @xmath48 , with @xmath49+\mathcal{d}_{\mathscr{a\!b}}^{\ddagger}(\star)$ ] and @xmath50+\mathcal{d}^{\ddagger}_{v}(\star)$ ] , we can write @xmath34 as @xmath51 where @xmath52 . the structure of eq .
can be greatly simplified by induction .
in particular , suppose that @xmath53 satisfies the following two conditions : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ \(a ) @xmath53 is supported on @xmath40 .
( b ) @xmath53 is diagonal on @xmath22 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ note that both conditions are satisfied trivially for @xmath54 , since we have assumed that @xmath33 is fully supported on @xmath21 .
we will show that the operator @xmath55 also satisfies these conditions , thereby proving that all @xmath56 do . toward this end
, we first show that @xmath57+\mathcal{d}^{\ddagger}_v(\hat{o}_n)$ ] vanishes by the assumptions .
the first term vanishes because @xmath43 is diagonal and supported on @xmath44 , while @xmath53 is diagonal on @xmath58 by conditions ( a , b ) , giving @xmath59=0 $ ] .
the second term vanishes because @xmath60 only contains jump operators supported on @xmath23 , implying ( together with condition a ) that @xmath61 .
thus @xmath62 as claimed , giving @xmath63 .
now we note that , because @xmath42 and the jump operators in @xmath46 are supported on @xmath40 , @xmath64 must also be supported on @xmath40 , and @xmath65 therefore satisfies condition ( a ) . furthermore , because @xmath66 is diagonality preserving , @xmath65 continues to satisfy condition ( b ) .
therefore , by induction , we deduce that @xmath67 , and is supported on @xmath40 . plugging this result into eq . and using the isolated support of @xmath56 to carry out the trace over @xmath23 , we obtain @xmath68 where @xmath69 is the reduced density matrix of the initial state upon tracing out all sites in @xmath23 . returning to the schrdinger picture
we find our primary result , @xmath70 equation can be read as follows : the time - dependent expectation value of @xmath33 can be computed by first calculating the reduced density matrix on @xmath40 , and then evolving it using a hamiltonian and jump operators supported only on @xmath40 .
it follows immediately that correlations are localized for finite - ranged hamiltonians . for example , consider a correlation function @xmath71 ( @xmath72 ) , and suppose the system starts in a product state .
if the hamiltonian is of finite range @xmath73 , and @xmath8 and @xmath74 are separated by a distance @xmath75 ( so that sites @xmath8 and @xmath74 do not share any neighbors , and @xmath40 decomposes into two disjoint sets ) , then it follows from eq . that @xmath76 at all times @xcite . _ applications_.just as many - body ground states can suddenly change in character as a parameter in the hamiltonian is continuously adjusted , signaling a quantum phase transition , the steady state of a master equation can exhibit a sudden change when a parameter in the liouvillian is adjusted continuously , signaling a dissipative phase transition .
while a quantum phase transition is associated with the closing of an energy gap in the spectrum of the hamiltonian , a dissipative phase transition is associated with the closing of a _ dissipative gap _ in the spectrum of the liouvillian .
consider the dissipative transverse - field ising model studied in refs.@xcite , @xmath77 in which fluctuations due to both dissipation ( strength @xmath78 ) _ and _ a transverse field ( strength @xmath79 ) are considered ; note that the state which minimizes the energy with respect to the transverse field ( for @xmath80 ) is identical to the steady state of the dissipation alone . in eq .
we have chosen to make the hamiltonian diagonal along the @xmath81 direction for consistency with ref .
@xcite .
however , for @xmath82 and upon making a rotation of the coordinate system that takes the ising part of @xmath83 into the form of eq . , it can easily be verified that this model satisfies eq .. in ref.@xcite , a careful analysis of the variational techniques developed in @xcite suggests that the system is disordered for any value of @xmath78 at @xmath82 and in high dimensions . here
, we can make this conclusion rigorous in _ any _ dimension : in the supplement , we prove that this model at @xmath82 [ and also a more general subclass of the models obeying eq . ]
has a finite dissipative gap @xmath84 .
since the model is disordered at @xmath85 and a gap exists at @xmath82 for any @xmath86 , it must be disordered along the entire @xmath82 axis . in light of the liouvillian stability results proved in ref.@xcite
, the existence of a dissipative gap at @xmath82 may actually imply something much stronger : the gap persists for small @xmath80 , ruling out a phase transition for @xmath87 and implying that the system is disordered for sufficiently small @xmath79 at any finite @xmath78 .
( 10 spins in 1d with periodic boundary conditions , and @xmath88 ) . ]
the solutions developed here can also be used to calculate dynamics of the fluctuating - transverse - field ising model , @xmath89 here , the transverse fields @xmath90 are gaussian random variables with white - noise spectrum @xmath91 .
this model interpreted such that observables are obtained by averaging dynamics induced by @xmath92 over realizations of the fluctuating transverse field can be viewed as a continuous unraveling of the master equation describing dephasing dynamics in the ising model @xcite , @xmath93+\gamma\sum_{j}(\hat\sigma^z_j\,\hat\rho\,\hat\sigma_j^z-\hat\rho).\end{aligned}\ ] ] [ see also refs .
@xcite for a constructive approach to eq . starting from eq . ] .
a master equation of this form arises naturally in ion trap experiments , where it captures the effects of qubit dephasing during ising dynamics induced by mlmer - srensen gates @xcite , and hence it plays a prominent role in the description of decoherence effects on trapped - ion based approaches to quantum computation .
naively one would not expect there to be an efficient solution to eq .
, given that its unraveling [ either continuous or discrete , with the continuous case leading back to eq . ]
requires one to solve the ising model in the presence of a time - dependent transverse field . nevertheless , after a rotation by @xmath94 about the @xmath95 axis eq .
falls into the class of solvable models studied here , and observables can be computed efficiently . in fig.[fig : applications](a ) , we show simulations of the stochastic schrdinger equation associated with @xmath96 ; while these simulations are computationally very expensive , and not feasible for more than @xmath97 spins , the average over sufficiently many trajectories ( lighter to darker red lines indicating more averaging ) can be seen to converge to the exact solution ( black - dashed line ) . _
outlook_. we have identified an exactly solvable class of driven - dissipative quantum spin models in which dissipation frustrates the ordering of a classical hamiltonian .
the underlying algebraic structure that enables an exact solution , ensured by eq .
, can also be identified in a much more general class of hamiltonians that are solvable in the absence of dissipation .
for example , the results can easily be generalized to systems with arbitrary finite - dimensional hilbert spaces on each site and time - dependent hamiltonians .
long - range interactions do spoil some aspects of the solution , but the absence of dissipative phase transitions can still be proven .
hamiltonians that are not diagonalized by a local choice of basis but can still be written as a sum of local commuting terms , such as the toric code @xcite , are also amenable to similar solution techniques , and will be the topic of future research .
we acknowledge hendrik weimer and vincent overbeck for helpful discussions .
m.f.m .
, j.t.y . , and a.v.g .
acknowledge support by arl cdqi , aro muri , nsf qis , aro , nsf pfc at jqi , and afosr .
mm acknowledges start - up funding from michigan state university .
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 link:\doibase 10.1103/revmodphys.69.315 [ * * , ( ) ] @noop _ _ ( , ) http://stacks.iop.org/1742-5468/2005/i=04/a=p04010 [ * * , ( ) ] link:\doibase 10.1103/physreva.78.010306 [ * * , ( ) ] http://stacks.iop.org/1367-2630/10/i=4/a=043026 [ * * , ( ) ] link:\doibase 10.1103/physreva.87.012108 [ * * , ( ) ]
\doibase http://dx.doi.org/10.1016/j.aop.2016.05.019 [ * * , ( ) ] http://iopscience.iop.org/article/10.1088/1742-5468/2010/05/l05002/meta [ * * , ( ) ] link:\doibase 10.1103/physreve.83.011108 [ * * , ( ) ] http://iopscience.iop.org/article/10.1088/1742-5468/2011/06/p06007 [ * * , ( ) ] http://iopscience.iop.org/article/10.1088/1367-2630/16/1/013042 [ * * , ( ) ] https://arxiv.org/abs/1609.07249 [ ( ) ] https://arxiv.org/abs/1609.07838 [ ( ) ] link:\doibase 10.1103/physrevlett.101.105701 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.217206 [ * * , ( ) ]
link:\doibase 10.1103/physrevlett.107.137201 [ * * , ( ) ] http://stacks.iop.org/1751-8121/48/i=37/a=373001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.257204 [ * * , ( ) ] http://stacks.iop.org/0034-4885/79/i=9/a=096001 [ * * , ( ) ] link:\doibase 10.1103/physreva.78.042307 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys1073 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys1342 [ * * , ( ) ] http://stacks.iop.org/1367-2630/16/i=1/a=013023 [ * * , ( ) ] link:\doibase 10.1126/science.1074464 [ * * , ( ) ] http://dx.doi.org/10.1038/nature05131 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.230602 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys3143 [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms11887 [ * * , ( ) ] link:\doibase 10.1126/science.aad9958 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys2630 [ * * , ( ) ] http://dx.doi.org/10.1038/nature11361 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.113901 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.113.023006 [ * * , ( ) ] link:\doibase 10.1103/physrevb.93.014307 [ * * , ( ) ] https://arxiv.org/abs/arxiv:1606.08863 [ ( ) ] link:\doibase 10.1103/physrevlett.114.040402 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.86.5188 [ * * , ( ) ] link:\doibase 10.1103/physreva.47.5138 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.82.1835 [ * * , ( ) ] http://stacks.iop.org/1464-4266/7/i=10/a=025 [ * * , ( ) ] http://dx.doi.org/10.1038/nature09071 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.117.060505 [ * * , ( ) ] @noop \doibase http://dx.doi.org/10.1016/s0003-4916(02)00018-0 " [ * * , ( ) ] @noop link:\doibase 10.1103/physreva.87.042101 [ * * , ( ) ] http://stacks.iop.org/1367-2630/15/i=11/a=113008 [ * * , ( ) ] @noop link:\doibase 10.1007/s00220 - 015 - 2355 - 3 [ * * , ( ) ] http://stacks.iop.org/0305-4470/26/i=9/a=019 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/physreva.64.052110 [ * * , ( ) ] https://arxiv.org/abs/1608.01317 [ ( ) ] link:\doibase 10.1103/physrevlett.116.160401 [ * * , ( ) ] link:\doibase 10.1103/physreva.90.021603 [ * * , ( ) ] |
a significant increase in the @xmath0c to @xmath6c isotope ratio was detected in japanese trees in ad 774/5 and a subsequent decrease for @xmath7 yr ( * ? ? ?
* henceforth m1 ) .
it is consistent with an increase in @xmath0c in american and european trees with @xmath8 to @xmath9 yr time resolution @xcite . if deposited within one year or less - best consistent with an atmospheric deposition model - the increase is 10 times larger than the average production due to galactic cosmic rays and 20 times larger than expected from the @xmath10 yr solar cycle ( m12 ) .
this requires a @xmath1-ray energy input of @xmath2 erg at earth ( m12 ) .
also , a @xmath11 increase in @xmath3be around ad 775 was observed in antarctica , but with lower time resolution @xcite .
solar or stellar flare were found to be unlikely because of the insufficient energetics and spectrum of such flares ( m12 ) . a normal supernova ( sn )
was also found to be unlikely from the lack of any historical sighting or a sn remnant ( snr ) ( m12 ) .
a strongly absorbed sn was not considered quantitatively , yet : absorption in the line - of - sight would not affect @xmath1-rays , but would decrease the observable optical flux of a sn . of the total energy output of a sn , e(event ) = @xmath12 erg , a fraction @xmath13 goes into @xmath1-rays @xcite .
the ratio between the @xmath1-ray energy emitted by a sn event spread homogeneously into the total area of a spherical shell around the sn ( @xmath14 with distance @xmath15 from the event to earth ) and the @xmath1-ray energy e(obs ) observed at earth is equal to the ratio between the surface area of that sphere and the earth solid angle @xmath16 ( with earth radius @xmath17 ) : @xmath18 therefore , a normal sn ( with @xmath13 ) , of which a @xmath1-ray flux of e(obs ) = @xmath2 erg was observed at earth , would have a distance @xmath19 pc , independent of absorption .
if the ad 774/5 event were one of the rare ( @xmath20 ) over - luminous sne , up to four times brighter than normal sne @xcite , then the expected distance is @xmath21 pc . from the peak absolute magnitude @xmath22 @xcite
, we can estimate the unabsorbed apparent peak magnitude : + m=@xmath23 mag , 124 pc , m=@xmath24 mag , sn ia , + m=@xmath25 mag , 124 pc , m=@xmath26 mag , sn ii , + m=@xmath27 mag , 260 pc , m=@xmath28 mag , sn ibc .
+ one would need absorption of at least @xmath29 mag to disable a historical sighting by naked eye ( limit @xmath30 mag for discovery of a new object , * ? ? ? * ) .
such a strong absorption within @xmath31 or @xmath32 pc is not possible , except in small areas towards dark clouds @xcite : the closest dark clouds with @xmath33 mag are lynds 183 at @xmath34 pc with up to a@xmath35 mag @xcite and @xmath36 oph at @xmath37 pc with up to a@xmath38 mag @xcite .
absorption of @xmath33 mag is limited to @xmath39 deg@xmath40 on the sky ( * ? ? ? * and k. dobashi priv .
comm . ) and less for distances within @xmath41 to @xmath32 pc @xcite .
the probability of an event within @xmath39 deg@xmath40 of the whole sky is @xmath42 .
even then , a large , young , and bright snr would be detectable by x - ray pointings , but can be excluded ( * ? ? ?
* chandra snr cataloggal.html ] )
. given the measurement precision achieved in @xmath0c for the @xmath43 peak in ad 774/5 ( m12 ) , potential @xmath0c from sne can be detected up to @xmath44 pc with @xmath45 .
indeed , there are no sne , pulsars , nor snr known within a few hundred pc with age of some 300 to 2000 yr ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* footnote 1 ; mcgill sgr / axp catalog ) there are eleven events with evidence ( historic observation , detected snr and/or known pulsar ) for a sn within 2000 yr and 5 kpc ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* footnote 1 & 2 ) and for all of them , a snr is detected @xcite , and at least eight were observed historically @xcite . while a missing historic observation is possible , a missing snr is very unlikely .
magnetar flares ( soft gamma - ray repeaters or anomalous x - ray pulsars ) were not yet considered : the largest flare observed was the x- and @xmath1-ray flare of sgr 1806 - 20 on 2004 dec 27 with peak energy ( @xmath46 erg / s at 15 kpc @xcite , or e(event ) = @xmath47 erg at @xmath48 kpc @xcite .
if the ad 774/5 event were such a flare , it would have taken place at @xmath49 pc ( equ . 1 with @xmath50 ) , but there is no neutron star known within such a small distance ( * ? ? ?
* footnote 2 ) even if a magnetar with @xmath51 g dipole field could produce an event with @xmath52 erg @xcite , the distance of such a neutron star to produce the ad 774/5 event would have to be @xmath53 pc .
a magnetar at that small distance would have been detected by the rosat all - sky x - ray survey : for a persistent bolometric luminosity ( mostly x - rays ) of @xmath54 erg / s with typical observed spectral components of magnetars ( blackbody with peak energy @xmath55 kev and power - law index @xmath56 , footnote 2 ) , we expect 150 to 800000 cts / s in the rosat energy band 0.1 - 2.4 kev at 10 to even 100 pc , i.e. easily detectable .
hence , we can exclude magnetar flares for the ad 774/5 event .
given that events on earth as well as solar and stellar flares ( m12 ) including neutron star flares ( see above ) as well as unabsorbed ( m12 ) and absorbed supernovae are very unlikely to be the cause for the ad 774/5 cosmic - ray event ( see above ) , we will now consider a gamma - ray burst ( grb ) .
the observed duration and spectral hardness of grbs allows to divide them into long ( @xmath57s ) and short ( @xmath58s ) grbs , the latter are harder regarding the spectrum ( power - law with exponential cutoff ) and are not related to sne nor snrs @xcite . while long grbs are caused by the core collapse of a very massive star , short grbs are explained by the merger of two compact objects @xcite .
a merger of two previously orbiting compact objects is the coalescence of a neutron star with either a black hole becoming a more massive black hole , or with another neutron star becoming either a relatively massive stable neutron star or otherwise a black hole , if the total mass exceeds the upper mass limit of neutron stars , somewhere between 2 and 3 m@xmath59 .
e.g. , the merger of two magnetized neutron stars can produce a spinning black hole launching a relativistic jet as observed in short grbs @xcite , if the earth is located in the jet .
let us now consider a short grb .
a short grb emits an isotropic equivalent energy of e(event ) = @xmath60 to @xmath61 erg in the observed energy range 10 kev to 30 gev @xcite , most or all in @xmath1-rays ( @xmath62 to 1 ) .
we estimate the distance towards a short grb from equ . 1 to @xmath63 to @xmath64 kpc , i.e. within our galaxy .
hence , the energetics of the @xmath0c peak on earth are consistent with a short grb .
effects of nearby long grbs on the earth biosphere due to the direct hit ( @xmath65 erg / s for 10s ) on one half - sphere were found to be lethal within @xmath66 kpc @xcite .
this can be scaled to a short grb with @xmath60 to @xmath61 erg .
hence , for a short grb within @xmath4 kpc , strong extinction effects are expected . because no extinction event was observed on earth for ad 774/5 , the short grb was more distant , probably @xmath4 to @xmath5 kpc .
a transient event is expected in the optical ( macronova ) from compact mergers with @xmath67 mag at peak @xcite .
this corresponds to @xmath68 ( 0.1 kpc ) or @xmath69 mag ( 4 kpc ) for negligible absorption .
hence , it may have been observable by naked eye , but only for up to one day , i.e. much shorter than a typical sn .
if reports about such a sighting remain missing , it can be due to the short time - scale , strong absorption , bad weather , and/or sky location near the sun and/or above unpopulated areas such as the pacific during the short visibility period .
a missing historical observation and a missing snr are fully consistent with a short grb .
since the peak of @xmath0c observed in ad 774/5 is consistent with a sharp increase within 0.1 to 1 yr ( m12 ) , a short grb typically lasting less than 2s @xcite and being undispersed in interstellar space is consistent with the observations regarding the short time - scale .
c and @xmath3be .
we plot the energy dependence of the cross - sections @xmath70 of the reactions @xmath0n(n , p)@xmath0c and @xmath0n(n , p+@xmath71)@xmath3be as full and dotted lines , respectively , cross - section in milli - barn ( mb , 1 barn is @xmath72 m@xmath40 ) versus energy in mev.,scaledwidth=35.0% ] ( with frequency @xmath73 ) versus energy in mev sampling the whole range of parameters observed , i.e. several typical spectra as blue and red lines for short and long grbs , respectively .
we plot the flux as log of @xmath74 in mev@xmath75 photons @xmath76 @xmath77 s@xmath78 mev@xmath79 ( plus arbitrary scaling due to normalization ) . at the observable low @xmath1-ray energies
, we see that short grbs are harder than long grbs , hence the devision and naming .
we use equ . 2 & 3 and the band function @xcite to estimate the production rate of @xmath0c to @xmath3be from the input spectra for short and long grbs .
because long grbs have a flat spectrum ( band function ) , they can not reproduce the observed production rate of @xmath0c to @xmath3be ( @xmath80 ) , while short grbs can reproduce the observed ratio , see fig .
[ fg3].,scaledwidth=35.0% ] given the cross - sections of the relevant reactions producing @xmath0c and @xmath3be ( fig .
[ fg1 ] ) and the full range of observed spectral parameters of short @xcite and long ( band function , * ? ? ?
* ) grbs ( fig .
[ fg2 ] ) , we computed the outcome ( fig . [ fg3 ] ) .
we assume that the peaks in @xmath0c ( @xmath81 atoms / cm@xmath40/s in @xmath83 yr , m12 ) , and @xmath3be ( @xmath84 increase with 10 yr time resolution , * ? ? ? * ) were both due to the same event , i.e. produced within one year .
then , with the known background rates for @xmath0c ( m12 ) and for @xmath3be @xcite , we conclude that @xmath85 ( 1@xmath70 error ) times more @xmath0c was produced than @xmath3be , above the respective backgrounds .
since some of the @xmath3be production may have been produced by some other effects in that decade ( lower time resolution ) , this ratio can be considered as a lower limit .
we can compare this to the expected outcome for a short grb : the typical energy spectrum of a short grb is @xmath86 with a power - law index @xmath71 and a cutoff energy e@xmath87 @xcite .
we obtained the cross - sections @xmath70 for the reactions @xmath0n(n , p)@xmath0c and @xmath0n(n , p+@xmath71)@xmath3be from e = 10 to 60 mev ( footnote 3 ) .
we show the dependence of the cross - sections @xmath70 on the energy e for both reactions in fig .
then , we integrated the cross - section @xmath70 at that energy range e multiplied by the typical short grb spectra sp(e ) over the energy e from 10 to 60 mev for both @xmath0c and @xmath3be to obtain the expected ratio : @xmath88 given the definition of cross - section , both the nominator and the denominator in this equation are proportional to the respective number of atoms produced .
we estimated the ratio as expected differential production rates for a grid of @xmath71 and e@xmath87 as observed for short grbs , i.e. power law index @xmath71 from @xmath89 to 1.4 and cutoff energy e@xmath87 from 49 to 1900 kev @xcite , using xspec .
the results are shown in fig .
we then also estimated the expected fraction of @xmath0c to @xmath3be production over the backgrounds for a long grb with a band function as spectrum @xcite , i.e. a smoothly broken power - law .
again , we used the full range of parameters observed for all those long grbs , where all three spectral parameters were obtained by good fits , namely for the 1st index @xmath71 from @xmath90 to 0.07 , for the 2nd index @xmath91 from @xmath92 to @xmath93 , and for the cutoff energy e@xmath87 from 52 to 2867 kev ( not redshifted ) for long grbs observed by fermi or batse @xcite .
[ fg2 ] already shows qualitatively that a long grb with a model of a smoothly broken power law ( band function ) can not explain that the @xmath0c production was so much larger than the @xmath3be production given the energy - dependence of the cross - sections of those reactions .
our detailed calculations show that a long grb with the band function would produce only up to 18 times more @xmath0c than @xmath3be .
only a short grb can explain this observable ( ratio @xmath80 ) .
the rate of @xmath0c peaks ( 1 in 3000 yr ) is not well constrained , but non - zero ; the error of the rate is fully unconstrained .
the rate of short grbs is @xmath94 gpc@xmath95 yr@xmath79 @xcite , all observed and pointed at us . using then @xmath96 mpc@xmath95 as number density of typical galaxies
@xcite , we have one short galactic grb in @xmath97 kyr beamed towards earth .
if we restrict this estimate to the galactic disk within 4 kpc , the observed rate is 10 times lower , i.e. one nearby galactic grb beamed at us in @xmath98 kyr . given the sensitivity limits of detectors like batse and swift , this is a highly uncertain lower limit .
while this is not consistent with 1 event in 3000 yr within @xmath99 , it is consistent within @xmath100 .
given that the error of the @xmath0c event rate is unknown , it is consistent with the rate of local short grbs even within @xmath101 . even though the possible connection between short grbs and compact mergers are not of central importance for our arguments , because we use only real observables of short grbs
, we will now consider the rates of mergers of compact objects .
from the three known double neutron stars , one can expect the rate of mergers per galaxy to be 3 to 190 myr@xmath79 ( @xmath99 error range ) with the mean being 13 myr@xmath79 @xcite . from the initial mass function and ,
hence , birth rate of massive stars that can become neutron stars , then taking into account the multiplicity rate , evolution , and interaction of massive stars , one can predict 0.3 to 50 mergers per galaxy myr@xmath79 ( @xmath99 error range ) with the mean being 15 myr@xmath79 @xcite . thus , at most we expect 190 mergers per galaxy myr@xmath79 or up to one merger in @xmath102 yr . if such a merger would be observable as short grb , one would have to correct the rate for the beaming fraction @xmath103 to 0.13 for short grbs @xcite .
for @xmath104 , one would then expect up to one merger in @xmath105 yr ( within @xmath99 error bars ) , pointed towards earth as short grb ( ten times less within 4 kpc ) . because of sensitivity limits of observational techniques , both the observed multiplicity rate of massive stars and the estimated number of double neutron stars and mergers are lower limits .
we can add the rates of mergers between neutron stars and black holes and between two black holes .
neither the highly uncertain rates of observed short grbs nor of neutron star mergers are consistent with the observed rate of the @xmath0c event ( one event in 3000 yr ) within @xmath99 , but all are consistent within @xmath101 .
furthermore , a short grb is the only known phenomenon that can provide correct energetics , correct spectrum , and correct time - scale for the observed event ; it also does not produce a typical sn light curve for several month nor a detectable snr nor a mass extinction event on earth , which are all missing .
if the ad 774/5 event was a short grb and if the probability to observe one galactic grb within 3000 yr is too small , one would have to conclude that there are more ( fainter ) short grbs than observed so far , and/or that there is another astrophysical population contributing to short grbs , which was not yet fully recognized .
short grbs with extended emission may partly be due to either an accretion - induced collapse of a white dwarf or the merger of two white dwarfs @xcite .
in such an event , a magnetar can form @xcite .
the rate of mergers of two white dwarfs with a total mass above the chandrasekhar mass limit of @xmath106 m@xmath59 has been estimated to be @xmath107 per m@xmath59 with @xmath108 error bars @xcite .
this corresponds to only about one tenth of the sn ia rate , so that the merger of two white dwarfs with super - chandrasekhar mass can not explain all sn ia events .
however , it is suggested that such super - chandrasekhar mergers can be observed as short grbs @xcite . for our galaxy with @xmath109
m@xmath59 , we obtain a rate of @xmath110 mergers of white dwarf binaries with super - chandrasekhar mass in 3000 yr .
if we assume that such an event can be observed as short grb with extended emission with a beaming factor of up to @xmath111 @xcite , and if we also restrict the rate to the disk within 4 kpc ( the maximum distance of a short grb to explain the ad 774/5 event ) , we expect @xmath112 such mergers in 3000 yr .
since the error of the rate of @xmath0c events ( as in ad 774/5 ) is unconstrained , one can not claim that the rates of white dwarf binary mergers with super - chandrasekhar mass ( pointing towards us a short grb ) and the @xmath0c event rate are inconsistent . since short grbs with extended emission may have lower total energies @xcite , which can be below the batse and swift sensitivity limits , they may often remain undetected , so that their observable rates could be underestimated .
there is evidence for short grbs with lower energies and their rate is probably much higher @xcite .
given the discussion of the rates , we can speculate that some short grbs , like possibly one in ad 774/5 , are due to an accretion - induced collapse of a white dwarf or the merger of two white dwarfs , and that such ( possibly frequent ) short grbs typically have energies below the current sensitivity limit of @xmath60 erg and possibly wide beaming angles ; they may produce a neutron star , but no snr .
a long grb can be accompanied by a sn and a snr , but none were observed for ad 774/5 ; this could be due to strong absorption ( no optical sighting of the sn ) and large distance ( faint snr with very small angular extension on sky ) . with the typical isotropic equivalent energy output e(event )
= @xmath61 to @xmath113 erg of a long grb in @xmath1-rays of 10 kev to 30 gev and @xmath62 to 1 @xcite , we estimate its putative distance towards earth from equ . 1 to @xmath114 to @xmath115 kpc , i.e. in our galaxy or the neighbouring canis major , sagittarius , or ursa major ii dwarf galaxies . to avoid a historical sighting of a sn brighter than @xmath30 mag @xcite at the minimum distance of 1 kpc
, one would need an absorption of @xmath116 mag for a peak absolute magnitude of @xmath117 mag for a collapsar / hypernova @xcite .
an area of @xmath118 deg@xmath40 has an absorption of @xmath119 mag @xcite , an even smaller area for clouds within 1 kpc .
there is no such snr detected behind these areas ( * ? ? ?
* footnote 1 ) . considering also that long grbs are
20 times less frequent than short grbs @xcite , a long grb behind such strong absorption is then @xmath120 times less likely than a short grb to explain the @xmath0c peak in ad 774/5 .
also , sampling the whole observed range of spectral parameters of long grbs with their smoothly broken power law or band function @xcite , we can not explain the differential @xmath0c to @xmath3be production observed ( fig .
[ fg1 ] ) .
hence , a short grb remains the only plausible explanation for the @xmath0c peak in ad 774/5 .
we list in table 1 all known neutron stars with characteristic age @xmath121 yr ( * ? ? ?
* footnote 2 ) at distances from 1 to 4 kpc , but without any known snr ( green 2009 ; footnote 1 ) .
the list includes two anomalous x - ray pulsars ( axps ) , one fermi gamma - ray large area telescope pulsar ( fgl ) , and one soft gamma - ray repeater ( sgr ) .
we use a larger pulsar age upper limit ( 25000 yr ) than the time since ad 774/5 , because characteristic ages are usually upper limits and can be 20 times larger than the true age @xcite .
we include sgr 0418 + 5729 with an upper limit for the period derivative and , hence , a lower limit on the age ( which is below 25000 yr ) . if a neutron star was formed in ad 774/5 , it is also possible that it was not yet discovered , e.g. because of misdirected pulsar beaming , or that distance and/or age have not yet been determined .
for the five pulsars listed here , one should obtain deep x - ray , @xmath1-ray , h@xmath71 , and radio observations to search for snrs : if a snr can be excluded in one of them , that pulsar may be a good candidate for the product of the ad 774/5 event .
three of the five neutron stars listed are axps or sgrs , which can form by a short grb with extended emission @xcite * table 1 : young neutron stars at 1 - 4 kpc without snr .
* we list pulsar name with position for j2000.0 , rotation / pulse period @xmath122 , period derivative @xmath123 , distance , and characteristic age @xmath124 , data from @xcite or mcgill sgr / axp catalog ( footnote 2 ) .
+ [ cols="<,>,>,^,^ , < " , ] in summary , all observables of the @xmath0c peak in ad 774/5 are consistent with a galactic short grb at 1 - 4 kpc : sufficient energetics , correct spectrum , and correct time - scale , also neither a sn nor a snr nor a mass extinction event .
the only assumptions made were the following : from comparing their @xmath0c tree ring data with a model of incorporation of @xmath0c into the biosphere , m12 concluded that the event duration was one year or shorter , and they could then derive the @xmath0c flux and energy deposited on earth which was also used in our work ; we also used the @xmath3be flux observed in the same decade @xcite and assumed that it was produced by the same event ( see e.g. * ? ? ?
* ) to derive the differential @xmath0c to @xmath3be production rate ; we derived a lower limit to the differential production rate , because some of the @xmath3be observed in that decade could have been produced by other effects .
the derived lower limit was then found to be inconsistent with long grbs , but fully consistent with the spectra of short grbs .
rates of short grbs and neutron star mergers are marginally consistent with one event in 3000 yr , but the error of the rate of the ad 774/5 event is unknown .
the merger of two white dwarfs with super - chandrasekhar mass or an accretion - induced collapse of a white dwarf producing a short grb ( with @xmath125 erg at @xmath83 kpc ) should also be considered .
the product could be a neutron star without snr , so that our conclusions are testable .
we would like to thank the german national science foundation dfg ( deutsche forschungsgemeinschaft ) for financial support through the collaborative research center sonderforschungsbereich sfb tr 7 gravitational wave astronomy sub - project c7 .
we used the online catalog of supernova remnants by d. green , the chandra supernova remnant catalog maintained by f. seward , the atnf online catalog of pulsars maintained by g.b .
hobbs and r.n .
manchester , and the mcgill online catalog of sgrs and axps maintained by the mcgill pulsar group .
we also thank kazuhito dobashi for information about the size of the sky area , where the extinction is @xmath33 or @xmath126 mag , considering the whole sky .
badenes c. & maoz d. , 2012 , apj 749 , l11 band d. , matteson j. , ford l. , et al . , 1993 , apj 413 , 281 berger e. , 2007 , apj 670 , 1254 berger e. , 2011 , newarev .
55 , 1 bibby j.l . ,
crowther p.a . , furness j.p . , clark j.s . , 2008 , mnras 386 , l23 bucciantini n. , metzger b.d . , thompson t.a . , quataert e. , 2012 , mnras 419 , 1537 burger g. & ebert h.g .
( eds . ) , 1981 , proc .
neutron dosimetry , munich coward d.m . , howell e.j . , piran t. , et al . , 2012 ,
mnras 425 , 2668 dimbylow p.j .
, 1980 , phys .
25 , 637 dobashi k. , 2011 , pasj 63 , 1 dominik m. , belczynski , k. , fryer , c. , holz , d.e . , berti , e. , bulik , t. , mandel , i. , oshaughnessy , r. , 2012 , apj 759 , 52 ghirlanda g. , nava l. , ghisellini g. , celotti a. , firmani c. , 2009 , a&a 496 , 585 green d.a . , 2009 , bull . astron .
india 37 , 45 horiuchi k. et al . , 2008 ,
. geochronol . 3 , 253 hurley k. , boggs s.e . ,
smith , d.m .
, et al . , 2005 , nat .
434 , 1098 kim c. , kalogera v. , lorimer d. , 2010 , newarev .
54 , 148 kramer m. , lyne a.g . , hobbs g. , lhmer o. carr p. jordan c. , wolszczan a. , 2003 , apj 593 , l31 levan a. & tanvir n. , 2005 , gcn circ .
3927 lombardi m. , lada c.j .
, alves j. , 2008 , a&a 489 , 143 manchester r.n . ,
hobbs g.b . , teoh a. , hobbs m. , 2005 , aj 129 , 1993 ( updates on http://www.atnf.csiro.au/people/pulsar/psrcat ) melott a.l . , et al . , 2004 ,
jbioa 3 , 56 metzger b.d . & berger e. , 2012 , apj 746 , 48 metzger b.d . , giannios d. , thompson t.a . , bucciantini n. , quataert e. , 2011 , mnras 413 , 2013 miyake f. , nagaya k. , masuda k. , nakamura t. , 2012 , nat 486 , 240 ( m12 ) nakar e. , 2007 , phys . rep .
442 , 166 nakar e. , gal - yam a. , fox d.b . , 2006 ,
apj 650 , 281 pagani l. , et al . , 2004 , a&a 417 , 605 piran t. , nakar e. , rosswog s. , 2012 , arxiv:1204.6242 reipurth b. , 2008 , handbook of low mass star forming regions , astron .
pacific rezzolla l. , giacomazzo b. , baiotti l. , granot j. , kouveliotou c. , aloy m.a .
, 2011 , apj 732 , l6 richardson d. , branch d. , casebeer d. , millard j. , thomas r.c . , baron e.a . , 2002 , aj 123 , 745 sadavoy s.i .
, di francesco j. , bontemps s. , et al . , 2010 , apj 710 , 1247 strom r.g . , 1994 , a&a 288 , l1 stuiver m. , et al . , 1998 ,
radiocarbon 40 , 1041 tanvir n. , chapman r. levan a. , priddey r. , 2005 , nat 438 , 991 thomas b.c . , melott , a.l . , jackman , c.h . , et al . , 2005 , apj 634 , 509 zhang , b. , zhang , b. , liang , e.w . , et al . , 2011 , apj 730 , 141 | in the last 3000 yr , one significant and rapid increase in the concentration of @xmath0c in tree rings was observed ; it corresponds to a @xmath1-ray energy input of @xmath2 erg at earth within up to one year in ad 774/5 @xcite .
a normal supernova and a solar or stellar flare are unlikely as cause @xcite , so that the source remained unknown . here
, we show that a short gamma - ray burst ( grb ) in our galaxy is consistent with all observables : such an event is sufficiently short and provides the necessary energy in the relevant spectral range of @xmath1-rays .
its spectral hardness is consistent with the differential production rates of @xmath0c and @xmath3be as observed .
the absence of reports about a historic sighting of a supernova in ad 774/5 or a present - day supernova remnant are also consistent with a short grb .
we estimate the distance towards this short grb to be @xmath4 to @xmath5 kpc - sufficiently far away , so that no extinction event on earth was triggered .
this is the first evidence for a short grb in our galaxy .
[ firstpage ] gamma - rays bursts , supernovae , neutron stars , white dwarfs |
relativistic viscous hydrodynamics has become the workhorse of dynamical modeling of ultra - relativistic heavy - ion collisions .
however , hydrodynamics does not become valid until the medium has reached a certain degree of local momentum isotropization @xcite . in an inhomogeneous system , collective flow ( i.e. space - momentum correlations ) begins , however , to develop already before hydrodynamics becomes valid .
the hydrodynamic stage thus starts with a non - vanishing pre - equilibrium flow @xcite .
in @xcite we therefore performed a systematic study of pre - equilibrium flow effects on heavy - ion collision observables .
we here summarize the main results of this study and add a few recent results that go beyond the work reported in @xcite .
we assume longitudinal boost - invariance and model the approach towards local thermal equilibrium in a heavy - ion collision very simply @xcite as a pre - equilibrium stage of noninteracting free - streaming massless partonic degrees of freedom , separated by a switching time @xmath1 from a strongly coupled quark - gluon plasma ( qgp ) stage in which frequent collisions isotropize the local momentum distribution sufficiently quickly that it can be described by viscous relativistic fluid dynamics .
the variable switching time @xmath1 parametrizes the duration of the thermalization process . for massless degrees of freedom
the evolution of the energy - momentum tensor during the free - streaming stage is independent of the initial transverse momentum distribution as long as it starts out locally azimuthally symmetric , and can be solved analytically @xcite . at @xmath1
we landau - match the analytically evolved pre - equilibrium energy - momentum tensor to viscous hydrodynamic form @xcite .
space - momentum correlations established by the free - streaming dynamics in the pre - equilibrium stage manifest themselves as non - zero initial values for the hydrodynamic flow velocity profiles , giving rise to non - zero initial radial and anisotropic flows .
local momentum anisotropies resulting from the free - streaming evolution of a spatially inhomogeneous initial density profile generate nonzero initial values for the shear stress . matching the traceless pre - equilibrium energy - momentum tensor of noninteracting massless degrees of freedom to a non - conformal , lattice qcd based equation of state @xcite at @xmath1 generates a non - zero initial bulk viscous pressure field in the hydrodynamic fluid .
all of these initial fields are then further evolved with the second - order viscous relativistic fluid dynamics code vish2 + 1 .
after hadronization of the qgp at a pseudocritical temperature @xmath12mev ( determined by the equation of state @xcite ) , the medium constituents are color neutral hadrons whose interactions are weaker than the gluon - mediated interactions in the earlier color - deconfined qgp , causing a rapid growth of their mean free paths and a subsequent breakdown of the fluid dynamic framework .
we therefore switch from vish2 + 1 to a microscopic description , using the hadron cascade code urqmd , on an isothermal switching surface of temperature @xmath4 which we here consider as an unknown parameter to be optimized by comparison with experimental data .
two additional important parameters affecting the evolution of collective flow and the @xmath5 distributions of the finally emitted hadrons are the shear and bulk viscosities during the liquid qgp stage . for simplicity
, we assume the qgp specific shear viscosity @xmath2 to be a temperature - independent , adjustable constant .
for the specific bulk viscosity @xmath13 , which is expected to develop a strong peak due to critical scattering close to the quark - hadron phase transition , we adopt the temperature - dependent parametrization given in@xcite ( which features a peak at @xmath14mev ) , but consider its normalization as a freely tunable parameter . in practice ,
peak values of @xmath13 exceeding 1 ( corresponding to bulk viscosity normalization factors @xmath15 ) cause the ( negative ) bulk viscous pressure to become very large near @xmath16 , leading to negative total pressures and mechanical instability of the medium against cavitation .
these physical instabilities also eventually render the hydrodynamic code numerically unstable .
we found in @xcite that an extended pre - equilibrium stage increases the final radial flow , leading to flatter @xmath5-spectra , while leaving the integrated charged hadron @xmath17 and @xmath18 almost unchanged unless @xmath1 significantly exceeds 2fm/@xmath19 .
bulk viscosity inhibits radial flow build - up , so the constraint from the measured @xmath5 distributions causes a strong positive correlation between @xmath1 and the bulk viscosity normalization factor . the experimental charged hadron elliptic flow constraint anticorrelates bulk and shear viscosity ; at low @xmath5 , the bulk viscous correction @xmath20 at @xmath4 acts as an effective positive chemical potential for massive hadrons @xcite , causing an anticorrelation between the bulk viscosity normalization factor and the chemical decoupling temperature @xmath4 .
distributions for pions , kaons and protons ( left ) , @xmath9 , @xmath10 and @xmath11 ( middle ) , and the differential elliptic flows for @xmath6 , @xmath7 , @xmath8 and @xmath9 ( right ) from 400 hydrodynamic simulations with fluctuating initial conditions , each oversampled 400 times with urqmd to obtain sufficient particle statistics .
solid lines : full vishnu runs . dashed lines : pure hydro runs with immediate freeze - out at @xmath4 , without subsequent hadronic rescattering using urqmd . experimental data from alice @xcite .
[ fig1],title="fig : " ] + distributions for pions , kaons and protons ( left ) , @xmath9 , @xmath10 and @xmath11 ( middle ) , and the differential elliptic flows for @xmath6 , @xmath7 , @xmath8 and @xmath9 ( right ) from 400 hydrodynamic simulations with fluctuating initial conditions , each oversampled 400 times with urqmd to obtain sufficient particle statistics .
solid lines : full vishnu runs .
dashed lines : pure hydro runs with immediate freeze - out at @xmath4 , without subsequent hadronic rescattering using urqmd . experimental data from alice @xcite .
[ fig1],title="fig : " ] distributions for pions , kaons and protons ( left ) , @xmath9 , @xmath10 and @xmath11 ( middle ) , and the differential elliptic flows for @xmath6 , @xmath7 , @xmath8 and @xmath9 ( right ) from 400 hydrodynamic simulations with fluctuating initial conditions , each oversampled 400 times with urqmd to obtain sufficient particle statistics .
solid lines : full vishnu runs .
dashed lines : pure hydro runs with immediate freeze - out at @xmath4 , without subsequent hadronic rescattering using urqmd .
experimental data from alice @xcite .
[ fig1],title="fig : " ] distributions for pions , kaons and protons ( left ) , @xmath9 , @xmath10 and @xmath11 ( middle ) , and the differential elliptic flows for @xmath6 , @xmath7 , @xmath8 and @xmath9 ( right ) from 400 hydrodynamic simulations with fluctuating initial conditions , each oversampled 400 times with urqmd to obtain sufficient particle statistics .
solid lines : full vishnu runs . dashed lines : pure hydro runs with immediate freeze - out at @xmath4 , without subsequent hadronic rescattering using urqmd . experimental data from alice @xcite .
[ fig1],title="fig : " ] these tendencies are illustrated in the top panel of fig .
[ fig1 ] for which we ran vishnu simulations with smooth ensemble - averaged mc - glauber initial conditions for 10 - 20% central pb - pb collisions at 2.76@xmath21gev for 1000 different quadruplets of the parameters listed in table [ t1 ] , calculated the experimental observables listed in the same table ( chosen for their insensitivity to initial - state event - by - event fluctuations ) , determined the @xmath22 for each quadruplet by comparing them with the experimental values , and then reconstructed the posterior likelihood distributions in the 4-dimensional parameter space using mcmc sampling @xcite .
for the set with the lowest @xmath22 , listed in the lower left panel of fig .
[ fig1 ] , we then ran 400 event - by - event vishnu simulations with fluctuating initial conditions , each oversampled 400 times in the urqmd stage for sufficient particle statistics , to predict the full transverse momentum distributions and their elliptic flows @xmath23 for the specific hadron species shown in the bottom row of fig .
[ fig1 ] .
._left : _ measured values of nine hadronic observables from 2.76@xmath21tev 1020% central pb+pb collisions used to constrain the model parameters listed on the right .
@xmath24 and @xmath0 are truncated means calculated from the data tables for the transverse momentum distributions listed in @xcite .
_ right : _ best - fit parameters and 95% confidence intervals ( c.i . ) using the results from markoff chain monte carlo sampling of the posterior parameter distribution shown in fig .
[ fig1 ] ( top ) .
[ t1 ] [ cols="<,^,^ " , ] to illustrate the effect of the microscopically simulated late hadronic rescattering stage on these observables we also added dashed lines illustrating their state at the end of the fluid stage at @xmath4 .
protons and @xmath9s experience a significant radial boost from urqmd , pushing their yields and elliptic flows to larger transverse momenta .
since @xmath9s in urqmd scatter with reduced cross sections relative to those of protons , this shift towards larger @xmath5 is weaker for @xmath9s than for protons ; as a result , the hydrodynamically predicted elliptic flow mass ordering between protons and @xmath9s at @xmath4 , clearly visible in the dashed lines in the lower right panel of fig .
[ fig1 ] and also in the experimental data @xcite , is basically eliminated after hadronic rescattering , although not inverted as previously observed in the vishnu model without pre - equilibrium dynamics @xcite . a shorter lifetime of the hadronic rescattering stage could help to keep this from happening . c. shen , z. qiu , h. song , j. bernhard , s. bass and u. heinz , comput .
commun .
* 199 * ( 2016 ) 61 .
u. heinz et al . , nucl .
a , in press [ arxiv:1509.05818 [ nucl - th ] ] .
p. f. kolb and r. rapp , phys .
c 67 ( 2003 ) 044903 ; p. f. kolb , http://www.bibliothek.iini-regensburg.de/opus/volltexte/2002/66/[phd thesis ] , universitt regensburg , 2002 .
w. broniowski , w. florkowski , m. chojnacki and a. kisiel , phys .
c 80 ( 2009 ) 034902 .
b. schenke , p. tribedy and r. venugopalan , phys .
lett . 108 ( 2012 ) 252301 .
w. van der schee , p. romatschke and s. pratt , phys .
lett . 111 ( 2013 ) 222302 .
p. romatschke , eur .
j. c 75 ( 2015 ) 429 . j. liu , c. shen and u. heinz , phys . rev .
c 91 ( 2015 ) 064906 .
a. bazavov et al .
[ hotqcd collaboration ] , phys .
d 90 ( 2014 ) 094503 .
g. s. denicol , t. kodama , t. koide and p. mota , phys .
c 80 ( 2009 ) 064901 .
r. preghenella [ alice collaboration ] , arxiv:1203.5904 [ hep - ex ] ; b. b. abelev et al .
[ alice collaboration ] , jhep 1506 ( 2015 ) 190 ; phys .
lett . 111 ( 2013 ) 222301 ; phys .
b 728 ( 2014 ) 216 .
k. dusling and t. schfer , phys .
c 85 ( 2012 ) 044909 . | to bracket the importance of the pre - equilibrium stage on relativistic heavy - ion collision observables , we compare simulations where it is modeled by either free - streaming partons or fluid dynamics .
these cases implement the assumptions of extremely weak vs. extremely strong coupling in the initial collision stage .
accounting for flow generated in the pre - equilibrium stage , we study the sensitivity of radial , elliptic and triangular flow on the switching time when the hydrodynamic description becomes valid . using the hybrid code iebe - vishnu @xcite
we perform a multi - parameter search , constrained by particle ratios , integrated elliptic and triangular charged hadron flow , the mean transverse momenta of pions , kaons and protons , and the second moment @xmath0 of the proton transverse momentum spectrum , to identify optimized values for the switching time @xmath1 from pre - equilibrium to hydrodynamics , the specific shear viscosity @xmath2 , the normalization factor of the temperature - dependent specific bulk viscosity @xmath3 , and the switching temperature @xmath4 from viscous hydrodynamics to the hadron cascade urqmd . with the optimized parameters , we predict and compare with experiment the @xmath5-distributions of @xmath6 , @xmath7 , @xmath8 , @xmath9 , @xmath10 and @xmath11 yields and their elliptic flow coefficients , focusing specifically on the mass - ordering of the elliptic flow for protons and lambda hyperons which is incorrectly described by vishnu without pre - equilibrium flow
. collective flow , pre - equilibrium dynamics , quark - gluon plasma , viscosity , model - to - data comparison , parameter optimization , uncertainty quantification |
galactic black hole binaries at high mass accretion rates show spectra which are often dominated by a soft , quasi - thermal component ( e.g. the review by tanaka & lewin 1995 ) .
this shape matches very well with that expected from an optically thick , geometrically thin accretion disc , as described in the seminal paper by shakura & sunyaev ( 1973 ) .
this identification is strengthened by observations of variability in which the temperature , @xmath4 , and luminosity , @xmath5 , of this quasi - thermal component change together in such a way as to indicate that the size of the emitting structure remains approximately constant ( @xmath6 ) .
the last stable orbit is the only obvious candidate for such a radius , so this gives compelling evidence for a key prediction of general relativity , as well as providing an observable diagnostic of the mass and spin of the black hole ( ebisawa et al 1994 ; kubota et al 2001 ; gierliski & done 2004 ; kubota & done 2004 ; hereafter kd04 ) .
the most recent models of accretion disc spectra are able to reproduce this behavior ( davis et al 2005 ; davis , done & blaes 2006 ) .
these supersede earlier models which showed strong deviations from the @xmath7 relation , but which did not include metal opacity nor treat the difference in vertical structure between the radiation and gas pressure dominated regimes ( merloni , fabian & ross 2000 ) .
the disc dominated spectra always contain some small fraction of emission in a tail extending to much higher energies .
however , some high luminosity spectra have a rather different spectral shape , where the tail contains a much larger fraction of the bolometric luminosity .
the disc emission in these spectra ( variously termed very high or steep power law dominant state , hereafter vhs ; miyamoto et al .
1991 ; mcclintock & remillard 2003 ) is less obvious due to the lower contrast against the strong continuum .
nonetheless , it can be identified in spectral fitting , but its derived parameters ( temperature and luminosity ) are somewhat dependent on how the tail is modelled , which in turn depends on the radiation processes involved .
a power law or broken power law description for the tail might be appropriate if the emission is from nonthermal electrons in the jet , but these models require fine tuning in order to match all the observational constraints .
conversely , comptonisation models , where the disc is the source of seed photons , rather naturally reproduce both multiwavelength spectral constraints and x - ray variability patterns ( see 3 ) , so here we explore the consequences of such models .
firstly , the disc temperature in gbh is close to ( or within ! ) the observed x - ray bandpass , so the low energy break in the comptonised component is predicted to be present , giving a rather different spectrum at low energies to that of a power law which extends unbroken through the disc emission .
thus using proper comptonisation models can significantly change the parameters of the inferred disc component ( done , @xmath8ycki & smith 2002 ) and distort the measured luminosity - temperature relation ( kubota et al 2001 ; kubota & makishima 2004 ; kd04 ) .
this helps considerably in interpreting data from the weak vhs , where the disc contributes more than half of the total bolometric luminosity .
a power law model for the tail in these data gives derived disc temperatures which are significantly higher than those seen from the same source at the same disc luminosity in the high / soft state .
using proper comptonisation models instead of the power law approximately recovers the @xmath7 behavior ( kubota et al 2001 ; kubota & makishima 2004 ; kd04 ) .
the match to the high / soft state data is even better with an additional correction to the observed disc luminosity to account for the photons which are scattered into the comptonised spectrum ( kubota & makishima 2004 ; kd04 ) .
however , the most strongly comptonised vhs ( where the disc is less than half of the bolometric luminosity ) have disk spectra which show significantly lower temperatures than those seen at the same disc luminosity in the high / soft state .
these are inconsistent with a constant disc radius even with proper comptonisation models and correction for scattering ( kd04 ) .
this could indicate that the disc is truncated above the last stable orbit ( kd04 ) , but here we explore whether this conclusion is required with a more sophisticated approach to comptonisation .
the parameters derived for the hot electrons indicate that the compton cloud is marginally optically thick , with @xmath1 23 .
this material plainly must intercept the disc photons in order to comptonise them , so will also distort our view of the disc ( kd04 ) .
an optically thick corona over the inner disc could simply hide the hottest part of the disc from view .
more likely , the power dissipated in the corona can _ reduce _ the disc emissivity ( e.g. svensson & zdziarski 1994 ) since the energy in the corona must derive ultimately from the accretion flow .
we build explicit models of an inner disc- corona to explore these ideas , and fit them to two strong vhs spectra ( simultaneous 0.7 200 kev asca and rxte data ) of the galactic black hole binary xte j@xmath0 to constrain the possible geometries of the disk - corona in this state . in 2
, we review the spectral parameters ( luminosity and temperature ) of the geometrically thin disk in the disk dominated high / soft state . in
3 we describe the simultaneous asca - rxte observations of the broadband spectra .
we compare the derived disc emission in the vhs to that seen in the high / soft state , showing that it implies a very different disc area and/or emissivity ( 4 ) . in 5 we describe our model of an inner disc corona above a standard emissivity disc .
fitting this to the data shows explicitly that optical depth effects alone can not hide the inner disc , and that this model requires a large change in disc area compared to the high / soft state .
we modify this in 6 to incorporate energetic coupling of the inner disc and corona as in svensson & zdziarski ( 1994 ) .
the reduction in disc emissivity and temperature under the corona means this model can fit the observed spectra with only a small - or possibly no - increase in disc area .
we review all the results of the models in 7 , and then incorporate additional constraints from other spectral and timing characteristics of this state in 8 .
we show that an optically thick comptonisation in an inner disc corona is not inconsistent with the presence of high frequency qpo s .
however , it does conflict with claims of a strong , extremely broad iron line in these spectra ( miller et al 2004 ) .
we show that there is strong evidence supporting comptonisation as the emission mechanism for the tail ( 9 ) , whereas there is an alternative explanation for the iron line , in which its width is overestimated due to the unrecognized presence of absorption lines ( done & gierlinki 2006 ) .
all the comptonisation models considered here _ require _ that the inner disc changes in extent and/or emissivity in response to the presence of the corona .
there are no solutions in which the inner disc has the same radial structure as that seen in the high / soft state .
in all the following we use the binary system parameters determined by the optical studies of orosz et al . ( 2002 ) i.e. mass of 8.411.2 @xmath9 , distance , @xmath10 , of @xmath11 5 kpc and binary inclination angle , @xmath12 , of @xmath13 ( orosz et al .
2002 ) . however , we are primarily interested in _ changes _ in the disc emitting area rather than absolute values i.e. in comparing the disc structure in the vhs with that seen in the disc dominated high / soft states .
the high / soft states in xte j@xmath14 have been well fit using the diskbb model ( mitsuda et al .
1984 ) in xspec ( kubota & makishima 2004 ; gierliksi & done 2004 ; kd04 ) , giving a clear @xmath7 relationship , implying constant area , over a factor 100 change in disc luminosity .
such an obviously thermal relationship between luminosity and temperature from diskbb fits is now predicted by the best current models of the disc spectra which include radiative transfer with full metal opacities and self consistent vertical structure as well as relativistic effects ( davies et al 2005 ) .
the apparent inner disc radius of this source , @xmath15 , is obtained from the normalization of diskbb model , defined as @xmath16 .
the average value inferred from all the high / soft state constant area diskbb results is @xmath17 km .
however , the true inner disc radius is related to this apparent radius via a series of correction factors ( for the stress free inner boundary condition , effective temperature not equal to colour temperature and relativistic effects ; e.g. kubota et al 1998 ; 2001 ) . however , the fractional _ change _ in inferred disc radius is unaffected by uncertainties in distance , inclination and correction factors , so we give the apparent inner radius , @xmath15 as a _ ratio _ of diskbb normalisation seen in these fits to that derived from all the high / soft state data in kubota & makishima ( 2004 ) .
these ratios are not sensitive to any of the binary system uncertainties , so can sensitively show the _ change _ in the disc properties between the high / soft and vhs .
asca observations of this source were performed three times , on 1998 september 12 , 23 , and 1999 march 17 .
there are simultaneous _ rxte
_ observations corresponding to each asca observation .
the second data set shows a strongly comptonised vhs spectrum , and was included in the analysis of kd04 .
here we also study the first simultaneous data set , which is at the boundary between the low / hard state and the strongly comptonised vhs .
we do not use the third asca dataset as the gain was changed on the corresponding _ rxte _ data dataset midway through the observation , and the spectrum is only a weakly comptonised vhs .
the observational i d s of the corresponding rxte data are 30188 - 06 - 05 - 00 and 30191 - 01 - 10 - 00 , for the first and second asca observations , respectively .
we use the data as extracted by kd04 ( standard procedures for selecting good pca and hexte data , with pca responses derived for each observation using _ pcarsp _ v8.0 : see appendix for a comparison with _ pcarsp _
we use the pca data from 320 kev , and hexte from 20 - 200 kev .
the asca gis events were extracted from circular regions of @xmath18 radius centered on the image peak , after selecting good time intervals in a standard procedure .
net exposures of 1.4 ks and 2.4 ks were obtained for the simultaneous pointing on the first and second observations , respectively .
the dead time correction fractions for the first asca data set are 76.1% for gis2 and 77.9% for gis3 , determined by reference to the count - rate monitor data with little dead time , using the method described in makishima et al ( 1996 ) .
similarly , for the second observation , these are calculated as 85.6% and 87.3% for gis2 and gis3 , respectively .
we only use the gis data ( 0.710 kev ) since the sis is strongly affected by pileup . in order to take into account the calibration uncertainties , we add 1% systematic errors to the gis spectra ( makishima et al 1996 ) and 0.5% systematic errors to the pca .
we also include 0.5% systematic uncertainty on the hexte data to be conservative , though this does not significantly change the fit ( see appendix ) .
we allow @xmath19 to be free ( described using wabs ) as the low energy bandpass of the asca gis constrains this parameter .
the data from different instruments are fitted with same model parameters except that relative normalization factors are free .
hereafter , we use the normalization of the pca in the fitting results so as to be able to directly compare with the pca analysis of kubota & makishima ( 2004 ) for the disc spectra . using this
we find the observed ( i.e. absorbed ) 0.7 200 kev fluxes of @xmath20 and @xmath21 for the first and second observations , respectively , corresponding to observed luminosities of @xmath22 and @xmath23 for isotropic emission .
correcting for absorption and the energy band gives a bolometric luminosity of @xmath24 and @xmath25 i.e. about 35 per cent of the eddington limit .
= 0.65 these vhs spectra are dominated by a strong , steep continuum rather than the disc , though the disc can still be seen .
the continuum shape is complex : fits using a single power law ( together with the disc , and including a smeared edge and gaussian to phenomenologically model the reflection components ) are completely statistically unacceptable , with @xmath26 of @xmath27 and @xmath28 , for the first and second datasets , respectively .
this was also shown by sobczak et al ( 1999 ) , who had to fit the pca and hexte data separately in order to get acceptable fits with a power law continuum .
however , the continuum curvature is complex : a single thermal comptonisation model for the tail ( with disc , smeared edge and gaussian line ) gives a much improved fit over a power law , but the fits are not formally acceptable at @xmath26 of @xmath29 and @xmath30 , respectively .
the complex form of the tail is shown in more detail by gierliski & done ( 2003 ) , who used osse data together with the pca and hexte to extend the bandpass beyond 500 kev for times around that of the second dataset used here .
these clearly showed that the curvature seen in the tail can not be matched by comptonisation from thermal or non - thermal electrons ( see fig .
5 of gierliski & done 2003 ) .
instead the continuum shows features of _ both _ thermal _ and _ non - thermal electrons ( see also zdziarski et al 2001 for similar spectra from grs 1915 + 105 ) . here
we follow kd04 and approximate this by using the thcomp model ( zdziarski , johnson & magdziarz 1996 ; now publically available as a local model in xspec ) to describe the thermal compton scattering , together with a weak power law of @xmath31 fixed at 2 to approximate the higher energy ( non- thermal ) emission .
we have checked that our results on the disc temperature and luminosity are not dependent on this approximation for the shape of the high energy spectrum by using the full hybrid thermal - nonthermal comptonisation code eqpair ( coppi 1999 ) on our final model of the comptonised disc - corona described in 6 .
the seed photons in thcomp are set to be from the disc ( giving a slightly different low energy rollover to that produced by the comptt model in the xspec general release which assumes a wien distribution ) .
this is equivalent to assuming that the comptonising corona covers the whole disc , as shown schematically in fig .
[ fig : sketch1]a . following kd04 we use several different descriptions of the reflected emission , to try to assess the effect on the derived disc parameters of systematic uncertainties in current reflection models .
firstly we use the gaussian line and smeared edge to give a phenomenological description of reflection ( seg ) .
then we replace these components with the physical reflection model incorporated in thcomp ( ref ) which includes both full reflected continuum and self consistent iron line emission for arbitrary ionization state of the reflected material ( @xmath8ycki , done & smith 1999 ) .
ionization is important as the lack of photoelectric opacity in ionized material means reflection contributes to the low energy spectrum so can affect the derived disc temperature and luminosity .
however , the ionised reflection in thcomp does not include compton up and downscattering of the reflected emission in the disc itself .
this is important for highly ionized and/or hot discs , and leads to strong distortions on the predicted line and edge shape in the reflected emission which can affect the spectral fitting , either by distorting the inferred solid angle of reflecting material , or its ionization or amount of relativistic smearing ( ross , fabian & young 1999 ; ballantyne , ross & fabian 2001 ; done & gierliksi 2006 ) . [ cols="^,^ " , ]
the conflict with the broad iron line motivates us to review the evidence for comptonisation as the emission mechanism for the tail seen in these data .
the only feasible alternative appears to be the jet .
x - ray emission from the jet in the low / hard state has been proposed by e.g. markoff et al ( 2001 ) , so here we use this as a guide to what might be expected from these models in the vhs .
direct synchrotron emission might give rise to a power law , or broken power law , or cutoff power law tail ( markoff et al 2001 ; wilms et al 2005 ; markoff , nowak & wilms 2005 ) . in
3 we showed that a power law gave a completely unacceptable fit to these data , but a broken power law is as good a description of the continuum as our thermal+nonthermal comptonisation model ( see appendix ) .
however , such models run into difficulties when extrapolated to lower energies as the steep spectrum dominates the bolometric luminosity .
[ fig : opt_x]a shows this for the best fit broken power law and disc continuum model for the second dataset .
the extrapolated x - ray continuum overpredicts the simultaneously observed optical v and b points ( dereddedned using @xmath32 ) of jain et al ( 1999 ) by a factor of over 1000 .
clearly the data strongly require that there is a low energy break in the continuum between the soft x - ray and optical wavebands .
such a break can be produced from jet models at the point where the jet synchrotron emission becomes optically thick to self - absorption ( e.g. markoff , nowak & wilms 2005 ; giannios 2005 ) . however , there are some constrains on the energy of this break from variability arguments . the short timescale r.m.s .
variability spectra of these vhs pca data show a monotonic increase in fractional variability amplitude with energy form 350 kev , consistent with pivoting about a fixed point at @xmath33 kev ( gierliski & zdziarski 2005 ) .
there is no obvious reason for jet models to pick out such an energy .
this contrasts with models in which the x - ray continuum is produced by compton scattering of disc photons .
this naturally produces a break in the comptonised continuum at energies below the seed photon energy ( see fig .
[ fig : opt_x]b ) , and since this is at @xmath34 kev there is no fine tuning required to explain this energy being picked out as the spectral pivot .
gierliski & zdziarski ( 2005 ) show that for the second dataset used here , the r.m.s .
variability spectrum from the pca is well matched by thermal compton scattering of the disc photons ( see also rodrigues et al 2004 ; zdziarski et al 2005 ; sobolewska & @xmath8ycki 2006 for the same results for other vhs spectra ) .
the comptonisation models actually underpredict the optical flux , but this is expected as the disc model here only includes direct emission , while reprocessing is likely to be an important contributor to the optical emission ( e.g. esin et al 2000 ; hynes 2005 ) , and optical synchrotron emission from the jet may also be present ( see turler et al 2004 for a model of the plateau / state c / class @xmath35 state of grs 1915 + 105 with a turndown in the optical ) . given the strength of the disc emission in the vhs then it is also rather natural to assume that this is the dominant contributor to the seed photons .
again this contrasts with the low / hard state , where the disc is much weaker , and the jet might be expected to be a more important contributor ( markoff et al 2001 ) . however , even for the low / hard state , the jet models fail to reproduce the detailed shape of the tail ( zdziarski et al 2003 ) , so that the current version of these jet models now have the majority of the x - ray emission produced by comptonisation of the disc and cyclo synchrotron photons by thermal electrons at the jet base ( markoff et al 2005 ) .
since disc comptonisation appears to be a component of the x - ray continuum even in the low / hard state , this process should be even more dominant with the stronger disc emission seen in the vhs .
thus both the multiwavelength spectral shape and energy dependence of the x - ray variability strongly support a disc comptonisation interpretation of the x - ray tail .
we show in 7 that this is not inconsistent with the presence of high ( or low ) frequency qpo s .
a hot ( and so presumably geometrically thick ) inner disc corona may actually be required to produce the strong jet ( mckinney 2005 ) traced by radio emission during this outburst of xte j@xmath0 ( corbel et al 2001 ) .
the only obvious inconsistency is with the interpretation of the iron line in these data given by miller et al ( 2004 ) .
however , unlike the comptonisation interpretation of the tail , which seems to be the only viable mechanism , there does exist an alternative explanation for the line shape .
done & gierliski ( 2005 ) show that the asca spectrum in the second vhs observation used here can be fit equally well by much less extreme line smearing if there is also some ionized absorption present .
the underlying iron line width is then reduced , consistent with originating from reflection in the outer disc , which allows the inner disc to be mostly covered by the corona .
thus there are no data which unambiguously rule out an optically thick inner disc corona , while there are many observations which strongly support its existence .
such a corona will inevitably distort the disc spectrum in the ways explored in this paper .
we investigate the broadband continuum spectra of two vhs observations covering 0.8200 kev .
these have a strong x - ray tail , contributing more than half of the total bolometric luminosity of the source .
the disc in these spectra has an apparent temperature which is much lower than that seen from the same source at the same luminosity in the disc dominated high / soft state .
we show that this is not simply produced by the effect of an optically thick corona hiding the emission from the inner disc , but instead requires that the disc changes in some way from that seen in the high / soft state , either in area ( fig . [ fig : sketch2]a ) , emissivity ( fig . [ fig : sketch2]b ) or both ( fig .
[ fig : sketch2]c ) .
the most obvious change in these spectra is that there is now a corona .
energetically , this must be powered ultimately by the accretion flow , so there is less power available for the disc itself .
we build a specific example of disc - corona coupling , where the corona over the inner disc drains power from the disc material under it , changing its emissivity ( svensson & zdziarski 1996 ) .
we show that this reduces the amount of disc truncation required , to a level where it is probably not significant given the model uncertainties e.g. if some fraction of the accreting mass goes through the corona as well as some fraction of the power , or if the geometry of the corona is more quasi - spherical than slab - like
. however , we marginally favour the coupled disc - corona system with a slightly truncated disc ( fig .
[ fig : sketch2]c ) as this can simultaneously explain the lf qpo in the relativistic disc precession model .
however , since there is as yet no clear consensus on models of qpo formation , this can not give unambiguous information on disc radii .
thus fig .
[ fig : sketch2]b , where the disc area ( though not its properties ! ) is unchanged from the disc dominated states is also consistent with the spectra , though this leaves the change in lf qpo as an open question .
conversely , a large truncation radius with continuous corona ( fig . [ fig : sketch2]a ) _ is _ ruled out by the _
steep _ comptonised spectrum seen . however , we stress that irrespective of the specific model , the strong comptonisation seen in these very high state data clearly requires that the underlying disc properties are rather different to those of a disc without a corona .
unsurprisingly , the data require that the disc structure _ changes _ in response to whatever changing dissipation mode powers the growth of the coronal structure .
we also stress that these vhs spectra are very similar to those seen from several ulx in terms of having a very low apparent disc temperature for their luminosity .
this has been interpreted as evidence for intermediate mass black holes in the ulx , of order @xmath36 ( miller , fabian & miller 2004 ) .
obviously this is not an option in xte j@xmath0 , where the black hole mass is constrained by dynamical studies to be a normal stellar remnant black hole mass of order @xmath37 ( orosz et al 2002 ) .
we suggest that the same process , of a corona draining energy from the inner disc , may be operating in these ulx .
many of these would then be compatible with fairly normal binary systems , where the black hole mass is at the upper end of that which can reasonably form from models of stellar collapse ( typically @xmath38 or up to @xmath39 from recent calculations of binary evolution ; belczynski et al .
2004 ) accreting at high mass accretion rates ( 0.1few times the eddington limit )
. we will explore this possibility in depth in a subsequent paper .
the present work is supported in part by jsps grant of no.13304014 and no.14079101 .
cd thanks isas and riken for their hospitality during visits at which this work was carried out .
ak is supported by special postdoctoral researchers program in riken .
99 abramowicz m. a. , kluzniak w , 2001 , a & a , 374 , 19 agol e. , krolik j. h. , 2000 , apj , 528 , 161 ballantyne d. r. , ross r. r. , fabian a. c. , mnras , 2001 , mnras , 327 , 10 blaes o. , arras p. , fragile c. , 2006 , mnras , submitted , ( astro - ph/0601379 ) belczynski k. , sadowski a. , rasio f. a. , 2004 , apj , 611 , 1068 corbel s. , et al , apj , 554 , 43 coppi p. s. , 1999 , aspc , 161 , 375 di matteo t. , psaltis d. , 1999 , apj , 526 , 101 davis s. w. , blaes o. m. , hubeny i. , turner n. j. , 2005 , apj , 621 , 372 davis s. w. , done c. , blaes o. m. , 2006 , apj , in press done c. , gierliski m. , 2003 , mnras , 342 , 1041 done c. , madejski g. m. , @xmath8ycki p. t. , 2000 , apj , 536 , 213 done c. , @xmath8ycki p. , smith d. a. , 2002 , mnras , 331 , 453 done c. , gierliski m. , 2006 , mnras , 367 , 659 ebisawa k. et al , 1994 , pasj , 46 , 375 esin a. a. , kuulkers e. , mcclintock j. e. , narayan r. , 2000 , apj , 532 , 1069 giannios d. , 2005 , a&a , 437 , 1007 gierliski m. , done c. 2003 , mnras , 342 , 1083 gierliski m. , done c. 2004 , mnras , 347 , 885 ( gd04 ) gierliski m. , zdziarski a. a. , 2005 , mnras , 363 , 1349 gilfanov m. , revnivtsev m. , molkov s. , 2003 , a & a , 410 , 217 haardt f. , maraschi l. 1993 , apj , 413 , 507 hynes r. i. , 2005 , apj , 623 , 1026 jain , r.k .
, bailyn c.d . , orosz j.a . ,
remillard r.a .
, mcclintock j.e . , 1999 ,
apjl , 517 , 131 kubota a. , et al , 1998 , pasj , 50 , 667 kubota a. , done c. , 2004 , mnras , 353 , 980 ( kd04 ) kubota a. , makishima k. , ebisawa k. , 2001 , apj , 560 , l147 kubota a. , makishima k. , 2004 , apj , 601 , 428 makishima k. , et al 1996 , pasj , 48 , 171 markoff s. , falcke h. , fender r. , 2001 , a&a , 372 , l25 markoff s. , nowak m. a. , wilms j. , 2005 , apj , 635 , 1203 mcclintock j. e. , remillard r. a. 2003 , in compact stellar x - ray sources , eds .
lewin w. h. g. , & van der klis , m. , ( cambridge university press , cambridge ) , in press ( astro - ph/0306213 ) mckinney j.c .
, 2005 , apj , 630 , 5 merloni a. , fabian a. c. , ross r. r. , 2000 , mnras , 313 , 193 miller j. m. , fabian a. c. , miller m. c. , 2004 , apj , 614 , 117 miller j.m . , et al .
2003 , mnras , 338 , 7 miller j.m .
, fabian a.c . , nowak m.a .
, lewin w.h.g . , 2004 , to appear in proc .
10th anual marcel grossmann meeting on general relativity ( astro - ph/0402101 ) mitsuda k. , et al 1984 , pasj , 36 , 741 miyamoto s. , kimura k. , kitamoto s. , dotani t. , ebisawa k. 1991 , apj , 383 , 784 orosz j. a. et al 2002 , apj , 568 , 845 pozdnyakov l. a. , sobol i. m. , sunyaev r. a. 1983 , aspr , 2 , 189 remillard r. a. , sobczak g. j. , muno m. p. , mcclintock j. e. , 2002 , apj , 564 , 962 remillard r. a. , 2005 , an , 326 , 804 rodriguez j. , corbel s. , hannikainen d. c. , belloni t. , paizis a. , vilhu o. , 2004 , apj , 615 , 416 ross r. r. , fabian a. c. , young a. j. , 1999 , 306 , 461 shakura n. , sunyaev r. , 1973 , a&a , 24 , 337 ycki p. t. , sobolewska m. a. , 2005 , mnras , 364 , 891 sobczak g. j. , mcclintock j. e. , remillard r. a. , levine a. m. , morgan e. h , bailyn c. d. , orosz j. a. 1999 , apj 517 , 121 stella l. , mario v. , morsink s. m. , 1999 , apj .
524 , l63 svensson r. , zdziarski a. a. , 1994 , 436 , 599 tanaka y. , & lewin w. h. g. 1995 , in x - ray binaries , eds .
lewin w. h. g. , van paradijs j. and van den heuvel w. p. j. , ( cambridge university press , cambridge ) , p126 trler m. , courvoisier t.
j .- l . , chaty s. , fuchs y. , 2004 , a&a , 415 , l35 van der klis m. , 2000 , ara&a , 38 , 717 wilms j. , nowak m. a. , pottschmidt k. , pooley g. g. , fritz s. , 2006 , a&a , 447 , 245 zdziarski a. a. , johnson w. n. , magdziarz p.,1996 , mnras , 283 , 193 zdziarski a. a. , grove j. e. , poutanen j. , rao a. r. , vadawale s. v. , 2001 , apj , 554 , 45 zdziarski a. a. , gierliski m. , rao a. r. , vadawale s. v. , mikoajewska j. , 2005 , mnras , 360 , 825 zdziarski a. a. , lubiski p. , gilfanov m. , revnivtsev m. , 2003 , mnras , 342 , 355 @xmath8ycki , p. t. , done , c. & smith , d. a. , 1999 , mnras , 305 , 231
we use the second dataset to quantify some aspects of the fits , with a baseline model of a broken power law tail ( together with disc and seg reflection gives @xmath40 for @xmath41 kev .
by contrast , using the newest response ( pcarsp 10.1 ) gives @xmath42 for @xmath43 kev .
the shift is within the uncertainties , but more important is systematic : representative high / soft state spectra also have similarly shifted temperatures so that the difference between the disc properties in the vhs and high / soft state is robust . replacing the hexte data with a file without
the 0.5 per cent systematic error gives @xmath44 and @xmath45 kev , nearly identical as the statistical uncertainties are all larger than 1 per cent .
cccccccccc model&@xmath46 & @xmath47 & @xmath48 & @xmath49 & @xmath50 & @xmath51(slab ) & @xmath52 & smedge , line ( a ) & @xmath53 + & @xmath54 & kev & & kev & ( @xmath55 ) & ( @xmath56 ) & & reflection ( b e ) & + + seg&@xmath57 & @xmath58&@xmath59 & @xmath60 & 1.03 & 3.14&0.86 & @xmath61 kev & 201.4/192 + & & & & & ( 37.6 ) & ( 58.0 ) & & @xmath62 & + & & & & & & & & width@xmath63 kev & + & & & & & & & & @xmath64 kev & + & & & & & & & & ew=@xmath65 ev & + ref&@xmath66 & @xmath67 & @xmath68 & @xmath69 & @xmath70 & 3.05/2.64 & 0.76 & @xmath71 & 230.2/194 + & & & & & ( @xmath72 ) & ( 51.0/44.9 ) & & @xmath73 & + & & & & & & & & @xmath74 & + xi1&@xmath75 & @xmath76 & @xmath77 & @xmath78 & @xmath79 & 3.52/2.95 & 0.76 & @xmath80 & 146.3/146 + & & & & & ( 31.0 ) & ( 80.2/64.4 ) & & ( @xmath81 ) & + xi2&@xmath82&@xmath83 & @xmath84 & @xmath85 & @xmath86 & 3.14/2.69 & 0.81 & @xmath87 & 161.2/146 + & & & & & ( 29.8 ) & ( 57.0/48.5 ) & & ( @xmath88 ) & + xi3&@xmath89&@xmath90 & @xmath68 & @xmath91 & 0.97 & 3.10/2.41 & 0.81 & @xmath92 & 157.8/146 + & & & & & ( 33.5 ) & ( 53.5/42.7 ) & & ( @xmath93 ) & + + seg&@xmath94 & @xmath95&@xmath96 & @xmath97 & 1.19 & 4.09 & 0.69 & @xmath98 kev & 157.8/192 + & & & & & ( 36.1 ) & ( 80.9 ) & & @xmath99 & + & & & & & & & & width@xmath100 kev & + & & & & & & & & @xmath101 kev & + & & & & & & & & ew=@xmath102 ev & + ref&@xmath103 & @xmath104 & @xmath105 & @xmath106 kev . ]
& @xmath107 & 5.87/5.32 & 0.29 & @xmath108 & 226.2/194 + & & & & & ( @xmath107 ) & ( 163.9/161.9 ) & & @xmath109 & + & & & & & & & & @xmath110 & + xi1&@xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 & 5.20/4.00 & 0.46 & @xmath116 & 113.5/146 + & & & & & ( 27.6 ) & ( 139.5/98.5 ) & & ( @xmath81 ) & + xi2&@xmath117&@xmath118 & @xmath119 & @xmath120 & @xmath121 & 4.10/3.29 & 0.62 & @xmath122 & 131.6/146 + & & & & & ( 32.9 ) & ( 80.9/64.4 ) & & ( @xmath88 ) & + xi3&@xmath123&@xmath124 & @xmath125 & @xmath126 & 1.13 & 4.18/3.16 & 0.64 & @xmath127 & 127.9/146 + & & & & & ( 34.2 ) & ( 80.8/63.1 ) & & ( @xmath93 ) & + cccccccccc model&@xmath46 & @xmath47 & @xmath15 & @xmath128&@xmath48 & @xmath49 & disk - corona , is calculated for inclination @xmath13 by integrating the model spectrum by 0.001300 kev .
the intrinsic disc luminosity , @xmath129 , is calculated by referring to observed @xmath15 and @xmath47 as @xmath130 . ] & smedge & @xmath53 + & @xmath54 & kev & [ km ] & & & kev & & reflection & + + seg&@xmath131 & @xmath132&@xmath133 & @xmath134 & @xmath59 & @xmath135 & @xmath136 & @xmath137 & 204.9/192 + & & & & & & & @xmath138 & @xmath139 & + & & & & & & & @xmath140&@xmath141 & + & & & & & & & & @xmath142 & + & & & & & & & & @xmath143 [ ev ] & + ref & @xmath144&@xmath145&@xmath146&@xmath147&@xmath148&@xmath149&@xmath150 & @xmath151 & 227.3/194 + & & & & & & & @xmath152 & @xmath153 & + & & & & & & & @xmath154&@xmath155 & + xi1 & @xmath94&@xmath156&@xmath157&@xmath158&@xmath159&@xmath78&@xmath160&@xmath161&148.9/146 + & & & & & & & @xmath162 & & + & & & & & & & @xmath163 & & + xi2 & @xmath164&@xmath165&@xmath166&@xmath167&@xmath168&@xmath169&@xmath150&@xmath170&159.6/146 + & & & & & & & @xmath171 & & + & & & & & & & @xmath154 & & + xi3 & @xmath172&@xmath173&@xmath174&@xmath175&@xmath176&@xmath177&@xmath178&@xmath179&157.0/146 + & & & & & & & @xmath180 & & + & & & & & & & @xmath181 & & + + seg&@xmath182 & @xmath183&@xmath184 & @xmath185 & @xmath96 & @xmath97 & @xmath186 & @xmath187 & 151.1/192 + & & & & & & & @xmath188&@xmath189 & + & & & & & & & @xmath190&@xmath191 & + & & & & & & & & @xmath192 & + & & & & & & & & @xmath193[ev ] & + ref&@xmath194&@xmath195&@xmath196&@xmath197&@xmath198&@xmath199&@xmath200 & @xmath201 & 228.5/194 + & & & & & & & @xmath202 & @xmath203 & + & & & & & & & @xmath204&@xmath205 & + xi1 & @xmath206&@xmath207&@xmath208&@xmath209&@xmath210&@xmath211&@xmath212&@xmath213&127.0/146 + & & & & & & & @xmath214 & & + & & & & & & & @xmath215 & & + xi12 & @xmath216&@xmath217&@xmath218&@xmath219&@xmath220&@xmath221&@xmath222&@xmath223&121.2/146 + & & & & & & & @xmath224 & & + & & & & & & & @xmath225 & & + xi13 & @xmath226&@xmath227&@xmath228&@xmath229&@xmath230&@xmath231&@xmath186&@xmath232&124.2/146 + & & & & & & & @xmath233 & & + & & & & & & & @xmath225 & & + cccccccccc model&@xmath46 & @xmath234 & @xmath15&@xmath128&@xmath48 & @xmath49 & disc - corona , is calculated for inclination @xmath13 by integrating the model spectrum by 0.001300 kev . in this fit , the intrinsic disc luminosity calculated as @xmath235 & smedge & @xmath53 + & @xmath54 & kev & [ km ] & & kev & & & reflection & + + seg&@xmath236 & @xmath237&@xmath238 & @xmath239 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & 204.7/192 + & & & & & & & @xmath244&@xmath245 & + & & & & & & & @xmath140 & @xmath246 & + & & & & & & & & @xmath142 & + & & & & & & & & @xmath247 & + ref&@xmath248&@xmath249&@xmath250&@xmath251&@xmath252&@xmath253&@xmath254 & @xmath255 & 230.9/194 + & & & & & & & @xmath256 & @xmath257 & + & & & & & & & @xmath154 & @xmath258 & + xi1&@xmath236&@xmath259&@xmath260&@xmath261&@xmath262&@xmath263&@xmath264 & @xmath265 & 150.3/146 + & & & & & & & @xmath266 & & + & & & & & & & @xmath190 & & + xi2&@xmath248&@xmath267&@xmath268&@xmath269&@xmath270&@xmath271&@xmath272 & @xmath273 & 159.3/146 + & & & & & & & @xmath274 & & + & & & & & & & @xmath154 & & + xi3&@xmath248&@xmath275&@xmath276&@xmath277&@xmath278&@xmath279&@xmath280 & @xmath281 & 157.9/146 + & & & & & & & @xmath256 & & + & & & & & & & @xmath181 & & + + seg&@xmath282 & @xmath283&@xmath284 & @xmath285 & @xmath286 & @xmath97 & @xmath287 & @xmath187&154.8/192 + & & & & & & & @xmath288&@xmath189 & + & & & & & & & @xmath190 & @xmath289 & + & & & & & & & & @xmath192 & + & & & & & & & & @xmath193 & + ref&@xmath290&@xmath291&@xmath292&@xmath293&@xmath294&@xmath295&@xmath212 & @xmath296 & 207.2/194 + & & & & & & & @xmath297 & @xmath298 & + & & & & & & & @xmath299 & @xmath300 & + xi1&@xmath301&@xmath302&@xmath303&@xmath304&@xmath305&@xmath306&@xmath307 & @xmath308 & 114.8/146 + & & & & & & & @xmath309 & & + & & & & & & & @xmath310 & & + xi2&@xmath311&@xmath312&@xmath313&@xmath314&@xmath315&@xmath231&@xmath316 & @xmath317 & 123.2/146 + & & & & & & & @xmath318 & & + & & & & & & & @xmath225 & & + xi3&@xmath311&@xmath319&@xmath320&@xmath321&@xmath322&@xmath323&@xmath324 & @xmath325 & 128.1/146 + & & & & & & & @xmath326 & & + & & & & & & & @xmath225 & & + | the x - ray spectra of galactic binary systems dominated by a quasi - thermal component ( disc dominated or high / soft state ) are well described by a standard shakura - sunyaev disc structure down to the last stable orbit around the black hole .
this is not the case in the very high ( or steep power law ) state , where the x - ray spectra show both a strong disc component and strong , steep tail to higher energies .
we use simultaneous optical - asca - rxte data from the black hole transient xte j@xmath0 as a specific example of two such spectra , where the power emitted in the tail is more than 50 per cent of the bolometric luminosity . these have disc spectra which are significantly lower in temperature than those seen from the same source at the same luminosity in the high / soft state .
if these give a true picture of the disc then either the disc emissivity has reduced , and/or the disc truncates above the last stable orbit .
however , it is often assumed that the tail is produced by compton scattering , in which case its shape in these spectra requires that the comptonising region is marginally optically thick ( @xmath1 23 ) , and covers a large fraction of the inner disc .
this will distort our view of the disc , especially of the hottest temperature material .
we build a theoretical model of a comptonising corona over an inner disc , and fit this to the data , but find that it still requires a large increase in inner disc radius for a standard disc emissivity .
instead it seems more probable that the disc emissivity _ changes _ in the presence of the corona .
we implement the specific inner disc - corona coupling model of svensson & zdziarski ( 1994 ) , in which some fraction @xmath2 of the accretion power is dissipated in the corona leaving only a fraction @xmath3 to be dissipated in the optically thick disc .
we show that this can explain the low temperature / high luminosity disc emission seen in the very high state with only a small increase in radius of the disc . while this inferred disc truncation is probably not significant given the model uncertainties , it is consistent with the low frequency qpo and gives continuity of properties with the low / hard state spectra .
accretion , accretion discs black hole physics radiation mechanisms : thermal radiative transfer x - rays : binaries x - rays : individual : xte j@xmath0 |
there have been several studies on simple semirings , e.g. , in @xcite . amongst other things a complete classification of finite commutative simple semirings
was presented in @xcite .
but there exists so far no classification of all finite simple semirings .
monico showed in @xcite that every proper finite simple semiring with more than two elements and nontrivial addition is additively idempotent .
thereupon additively idempotent semirings have been studied in @xcite . in this work
we aim to describe all finite simple additively idempotent semirings .
we did not succeed to characterize all these semirings , but our approach covers many cases . before we go into more detail we state the most important definitions .
let @xmath0 be a nonempty set and @xmath1 and @xmath2 two binary operations on @xmath0 .
then @xmath3 is called a _ semiring _
if @xmath4 is a commutative semigroup , @xmath5 is a semigroup , and both distributive laws @xmath6 and @xmath7 hold for all @xmath8 .
if @xmath3 is a semiring , then we write for the multiplication mostly @xmath9 for @xmath10 .
let @xmath3 be a semiring and @xmath11 .
we call @xmath12 _ right _ ( _ left _ ) _ absorbing _ if it is multiplicatively right ( left ) absorbing , i.e. , if @xmath13 ( @xmath14 ) holds for every @xmath15 . if @xmath12 is left and right absorbing then it is called _
absorbing_. if @xmath4 has a neutral element which is absorbing , then it is called the _ zero _ of the semiring @xmath3 . a _ congruence _ of a semiring @xmath3
is an equivalence relation @xmath16 on @xmath0 such that for every @xmath17 : @xmath18 the semiring @xmath3 is called _ simple _ if its only congruences are @xmath19 and @xmath20 . to present the main result from @xcite ,
we need the following theorem about simple semigroups , which can be found in ( * ? ? ?
* theorem 3.7.1 ) .
[ simple_semigroups ] let @xmath21 , and @xmath22 an @xmath23 matrix with @xmath24 for all @xmath25 such that no row or column is identically zero , no two rows are identical , and no two columns are identical .
let @xmath26 and define a binary operation on @xmath27 by @xmath28 then @xmath29 is a simple semigroup of order @xmath30 .
conversely , every finite simple semigroup with an absorbing element is isomorphic to one of this kind .
the main result in @xcite is the following : let @xmath3 be a finite simple semiring .
then one of the following holds : 1 .
@xmath31 , 2 .
@xmath32 for some finite field @xmath33 and some @xmath34 , 3 .
@xmath3 is a zero multiplication ring of prime order , 4 .
@xmath5 is a semigroup as in theorem [ simple_semigroups ] with absorbing element @xmath35 and @xmath36 , 5 .
@xmath4 is idempotent[additively_idempotent ] .
of course every semiring in the first four cases is simple but not every additively idempotent semiring is simple .
hence , if one wants to classify all finite simple semirings , then it remains to describe all finite simple additively idempotent semirings .
the case , where such a semiring has a zero , was studied in @xcite .
we need some preparation to state the main result of it . a _ lattice _
@xmath37 is an ordered set where for every two elements @xmath38 the supremum @xmath39 and the infimum @xmath40 in @xmath41 exists .
the lattice @xmath42 is called _ complete _ if for every subset @xmath43 the supremum @xmath44 and the infimum @xmath45 in @xmath41 exists .
a complete lattice has a greatest element @xmath46 and a least element @xmath47 .
if there is no confusion , then we write just @xmath48 and @xmath49 .
a mapping @xmath50 between two lattices @xmath37 and @xmath51 is called _ join - morphism _ if @xmath52 holds for every @xmath38 . by @xmath53
we denote the set of all join - morphisms from @xmath42 to @xmath42 .
if @xmath42 and @xmath54 are complete and @xmath55 fulfills @xmath56 for every subset @xmath43 , then @xmath55 is called _ residuated _ ( or _ complete join - morphism _ ) .
if @xmath42 and @xmath54 are finite , then @xmath55 is residuated iff it is a join - morphism and fulfills @xmath57 . by @xmath58
we denote the set of all residuated mappings from @xmath42 to @xmath42 .
the structure @xmath59 , where @xmath60 denotes the pointwise supremum and @xmath61 the composition of two mappings , is a semiring for a complete lattice @xmath42 . for @xmath62 define the mapping @xmath63 by @xmath64 more information about lattices can be found in @xcite and about residuated mappings in @xcite .
the main result from @xcite can be stated as follows : [ theorem_zum ] let @xmath37 be a finite lattice and @xmath65 a subsemiring of @xmath66 such that @xmath67 for every @xmath62 . then @xmath68 is a finite simple additively idempotent semiring with zero .
conversely , every finite simple additively idempotent semiring @xmath69 with @xmath70 and a zero is isomorphic to such a semiring .
we use here the same approach as in @xcite , i.e. , we try to characterize every finite simple additively idempotent semiring as a semiring of join - morphisms of a semilattice . for this
we have to distinguish between several cases . by the _ greatest element _ of a finite additively idempotent semiring @xmath3 or a finite commutative semigroup @xmath4 we mean the greatest element of @xmath71 , where the order @xmath72 on @xmath0 is defined by @xmath73 for @xmath10 ; this element is @xmath74 .
we consider the cases where the greatest element of a semiring is 1 .
neither right nor left absorbing , 2 .
right but not left absorbing , 3 .
left but not right absorbing , 4 .
absorbing and the semiring possesses a finite idempotent irreducible semimodule ( see section [ chapter_semimodules ] for definitions ) which 1 .
satisfies property @xmath75 below , 2 .
does not satisfy property @xmath75 . [ case_remaining ] for a finite idempotent semimodule @xmath76 with greatest element @xmath77 we consider the property : @xmath78 this property is satisfied , e.g. , if @xmath76 has a neutral element or if @xmath77 is _ join - irreducible _ ,
i.e. , if @xmath79 for all @xmath80 .
we succeed with this approach for every case , except case [ case_remaining ] , for which we have a conjecture . as semirings in case [ case_remaining ] have no additively neutral element we complete with our characterization theorems the classification of finite simple semirings with additively neutral element , which will be summarized in theorem [ classification_add_neutr ] .
this paper is structured as follows .
section [ chapter_semimodules ] contains a comprehensive study of semimodules , especially idempotent irreducible semimodules .
these semimodules are necessary to describe the embedding of a finite simple additively idempotent semiring into the semiring of join - morphisms of a semilattice , which is done in section [ section_embedding ] . in section [ chapter_subsemirings ]
we study simple subsemirings of a semiring of join - morphisms of a semilattice .
the main results are stated in section [ chapter_main_results ] , which are the characterization theorems that characterize a finite simple additively idempotent semiring as a semiring of join - morphisms of a semilattice .
section [ chapter_isomorphic_semirings ] clarifies that if two semirings considered in the main results are isomorphic , then also the corresponding semilattices have to be isomorphic .
the question when a semiring has an additively or multiplicatively neutral element is answered in section [ chapter_neutral_elements ] , where we also state the complete classification of finite simple semirings with an additively neutral element . in section [ chapter_remaining_case ]
we discuss the remaining case [ case_remaining ] , and at last we present some examples in section [ chapter_example ] .
in the following let @xmath3 be a semiring .
an @xmath0__-semimodule _ _ is a commutative semigroup @xmath76 together with an @xmath0-multiplication @xmath81 , @xmath82 , such that for all @xmath83 and @xmath84 it holds : @xmath85 for an @xmath0-semimodule @xmath76 and a subset @xmath86 we define @xmath87 , and for @xmath88 we define @xmath89 .
let @xmath76 be an @xmath0-semimodule .
( @xmath0-)subsemimodule _ of @xmath76 is a subsemigroup @xmath90 of @xmath76 such that @xmath91 .
if @xmath76 is an @xmath0-semimodule and @xmath88 , then @xmath92 is clearly a subsemimodule of @xmath76 .
let @xmath76 be an @xmath0-semimodule .
a _ ( semimodule ) congruence _ on @xmath76 is an equivalence relation @xmath16 on @xmath93 such that @xmath94 for all @xmath95 and all @xmath11 . for a congruence @xmath96 on an @xmath0-semimodule @xmath76 note
that @xmath97 with @xmath98+[y]:=[x+y]$ ] and @xmath99:=[rx]$ ] is again an @xmath0-semimodule , which is called _
quotient semimodule_. if @xmath76 is an @xmath0-semimodule then we denote by @xmath100 the set of all semigroup endomorphisms of the semigroup @xmath76 .
[ def_t ] an @xmath0-semimodule @xmath76 is called _ faithful _ if the semiring homomorphism @xmath101 is injective . furthermore , @xmath76 is said to be _ faithful of smallest cardinality _
if @xmath76 is faithful and any @xmath0-semimodule @xmath90 with cardinality @xmath102 is not faithful .
an @xmath0-semimodule @xmath76 is called _ trivial _ if @xmath103 . furthermore ,
following @xcite , we call an @xmath0-semimodule @xmath76 _ quasitrivial _ if the homomorphism @xmath104 from definition [ def_t ] is constant , i.e. , if @xmath105 for all @xmath106 , @xmath107 , and we call @xmath76 _ id - quasitrival _ if @xmath108 for all @xmath11 , i.e. , if @xmath109 for all @xmath11 , @xmath107 . clearly , a trivial semimodule is id - quasitrivial , and an id - quasitrivial semimodule is quasitrivial .
[ remark_embed ]
let @xmath3 be simple , and let @xmath76 be a non - quasitrivial @xmath0-semimodule . then @xmath76 is faithful and @xmath3 is isomorphic to the subsemiring @xmath110 of @xmath111 .
an @xmath0-semimodule @xmath76 is called _ sub - irreducible _ , if it is non - quasitrivial and it has only id - quasitrivial proper subsemimodules ; it is called _ quotient - irreducible _ if it is non - quasitrivial and possesses only the trivial proper quotient semimodule , i.e. , if its only congruences are @xmath112 and @xmath113 . if an @xmath0-semimodule is both sub- and quotient - irreducible it is called _
irreducible_. let @xmath3 be finite , simple , and additively idempotent and @xmath76 a finite idempotent @xmath0-semimodule .
then @xmath76 is sub - irreducible iff it is quotient - irreducible .
evidence for this conjecture is given by the fact that all semimodules considered in experiments satisfy this equivalence .
let @xmath3 be in this section a finite simple semiring .
the main result of this section is proposition [ existence_of_irreducible_semimod ] , which states that @xmath3 admits a finite idempotent irreducible semimodule if @xmath3 is additively idempotent and fulfills @xmath114 . if a commutative semigroup @xmath115 ( e.g. , a semimodule or the additive semigroup of a semiring ) is idempotent , then it is a semilattice .
hence , it can be regarded as an ordered set @xmath116 , where @xmath117 for all @xmath118 .
if @xmath115 has a neutral element @xmath119 , then @xmath119 is the least element in @xmath116 .
conversely , a least element in @xmath116 is a neutral element in @xmath115 .
if @xmath27 is finite then @xmath116 has a least element iff @xmath116 is a lattice .
a finite semilattice has a greatest element . to avoid confusion with multiplicatively neutral elements ,
we denote the greatest element of a semilattice @xmath115 , which is an idempotent semimodule or the idempotent additive structure of a semiring , by @xmath120 or just by @xmath121 if it is clear to which semilattice @xmath121 belongs .
if for elements @xmath122 with @xmath123 in an ordered set @xmath124 it holds that @xmath125 then @xmath126 is called a _
lower neighbor _ of @xmath127 and @xmath127 an _ upper neighbor _ of @xmath126 .
an element @xmath128 is called _ minimal _ in @xmath124 if there is no element @xmath129 with @xmath130 .
let @xmath131 denote the set of minimal elements in @xmath124 .
furthermore , for @xmath132 denote @xmath133 and @xmath134 .
the following lemma is ( * ? ? ?
* lemma 6 ) .
[ monico ] if the multiplication table of @xmath3 has two identical rows or two identical columns , then @xmath135 or @xmath136 . [ lemma_existence_of_faithful_semimod ]
let @xmath137 and @xmath114 .
then @xmath4 as an @xmath0-semimodule is not quasitrivial .
therefore , there exists a faithful @xmath0-semimodule @xmath76 of smallest cardinality and @xmath138 .
suppose @xmath4 is quasitrivial , then @xmath139 for all @xmath140 .
thus , every two rows in the multiplication table of @xmath3 are identical , in contradiction to lemma [ monico ] . therefore , @xmath4 is faithful and the rest is clear . if @xmath76 is a faithful @xmath0-semimodule of smallest cardinality , then all proper sub- and quotient - semimodules of @xmath76 are non - faithful , and therefore quasitrivial
. we will show in proposition [ prop_m_is_irreducible ] that @xmath76 is even irreducible .
[ lemma_ra = sa ] let @xmath137 , @xmath114 , let @xmath76 be an @xmath0-semimodule and let @xmath88 such that the subsemimodule @xmath92 is quasitrivial
. then @xmath141 . on the @xmath0-semimodule
@xmath4 consider the congruence @xmath142 defined by @xmath143 for @xmath106 .
since @xmath92 is quasitrivial , for all @xmath140 it holds that @xmath144 , so that @xmath145 , and @xmath142 is even a semiring congruence on @xmath3 .
supposing @xmath146 , we have @xmath139 for all @xmath8 , contradicting lemma [ lemma_existence_of_faithful_semimod ] . by simplicity of @xmath3 ,
then @xmath147 , so that @xmath141 .
[ lemma_ra = m ] let @xmath137 , @xmath114 , and let @xmath76 be a faithful @xmath0-semimodule of smallest cardinality .
then there exists @xmath88 with @xmath148 .
assume on the contrary that for any @xmath88 it holds that @xmath149 , then the @xmath0-semimodule @xmath92 is not faithful and , by simplicity of @xmath3 , it is quasitrivial , hence lemma [ lemma_ra = sa ] implies that @xmath141 .
this means that @xmath76 is quasitrivial , which contradicts the assumption that @xmath76 is faithful .
[ lemma_quotient - irr ]
let @xmath137 , @xmath114 , and let @xmath76 be a faithful @xmath0-semimodule of smallest cardinality .
then @xmath76 is quotient - irreducible .
let @xmath96 be a semimodule congruence on @xmath76 distinct from @xmath112 and let @xmath150 .
then @xmath90 is not faithful and therefore it holds that @xmath151=r[x]=s[x]=[sx]$ ] for all @xmath106 and @xmath107 . by lemma [ lemma_ra = m ] there
exists @xmath88 with @xmath148 . since @xmath152=[sa]$ ] for all @xmath106 , we see that @xmath153 and the statement follows .
[ prop_m_is_irreducible ] let @xmath137 , @xmath114 , and let @xmath76 be a faithful @xmath0-semimodule of smallest cardinality .
then @xmath76 is irreducible . from lemma [ lemma_quotient - irr ] we know that @xmath76 is quotient - irreducible
. now let @xmath90 be a proper subsemimodule of @xmath76 , which has to be quasitrivial , so that @xmath154 for every @xmath155 .
define the equivalence relation @xmath16 on @xmath93 by @xmath156 for all @xmath84 .
let @xmath95 with @xmath157 and let @xmath106 .
it holds that @xmath158 , i.e. , @xmath159 .
it also holds that @xmath160 , i.e. , @xmath161 .
thus , @xmath96 is a semimodule congruence on @xmath76 .
lemma [ lemma_quotient - irr ] implies that @xmath162 or @xmath163 .
assume that @xmath162 .
then for all @xmath84 and all @xmath11 it holds that @xmath164 , so in particular @xmath165 . by lemma [ lemma_ra = m ] there exists @xmath88 with @xmath148 .
then @xmath166 for all @xmath155 , and in particular @xmath167 , a contradiction .
it must hold that @xmath163 . for every @xmath155 there is @xmath168 such that @xmath169 .
hence if @xmath170 for some @xmath171 then @xmath172 for all @xmath11 , so that @xmath173 and thus @xmath174 . now for any @xmath155 and @xmath175 it holds that @xmath176 , so that @xmath177 follows .
thus for every @xmath155 it holds that @xmath178 , which means that @xmath90 is id - quasitrivial .
so we have proven that @xmath76 is sub - irreducible .
[ |rr|>1 ] let @xmath3 be additively idempotent and @xmath114 .
then it holds that @xmath137 .
let @xmath179 and let @xmath96 be the equivalence relation on @xmath0 with the equivalence classes @xmath180 and @xmath181 .
it is easy to check that @xmath16 is a nontrivial congruence of the semigroup @xmath4 .
if @xmath135 would hold , then every equivalence relation on @xmath0 would be a congruence of @xmath5 .
consequently , @xmath96 would be a nontrivial congruence of @xmath3 , contradicting the simplicity of @xmath3 . by this last result ,
when considering a semiring @xmath3 that fulfills @xmath114 and @xmath182 , we can drop the condition @xmath137 in the case of an additively idempotent semiring .
[ lemma_m_is_idempotent ] let @xmath3 be additively idempotent , @xmath114 , and let @xmath76 be a faithful @xmath0-semimodule of smallest cardinality .
then @xmath76 is idempotent .
lemma [ lemma_ra = m ] yields the existence of an element @xmath88 with @xmath148 .
let @xmath183 .
then there exists @xmath11 with @xmath184 and it follows that @xmath185 .
thus , @xmath76 is idempotent .
[ existence_of_irreducible_semimod ] let @xmath3 be additively idempotent with @xmath114 .
then there exists a finite idempotent irreducible @xmath0-semimodule . by lemma [ lemma_existence_of_faithful_semimod ]
there exists a faithful @xmath0-semimodule of smallest cardinality , which is irreducible by proposition [ prop_m_is_irreducible ] and idempotent by lemma [ lemma_m_is_idempotent ] .
let throughout this section @xmath3 be a finite simple additively idempotent semiring with @xmath114 and @xmath76 a finite idempotent sub - irreducible @xmath0-semimodule .
we will study the properties of the @xmath0-semimodule @xmath76 , depending on the properties of @xmath3 ( @xmath186 is absorbing , @xmath187 exists and is left absorbing etc . ) .
these properties are needed to describe the embedding of @xmath3 into @xmath188 for a suitable semilattice @xmath42 , what will be done in section [ section_embedding ] .
we start with a preliminary lemma .
let @xmath69 be an additively idempotent semiring and @xmath90 an idempotent @xmath27-semimodule . for all @xmath189 and @xmath190
it holds : @xmath191 if @xmath192 , i.e. , @xmath193 , then @xmath194 , i.e. , @xmath195 .
similarly , if @xmath196 , i.e. , @xmath197 , then @xmath198 , i.e. @xmath199 . [ lemma_a = b ] let @xmath93 be an idempotent sub - irreducible @xmath0-semimodule , and let @xmath200 such that @xmath201 .
then @xmath202 , and @xmath126 is either an absorbing or a neutral element of @xmath76 .
consider the set @xmath203 , which contains the element @xmath126 .
then @xmath90 is a subsemimodule of @xmath76 .
indeed , let @xmath190 and let @xmath15 . for all @xmath11 we have @xmath204 and @xmath205 ,
hence @xmath206 and @xmath207 , as desired . since @xmath76 is non - quasitrivial we have @xmath208 , so that @xmath209 , and hence @xmath90 is id - quasitrivial .
in particular , for @xmath210 we have @xmath211 , so that @xmath202 .
now consider the sets @xmath212 and @xmath213 , which form @xmath0-subsemimodules @xmath214 and @xmath215 of @xmath76 .
we have to show that either @xmath216 , in which case @xmath126 is an absorbing element , or @xmath217 , in which case @xmath126 is a neutral element .
suppose then that @xmath218 and @xmath219 .
since @xmath76 is sub - irreducible we have that @xmath214 and @xmath215 are id - quasitrivial .
we claim that @xmath220 is an @xmath0-subsemimodule of @xmath76 as well .
let @xmath84 , @xmath221 and let @xmath11 ; then clearly @xmath222 .
suppose that @xmath223 , i.e. , @xmath224 .
then , since @xmath225 , it holds that @xmath226 , so that @xmath227 , contradicting @xmath228 . hence @xmath229 is an @xmath0-subsemimodule of @xmath230 ) , which is proper and thus id - quasitrivial . from this and because @xmath231 it follows that @xmath76 is id - quasitrivial , which contradicts a requirement for sub - irreducibility .
[ corollary_id - quasitrivial ] let @xmath90 be a proper , hence id - quasitrivial , subsemimodule of @xmath76 .
if @xmath76 has no neutral element then @xmath232 . if @xmath76 has a neutral element @xmath49 then it holds that @xmath233 .
[ prop_ra = m ] there exists @xmath88 with @xmath148 .
furthermore , if @xmath107 satisfies @xmath234 then either @xmath235 and @xmath236 , or a neutral element @xmath237 exists , @xmath238 , and @xmath239 .
let @xmath107 be such that @xmath234 .
then the subsemimodule @xmath240 of @xmath76 is id - quasitrivial , and by lemma [ lemma_ra = sa ] it follows that @xmath241 .
now lemma [ lemma_a = b ] implies that either @xmath242 and @xmath243 , or @xmath238 and @xmath244 , which proves the last statement .
for the first statement , suppose that for every @xmath88 it holds that @xmath245 , so that @xmath141
. then @xmath76 is quasitrivial , which is a contradiction .
[ prop_greatest_element ] the following holds : 1 .
if @xmath246 for some @xmath107 , then @xmath76 has a neutral element @xmath247 , @xmath238 , and @xmath244 , [ infty_1 ] 2 .
the element @xmath186 is right absorbing iff @xmath243,[infty_2 ] 3 . if @xmath186 is not left absorbing then @xmath237 and @xmath244,[infty_3 ] 4 .
if @xmath186 is left absorbing , @xmath76 is quotient - irreducible , @xmath248 , and @xmath249 then @xmath250.[infty_4 ] [ infty_1 ] . :
if @xmath107 satisfies @xmath251 then there exists @xmath252 with @xmath253 and hence @xmath254 .
thus the statement follows from proposition [ prop_ra = m ] .
[ infty_2 ] . : if @xmath186 is right absorbing then @xmath255 for all @xmath11 .
conversely , if @xmath236 , let @xmath107 . if @xmath256 holds then @xmath257
. otherwise , by [ infty_1 ] .
we must have @xmath237 , @xmath238 , and @xmath244 , and thus @xmath258 .
hence , @xmath259 for all @xmath107 , so that , since @xmath76 is faithful , @xmath186 is right absorbing .
[ infty_3 ] . : if @xmath256 for all @xmath107 then @xmath260 for all @xmath261 , and , since @xmath76 is faithful , @xmath186 would be left absorbing , contradicting the precondition .
therefore , by [ infty_1 ] .
we have @xmath237 and @xmath239 . [ infty_4 ] . :
assume on the contrary that @xmath262 , so that @xmath239 by proposition [ prop_ra = m ] , and hence @xmath263 .
by [ infty_1 ] .
, for all @xmath264 we have @xmath265 , so that @xmath266 .
it follows easily that the equivalence relation on @xmath93 with classes @xmath267 and @xmath268 is a nontrivial semimodule congruence , contradicting the quotient - irreducibility .
[ prop_semirings_with_neutral_element ] let @xmath4 have a neutral element @xmath187
. then @xmath76 has a neutral element
. by proposition [ prop_ra = m ] there exists @xmath88 with @xmath148 . for all @xmath11
it holds that @xmath269 and thus @xmath270 , therefore @xmath271 is the least element , i.e. , the neutral element , in @xmath76 .
[ prop_has_a_zero ] let @xmath186 be neither left nor right absorbing .
then @xmath3 has a zero . since @xmath186 is not left absorbing it follows from proposition [ prop_greatest_element]-[infty_3 ] .
that @xmath76 has a neutral element @xmath247 such that @xmath244 .
since @xmath186 is not right absorbing , proposition [ prop_greatest_element]-[infty_2 ] . implies @xmath272 and proposition [ prop_ra = m ] implies @xmath273 .
hence , there exists @xmath11 such that @xmath274 .
then we have @xmath275 for all @xmath107 .
since @xmath76 is faithful , @xmath276 must be the neutral element of @xmath4 , and @xmath277 for all @xmath107 .
for all @xmath11 and @xmath107 it follows @xmath278 and @xmath279 , so that , since @xmath76 is faithful , @xmath280 . hence , @xmath187 is left and right absorbing and therefore a zero .
let first @xmath3 be any semiring .
[ lemma_uv ] let @xmath76 be an idempotent @xmath0-semimodule and let @xmath281 be such that @xmath282 is minimal , @xmath283 , and @xmath284 for all @xmath285 . define the ( reflexive , symmetric ) relation @xmath286 on @xmath93 by @xmath287 and let @xmath96 be the transitive hull of @xmath286 .
then the equivalence relation @xmath96 is a congruence on @xmath93 .
let @xmath288 with @xmath289 , then we claim that @xmath290 .
otherwise , there exists @xmath11 with @xmath291 .
if , say , @xmath292 and @xmath293 , it follows from the minimality of @xmath282 that @xmath294 and from the condition on @xmath295 that @xmath296 , hence @xmath297 , contradicting @xmath289 .
similarly , if @xmath298 and @xmath299 we infer that @xmath300 and @xmath301 , again a contradiction . from this it follows that @xmath302 implies @xmath303 , for all @xmath288 .
now let @xmath200 with @xmath289 and let @xmath15 .
then @xmath304 , since for all @xmath11 it holds that @xmath305 .
therefore , for all @xmath200 and @xmath15 we have that @xmath302 implies @xmath306 . [ prop_uv ]
let @xmath76 be a quotient - irreducible idempotent @xmath0-semimodule and let @xmath281 be such that @xmath282 is minimal , @xmath283 , and @xmath307 for all @xmath308 .
suppose that either 1 .
@xmath309 and @xmath310 for all @xmath311 , or 2 .
@xmath312 and @xmath313 for all @xmath308 .
then for all @xmath200 with @xmath314 there exists @xmath11 such that @xmath315 and @xmath296 .
consider the congruence @xmath96 on @xmath93 of lemma [ lemma_uv ] .
suppose that condition 1 . holds and assume that @xmath316 for some @xmath317
. then also @xmath318 for some @xmath319 . by the condition there
exists @xmath11 such that @xmath320 and @xmath321 , hence @xmath322 , contradicting @xmath323 .
therefore , @xmath162 can not hold , and by quotient - irreducibility of @xmath93 it follows that @xmath163 . similarly , @xmath324 does not hold under condition 2 . for any @xmath325 , and it follows that @xmath163 in this case as well .
now let @xmath200 with @xmath314 . then @xmath326 and since @xmath327 there exists @xmath11 such that @xmath328 .
since @xmath329 we have @xmath315 and @xmath330 , and from the condition on @xmath295 it follows that @xmath296 .
now let @xmath3 be a finite simple additively idempotent semiring with @xmath114 , and let @xmath76 be a finite idempotent irreducible @xmath0-semimodule .
the following two propositions are density results akin to ( * ? ? ?
* proposition 3.13 ) .
let @xmath200 .
if there exists an element @xmath11 , with @xmath331 then it is unique , since @xmath76 is faithful , and we denote it by @xmath332 . [ prop_r_a0 ] let @xmath186 be not left absorbing .
then @xmath333 for every @xmath334 . by proposition
[ prop_greatest_element ] , @xmath76 has a neutral element @xmath247 and it holds that @xmath244 . also , @xmath335 for all @xmath336 by proposition [ prop_ra = m ] and @xmath337 , hence @xmath338 for all @xmath339
. therefore , we can apply proposition [ prop_uv ] using condition 1 . with @xmath340 and @xmath341 .
for fixed @xmath342 we conclude that for any @xmath107 with @xmath343 there exists @xmath344 such that @xmath345 and @xmath346 .
define now @xmath347 and let @xmath348 . if @xmath349 , then @xmath350 for every @xmath351 with @xmath343 , i.e. , @xmath352 .
if @xmath353 , then @xmath354 , i.e. , @xmath355 .
thus , @xmath356 .
[ prop_r_au ] suppose there exists @xmath357 such that @xmath358 for all @xmath359 and @xmath243 .
then @xmath360 for every @xmath334 . by proposition [ prop_ra = m ] ,
if @xmath76 has no neutral element then @xmath251 for all @xmath359 , and if @xmath76 has a neutral element @xmath340 then @xmath251 for all @xmath361 and @xmath362 .
hence @xmath313 for all @xmath359 and we can apply proposition [ prop_uv ] using condition 2 . with @xmath282 and @xmath341 .
for fixed @xmath342 we conclude that for any @xmath107 with @xmath343 there exists @xmath344 such that @xmath363 and @xmath346 . as in the proof of proposition [ prop_r_a0 ] , we define @xmath364 and we have @xmath365 for @xmath349 and @xmath355 otherwise .
thus , @xmath366 .
in this section let @xmath3 be again a finite simple additively idempotent semiring with @xmath114 . we are going to embed @xmath3 into the semiring @xmath188 for a suitable finite semilattice @xmath37 .
the subsemiring @xmath367 of @xmath188 corresponding to @xmath3 fulfills then certain conditions , depending on the properties of @xmath3 .
we list in the beginning of this section all conditions that may arise for @xmath367 and which may be necessary for the characterization of @xmath3 .
first , we need two notations . for @xmath62 ,
let @xmath368 be the mapping from @xmath41 to @xmath41 that maps constantly to @xmath126 , and let @xmath369 be the mapping defined by @xmath370 the semiring @xmath367 may fulfill some of the following conditions : @xmath371 if @xmath42 is a lattice then @xmath367 may also fulfill : @xmath372 we also need the following notations .
let @xmath42 be a finite semilattice and @xmath54 a finite lattice .
then we denote : @xmath373 [ lemma1 ] let @xmath37 be a finite lattice , @xmath374 and @xmath375 .
then there exists an element @xmath376 such that @xmath377 .
define @xmath378 .
then @xmath379 holds for every @xmath380 and it follows that @xmath377 . because of @xmath381 and @xmath382 , it can not hold that @xmath383 .
[ lemma_e_a1_leq_f ] let @xmath37 be a finite lattice and @xmath367 a simple subsemiring of @xmath384 that fulfills ( [ condition_e_a1 ] )
. then it also fulfills ( [ condition_e_a1_leq_f ] ) .
define the set @xmath385 and the equivalence relation @xmath96 on @xmath27 with the equivalence classes @xmath386 and @xmath387 for every @xmath388 .
let @xmath389 with @xmath390 and @xmath391 .
consequently , @xmath55 and @xmath392 must be contained in @xmath386 and hence there exist @xmath393 with @xmath394 and @xmath395 .
one can easily show that @xmath396 , @xmath397 , and @xmath398 for some @xmath399 holds for every @xmath400 , what yields @xmath401 , @xmath402 , @xmath403 .
analogously , one can show @xmath404 , @xmath405 , @xmath406 and it follows that @xmath407 , @xmath408 , @xmath409 .
thus , @xmath96 is a congruence .
since @xmath96 must be trivial and @xmath386 is a class with more than one element , @xmath410 follows . hence , @xmath411 .
note in the following that @xmath412 holds for a finite idempotent semimodule @xmath76 .
[ prop_right_but_not_left_greatest_el ] let @xmath186 be right but not left absorbing .
then there exists a finite lattice @xmath42 with more than two elements such that @xmath3 is isomorphic to a subsemiring of @xmath384 which fulfills ( [ condition_e_a1 ] ) , ( [ condition_e_a1_leq_f ] ) , and ( [ condition_fb = x ] ) . by proposition [ existence_of_irreducible_semimod ] , there exists a finite idempotent irreducible @xmath0-semimodule @xmath76 , and @xmath3 is isomorphic to the subsemiring @xmath110 of @xmath413 , by remark [ remark_embed ] . from proposition [ prop_greatest_element ]
it follows that @xmath76 has a neutral element @xmath247 , i.e. , @xmath414 is a lattice , and @xmath244 holds , as well as @xmath243 .
thus , @xmath415 is even a subsemiring of @xmath416 .
the lattice @xmath417 must have more than two elements , because of @xmath114 .
now , ( [ condition_e_a1 ] ) follows by proposition [ prop_r_a0 ] , ( [ condition_e_a1_leq_f ] ) by lemma [ lemma_e_a1_leq_f ] , and ( [ condition_fb = x ] ) by proposition [ prop_ra = m ] . to achieve a similar result for the case that @xmath186 is left but not right absorbing , we need some preparation .
let @xmath37 be a finite lattice with supremum @xmath60 and infimum @xmath418 .
then @xmath419 is the dual lattice of @xmath42 with supremum @xmath420 , infimum @xmath421 , least element @xmath422 , and greatest element @xmath423 . for two mappings @xmath424 on a set @xmath27
, we define @xmath425 . for a residuated mapping @xmath426
there exists a unique isotone mapping @xmath427 with @xmath428 and @xmath429 , which is called the _ residual _ of @xmath55
. we will denote the residual of a residuated mapping @xmath55 by @xmath430 and we define @xmath431 and @xmath432 for @xmath433 . for @xmath434 and @xmath435
it holds that @xmath436 .
it further holds that @xmath437 and @xmath438 , where @xmath439 is an isomorphism between @xmath59 and @xmath440 ( see @xcite ) .
[ s_plus ] let @xmath37 be a lattice .
then it holds that @xmath441 . moreover , if @xmath367 is a subsemiring of @xmath384 , then @xmath442 is a subsemiring of @xmath443 and it holds that @xmath444 .
let @xmath375 and @xmath435 . since the set @xmath445 is closed under @xmath446 , we have that @xmath447 implies @xmath448 , i.e. , @xmath449 . because of @xmath450 , we have @xmath451 .
now let @xmath452 such that @xmath453 , i.e. , @xmath454 implies @xmath455 .
it follows that @xmath456 .
thus , @xmath457 .
hence , @xmath458 is an isomorphism between @xmath384 and @xmath459 , and for every subsemiring @xmath367 of @xmath384 , we have @xmath460 and @xmath461 .
let @xmath37 be a finite nontrivial lattice and define @xmath462 and @xmath463 .
then let @xmath464 be the mapping defined by @xmath465 the following lemma is easy to prove .
[ psi_s_plus ] let @xmath466 be a finite nontrivial lattice .
then @xmath464 is an isomorphism between @xmath467 and @xmath468 . in particular
, @xmath469 is a subsemiring of @xmath468 and @xmath470 holds for every subsemiring @xmath367 of @xmath467 . [ lemma_derive_semiring ]
let @xmath51 be a finite lattice and @xmath471 .
moreover , let @xmath367 be a subsemiring of @xmath472 which fulfills ( [ condition_e_a1 ] ) , ( [ condition_e_a1_leq_f ] ) , and ( [ condition_fb = x ] ) .
then @xmath473 is a subsemiring of @xmath474 ( where @xmath475 refers to the supremum in @xmath476 ) , which fulfills ( [ condition_k_a ] ) , ( [ condition_k_a_leq_f ] ) , and ( [ condition_6 ] ) . by lemma [ s_plus ]
, @xmath442 is a subsemiring of @xmath477 and therefore @xmath478 is a subsemiring of @xmath479 . by lemma [ psi_s_plus ] ,
@xmath473 is a subsemiring of @xmath474 . by ( [ condition_e_a1 ] )
, we have @xmath480 for every @xmath481 and therefore @xmath482 for every @xmath483 , where @xmath484 because of @xmath485 , we get @xmath486 . consequently , condition ( [ condition_k_a ] ) is fulfilled .
now let @xmath487 and @xmath488 .
then by ( [ condition_fb = x ] ) , there exists an @xmath489 with @xmath490 and @xmath491 holds .
hence , @xmath492 implies @xmath493 for every @xmath494 .
let @xmath494 with @xmath495 and assume @xmath496 .
it follows that @xmath497 , what is a contradiction , and we derive the equivalence @xmath498 for every @xmath494 .
if we use the order relation @xmath499 in @xmath476 , then we have @xmath500 if @xmath501 and @xmath502 otherwise .
hence , @xmath503 if @xmath501 and @xmath504 otherwise .
the mapping @xmath505 is then the required mapping for @xmath126 and @xmath127 in condition ( [ condition_6 ] ) .
condition ( [ condition_k_a_leq_f ] ) is satisfied by @xmath473 , because @xmath367 fulfills ( [ condition_e_a1_leq_f ] ) and @xmath506 is isomorphic to @xmath507 .
[ proposition_first_part_left_but_not_right_greatest ] let @xmath186 be left but not right absorbing .
then there exists a finite nontrivial semilattice @xmath42 such that @xmath3 is isomorphic to a subsemiring of @xmath188 that fulfills ( [ condition_k_a ] ) , ( [ condition_k_a_leq_f ] ) , and ( [ condition_6 ] ) .
define @xmath508 for every @xmath106 .
then @xmath509 is a finite simple additively idempotent semiring such that @xmath186 is right but not left absorbing . by proposition [ prop_right_but_not_left_greatest_el ] , there exists a finite lattice @xmath51 with @xmath510 such that @xmath509 is isomorphic to a subsemiring @xmath367 of @xmath472 that fulfills conditions ( [ condition_e_a1 ] ) , ( [ condition_e_a1_leq_f ] ) , and ( [ condition_fb = x ] ) .
let @xmath471 , then by lemma [ lemma_derive_semiring ] , @xmath473 is a subsemiring of @xmath474 , which fulfills ( [ condition_k_a ] ) , ( [ condition_k_a_leq_f ] ) , and ( [ condition_6 ] ) .
clearly , @xmath511 is nontrivial . because of @xmath512 by lemma [ s_plus ]
, it holds that @xmath513 by lemma [ psi_s_plus ] .
the following proposition is ( * ? ? ?
* theorem 2.2 ) for finite semilattices .
note that for a finite semilattice @xmath37 the mappings @xmath369 with @xmath374 and @xmath514 are exactly the mappings of range at most two in @xmath515 . [ prop_f_ab_leq_f ] let @xmath37 be a finite nontrivial semilattice and @xmath367 a subsemiring of @xmath516 that fulfills ( [ condition_f_ab ] ) .
then @xmath367 is simple iff it fulfills ( [ condition_f_ab_leq_f ] ) .
recall that a finite idempotent semimodule @xmath76 satisfies @xmath75 if for its greatest element @xmath77 there exists @xmath517 with @xmath79 for all @xmath359 .
[ prop_irr_semimod_with_join - irr_greatest ] let @xmath186 be absorbing and let @xmath76 be a finite idempotent irreducible @xmath0-semimodule satisfying @xmath75 . then @xmath3 is isomorphic to a subsemiring of @xmath518 that fulfills ( [ condition_f_ab ] ) and ( [ condition_f_ab_leq_f ] ) . by remark [ remark_embed ] , the semiring @xmath3 is isomorphic to a subsemiring of @xmath519 , and because of @xmath243 by proposition [ prop_greatest_element ] even of @xmath518 . since @xmath114
we must have that @xmath248 as well .
there exists @xmath517 with @xmath358 for all @xmath520 , and without loss of generality we may assume that @xmath521 .
by proposition [ prop_r_au ] we have that @xmath360 for every @xmath342 . by proposition [ prop_ra = m ] and proposition [ prop_greatest_element]-[infty_4 ] .
it follows that @xmath522 .
hence , for all @xmath183 there exists @xmath15 such that @xmath523 and therefore @xmath524 .
thus , ( [ condition_f_ab ] ) is fulfilled and ( [ condition_f_ab_leq_f ] ) follows by proposition [ prop_f_ab_leq_f ] .
in this section we consider the other direction , i.e. , we start with a semilattice @xmath42 and show that certain subsemirings of @xmath188 are simple .
[ prop_second_part_right_but_not_left_greatest ] let @xmath37 be a finite lattice with more than two elements and let @xmath68 be a subsemiring of @xmath384 , which fulfills ( [ condition_e_a1 ] ) , ( [ condition_e_a1_leq_f ] ) , and ( [ condition_fb = x ] ) .
then @xmath68 is a finite simple additively idempotent semiring and the greatest element is right but not left absorbing .
it is clear that @xmath68 is a finite additively idempotent semiring ; its greatest element is @xmath525 .
it is easy to see that each element @xmath526 , where @xmath527 , is right absorbing , hence in particular @xmath528 is right and not left absorbing . to prove simplicity ,
let @xmath96 be a congruence on @xmath68 , and suppose that @xmath529 , i.e. , there are @xmath530 such that @xmath531 and @xmath390
. hence there exists @xmath380 such that @xmath532 , and we may assume that @xmath533
. then we have @xmath526 , so that @xmath534 , and there are @xmath535 such that @xmath536 and @xmath537 . furthermore , @xmath538 and @xmath539 , so that @xmath540 .
we have shown that there are elements @xmath535 with @xmath540 such that @xmath541 .
now we show that for all @xmath542 there exists @xmath543 , @xmath544 such that @xmath545 .
so let @xmath546 and let @xmath547 such that @xmath548 .
considering @xmath549 it is easy to see that @xmath550 . on the other hand
there is @xmath435 such that @xmath551 , and it holds that @xmath552 . from this and since @xmath553 it follows @xmath544 .
furthermore , we have @xmath554 as desired . by applying the last paragraph
repeatedly we see that for all @xmath555 it holds that @xmath556 .
now let @xmath557 be arbitrary and let @xmath558 such that @xmath559 .
then we get @xmath560 .
hence @xmath561 , as desired .
[ proposition_second_part_left_but_not_right_greatest ] let @xmath37 be a nontrivial finite semilattice and @xmath68 a subsemiring of @xmath188 , which fulfills ( [ condition_k_a ] ) , ( [ condition_k_a_leq_f ] ) , and ( [ condition_6 ] ) .
then @xmath68 is a finite simple additively idempotent semiring and the greatest element is left but not right absorbing .
it is clear that @xmath68 is a finite additively idempotent semiring ; its greatest element is @xmath562 .
each element @xmath563 , where @xmath564 , is left absorbing , hence in particular @xmath565 is left but not right absorbing . to prove simplicity , let @xmath96 be a congruence on @xmath68 , and suppose @xmath529 , i.e. , there are @xmath530 such that @xmath531 and @xmath390
. there exists @xmath380 such that @xmath566 , and we may assume @xmath567 .
then we have @xmath568 .
now for all @xmath558 there exists @xmath435 , @xmath569 such that @xmath570 .
indeed , let @xmath547 such that @xmath571 if @xmath572 , and @xmath573 otherwise .
then in particular @xmath574 , and @xmath575 . by applying the last paragraph
repeatedly we see that @xmath576 for all @xmath577 .
now let @xmath578 be arbitrary and let @xmath577 such that @xmath579 . then @xmath580 .
consequently @xmath581 , as desired .
[ prop_semiring_with_absorbing_greatest ] let @xmath42 be a finite nontrivial semilattice and let @xmath68 be a subsemiring of @xmath516 , which fulfills ( [ condition_f_ab ] ) and ( [ condition_f_ab_leq_f ] ) .
then @xmath68 is a finite simple additively idempotent semiring with absorbing greatest element . clearly , @xmath68 is finite and additively idempotent .
the simplicity holds by proposition [ prop_f_ab_leq_f ] .
the greatest element is @xmath582 , for arbitrary @xmath374 , which is absorbing .
[ lemma_l_is_irr_semimod ] let @xmath37 be a finite nontrivial semilattice and @xmath68 a simple subsemiring of @xmath516 , which fulfills ( [ condition_f_ab ] ) and @xmath114 .
then @xmath583 is an irreducible @xmath0-semimodule .
clearly , @xmath583 is an @xmath0-semimodule , which is faithful and hence non - quasitrivial .
let @xmath584 be an @xmath0-subsemimodule of @xmath583 with @xmath585 .
then there exists @xmath487 with @xmath586 and it follows that @xmath587 for every @xmath588 . thus @xmath589 and @xmath583 is consequently sub - irreducible .
let now @xmath96 be a semimodule congruence on @xmath583 with @xmath590 , i.e. , there exist @xmath62 with @xmath591 and @xmath302 . without loss of generality we can say @xmath314 .
it follows @xmath592 .
choose @xmath593 arbitrarily .
then @xmath594 .
hence , @xmath595 for every @xmath596 .
thus , @xmath597 must hold .
we conclude that @xmath583 is quotient - irreducible .
now we are ready to establish the characterization theorems for finite simple additively idempotent semirings of all cases mentioned in the introduction , except case [ case_remaining ] .
the first theorem states that the finite simple additively idempotent semirings with greatest element that is neither left nor right absorbing are exactly the finite simple additively idempotent semirings with zero .
it follows from proposition [ prop_has_a_zero ] ; the second part of the theorem is obvious .
[ main_neither_left_nor_right ] let @xmath3 be a finite simple additively idempotent semiring with @xmath114 and such that @xmath186 is neither left nor right absorbing
. then @xmath3 is isomorphic to a semiring as in theorem [ theorem_zum ] .
conversely , every semiring in theorem [ theorem_zum ] has a greatest element , which is neither left nor right absorbing .
we get the following theorem from proposition [ prop_right_but_not_left_greatest_el ] and proposition [ prop_second_part_right_but_not_left_greatest ] .
[ main_right_but_not_left ] let @xmath42 be a finite lattice with more than two elements and let @xmath68 be a subsemiring of @xmath384 , which fulfills ( [ condition_e_a1 ] ) , ( [ condition_e_a1_leq_f ] ) , and ( [ condition_fb = x ] ) .
then @xmath68 is a finite simple additively idempotent semiring and the greatest element is right but not left absorbing .
conversely , every finite simple additively idempotent semiring @xmath69 with @xmath70 and with right but not left absorbing greatest element is isomorphic to such a semiring
. proposition [ proposition_first_part_left_but_not_right_greatest ] and proposition [ proposition_second_part_left_but_not_right_greatest ] yield the following result .
[ main_left_but_not_right ] let @xmath42 be a finite nontrivial semilattice and @xmath68 a subsemiring of @xmath188 , which fulfills ( [ condition_k_a ] ) , ( [ condition_k_a_leq_f ] ) , and ( [ condition_6 ] ) .
then @xmath68 is a finite simple additively idempotent semiring and the greatest element is left but not right absorbing .
conversely , every finite simple additively idempotent semiring @xmath69 with @xmath70 and with left but not right absorbing greatest element is isomorphic to such a semiring .
the next theorem holds by proposition [ prop_irr_semimod_with_join - irr_greatest ] , proposition [ prop_semiring_with_absorbing_greatest ] , and proposition [ lemma_l_is_irr_semimod ] .
recall that we say that a finite idempotent semimodule ( or a finite semilattice ) @xmath76 satisfies property @xmath75 if for its greatest element @xmath77 there exists @xmath517 with @xmath79 for all @xmath598 [ main_absorbing ] let @xmath42 be a nontrivial finite semilattice satisfying @xmath75 and let @xmath68 be a subsemiring of @xmath516 , which fulfills ( [ condition_f_ab ] ) and ( [ condition_f_ab_leq_f ] ) . then @xmath68 is a finite simple additively idempotent semiring with absorbing greatest element and it possesses an idempotent irreducible @xmath0-semimodule satisfying @xmath75 .
conversely , every finite simple additively idempotent semiring @xmath69 with @xmath70 , with absorbing greatest element , and which possesses an idempotent irreducible @xmath27-semimodule satisfying @xmath75 is isomorphic to such a semiring .
in this section we show that if we have two semirings as in theorem [ main_right_but_not_left ] , theorem [ main_left_but_not_right ] , or theorem [ main_absorbing ] that are isomorphic , then the corresponding semilattices have to be isomorphic as well . in @xcite
the same was done for semirings as in theorem [ theorem_zum ] ( theorem [ main_neither_left_nor_right ] ) .
an _ order isomorphism _ ( resp . _ dual order isomorphism _ ) between two ordered sets @xmath124 and @xmath599 is a surjective mapping @xmath600 with @xmath601 ( resp .
@xmath602 ) for every @xmath603 . note that a ( dual ) oder isomorphism is necessarily bijective .
an _ order automorphism _ of @xmath124 is an order isomorphism from @xmath124 to @xmath124 .
[ lemma_f_00_r ] let @xmath37 be a finite lattice and @xmath68 a subsemiring of @xmath384 that fulfills ( [ condition_e_a1 ] ) .
then @xmath604 is a dual order isomorphism between @xmath605 and @xmath606 .
first we verify @xmath607 .
the inclusion `` @xmath608 '' holds by lemma [ lemma1 ] .
now let @xmath374
. then @xmath609 .
this proves the equality and it follows that @xmath610 is well - defined and surjective . because of @xmath611 for all @xmath393 , @xmath610 is a dual order isomorphism .
[ lemma_r_k_1 ] let @xmath37 be a finite semilattice and @xmath68 a subsemiring of @xmath188 that fulfills ( [ condition_k_a ] ) .
then @xmath612 is an order isomorphism between @xmath42 and @xmath613 .
first we verify @xmath614 .
let @xmath578 . then @xmath615 .
now let @xmath564 .
then @xmath616 .
this proves the equality and it follows that @xmath617 is well - defined and surjective . because of @xmath618 for all @xmath62 , @xmath617 is an order isomorphism .
[ lemma_r_f_ab ] let @xmath37 be a finite semilattice , @xmath68 a subsemiring of @xmath516 that fulfills ( [ condition_f_ab ] ) and let @xmath393 .
then @xmath619 is an order isomorphism between @xmath42 and @xmath620 .
first we verify @xmath621 .
let @xmath557 . then @xmath622 .
now let @xmath593 .
then @xmath623 .
this proves the equality and it follows that @xmath624 is well - defined and surjective . since @xmath625 for all @xmath626 , @xmath624 is an order isomorphism . let @xmath627 be a finite lattice and @xmath628 a subsemiring of @xmath629 as in theorem [ main_right_but_not_left ] for @xmath630 . if @xmath631 and @xmath632 are isomorphic , then @xmath633 and @xmath634 are also isomorphic .
let @xmath631 and @xmath632 be isomorphic and let @xmath635 be an isomorphism .
let @xmath636 for @xmath630 . since @xmath637 is the greatest element in @xmath638 , we have @xmath639 .
it follows that @xmath640 .
hence , @xmath641 .
with lemma [ lemma_f_00_r ] we find that @xmath642 .
it trivially follows that @xmath643 .
let @xmath627 be a finite semilattice and @xmath628 a subsemiring of @xmath644 as in theorem [ main_left_but_not_right ] for @xmath630 .
if @xmath631 and @xmath632 are isomorphic , then @xmath633 and @xmath634 are also isomorphic .
let @xmath631 and @xmath632 be isomorphic and let @xmath635 be an isomorphism .
let here @xmath645 for @xmath630 . since @xmath646 is the greatest element in @xmath638 , we have @xmath647 .
it follows that @xmath648 .
hence , @xmath649 . from lemma [ lemma_r_k_1 ] follows that @xmath650 .
[ prop_isomorphic ] let @xmath627 be a finite semilattice and @xmath628 a subsemiring of @xmath651 as in theorem [ main_absorbing ] for @xmath630 .
if @xmath631 and @xmath632 are isomorphic , then @xmath633 and @xmath634 are also isomorphic .
an element @xmath126 in a finite semilattice @xmath42 is called _ coatom _ of @xmath42 if it is a lower neighbor of @xmath48 . with @xmath652
we denote the set of coatoms in @xmath42 .
let @xmath631 and @xmath632 be isomorphic and let @xmath635 be an isomorphism .
one can easily show that @xmath653 holds .
thus for @xmath654 , @xmath655 there exist @xmath656 , @xmath657 with @xmath658 .
we find that @xmath659 .
hence , @xmath660 . from lemma [ lemma_r_f_ab ] follows that @xmath661 .
we remark that , along similar lines as in this section , one can also prove that for every semiring characterized in section [ chapter_main_results ] there exists up to isomorphism a unique idempotent irreducible semimodule ( with property ( @xmath662 ) , in the case of theorem [ main_absorbing ] ) .
if the greatest element @xmath48 of a finite lattice is join - irreducible , then we denote the unique lower neighbor of @xmath48 by @xmath663 .
let @xmath37 be a finite lattice and @xmath68 a semiring as in theorem [ main_right_but_not_left ] .
then @xmath664 has a neutral element iff @xmath48 is join - irreducible .
if the neutral element exists , then it is right but not left absorbing .
if @xmath48 is join - irreducible , then @xmath665 is clearly a neutral element in @xmath664 .
if @xmath664 has a neutral element @xmath666 then it must fulfill @xmath667 for every @xmath527 . for all @xmath393 we have that @xmath668
because of @xmath669 , i.e. , @xmath48 is join - irreducible .
the element @xmath665 is right absorbing , since @xmath670 and @xmath381 for all @xmath578 .
but it is not left absorbing , since @xmath671 for all @xmath374 .
let @xmath37 be a finite semilattice and @xmath68 a semiring as in theorem [ main_left_but_not_right ] .
then @xmath664 has a neutral element iff @xmath42 is a lattice .
if the neutral element exists , then it is left but not right absorbing .
if @xmath42 is a lattice , then @xmath672 is clearly a neutral element in @xmath664 .
if @xmath664 has a neutral element @xmath666 , then it must fulfill @xmath673 for every @xmath674 .
thus for all @xmath380 , @xmath675 is the least element in @xmath42 , i.e. , @xmath42 is a lattice and it holds that @xmath676 . clearly , @xmath672 is left absorbing , but it is not right absorbing because of @xmath677 .
let @xmath42 be a finite semilattice and @xmath68 a semiring as in theorem [ main_absorbing ] .
then @xmath664 has a neutral element iff @xmath48 is join - irreducible and @xmath42 is a lattice .
if the neutral element exists , then it is neither left nor right absorbing .
if @xmath48 is join - irreducible and @xmath42 is a lattice , then @xmath665 is a neutral element in @xmath664 .
if @xmath664 has a neutral element @xmath666 , then it must fulfill @xmath678 for every @xmath564 and @xmath679 .
thus for @xmath680 , @xmath675 is the least element in @xmath42 , i.e. , @xmath42 is a lattice and @xmath681 holds .
also , for all @xmath682 we have that @xmath668 because of @xmath683 , i.e. , @xmath48 is join - irreducible .
since @xmath684 is absorbing , @xmath666 can not be left or right absorbing .
when considering finite simple additively idempotent semirings with an additively neutral element , any finite idempotent irreducible semimodule over such a semiring has a neutral element by proposition [ prop_semirings_with_neutral_element ] , and thus satisfies @xmath75 .
hence , semirings of this kind with an absorbing greatest element are already characterized by theorem [ main_absorbing ] .
therefore the classification of finite simple semirings with additively neutral element is complete and can be summarized as in the next theorem .
[ classification_add_neutr ] let @xmath3 be a finite semiring with additively neutral element .
then @xmath3 is simple iff one of the following holds : 1 .
@xmath685 , 2 .
@xmath32 for some finite field @xmath33 and some @xmath686 , 3 .
@xmath3 is a zero multiplication ring of prime order , 4 .
@xmath3 is isomorphic to a semiring as in theorem [ theorem_zum ] , 5 .
@xmath3 is isomorphic to a semiring as in theorem [ main_right_but_not_left ] , where @xmath48 is join - irreducible , 6 .
@xmath3 is isomorphic to a semiring as in theorem [ main_left_but_not_right ] , where @xmath42 is a lattice , 7 .
@xmath3 is isomorphic to a semiring as in theorem [ main_absorbing ] , where @xmath48 is join - irreducible and @xmath42 is a lattice .
let @xmath37 be a lattice and @xmath68 a semiring as in theorem [ main_right_but_not_left ] .
then @xmath687 has a neutral element iff @xmath688 . if @xmath688 then @xmath46 is join - irreducible . if @xmath688 then it is clearly a neutral element of @xmath687 .
let @xmath687 have a neutral element @xmath689 and let @xmath380 .
for @xmath690 there exists @xmath578 with @xmath691 .
it follows that @xmath692 , i.e. , @xmath693 . if @xmath688 then there exists @xmath374 with @xmath694 , i.e. , @xmath343 implies @xmath695 , for every @xmath380 .
hence , @xmath126 is the unique lower neighbor of @xmath48 , i.e. , @xmath48 is join - irreducible .
let @xmath37 be a semilattice and @xmath68 a semiring as in theorem [ main_left_but_not_right ] .
then @xmath687 has a neutral element iff @xmath688 . if @xmath688 then @xmath42 is a lattice .
if @xmath688 then it is clearly a neutral element of @xmath687 . if @xmath687 has a neutral element @xmath689 then @xmath696 for every @xmath380 , i.e. , @xmath693 . if @xmath688 then there exists @xmath564 with @xmath697 .
thus @xmath698 for all @xmath380 .
hence , @xmath126 is the least element in @xmath42 , i.e. , @xmath42 is a lattice
. let @xmath37 be a semilattice and @xmath68 a semiring as in theorem [ main_absorbing ] .
then @xmath687 has a neutral element iff @xmath688 .
if @xmath688 then @xmath699 is join - irreducible and @xmath42 is a lattice . if @xmath688 then it is clearly a neutral element in @xmath687 .
let @xmath687 have a neutral element @xmath689 and let @xmath380 . for @xmath374 , the equality @xmath700 holds , i.e. , @xmath693 . if @xmath688 then there exists @xmath374 and @xmath588 with @xmath701 .
thus @xmath343 implies @xmath695 , and @xmath227 implies @xmath702 , for every @xmath380 .
hence , @xmath126 is the unique lower neighbor of @xmath48 , i.e. , @xmath48 is join - irreducible .
also , it follows that @xmath702 for any @xmath703 , so that @xmath127 is the least element and @xmath42 is a lattice .
> from the results in this section it also follows that the existence of a multiplicatively neutral element implies the existence of an additively neutral element , for all semirings in theorem [ main_right_but_not_left ] , theorem [ main_left_but_not_right ] , and theorem [ main_absorbing ] .
the semirings that elude our characterizisation theorems are the finite simple additively idempotent semirings with absorbing greatest element , which possess a finite idempotent irreducible semimodule @xmath93 _ without _ property @xmath75 , so that for all @xmath517 there is @xmath598 such that @xmath704 . for this case
we have a construction of semirings of join - morphisms of semilattices .
in fact , we conjecture that this construction covers these semirings .
we need some preparation for it .
let @xmath705 and @xmath706 be finite semilattices and let @xmath707 . then define @xmath708 where @xmath709 and @xmath710 iff @xmath711 and @xmath712 , for all @xmath713 , @xmath714 .
note that every equivalence class in @xmath715 , except @xmath716 , has just one element , i.e. , @xmath717 .
see figure [ figure_amalgam ] for an example .
( 300,100 ) ( 0,30 ) 2020 ( 30,40)@xmath718 ( 50,20 ) 530 ( 60,40)@xmath719 ( 80,20 ) 6040 ( 10,0)@xmath42 ( 45,0)@xmath54 ( 90,0)@xmath720 ( 80,60 ) ( 120,60 ) ( 100,40 ) ( 100,60 ) ( 100,80 ) ( 180,30 ) 2020 ( 210,40)@xmath721 ( 230,20 ) 530 ( 240,40)@xmath719 ( 260,20 ) 530 ( 190,0)@xmath42 ( 225,0)@xmath54 ( 270,0)@xmath722 let @xmath705 and @xmath706 be finite semilattices , and let @xmath723 and @xmath724 . then let @xmath725 be the mapping in @xmath726 defined by @xmath727 \big ) = [ f(x ) , g(y ) ] \ ] ] for every @xmath728 , where @xmath729 $ ] denotes the class of @xmath730 in @xmath715 .
since @xmath723 and @xmath724 the map @xmath725 is clearly well - defined .
note that for @xmath731 and @xmath732 the rules @xmath733 and @xmath734 apply .
with @xmath735 we denote the set of the automorphisms of a semilattice @xmath54 .
we consider in particular the case where @xmath706 is the semilattice @xmath736 where @xmath737 , for some @xmath738 ; that is , different elements are comparable only if one equals @xmath121 . in this case , @xmath735 consists of all bijective maps @xmath739 such that @xmath740 , and thus the group @xmath741 is isomorphic to the symmetric group @xmath742 .
any subgroup @xmath743 of @xmath741 acts in this sense faithfully on the set @xmath744 .
[ construction ] let @xmath705 be a semilattice and let @xmath745 be the semilattice , where @xmath746 , @xmath747 and @xmath748 . further let @xmath743 be a subgroup of @xmath741 with @xmath749 for every @xmath750 with @xmath531 , let @xmath751 , and let @xmath68 be a subsemiring of @xmath752 with @xmath753 if @xmath754 then @xmath755 and @xmath68 corresponds to a subsemiring @xmath367 of @xmath516 , which fulfills ( [ condition_f_ab ] ) and ( [ condition_f_ab_leq_f ] ) . if @xmath42 does not satisfy @xmath75 then @xmath68 possesses also a finite irreducible idempotent @xmath0-semimodule which does not satisfy @xmath75 , namely @xmath583 ( see proposition [ lemma_l_is_irr_semimod ] ) .
if @xmath756 then @xmath757 and @xmath68 belongs to a class of finite simple semirings with absorbing greatest element , which are also known .
these semirings have been presented in the case of commutative semirings in @xcite and for not necessarily commutative semirings in @xcite : let @xmath758 be a finite group and define @xmath759 . extend the multiplication of @xmath760 to @xmath761 by the rule @xmath762 for every @xmath763 and define the addition on @xmath761 by @xmath764 and @xmath765 for every @xmath766 with @xmath767 .
then @xmath768 is a finite simple additively idempotent semiring with absorbing greatest element and @xmath769 is a finite irreducible idempotent semimodule without property @xmath75 if @xmath770 .
construction [ construction ] is a combination of those two types of semirings . as shown in the next proposition , these semirings are also simple .
let everything as in construction [ construction ] .
then @xmath68 is a finite simple additively idempotent semiring with absorbing greatest element .
clearly , @xmath68 is a finite additively idempotent semiring .
its greatest element is @xmath771 , which is obviously absorbing .
let @xmath96 be a congruence on @xmath68 with @xmath772 , i.e. , there exist @xmath773 with @xmath774 and @xmath775 . by ( [ condition_a ] ) there exist @xmath776 , @xmath777 with @xmath778 and @xmath779 .
without loss of generality we can assume @xmath780 .
it follows that @xmath781 .
choose @xmath782 arbitrarily .
we will show that @xmath783 holds . from this
it follows that @xmath784 and therefore the simplicity
. again there exist @xmath785 , @xmath786 with @xmath787 . by ( [ condition_c ] )
there exists @xmath788 , @xmath514 and @xmath789 such that @xmath790 .
we have @xmath791 and thus @xmath792 for some @xmath793 . for all
@xmath380 it follows @xmath794 \le [ \lambda_1(x ) , \lambda_2(y)]$ ] , so that @xmath795 ; hence @xmath796 . because of @xmath781 it holds that @xmath797 and thus there exists @xmath380 with @xmath798 .
it follows that @xmath799 .
it also must hold that @xmath800 , i.e. , @xmath801 . since @xmath743 is a group there exists @xmath802 with @xmath803 .
we make a distinction of cases .
case 1 : @xmath804 .
it holds that @xmath805 and @xmath806 and because of @xmath807 it follows that @xmath808 .
case 2 : @xmath809 and @xmath810 .
it must hold that @xmath811 and it follows that @xmath812 , i.e. , @xmath813 . as in the previous case one can show the equality @xmath814 .
additionally it holds in this case that @xmath815 and we find again that @xmath783 .
case 3 : @xmath809 and @xmath816 .
there exists @xmath435 with @xmath817 .
it also holds that @xmath818 , i.e. , @xmath819 . consequently there exists @xmath820 with @xmath821 .
it follows that @xmath822 .
further it holds that @xmath823 and it holds again that @xmath783 .
case 4 : @xmath809 and @xmath824 . in this case
there exists @xmath577 with @xmath825 .
analogously to the previous case one can show that @xmath826 and @xmath827 holds and we find again @xmath783 .
let everything as in construction [ construction ] .
additionally , let @xmath828 , and let @xmath829 or let @xmath42 without property @xmath75
. then @xmath830 is a finite idempotent irreducible @xmath0-semimodule without property @xmath75 .
it is easy to see that @xmath831 is a finite idempotent @xmath0-semimodule without property @xmath75 .
further it is an @xmath0-nonidentity semimodule and it fulfills @xmath832 .
considering the action of the group @xmath743 on the set @xmath744 it follows from the conditions in construction [ construction ] that for every @xmath833 the orbit map @xmath834 , @xmath835 is injective ; now , since @xmath828 , this map is even bijective .
let @xmath836 be an @xmath0-subsemimodule of @xmath831 with @xmath837 , i.e. , there exist @xmath838\in m$ ] with @xmath838\ne a$ ] .
hence @xmath839 and @xmath840 .
choose @xmath841\in l\boxtimes k$ ] arbitrarily .
since the orbit map @xmath842 is bijective it follows that there exits @xmath843 with @xmath844 .
it follows that @xmath845)=[c , d]$ ] .
thus , @xmath846 = l\boxtimes k$ ] and @xmath831 is consequently sub - irreducible .
let @xmath96 be a semimodule congruence on @xmath831 with @xmath847 , i.e. , there exist @xmath838,[c , d]\in l\boxtimes k$ ] with @xmath838\sim[c , d]$ ] and @xmath838\ne[c , d]$ ] .
let @xmath848 , @xmath849 .
we will show that @xmath850\sim a $ ] holds . from this
it follows that @xmath851 , i.e. , @xmath831 is quotient - irreducible . if @xmath838 = a$ ] then @xmath841\ne a$ ] , i.e. , @xmath852 .
hence , there exists @xmath843 with @xmath853 and it follows that @xmath854 ) = [ e , f]$ ] .
the case @xmath841=a$ ] works analogously .
so from now on we can consider the case that @xmath838,[c , d]\ne a$ ] .
if @xmath855 then it holds that @xmath856 and it follows that @xmath838=[a , b]\vee [ a , b]\sim [ c , d]\vee [ a , b]=[a,1_{\mathbf{k}}]=a$ ] .
we find @xmath857 $ ] as before .
now consider the case @xmath858 .
there exists @xmath400 with @xmath859 and it follows that @xmath850 = ( f_{a , e}\boxtimes h)([a , b ] ) \sim ( f_{a , e}\boxtimes h)([c , d])=[1_{\mathbf{l}},h(d)]=a$ ] . the case @xmath860 works analogously .
hence @xmath831 is irreducible .
[ corollary_conjecture ] let everything as in construction [ construction ] .
additionally , let @xmath861 , and let @xmath829 or let @xmath42 be without property @xmath75 .
then @xmath68 is a finite simple additively idempotent semiring with absorbing greatest element , which possesses a finite idempotent irreducible semimodule without property @xmath75 .
we computed all finite simple additively idempotent semirings with cardinality at most @xmath862 . from these semirings ,
every finite simple additively idempotent semirings with absorbing greatest element , which possesses a finite idempotent irreducible semimodule not satisfying @xmath75 , is isomorphic to a semiring in corollary [ corollary_conjecture ] .
for this reason we have the following conjecture .
let @xmath68 be a finite simple additively idempotent semiring with absorbing greatest element , which possesses a finite idempotent irreducible semimodule without property @xmath75
. then @xmath68 is isomorphic to a semiring in corollary [ corollary_conjecture ] .
let @xmath863 be the total order with @xmath864 elements .
the following semiring , consisting of the mappings @xmath865 , is the unique finite simple additively idempotent semiring with right but not left absorbing greatest element , induced by @xmath42 : @xmath866 the following semirings are all finite simple additively idempotent semirings with right but not left absorbing greatest element , induced by @xmath867 : @xmath868 for space reasons we just show the addition and multiplication table for the semiring @xmath869 : @xmath870 the following semiring is the unique finite simple additively idempotent semiring with left but not right absorbing greatest element , induced by @xmath871 : @xmath872 the following semirings are all finite simple additively idempotent semirings with absorbing greatest element , induced by @xmath873 : @xmath874 @xmath875
the authors would like to thank the anonymous referee for many valuable suggestions which have increased the quality of this work considerably .
k.al-zoubi , t.kepka , p.nmec : _ quasitrivial semimodules i _ , acta univ . carolin .
phys . 49 ( 2008 ) , no . 1 , 316 r.el bashir , j.hurt , a.janak , t.kepka : _ simple commutative semirings _ , journal of algebra 236 ( 2001 ) , 277306 r.el bashir , t.kepka : _ congruence - simple semirings _ , semigroup forum 75 ( 2007 ) , 588608 g.birkhoff : _ lattice theory _ , 3rd ed . ,
providence , 1967 t.s.blyth , m.f.janowitz : _ residuation theory _ , pergamon press , oxford , 1972 j.howie : _ fundamentals of semigroup theory _ , oxford univ . press , oxford , 1995 j.jeek , t.kepka : _ the semiring of 1-preserving endomorphisms of a semilattice _ , czechoslovak mathematical journal 59 ( 134 ) , 2009 , 9991003 j.jeek , t.kepka , m.marti : _ the endomorphism semiring of a semilattice _ , semigroup forum 78 ( 2009 ) , 2126 s.s.mitchell , p.b.fenoglio : _ congruence - free commutative semirings _ , semigroup forum 37 ( 1988 ) , 7991 c.monico : _ on finite congruence - simple semirings _ , journal of algebra 271 ( 2004 ) , 846854 j.zumbrgel : _ classification of finite congruence - simple semirings with zero _ , journal of algebra and its applications 7(3 ) , 2008 , 363377 | since for the classification of finite ( congruence-)simple semirings it remains to classify the additively idempotent semirings , we progress on the characterization of finite simple additively idempotent semirings as semirings of join - morphisms of a semilattice .
we succeed in doing this for many cases , amongst others for every semiring of this kind with an additively neutral element . as a consequence
we complete the classification of finite simple semirings with an additively neutral element . to complete the classification of all finite simple semirings
it remains to classify some very specific semirings , which will be discussed here .
our results employ the theory of idempotent irreducible semimodules , which we develop further .
semirings , semimodules , semilattices , join - morphisms 16y60,06a12 |
first we consider the renormalization of the self energy of the order parameter field , arising from the diagrams shown in fig .
[ opself ] .
the one loop correction of the bosonic self energy due to its coupling with the gauge field ( see fig . [ opself](a ) ) reads @xmath139,\ ] ] where @xmath140 } = \int^{1}_0 dx \int \frac{d^d p}{(2 \pi)^d } \ : \frac{(2-x)^2 k^2+p^2}{\left [ p^2 + x(1-x)k^2 + x m^2 \right]^2}\ ] ] and @xmath141}= \int^{1}_0 dx \ : 2 ( 1-x ) \int \frac{d^d p}{(2 \pi)^d } \ : \frac{\left ( \left ( p - x k\right)^2 + 2 k \cdot p - 2 x k^2 \right)^2}{\left [ p^2 + x(1-x ) k^2+m^2\right]^3}\ ] ] after setting @xmath142 in both the integrals . performing some standard integrals
we arrive at @xmath143 therefore the correction to the bosonic self energy due to its coupling with the gauge field is @xmath144 besides the above renormalization , the bosonic self energy acquires additional correction from the interaction vertex ( @xmath145 ) , as shown in fig .
[ opself](b ) .
this contribution is @xmath146 apart from the above two contributions , additional renormalization of the bosonic self energy comes from its coupling with the massless dirac fermions , as shown in fig .
[ opself](c ) , yielding @xmath147 where @xmath148 .
these three diagrams together give the following renormaization conditions @xmath149 leading to @xmath150 in eq .
( 8) and mass renormalization in eq .
( 12 ) in the main text .
next we present the computation of the self energy correction of the gauge fields .
it arises from the diagrams shown in fig .
[ gaugefieldself ] . contribution from the diagram ( a ) is @xmath151 \ : \left [ ( q+k)^2 + m^2 \right ] } = \left ( 4 \ ; n_b \ ; e^2 \right ) \int^1_0 dx \int \frac{d^d q}{(2 \pi)^d } \frac{\frac{4}{d}\ ; q^2 \delta_{\mu \nu } + ( 1 - 2 x)^2 k_\mu k_\nu}{\left [ q^2 + x ( 1-x ) k^2 + m^2\right]^2},\ ] ] once we take @xmath152 . performing some standard integrals gives @xmath153 upon taking the contribution from the diagram in fig . [ gaugefieldself](c ) , the term proportional to @xmath57 in the last expression exactly cancels out and the gauge field remains transverse .
the non - trivial coupling of the gauge field with dirac fermions also gives a correction to the gauge field propagator , as shown in fig .
[ gaugefieldself](b ) .
contribution from this diagram reads as @xmath154 \ : i_{\rho \sigma } \;=\ ; e^2 4 n_f \ ; \left ( \delta_{\mu \rho } \delta_{\nu \sigma}- \delta_{\mu \nu } \delta_{\rho \sigma } + \delta_{\mu \sigma } \delta_{\nu \rho } \right ) \ : i_{\rho \sigma},\end{aligned}\ ] ] where @xmath155 ^ 2},\ ] ] once we take @xmath156 . a tedious , but otherwise straightforward calculation yields @xmath157 therefore the correction to the gauge propagator from the diagram in fig .
[ gaugefieldself](b ) is @xmath158 \ : \frac{n_d \ ; k^{-\epsilon}}{\epsilon } \;+\ ; { \cal o } ( 1).\ ] ] collecting the contributions from these three diagrams gives the renormalization condition for @xmath159 as @xmath160 leading to the announced result in the main text , eq .
let us now compute the fermionic self - energy corrections to the one loop order .
fermions coupling with the gauge field , shown in fig .
[ fermionself](a ) , leads to the following correction to its self - energy @xmath161 \\ & = & i \ ;
e^2 \int^1_0 dx \int \frac{d^d p}{(2 \pi)^d } \left [ ( 2-d ) \gamma_\rho \ : \frac{p_\rho + ( 1- x ) k_\rho}{\left [ p^2 + x ( 1-x ) k^2 \right]^2 } - ( 1-x ) \gamma_\mu \gamma_\rho \gamma_\nu \ ; \frac{(p - x k)_\mu ( p - x k)_\nu \left ( p+(1-x)k \right)_\rho}{\left[p^2+x(1-x)k^2 \right]^3 } \right],\end{aligned}\ ] ] where @xmath162 .
after performing the integrals we find a trivial contribution @xmath163 \frac{\slashed{k}}{\epsilon } \ ; k^{-\epsilon } \ ; \frac{1}{(4 \pi)^2 } + { \cal o}(1)\equiv 0 + { \cal o}(1),\ ] ] as @xmath164 .
however , the fermionic self energy acquires nontrivial correction due to its coupling with the order - parameter field , shown in fig .
[ fermionself](b ) , yielding @xmath165 \ : \frac{1}{(k - p)^2+m^2 } \;=\ ; g^2 \ ; \left ( \frac{1}{2 \epsilon } \right ) \ ; i \slashed{p } \ ; n_d m^{-\epsilon } + { \cal o } ( 1).\ ] ] therefore @xmath166 as shown in eq .
( 8) in the main part of the paper .
to the one loop order the boson - fermion vertex also gets renormalized .
its renormalization comes from the diagrams shown in fig [ bosfermvert ] .
let us first consider the contribution of the diagram ( a ) , renormalizing @xmath167 vertex @xmath168,\end{aligned}\ ] ] where @xmath169 ^ 3 } , \nonumber \\\end{aligned}\ ] ] and @xmath170_\rho \left [ q- y p+ xy k \right]_\sigma}{\left[q^2-\left((1-y)p + xy k\right)^2 + ( 1-y)p^2 + xy k^2 \right]^4 } \right ] , \nonumber \\\end{aligned}\ ] ] after shifting the momentum as @xmath171 in @xmath172 , @xmath173 , and @xmath174 . for our renormalization group calculation , since we use the minimal - subtraction scheme , we only need to keep the divergent pieces from these two integrals , which read @xmath175 } \nonumber \\ & = & \frac{n_d}{24 \ ; \epsilon } \:k^{-\epsilon } \ : \left ( \delta_{\mu \nu } \delta_{\rho \sigma } + \delta_{\mu \rho } \delta_{\nu \sigma } + \delta_{\mu \sigma } \delta_{\rho \nu}\right).\end{aligned}\ ] ] hence the renormalization of the boson - fermion vertex is @xmath176 + { \cal o}(1)= - p_+ \ ; \frac{3 \ ; e^2 } { \epsilon } \ ; n_d + { \cal o}(1 ) .
\label{vertexcorrection}\ ] ] the renormalization of the other boson - fermion vertex , proportional to @xmath177 , comes from the diagram ( b ) of fig .
[ bosfermvert ] , which gives an identical contribution as in eq .
[ [ vertexcorrection ] ] . in principle
, the boson - fermion vertex can also be renormalized from the diagram fig .
[ bosfermvert ] ( c ) and ( d ) .
they respectively renormalizes the @xmath167 and @xmath177 vertex .
however , it is sufficient to consider one of them , for example ( c ) .
its contribution reads as @xmath178,\end{aligned}\ ] ] where @xmath179 \;(p+k')^2 } \nonumber \\ & = & \int^1_0 2 y \ ; dy \ ; \int^1_0 \ ; dx \ ; \int \frac{d^d p}{(2 \pi)^d } \frac{\left ( p+(1+y)q - xyk \right)_\nu \ : \left(p-(1-y)q+(1- xy)k ' \right)_\alpha}{\left [ p^2 + \delta\right]^3},\end{aligned}\ ] ] where , @xmath180 and @xmath181 ^ 2 $ ] .
we shift the momentum @xmath182 , while writing the last equation .
the other integral @xmath183 reads as @xmath184 \;(q+k')^2 } \nonumber \\ & = & 6 \int^1_0 y^2 \ ; dy \int^1_0 ( 1-x ) \ ; dx \ : \frac{h_\mu(x , y , q , p , k ' ) h_\nu(x , y , q , p , k ' ) \left ( p+(1+y)q - xy k ' \right)_\mu \left ( p-(1-y)q-(xy-1 ) k ' \right)_\alpha}{\left [ p^2 + \delta \right]^4 } , \nonumber \\\end{aligned}\ ] ] after shifting the momentum @xmath185 as before , where @xmath186 .
we are however interested only in the divergent pieces of these two integrals which read as @xmath187 hence the renormalization of the @xmath167 vertex from the diagram ( c ) in fig .
[ bosfermvert ] reads as @xmath188 \ :
\frac{n_d}{\epsilon } + { \cal o } ( 1 ) = 0 + { \cal o } ( 1),\ ] ] as @xmath164 .
hence this diagram provides trivial renormalization of the boson - fermion vertex ( @xmath167 ) .
similarly , one can show that renormalization of the other boson - fermion vertex ( @xmath177 ) arising from the diagram fig .
[ bosfermvert ] ( d ) is also trivial .
renormalization of the yukawa vertex can also arise from boson - fermion interaction , as shown in fig .
[ bosfermvert ] ( e ) , ( f ) , ( g ) , ( h ) .
the combined contribution of these diagram is however @xmath189 these contributions yield the announced renormalization condition of the yukawa coupling ( g ) in the main text , namely @xmath190
finally we consider the renormalization of the @xmath191 vertex .
renormalization of this vertex from the yukawa coupling , shown in fig .
[ lambdvert](a ) , reads @xmath192 the renormalization of this vertex arising from the diagrams in fig .
[ lambdvert](b ) is @xmath193 moreover , the coupling of the order parameter field with the gauge field also provides a renormalization of the @xmath191 vertex , as shown in fig .
[ lambdvert](c ) .
its contribution reads @xmath194 \ : \left [ \delta_{\mu \nu } \;-\ ; \frac{q_\mu q_\nu}{q^2}\right ] \;=\ ; -32 \ ; e^4 \ ; \frac{3 n_d}{\epsilon}.\end{aligned}\ ] ] upon collecting contributions from all the diagrams in fig .
[ lambdvert ] , one gets the renormalization condition of the coupling constant @xmath70 , as in eq . (
11 ) in the main part of the paper . | the field theory of the semimetal - superconductor quantum phase transition for graphene and surface states of topological insulators is presented .
the lagrangian possesses the global @xmath0 symmetry , with the self - interacting complex bosonic order - parameter and the massless dirac fermions coupled through a yukawa term .
the same theory also governs the quantum critical behavior of graphene near the transition towards the bond - density - wave ( kekule ) insulator .
the local @xmath0 gauged version of the theory which describes the quantum semimetal - superconductor transition in the ultimate critical regime is also considered . due to
the yukawa coupling the transitions are found to be always continuous , both with and without the fluctuating gauge - field .
the critical behavior is addressed within the dimensional regularization near four space - time dimensions , and the calculation of various universal quantities , including critical exponents and the universal mass - ratio is reported .
dirac quasiparticles represent low - energy excitations in various low - dimensional condensed - matter systems , such as graphene and topological insulators ( tis ) . in graphene ,
pseudorelativistic dirac quasiparticles emerge from hopping of the electrons on the underlying honeycomb lattice,@xcite while on a surface of a ti , they result from an odd number of band inversions in the bulk of the system leading to a topologically non - trivial @xmath1 invariant.@xcite in both cases , the linearly dispersing dirac quasiparticles give rise to a semimetallic ground state , stable against weak electron - electron interactions.@xcite when the repulsive interactions are sufficiently strong , however , a plethora of insulating phases can in principle be realized in graphene.@xcite furthermore , dirac fermions in graphene can also condense into _ four _ different gapped superconducting states , if the net interaction acquires an attractive component.@xcite the simplest of them , which will be the subject of the present study , is the uniform , spin - singlet @xmath2-wave pairing , favored by a sufficiently strong on - site attractive interaction.@xcite two of the remaining pairing gaps are spatially inhomogeneous spin triplets , which break the translational symmetry of the honeycomb lattice into kekule patterns.@xcite they are favored by the sufficiently strong nearest - neighbor attraction .
finally , yet another triplet pairing with an @xmath3-wave symmetry can be stabilized by a strong second - neighbor attraction.@xcite on the other hand , due to their reduced number of fermionic components , the massless dirac fermions residing on the surface of tis can acquire a superconducting gap only by pairing into an @xmath2-wave superconducting state .
possible inhomogeneous @xcite and chiral @xmath4 superconducting states in doped graphene,@xcite as well as proposed realizations of majorana fermions in graphene and tis , @xcite make the study of superconducting instabilities of dirac fermions in low - dimensional condensed - matter systems theoretically and experimentally interesting and timely .
@xcite the _ bosonic _ order parameters ( ops ) , characterizing both the insulating and the superconducting states , are composite objects of dirac fermions , and may exhibit different symmetries .
besides the usual self - interaction , the ops here are also coupled to the massless dirac fermions via yukawa term.@xcite previously , we studied the ising and heisenberg universality classes of the transition into charge - density - wave and spin - density - wave insulators , respectively .
the cooper pairs are of course _ charged _ , and consequently the corresponding field theory possesses a global @xmath0 symmetry .
we therefore here develop the @xmath0- symmetric field - theoretical description of the quantum semimetal - superconductor transition in graphene and surfaces of tis .
all the coupling constants in this effective theory are _ marginal _ in @xmath5 space - time dimensions , which enables us to perform the @xmath6 expansion with @xmath7 to address its critical behavior .
we find that the semimetal - superconductor transition is always continuous , and compute the critical exponents associated with the transition ; in particular , the correlation - length exponent and the anomalous dimensions for both the op and the fermion fields are obtained .
besides describing the semimetal - superconductor quantum phase transition , the @xmath0 field theory should also pertain to the quantum phase transition from the semimetallic into an insulating state with the dynamically generated kekule mass , which breaks the translation symmetry of the lattice.@xcite motivated by this physical problem , as well as by the theoretical possibility of a fully lorentz invariant semimetal - superconducting transition in these dirac systems , we extend our theory to include a fluctuating @xmath0 gauge field , that would describe the coupling of the electromagnetic field to the bosonic op and the dirac fermions near the critical point.@xcite the lorentz - symmetric critical point describing the continuous transition in this theory is _ charged _ , and also turns out to be stable for any physical number of fermion flavors ( that is , for @xmath8 ) . to set up the problem
, we first consider the pairing of the the gapless excitations in graphene , around the two inequivalent dirac points , at @xmath9 , described by an eight - component dirac - nambu spinor @xmath10 $ ] , where @xmath11$].@xcite here , @xmath12 , with @xmath13 being the lattice constant .
@xmath14 is the three - momentum and @xmath15 , @xmath16 .
@xmath17 is the spin projection along the @xmath18-axis .
@xmath19 and @xmath20 are the grassmanian fields on two sublattices .
the free dirac lagrangian in this representation assumes relativistically invariant form @xmath21 , where @xmath22 , @xmath23 , @xmath24 with @xmath25 as the imaginary time and summation over the repeated indices assumed .
the @xmath26-matrices are defined as @xmath27 , @xmath28 , @xmath29 , and @xmath30 , where @xmath31 form the pauli basis for two - dimensional matrices , and we take @xmath32 , as usual .
the @xmath2-wave superconducting op reads @xmath33 with @xmath34 as the _ phase _ of the superconducting op .
the op anticommutes with the particle number operator @xmath35 and commutes with all the three generators of the spin rotations , @xmath36 , and hence represents a spin - singlet .
moreover , it is even under the exchange of the sublattices , or of the dirac points . in terms of nambu s ( particle - hole doubled ) spinor basis , @xmath37
, the lagrangian for the gapless surface states of tis also adopts the relativistic form @xmath38 , with the @xmath26-matrices defined as @xmath39 .
the @xmath2-wave superconducting op and the number operator for the surface states of tis assume identical form as for graphene , only without the @xmath40 in the first block . due to nambu s particle - hole doubling ,
the true number of fermionic degrees of freedom on the surface of ti is a _
quarter _ of the one in graphene , however . in this special case
the critical theory acquires the supersymmetry , and the one - loop @xmath41-expansion is known to be exact.@xcite next we wish to study the quantum phase transition from the semimetallic into the @xmath2-wave superconducting phase . since we want to formulate an @xmath42-expansion near _ four _ space - time dimensions , we need first to define a spinor basis in which the theory can be formally extended from the physical three , to four space - time dimensions .
we therefore rotate the spinor @xmath43 , where , in graphene , @xmath44 $ ] . after this unitary transformation ,
the @xmath2-wave op reads @xmath45 while leaving the relativistically invariant free dirac lagrangian , @xmath46 , unchanged .
the number operator is then @xmath47 .
similarly , in ti , the analogous transformation is performed by choosing the simpler @xmath48 $ ] . for generality ,
we consider the @xmath0 gauge theory for @xmath49 flavors of four - component dirac fermions coupled to the bosonic op with @xmath50 complex components via the yukawa coupling in the presence of a fluctuating gauge field , with the complete lagrangian @xmath51 .
the coupling of the fermions to the @xmath0 gauge field reads @xmath52 the matrix @xmath53 appearing in the minimal coupling is then the number operator , and @xmath54 is the @xmath0 charge .
the coupling of the op to the massless fermions has the yukawa form @xmath55.\ ] ] on the other hand , the dynamics of the op coupled to the @xmath0 gauge field can be described by the standard ginzburg - landau lagrangian @xmath56 where @xmath57 is the tuning parameter of the transition .
the @xmath0 gauge field is described by the usual maxwell lagrangian @xmath58 with @xmath59 .
we will use the transverse ( landau ) gauge @xmath60 , in which the general gauge - invariant op becomes local.@xcite the above theory is constructed to be invariant under the following local @xmath0 gauge transformation @xmath61 . without the gauge field @xmath62 , the above field theory for @xmath63 and @xmath64 also governs the critical behavior of graphene close to the transition to spin - singlet kekule insulator .
however , in the latter case , the dirac spinor needs to be redefined as @xmath65^\top$ ] , where @xmath66 $ ] , with @xmath67 as the spin projections , and with the frequency label suppressed.@xcite we have set the fermi velocity and the velocity of the bosonic excitations to be equal , since a weak anisotropy in the velocities is irrelevant.@xcite the local ( @xmath68 ) @xmath0 gauge theory describes the ultimate critical behavior at the superconducting transition , at which _ all _ the velocities in the theory are equal to the velocity of light . in graphene and on a surface of tis , however , such a fixed point is experimentally inaccessible , since the bare fermi velocities are @xmath69 m/s .
next , we proceed with the analysis of the @xmath0-symmetric yukawa field theory .
the couplings @xmath70 , @xmath54 and @xmath71 are all dimensionless in ( 3 + 1 ) space - time dimensions , suggesting the @xmath41-expansion about @xmath72 as a tool of choice for a study of the quantum critical behavior .
define then the action @xmath73 , where the renormalized lagrangian is @xmath74 the computation of the self - energy diagrams for the fermions , the order - parameter , and the gauge field using minimal - subtraction scheme then yields the renormalization constants to the one - loop order @xmath75 @xmath76 where @xmath7 and the dimensionless couplings @xmath77 are rescaled as @xmath78 , with@xmath79 .
( see supplementary material ) .
the computation of the vertex diagrams to the same order gives the following renormalization conditions for the coupling constants : @xmath80 @xmath81 the renormalization of the tuning parameter @xmath82 can be extracted from the self - energy diagrams of the op , leading to @xmath83 here , the couplings with subscript `` 0 '' are the bare couplings , and the ones without the subscript are the renormalized couplings , and @xmath84 is the renormalization scale .
dimensional regularization explicitly preserves gauge invariance of the theory implying @xmath85 , to _ any _ order .
@xcite in conjunction with this identity and eq .
( [ z - a ] ) , one can write the ( ultraviolet ) beta - function of the charge as @xmath86 the renormalization group flow of the remaining two couplings can be obtained from eqs .
( [ rc - g ] ) and ( [ rc - lambda ] ) @xmath87 the above beta - functions , besides the trivial , yield the following _ neutral _ ( @xmath88 ) fixed points : ( i ) the wilson - fisher fixed point at @xmath89 and @xmath90 ; ( ii ) the neutral gross - neveu fixed point : @xmath91 where @xmath92 , with @xmath93 . ignoring the gauge coupling @xmath54 for the moment , this fixed point is _ critical _ , and it controls the transition towards the @xmath2-wave superconducting state , or into the spin - singlet kekule insulator .
weak charge @xmath94 is , however , a relevant coupling at this critical point .
( iii ) the bicritical point in the @xmath95-plane : @xmath96 , located in the unphysical region ( @xmath97 ) of the @xmath98-interaction .
therefore , our one - loop results suggest that the semimetal - superconducting transition is of the second order in the absence of the fluctuating gauge field . the result is qualitatively similar to the insulating ising and heisenberg universality classes .
@xcite the correlation - length exponent ( @xmath99 ) can readily determined from eqs .
( [ z - phi - psi ] ) and ( [ corr - length - exp ] ) , yielding @xmath100 with @xmath101 and @xmath102 corresponding to the neutral gross - neveu critical point .
since the lorentz - symmetry breaking perturbations are irrelevant near the critical point@xcite the dynamical critical exponent is @xmath103 , and the fermi velocity ( @xmath104 ) is non - critical . near the neutral critical point both the order - parameter and the fermion fields acquire a non - trivial anomalous dimension which , respectively , reads @xmath105 the residue of the quasiparticle pole of the fermions @xmath106 , and they cease to exist as sharp excitations at the critical point
moreover , as the system approaches the critical point from the superconducting side both the mass of the superconducting op and the fermion mass vanish with the universal ratio @xmath107 in order to extract the critical exponents and amplitudes for graphene one needs to substitute @xmath63 , @xmath108 , while for surface states of tis one should use @xmath109 , @xmath110 . in the latter case ,
we obtain @xmath111 and @xmath112 in agreement with ref . .
in the fully gauged theory with @xmath68 , the _ charged _ wilson - fisher fixed points are at @xmath113 , where @xmath114 . on the other hand ,
previously discussed neutral gross - neveu fixed point is unstable in the charge direction , and a pair of charged fixed points is located at @xmath115}{2 x^2 y^2 w } \right ) \epsilon,\ ] ] where @xmath116 , \delta_2=-4 w x^2 y^2 \left [ 54 x^2 - 2n_f ( 13.5 + y)^2 \right]$ ] .
however , only the fixed point with @xmath117 is stable in the critical plane ( @xmath118 ) .
this fixed - point is therefore critical , and controls the behavior in the vicinity of the quantum phase transition in the full lorentz - invariant @xmath0 gauge theory .
furthermore , for any physical number of flavors this critical point describes the second - order phase transition , since the quantity @xmath119 is _ positive _ for any @xmath8 .
however , since all the velocities in this theory are set to be equal to the velocity of light , this critical point may be reached only in the deep infrared regime.@xcite the other fixed - point at @xmath120 lies in the unphysical region ( @xmath121 ) of the @xmath98-interaction for any @xmath49 . on the other hand ,
when @xmath122 , we obtain the standard one - loop result for the critical number of the complex components of the op above which the superconductor transition is of the second order , @xmath123 , and the transition is controlled by the charged wilson - fisher fixed point , with @xmath124.@xcite it is worth observing that without the yukawa interaction , @xmath125 reduces to @xmath126 if one takes only into account the coupling of fluctuating gauge fields with the massless dirac fermions.@xcite the yukawa coupling therefore appears to be crucial for the stabilization of criticality in the theory , and the suppression of the possible discontinuous transition , which occurs in related theories .
@xcite the superconducting coherence length ( @xmath127 ) diverges as @xmath128 , and the correlation length exponent ( @xmath99 ) can be computed readily from eq .
( [ correlationlength ] ) .
the boson and the fermion fields in the vicinity of this charged critical point acquire anomalous dimensions , which can be found from eq .
( [ anomalousdimension ] ) .
one can also compute the flow for the ginzburg - landau parameter @xmath129 characterizing the transition , reads @xmath130.\end{aligned}\ ] ] at the _ charged _ gross - neveu critical point , this flow equation has fixed points at @xmath131 and @xmath132 , for arbitrary @xmath50 and @xmath8 .
the residue of the quasi - particle pole vanishes at the charged critical point as @xmath133 .
the optical conductivity in the entire semi - metallic phase remains constant and universal , while it becomes infinite ( zero ) in the superconducting ( kekule ) phase .
right at the quantum critical point it is also expected to be universal , but different from the one in the semi - metallic phase.@xcite the universal conductivity at the gross - neveu ( neutral or charged ) critical point is also expected to be different from one found in a pure bosonic theory . @xcite the computation of its value is an interesting problem left for future research . to summarize , by employing a @xmath0-symmetric gross - neveu - yukawa theory , we here studied the zero - temperature semimetal - superconductor ( kekule insulator ) transition in graphene and surface of tis , and showed that it is continuous for any number of dirac flavors .
the full @xmath0 gauge theory exhibits a _
charged _ critical point also for an arbitrary number of dirac flavors , and may be relevant for the semimetal - superconducting transition in the deep infrared regime . b.r . wishes to acknowledge the support by national science foundation cooperative agreement no .
dmr-0654118 , the state of florida , and the u.s .
department of energy .
v. j. acknowledges the support of the netherlands organization for scientific research ( nwo ) .
i. f. h. is supported by the nserc of canada .
99 p. r. wallace , phys . rev . *
71 * , 622 ( 1947 ) .
g. w. semenoff , phys .
rev .
lett . * 53 * , 2449 ( 1984 ) .
l. fu and c. l. kane , phys .
rev .
lett .
* 98 * , 106803 ( 2007 ) .
i. f. herbut , phys .
rev .
lett . * 97 * , 146401 ( 2006 ) .
i. f. herbut , v. jurii , and b. roy , phys .
rev .
b * 79 * , 085116 ( 2009 ) .
s. ryu , c. mudry , c. y. hou and c. chamon , phys .
rev .
b * 80 * , 205319 ( 2009 ) .
b. roy and i. f. herbut , phys .
rev .
b * 82 * , 035429 ( 2010 ) .
e. zhao , a. paramekanti , phys .
rev .
lett . * 97 * , 230404 ( 2006 ) . c. honerkamp , phys .
rev .
lett . * 100 * , 146404 ( 2008 ) .
r. nandkishore , l. levitov , and a. chubukov , nat phys .
* 8 * , 158 ( 2012 ) .
l. fu and c. l. kane , phys .
rev .
lett . * 100 * , 096407 ( 2008 ) .
j. linder , t. tanaka , t. yokoyama , a. sudbo , and n. nagaosa , phys .
rev .
lett . * 104 * , 067001 ( 2010 ) .
p. hosur , p. ghaemi , r. k. s. mong , a. vishwanath , phys .
rev .
lett . , * 107 * , 097001 ( 2011 ) .
p. strack , s. takei , and w. metzner , phys .
rev .
b * 81 * , 125103 ( 2010 ) . j. r. williams , _ et .
al .
_ , phys .
rev .
lett . * 109 * , 056803 ( 2012 ) .
i. f. herbut , v .
jurii ' c , and o. vafek , phys .
rev .
b * 80 * , 075432 ( 2009 ) ; b. roy , phys .
rev .
b * 84 * , 113404 ( 2011 ) . c. -y .
hou , c. chamon , c. mudry , phys .
rev .
lett . * 98 * , 186809 ( 2007 ) .
i. herbut , _ a modern approach to critical phenomena _
( cambridge university press , cambridge , 2007 ) . i. f. herbut , phys .
rev .
lett . * 104 * , 066404 ( 2010 ) .
s .- s .
lee , phys .
rev .
b * 76 * , 075103 ( 2007 ) . here
the gamma matrices are defined as @xmath134 .
i. f. herbut and z. teanovi , phys .
rev .
lett .
* 76 * , 4588 ( 1996 ) ; _
ibid_. * 78 * , 980 ( 1997 ) .
m. vojta , y. zhang , and s. sachdev , phys .
rev .
lett .
* 85 * , 4940 ( 2000 ) .
h. kleinert and f. nogueira , phys .
rev .
b * 66 * , 012504 ( 2002 ) .
i. f. herbut , phys .
rev .
lett .
* 79 * , 3502 ( 1997 ) ; ibid . *
81 * , 3196 ( 1998 ) .
* supplementary material of `` quantum superconducting criticality in graphene and topological insulators '' * + bitan roy@xmath135 , vladimir jurii ' c@xmath136 , and igor f. herbut@xmath137 @xmath135 national high magnetic field laboratory , florida state university , tallahassee , florida 32306 , usa + @xmath136 instituut - lorentz for theoretical physics , universiteit leiden , p.o .
box 9506 , 2300 ra leiden , the netherlands + @xmath137department of physics , simon fraser university , burnaby , british columbia , canada v5a 1s6 we here present details of the renomalization group calculation to the one - loop order using the euclidean partition function .
@xcite for the ease of calculation , we here rewrite the coupling of the order parameter to the massless dirac fermions , shown in eq . ( 4 ) of the main text as @xmath55 \
: \equiv \ : g(\phi{\bar\psi}p_+\psi+\phi^*{\bar\psi}p_-\psi),\ ] ] where the projectors are @xmath138 . |
the notion of poissonian ensembles of markov loops ( loop soups ) was introduced by lawler and werner in @xcite in the context of two dimensional brownian motion ( it already appeared informally in @xcite ) : the loops of a brownian loop soup on a domain @xmath3 are the points of a poisson point process with intensity @xmath4 where @xmath5 is a positive real and @xmath6 is the brownian loop measure on @xmath7 .
loop clusters induced by a brownian loop soup were used to give a construction of the conformal loop ensembles ( cle ) in @xcite and @xcite .
they are defined as follows : _ two loops @xmath8 and @xmath9 are said to be in the same cluster if one can find a finite chain of loops @xmath10 such that @xmath11 , @xmath12 and for all @xmath13 , @xmath14_. poissonian ensembles of markov loops can also be defined on graphs . they were studied in details in @xcite and @xcite . in particular poissonian ensembles of loops on the integer lattice @xmath15 induced by simple ( nearest neighbor )
random walks give a discrete version of brownian loop soup ( see @xcite ) .
the aim of this paper is to study loop clusters induced by markov loop ensembles on graphs .
the facts presented here , which are built on the results presented in @xcite , are more elementary than in the brownian loop soup theory , but they point out that poissonian ensembles of markov loops and related partitions are of more general interest . the examples we develop show relations with several theories : coalescence , percolation and random graphs . in section 1 , we recall the different objects needed to define the markovian loop clusters on a graph and the coalescent process induced by the partition of vertices they define . in section 2 ,
we state general properties of the clusters and establish some formulae useful to study the semigroup of the coalescent process . in the last three sections we address several aspects of the loop clusters induced by a simple random walk killed at a constant rate @xmath16 on different graphs : * on @xmath17 , the loop cluster model reduces to a renewal process .
we establish a convergence result by rescaling space by @xmath18 , killing rate by @xmath19 and let @xmath19 converge to @xmath20 . *
the loop cluster model can be seen as a percolation model with two parameters @xmath5 and @xmath16 .
bernoulli percolation appears as a limit if @xmath5 and @xmath16 tend to @xmath21 so that @xmath22 converges to a positive real .
we show that a non - trivial percolation threshold occurs for the loop cluster model on the lattice @xmath23 with @xmath2 . * as a last example , we consider the complete graph . we give a simple construction of the coalescent process on the complete graph and we deduce a construction of a coalescent process on the interval @xmath24 $ ] .
letting the size of the graph increase to @xmath21 , we determine the asymptotic distribution of the coalescence time : it appears to be essentially the same as in the erds - rnyi random graph model ( see @xcite ) though the lack of independence makes the proof significantly more difficult .
we consider a finite or countable simple graph @xmath25 endowed with positive conductances @xmath26 and a positive measure @xmath27 ( called the _ killing measure _ ) .
we denote by @xmath28 the conductance of the edge between vertices @xmath29 and @xmath30 and set @xmath31 for every @xmath32 .
the conductances and the killing measure @xmath16 induce a sub - stochastic matrix @xmath33 : @xmath34 , @xmath35 .
we assume that @xmath33 is irreducible .
a discrete based loop @xmath8 of length @xmath36 on @xmath37 is defined as an element of @xmath38 ; it can be extended to an infinite periodic sequence .
two based loops of length @xmath39 are said equivalent if they only differ by a shift of their coefficients ( _ i.e. _ the based loops of length @xmath39 , @xmath40 is equivalent to the based loop of length @xmath39 @xmath41 for every @xmath42 ) .
a discrete loop is an equivalent class of based loops for this equivalent relation ( an example is drawn in figure [ figloop ] ) .
let @xmath43 ( resp .
@xmath44 ) denote the set of discrete loops ( resp .
discrete based loops ) of length at least 2 on @xmath45 .
we associate to each based loop @xmath46 of length @xmath47 the weight @xmath48 .
this defines a measure @xmath49 on @xmath50 which is invariant by the shift and therefore induces a measure @xmath6 on @xmath43 . if @xmath51 is a positive function on @xmath45 such that @xmath52 , a new set of conductances @xmath53 and a new killing measure @xmath54 can be defined as follows : @xmath55 for every @xmath56 and @xmath57(x)\lambda_x$ ] for every @xmath32 .
this modification corresponds to the doob s @xmath51-transform : the associated transition matrix @xmath58 verifies @xmath59 , @xmath35 .
it is a self - adjoint operator on @xmath60 since @xmath61 with @xmath62 the loop measure @xmath6 is invariant under doob s @xmath51-transform .
+ the loop measure can be defined without assuming that @xmath33 is a substochastic matrix
. a weaker condition would be to assume that @xmath33 is a positive and contractive matrix .
taking such matrices does not extend the set of loop measures .
indeed , perron - frobenius theorem states that if @xmath63 is a positive and irreducible matrix and if its spectral radius @xmath64 is smaller or equal to 1 , then there exists a positive function @xmath51 on @xmath45 such that @xmath65 .
the @xmath51-transform of @xmath63 is then a substochastic matrix .
let @xmath66 be a poisson point process with intensity @xmath67 defined on @xmath68 .
for @xmath69 , let @xmath70 denote the projection of the set @xmath71\times{\mathcal{dl}}(x))}$ ] on @xmath43 ; @xmath72 is an increasing family of loop sets which has stationary independent increments .
it coincides with the family of non - trivial discrete loop sets induced by the poisson point process of continuous - time loops defined in @xcite .
an edge @xmath73 is said to be _ open at time @xmath5 _ if @xmath74 is traversed by at least one loop of @xmath75 .
the set of open edges defines a subgraph @xmath76 with vertices @xmath45 .
the connected components of @xmath76 define a partition of @xmath45 denoted by @xmath77 .
the elements of the partition @xmath77 are the loop clusters defined by @xmath75 ( an example is drawn in figure [ figcluster ] ) .
this paper is devoted to the study of @xmath77 .
+ if @xmath78 is a subset of @xmath45 , let us define the @xmath70-neighborhood of @xmath78 : @xmath79 given any @xmath80 , set @xmath81 ( in the example drawn figure [ figcluster ] , @xmath82 and for every @xmath83 , @xmath84 for instance ) .
this defines an equivalence relation and the associated partition is @xmath77 .
positive integer powers of a based discrete loops is defined by concatenation : if @xmath85 is a based discrete loop of length @xmath39 then @xmath86 ^ 1=\ell$ ] , @xmath86^{2}$ ] is the discrete based loop @xmath87 and so on
. as the @xmath88-th power of equivalent based loops are also equivalent , powers of discrete loops are well - defined .
a discrete loop @xmath89 is said to be _ primitive _ if there is no integer @xmath90 and no discrete based loop @xmath8 such that @xmath89 is the equivalent class of @xmath86^{m}$ ] ; any discrete loop @xmath91 can be represented as a power of a unique primitive discrete loop denoted by @xmath92 .
let @xmath93 denote the set of primitive discrete loops @xmath45 of length at least 2 and let @xmath94 denote the set of primitive discrete loops defined by @xmath75 .
clearly @xmath77 depends only on @xmath94 .
[ primitivedistrib ] the probability distribution of @xmath95 is a product measure @xmath96 on @xmath97 defined by @xmath98 if @xmath99 are distinct primitive discrete loops , the sets @xmath100 are disjoint , hence the @xmath101 events @xmath102 for @xmath103 are independent .
therefore , the law of @xmath94 is a product measure on @xmath104 and for every @xmath105 , @xmath106^m)).\ ] ] as @xmath107^m)=\frac{1}{m}\mu(\eta)^m$ ] for every @xmath108 , we deduce that@xmath109 it follows from proposition [ primitivedistrib ] that the harris inequality ( and also the b - k inequality ) holds on increasing events ( see e.g. @xcite ) . in particular
let us say that _ a subset @xmath78 of @xmath45 is connected at time @xmath5 _ if it is contained in a cluster of @xmath77 . then by the harris inequality , @xmath110 in a similar way
, we say that _ a subgraph is open at time @xmath5 _ if all its edges are traversed by jumps of loops in @xmath75
. then the same inequality holds .
+ these inequalities can be extended to any number of increasing events .
in particular , * for every partition @xmath111 of @xmath45 into non - empty blocks , the probability that @xmath77 is a coarser and @xmath112 of @xmath45 , @xmath113 is said to be _ coarser _ than @xmath112 ( denoted by @xmath114 ) or @xmath112 is said to be _ finer _ than @xmath113 ( denoted by @xmath115 ) if every block of @xmath112 is a subset of a block of @xmath113 . ]
partition than @xmath112 satisfies : @xmath116 * if @xmath117 , then @xmath118 .
the evolution in @xmath5 of @xmath77 defines a partition - valued markov chain .
let @xmath112 be a partition of @xmath45 into non - empty blocks @xmath119 . from state @xmath112 ,
the only possible transitions are to a partition @xmath120 obtained by merging blocks indexed by some subset @xmath121 of @xmath122 to form one block @xmath123 and leaving all other blocks unchanged .
the transition rate from @xmath112 to @xmath120 is @xmath124 where @xmath125 is the set of @xmath126-tuples of elements of @xmath127 having at least one element that belongs to each block @xmath128 : @xmath129 if the graph @xmath37 is finite , these transition rates can be expressed with determinants of green s functions of subgraphs . in order to describe the formula ,
let us introduce some notations ; let @xmath130 denote the green s function of @xmath37 : @xmath131 where @xmath132 , @xmath122 is the identity matrix and @xmath133 is the conductance matrix . for every subset @xmath134 of vertices , set @xmath135 and let @xmath136 denote the green s function of the subgraph @xmath137 of @xmath37 : @xmath138 is the inverse of the matrix @xmath139 .
let us assume that @xmath25 is a finite graph .
+ for every partition @xmath140 of @xmath45 and every subset @xmath121 of @xmath122 with at least two elements : @xmath141 by the inclusion - exclusion principle , @xmath142 to conclude , we use that : * for a subset @xmath134 of vertices,@xmath143 * for every family of reals @xmath144 indexed by a finite subset @xmath121 with at least two elements , @xmath145 [ extranskn ] let us consider the complete graph @xmath146 with unit conductances and a uniform killing measure with intensity @xmath16 .
the transition matrix @xmath33 verifies : @xmath147 for every vertices @xmath29 and @xmath30 .
thus for every subset @xmath134 of vertices , @xmath148 ( the computation of this determinant is detailed in lemma [ detmatrix ] ) . using and ,
we obtain : @xmath149 in this section , we assume that the graph @xmath37 is finite . for a partition @xmath112 of @xmath45 ,
let @xmath150 denote the conditional probability @xmath151 .
the probability that @xmath77 is finer than a given partition of @xmath45 has a simple expression : [ semigrouppartition ] let us assume that the graph @xmath25 is finite .
let @xmath112 be a partition of @xmath45 into non - empty blocks @xmath119 .
for every partition @xmath152 of @xmath45 , @xmath153 let us assume that @xmath112 is coarser than @xmath152 .
the event ` _ _ @xmath77 is finer than @xmath112 _ _ ' means that every loop of @xmath154 is included in a block of @xmath112 .
therefore , @xmath155 since the set of loops in each block @xmath156 of @xmath112 at time @xmath5 defines independent poisson point processes . to conclude , we use that @xmath157 and @xmath158 an explicit formula for @xmath159 can be derived from lemma [ semigrouppartition ] .
let us first introduce some notations . for a partition @xmath112 ,
let @xmath160 denote the number of non - empty blocks of @xmath112 .
for a subset @xmath78 , let @xmath161 denote the restriction of @xmath112 to @xmath78 : @xmath161 is a partition of @xmath78 , the blocks of which are the intersection of the blocks of @xmath112 with @xmath78 .
let us consider a finite graph @xmath25 .
let @xmath152 and @xmath112 be two partitions of @xmath45 .
if @xmath112 has @xmath126 non - empty blocks denoted by @xmath162 then @xmath163 let us first assume that @xmath164 . to obtain equation , it is sufficient to prove the following identity : @xmath165 where @xmath166 denotes the set of partitions of @xmath45 with @xmath8 non - empty blocks
let us assume that @xmath77 is a partition with @xmath167 non - empty blocks . for @xmath168
, we can construct a partition coarser than @xmath77 with @xmath8 non - empty blocks by choosing how to merge some blocks of @xmath77 , that is by choosing a partition of @xmath169 with @xmath8 blocks .
therefore @xmath170 and the right - hand side of is equal to @xmath171 . by an identity on the stirling numbers of the second kind ( see for example @xcite equation ( 1.30 ) page 22 ) , this sum is equal to 1 if @xmath172 and @xmath20 if @xmath173 .
+ to prove for a partition @xmath112 with @xmath126 non - empty blocks @xmath174 , we consider @xmath77 as a sequence of partitions on @xmath175 respectively and apply : @xmath176 where @xmath177 .
therefore , @xmath178 let us consider the complete graph @xmath146 endowed with unit conductances and a uniform killing measure of intensity @xmath16 .
if @xmath112 is a partition of the set of vertices @xmath45 with @xmath126 blocks @xmath179 then @xmath180 and @xmath181 let us note that @xmath182 is the @xmath167-th moment @xmath183 of the random variable @xmath184 where @xmath185 denotes a gamma@xmath186-distributed random variable
. thus @xmath187 let @xmath188 denote the @xmath39-th cumulant of @xmath189 .
formula and the expression of cumulants in terms of moments ( see formula ( 1.30 ) in @xcite for example ) yield @xmath190 . for a subset @xmath7 of @xmath45 ,
let @xmath191 denote the loops of @xmath75 contained in @xmath7 and @xmath192 the partition of @xmath7 induced by @xmath191 .
in general @xmath192 is finer than the restriction of @xmath77 to @xmath7 but coincides with it if @xmath7 is a union of elements of @xmath77 .
+ for every partition @xmath112 of @xmath45 with @xmath126 non - empty blocks @xmath179 , the loop sets @xmath193 , , @xmath194 , @xmath195 are independent .
this yields the following equality : @xmath196 let us note also that if @xmath197 then @xmath198 ( see @xcite page 2 for a related result in schramm - loewner evolution ( sle ) context ) . the second term on the right - hand side equals @xmath199 if @xmath45 is finite , it can be written as @xmath200 the probability that @xmath201 is finer than a partition @xmath112 can also be expressed using exit distributions .
the formula obtained is easier to use than for graphs such as tori , trees
let us first introduce some notations .
for a subset @xmath7 of @xmath45 , let @xmath202 denote the inner boundary of @xmath7 : @xmath203 and let @xmath204 denote the exit distribution ( or poisson kernel ) from @xmath7 : for @xmath32 and @xmath205 , @xmath206 is the probability that a markov chain with transition matrix @xmath33 starting from @xmath29 exits from @xmath7 at @xmath30 .
[ semigroupexit ] let @xmath207 be a partition of @xmath45 .
let @xmath208 denote the union of the boundary points of the blocks of @xmath112 .
let @xmath209 denote the matrix indexed by @xmath210 defined by : @xmath211 if @xmath212 is finite then the probability for @xmath77 to be finer than @xmath112 is @xmath213 * first , let us assume that @xmath45 is finite .
let @xmath214 denote the product of the block diagonal matrix @xmath215 by the matrix @xmath216 .
we can rewrite the expression of @xmath217 given by lemma [ semigrouppartition ] as @xmath218 .
the restriction of @xmath214 to @xmath219 is the identity . the exit distribution from a subset @xmath7 verifies : @xmath220 for every @xmath221 and @xmath205 .
therefore , @xmath222 where @xmath223 let @xmath224 denote a markov chain with transition matrix @xmath33 .
the trace of @xmath224 on @xmath210 defines a markov chain denoted by @xmath225 on @xmath210 and thus a poisson point process @xmath226 on @xmath227 ( see @xcite chap .
7 ) : the discrete loops set at time @xmath5 is @xmath228 let @xmath229 be the partition of @xmath210 induced by @xmath230 .
the non - empty subsets @xmath231 , @xmath232 define a partition of @xmath212 denoted by @xmath233 . as @xmath234 is the event ` _ _ @xmath235 such that @xmath236 , @xmath237 , @xmath238 , @xmath239 is not crossed by a loop of @xmath70 at time @xmath5 _ _ ' , it only depends on the restriction of the loops on @xmath210 , hence @xmath240 .
+ for a subset @xmath134 of @xmath212 , let @xmath241 denote the exit distribution from @xmath134 for @xmath225 .
it follows from formula applied to @xmath229 that @xmath242 where @xmath243 to conclude , it remains to note that @xmath244 for every @xmath245 and @xmath246 .
* let us now assume that @xmath45 is a countable set .
let @xmath247 be an increasing sequence of finite subsets of @xmath45 such that @xmath248 .
as @xmath212 is assumed to be finite , there exists an integer @xmath249 such that @xmath212 is included in @xmath250 .
for @xmath251 , a loop in @xmath252 that passes through two different blocks of @xmath112 , passes through two different blocks of @xmath253 the restriction of @xmath112 to @xmath254 .
therefore , the probability that @xmath77 is finer than @xmath112 is : @xmath255 by the first part of the proof , @xmath256 .
it remains to note that the matrix @xmath257 coincides with @xmath209 for every @xmath251 to deduce that @xmath258 .
an edge is said to be _ closed _ at time @xmath5 if it is not crossed by a loop of @xmath154 .
more generally , a family of edges @xmath259 is said to be closed at time @xmath5 if every edge of @xmath259 is closed at time @xmath5 .
+ let us assume that @xmath37 is a finite graph . to compute the probability for a family of edges @xmath259 to be closed , we modify the conductances and the killing measure so that the conductance of every edge in @xmath259 is @xmath20 and the measure @xmath260 is unchanged : @xmath261 @xmath262 and @xmath263 @xmath264 .
let @xmath265 denote the green function associated with @xmath266 .
[ closedset ] let us assume that @xmath25 is a finite graph .
the probability for a family of edges @xmath259 to be closed at time @xmath5 is @xmath267 . for a loop @xmath8 and an edge @xmath74 ,
let @xmath268 denote the number of jumps of @xmath8 across @xmath74 .
set @xmath269 .
@xmath270 recall that @xmath271 . for @xmath272
$ ] , let @xmath273 denote a perturbation of @xmath33 defined by @xmath274_{x , y}=\left\{\begin{array}{ll}sp_{x , y } & \text{if}\ ; \{x , y\}\in e\\ p_{x , y } & \text{if}\ ; \{x , y\}\not\in e. \end{array}\right.\ ] ] similarly , @xmath275 . in particular , @xmath276 lemma [ closedset ] follows since @xmath277 is the transition matrix associated with @xmath278 .
we can deduce from lemmas [ closedset ] and [ semigrouppartition ] that if @xmath259 is a set of edges of a finite graph with extremities in different blocks of a partition @xmath140 then @xmath279
on the graph @xmath17 , the clusters @xmath77 are intervals between closed edges at time @xmath5 ( namely edges which are not crossed by any loop of @xmath75 ) .
the poisson loop sets induced by a simple random walk killed at constant rate @xmath16 have the following properties : let us consider the graph @xmath17 endowed with unit conductances and a uniform killing measure with intensity @xmath16 .
set @xmath280 [ proprenewal ] * the midpoints of the closed edges at time @xmath5 form a renewal process .
moreover , * * the probability that @xmath281 is closed at time @xmath5 is equal to @xmath282 for every @xmath283 ; * * given that @xmath284 is closed at time @xmath5 , the probability that @xmath281 is also closed is equal to @xmath285 for every @xmath286 . *
assume that @xmath2870 , 1[$ ] .
let @xmath288 denote the law of this renewal process i.e. the law of the distance between two consecutive closed edges at time @xmath5
. for @xmath289 let @xmath290 denote a sequence of independent random variables with distribution @xmath291 .
+ for every @xmath292 , as @xmath19 converges to @xmath20 , @xmath293}y_{{\varepsilon},i}}$ ] converges in law to the value at @xmath294 of a subordinator with potential density @xmath295 .
this subordinator is associated with the poisson covering ( cf .
@xcite , chapter 7 ) defined by the infinite measure on @xmath296 with density @xmath297 these covering intervals can be viewed as images of the `` loop soup '' of intensity @xmath5 associated with the brownain motion killed at constant rate @xmath16 . *
an edge @xmath298 is closed at time @xmath5 if and only if @xmath299 . the next closed edge is the first edge which is not crossed by any loop of @xmath300 , and previous closed edges are defined by @xmath301 which is independent of @xmath300 .
stationarity is obvious as @xmath302 is distributed like @xmath303 . for @xmath286 ,
let @xmath304 denote the probability that @xmath281 is closed at time @xmath5 .
by lemma [ semigroupexit ] , @xmath305 where @xmath206 denote the probability that a simple random walk killed at constant rate @xmath16 and starting at @xmath29 exits @xmath7 at point @xmath30 . the functions @xmath306 can be computed from the solutions of equation @xmath307 which are linear combinations of @xmath308 and @xmath309 .
+ we obtain @xmath310 for every @xmath311 . + for @xmath36 , let @xmath312 denote the probability that @xmath281 is closed at time @xmath5 given that @xmath284 is closed at time @xmath5 .
to compute @xmath312 , we consider a simple random walk on @xmath313 killed at rate @xmath16 and at point @xmath20 ; we denote it @xmath314 .
as @xmath315 is the poisson loop sets associated with @xmath314 , by lemma [ semigroupexit ] we obtain : @xmath316 where @xmath317 is the exit distribution from a subset @xmath7 of @xmath313 for @xmath314 : @xmath318 and @xmath319 for @xmath320 .
therefore @xmath321 * let @xmath322 denote the laplace transform of the function @xmath323 : @xmath324 let @xmath325 denote the laplace transform of @xmath288 .
we shall prove that for every @xmath326 , @xmath327 converges towards @xmath328 as @xmath329 tends to @xmath20 which yields ( ii ) .
+ as @xmath330 for every @xmath36 , we have the laplace transforms identity : @xmath331 .
therefore , it is sufficient to show that @xmath332 converges to @xmath333 as @xmath19 tends to @xmath20
. let @xmath334 and @xmath335 be two positive reals such that @xmath336 . for @xmath19 small enough , @xmath337 and @xmath338 .
for @xmath339 let @xmath340 be the function defined by @xmath341 for @xmath342 .
therefore , for @xmath19 small enough , @xmath343 for every @xmath286 . by lebesgue s dominated convergence theorem applied to the function @xmath344 defined by @xmath345
we obtain : @xmath346 ( for @xmath347 , @xmath344 is dominated by @xmath3480 , \frac{1}{2a}]}(x)+\frac{e^{-sx}}{(1-e^{-1/2})^{\alpha}}\operatorname{1\hspace{-.29em}i}_{]\frac{1}{2a},+\infty[}(x)$ ] ) .
this implies that for every @xmath326 @xmath349 and @xmath350 these inequalities hold for every @xmath336 .
therefore for every @xmath326 , @xmath332 converges to @xmath333 which ends the proof of ( ii ) . in the case of the simple random walk on @xmath313 killed at @xmath20 , a similar argument ( detailed in @xcite )
shows that : * for @xmath351 , the midpoints of the closed edges at time @xmath5 form a renewal process with holding times @xmath352 ; the generating function of @xmath353 is @xmath354 where li denotes the polylogarithm : @xmath355 , @xmath356 .
set @xmath357 for @xmath358 . as @xmath329 tends to @xmath20
, @xmath359 converges in law towards a stable subordinator @xmath360 with index @xmath361 . in the case of a finite interval
@xmath362 $ ] , we obtain a renewal process conditioned to jump at point @xmath363 . * for @xmath364
, there are only a finite number of clusters .
in particular , @xmath365 .
let @xmath366 be a family of coefficients in @xmath367 $ ] . in the bernoulli percolation model of parameter @xmath368 on the graph @xmath25 ,
every edge @xmath74 is , independently of each other , called ` open ' with probability @xmath369 and ` closed ' with probability @xmath370 .
vertices connected by open paths define a partition of @xmath45 denoted by @xmath371
. we can compare @xmath371 to the partition induced by the set of primitive discrete loops of length @xmath372 for a finite graph @xmath373 .
[ bernpercolation ] let us consider a finite graph @xmath373 .
for @xmath374 and @xmath375 , set @xmath376 . + the partition induced by the set of primitive discrete loops of length @xmath372 at time @xmath5 has the same law as @xmath377 .
let @xmath378 denote the set of primitive discrete loops of length @xmath372 at time @xmath5 .
the law @xmath378 for every @xmath5 is a product measure @xmath96 on @xmath379 : for every edge @xmath380 , let @xmath381 denote the class of the based loop @xmath382 .
the probability that a loop in @xmath75 has @xmath239 as support is @xmath383 and @xmath384 bernoulli percolation clusters on a graph appear to be a limiting case of partitions defined by loop clusters in which only two points loops contribute asymptotically .
let us consider a finite graph @xmath25 endowed with unit conductances and a uniform killing measure with intensity @xmath385 .
let @xmath386 be the partition induced by the poisson loop set on @xmath37 at time @xmath5 .
fix @xmath387 .
+ if @xmath16 and @xmath5 tend to @xmath21 such that @xmath388 converges to @xmath389 , then @xmath386 converges in law towards the bernoulli percolation of parameter @xmath390 . for a partition @xmath140 of @xmath45 ,
let @xmath391 denote the set of edges of @xmath392 linking different blocks of @xmath112 .
the law of the bernoulli percolation of parameter @xmath390 denoted by @xmath393 is characterized by the identities:@xmath394 to prove the convergence of @xmath395 , we apply lemma [ semigroupexit ] : @xmath396 where @xmath209 is defined as in lemma [ semigroupexit ] .
we note then that @xmath397 is equivalent to @xmath398 if @xmath239 belongs to @xmath391 and of order less or equal to @xmath399 otherwise . indeed , if @xmath400 and @xmath401 for @xmath236 , @xmath402 and @xmath403 for all @xmath404 .
+ a second - order taylor expansion shows that @xmath405 is equivalent to @xmath406 , with @xmath407 .
let us consider the poisson loop process induced by the simple random walk on @xmath408 , killed at a constant rate @xmath385 : @xmath409 for every @xmath410 and @xmath411 .
let @xmath412 denote the probability of percolation at time @xmath5 _ i.e. _ the probability of any fixed point to be connected to infinity by an open path at time @xmath5 .
the following proposition presents some properties of the function @xmath413 : [ probapercol ] let @xmath414 denote the critical probability for bond percolation on @xmath415 ( @xmath2 ) .
* @xmath412 is an increasing function of @xmath5 and a decreasing function of @xmath16 .
* @xmath416 for every @xmath374 and @xmath385 such that @xmath417 . * for any @xmath374 , @xmath412 vanishes for @xmath16 large enough .
* @xmath412 is an increasing function of @xmath5 since @xmath418 is increasing .
+ to show that @xmath412 is a decreasing function of @xmath16 , we use an independent thinning procedure .
let @xmath419 .
the corresponding measures on based loops satisfy : @xmath420 for @xmath421 , @xmath83 and @xmath422 with the convention @xmath423 . by erasing independently each based loop @xmath424 of length @xmath83 with probability @xmath425
, we obtain a discrete loop set having the same distribution as @xmath426 .
* by lemma [ bernpercolation ] , the partition induced by the set of primitive discrete loops of length 2 at time @xmath5 has the same law as the bernoulli percolation with parameter @xmath427 .
it follows from bernoulli percolation on @xmath415 that if @xmath428 then @xmath429 . * to prove that @xmath412 vanishes for @xmath16 large enough , we show that there exists a finite real @xmath430 such that any self - avoiding path @xmath431 of length @xmath432 is open at time @xmath5 with probability less than @xmath433 .
we can then conclude with the usual path - counting argument : for every @xmath434 , @xmath412 is bounded above by the probability that there exists an open self - avoiding path of length @xmath363 at time @xmath5 starting from the origin , hence @xmath435 let us first introduce some notations .
* * let @xmath436 consist of partitions of @xmath437 in which all blocks have at least two elements ( the number of blocks of such a partition @xmath112 is denoted by @xmath160 and blocks are denoted by @xmath438 , @xmath439 , ... ) .
* * for a vertex @xmath440 and a loop @xmath441 , let @xmath442 denote the number of times @xmath8 passes through @xmath440 : @xmath443 if @xmath8 is the class of the based loop @xmath444 .
+ let us consider a self - avoiding path of length @xmath432 denoted by @xmath445 and let @xmath446 denote the set of edges @xmath447 .
+ to be open , the edges of @xmath29 have to be covered by the edges of @xmath448 loops of @xmath75 . among these loops , those that cover at least one edge
@xmath449 can be used to define a partition @xmath450 as follows : let @xmath451 be a loop in @xmath70 covering edge @xmath452 .
the first block @xmath438 of @xmath112 consists of the indices of the endpoints of @xmath449 covered by @xmath451 . if @xmath453 , let @xmath167 be the smallest integer @xmath454 such that @xmath455 is not in @xmath438 and let @xmath456 be a loop covering @xmath457 .
the second block @xmath439 of @xmath112 is defined as the set of indices of endpoints of @xmath449 covered by @xmath456 that are not in @xmath438 , and so on until all elements of @xmath437 belong to a block written down ( an example is presented in figure [ path ] ) . therefore , @xmath458 + by campbell s formula ( equation ( 4.2 ) page 36 in @xcite ) , for every @xmath459 and for every positive functions @xmath460 @xmath461 therefore , @xmath462 for a finite set @xmath78 ,
let @xmath463 denote the set of permutations of elements of @xmath78 .
by proposition 5 page 20 in @xcite , for every @xmath83 and for every vertices @xmath464 , @xmath465 therefore , @xmath466 where @xmath467 the blocks of a partition @xmath468 in can be seen as the orbits of a permutation without fixed point . since a circular order on @xmath126 integers corresponds to @xmath126 different permutations of these integers , @xmath469 can be rewritten as follows : @xmath470 where * * @xmath471 denotes the number of cycles in a permutation @xmath113 , * * @xmath472 denotes the set of permutations of @xmath437 without fixed point ( such a permutation corresponds to a configuration without isolated points ) .
+ ( see lemma [ permanent ] in the appendix for details . )
the right - hand side of equality is nothing but @xmath473 defined in @xcite page 41 . to conclude we use that @xmath474 since the vertices @xmath455 are pairwise distinct . as @xmath475 for every @xmath476 , @xmath477 thus @xmath478 which ends the proof .
* it follows from proposition [ probapercol ] that for every @xmath374 , there is a finite value of @xmath16 , which we denote by @xmath479 , above which @xmath412 vanishes ; morever @xmath480 is an increasing function that converges to @xmath21 as @xmath481 .
* as the simple random walk on @xmath15 is recurrent , for @xmath482 the probability that a fixed edge is open at time @xmath374 converges to @xmath483 as @xmath16 tends to @xmath20 .
indeed , let us first note that @xmath484 for every @xmath485 ( this equality has been proven for a finite graph , @xmath486 is equal to @xmath487 ] , so we apply it to the restriction of the random walk to @xmath488 ^ 2 $ ] and let @xmath489 tend to @xmath21 ) .
therefore @xmath490 converges to @xmath483 as @xmath16 tends to @xmath20 .
+ fix @xmath485 and @xmath491 . by symmetry , @xmath492 since @xmath493 has independent stationary increments , @xmath494 therefore , @xmath495 for a fixed @xmath496 , let us choose @xmath39 such that @xmath497 .
there exists @xmath498 such that for @xmath499 , @xmath500 .
it follows from that @xmath501 .
this section is devoted to the study of poisson loop sets on the complete graph @xmath146 endowed with unit conductances and a uniform killing measure .
the set of vertices is identified with @xmath502 and the set of partitions of @xmath502 is denoted by @xmath503 .
the intensity of the killing measure is denoted by @xmath504 with @xmath505 , hence the coefficients of the transition matrix @xmath33 are : @xmath506 with @xmath507 for every @xmath508 .
as @xmath39 will vary , the loop set and the partition defined by the loop clusters at time @xmath5 will be denoted by @xmath509 and @xmath510 respectively . in the first part , @xmath39 is fixed ; we present another construction of the coalesecent process @xmath511 and use this construction to define a similar coalescent process on the interval @xmath367 $ ] . in the second part ,
we let @xmath39 tend to @xmath21 and describe the distribution of the first time when @xmath512 has no block of size one ( cover time ) and the first time when @xmath513 has only one block ( coalescent time ) .
[ transitionkn ] from state @xmath514 , the only possible transitions of @xmath515 are to partitions @xmath120 where @xmath516 is a subset of @xmath122 with @xmath517 elements .
its transition rate from @xmath112 to @xmath120 is equal to : @xmath518 where @xmath519 is the set of @xmath126-tuples of @xmath121 in which each element of @xmath121 appears .
let us recall the expression of @xmath520 obtained in example [ extranskn ] : @xmath521 by expanding the logarithm , we obtain : @xmath522 to establish @xmath523 which ends the proof of , we make use of the inclusion - exclusion principle .
let us consider the following random experiment : _ ` we choose uniformly at random @xmath126 points in the set @xmath524 ' _ and for @xmath525 , let @xmath526 denote the event _ ` at least one of the @xmath126 points falls into @xmath527'_. by the inclusion - exclusion principle , the coefficient @xmath528 divided by @xmath529 is equal to @xmath530 .
the probability @xmath530 can also be decomposed by introducing the events @xmath531 _ ` the @xmath454-th point falls into @xmath527 ' _ for every @xmath532 and @xmath525 : @xmath533 .
this corresponds to the right - hand side of divided by @xmath529 .
equation is obtained by rearranging the terms of the product @xmath534 .
formula of the transition rate has a simple interpretation : if we choose @xmath535 points uniformly at random in @xmath502 where @xmath535 has the logarithmic series distribution is defined as the measure + @xmath536 . ] with parameter @xmath537 , the probability that at least one point falls into each block @xmath538 and none outside of @xmath539 is equal to @xmath540 . from this remark , we derive a simpler construction of @xmath511 : * @xmath542 is a random variable with values in @xmath503 ; * let @xmath311 . given @xmath543 , * *
we choose an integer @xmath544 independent of @xmath545 , following the probability distribution @xmath546 * * we choose @xmath535 points @xmath547 uniformly at random on @xmath502 and independently of @xmath548 ; * * @xmath549 is obtained from @xmath550 by merging blocks of @xmath543 that contain at least one of the @xmath535 points @xmath551 . * the only possible transitions from state @xmath111 are to partitions @xmath120 where @xmath121 is a subset of @xmath122 with at least two elements and its transition rate from @xmath112 to @xmath120 is @xmath520 defined in proposition [ transitionkn ] . *
if @xmath112 is a partition of @xmath502 with @xmath126 non - empty blocks @xmath557 then @xmath558 the algorithm described in proposition [ construction ] can be adapted to define a coalescent process @xmath559 on the interval @xmath367 $ ] such that for every partition @xmath112 of @xmath560 $ ] and every @xmath556 , @xmath561 converges to @xmath562 if @xmath563 converges to @xmath289 and the partition @xmath564 of @xmath502 converges to @xmath112 as @xmath39 tends to @xmath21 . for @xmath565 ,
let @xmath566)$ ] denote the set of partitions of @xmath367 $ ] with @xmath126 blocks having a positive lebesgue measure and let @xmath567)$ ] denote @xmath568)$ ] .
for a partition @xmath569)$ ] and a subset @xmath121 of @xmath122 , let @xmath120 denote the partition obtained from @xmath112 by merging blocks @xmath570 for @xmath571 .
let @xmath19 be a positive real .
let us define a @xmath567)$]-valued sequence @xmath572 iteratively as follows : * @xmath542 is a random variable with values in @xmath567)$ ] ; * let @xmath573 .
given @xmath574 , * * we choose an integer @xmath544 independent of @xmath545 following the probability distribution @xmath575 * * we choose @xmath535 points @xmath547 uniformly at random in the interval @xmath367 $ ] and independently of @xmath576 ; * * @xmath577 is obtained from @xmath578 by merging blocks of @xmath574 that contain at least one of the @xmath535 points @xmath551 . *
its positive transition rates are from a partition @xmath582 to a partition @xmath120 with @xmath583 having at least two elements ; the value of such a transition rate is @xmath584 * if @xmath112 is a partition of @xmath367 $ ] with @xmath126 non - empty blocks @xmath585 then @xmath586 by construction @xmath572 is a @xmath567)$]-valued markov chain such that for every @xmath286 , @xmath587 or @xmath577 is a coarser partition of @xmath367 $ ] than @xmath574 obtained by merging several blocks of @xmath574 in one block .
+ let @xmath582 be a partition of @xmath367 $ ] and let @xmath121 be a subset of @xmath122 with at least two elements .
set @xmath588 .
given the event @xmath589 , @xmath577 is equal to @xmath120 if among the @xmath535 points @xmath551 uniformly distributed in the interval @xmath367 $ ] at least one point falls in every block @xmath570 for @xmath571 and none falls in @xmath367\setminus b_{j}$ ] .
therefore , @xmath590 is equal to @xmath591 given the event @xmath589 , @xmath577 is equal to @xmath112 if @xmath551 fall in one block of the partition @xmath112
. therefore , @xmath592 the generator of @xmath580 is then @xmath593 where @xmath63 is the transition matrix of @xmath572 .
+ let us now compute @xmath594 where @xmath112 is a partition of @xmath367 $ ] with @xmath126 non - empty blocks @xmath585 coarser than or equal to @xmath152 .
@xmath595 means that all points that are chosen simultaneously in the interval @xmath367 $ ] before time @xmath294 belong to the same block of @xmath112 .
therefore , by decomposing @xmath595 according to the value of @xmath596 and the number of points falling at the same time , we obtain that @xmath594 is equal to @xmath597 we expand the product @xmath598 and use that @xmath599 to obtain that @xmath594 is equal to @xmath600 the second sum is also equal to : @xmath601 therefore , @xmath594 is equal to @xmath602 the expression inside the braces is equal to @xmath603 this yields @xmath604 a vertex @xmath29 is said to be _ isolated at time @xmath5 _ if no loop @xmath605 passes through @xmath29 .
we call _ ` cover time ' _ the smallest @xmath5 such that @xmath146 has no isolated vertex at time @xmath5 and we denote it @xmath606
. + let us assume that the intensity of the killing measure is proportional to the size of the graph : @xmath607 with @xmath289 .
if we use the algorithm described in proposition [ construction ] to define @xmath608 , then we need to choose uniformly at random an average of @xmath609 points in @xmath502 .
if the points are drawn one by one and not by packs of random sizes , then the solution of the classical coupon collector s problem provides that the values of @xmath606 is around @xmath610 for large @xmath39 .
this analogy holds : [ nbisolated ] let us assume that the intensity of the killing measure on @xmath146 is @xmath607 with @xmath289 . for every @xmath611
, the number of isolated vertices at time @xmath612 converges in law to the poisson distribution with parameter @xmath613 .
in particular , @xmath614 converges in law to the gumbel distribution . ] . for each vertex @xmath29 ,
let @xmath615 denote the indicator of the event ` _ _
@xmath29 is isolated at time @xmath616 _ _ ' .
the number of isolated vertices at time @xmath616 is @xmath617 .
for every @xmath565 , let @xmath618 denote its @xmath126-th factorial moment ; @xmath619 is the sum of @xmath620 over all ( ordered ) @xmath126-tuples of distinct vertices @xmath621 . + as @xmath620 is the probability that no loop in @xmath622 intersects the subset @xmath623 , @xmath624 and @xmath625 therefore , @xmath626 . + in our setting , for every @xmath627 , @xmath628 and for every @xmath629 , @xmath630 thus @xmath631 as @xmath632 , we deduce that @xmath619 converges to @xmath633 for every @xmath565 .
the convergence to the poisson distribution with parameter @xmath613 follows from the theory of moments . * the distribution of the number of isolated vertices at a time @xmath5
can also be obtained by the inclusion - exclusion principle ; the probability that there exists @xmath101 isolated vertices at time @xmath5 is @xmath634 * in the statement of proposition [ nbisolated ] , we assume that the intensity of the killing measure @xmath635 is proportional to the number of vertices @xmath39 : @xmath607 .
in fact the same proof shows that proposition [ nbisolated ] also holds if we only assume that @xmath636 converges to @xmath21 and replace @xmath19 with @xmath637 in the statement .
let us note that if @xmath636 converges to a constant @xmath638 then @xmath619 converges to @xmath639 for every @xmath459 and @xmath640 tends to @xmath21 for any @xmath292 .
let @xmath641 denote the coalescence time , _
i.e. _ the first time at which @xmath510 has only one block .
the following theorem shows that the cover time and the coalescent time have the same asymptotic distribution : [ coaltime ] let us assume that @xmath607 for every @xmath36 with @xmath289 .
for every @xmath36 , set @xmath612 where @xmath642 is a fixed real .
+ for every @xmath311 , the probability that @xmath643 consists only of a component of size greater or equal to 2 and @xmath126 isolated points converges to @xmath644 as @xmath39 tends to @xmath21 .
+ in particular , @xmath645 converges in law to the gumbel distribution .
* theorem [ coaltime ] shows that @xmath646 is a sharp threshold function for the connectedness of the random graph process @xmath647 defined by the loop sets @xmath648 . *
such properties have been proven for the first time by erds and rnyi in @xcite for the random graph model they have introduced . to facilitate comparison ,
the following theorem states some of their results in a slightly different way from that used in @xcite : + let @xmath649 denote a random graph obtained by forming @xmath650 links between @xmath39 labelled vertices , each of the @xmath651 graphs being equally likely .
+ for every @xmath652 and every @xmath653 , the probability that @xmath654 contains a connected component of size @xmath655 and @xmath126 isolated points converges to @xmath656 as @xmath39 tends to @xmath21 .
* let us note that the ratio @xmath657 converges to @xmath658 as @xmath39 tends to @xmath21 .
the limit is an increasing function of @xmath19 that converges to @xmath483 as @xmath19 tends to @xmath21 .
therefore , it is not surprising to obtain properties similar to the erds - rnyi model for large @xmath19 .
+ in fact , we can show using the same proof that theorem [ coaltime ] holds if we replace @xmath19 with a positive sequence @xmath659 such that @xmath660 .
+ by contrast , if we let @xmath19 converge to @xmath20 as @xmath39 tends to @xmath21 , the loop sets can have very large loops : a result of y. chang in @xcite implies that if @xmath661 with @xmath662 then @xmath663 the rest of the section is devoted to the proof of theorem [ coaltime ] .
+ let @xmath664 denote the event ` _ _ @xmath643 consists only of a component of size greater or equal to 2 and @xmath126 isolated points _ _ ' , let @xmath665 be the number of isolated vertices in @xmath643 and let @xmath666 be the event ` _ _ @xmath643 has at least two components of size greater or equal to @xmath372 _ _ ' . for @xmath667 , @xmath668 by proposition
[ nbisolated ] , @xmath669 converges to @xmath670 .
we shall prove that @xmath671 converges to @xmath20 .
for a subset @xmath134 , let @xmath672 denote the probability that @xmath134 is a block at time @xmath616 .
@xmath673 as @xmath674 is independent of @xmath675 , we have @xmath676 for a sufficiently large set @xmath134 , we shall simply bound @xmath672 from above by @xmath677 for small @xmath134 , the probability @xmath678 is small . instead of computing it , we shall consider its upper bound by the probability that the total length of non - trivial loops on @xmath679 is greater or equal to @xmath680 , that is @xmath681 where @xmath682 denotes the number of crossings of a vertex @xmath29 by the set of non - trivial loops included in @xmath134 at time @xmath5 .
therefore for a small set @xmath134 , we shall use that @xmath683 we start by stating two lemmas : lemma [ intersblock ] provides an upper bound for the probability that no loop in @xmath643 intersects @xmath134 and @xmath684 .
lemma [ lengthloop ] gives an exponential inequality for @xmath681 based on markov s inequality .
to shorten the notations let @xmath685 denote the power set of @xmath502 and let @xmath686 consist of subsets of @xmath502 of cardinality @xmath101 .
[ intersblock ] for every nonempty proper subset @xmath134 of @xmath502 , let @xmath687 denote the event ` no loop in @xmath643 passes through both @xmath134 and @xmath684 ' where @xmath688 .
+ for every @xmath6890,1[$ ] there exists @xmath690 such that for every @xmath691 and for every @xmath692 , @xmath693 let @xmath694 and let @xmath134 be a subset of @xmath502 with @xmath101 elements . @xmath695 in our setting , @xmath696 for every subset @xmath7 .
therefore , @xmath697 . using that @xmath698
( see for example @xcite , formula 1.5 page 4 ) , we obtain : @xmath699 where @xmath700 is the function on @xmath7010 , 1/2]$ ] defined by : @xmath702 for @xmath7030 , 1/2]$ ] , with @xmath704 . + the function @xmath700 is of class @xmath705 in the interval @xmath7010,1/2]$ ] and the first two derivatives of @xmath700 are : a @xmath705 function in the interval @xmath7010,1/2]$ ] that verifies conditions @xmath715 , @xmath716 , @xmath717 has the following property : for every @xmath7180 , 1/2]$ ] such that @xmath719 is positive , the minimum of @xmath700 on the interval @xmath720 $ ] lies on @xmath721 or @xmath722 ( see lemma [ minfunct ] in the appendix ) .
in fact , @xmath723 converges to @xmath21 , whence for sufficiently large @xmath39 , @xmath724}f_n(x)=f_n(c)$ ] . to conclude ,
let us consider @xmath700 at point @xmath725 for some @xmath6890 , 1[$ ] .
@xmath726 therefore @xmath727 is equivalent to @xmath728 .
we deduce that for sufficiently large values of @xmath39 and for every @xmath729 $ ] , @xmath730 which ends the proof .
[ lengthloop ] let @xmath731 and @xmath732 be two positive reals such that @xmath733 .
assume that @xmath607 with @xmath289 and set @xmath612 with @xmath734 for every @xmath573 .
+ there exists @xmath735 such that for every @xmath736 , and @xmath737 with @xmath738 , @xmath739 the generating function of the vector @xmath742 has been computed for any finite graph @xmath743 in @xcite in the left - hand side term of the equation has to be replaced by @xmath5 . ] , page 37 : @xmath7440 , 1]^{\mathcal{v}},$ ] @xmath745_{x , y\in f}\big)\right)^{-\alpha}.\ ] ] in our setting , the generating function of @xmath746 satisfies : @xmath747 where @xmath748 and @xmath749 for @xmath750 .
the computation of that determinant is detailed in the appendix ( lemma [ detmatrix ] ) .
we obtain : @xmath751 equality can be extended to @xmath752 and the determinant is positive for every @xmath753 . by markov s inequality , for every @xmath754 , @xmath755 where @xmath756 .
the expression of the generating function of @xmath757 shows that the value of @xmath758 depends on @xmath134 only via @xmath680 ; we denote it @xmath759 .
the value of @xmath760 is @xmath761 for every @xmath754 such that @xmath762 . + set @xmath763 and @xmath764 for every integer @xmath765
. we shall use inequality for @xmath766 . before , let us note that @xmath767 converges to @xmath21 and @xmath768 converges to @xmath20 .
therefore , for sufficiently large values of @xmath39 and @xmath769 , the two conditions @xmath770 and @xmath771 are satisfied .
+ it remains to bound from below @xmath772 .
@xmath773 using the classical lower bounds @xmath774 along with @xmath775 , we obtain @xmath776 where , for sufficiently large values of @xmath39 therefore there exists a constant @xmath489 such that for sufficiently large values of @xmath39 and for every @xmath780 , @xmath781 in particular for @xmath782 , @xmath783 .
this shows that for every @xmath784 and sufficiently large values of @xmath39 , @xmath785 which ends the proof .
+ let us note that a study of @xmath786 shows that for every @xmath787 , and sufficiently large values of @xmath39 , @xmath786 has a maximum at a point which is equivalent to @xmath788 as @xmath39 tends to @xmath21 . we can now complete the proof of theorem [ coaltime ] .
set @xmath789 .
we shall prove that the upper bound of @xmath671 , @xmath790 converges to @xmath20 as @xmath39 tends to @xmath21 . by lemma [ intersblock ] , @xmath791
let us consider the case of subsets @xmath134 with @xmath792 elements . using the notations introduced in lemmas [ intersblock ] and [ lengthloop ]
, we have @xmath793 by lemma [ lengthloop ] , for every @xmath7940 , \delta[$ ] , sufficiently large @xmath39 and @xmath769 , @xmath795 let us study the function @xmath796 .
by computations we obtain that @xmath797 is equivalent to @xmath798 as @xmath39 tends to @xmath21 .
the study of @xmath700 made in the proof of lemma [ intersblock ] shows that , for sufficiently large values of @xmath39 , @xmath799 is greater than @xmath797 in @xmath800 $ ] .
let us introduce a real @xmath801 such that @xmath802 .
we have shown that for @xmath39 large enough and every @xmath769 , @xmath803 and thus @xmath804 by taking @xmath805 and @xmath806 for instance we obtain that for sufficiently large values of @xmath39 , @xmath807 .
this ends the proof of theorem [ coaltime ] .
[ permanent ] for a finite set @xmath259 , let @xmath808 denote the set of permutations of @xmath259 and let @xmath809 consist of permutations of @xmath259 without fixed point . for two integers @xmath810 and @xmath811 , let @xmath812 denote the partitions of @xmath813 with @xmath126 blocks , each of them having at least two elements .
+ for a @xmath814 matrix @xmath78 and a real @xmath5 , set @xmath815 where @xmath471 denotes the number of cycles in a permutation @xmath113 ( @xmath816 introduced in @xcite page 41 , can also be defined as the @xmath5-permanent of @xmath78 with the diagonal elements of the matrix set to zero ) . +
another expression of @xmath817 is @xmath818 for a finite set @xmath212 , let @xmath819 be the subset of @xmath820 which consists of cycles of length @xmath821 .
the decomposition of permutations into disjoint cycles entails that @xmath822 therefore , it remains to prove that for every @xmath83 @xmath823 starting from a permutation @xmath824 , we define a cycle of length @xmath126 , @xmath825 such that @xmath826 by setting @xmath827 conversely , starting from a cycle @xmath113 of length @xmath126 , we can construct exactly @xmath126 different permutations @xmath828 ,
, @xmath829 such that @xmath830 by setting @xmath831 this shows equality , completing the proof .
* there exists @xmath842a , b[$ ] such that @xmath833 is a decreasing function on @xmath701a , \rho[$ ] and an increasing function on @xmath843 $ ] ; * there exists two reals @xmath844 in @xmath701a , b[$ ] such that @xmath833 is a decreasing function on @xmath701a , \rho_{-}[$ ] , an increasing function on @xmath845 $ ] and a decreasing function on @xmath846 $ ] .
[ detmatrix ] let @xmath642 and @xmath841 be two reals and let @xmath39 be a positive integer .
let @xmath850 denote the identity matrix of size @xmath39 and let @xmath851 denote the @xmath852 matrix with all entries equal to one .
the determinant of the matrix @xmath853 is @xmath854 .
first , @xmath855 .
let @xmath47 . if we substract the column @xmath856 to the column @xmath39 and expand the determinant along the column @xmath39 , we obtain : + @xmath857 where @xmath858 denotes the matrix defined by @xmath859 by applying the same transformations to @xmath860 , we obtain @xmath861 the expression of @xmath862 follows by induction on @xmath39 . 10 jean bertoin .
subordinators : examples and applications . in _
lectures on probability theory and statistics ( saint - flour , 1997 ) _ , volume 1717 of _ lecture notes in math .
_ , pages 191 .
springer , berlin , 1999 . | we study the loop clusters induced by poissonian ensembles of markov loops on a finite or countable graph ( markov loops can be viewed as excursions of markov chains with a random starting point , up to re - rooting ) .
poissonian ensembles are seen as a poisson point process of loops indexed by ` time ' .
the evolution in time of the loop clusters defines a coalescent process on the vertices of the graph . after a description of some general properties of the coalescent process
, we address several aspects of the loop clusters defined by a simple random walk killed at a constant rate on three different graphs : the integer number line @xmath0 , the integer lattice @xmath1 with @xmath2 and the complete graph .
these examples show the relations between poissonian ensembles of markov loops and other models : renewal process , percolation and random graphs . |
interacting many - body systems in one spatial dimension , once considered theoretical playgrounds for the more involved `` physical '' three - dimensional systems , hold currently , and for quite some time , the status of physically relevant theories .
effective reduced - dimensional systems are indeed routinely prepared and studied experimentally .
prominent examples include ultracold atomic gases @xcite , organic conductors @xcite and nanotubes @xcite . from the theoretical point of view
, one - dimensional systems are very appealing , for a number of reasons .
firstly , there are a number of exactly solvable models in one spatial dimension .
examples include those models solved via the bethe ansatz technique @xcite , such as the spinless bose gas with dirac delta interactions ( lieb - liniger model ) @xcite and its spin-1/2 fermionic counterpart @xcite , and models with unbroken susy , which admit jastrow - product ground state wave functions , such as sutherland s model with inverse - square interactions and variations thereof @xcite .
more recently , strongly - coupled mesoscopic multicomponent systems in arbitrary external potentials have been the subject of intense investigation @xcite .
moreover , some numerical methods work especially well for one - dimensional systems . for instance
, the density matrix renormalization group ( dmrg ) @xcite is extremely powerful and flexible to use in most one - dimensional many - body problems on a lattice .
the low - energy properties of many of the above systems can be described by the universality class of luttinger liquids @xcite . in brief , the properties of these systems , such as the collective excitation spectrum , correlation functions , or density of states , at energies close to the fermi energy , exhibit universal behaviour that is described by the exact solution of tomonaga - luttinger s model @xcite via bosonization @xcite .
for instance , the density of states behaves as a power law , where the exponent and the proportionality constant are the only microscopic , non - universal details , around the fermi points @xcite . historically , there are two fundamentally different , yet complementary approaches for the introduction of luttinger liquid phenomenological models for one - dimensional quantum gases .
the first one , called constructive bosonisation @xcite , naturally arises for fermionic systems upon linearisation of the kinetic energy dispersion , as introduced by tomonaga @xcite . the second ,
more phenomenological , and also more general approach , is field - theoretical bosonisation , introduced by haldane @xcite . in this article , we consider one - dimensional spinless fermi gases in the fermionic luttinger liquid regime , that is , when their low - energy behaviour can be modelled by luttinger s model .
we will follow the constructive , rather than the field - theoretical route , which is most convenient for weakly - interacting fermions .
in particular , we study the microscopic origin of the interaction coupling constants in luttinger s model .
we first address the main forward - scattering channel in the microscopic theory , corresponding to the collision of two fermions with incident momenta close to the two opposite fermi points . in the very weak - coupling regime , these interbranch processes are well understood , and we show how to renormalise luttinger s interbranch interaction in favour of scattering data ( i.e. the two - body phase shift ) . from our analysis of the interbranch collision channel ,
we show how luttinger s model naturally arises from tomonaga s model upon renormalisation of the fermi points , which are to be considered as bare parameters of the theory . in this way
, we also explain the reason behind the interbranch coupling constant remaining `` unrenormalised '' in luttinger s model , a well known fact , as opposed to tomonaga s model , where the theory requires non - trivial renormalisation .
we then consider intrabranch interactions , i.e. interactions between particles moving in the same direction . in this case
, symmetry dictates that the intrabranch coupling constant must be identical to the interbranch one @xcite .
however , the current understanding of the microscopic origin of the intrabranch interaction in luttinger s model is quite unsatisfactory .
usually @xcite , its origin is said to be linked to the forward scattering of fermions near the same fermi point , i.e. in the same branch . even though this picture sounds reasonable , and easy to grasp physically
, it has some major issues . here , we carefully illustrate these issues , and find that the effective _
intra_branch interaction in luttinger s model corresponds , in the microscopic fermionic model , to an energy shift due to _ inter_branch interactions .
we finally apply our results in order to obtain the luttinger s parameter and the speed of excitations in the tonks - girardeau regime @xcite of the lieb - liniger gas , corresponding to the weak - coupling limit of its fermionic dual @xcite .
we consider a system of @xmath0 non - relativistic spinless fermions of mass @xmath1 interacting via pairwise potentials @xmath2 in one spatial dimension ( 1d ) .
the hamiltonian in the first quantisation has the form @xmath3 where @xmath4 is the position of the @xmath5-th particle . in the momentum representation ,
the two - body interaction is given by @xmath6 where we have defined @xmath7 .
with this notation , we can write down the second - quantised hamiltonian in the momentum representation , @xmath8 above , @xmath9 is the single - particle kinetic energy dispersion , @xmath10 is the system s size , @xmath11 ( @xmath12 ) annihilates ( creates ) a fermion with momentum @xmath13 , and therefore satisfies canonical anticommutation relations @xmath14 in hamiltonian ( [ hamiltonianmomentum ] ) we have assumed periodic boundary conditions , that is , @xmath15 with @xmath16 being integer numbers .
since we are working with fermions , it is convenient to introduce even and odd wave components in the interaction as @xmath17 .
the even ( odd ) wave component , acting only on bosonic ( fermionic ) wave functions , sometimes called `` s - wave '' ( `` p - wave '' ) , is given by the following symmetric ( antisymmetric ) combination @xmath18,\\ v_p(k , k')&=\frac{1}{2}\left[v(k , k')-v(k ,- k')\right].\label{pwave}\end{aligned}\ ] ] in the following , we will use @xmath19 or @xmath20 at our discretion , depending on whichever is more convenient in each situation .
in many applications , it is advantageous to construct effective interactions that mimic the exact scattering properties of a system at the relevant energy scales .
this can greatly simplify the calculations while keeping a correct description of collisional processes , and allow for universal ( i.e. independent of microscopic details ) predictions of many quantities of interest .
for instance , low - energy scattering is adequately described by means of `` pionless '' effective field theory ( eft ) @xcite , in which the interaction is taylor - expanded around @xmath21 as @xmath22 the lowest - order contribution to the p - wave potential above is given by @xmath23 where @xmath24 is the bare coupling constant that is to be renormalised in favour of low - energy scattering data . in this article
, we shall be interested in collisions of particles with relative momenta of @xmath25 , where @xmath26 is the fermi momentum .
the eft interaction ( [ v4 ] ) can be used if the particles have small relative momenta @xmath27 . if , instead , @xmath28 , the lowest order interaction is given by @xmath29.\label{vp1}\ ] ] the resulting low - energy interaction is sometimes referred to as `` constant '' @xcite .
this is a misconception that can lead to contradictions @xcite . if we wish to use eq .
( [ vp1 ] ) for all momenta , this must satisfy @xmath30 ( see eq .
( [ pwave ] ) ) .
this gives a model interaction of the form @xmath31\mathrm{sgn}(k)\mathrm{sgn}(k')\nonumber \\ & \equiv \tilde{g}_2\mathrm{sgn}(k)\mathrm{sgn}(k')\equiv v_p^{(2)}(k , k').\label{vp2}\end{aligned}\ ] ] the collisions due to the above interactions , and its corresponding diagram , are depicted in fig .
[ fig : g2-process ] . for the sake of completeness
, we write down the interaction @xmath32 in the second - quantised form @xmath33 where we have relabelled the momenta in the sum as @xmath34 , @xmath35 and @xmath36 .
all relevant scattering properties of the system can be extracted from the transition matrix ( @xmath37-matrix ) @xmath38 .
after separation of centre of mass and relative coordinates , the @xmath37-matrix for an odd - wave interaction satisfies the following lippmann - schwinger equation @xmath39 where @xmath40 is a complex ( relative ) energy , and @xmath41^{-1}$ ] is the non - interacting green s function , with @xmath42 the non - interacting relative hamiltonian .
the scattering states of a hamiltonian of the form @xmath43 correspond to an incident wave @xmath44 with non - interacting energy @xmath45 and a scattered wave @xmath46 .
this is given by @xmath47 above , the @xmath48 signs correspond to waves propagating forward ( @xmath49 ) and backwards ( @xmath50 ) in time , as dictated by time - dependent scattering theory @xcite .
the non - interacting green s function on the energy shell , @xmath51 , can be written as @xmath52,\ ] ] where @xmath53 denotes cauchy s principal value .
we now define the principal value @xmath37-matrix @xmath54 and the principal value non - interacting green s function @xmath55 as @xmath56,\\ \mathcal{t}(\epsilon(k))&=\frac{1}{2}\left[t(\epsilon(k)+i0^+)+t(\epsilon(k)-i0 ^ -)\right].\end{aligned}\ ] ] it is easy to see that the principal value @xmath37-matrix - matrix ) in the literature . ] for real positive energies @xmath57 satisfies the lippmann - schwinger equation @xmath58 the advantage of using the principal value @xmath37-matrix is that it avoids the use of imaginary and irrelevant except for bound states
parts , and it completely determines the scattering properties of a system of two identical particles .
we now solve the lippmann - schwinger equation , eq .
( [ pvt ] ) , for the interaction @xmath32 in eq .
( [ vp2 ] ) .
the @xmath37-matrix is readily calculated and reads @xmath59 after straightforward but tedious integrations , we find the position - represented fermionic wave functions in the relative coordinate @xmath60 for relative momentum @xmath13 and energy @xmath61 ( @xmath62 is the reduced mass ) , @xmath63\sin(kx)-\frac{m \tilde{g}_2}{\pi\hbar^2|k|}\mathrm{si}(kx)\cos(kx).\label{wf}\ ] ] above , @xmath64 ( @xmath65 ) is the cosine ( sine ) integral .
the only relevant information in the above wave function is the asymptotic ( @xmath66 ) behaviour . at long distances , the wave function in eq .
( [ wf ] ) behaves as @xmath67 we compare the above asymptotic wave function with the general asymptotic form of a fermionic scattering state in 1d , @xmath68 , and we immediately identify the relation between the coupling constant @xmath69 and the scattering phase shift @xmath70 , as @xmath71 we still need to decide upon a renormalisation condition , that is , the collisional energy at which we wish to renormalise the interaction , in which case the phase shift at that specific energy is exact . naturally , we should renormalise the interbranch scattering phase shift for particles near opposite fermi points @xmath72 , that is , with relative momentum @xmath26 .
this yields the following value for the bare coupling constant @xmath73 where we have defined the fermi velocity @xmath74 .
we now consider the approximate linearised spectrum around the fermi momenta @xmath75 , which is the starting point for building luttinger liquid theories for fermions .
we have included the @xmath76 superscript since we would like to treat , in general , the fermi points as _ bare _ quantities of the theory .
this might sound unusual , but notice that ( i ) upon _ any _ approximation all coupling constants of the theory may need to be renormalised and in this case it is the dispersion itself that has been expanded ( and truncated ) as a power series ; and ( ii ) renormalisation of the fermi points is _ always _ implicitly assumed in luttinger liquid phenomenology @xcite , as we shall see below .
the single - particle energy dispersion is well approximated by @xmath77 provided that @xmath78 . in eq .
( [ lineardispersion ] ) we have defined the ( bare ) fermi energy @xmath79 ^ 2/2m$ ] .
the many - fermion system with the linearised dispersion ( [ lineardispersion ] ) corresponds to tomonaga s model @xcite , where @xmath80 . in order to connect with the phenomenological model later
, we will call right ( left ) movers particles with @xmath81 ( @xmath82 ) . for interbranch collisions
we need the relative energy @xmath83 for a right and a left mover with relative momentum @xmath13 . this is given by @xmath84 for the interaction ( @xmath85 ) ,
the principal value @xmath37-matrix is readily calculated .
notice that the inverse @xmath37-matrix is ultraviolet ( uv ) divergent , and we therefore place a momentum cutoff @xmath86 .
the bare @xmath37-matrix takes the form @xmath87 , with @xmath88 we now have two options .
we can renormalise either the interaction coupling constant @xmath69 while letting the fermi momentum take its physical value ( @xmath89 ) , or we may renormalise the bare fermi momentum @xmath90 .
the renormalisation of the interaction coupling constant is straightforward .
applying the renormalisation condition @xmath91 , we obtain @xmath92 and then we have @xmath93 the inverse @xmath37-matrix is now well - defined at almost all energies .
it still retains an infrared ( ir ) divergence , which implies low - energy scattering is suppressed in this approximation .
however , this is unimportant , since we will only be concerned with high - energy ( of @xmath94 ) processes .
the other possibility , namely the renormalisation of the fermi points , is quite easy to implement , too .
notice that for @xmath69 to be finite ( and equal to @xmath95 ) , all we need is @xmath96 , with @xmath26 the renormalised ( i.e. physical ) fermi momentum . in this case , the inverse @xmath37-matrix , eq .
( [ invt ] ) , at energy @xmath97 , is regular and equal to @xmath69 when @xmath98 .
the @xmath37-matrix then takes the form @xmath99 written in the above form , the @xmath37-matrix appears to be ill - defined .
however , if we are interested in relative momenta around the ( bare ) fermi points , @xmath100 , with @xmath101 , we have a constant @xmath37-matrix @xmath102 as we take the cutoff to infinity . for momenta
far away from the ( bare ) fermi points , we get a logarithmically suppressed effective interaction instead .
all we need to do now is to obtain the renormalisation condition in terms of the phase shift at @xmath103 . in order for the renormalisation of the fermi points to make sense
, we need to assign the value of the phase shift at the physical value of @xmath26 to scattering at energy @xmath97 . from the asymptotic analysis of the scattering wave function we obtain the renormalisation condition @xmath104 which is exactly the same renormalisation prescription as for the quadratic dispersion , eq .
( [ g2quad ] ) , showing that @xmath69 is unrenormalised when passing from the quadratic to the linear approximation .
the renormalisation of the fermi points , while leaving the interaction coupling constant be renormalised without infinities , naturally leads to luttinger s model . to see this , shift the momentum of right ( left ) movers as @xmath105 ( @xmath106 ) , and
rescale the cutoff as @xmath107 .
then , the momenta of right and left movers both belong to the interval @xmath108 . the non - interacting hamiltonian becomes @xmath109,\label{luttingernonint}\ ] ] where @xmath110 and @xmath111 is the total number of particles , and where we have defined @xmath112 ( @xmath113 ) as the annihilation operator for a right ( left ) mover with momentum @xmath114 ( @xmath115 ) .
the above hamiltonian , eq .
( [ luttingernonint ] ) , exactly corresponds , after taking the cutoff @xmath116 , to the non - interacting luttinger hamiltonian .
this shows that , firstly , the renormalisation of the fermi points is a usual , implicit assumption in most treatments of luttinger liquid phenomenology and , secondly , that the interaction coupling constant remaining `` unrenormalized '' @xcite ( that is , renormalised without infinities ) , is a consequence of the renormalisation of the fermi points , and taking the limit @xmath98 _ before _ allowing the coupling constant to change at all .
the intrabranch coupling constant , typically denoted by @xmath117 @xcite , is very simple to obtain as a function of the interbranch coupling constant , since it must obey a natural physical constrain , corresponding to the conservation of charge ( particle number ) @xcite . in our notation
, it corresponds to setting @xmath118 .
however , the understanding of the microscopic origin of @xmath117 is far from satisfactory , and we study this in detail in this section .
it is customary to consider intrabranch interactions as two - particle scattering processes @xcite ( see fig .
[ fig - g4 ] ) , and then use a phenomenological coupling constant , namely @xmath117 , that simply adds to the fermi velocity and renormalises the bosonic excitation energies in the luttinger liquid @xcite .
there are , however , several problems associated with this collisional interpretation of intrabranch processes .
firstly , before linearising the single - particle dispersion , one can define two - fermion scattering states in , say , the right - moving branch ( we will consider right - moving fermions throughout this section unless otherwise stated ) . since the relevant scattering processes occur around the fermi point , the energy at which one should renormalise the interactions should be twice the fermi energy @xmath119 .
however , the fermionic @xmath37-matrix ( ergo the phase shift ) tends to zero quadratically with relative momentum , see eq .
( [ v4 ] ) .
this would mean that the renormalised fermi velocity would not be changed to zero - th order in the momentum of excitations , contrary to well - known results and phenomenology @xcite .
the second , even more severe problem , is that even if we wanted to renormalise the scattering properties in the linearised approximation to match the exact scattering properties ( obtained using the full quadratic dispersion ) around the fermi point , we would simply not be able to do so .
the fact , seemingly unnoticed so far , is that two particles ( fermions or bosons ) in the same branch of a linearised dispersion do not scatter .
we begin by considering this issue in the following subsection .
consider the lippmann - schwinger equation off the energy shell @xmath120 in the relative coordinate , @xmath121 where @xmath44 is regarded here as a fermionic ( antisymmetric ) state , that is @xmath122 .
as we go on - shell , @xmath123 , we see that the principal value @xmath37-matrix must vanish linearly with @xmath57
. therefore , we define @xmath124 for @xmath123 .
as we take the limit @xmath123 , we obtain the following integral equation @xmath125 the above equation can be written as an operator identity , which reads @xmath126 the simplest way to solve eq .
( [ opeq ] ) is by diagonalising @xmath19 .
we define a unitary operator @xmath127 can be quite degenerate , the operator @xmath127 is not necessarily unique . ]
that diagonalises @xmath19 , together with its associated orthogonal set of eigenfunctions @xmath128 and their corresponding eigenvalues @xmath129 , such that @xmath130 upon transformation of @xmath19 and @xmath131 with @xmath127 , eq . ( [ opeq ] ) reduces to @xmath132 for a strict finite - range potential @xmath19 ( which we can always consider as a good approximation ) , most of its eigenvalues will be degenerate and equal to zero .
we must therefore deal with eq .
( [ equalpha ] ) with care .
let us define the set of values of @xmath133 for which @xmath134 , that we define as @xmath135 , that is @xmath136 while we denote its complementary set by @xmath137 . if @xmath138 , then for each @xmath139 we must have either @xmath140 or @xmath141 .
if @xmath142 as well , then from eq .
( [ equalpha ] ) we see that @xmath143 is arbitrary , which is a natural consequence of degeneracy . in this case , we shall fix its value to a constant @xmath144 , i.e. @xmath145 for @xmath146 , we are obviously forced to set @xmath141 .
we now consider the case @xmath147 . from eq .
( [ equalpha ] ) we have that if @xmath148 , then @xmath143 is arbitrary and we set it to the constant @xmath144 as above if @xmath142 .
the case @xmath149 yields @xmath150 . )
, to raise the excitation energies of quasi - particles ( see text).,scaledwidth=100.0% ] in the discussion above we have completely solved the lippmann - schwinger equation . we still need to calculate the scattering states .
these have the form @xmath151 inserting a resolution of the identity , we find the total scattering state as @xmath152.\label{totscat}\ ] ] it is very illustrative to choose @xmath153 above . using the resolution of the identity @xmath154 in eq .
( [ totscat ] ) , and identifying @xmath155 where @xmath156 is the projector onto @xmath135 , we find that @xmath157 the above result has a clear interpretation : the scattering state of the system associated with the incident wave @xmath44 is nothing but the projection of the incident wave onto the zero potential eigenstates .
an immediate consequence of eq .
( [ projector ] ) is that our system ( or any other flat - banded system ) only supports scattering states if the potential has at least one vanishing eigenvalue .
the above analysis proves our claim that intrabranch interactions can lead to either ( i ) energy shifts if the potential has no zero eigenvalues or ( ii ) a superposition of non - interacting waves that avoid potential overlap if the interaction has vanishing eigenvalues .
it is clear from the discussion in the previous subsection that the @xmath117-processes do not correspond to particle - particle intrabranch scattering .
luttinger liquid theory , on the other hand , predicts that particle - hole excitation energies are shifted due to intrabranch interactions .
we will see in the following that studying the energy shifts of particle - hole excitations in the chiral ( right- and left - moving ) branches separately allows for some sort of renormalisation of the @xmath117 coupling constant in the weak _
inter_-branch interaction regime .
we will show , however , that microscopic fermion - fermion intrabranch interactions _ do not _ correspond to the intrabranch interactions appearing in luttinger liquid phenomenology .
these will be considered in the next subsection .
we begin by defining the chiral right - moving problem , with single - branch hamiltonian @xmath158 given by @xmath159 where @xmath160 hamiltonian ( [ hrr ] ) conserves total momentum .
since the fermi sea @xmath161 for @xmath162 right - moving fermions , given by @xmath163 is the unique state with lowest total momentum , it is an eigenstate of the interacting hamiltonian @xmath158 . for quite general repulsive interactions we can also assume it is the ground state of the chiral system . in order to obtain particle - hole excitation energies , it is convenient to rewrite the interaction ( [ vrr1 ] ) in the following way ( throwing away constant energy shifts ) @xmath164 above , we have defined the displacement operators @xmath165 and @xmath166 as @xmath167 since the first term on the r.h.s . of eq ( [ vrr2 ] )
is bosonizable , we shall first deal with the second term , which we denote by @xmath168 .
in particular , we are interested in the difference in energy between particle - hole excitations and the fermi sea due to this interaction term . we will denote particle - hole states by @xmath169 , with @xmath170 where obviously @xmath171 and @xmath172 .
the action of @xmath168 on a particle - hole state , throwing away a trivial constant term so that it gives the excitation energies , is given by @xmath173{|{\tilde{k}+q;\tilde{k}}\rangle},\ ] ] which means particle - hole states are eigenvectors of @xmath168 .
since we are interested in luttinger liquid phenomenology , we will consider only excitation energies of @xmath174 . to this order
, we can obtain these analytically by expanding @xmath175 as @xmath176 we obtain @xmath177q\right.\left.-\frac{1}{2\pi}\int_{\tilde{k}}^{\tilde{k}+q}dk v(\tilde{k}-k)\right\}{|{\tilde{k}+q;\tilde{k}}\rangle}+\mathcal{o}(q^2).\ ] ] in order to obtain the leading order in @xmath178 above , we set @xmath179 and finally arrive at @xmath180{|{\tilde{k}+q;\tilde{k}}\rangle}.\ ] ] above , we see that the excitation energy to linear order in @xmath178 due to the part @xmath168 of the intrabranch interaction is given by @xmath181 .
we now consider the first term on the r.h.s . of eq .
( [ vrr1 ] ) , which we will denote by @xmath182 , i.e. the part of the interaction that is straightforwardly bosonizable . for sufficiently small @xmath133 ( @xmath183 ) , the usual commutation relations for the displacement operators @xcite are valid , @xmath184=\frac{lq}{2\pi}\delta_{q , q'}.\ ] ] we define bosonic operators @xmath185 ( and equivalently for @xmath186 ) , and the term @xmath182 is rewritten as @xmath187 the eigenstates of @xmath182 including only single particle - hole pairs are given by @xmath188 and their respective eigenvalues are given by @xmath189 . in order to establish the excitation energies solely in terms of a quadratic bosonic hamiltonian , we need to make two approximations at this point .
one of them is the linearisation of the single - particle dispersion , so that the excitation kinetic energy of a particle - hole state @xmath169 , which is given by @xmath190 , is approximated by @xmath191 for all @xmath192 .
the second one corresponds to approximating the energy shift due to @xmath168 as @xmath193 for all @xmath192 , a reasonable approximation for small @xmath178 . in that approximation ,
@xmath194 are also eigenstates of @xmath168 , and we can finally write down the intrabranch interaction , approximately , as @xmath195a_{qr}^{\dagger}a_{qr}.\ ] ] the above intrabranch interaction would give a ( weak - coupling ) value of @xmath196 .
this is incorrect , since in the weak - coupling limit the result should be @xmath197 .
what has gone wrong here is that the intrabranch interaction in the _ effective luttinger s model _ does not correspond to an intrabranch process in the original fermionic system .
we will see how this is the case in the following subsection .
we consider now the weak - coupling limit of the intrabranch interaction .
the relevant state to be considered is a particle - hole excitation from the static fermi sea in the original fermionic system , @xmath198 where @xmath199 and @xmath200 , while @xmath201 we denote the relevant part of the interaction by @xmath202 , which is given by @xmath203c_k^{\dagger}c_k c_{\tilde{k}+q}^{\dagger}c_{\tilde{k}}.\ ] ] all other interaction processes give contributions of @xmath204 or higher .
we obtain @xmath205 { |{\tilde{k}+q,\tilde{k}}\rangle}.\ ] ] subtracting the energy shift corresponding to the fermi sea , i.e. that obtained by setting @xmath206 above , and after using @xmath207 , we obtain the energy shift @xmath208 to @xmath174 , with @xmath209 which is the correct answer we were looking for .
with the notations used in the previous sections , we can write down the effective luttinger liquid hamiltonian of the system in bosonized form , i.e. in terms of the boson operators for right ( left ) movers @xmath210 ( @xmath211 ) . this is given by @xmath212+\frac{g_2}{2\pi}\sum_{q>0}q\left[a_{qr}^{\dagger}a_{ql}^{\dagger}+a_{ql}a_{qr}\right],\label{luttingerbosonic}\ ] ] where @xmath213 , and where @xmath214 , see eq .
( [ g2r ] )
. the last statement might not be so easy to grasp , so we prove it as follows , using the notation of ref .
the original interbranch interaction is given by @xmath215 \nonumber \\ & \times c_{kr}^{\dagger}c_{k'+q , r}c_{k',l}^{\dagger } c_{k'-q , l } + ( \mathrm{r}\leftrightarrow \mathrm{l}).\label{rlthing}\end{aligned}\ ] ] to see how the p - wave interaction arises due to the exchange above , we define @xmath216 and @xmath217 , in which case @xmath218 and @xmath219 . therefore , we have @xmath220 . replacing now the p - wave potential by the `` constant '' interaction , we obtain @xmath221 then for the first term in eq .
( [ rlthing ] ) we have @xmath222 and , for small @xmath133 ( forward scattering ) , we have @xmath223 , and therefore @xmath224 , which leads to the @xmath225-interaction term in eq .
( [ luttingerbosonic ] ) .
hamiltonian ( [ luttingerbosonic ] ) corresponds to the notation in ref .
@xcite , and the luttinger parameter @xmath226 and the speed of excitations @xmath227 are given by @xmath228^{1/2},\label{luttingerk}\\ u&=v_f\left[\left(1+\frac{g_4}{2\pi \hbar v_f}\right)^2-\left(\frac{g_2}{2\pi\hbar v_f}\right)^2\right]^{1/2}.\label{luttingeru}\end{aligned}\ ] ] in the weak - coupling limit , we can expand the luttinger parameter and the speed of excitations as follows @xmath229 our simple renormalisation prescription gives the first order results above exactly .
we have to notice that the first order results in @xmath225 ( or equivalently @xmath117 ) , obtained from our renormalisation prescription are not perturbative , since they represent the exact two - body phase - shifts of the system and not their first born approximation @xmath230 .
we apply now the results of the previous sections to a particular example of an exactly solvable model , namely the ( bosonic ) lieb - liniger gas @xcite , or its fermionic dual , the cheon - shigehara model @xcite . for the sake of clarity , and to avoid confusion with other notations in the literature ,
we write down the lieb - liniger hamiltonian @xmath231 in first quantised form @xmath232 the dual fermion - fermion interaction in cheon - shigehara s model , however , has a formidable form in the position representation .
its momentum representation , on the other hand , is very simple , and does not appear to be widely known .
this is given in ref .
@xcite , and has the form @xmath233 which corresponds to a p - wave interaction ( see eq .
( [ pwave ] ) ) . from the interaction in eq .
( [ vcs ] ) , we obtain the following phase shift @xcite @xmath234 the dual interaction in lieb - liniger s model is obtained by equating the @xmath235 to the lieb - liniger phase shift @xcite , obtaining @xmath236 the above discussion immediately yields the value of @xmath225 ( and @xmath117 ) in terms of the lieb - liniger coupling constant @xmath237 , @xmath238 defining now the dimensionless lieb - liniger parameter @xmath239 , @xmath240 we obtain the desired first order result in the tonks - girardeau limit @xmath241 in agreement with the known result ( see , e.g. @xcite ) .
it is worth noting that in this model _ only _ , the first born approximation to @xmath225 is identical to the result using the phase shift .
this is due to the fact that lieb - liniger s model is a zero - range @xmath242-matrix theory , i.e. the interaction has zero range and the model is solvable via the bethe ansatz . in collision - theoretic language ,
the interaction potential in this model has already been renormalised to give the correct phase shift . to see this , we note that the following interaction has the same action on _ fermions _ as the cheon - shigehara interaction ( [ vcs ] ) , @xmath243 which implies @xmath244 , which is identical to the result obtained using the phase - shift , eq .
( [ g2ll ] ) .
we illustrate now the fact that the first born approximation to @xmath225 being identical to the non - perturbative result only occurs in lieb - liniger s model , where the interaction has been renormalised already for that purpose .
we consider a full - blown two - particle interaction of the following form in the position representation @xmath245 where @xmath60 is the relative coordinate , @xmath246 is the strength of the interaction , and @xmath247 controls its range . for two fermions , the phase shift is readily calculated from the relation @xmath248 where the fermionic scattering state @xmath249 satisfies the lippmann - schwinger equation @xmath250 the results for the phase shift ( which give the value of @xmath225 ) are shown in fig .
[ fig : phaseshift ] , where we have set @xmath251 . there , we observe that the weak - coupling result is only valid in a small window of weak interaction strength @xmath252
. however , the gas phase of the system can be weakly - coupled for strong attractive interactions , to the left of the shape resonance in fig .
[ fig : phaseshift ] .
in fact , the strongly - attractive side of lieb - liniger and cheon - shigehara models ( the super - tonks regime ) , corresponds to this region in the exponential potential s strength , while the weakly attractive regime in cheon - shigehara model corresponds to the weakly attractive regime of the exponential potential , and so does the strongly repulsive ( tonks regime ) lieb - liniger model .
we also note that there is no such thing as _ repulsive _ cheon - shigehara model . when @xmath253 ( @xmath254 ) we still have a strongly - attractive interaction due to renormalisation @xcite .
cheon - shigehara model is everywhere attractive : strongly attractive ( there are bound states ) for @xmath253 ( @xmath254 ) and weakly attractive ( there are no bound states ) for @xmath255 ( @xmath256 ) . for two fermions interacting via an exponential potential , eq .
( [ exppot ] ) , compared to its first born approximation ( dashed blue line ) , for @xmath251 . the vertical red line denotes the position of a shape resonance.,scaledwidth=100.0% ]
in this article we have studied how the phenomenological constants of luttinger liquid theories relate to the microscopic interactions of spin - polarised one - dimensional fermi gases .
we have shown that interbranch interactions rule both phenomenological inter- and intrabranch interactions in luttinger s model .
this has also been shown to arise in a completely natural way from the renormalisation of the fermi points in tomonaga s model , upon an appropriate redefinition of the wave numbers for right- and left - movers . as an important side result ,
our renormalisation programme explains , in very simple terms , the well - known fact of unrenormalisabilty of the interbranch coupling constant .
our theoretical framework can be generalised to luttinger liquids with spin , chiral luttinger liquids where the results of section [ section52 ] are directly applicable and helical luttinger liquids @xcite .
many of the concepts studied here can be relevant for the study of non - linear luttinger liquids @xcite , i.e. when the non - linearity in the kinetic energy dispersion becomes important .
m.v . and p. .
acknowledge support from epsrc grant no .
ep / j001392/1 , l.g.c . acknowledges support from the epsrc cm - dtc .
s. eggert , _ lecture notes from the a3 foresight summer school , korea _ in : `` theoretical survey of one dimensional wire systems '' , y. kuk , et al .
( eds . ) , ( sowha publishing , seoul , 2007 ) ; arxiv:0708.0003 . | we consider interacting one - dimensional , spinless fermi gases , whose low - energy properties are described by luttinger liquid theory .
we perform a systematic , in - depth analysis of the relation between the macroscopic , phenomenological parameters of luttinger liquid effective field theory , and the microscopic interactions of the fermi gas .
in particular , we begin by explaining how to model effective interactions in one dimension , which we then apply to the main forward scattering channel the interbranch collisions common to these systems .
we renormalise the corresponding interbranch phenomenological constants in favour of scattering phase shifts .
interestingly , our renormalisation procedure shows ( i ) how luttinger s model arises in a completely natural way and not as a convenient approximation from tomonaga s model , and ( ii ) the reasons behind the interbranch coupling constant remaining unrenormalised in luttinger s model .
we then consider the so - called intrabranch processes , whose phenomenological coupling constant is known to be fixed by charge conservation , but whose microscopic origin is not well understood .
we show that , contrary to general belief and common sense , the intrabranch interactions appearing in luttinger liquid theory do not correspond to an intrabranch scattering channel , nor an energy shift due to intrabranch interactions , in the microscopic theory . instead , they are due to interbranch processes .
we finally apply our results to a particular example of an exactly solvable model , namely the fermionic dual to the lieb - liniger model in the tonks - girardeau and super - tonks - girardeau regimes .
luttinger liquids , renormalisation , effective field theory , one - dimensional systems |
in spite of their highly idealized character , ` sticky hard sphere ' ( shs ) models continue to attract considerable attention in the recent literature on mesoscopic fluids .
the reasons are the following . on the one hand ,
these models are simple enough to allow analytic solution of the ornstein - zernike ( oz ) integral equation of the liquid state theory within some appropriate approximation . on the other hand , their potentials , albeit crude , exhibit realistic physical features ( repulsion plus surface attraction ) typical of the interactions between particles of many macromolecular fluids .
furthermore , the shs models have already been successfully employed in studies on colloidal suspensions , micelles , protein solutions , microemulsions , flocculation , and crystallization stell91,jamnik96,santos98,jamnik01 .
the first and most known model ( shs1 ) was proposed by baxter baxter68,baxter71a , baxter71b , barboy74 . its pair potential adds to a hard sphere ( hs ) repulsion an attractive square - well ( sw ) , which becomes infinitely deep and narrow according to a limiting procedure ( baxter s ` sticky limit ' ) that keeps the second virial coefficient finite baxter68 . for pure fluids
the starting potential is @xmath0 @xmath1 where @xmath2 is the distance between particles , @xmath3 the hs diameter , @xmath4 and @xmath5 denote the depth and width of the well , respectively , and @xmath6 ( @xmath7 being boltzmann s constant , and @xmath8 the temperature ) .
the dimensionless parameter @xmath9 is an _
arbitrary _ monotonically increasing function of @xmath10 baxter71b , and @xmath11 is a measure of ` stickiness ' between particles . in the limit
@xmath12 @xmath13 becomes the hs potential with diameter @xmath14 the logarithmic form of @xmath15 serves to generate a simple expression for the boltzmann factor , @xmath16 $ ] , i.e. @xmath17 = \left\ { \begin{array}{lll } 0 , & & 0<r<\sigma \\
\alpha r\ \left ( r-\sigma \right ) ^{-1 } , & & \sigma \leq r\leq r \\ 1 , & & r >
r% \end{array}% \right . , \label{i3}\ ] ] the ` sticky limit ' amounts to the limit @xmath18 , and the second virial coefficient , @xmath19 @xmath20 , $ ] becomes @xmath21 $ ] . the oz equation for the shs1 model admits analytical solution within the _ percus - yevick _ ( py ) _ approximation _ baxter68,perram75,barboy79 .
however , the difficulty in applying the multi - component shs1-py solution to polydisperse colloidal mixtures with a large number @xmath22 of components ( due to the presence of @xmath23 coupled quadratic equations @xcite to be solved numerically ) has led to two more recent attempts of finding an alternative shs model , which could be analytically tractable even in the general multi - component case brey87,mier89,ginoza96,tutschka98,tutschka00,tutschka01,tutschka02,gazzillo00,gazzillo02a , gazzillo02b , gazzillo03a , gazzillo03b , timoneda89,ginoza01,jamnik91,hoye93 .
we shall refer to them as shs2 @xcite and shs3 models @xcite . in shs2 the adhesive part of the potential
is generated from an attractive yukawa tail , whose amplitude and inverse range become infinite in the sticky limit , while their ratio remains constant .
the shs3 model has been proposed to replace shs2 , overcoming a recently discovered pathology while retaining some useful features of the previous model . as remarked in ref .
@xcite , the crucial difference between shs1 and shs2 lies in the attractive part of the interaction , since the area of the shs1 well vanishes as @xmath24 when the width @xmath25 goes to zero , while the corresponding integral remains finite in the shs2 case . for shs2 , the oz equation was analytically solved within the _ mean spherical approximation _ ( msa ) @xcite , and the multi - component shs2-msa solution is indeed much simpler than its shs1-py counterpart . on the other hand ,
baxter s shs1 model was never solved analytically within the msa or other closures different from the py one .
note that in the msa the main input for the direct correlation function is the potential sw1 itself , which has , in the sticky limit , a weaker ( logarithmic ) divergence than the boltzmann factor .
thus it would be interesting to investigate other approximations which consider , more correctly , the boltzmann factor rather than the potential , in a way similar to the py case .
motivated by this scenario , in this paper we would like to revisit baxter s model within a new framework . we consider a general class of closures which , as the py , hinge on the vanishing of the direct correlation function beyond a finite range @xmath26 .
each component of the class is identified by a different approximation to the cavity function in the intermediate region @xmath27 .
we shall show that this difference survives even in the limit @xmath28 and hence it gives rise to different solutions , each of them being characterized by a different expression for a single parameter @xmath29 . in this class
we first consider a subset of closures which is based on density expansion of the cavity function .
all these closures yield a simpler solution than the py one , and yet they provide a satisfactory description on regimes frequently encountered in experiments on colloids and globular proteins . the solutions corresponding to these closures could be straightforwardly extended to multi - component fluids , unlike the py case . in addition , we identify a second subclass of closures as a possible route to improve over the py approximation . finally , we provide a careful discussion on the equation of state derived from this class of solutions . in this process , we critically discuss the correct way of taking the ` sticky limit ' and its influence on the predicted structural and termodynamic properties of the model .
a short - cut , but not fully equivalent , procedure may in fact generate an incorrect result for the virial equation of state ( eos ) .
we point out that pitfalls of this sort have already occurred in the case of shs1 mixtures @xcite .
such a problem should be clearly borne in mind when investigating thermodynamic properties of more complex fluid models which include sticky interaction terms in the pair potential .
for a one - component fluid of @xmath30 molecules in a volume @xmath31 with spherically symmetric interactions , the oz equation is the convolution relationship @xmath32 which defines the ` direct correlation function ' ( dcf ) @xmath33 in terms of the ` total correlation function ' @xmath34 , deviation of the ` radial distribution function ' ( rdf ) @xmath35 from its ideal gas value of @xmath36 , i.e. @xmath37 . here , the asterisk represents three - dimensional convolution , i.e. @xmath38 @xmath39 .
moreover , @xmath2 refers to the distance between particles , @xmath40 is the number density , and the convolution term @xmath41 is the ` indirect correlation function ' , continuous with its first three derivatives everywhere , even when the potential has a hard - core part . the oz equation can be cast into the baxter form @xcite@xmath42where the ` factor correlation function ' @xmath43 is an auxiliary function , and the prime denotes differentiation with respect to @xmath2 .
solving the baxter equations is tantamount to determining @xmath43 , from which @xmath44 and @xmath45 can be easily calculated . to solve the oz equation , some approximate ` closure ' relating @xmath44 , @xmath45 and the potential @xmath46 must be added . for hard - core fluids
, whose @xmath46 may be written as the sum of a hs contribution and a tail , i.e. , @xmath47 ( with @xmath48 for @xmath49 and @xmath50 for @xmath51 an approximation is required only for @xmath52 in fact , in this case both @xmath33 and @xmath35 are exactly defined inside the core , i.e. , @xmath53 , \qquad 0<r<\sigma .
\label{clos0e}\end{aligned}\ ] ] let us consider some known approximations to the exact dcf : \i ) the _ mean spherical approximation _
( msa ) , @xmath54 \ii ) a _ modified mean spherical approximation _ ( mmsa ) gazzillo03b , huang84,pini02 @xmath55 where @xmath56 is the mayer function .
\iii ) the _ percus - yevick _ ( py ) closure , @xmath57 , \qquad r\geq 0 . \label{clos3}\ ] ] it is well known that the msa assumption is rather accurate at large @xmath2 values , but not in the region just outside the hard core . for short - ranged potentials ,
one thus expects that the msa performance could be improved by using the mmsa which replaces @xmath58 with @xmath59 .
moreover , the mmsa yields the correct zero - order term in the density expansion of the dcf , since @xmath60 when @xmath61 .
note that , when applied to baxter s sw1 potential , all these closures imply that @xmath33 vanishes beyond the range of the potential . in the present paper we will thus focus , for this model , on the class of _ all _ closures which can be written as@xmath62 , & \qquad 0<r<\sigma , \\
c_{{\rm shrink}}\left ( r\right ) , & \qquad \sigma < r < r , \\ 0 , & \qquad \text{\ } r > r.%
\end{array}% \right . \label{clos0f}\]]we shall refer to this expression as _ generalized py _ ( gpy ) _ approximation .
_ we remark that in eq .
( @xmath63 ) the novelty lies on the fact that it does not univocally define a single closure , since different approximations can be obtained by different definitions of @xmath64 , the dcf in the shrinking interval ( @xmath65 ) . as it turns out ,
these differences persist _ even in the sticky limit _
@xmath66 !
strictly speaking , all closures of the gpy class , when different from the py one , are _ mixed closures _ , being a combination of the py approximation for @xmath67 with a different approximation to @xmath33 for @xmath68 .
for instance , @xmath69 yields a msa / py closure , while one recovers the py approximation ( py / py ) when @xmath70 $ ] .
we recall that combinations of closures , such as hypernetted chain - mean spherical ( hnc / msa ) and percus - yevick - mean spherical ( py / msa ) approximations have already been employed in studies on electrolytes with sticky interactions rasaiah85,herrera91 .
let us apply the gpy approximation to the model defined by baxter s sw1 potential .
different approximations to @xmath64 generate , of course , different solutions ( which we are not able to write explicitly in analytic terms ) .
it is however surprising that , as shown in appendix a , the differences among all these solutions drastically reduce in the sticky limit all being included in one common parameter , @xmath71 , while the @xmath2 dependence becomes identical .
as a consequence of this degeneracy , the oz equation for the shs1 model admits , for _ all _ members of the gpy class of ( mixed ) closures , the following general _ analytic _ solution @xmath72 \theta \left ( \sigma -r\right ) , \qquad \text { for \ } r\geq 0 , \label{so1}\]]@xmath73@xmath74where @xmath75 denotes the cavity function at contact and , for simplicity , the subscript shs1-gpy has been omitted in @xmath76 , @xmath77 , and @xmath29 .
the gpy solution includes and extends the shs1-py expression found by baxter @xcite , and formally resembles the msa one for the shs2 model ginoza96 .
in fact , as regards the dependence on @xmath2 , the functional form is identical in both the mentioned cases .
the distinguishing feature of each element of the gpy class lies , however , in the parameter dependence of @xmath78 , that mantains a ` memory ' of the situation before the sticky limit is taken , depending upon the closure chosen for @xmath64 ( and upon the model ) @xcite .
note that the sticky limits of @xmath79 and @xmath80 are in general different , since different approximations to @xmath81 may be used for @xmath68 and @xmath67 , respectively .
the function @xmath82 is discontinuous at @xmath83 since @xmath84 and this jump leads to @xmath85 \theta \left ( \sigma -r\right ) -q_{\sigma } \sigma ^{2}\delta ( r-\sigma ) $ ] .
different approximations correspond to different expressions for @xmath86 .
it is also of some interest to write down an explicit expression for the dcf inside the core , which can be derived by using eq .
( @xmath87 ) , and reads@xmath88 \frac{r}{% \sigma } -\frac{\eta } { 2}a^{2}\left ( \frac{r}{\sigma } \right ) ^{3}\right\ } \theta ( \sigma -r ) \nonumber \\ & & + \ q_{\sigma } \ \sigma \delta ( r-\sigma ) ,
\qquad \qquad 0\leq r<\sigma .
\label{so4}\end{aligned}\]]each solution @xmath89 of the gpy class is fully characterized by its approximation to @xmath90 . in the following we thus discuss some particular solutions focusing on the cavity function .
the cavity function @xmath81 is defined by@xmath91sometimes it is also expressed as @xmath92 which defines the ` tail function ' @xmath93 @xcite , involved also in the exact relationship @xmath94 + e(r)d(r ) .
\label{y3}\ ] ] note that the py approximation amounts to set @xmath95 @xmath96 the exact @xmath81 is a well - defined continuous function of @xmath2 with at least 2 continuous derivatives , even when the potential @xmath46 itself has a hard - core region or other discontinuities @xcite .
since in baxter s model the cavity function at the contact @xmath97 plays a crucial role in characterizing a particular approximation within the gpy class , it is important to know , for shs1 , the first three coefficients of the expansion @xmath98 ( @xmath99 , @xmath100 being the packing fraction and the particle volume , respectively ) , which can be derived from the shs1 _ exact _ virial coefficients @xcite ( see section v.b and eq .
( @xmath101 ) ) .
one finds @xmath102 where@xmath103 note that , within the gpy class , _ mixed _ closures imply that @xmath104 when @xmath105 , whereas @xmath106 for @xmath68 may largely differ from the py expression .
a second crucial remark is that , while @xmath107 is continuous everywhere , a cavity function derived from an approximate closure may be discontinuous .
this drawback occurs , for instance , in the msa ( see eq .
( @xmath108 ) below ) when the potential has some discontinuities outside the core . in the case of sw fluid ,
@xmath109 may thus be considered continuous at @xmath110 , but it is discontinuous at @xmath111 ( a fact which has remarkable consequences on the derivation of the corresponding msa virial equation of state @xcite ) . before ending this section , we would like to stress also a general methodological point related to the way the sticky limit is performed , in view of its importance in the evaluation of the virial eos .
the correct procedure - hereafter referred to as _
sticky limit at the end _ * - * involves three steps : * * first , solve the oz equation , within some approximate closure , for the ` starting ' potential @xmath13 .
second , compute structural and thermodynamic properties of the sw1 fluid .
finally , perform the sticky limit . in baxter
s case , an immediate difficulty arises from the lack of analytic solution for sw fluids .
remarkably , one can find the correct shs1-py solution even in the absence of a complete solution of the sw1-py model , by using perturbative expansions in powers of @xmath5 .
the starting point is @xmath112 \
y_{{\rm sw1}}(r ) , \label{sl1}\ ] ] where the expression ( @xmath113 ) for @xmath114 has been inserted into eq .
( @xmath115)@xmath116 note that no approximation is contained in eq .
( @xmath117 ) , if @xmath118 is exact .
the use of @xmath119 yields @xmath120 .
on inserting this expression into eq .
( @xmath121 ) , one can determine the necessary properties of the unknown @xmath122 ( see appendix a ) and hence , in the sticky limit , of @xmath123 . however , the evaluation of structural and thermodynamic properties involves operations such as derivatives and integrals , which do not necessarly commute with the sticky limit .
an underestimation of this fact has led to the existence of some incorrect results in the literature when the sticky limit was carried out already at the level of the boltzmann factor ( _ sticky limit from the outset _ ) , i.e. @xmath124 which implies that @xcite @xmath125 @xmath126 ( note that in eq .
( @xmath127 ) one can not write @xmath128 $ ] , since @xmath129diverges logarithmically ) .
this short - cut procedure of ` sticky limit from the outset ' is rather common in the literature on systems with sticky terms in the pair potential ( not only shs fluids ) .
we stress that this approach , albeit computationally simpler , is not always equivalent to the ` sticky limit at the end ' ( which remains the unique correct procedure ) and may be dangerous in calculations of thermodynamic properties . in view of this scenario ,
we have performed an accurate re - analysis of baxter s work and found that the way of taking the sticky limit does not affect @xmath43 , as well as any quantity computed from it without @xmath2-derivatives , i.e. , for instance , the structure factor @xmath130 the compressibility and energy eos s , the internal energy and the helmholtz free energy .
this is _ not _ the case of the virial eos as it will be shown later .
being aware of the danger hidden in the procedure of ` sticky limit from the outset ' , baxter used it only whenever possible @xcite . in the present paper , we follow only the surely correct procedure of ` sticky limit at the end ' .
in addition to the py approximation , we now consider other gpy solutions relevant to _ mixed _
closures @xmath131/py , where the dots refer to the approximation chosen for @xmath132 , the dcf in the shrinking interval ( @xmath3,@xmath26 ) .
let us start from @xmath133 . for any square - well potential , and in particular for baxter s sw1 ,
this msa / py mixed closure coincides with the pure msa one . in terms of cavity function , one can write @xmath134 \ \left [ 1+\gamma \left ( r\right ) -\beta \phi _ { { \rm tail}% } \left ( r\right ) \right ] \text { \qquad for \ } r>\sigma , \label{ymsa}\ ] ] and thus @xmath135 , & \qquad \sigma < r < r , \\ 1+\gamma _ { { \rm sw1-msa}}\left ( r\right ) , & r >
r,% \end{array}% \right.\]]which is clearly discontinuous at @xmath111 . in the @xmath66 limit one gets : @xmath136 since the factor @xmath137 remains finite , while the remaining term in @xmath138 , involving the product of the well - width @xmath5 times the potential at contact , vanishes due to the weaker ( logarithmic ) divergence of @xmath139 ( on the contrary , @xmath5 times the boltzmann factor would yield a non - zero finite result ) . as a consequence , one finds @xmath140 this means that _ the shs1-msa factor correlation function differs strongly from its shs2-msa counterpart and coincides with the hs - py one . _ in other words , for baxter s potential , the msa is a far too poor approximation , unable to include into @xmath141 any effect due to the surface adhesion . on the other hand
, we can anticipate that the shs1-msa virial pressure does not coincide with its hs - py counterpart , but diverges ! as discussed in detail later on ( section v.c and appendix b ) , this is due to the discontinuity at @xmath111 in the cavity function @xmath142 within the msa approximation .
hence a particularly careful treatment of the jump in the cavity function , appearing in the virial equation of state , is required in this case @xcite , unlike all those closures which preserve the continuity of @xmath81 ( e.g. py ) . as a result
, we find the somewhat paradoxical situation of having the limit of the factor correlation function @xmath143 identical with the hard sphere one @xmath144 , whereas the corresponding limits of the virial equations of state are different .
this unexpected result may appear puzzling at first sight .
however we attribuite it to the strong thermodynamic inconsistency generated by the msa in the sticky limit .
henceforth the mmsa will be denoted as c0_-approximation _ , to emphasize that it corresponds to the _ exact _ zero - order approximation , @xmath145 , for the dcf in the shrinking interval outside the core .
now @xmath146 $ ] for @xmath105 , and @xmath147this implies that @xmath148 @xmath149 which depends on temperature , but is _ density - independent . _
it is interesting to remark that the resulting @xmath150 _ _ solution coincides with the mmsa solution of the shs3 model gazzillo03b , provided that the reduced temperature @xmath151 of that potential is replaced with baxter s @xmath152 .
it is also worth noting that the shs3-mmsa solution is formally coincident with the shs2-msa one @xcite . as a consequence of this mapping ,
most of the msa results obtained in the past for shs2 ( which turns out to be _ ill - defined _ from a thermodynamical point of view ) are found to still hold within a correct framework @xcite .
the most natural extension of the previous approximation is the inclusion of the next order correction in the density expansion .
we shall denote this as c1_-approximation _ , _ _ which is defined by @xmath153 , \]]where @xmath154 is the first - order coefficient in the density expansion of @xmath155 .
this approximation is _ exact up to first - order _ in density . as a consequence ,
we find @xmath156 with @xmath157 being the exact first - order coefficient given by eq .
( @xmath158 ) .
the corresponding @xmath159 follows immediately .
clearly , higher orders terms in the density expansion of @xmath160 could be considered . however
, even the inclusion of the next order correction would require a much more elaborate analysis compared to the first - order one .
furthermore it is reasonable to expect that higher order , @xmath161-dependent , terms should be introduced in a way balanced with the hs part , in view of the slow convergence of the perturbative series . for these reasons , in this paper ,
our analysis will be limited to the first correction only .
next we consider the py closure , which can not be deduced from c0 by adding a finite number of perturbative terms . in this case
the approximation to @xmath81 is ` continuous ' , in the sense that the same closure is employed both in @xmath27 and @xmath162 .
( @xmath163 ) then becomes @xmath164 where the contact value of the py cavity function can be determined analytically by solving a quadratic equation , usually written in terms of @xmath165 .
only the smaller of the two real solutions ( when they exist ) is physically significant @xcite .
the result reads@xmath166 } , \label{b2}\]]where @xmath167 is the hs - py result @xmath168it is interesting to compare the first terms of the expansion of @xmath169 in powers of @xmath170 with the exact results given by eq .
( @xmath158 ) .
one finds that @xmath171 , @xmath172 , while @xmath173 note that in @xmath174 there is a @xmath175-term , missing here.@xmath176 by exploiting the baxter - oz equations , one can evaluate @xmath177 and find ( see appendix b)@xmath178an improvement over the py approximation can be obtained by recalling eq .
( @xmath179 ) and writing@xmath180 ^{2}+d(\sigma ) .
\label{so6}\end{aligned}\]]we stress that in eq.([so6 ] ) the presence of @xmath181 is our key for improving over the py closure .
if we are able to give an explicit approximation for @xmath182 , then @xmath75 can be computed ( analytically or numerically ) from eq .
( @xmath183 ) . in the following
we shall attempt to motivate a particular form of @xmath184 using a combination of exact heuristic arguments .
our guidance in this task hinges on two observations .
first , we note that density expansion of eq .
( [ y4 ] ) suggests that @xmath182 starts with a @xmath185-term , and hence it is negligible at low density .
the py case indeed corresponds to set @xmath186 . on the other hand
, it is well known that the py approximation yields a description of the cavity function for hs which is far from being fully accurate .
this can be quickly established from the density expansion of eq .
( [ b5 ] ) which yields @xmath187 , markedly different from @xmath188 ( see eq .
( [ y5 ] ) ) . in order to cope with both the above remarks , we propose the following simple expression for @xmath189 , which will be referred to as gpy1 , @xmath190 substitution in eq .
( [ e1 ] ) yields eq .
( [ b2 ] ) again , but now with @xmath191the hs part of @xmath192 has been chosen in order to get @xmath193 the well - known carnahan - starling expression , which is more accurate than @xmath194@xcite .
on the other hand , the @xmath195-dependent part of @xmath192 ensures that now the @xmath196expansion of @xmath197 yields@xmath198where the first three terms are closer to the exact ones than their py counterparts . in spite of the fact that the approximation given in eq . ( [ gpy1 ] )
is not fully physically motivated , it nevertheless does represent an improvement over the py one , as it will be shown in section vi where numerical tests are reported .
the structure of the fluid is described by @xmath35 , or by the structure factor , @xmath199 ^{-1}=\left [ \widehat{q}\left ( -k\right ) \widehat{q}\left ( k\right ) \right ] ^{-1 } , \label{sfact}\]]where @xmath200 is the magnitude of the scattering wave vector,@xmath201 and @xmath202 denote the three - dimensional fourier transform of @xmath45 and @xmath33 , respectively , while @xmath203 with @xmath204 being the unidimensional fourier transform of @xmath141 .
( [ so1 ] ) , the calculation of @xmath205 is straightforward and yields @xmath206 ^{2}+\left [ 12\eta \ \widehat{q}_{\sin } \left ( k\sigma \right ) \right ] ^{2}},\ ] ] where @xmath207@xmath208 ^{-1}+q_{\sigma } \,,\]]and @xcite @xmath209
the thermodynamic properties can be obtained through the usual compressibility , energy and virial routes to the eos . in any approximate theory , however , these results do not agree with each other , due to thermodynamic inconsistency .
the _ exact _ values of the first virial coefficients in the @xmath170-expansion of the shs1 pressure @xmath210 are @xcite @xmath211 in the following we will discuss results for the msa , c0 and c1 closures , and compare them with the py ones .
thermodynamics of the gpy1 approximation will not be considered in this paper . in general , the _ compressibility _ @xmath212 _ equation
_ reads @xmath213 where the factorization given in eq.([sfact ] ) as well as the definition of @xmath76 have been exploited .
integration with respect to density then yields the compressibility eos , i.e. for @xmath214 @xmath215 for our model and within the gpy class of closures , eqs .
( @xmath216 ) yields @xmath217 in particular , for the msa one has the simple result @xmath218 which is consistent with that reported in eq .
( [ b4 ] ) . then , for the @xmath219_-approximation _
( mmsa ) compressibility eos one finds@xmath220while the @xmath221_-approximation _ gives a more complicated expression @xmath222 where @xmath223 .
these results can be compared with the well known py ones@xmath224explicit numerical tests will be reported in section vi .
it is worthwhile noticing that use of eq .
( @xmath225 ) into eq .
( @xmath226 ) does _ not _ yield back eq .
( @xmath227 ) . the helmholtz free energy @xmath228 can be computed from the cavity function at contact .
in fact , following baxter @xcite , it can be shown that @xmath229 we stress that this relationship is _ exact _ , independent of any closure .
once that @xmath228 is known , the energy eos can be determined via @xmath230 . inserting the analogue of the @xmath170-expansion eq .
( [ cexp ] ) for @xmath231 into eq .
( @xmath232 ) , one gets the exact relationship between virial coefficients and @xmath233 , namely @xmath234which we have exploited to calculate the exact first @xmath233 s from the exact first @xmath235 s given by post and glandt @xcite .
> from the contact values calculated in section iii.b one finds @xmath236 @xmath237 @xmath238 and , the corresponding expression for @xmath239 are : @xmath240 @xmath241 within the py approximation , analytic expressions for @xmath242 and @xmath243 are also available @xcite .
it is worth noting that the c0-energy results are mean - field ones : @xmath244 predicts only the second virial coefficient correctly , whereas its higher - order virial coefficients coincide with the hs ones .
on the other hand , the c1 and py closures give also the exact @xmath245 .
the determination of the virial _ _ @xmath246 eos @xmath247 .
\label{eosv}\ ] ] is the most delicate one .
the py result is @xcite @xmath248 \nonumber \\ & & -\frac{\eta } { 3\tau ( 1-\eta ) ^{3}}\ \left [ 1 - 5\eta -5\eta ^{2}+6\eta \lambda _ {
0}-\lambda _ { 0}^{2}+\frac{1}{24}\eta ^{-1}\lambda _ { 0}^{3}\right ] \ .
\label{pyv}\end{aligned}\ ] ] with the shorthand @xmath249 as remarked in section ii.e , a special care must be exerted in evaluating the sticky limit of this case . as no explicit derivation of eq .
( [ pyv ] ) has been reported in the literature , to the best of our knowledge , we shall discuss this point in some detail here .
the starting point is the exact expression @xcite , @xmath250 \right\ } .
\label{swvir}\ ] ] the evaluation of the last term is the _ crucial _ point . to our knowledge ,
bravo yuste and santos @xcite were the first to clearly stress that , in general , @xmath251 can not be replaced with @xmath252_{r=\sigma } $ ] , the analogous quantity corresponding to the procedure of ` sticky limit from the outset ' , since sticky limit and @xmath253differentiation do not commute ( see appendix b ) .
in fact , on replacing the last term in eq .
( [ swvir ] ) with @xmath254 one obtains an incorrect virial eos , which differs from eq .
( @xmath255 ) ( in particular , the @xmath256-term is missing ) . in some sense , a ` memory ' of the original potential
is retained even after the sticky limit has been performed .
the impossibility of getting the correct derivative from @xmath81 once the sticky limit has been taken was also recognized by seaton and glandt @xcite .
in short , _ the correct py virial eos can not be obtained if the sticky limit is taken ` from the outset ' .
_ on the other hand , no such drawbacks occur when evaluating other structural or thermodynamic properties in the absence of @xmath253differentiation .
this important point has sometimes been overlooked in the past literature .
for instance , we mention here the extension of eq .
( @xmath255 ) to mixtures as proposed by barboy and tenne @xcite .
their multi - component py virial equation of state , based upon the ` sticky limit from the outset ' , is not correct , since it does not reduce to eq .
( [ pyv ] ) in the one - component limit , as it should .
a similar pitfall appears in the virial pressure of ref .
@xcite . in appendix
b it is also shown that @xmath257 i.e. , in the sticky limit the msa virial eos differs from @xmath258 and diverges .
although this result may appear odd at first sight , it is nevertheless consistent with our previous discussion .
the virial eos depends not only on @xmath259 after the sticky limit , but also on the ` tail ' of the original potential .
it is also worth stressing that the evaluation of the msa virial eos for sw fluids embodies a further subtle point not present in the py virial eos , because of the discontinuity of @xmath142 @xcite ( see appendix b ) .
one could reasonably expect the above divergence to be a consequence of the poorness of the msa closure , which should disappear by resorting to more accurate approximations .
unfortunately , the virial eos diverges within the c0-approximation too ( see appendix b ) : @xmath260 and thus higher - order closures should be considered .
in order to give a flavor of the difference appearing in the different approximations , we present here some results and an interesting comparison with monte carlo ( mc ) simulation data @xcite . figures 1 and 2 depict msa , c0 , c1 and py results for the structure factor @xmath205 , compared with the corresponding mc data @xcite .
four thermodynamic states are considered and we have displayed the corresponding results in order of decreasing @xmath152 .
the simulation data for @xmath261 stem from a recent improved re - analysis @xcite of the calculations by kranendonk and frenkel @xcite .
these data are less susceptible to finite size effects than the original ones kranendonk88 . as msa curves coincide with the hs ones
, one can clearly appreciate the effects of surface attraction present in the shs1 model . as is known gazzillo00,gazzillo02a , at high values of @xmath152 repulsive forces predominate and the structure factor of sticky hard spheres resembles that of hard spheres , although heights and positions of peaks are quite different ( see figure 1 ) . lowering @xmath152 ( i.e. decreasing temperature or increasing stickiness ) leads to a different behaviour characterized by a strong increase of @xmath262 , which is related to the isothermal compressibility and the density fluctuations .
the observed development of such a new peak at @xmath263 ( see figure 2 ) signals the approach to a gas - liquid phase transition ( @xmath264 diverges at the spinodal curve ) . while there is a good overall agreement between mc data and py predictions
, one can observe that the c0 and c1 curves lie rather close to the py ones .
in addition , in figures 1a ) , 1b ) and 2c ) one clearly sees an improvement in going from the c0 to the c1 approximation , whereas in fig .
2d ) the c0 results are nearly coincident with the py ones , at least in the considered interval of parameters . in figure 2
the largest discrepancies among the different approximations occur near the origin ( see 2c ) ) .
figures 3 and 4 illustrate the behaviour of the compressibility pressure , @xmath265 .
in all the considered cases , the performance of the c0 and c1 approximations is quite good ( again , c1 is slightly better than c0 ) .
the c0 , c1 and py results are nearly coincident in the interval @xmath266 , which covers the concentration range typical of many colloidal or protein solutions .
figures 5 and 6 show the predictions for pressure obtained from the energy route , @xmath267 .
now , there are much larger differences with respect to the py and mc results .
it is somewhat surprising that , unlike the py case @xcite , within the c0 and c1 closures the compressibility route appears to be superior than the energy one . as a final illustrative example ,
we report here a sample of gpy1 results . in figure 7 gpy1 and py predictions for @xmath261
are compared with the mc simulation for the state @xmath268 . in general the py theory tends to underestimate the amplitude of the oscillations in @xmath261 .
now the improvement present in @xmath269 is reflected in the behaviour of @xmath261 near the origin . in this region , while the py curve lies below the mc data , the gpy1 one has a more correct trend .
on the other hand , since the py critical point evaluated via the compressibility route ( i.e. via @xmath270 ) is known to be underestimated with respect to the exact one @xcite , one can reasonably expect that @xmath271 may be underestimated too .
we have not considered thermodynamics corresponding to the gpy1 closure , since such a task would probably require analytical / numerical calculations which go beyond the scope of the present paper .
one could expect , however , results not significantly different from the py ones .
we have shown that baxter s model ( hard spheres with an infinitely deep and narrow attractive square - well ) is analytically solvable not only within the py approximation but also within a general class of closures .
we have denoted this class as ` generalized py ' approximation , since all its elements are ` mixed closures ' , which coincide with the py one beyond @xmath26 , the upper boundary of the well , but may employ a different approximation within the well region @xmath27 .
it has been argued that , even in the limit of vanishing width of the well , there is a ` memory ' of the approximation to @xmath81 chosen within the well , and this has been identified as the distinguishing feature of each closure of the gpy class . in particular , we have shown that : i ) the shs1 model can be solved analytically within the msa closure , but its shs1-msa solution has nothing to do with the shs2-msa one and is equal to @xmath144 ; ii ) an interesting analytic solution can be obtained also within the mmsa ( or c0-approximation ) .
the resulting factor correlation function @xmath272 is formally identical with @xmath273 ( if @xmath152 is replaced with the reduced temperature @xmath274 ) , obtained for the shs3 model @xcite .
although the solution @xmath43 of the oz integral equation , and all properties which can immediately derived from it , can be carried out within the unified gpy treatment , the calculation of other properties may require an analysis starting from the original sw potential , i.e. _ before _ the sticky limit
. a detailed analysis of the non - commutativity of the sticky limit with other operations in some instances , notably for the virial eos , has been provided . in principle
, one could extend the above remarks from the shs case to any general model with sticky terms in the pair potential . within this more general viewpoint ,
the presence of dirac @xmath275-functions in any of the functions @xmath276 , @xmath33 , @xmath35 could be regarded as an indication that the procedure of ` sticky limit from the outset ' is implicitly adopted .
the experience of the present paper on shs systems suggests that special care should be exerted in evaluating properties where @xmath2-differentiation is required , as in the virial eos , since the corresponding results might be altered by the non - commutativity of the sticky limit . the gpy class of closures may be roughly split into two subclasses having the py approximation as a separating borderline .
the first subclass is originated by the msa , which is commonly believed as one of the simplest possible closures in view of its mean - field character .
the main drawback of this closure lies in its linear relationship with the potential , which prevents a correct zero - density limit of the correlation functions .
we have cured this by modifying the closure so that the c0-direct correlation function is related to the mayer function rather than to the potential , and in this case the correct zero - density limit is recovered .
finally , an explicit density - dependence can be perturbatively included in the closure , order by order , through a density expansion .
we have discussed in detail the analytical results stemming from the zero ( c0 ) and first ( c1 ) order terms in density expansion and a comparison with monte carlo simulations shows that , in spite of the crudity of the approximations , the resulting structure factors are rather close to the mc data for a wide range of the parameters .
our main interest in the closures of this first subclass is that they are simple enough to allow an easy extension to mixtures and application to polydisperse fluids , unlike the py case .
the main disadvantage is , on the other hand , represented by strong thermodynamic inconsistencies , which can not be coped with in a simple way .
the closures of the second subclass have in fact been devised to the aim of reducing these inconsistencies , by exploiting the additional contribution @xmath189 to @xmath97 .
we have given only an illustrative example ( called gpy1 ) to show how small differences in the contact value of the cavity function can produce , for instance , a different behaviour of @xmath261 near the origin .
on the other hand , the _ exact _ dcf is expected to possess also a ` tail ' for @xmath67 , which all gpy closures neglect .
the presence of this tail ( which could be approximated , for instance , by a yukawa function ) implies that the _ exact _ solution @xmath277 can not coincide with @xmath278 .
these slight differences can be expected to be essential in order to achieve thermodynamic consistency , so that the use of the mentioned @xmath182 is useful but probably not sufficient . in this framework
, it would be interesting to compare our @xmath189-based closures with the completely different approach by santos _
@xcite , who proposed , for shs1 , to go beyond the py closure through a rational function approximation . finally , we wish to mention that work along the lines laid down in the present paper is in progress to investigate , by the multi - component extension of the c0-approximation , stability boundaries and percolation threshold of polydisperse colloidal fluids .
we are particularly grateful to daan frenkel and mark miller for having provided us with their very accurate monte carlo results before publication @xcite .
the italian miur ( ministero dellistruzione , delluniversit e della ricerca ) , the infm ( istituto nazionale di fisica della materia ) are gratefully acknowledged for financial support .
the condition @xmath279 when @xmath67 implies that @xmath280 for @xmath67 , and @xmath281 .
thus , eq . ( @xmath121 ) may be cast in the form @xmath282 , \label{ie5}\]]where the definitions @xmath283 , @xmath284 , @xmath285 , as well as the core condition , eq .
( @xmath286 ) , have been employed . here
, @xmath287 denotes the smaller between @xmath26 and @xmath288 .
> from eqs .
( @xmath289 ) , one finds that @xmath43 splits into three parts : @xmath290 for @xmath291 @xmath292 for @xmath293 and @xmath294 for @xmath295 the function @xmath43 is continuous at @xmath296 @xmath297 and @xmath298 while @xmath292 is always represented by the second degree polynomial @xmath299 @xcite . as @xmath18 ,
the intervals ( @xmath300 ) and ( @xmath65 ) vanish , and @xmath43 reduces to the sticky limit of @xmath301 to obtain @xmath302 as well as @xmath303 it is indifferent to take the sticky limit ` at the end ' or ` from the outset ' ( see section ii.e ) , since such a limit commutes with the @xmath304-integration included in the definition of @xmath76 and @xmath77 . because of
the continuity condition @xmath305 the sticky limit of @xmath306 can be evaluated from @xmath307 although the explicit solution is not known , integrating eq .
( @xmath289 ) yields a formally exact expression of @xmath308 from which we get @xmath309 @xmath310@xmath311 it is reasonable to assume that @xmath76 and @xmath77 tend to finite values , and the same occurs for @xmath312 , whereas @xmath313 and @xmath314 may diverge .
consequently , @xmath315 is determined by @xmath316since all remaining terms vanish as @xmath66 ( @xmath317 could also be replaced with @xmath318 since the difference becomes zero in the sticky limit ) .
( @xmath115 ) , along with eq .
( @xmath113 ) , then yields @xmath319 , and eq .
( @xmath320 ) follows immediately . that simple general relationship allows to determine @xmath43 of the shs1 model for _ any _ closure with @xmath279 when @xmath67
\1 ) _ exact case and py approximation . _ _ _ for any sw potential , the virial theorem is awkward since both @xmath35 and @xmath321 are discontinuous at @xmath110 and @xmath111 .
usually , one rewrites @xmath322 $ ] as @xmath323 in eq .
( @xmath324 ) , and uses @xmath327 \label{a2 } \\ &
= & 1 + 4\eta \left\ { g_{{\rm sw}}(\sigma ^{+})+\left ( r/\sigma \right ) ^{3}\ % \left [ g_{{\rm sw}}\left ( r^{+}\right ) -g_{{\rm sw}}(r^{-})\right ] \right\ } , \nonumber\end{aligned}\ ] ] @xmath328 \right\ } .\ ] ] as the exact ( as well as the py ) @xmath81 and @xmath329 are finite and continuous everywhere , one can expand @xmath330 about @xmath331 @xmath332_{r=\sigma } ^{\prime } \ ( r-\sigma ) + { \cal o}((r-\sigma ) ^{2}).\ ] ] @xmath335_{r=\sigma } ^{\prime } \right\ } \nonumber \\ & = & 1 + 4\eta \left\ { \left ( 1 - 3\alpha \right ) y_{{\rm shs1}}(\sigma ) -\alpha \ \sigma \ \lim_{r\rightarrow \sigma } [ y_{{\rm sw1}}(r)]_{r=\sigma } ^{\prime } \right\ } .\end{aligned}\ ] ] in contrast to what happens for one - dimensional systems of sticky hard rods @xcite , the order of sticky limit and differentiation is essential in the three - dimensional case , since these operators do not commute now , i.e. , to highlight the source of such a difference , we give a detailed demonstration of this result .
let us define @xmath338 and consider its py approximation , @xmath339 .$ ] since @xmath340 equals @xmath341 , this function can be obtained by subtracting eq .
( @xmath87 ) from eq .
( @xmath342 ) . after expanding all integrals @xmath343 in powers of @xmath5 and using the relationship @xmath344 ,
one takes the sticky limit and obtains@xmath345 to evaluate @xmath347 one gets @xmath348 from eq .
( @xmath289 ) for @xmath349 . after expanding @xmath350 up to the @xmath351-term ( again with the help of @xmath352 ) , the sticky limit can be performed and yields the derivation of the msa virial eos is not plain and thus deserves some remarks .
in fact , one might believe that the sw - msa virial eos can be simply obtained by inserting @xmath357 into eq .
( @xmath358 ) , which is exact for a generic sw fluid .
unfortunately , this opinion is incorrect .
in fact , as pointed out by smith _
@xcite , in the sw - msa case the procedure used for eq .
( @xmath358 ) becomes meaningless , because @xmath359 given by eq .
( @xmath360 ) is continuous at @xmath83 but discontinuous at @xmath111 where the sw potential is discontinuous ( on the contrary , @xmath107 is continuous everywhere ) .
consequently , eq . ( @xmath358 ) can not be valid in the msa .
@xmath361 , & \qquad \text{if \ \ \ } a <
x_{0}<b , \\ 0 , & \text{otherwise,}% \end{array}% \right.\]]which extends the usual definition of dirac s @xmath275 to @xmath362-functions discontinuous at @xmath363 , one can demonstrate that the proper msa expression is _ _
@xcite @xmath364 \right\ } \nonumber \\ & = & 1 + 4\eta \left\ { 1+\gamma _ { { \rm sw - msa}}(\sigma ) + \beta \varepsilon -\beta \varepsilon \left ( r/\sigma \right ) ^{3}\left [ 1+\gamma _ { { \rm sw - msa}% } ( r)+\frac{1}{2}\beta \varepsilon \right ] \right\ } .\end{aligned}\ ] ] we adopt the following terminological distinction .
a shs _ model _ is univocally defined by its hamiltonian , whose specification requires _ two _ elements : the starting potential , and the ` sticky limit ' procedure . on the other hand , a particular _ solution _
is characterized by _ three _ elements : the same two of the model , plus an approximate ` closure ' for the oz equation . | we discuss structural and thermodynamical properties of baxter s adhesive hard sphere model within a class of closures which includes the percus - yevick ( py ) one .
the common feature of all these closures is to have a direct correlation function vanishing beyond a certain range , each closure being identified by a different approximation within the original square - well region .
this allows a common analytical solution of the ornstein - zernike integral equation , with the cavity function playing a privileged role . a careful analytical treatment of the equation of state is reported .
numerical comparison with monte carlo simulations shows that the py approximation lies between simpler closures , which may yield less accurate predictions but are easily extensible to multi - component fluids , and more sophisticate closures which give more precise predictions but can hardly be extended to mixtures . in regimes typical for colloidal and protein solutions
, however , it is found that the perturbative closures , even when limited to first - order , produce satisfactory results . |
a kummer cover is a map @xmath0 given by @xmath1):=[x^n : y^n : z^n]$ ] .
kummer covers are a very useful tool in order to construct complicated algebraic curves starting from simple ones .
since kummer covers are finite galois covers of @xmath2 with @xmath3 , topological properties of the new curves can be obtained : alexander polynomial , fundamental group , characteristic varieties and so on ( see @xcite for papers using these techniques ) .
on the other hand , the braid monodromy of a plane projective curve is a powerful invariant that provides a way to compute the fundamental group of its complement and was originally described as a formalization of the zariski - van kampen method @xcite ( see @xcite also @xcite and references therein for a detailed exposition on the subject ) .
however , braid monodromy is a much more powerful invariant , since in fact it encodes the topology of the embedding of the curve , as well as the isomorphism problem for surfaces whose branching locus over @xmath4 is a given curve ( see @xcite to name only a few ) . in this work
we focus our attention on a method to reconstruct the braid monodromy of the preimage of a curve by a kummer cover .
this method is interesting in and of itself , since it combines two techniques , namely , the reconstruction of a highly non - generic braid monodromy with a systematic method to go from a non - generic to a generic braid monodromy .
this `` _ _ generification _ _ '' method is independent from kummer covers and can be applied in more general circumstances ( see also @xcite ) .
the reason for this is that oftentimes , non - generic braid monodromies are much easier to compute .
the importance of finding theoretical arguments to obtain braid monodromies lies in the fact that any direct computation requires polynomial root approximation , whose exactness can only be justified in particularly simple cases , such as curves whose equation can be given in @xmath5 $ ] ( @xcite ) , strongly real curves ( @xcite and references therein ) , or line arrangements ( see @xcite ) .
the layout of the paper is as follows : after describing in [ sec - settings ] the main objects to be used such as generic and non - generic braid monodromies , kummer covers and extended braid monodromies are described in [ sec - pencils ] as the main tool to recover a non - generic braid monodromy of a kummer cover from a braid monodromy of the base . useful _ generification _
techniques are described in [ sec - deformation ]
. sections
[ sec - cover - braids ] and [ sec - trans ] are more technical and describe useful systems of artin generators for the braids obtained after a kummer cover and the properties and singularities of kummer transforms .
finally , a detailed account of numerous examples ( some classical and some new ) of the power of kummer transforms is given in
[ sec - ex ] , where generic braid monodromies of smooth curves , sextics with six cusps ( on a conic and otherwise ) , sextics with nine cusps , hesse , ceva , and maclane arrangements are provided .
after the work of zariski , braid monodromy was defined by chisini @xcite but it was necessary to wait for moishezon @xcite in order to get further applications of this invariant .
let us fix a curve @xmath6 of degree @xmath7 , a point @xmath8 and a line @xmath9 such that @xmath10 .
we say that the curve is _ horizontal _ with respect to @xmath11 if it does not contain any line through @xmath11 ; we assume @xmath12 to be horizontal .
we consider a system of coordinates @xmath13 $ ] such that @xmath14 $ ] and @xmath15 .
we identify @xmath16 with affine coordinates @xmath17 $ ] .
let @xmath18 be a reduced equation of @xmath12 , @xmath19 @xmath20 normalized such that the coefficient of the term of higher degree of @xmath21 in @xmath22 is @xmath23 .
the fact that @xmath12 is horizontal is equivalent to @xmath24 .
the pencil of lines through @xmath11 is identified with @xmath25 , where @xmath26 corresponds with @xmath9 . following the previous notation the lines in the pencil
are denoted by @xmath27 .
let us restrict our attention to the affine part .
let @xmath28 and @xmath29 ; the line @xmath30 has equation @xmath31 while @xmath32 has equation @xmath33 , where @xmath34 let @xmath35 ; this is a finite set which contains the roots of @xmath36 ( if any ) and the values @xmath37 such that @xmath38 .
the set @xmath39 is the zero locus of the product of @xmath36 and the discriminant of @xmath40 with respect to @xmath41 .
let @xmath42 be a configuration space of @xmath43 ; for any @xmath44 the fundamental group @xmath45 is isomorphic to the braid group @xmath46 with the usual artin presentation @xmath47=1},\quad \underset{1\leq i < k-1}{\sigma_i\cdot\sigma_{i+1}\cdot\sigma_i=\sigma_{i+1}\cdot\sigma_i\cdot\sigma_{i+1 } } \right . \right\rangle.\ ] ] for the next sections we need to describe a canonical identification between @xmath46 and @xmath48 ; the group @xmath49 is identified with the homotopy classes of sets of arcs @xmath50\to{{\mathbb c}}$ ] starting and ending in @xmath51 and such that @xmath52 , @xmath53 $ ] .
let us order the points of @xmath51 , say @xmath54 and consider a set @xmath55 of simple segments @xmath56 , @xmath57 , such that @xmath58 , @xmath59 and the other intersections are empty ; such a collection @xmath55 will be called a _ diagram system _ for @xmath60
. then we associate to @xmath61 the braid which is constant for @xmath62 and behaves in a small small closed ( topological ) disk @xmath63 such that @xmath56 is a diameter as follows .
the points @xmath64 and @xmath65 counterclockwise exchange along @xmath66 .
these generators may be understood as _ half - twists _
around @xmath56 . .
] there is also a basis of the free group @xmath67 if one chooses a simple edge @xmath68 from @xmath69 to @xmath70 intersecting @xmath71 only at @xmath70 .
this basis @xmath72 is obtained as follows : take small disks @xmath73 centered at @xmath64 and assume that their intersection with @xmath74 are diameters with ends @xmath75 .
then @xmath76 is defined as follows : take a path @xmath77 from @xmath69 to @xmath78 running along @xmath79 outside the interior of the disk @xmath80 and goes counterclockwise along @xmath81 from @xmath82 to @xmath83 , @xmath84 .
let @xmath85 be the closed path obtained by running counterclockwise along @xmath81 with base point @xmath78 and define @xmath86 .
the group @xmath46 acts geometrically on the group @xmath87 as follows : @xmath88 [ ntc - accion ] as usual , @xmath89 is the conjugation of @xmath90 by @xmath91 . also , for short , we will write @xmath92 . on @xmath93 .
] there are two important facts in these definitions ; the element @xmath94 is a meridian of the point at infinity and @xmath95 is a fixed point by the action of @xmath46 .
we say that @xmath96 is an _ ordered geometric basis _ of @xmath67 . as a general notation , if @xmath97 is a group and @xmath98 we define the _ pseudo - coxeter element _ of @xmath99 as @xmath100 .
after this digression , note that @xmath101 defines a map @xmath102 .
[ def - bm ] the _ braid monodromy _ of the triple @xmath103 is the morphism @xmath104 defined by @xmath105 on the fundamental group .
[ rem - bm ] consider a geometric basis @xmath106 of @xmath107 and let @xmath108 be its pseudo - coxeter element .
note that @xmath109 is determined by @xmath110 having as pseudo - coxeter element @xmath111 .
the braid monodromy measures the motions of the points of @xmath32 along the affine lines @xmath30 ( identified with @xmath43 ) .
there are a lot of choices in order to obtain an element of @xmath112 from @xmath113 .
it is not hard to check that these choices are given by the orbits of an action of @xmath114 on @xmath112 .
the action of @xmath46 is given by simultaneous conjugation .
the action of @xmath115 is defined as follows ; let @xmath116 an artin system of generators of @xmath115 . then , if @xmath117 , then : @xmath118 @xmath119 is called a _ hurwitz move_. in particular for a choice of @xmath113 two objects are unique and well - defined : the conjugacy classes of the pseudo - coxeter element and of the _ monodromy group _ ,
i.e. , the group generated by @xmath120 .
in light of the previous discussion , a braid monodromy @xmath109 of a triple @xmath113 will sometimes be considered as a morphism ( see ) or as a list of braids @xmath121 , where @xmath122 is a geometric basis .
the first application of braid monodromy is the computation of the fundamental group , see @xcite . given @xmath109 the braid monodromy of a triple @xmath123 .
consider @xmath124 , @xmath125 , and @xmath30 ( @xmath126 ) the non - generic fibers of the pencil of lines through @xmath11 as above .
one has the following result .
[ thm - zvk1 ] the group @xmath127 has the following presentation : @xmath128 a triple @xmath123 ( or simply a curve @xmath32 ) is said to be _ fully horizontal _ if @xmath32 has no vertical asymptotes , i.e. , if @xmath129 . in that case , if @xmath130 ( maybe empty ) , corresponding to the generators @xmath131 , one has a new version of zariski - van kampen theorem , due to the fact that the above generators @xmath132 are meridians of the corresponding lines .
[ thm - zvk2 ] the group @xmath133 has the following presentation : @xmath134 there is also a statement like for curves with vertical asymptotes but it is more technical , see @xcite or @xcite . let us assume now that @xmath135 . in this case
the triple @xmath123 is automatically fully horizontal and @xmath136 . in this case
another version can be stated , which is closer to the original zariski - van kampen theorem .
[ thm - zvk3 ] the group @xmath137 has the following presentation : @xmath138 there is also a procedure to compute @xmath137 when @xmath139 , but the general formula is not so closed since it depends heavily on the local singularity @xmath140 .
we finish this sequence of versions of zariski - van kampen theorem with the generic case .
we assume @xmath135 , @xmath141 ( i.e they intersect at @xmath7 distinct points ) , and moreover for each @xmath142 there is exactly one point @xmath143 where the intersection is not transversal and satisfies @xmath144 where @xmath145 is the multiplicity of the germ @xmath146 . in the singular case ,
it means that @xmath147 is not in the tangent cone of @xmath146 and , in the smooth case , that @xmath148 is not an inflection point . in order to state the final form of zariski - van kampen theorem
, we need some notation .
the above conditions imply that for each @xmath149 , @xmath150 , we can express @xmath151 , where @xmath152 and @xmath153 is a positive word in @xmath154 , where @xmath155 and @xmath156 . for each @xmath157 , ( @xmath150 )
the elements @xmath158 represent another basis of @xmath159 .
let us denote by @xmath160 the action written in the basis @xmath161 .
[ thm - zvk4 ] the group @xmath137 has the following presentation : @xmath162 in this case , the pseudo - coxeter element of @xmath121 is @xmath163 , the positive generator of the central element of @xmath164 .
this is why in the literature braid monodromy is also referred to as a _ factorization _ of @xmath163 .
the main point is that under these genericity conditions , braid monodromy is an invariant of @xmath12 independent of the ( generic ) choice of @xmath11 and @xmath9 . moreover ,
if we can connect two curves by a path of equisingular curves , then they share the braid monodromy .
a more general type of genericity occurs simply when @xmath135 , @xmath141 .
we will refer to such a triple as _ generic at infinity_. in this case is still true , but the sets @xmath165 have to be replaced by a finite number of disjoint subsets @xmath166 .
an interesting case of genericity at infinity occurs when condition holds , but each fiber @xmath30 , @xmath126 is allowed to have more than one ramification point .
we refer to this case as _
local genericity_. from a locally generic braid monodromy , a generic braid monodromy might easily be reconstructed ( see ) .
following notation from [ sec - settings ] , let us fix a curve @xmath6 of degree @xmath167 .
we assume that @xmath14\notin\bar{{{\mathcal c}}}$ ] .
let @xmath168 be its kummer transform .
denote by @xmath27 the pencil of lines through @xmath11 .
most of the results of this section also hold if @xmath169 and the tangent cone of @xmath170 is @xmath171 ( or alternatively , if a change of variables can be performed to match these conditions ) .
let @xmath172 and let @xmath173 be its affine part .
we consider the set of lines through @xmath11 which are not transversal to @xmath174 and to which we include the lines @xmath175 and @xmath9 if necessary ( due to their special relation with @xmath176 ) .
let @xmath177 the braid monodromy of @xmath174 with respect to @xmath11 and @xmath9 is the morphism @xmath178 note that the image of @xmath109 is contained in @xmath179 , the subgroup of @xmath180 given by all the braids whose associated permutation fixes a given point ( see [ sec - cover - braids ] for a description of this group ) .
[ dfn - extended - bm ] the map @xmath181 is called the _ extended braid monodromy _ of the triple @xmath182 with respect to @xmath173 . note that if @xmath183 then the image by @xmath184 of a meridian around @xmath185 is the trivial braid .
we define the analogous objects for @xmath186 : @xmath187 for @xmath188 , an @xmath189-th root of @xmath190 .
the following lemmas are easy consequences of the properties of a cover .
[ lema - aux1 ] let @xmath191 be the map defined by @xmath192 .
then , @xmath193 and @xmath194 .
[ lema - aux2 ] the map @xmath176 induces degree @xmath189 coverings @xmath195 .
moreover , the preimage of a line @xmath196 by @xmath176 is the disjoint union of @xmath30 , @xmath197 , each one inducing a covering @xmath198 as above . as a consequence of these lemmas ,
the braid monodromy @xmath199 can be thought as follows : consider the loops @xmath200 in @xmath201 which can be lifted as loops by @xmath202 and replace the braid @xmath203 , where @xmath167 points move in @xmath204 , by the braid obtained by the constant string @xmath185 and the @xmath189-roots of these @xmath167 points .
this construction defines a map @xmath205 which is described in [ sec - cover - braids ] .
let us summarize the results .
if @xmath206 , we denote by @xmath207 the free group @xmath208 .
analogously , we denote by @xmath209 the free group @xmath210 . using covering theory , the map @xmath211 fits in a short exact sequence @xmath212 the following diagram holds : @xmath213 \matrix ( m ) [ matrix of math nodes , row sep=3em , column sep=2.5em , text height=1.5ex , text depth=0.25ex ] { { { \mathbb f}}_{r+1 } & { { \mathbb b}}_{k,1 } \\ { { \mathbb f}}_{n r+1 } & { { \mathbb b}}_{n k,1 } \\ } ; \path[->,>=angle 90](m-1 - 1 ) edge node[auto ] { $ \tilde{\nabla } $ } ( m-1 - 2 ) ; \path[->,>=angle 90](m-2 - 1 ) edge node[auto ] { $ \tilde{\nabla}_n $ } ( m-2 - 2 ) ; \path[<-right hook,>=angle 90](m-1 - 1 ) edge node[auto ] { } ( m-2 - 1 ) ; \path[right hook->,>=angle 90](m-1 - 2 ) edge node[auto ] { } ( m-2 - 2 ) ; \end{tikzpicture } \ ] ] using this diagram , one can recover geometric bases and relate the central elements of the corresponding braid groups .
the following makes this construction useful . summarizing ,
let @xmath214 be an extended braid monodromy for @xmath113 with respect to @xmath173 , and @xmath215 is the braid corresponding to @xmath175 .
consider @xmath216 $ ] the @xmath189-th kummer cover of @xmath4 and denote by @xmath217 the preimage of @xmath218 by @xmath176 .
the braid monodromy @xmath219 produces a list of braids @xmath220 as described in diagram .
one has the following [ prop - kummer - bm ] the element @xmath221 described above is an extended braid monodromy for the triple @xmath222 with respect to @xmath223 .
it is an immediate consequence of the previous construction and lemmas [ lema - aux1 ] and [ lema - aux2 ] .
note that the braid monodromy obtained from @xmath221 via the forgetful map @xmath224 is ( for the interesting cases ) a highly non - generic braid monodromy .
[ lema - geometric - basis - cover ] let @xmath225 be a geometric basis of @xmath207 such that @xmath226 is a meridian of @xmath227 .
then , the ordered list @xmath228 where @xmath229 ( see notation [ ntc - accion ] ) , forms a geometric basis of @xmath209 .
[ lema - image - central ] let @xmath230 be the positive generator of the center of @xmath180 .
then @xmath231 via the inclusion @xmath232 .
the purpose of the upcoming sections will be to describe an effective way to obtain a generic braid monodromy for @xmath233 from the given computation of @xmath199 .
this is an outline of the general strategy , which will be developed in what follows : * if @xmath234 , we compose @xmath199 with the natural map @xmath235 obtained by forgetting the constant string @xmath185 . * if @xmath236 then the map factorizes through @xmath237 whose kernel is generated by the meridians around @xmath185 . *
if @xmath238 we change the line at infinity ; the image @xmath239 of a meridian at infinity is obtained as follows .
let @xmath240 be the pseudo - coxeter of the braid monodromy ; then @xmath241 , where @xmath242 is the positive generator of the center of the braid group .
* slightly move the projection point @xmath11 to obtain a generic projection .
as was introduced in [ sec - pencils ] , let @xmath179 be the subgroup of @xmath180 given by all the braids whose associated permutation fixes a given point in @xmath43 , say @xmath185 .
we identify @xmath180 with the group of braids with ends in @xmath243 non - negative real points @xmath244 and we consider the artin generators @xmath245 in @xmath180 which are geometrically associated with the diagram system obtained from the paths joining these points in the real line , see [ sec - settings ] .
recall that any choice of a _ diagram system _
@xmath55 of @xmath243 points in @xmath43 induces an artin system of generators of @xmath180 .
note that the only half - twist @xmath61 that moves @xmath185 is @xmath246 . the following result is well known .
the group @xmath179 is generated by @xmath247 ( as a subgroup of @xmath180 ) and it is naturally isomorphic to the group of braids of @xmath167 strands in @xmath204 .
let us consider the map @xmath248 , given by @xmath249 .
note that this map induces a morphism @xmath250 via the multivalued function @xmath251 .
the goal of this section is to give explicit formul for this morphism . in order to do so
, one needs to explicitly choose systems of generators for @xmath180 and @xmath252 .
consider the diagram system @xmath55 for @xmath179 described above .
the image of the half twist @xmath61 , associated with @xmath56 , @xmath57 , is a product of half - twists associated with the pairwise disjoint arcs @xmath253 such that @xmath254 .
the image of @xmath255 is more complicated since @xmath256 intersect at @xmath185 .
however , note that the natural system of arcs @xmath257 does not produce a diagram system for @xmath252 . in order to solve this situation
one needs to define diagram systems for @xmath258 that allow to describe the arcs @xmath257 naturally . for different purposes , different diagram systems
might be more suited . here
we will concentrate our attention in two particular diagram systems : the circular and the radial .
let us denote @xmath259 .
we identify @xmath260 with @xmath261 , where @xmath262 is the non - negative @xmath189-th root of @xmath263 and @xmath264 .
consider the following arcs : * @xmath265 is the counterclockwise arc joining @xmath266 to @xmath267 in the circle centered at @xmath185 , @xmath268 ; * @xmath269 is the segment joining @xmath270 and @xmath271 .
these arcs and segments are illustrated in .
the list of arcs @xmath272 with a left - lexicographic order produces a diagram system called the _ circular diagram system_. the half twist produced by an arc @xmath265 will be denoted by @xmath273 .
the half - twists associated with @xmath257 will be denoted by @xmath274 .
figure [ fig : dibujo2_1 ] is obtained from figure [ fig : dibujo1 ] after unwinding the circular diagram system into a straight line to get the usual projections .
this will help visualize the rewriting of the half - twists @xmath274 in terms of the @xmath275 .
[ fig1 ] consider @xmath276 , where @xmath277 represents the left - lexicographic order ( that is , either @xmath278 or @xmath279 and @xmath280 ) .
we will define @xmath281 [ rem - p ] note that @xmath282 , @xmath283 .
also , note that @xmath284 represents a half - twist ( resp . a negative half - twist ) interchanging @xmath285 and @xmath286 along a spiral arc .
the arc corresponding to @xmath287 ( or @xmath288 ) is shown in .
[ lema - pm ] under the above conditions and for any @xmath289 one has 1 .
[ lema - pm1 ] @xmath290 , where @xmath291 denotes the element following @xmath292 in the left - lexicographic order .
2 . [ lema - pm2 ] @xmath293 .
[ lema - pm3 ] @xmath294 .
property is immediate by induction since @xmath295 where @xmath296 denotes the element immediately smaller than @xmath297 and @xmath298 denotes its second predecessor @xmath299 .
property is analogous . in order to show property note that @xmath274 is a half - twist that exchanges @xmath266 and @xmath300 .
therefore , according to one has @xmath301 a similar induction argument shows that @xmath302 the other equalities are a consequence of and properties and .
[ prop - motion1 ] the map @xmath303 is given by @xmath304 as mentioned at the beginning of this section , if @xmath305 , the preimage of @xmath56 by @xmath306 is a disjoint union of segments @xmath307 . therefore the image of the half - twist @xmath61 associated with @xmath56 is the product of the @xmath189 half - twists @xmath308 .
, @xmath309 . ]
note that @xmath255 corresponds to a full turn around @xmath185 , its preimage by @xmath306 being a counterclockwise rotation of angle @xmath310 of the points @xmath311 .
this is nothing but @xmath312 ( see figure [ fig : sk2 ] ) .
[ ex-0 ] let us consider the case @xmath313 and the composition of @xmath314 with the natural projection of @xmath315 given by forgetting the string @xmath185 .
then we have a map @xmath316 such that @xmath317 . [ ex-1 ] next consider the case @xmath318 and the map @xmath319 as in .
then , according to lemma [ lema - pm ] @xmath320 and @xmath321 for some applications it is more useful to consider a different diagram system which will be called _ radial _ , see .
[ fig2 ] we call the artin generators for this system @xmath322 , @xmath323 , @xmath324 . according to this choice , these are the half - twists @xmath274 associated with the segments @xmath257 for @xmath325 , that is , @xmath326 however , the element @xmath327 corresponds to an arc joining @xmath328 with @xmath329 as shown in figure [ fig:2bis ]
. denote @xmath330 then we have @xmath331 moreover , @xmath332 .
the following result holds .
[ prop - motion2 ] the map @xmath303 is given by @xmath333 where @xmath334 represents the right - lexicographic order .
the result follows from .
the formula for the image of @xmath61 is a direct consequence of .
the formula for the image of @xmath255 can be obtained using and the fact that @xmath335 commutes with @xmath336 and with @xmath337 for @xmath338 and @xmath339 ( see ) . by . ]
[ ex - cover2 ] let us consider the case @xmath340 and @xmath341 the composition of @xmath342 with the forgetful map as in .
according to one has @xmath343 in fact , a more convenient description of these maps can be given for another choice of generators @xmath344 coming from the natural diagram system on the real line joining the preimages of @xmath345 ( which they all lie on the real real line since @xmath340 ) . @xmath346
we are mostly interested in computing generic braid monodromies of curves ( see [ sec - generic ] ) .
however , it is usually easier ( and oftentimes more natural ) to compute non - generic braid monodromies .
for instance , those obtained after kummer covers ( even if the starting monodromy was generic ) are typically non generic . in this section
we show how to deal with non - generic braid monodromies .
let us consider the braid monodromy of a fully horizontal triple @xmath347 , where @xmath348 ( e.g. the hypothesis in ) , where @xmath349 and @xmath218 are not necessarily transversal .
[ prop - gen1 ] assume @xmath350 , let @xmath351 be a braid monodromy factorization for @xmath347 , and consider a line @xmath352 such that @xmath353 and @xmath354 . then @xmath355 is a braid monodromy factorization for @xmath356 . note that @xmath357 , where @xmath39 ( resp .
@xmath358 ) is the ramification set for @xmath347 ( resp . for @xmath356 ) . since the line @xmath352 is transversal , the pseudo - coxeter of its braid monodromy factorization is @xmath163 and the result follows .
this will be a common situation when studying kummer covers for curves which are not transversal to the axes . in that case
we can be more explicit .
let @xmath359 be an extended braid monodromy factorization for @xmath360,\bar{l}_\infty)$ ] with respect to @xmath361 .
assume that @xmath362 is not transversal to @xmath218 , and that @xmath215 is the braid corresponding to @xmath175 .
let us denote by @xmath363 its pseudo - coxeter element .
consider @xmath216 $ ] the @xmath189-th kummer cover of @xmath4 and denote by @xmath217 the preimage of @xmath218 by @xmath176 .
the extended braid monodromy @xmath364 produces an extended braid monodromy @xmath365 for @xmath366 ( see ) .
[ prop - gen2 ] under the above conditions , a generic at infinity extended braid monodromy factorization is obtained by adding to @xmath367 the braid @xmath368 , using the inclusion @xmath369 ( see diagram ) .
it is a consequence of and . as announced in
[ sec - settings ] , some types of non - generic braid monodromies can be easily deformed into generic braid monodromies .
this is the case with locally generic triples .
let @xmath103 be a locally generic triple ( see [ sec - generic ] ) and @xmath39 the set of ramification values of the projection . as already mentioned in
[ sec - generic ] , for each @xmath126 , there exists a finite number of points @xmath370 satisfying .
[ prop - gen3 ] under the above conditions , let @xmath371 be the braid associated with @xmath30 , @xmath126 . then there is a factorization @xmath372 , where @xmath373 are pairwise commuting braids and each one corresponds to one non - transversal point in @xmath30 .
moreover , replacing @xmath240 by @xmath374 ( the order does not matter ) in the braid monodromy factorization of @xmath103 for each @xmath142 produces a generic braid monodromy factorization of @xmath12 .
since @xmath375 the triple @xmath103 is fully horizontal , that is , there are no vertical asymptotes
. consider @xmath370 .
due to the conic structure of @xmath376 one can find disjoint milnor polydisks @xmath377 around each @xmath378 such that @xmath379 ( we need condition to ensure this ) .
let us denote by @xmath380 the local monodromy around each @xmath378 based at @xmath381 ( close enough to @xmath37 ) .
note that @xmath382 , where @xmath383 if and only if @xmath378 is a transversal intersection of @xmath32 and @xmath30 .
if @xmath378 is singular in @xmath32 , then the link produced by closing @xmath380 is the link of the singularity at @xmath378 .
if @xmath378 is a single tangency , then @xmath380 is a half twist .
therefore , @xmath380 is trivial if and only if @xmath378 is a transversal intersection of @xmath32 and @xmath30 .
we disregard these trivial braids , denote by @xmath384 the number of non - trivial braids , and denote them by @xmath385 after reordering .
define by @xmath386 a path that joins @xmath381 and @xmath37 consider @xmath387 the open braid associated with @xmath386 and define @xmath388 ( here `` @xmath389 '' just means juxtaposition ) . since the strings inside each milnor polydisk are different , the braids @xmath390 automatically commute .
for the _ moreover _ part , note that moving @xmath11 generically on @xmath391 causes just a small change of coordinates @xmath392 and hence each @xmath30 splits into @xmath384 non - transversal vertical lines satisfying the genericity condition at only one point .
this shows the statement . in the more general case when triples @xmath103 are just generic at infinity , the change of projection might lead to more complicated situations , but in general the result is of the same nature ,
that is , some braids produced by the monodromy around @xmath147 in the non - generic braid monodromy need to be replaced by an appropriate factorization .
we state some interesting particular cases .
[ prop - cusp ] let us assume that @xmath30 intersects @xmath12 transversally at @xmath393 points and it is tangent to an ordinary cusp .
after conjugation the monodromy around @xmath394 produces the braid @xmath395 .
moreover , @xmath396 can be replaced by the factorization @xmath397 so that the monodromy becomes generic in a regular neighborhood of @xmath30 .
the local equation of @xmath32 at @xmath398 is @xmath399 .
note that the local monodromy around @xmath400 is given by the braid parametrized by @xmath401 , @xmath402 $ ] , that is , a rotation of angle @xmath403 of the 3-rd roots of unity .
this produces the braid @xmath404 . a small perturbation of the projection point along @xmath9 produces a change of variable @xmath405 .
the discriminant of @xmath406 with respect to @xmath41 is @xmath407 , which implies that the non - transversal vertical line @xmath175 splits into @xmath408 and @xmath409 , where @xmath410 .
it is straightforward to check that there is a cusp at @xmath411 ( whose tangent @xmath412 is not vertical if @xmath413 ) , a simple point at @xmath414 , a vertical tangency at @xmath415 , and a simple point at @xmath416 ( see ) .
regardless of the value of @xmath417 , there is a geometric basis on a local disk @xmath418 centered at @xmath400 and a choice of generators @xmath419 and @xmath420 in @xmath421 such that the local braid monodromy around @xmath400 is given by @xmath422 .
therefore , since the product should be @xmath404 and @xmath423 , then the local monodromy around @xmath424 should be given by @xmath425 , which concludes the proof .
similar computations provide the following results for the simplest cases : [ prop - double ] let us assume that @xmath30 intersects @xmath12 transversally at @xmath393 points and it is tangent at one of the local branches of an ordinary double point .
after conjugation the monodromy around @xmath394 produces the braid @xmath426 .
moreover , @xmath396 can be replaced by the factorization @xmath427 so that the monodromy becomes generic in a regular neighborhood of @xmath30 .
[ prop - tangency ] let us assume that @xmath30 intersects @xmath12 transversally at @xmath428 points and it is tangent to an inflection point of order @xmath429 .
after conjugation the monodromy around @xmath394 produces the braid @xmath430 .
moreover , @xmath396 can be replaced by the factorization @xmath431 so that the monodromy becomes generic in a regular neighborhood of @xmath30 .
we are also interested in the case where @xmath12 is the union of a fully horizontal curve @xmath432 ( such that @xmath135 ) and some lines @xmath433 passing through @xmath11 and different from @xmath9 . in order to avoid technical problems ,
we restrict our attention to the case of line arrangements , which admits two approaches .
one of them is to use wiring diagrams @xcite ( which is an invariant equivalent to braid monodromy ) .
the other is more direct but we need some definitions .
let @xmath434 and let @xmath435 .
the @xmath436-_shift _ of @xmath437 into @xmath438 is the inclusion @xmath439 such that @xmath440 , @xmath268 .
the _ partial garside element _ of the strings @xmath441 , @xmath442 , is the image @xmath443 of the garside element by the @xmath444-shift of @xmath445 into @xmath437 .
let @xmath446 be a line arrangement such that @xmath11 is a point of multiplicity @xmath167 and @xmath447 .
let @xmath9 be a generic line through @xmath11 .
we want to construct a generic braid monodromy for @xmath12 starting from the one of @xmath448 , where @xmath449 . recall that this braid monodromy is obtained as a representation @xmath450 where @xmath39 is the set of @xmath22-coordinates of the multiple points of @xmath451 .
let @xmath452 , where @xmath453 ; since the braids associated with the meridians around the points in @xmath454 are trivial , the above mapping defines a representation @xmath455 ( which will be referred to as the _ augmented braid monodromy _ ) .
a choice of an ordered geometric basis @xmath106 allows to represent this braid monodromy by @xmath456 ; let @xmath457 be the indices of the braids corresponding to the vertical lines @xmath458 .
[ prop - def - lines ] under the above conditions , let us decompose @xmath153 as @xmath459 where @xmath460 is a product of squares of partial garside elements of @xmath437 .
then , a generic braid monodromy for @xmath12 is obtained by replacing the braids @xmath153 as follows : 1 .
[ prop - def - lines1 ] if @xmath461 ( by convention @xmath462 , @xmath463 ) then replace @xmath153 by the sequence @xmath464 and we take out the trivial braids ( for @xmath465 such that @xmath466 ) .
[ prop - def - lines2 ] if @xmath467 then replace @xmath153 by @xmath468 3 .
[ prop - def - lines3 ] finally , add @xmath469 . for simplicity , we can assume that @xmath470 , since we may add vertical lines to the arrangement and once the generic braid monodromy is obtained we may forget the strings corresponding to the added lines ; it is easily seen that this does not affect to the final result .
the augmented braid monodromy @xmath471 described above may be computed as follows .
fix a generic line @xmath472 through @xmath11 close to @xmath473 and fix a base point @xmath474 close to @xmath11 .
we fix an ordered geometric basis @xmath475 of @xmath476 such that @xmath76 is a meridian of @xmath477 and its pseudo - coxeter element is the the boundary of a disk @xmath478 in @xmath479 surrounding @xmath480 ( which is also the negative boundary of a disk in @xmath472 centered at @xmath11 ) . in the affine plane @xmath481 , we consider the horizontal line @xmath482 passing through @xmath483
. we can choose @xmath483 such that there is a disk @xmath484 containing @xmath485 in its interior .
we fix an ordered geometric basis @xmath486 of @xmath487 such that : * the meridians @xmath488 and @xmath76 are equal in @xmath489 ( and we identify them from now on ) .
the product @xmath490 is the boundary of a disk @xmath491 which is isotopic to @xmath478 ( in @xmath492 ) and disjoint to @xmath484 ( see ) . *
the meridians @xmath493 are obtained as follows .
pick a base point in @xmath494 ; all the meridians have as common part a path joining @xmath483 with this point avoiding ( counterclockwise ) the disk @xmath491 .
these meridians project onto the meridians @xmath149 .
it is then clear that @xmath495 in @xmath496 .
next step is to deform the projection point in @xmath473 which induces a family of coordinate changes @xmath497 for a fixed @xmath498 and @xmath499 $ ] . for each new projection point
we obtain a braid monodromy @xmath500 , @xmath499 $ ] .
we can fix small pairwise disjoint disks @xmath501 ( respectively centered at @xmath502 ) such that @xmath503 is contained in the interior of @xmath504 .
note that @xmath505 . in order to compute the braid monodromy @xmath506 we choose an ordered geometric basis @xmath507 , such that @xmath508 .
the generic base fiber is the new vertical line passing through @xmath483 . in this vertical line
we can construct disks isotopic to @xmath491 and @xmath484 containing respectively the intersections with @xmath509 and @xmath510 .
[ fig : deformacion - rectas ] we start with the computation of @xmath511 . the decomposition @xmath512 has a geometric meaning : @xmath335 is the braid starting from the vertical line through @xmath483 and ending in a vertical line close to @xmath513 while @xmath514 is the braid obtained by turning around @xmath513 as justifies .
let us produce a similar decomposition for @xmath515 in @xmath438 .
the motion producing @xmath335 induces now a braid @xmath516 as the product of two commuting braids : @xmath335 and @xmath517 , see .
the first one corresponds to the motion of the points in the disk @xmath491 and the second one is obtained by considering the motion of the @xmath518 point to the boundary ( behind the other ones ) , and then how this point approaches to @xmath491 . .
] the braid @xmath519 is decomposed as a product of conjugate of partial garside elements leading to the images of @xmath520 as in [ prop - def - lines2 ] , see . for the case
@xmath521 , the decomposition in [ prop - def - lines1 ] is obtained is the same one by forgetting the strings corresponding to the verticals line not in @xmath32 .
the final step consists in the choice of a new line at infinity close to @xmath473 ( without changing the projection point ) ; the last lines are no more parallel since the multiple point @xmath11 becomes affine and we must add the braid of [ prop - def - lines3 ] .
let us consider the line arrangement of . in order to obtain the augmented braid monodromy @xmath471 it is necessary to add an extra vertical line @xmath522 passing through the double point @xmath523 of @xmath32 .
it is easy to see that @xmath471 is given by @xmath524 .
the generic braid monodromy is given by : @xmath525
let @xmath12 be a ( reduced ) projective curve of degree @xmath167 of equation @xmath526 and let @xmath186 be its transform by a kummer cover @xmath176 , @xmath527 .
note that @xmath186 is a projective curve of degree @xmath528 of equation @xmath529 .
another obvious remark is that if @xmath12 is reducible so is @xmath186 .
the converse is not true as we will see in [ sec - ex ] .
we will briefly analyze the singularities of @xmath186 in terms of @xmath12 . for convenience
we distinguish three types of points in @xmath4 .
let @xmath530 such that @xmath531 $ ] .
we say that @xmath398 is a point of _ type @xmath532 _ ( or simply of _ type @xmath533 _ ) if @xmath534 . if @xmath535 but @xmath536 the point is said to be of _ type @xmath537 _
( types @xmath538 and @xmath539 are defined accordingly ) .
such points will also be referred to as _ type @xmath23 _ points . the corresponding line ( either @xmath540 , @xmath541 , or @xmath542 ) a type-@xmath23 lies on will be referred to as their _
axis_. the remaining points @xmath543 $ ] , @xmath14 $ ] , and @xmath544 $ ] will be called _ vertices _ ( or type 0 points ) and their axes are the two lines ( either @xmath545 , @xmath173 , or @xmath546 ) they lie on .
note that a point of type @xmath436 , @xmath547 in @xmath4 has exactly @xmath548 preimages under @xmath176 .
it is also clear that the local type of @xmath549 at any two points on the same fiber are analytically equivalent .
[ lema - local ] let @xmath530 be a point of type @xmath436 and @xmath550.there exist local coordinates @xmath551 and @xmath552 centered at @xmath553 and @xmath398 , respectively , such that : 1 . if @xmath554 , then @xmath555 .
2 . if @xmath556 , then @xmath557 , where @xmath558 and @xmath559 are the local equations ( at @xmath553 and @xmath398 , respectively ) of the axes containing the points . 3 . if @xmath560 , then @xmath561 , where @xmath562 and @xmath563 are the local equations of the axes containing @xmath564 . for the points of type @xmath533 ,
it is a straightforward consequence of fact that the cover @xmath176 is unramified at @xmath398 .
if @xmath398 is of type @xmath23 , let us assume that @xmath565 $ ] , @xmath566 , and then @xmath567 $ ] .
we choose @xmath568 and @xmath569{t^n+y})$ ] .
if @xmath398 is of type @xmath185 , we may assume @xmath570 $ ] ; we keep @xmath571 as coordinates for @xmath398 and @xmath553 .
[ prop - sing ] let @xmath530 be a point of type @xmath436 and @xmath550 .
one has the following : 1 . if @xmath554 , then @xmath572 and @xmath573 are analytically isomorphic .
2 . if @xmath556 , then @xmath573 is a singular point of type @xmath23 if and only if @xmath574 , where @xmath575 and @xmath576 is the axis of @xmath398 .
3 . if @xmath560 , then @xmath573 is a singular point .
it is a straightforward consequence of .
assume @xmath577 are points of type @xmath436 on @xmath578 and @xmath579 is a local equation of @xmath580 .
if @xmath554 , then @xmath581 is a local equation of @xmath582 . if @xmath556 then @xmath583 is a local equation of @xmath582 . finally ,
if @xmath560 , then @xmath584 is a local equation of @xmath582 .
[ ex - singpin ] in some cases , we can be more explicit about the singularity type of @xmath582 . 1 . if @xmath398 is of type @xmath23 , @xmath572 is smooth and @xmath575 then @xmath573 has the same topological type as @xmath585 . in particular ,
if @xmath340 , then @xmath573 is of type @xmath586 .
2 . if @xmath398 is of type @xmath185 , @xmath572 is smooth and transverse to the axes then @xmath573 is an ordinary multiple point of multiplicity @xmath189 .
in order to better describe singular points of type @xmath185 and @xmath23 of @xmath217 we will introduce some notation .
let @xmath530 be a point of type @xmath587 and @xmath550 a singular point of @xmath217
. denote by @xmath588 ( resp .
@xmath589 ) the milnor number of @xmath218 at @xmath398 ( resp .
@xmath217 at @xmath553 ) , also , denote by @xmath590 the local branches of @xmath218 at @xmath398 and consider @xmath591 .
define @xmath592 ( resp .
@xmath593 ) as the milnor number of the singularity @xmath594 ( resp .
@xmath595 ) . since @xmath587
, then @xmath398 and @xmath553 belong to either exactly one or two axes . if @xmath398 and @xmath553 belong to an axis @xmath576 , then @xmath596 and @xmath597 ( analogous notation for @xmath553 and @xmath598 ) .
more specific details about singular points of types @xmath185 and @xmath23 can be described as follows : [ prop - sing - kummer ] under the above conditions and notation one has the following : 1 .
[ prop - sing - kummer1 ] for @xmath556 , @xmath398 belongs to a unique axis @xmath576 and 1 .
@xmath599 , 2 .
@xmath600 , 3 .
if @xmath601 then @xmath602 , 4 . if @xmath603 , then @xmath598 has @xmath604 irreducible components . 2 .
[ prop - sing - kummer0 ] for @xmath560 , @xmath398 belongs to exactly two axes @xmath605 and @xmath606 1 .
@xmath607 , 2 .
@xmath608 , @xmath609 , 3 .
if @xmath601 then @xmath610 , 4 .
if @xmath611 , then @xmath598 has @xmath612 irreducible components which are analytically isomorphic to each other for any fixed @xmath157 .
for part note that @xmath613 the same proof applies to @xmath593 .
also @xmath614 where @xmath615 ( resp .
@xmath616 ) is a local equation for @xmath617 ( resp .
@xmath618 ) . finally , we can describe the irreducible branch @xmath617 as a puiseux factorization of type @xmath619 , where @xmath620 is a multiple of @xmath621 , the order of @xmath615 in @xmath41 .
note that @xmath622 .
assume @xmath623 is a common divisor , that is , @xmath624 , @xmath625 ( and hence @xmath626 ) then @xmath627 is invariant under the galois conjugation by @xmath628-roots of unity and hence it is a convergent series in @xmath629 , that is , a union of branches
. therefore the result follows .
the proof of part [ prop - sing - kummer0 ] is analogous .
in this section the results obtained in [ sec - pencils]-[sec - cover - braids ] will be applied to produce generic braid monodromies of curves .
we will follow the conventions introduced in @xcite . in the language of the previous section , diagram systems on @xmath43 for the starting curve @xmath32
will be chosen by joining the points with segments in a decreasing order according to the lexicographic order .
we start with the braid monodromy of a smooth curve considered as a fermat curve @xmath630 , via the @xmath189-th kummer cover of the line @xmath12 of equation @xmath631 . using the conventions of [ sec - pencils ]
, one can obtain an extended braid monodromy @xmath632 , where @xmath633 is a geometric basis of the free group @xmath208 and @xmath634 , @xmath635 as shown in . using and the commutative diagram
, one can obtain a braid monodromy @xmath636 for @xmath549 as follows : @xmath637 where the basis @xmath132 , @xmath638 is obtained as in , by forgetting the last term since @xmath175 is transversal .
however , note that @xmath639 is not a generic braid monodromy since the projection from @xmath11 contains @xmath189 special fibers @xmath640 , @xmath641 with an order @xmath189 tangency at @xmath642 $ ] .
still , @xmath639 is useful to compute the fundamental group of the complement of a smooth curve ( using ) .
this group is generated by @xmath643 with a relation @xmath644 and since @xmath645 the group is cyclic of order @xmath7 , as obtained by zariski in @xcite . in @xcite
zariski showed that the fundamental group of a sextic with six cusps on a conic is @xmath648 and the family of such sextics is irreducible , so they all have equivalent braid monodromies .
we are going to compute one in two steps : first a ( fully horizontal ) non - generic braid monodromy of such a sextic will be obtained from a simpler curve via a kummer covering and second , a deformation will be performed to compute the desired generic braid monodromy .
consider a conic as in .
it is tangent to two of the axes and transversal to the third one , e.g. , the conic @xmath649 . using the conventions of [ sec - pencils ] , the projection from @xmath11 has three non - generic fibers at @xmath650 .
consider @xmath651 and @xmath652 meridians around @xmath533 , @xmath23 , and @xmath185 resp .
( see ) forming an ordered geometric basis of the free group @xmath208 .
one can obtain an extended braid monodromy @xmath653 as follows @xmath654 let us consider the @xmath657-rd kummer cover of @xmath12 .
the curve @xmath658 is a curve of degree @xmath659 .
since @xmath12 is smooth and intersects transversally @xmath9 , the singular points are contained in @xmath660 ( ) . using , the curve @xmath658 possesses six ordinary cusps ( which are in a conic , namely the union of the lines @xmath661 and @xmath662 ) . using a radial system of generators as in
[ sec - radial ] , the map @xmath663 is given ( see ) by : @xmath664 where the generators @xmath665 on the right - hand side are defined as @xmath666 ( with the right - lexicographic order ) for simplicity .
let us conjugate the result by @xmath667 : @xmath668 as mentioned above , this braid monodromy is not generic .
however , the triple @xmath671 is fully horizontal and transversal at infinity
. one can deform @xmath672 as described in
[ sec - deformation ] .
first , a deformation is performed to obtain as in [ sec - gi - lg ] to produce a locally generic triple .
then , a move as described in [ sec - lg - gen ] will produce a generic braid monodromy . also note that the non - generic fibers of the projection from @xmath11 are @xmath673{2},t_2=1,t_3=\xi_3\sqrt[3]{2},t_4=\xi_3,t_5=\xi_3 ^ 2\sqrt[3]{2},t_6=\xi_3 ^ 2,t_7=0\},\ ] ] where @xmath674 is a meridian around @xmath262 and : 1 .
[ enum - lg1 ] @xmath675 , @xmath676 correspond to vertical lines containing three simple tangencies ( hence locally generic fibers ) .
[ enum - nlg ] @xmath675 , @xmath677 correspond to vertical lines passing tangent to an ordinary cusp and transversal to three smooth points of @xmath678 .
[ enum - lg2
] @xmath679 corresponds to a vertical line passing through three ordinary cusps ( hence a locally generic fiber ) .
this step is necessary at the fibers of type above . according to the braid obtained as image of @xmath680 ,
that is , @xmath681 has to be replaced by @xmath682 analogously can be done with @xmath683 and @xmath684 .
this is necessary at the locally generic fibers described above .
applying to the fibers of types and one obtains that , for instance , @xmath685 is replaced by @xmath686 analogously can be done with @xmath687 , @xmath688 , and @xmath689 .
however , instead of working out each @xmath690 separately we can note that @xmath691 , ( @xmath692 , @xmath693 ) .
therefore one can also work out the first set of braids coming from @xmath694 and @xmath695 and conjugate to obtain the rest of the braids coming from @xmath690 @xmath696 .
finally attach the braids @xmath697 obtained from @xmath689 . from and
one obtains @xmath698 using the sequence of hurwitz moves @xmath699 ( see ) one obtains the first block : this result is also obtained in ( * ? ? ?
* theorem 1(3 ) , p. 160 ) .
it is straightforward to compute the fundamental group of the complement of @xmath658 and to retrieve zariski s computation .
moreover , this braid monodromy allows us to compute the homotopy type of @xmath707 ( for a generic choice of line at infinity ) using the method in @xcite .
it is easy to see that @xmath708 consider a conic @xmath710 as in . projecting from @xmath711 $ ] as usual ,
one obtains two non - generic fibers @xmath473 and @xmath679 with tangencies @xmath712 $ ] and @xmath713 $ ] respectively .
also note that @xmath32 is tangent to @xmath714 at @xmath715 $ ] . using the conventions of [ sec - pencils ] , an extended braid monodromy @xmath653 , where @xmath716 , @xmath633 are meridians around @xmath717 and @xmath718 respectively forming an ordered basis of the free group @xmath208 and @xmath719 , @xmath720 . for simplicity , we conjugate this monodromy by @xmath721 and obtain @xmath722 the preimage of @xmath578 by the kummer cover of order @xmath657 is a sextic @xmath658 with nine cusps as it can be deduced from .
let us compute the braid monodromy @xmath723 .
following the ideas in [ subsec - zs ] one immediately obtains @xmath724 since the line @xmath473 is not generic , one can apply to obtain a braid monodromy that is generic at infinity .
the pseudo - coxeter element for the monodromy is @xmath725 and hence @xmath726 , whose image in @xmath727 via @xmath728 ( see ) is @xmath729 .
therefore @xmath730 is a braid monodromy factorization of @xmath731 generic at infinity for the zariski sextic @xmath658 . finally , in order to obtain a generic braid monodromy one needs to apply propositions [ prop - gen2 ] and [ prop - gen3 ] to slightly turn the projection generic and perform hurwitz moves to simplify the braids .
one can check that the final generic braid monodromy of @xmath732 is : @xmath733 the fundamental group of this curve complement has been extensively studied by zariski in @xcite .
he also used deformation arguments to recover the fundamental group studied in @xcite , see [ subsec - zs ] , as well as to study the fundamental group of sextics with six cusps not on a conic .
similar arguments can be used to describe the generic braid monodromy of any curve @xmath734 , @xmath189 odd , which is an irreducible curve of degree @xmath735 and @xmath736 singularities of type @xmath737 ( see @xcite ) .
we are going to use the classical ceva arrangement ( six lines joining four points in general position ) to find the braid monodromy of the @xmath709-ceva arrangement @xmath738 and it can be obtained as a kummer covering of three concurrent lines in a classical ceva arrangement , see .
another direct application is the monodromy of the maclane arrangement @xcite , which is obtained from @xmath732 by deleting one line .
this results in deleting one of the strings , that is , @xmath744 the computational difficulty of the braid monodromy of the @xmath709-ceva as well as the maclane arrangements comes from the fact that they can not be given by real equations ; maclane arrangement is the smallest one with this property . using our construction everything
is reduced to computing the braid monodromy of a very simple line arrangement .
braid monodromy for generalized ceva arrangements @xcite ( @xmath745 ) can also be obtained .
let us consider the following cubic @xmath746 , whose real picture appears in .
the usual projection from @xmath11 has four non - generic fibers @xmath747 at @xmath748 , where @xmath473 and @xmath679 are tangent lines at inflection points and @xmath605 is the vertical line through the node ( which is real , but whose branches are not ) . for this reason the real picture is not enough to recover the braid monodromy of @xmath749 .
however , the extra information required can be obtained from drawing the real part of because of the missing complex conjugate branches ( shown in as dotted lines ) . in order to find these dotted lines one can proceed as follows .
let @xmath750 $ ] and assume it has only one real root , say @xmath751 .
let @xmath752 be the remaining roots , @xmath753 .
note that their common real part is @xmath754 .
[ lema - real - roots ] under the above conditions , the polynomial @xmath755 contains exactly one real root , which is the common real part @xmath754 of the complex conjugate roots @xmath756 , @xmath757 of @xmath758 .
since the affine cubic @xmath765 satisfies the conditions of for any fixed @xmath766 the real parts of the complex roots of @xmath765 are given by the equation : @xmath767 this is enough to compute the required braid monodromy , see @xcite for details . in order to show the extended braid monodromy @xmath184 of @xmath32 note that @xmath173 is also tangent to @xmath32 at the inflection point @xmath768 . hence @xmath769 .
choose a geometric basis as in and a system diagram on the fiber given by decreasing lexicographic order on @xmath43 .
for instance , according to @xmath770 is the half - twist exchanging the two complex conjugated roots of @xmath771 , @xmath772 is the half - twist exchanging the complex conjugated root of @xmath771 of negative real part and @xmath185 .
then , @xmath773 is given by : @xmath774 in order to meet the hypothesis of [ sec - pencils ] one needs the system diagram to be such that the last string is the one that remains constant .
this means one needs to conjugate @xmath184 by @xmath775 .
combining this with a conjugation by @xmath776 , the hurwitz move @xmath777 ( changing @xmath778 to @xmath149 ) , and another conjugation by @xmath779 one obtains : the dual curve of a nodal quartic is a sextic curve with six cusps and four nodes . moreover
, this curve appears as a generic plane section of the discriminant of the polynomials of degree @xmath783 and its fundamental group has been computed by zariski @xcite : it is the braid group on @xmath784 with four strings
. this sextic can also be obtained from the cubic in [ subsec - useful ] using a kummer cover @xmath785 . each inflection point in the axes produces two cusps while the double point produces four nodes , see . in order to give a braid monodromy for @xmath786 from one
needs to combine @xmath787 ( see ) and .
we start with the first part of : @xmath788 using @xmath787 we obtain the braid monodromy of @xmath789 : @xmath790 we can compute the fundamental group of @xmath793 .
it is well known @xcite that this group is obtained as a central extension of the braid group of @xmath784 with four strings by @xmath794 .
more precisely , applying ( without the relation @xmath795 ) we obtain a group with generators @xmath796 and relators @xmath797=[\mu_2 ^ 2,\mu_5\mu_4^{-1}]=1,\mu_2\mu_4\mu_2=\mu_4\mu_2\mu_4,\mu_2\mu_5\mu_2=\mu_5\mu_2\mu_5,\\ \mu_6=\mu_2^{\mu_5},\mu_1=\mu_4*\mu_2,\mu_3=\mu_5^{\mu_2\mu_4},\underbrace{1=1}_{7\text { times}}. \end{split}\ ] ] this presentation is obtained using ` gap4 ` @xcite . from the original presentation
we obtain this one only using tietze moves of type i and ii .
hence , by libgober s result @xcite , one can verify that @xmath793 has the homotopy type of @xmath798 , where @xmath799 is the @xmath533-complex associated with the presentation @xmath800=[y^2,x z^{-1}]=1,x y
x = y x y , y z y = z y z \right.\right\rangle.\ ] ] it can be obtained from the cubic in [ subsec - useful ] using a kummer cover @xmath801 : it is the preimage of the cubic and the three ramification lines . it can be seen using and , and also directly from the equations : @xmath802 hence @xmath803 and the above factors give the four reducible members of the pencil of cubics generated by @xmath804 and @xmath803 .
the base points of this pencil are the nine common inflection points of the irreducible cubics of the pencil .
the reducible fibers split as product of lines : @xmath805 these are the lines joining the inflection points , i.e. , we obtain hesse arrangement .
we start with the braid monodromy with respect to @xmath806 : @xmath807 , and the braid at infinity is @xmath808 . in order to simplify the braid monodromy of the hesse arrangement we conjugate by @xmath809 and
we obtain : @xmath810 we proceed as in [ subsec - nodal ] but we perform a hurwitz move to the base of in order to obtain simpler braids : @xmath811 let us recall the map @xmath812 of : @xmath813 for convenience we rewrite this map in the usual generators : @xmath814 it is worthwhile to note that @xmath815
. let us denote @xmath816 and @xmath817 .
next step is to write down the braid monodromy of @xmath806 : @xmath818 this monodromy becomes generic at infinity adding @xmath819 .
let us denote @xmath820 . to compute a generic braid monodromy to this arrangement one has to apply where we have two vertical lines corresponding to the fifth and eighth braids : @xmath821
the braid monodromy was also obtained ( computer - aid ) in @xcite .
[ [ a - smooth - cubic - and - its - inflection - points - and - sextics - with - six - cusps - outside - a - conic ] ] a smooth cubic and its inflection points and sextics with six cusps outside a conic ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the kummer cover @xmath801 of the curve @xmath12 of equation @xmath822 produces a fermat cubic ( as the preimage of the line @xmath823 ) while the other three lines become the tangent lines to the nine inflection points
. the extended braid monodromy of @xmath749 is given by @xmath824 in order to meet the hypothesis of [ sec - pencils ] one needs both that the last generator in the ordered geometric basis corresponds to @xmath825 and that the system diagram is such that the last string is the one that remains constant .
one can meet the first requirement by using the hurwitz move @xmath777 : @xmath826 in order to switch the string @xmath827 to the last position one must perform the automorphism @xmath828 , but these braids are invariant by the morphism .
after these changes the braid around infinity is @xmath829 .
using one can apply . in @xcite sextics with six cusps not on a line have been constructed by means of 3rd kummer cover where the three lines are tangent lines to a smooth cubic at non - aligned inflection points .
hence we will be interested only in the three tangent lines with these property , namely , two preimages of @xmath830 , one preimage of @xmath831 and none of @xmath832 .
hence , we can forget the second string ( recall the automorphism ) and we obtain ( the first braid disappears since it is trivial ) : @xmath833 and hence the braid around infinity is @xmath721 .
we have to forget some strings in the mapping @xmath834 .
this causes the morphism to not be well defined in @xmath739 , but only in the pure braid group : @xmath835 hence the braid monodromy of the cubic with the tangent lines is given by : @xmath836 where the first two braids correspond to the tangent lines in the curve while the second string corresponds to the remaining tangent line .
some operations are need : conjugation by @xmath837 , cyclic permutation , permutation of the order of the strings and forgetting one braid ( at infinity ) : @xmath838 the braid at infinity being @xmath839 .
this result coincides with the one obtained using carmona s program @xcite .
we first consider the braid monodromy factorization induced by this braid monodromy after a double cover of the base , that is , @xmath840 : @xmath841 then perform the morphism @xmath342 ( see ) and a generic deformation , resulting in : @xmath842 a straightforward computation shows that @xmath843 . using @xcite one can check that @xmath844 f. catanese and b. wajnryb , _ the 3-cuspidal quartic and braid monodromy of degree 4 coverings _ , projective varieties with unexpected properties , walter de gruyter gmbh & co. kg , berlin , 2005 , pp .
113129 .
cogolludo - agustn and r. kloosterman , _ mordell - weil groups and zariski triples _ , geometry and arithmetic ( g. farkas c. faber and r. de jong , eds . ) , ems congress reports , europ . math .
soc . , 2012 , also available at ` arxiv:1111.5703 [ math.ag ] ` .
i. dolgachev and a. libgober , _ on the fundamental group of the complement to a discriminant variety _ , algebraic geometry ( chicago , ill . , 1980 ) , lecture notes in math .
862 , springer , berlin , 1981 , pp .
125 .
moishezon and m. teicher , _ braid group techniques in complex geometry .
iv . braid monodromy of the branch curve @xmath848 of @xmath849 and application to @xmath850 _ , classification of algebraic varieties ( laquila , 1992 ) , contemp .
162 , amer .
soc . , providence , ri , 1994 , pp . | in this work we describe a method to reconstruct the braid monodromy of the preimage of a curve by a kummer cover .
this method is interesting , since it combines two techniques , namely , the reconstruction of a highly non - generic braid monodromy with a systematic method to go from a non - generic to a generic braid monodromy .
this `` _ _ generification _ _ '' method is independent from kummer covers and can be applied in more general circumstances since non - generic braid monodromies appear more naturally and are oftentimes much easier to compute .
explicit examples are computed using these techniques . |
wide importance is now attached to the study of higher dimensional spaces as the arena of physical phenomena . however , as we go beyond spacetime , the topological possibilities introduce a new dimension to the difficulty of attaching physical significance to various manifolds . even at five dimensions , asymptotically flat stationary vacuum
black holes are not unique in the sense that horizons of topology @xmath0 @xcite in addition to @xmath1 @xcite are now known . however , in the simpler static case , it is known that asymptotically flat static vacuum black holes are unique @xcite and given by the tangherlini generalization of the schwarzschild vacuum @xcite .
these uniqueness properties suggest that these solutions represent the natural generalization of the schwarzschild spacetime . however , these are not the only asymptotically flat quasi - static vacua which can be considered spherically symmetric " . here
we are concerned with 5-manifolds which admit only the 2- sphere ( unlike the tangherlini vacua which admit the 3-sphere ) and , in addition , two hypersurface orthogonal killing vectors at least one of which is assumed timelike .
these spaces are widely studied in the literature from various physical approaches @xcite but the approach used here is purely geometrical and the results therefore applicable to any physical approach . the only assumption made , aside from the symmetries stated , is @xmath2 where @xmath3 is the 5-dimensional ricci tensor .
to set the notation let @xmath4 represent a killing vector field , @xmath5 and if , for the coordinate @xmath6 is adapted ( @xmath7 ) , represent the associated field by @xmath8 .
if , in addition to ( [ killing ] ) , @xmath4 also satisfies the hypersurface orthogonality condition @xmath9}=0 , \label{hyper}\ ] ] write the field as @xmath10 .
the traditional definition of a static region of spacetime is one which admits a timelike @xmath10 .
this terminology is also used here and we distinguish the adapted coordinate @xmath11 via timelike @xmath12 .
the spaces considered here admit an additional hypersurface orthogonal killing vector and we distinguish the adapted coordinate @xmath13 via @xmath14 .
( any internal properties of @xmath13 are therefore of no concern . ) in addition to these symmetries we assume 2-dimensional spherical symmetry .
that is , the spaces can be decomposed as @xmath15 where @xmath16 so that we also have the killing field @xmath17 and @xmath18 where @xmath19 and @xmath20 are non - zero constants and the other components of this @xmath4 are zero .
this @xmath4 of course includes @xmath21 . here
@xmath22 is independent of @xmath23 and @xmath24 and if @xmath25 we refer to this circumstance as an origin " .
with the foregoing symmetries it follows that ( [ ricci ] ) is satisfied by @xmath26 where @xmath27 and @xmath28 are non - zero constants , @xmath29 and @xmath30 are constants ( not both zero ) , @xmath31 ( assumed monotone ) , @xmath32 where @xmath33 is a non - zero constant , @xmath34 and @xmath35 where @xmath36 and @xmath37 are constants restricted by the relation @xmath38 to view ( [ sform ] ) as static we take @xmath39 ( unless otherwise noted ) and to view ( [ sform ] ) as an augmentation of spacetime ( with signature + 2 ) we take @xmath40 . we retain both signs of @xmath27 and
so allow the doubly static cases @xmath41 .
note that the forms of @xmath42 and @xmath43 remain unchanged under the interchange @xmath44 @xcite .
the reason for this symmetry here is that the adapted coordinates @xmath13 and @xmath11 are ( prior to setting @xmath39 ) interchangeable . since the magnitude of @xmath27 ( and of course @xmath28 )
can be absorbed into the scale of @xmath13 ( and @xmath11 ) , the form ( [ sform ] ) would appear at first sight to admit three independently specified constants , in addition to @xmath33 , via ( [ consistency ] ) .
this , however , is not the case as we can , without loss in generality , set @xmath45 thus simplifying ( [ s(b ) ] ) to the form @xmath46 this is shown in the appendix .
there are then only two specifiable parameters ( @xmath33 and @xmath29 or @xmath30 subject to ( [ consistency ] ) with ( [ onehalf ] ) ) .
clearly the form ( [ sform ] ) allows an infinite number of representations given a monotone function @xmath47 subject to the reality of the resultant metric coefficients .
another approach is to assume a form for @xmath48 and solve the differential equation ( [ ss(b ) ] ) .
for example , if we set @xmath49 we obtain the solution given by gross and perry @xcite and davidson and owen @xcite .
similarly , setting @xmath50 we obtain the solution given recently by millward @xcite .
these representations are discussed in a more direct way below .
whereas ( [ ss(b ) ] ) can be solved exactly for a wide variety of choices for @xmath48 , we can of course use @xmath51 as a coordinate and view ( [ ss(b ) ] ) as the required coordinate transformation . with @xmath51 as a coordinate ( [ sform ] ) takes the simple form @xmath52 where now @xmath53 @xmath54 and @xmath55 the coefficient in ( [ f(h ) ] ) has been redefined for convenience .
an important aspect of what follows is the distinction , on geometrical grounds , of various members of the solution locus ( [ condition ] ) . whereas the from ( [ hform ] ) is convenient ( and as is shown below , complete in most cases ) , coordinate transformations bring the spaces into more familiar forms .
for example , with the transformation @xmath56 we obtain the form given by kramer @xcite . similarly , with the transformation @xmath57 we obtain form given by given by gross and perry @xcite ( the notation ( here to there ) is related by @xmath58 , and @xmath59 ) and by davidson and owen @xcite ( the notation ( here to there ) is related by @xmath60 , and @xmath61 ) . the solution given recently by millward
@xcite is a very special case .
it is given by @xmath62 with @xmath63 and @xmath64 .
in this section we study the geometric properties of the solutions ( [ hform ] ) .
the various solutions are distinguished in figure [ locus ] via the parameters @xmath29 and @xmath30 .
first , however , we discuss the interchange symmetry associated with the solution locus ( [ condition ] ) .
the solution locus ( [ condition ] ) is obviously invariant to the interchange @xmath65 @xcite . to see why , consider the coordinate transformation @xmath66 under the transformation ( [ inversion ] ) we obtain ( [ hform ] ) but with @xmath67 , @xmath68 , and @xmath69 .
moreover , in terms of figure [ locus ] , we have the quadrant interchanges @xmath70 and @xmath71 along with the point interchanges @xmath72 and @xmath73 . as a result ,
the form ( [ hform ] ) duplicates all distinct classes of solutions . as a matter of convenience
we continue with this form here but with knowledge of this duplicity .
before the properties of the solutions ( [ hform ] ) are discussed in terms of the parameters @xmath29 and @xmath30 , it is useful to review the general nature of the constant @xmath33 as it is quite distinct from @xmath29 and @xmath30 .
this is similar to ( the reciprocal of ) the mass " in the four - dimensional schwarzschild vacuum as one might well guess from the other representations discussed above and as we explain in detail below . to see this
let @xmath74 represent the 4 dimensional schwarzschild vacuum ( of mass @xmath75 ) and subject this to a conformal transformation @xmath76 where @xmath77 is a constant .
this , of course , preserves ricci flatness ( in any dimension ) .
it is easy to show that under the transformation ( [ conformal ] ) @xmath78 it is clear from the form ( [ hform ] ) then that @xmath79 plays a role similar to @xmath75 ( using the freedom in the scale of @xmath13 and @xmath11 ) .
the parameters @xmath29 and @xmath30 , however , distinguish the solutions ( [ hform ] ) in a rather different way as we now examine .
as explained above , we refer to @xmath80 as an origin . from ( [ hform ] )
we have @xmath81 so that in all cases @xmath82
. moreover , @xmath83 has a minimum ( @xmath84 ) at @xmath85 which restricts the minima to quadrants @xmath86 and @xmath87 of figure [ locus ] . in quadrant
@xmath86 the minima occur for @xmath88 and in quadrant @xmath87 they exist for @xmath89 . for the exceptional solutions @xmath90 and @xmath91 clearly @xmath92 and for the exceptional solutions @xmath93 and @xmath94 , @xmath95 .
properties of @xmath96 are summarized in tables [ tab : regular ] and [ tab : exceptional ] .
.[tab : regular]regular " solutions [ cols="^,^,^,^,>",options="header " , ] for the spaces ( [ hform ] ) there is only one independent invariant derivable from the riemann tensor without differentiation and this can be taken to be @xmath97@xmath98 where @xmath97 is the 5-dimensional weyl tensor .
this is given by @xcite @xmath99 where @xmath100 first note that for the exceptional cases @xmath90 and @xmath91 @xmath101 and for the exceptional cases @xmath93 and @xmath94 @xmath102 in all cases @xmath103 moreover , @xmath104 except for the exceptional cases @xmath90 and @xmath91 for which @xmath105 and @xmath106 except for the exceptional cases @xmath93 and @xmath94 for which @xmath107 .
the limit ( [ weylsingzero ] ) shows that in all but the two cases indicated , the solution ( [ hform ] ) is singular at @xmath108 .
similarly , the limit ( [ weylsing ] ) shows that in all but the other two cases indicated , the solution ( [ hform ] ) is singular as @xmath109 .
the properties discussed above are summarized in tables [ tab : regular ] and [ tab : exceptional ] .
since , as shown above , @xmath110 and since @xmath111 where @xmath112 is the 5-dimensional riemann tensor , all solutions are ( locally ) asymptotically flat for @xmath113 . in any subspace
@xmath115 constant it follows from ( [ hform ] ) that @xmath116 only for @xmath117 , that is a subspace of the exceptional solutions @xmath91 and @xmath94 .
these are black string " solutions @xcite .
moreover , @xmath118 where @xmath75 is the usual schwarzschild mass ( as we discuss further below ) . with the aide of table [ tab :
exceptional ] we see that for the solution @xmath91 , @xmath88 covers the static part of the spacetime with @xmath119 and all of the spacetime with @xmath120 for @xmath89 .
similarly , the solution @xmath94 with @xmath89 covers the static part of the spacetime with @xmath119 and all of the spacetime with @xmath120 for @xmath88 .
( in an analogous way , we can consider any subspace @xmath121 constant with @xmath122 and interchange @xmath90 for @xmath91 , @xmath93 for @xmath94 and @xmath29 for @xmath30 in the foregoing .
the exceptional cases @xmath90 and @xmath93 we call soliton " solutions @xcite . ) hypersurfaces of constant @xmath51 in subspaces of constant @xmath13 can be categorized by way of the trajectories they contain . here
we distinguish the spacelike ( @xmath123 ) , timelike ( @xmath124 ) and null ( @xmath125 ) cases where @xmath126 and @xmath127 is tangent to a curve in the hypersurface ( @xmath128 , where @xmath129 is any parameter that distinguishes events along the curve ) . from ( [ hform ] )
we find @xmath130 the nature of the singularities thus categorized are summarized in tables [ tab : regular ] and [ tab : exceptional ] where @xmath131 means @xmath132 if exactly radial ( @xmath133 ) .
write the momenta conjugate to @xmath134 and @xmath21 as @xmath135 and @xmath136 and define the constants of motion @xmath137 and @xmath138 .
the null geodesic equation now follows as @xmath139 where @xmath140 for a suitably scaled affine parameter @xmath129 and without loss in generality we have set @xmath141 . in general then the singularities discussed above are directional " in their visibility as there exist turning points in @xmath51 . in the
radial " case @xmath142 we have @xmath143 whereas ( [ radialnull ] ) can be integrated in terms of special functions , this integration is not needed .
we simply observe that there are no turning points on @xmath144 or on @xmath145 . as a result , only in the exceptional solutions @xmath90 and @xmath91 on @xmath144 and @xmath93 and @xmath94 on @xmath146 are the forms ( [ hform ] ) incomplete and also free of naked singularities in the ranges given @xcite .
it is clear from table [ tab : exceptional ] that the exceptional solutions @xmath90 and @xmath91 are incomplete on @xmath144 and the exceptional solutions @xmath93 and @xmath94 are incomplete on @xmath145 . since there is no geometrical reason to terminate these cases , we consider here simple static extensions . in these cases
( but clearly not in general ) we can consider @xmath147 .
first note that the discussion on asymptotic flatness given above extends to @xmath148 in these cases .
moreover , the results given in table [ tab : exceptional ] extend directly to @xmath147 .
it follows then that the cases @xmath90 and @xmath91 on @xmath149 and @xmath93 and @xmath94 on @xmath150 are singular - free .
however , as we know from the 4-dimensional case , these static extensions can be incomplete . for the black string solutions @xmath91 and
@xmath94 we can write @xmath151 where @xmath152 is any regular completion of the schwarzschild manifold so the space is no longer static via @xmath12 . if @xmath122 the space remains static via @xmath14 .
similarly , if @xmath122 for the soliton solutions @xmath90 and @xmath93 we can write @xmath153 where again @xmath152 is any regular completion of the schwarzschild manifold so the space is no longer static via @xmath14 but remains static via @xmath12 .
of course , the past spacelike singularities in @xmath152 render these solutions also nakedly singular , but in a way quite unlike the foregoing . not all singularities can be considered equally serious . here
we define the geometrical mass associated with spherical symmetry in order to classify the singularities discussed above .
define the quantity @xmath75 via the sectional curvature of the two - sphere @xcite , @xmath154 ( in 4 dimensions the well known effective gravitational mass of spherically symmetric spacetimes is given by ( [ mass ] ) with @xmath155 replaced by @xmath156 . )
it follows from ( [ hform ] ) that @xcite @xmath157 so that in all cases @xmath158 however , only in the exceptional solutions is @xmath75 a constant and then , @xmath159 the properties of @xmath75 in the other solutions are summarized in table [ tab : regular ] . in what follows
we classify the geometrical strength " of a singularity on the basis of @xmath75 .
most of the foregoing generalizes in a straightforward way to higher dimensions . for example , at dimension 6 , introducing another hypersurface - orthogonal killing vector @xmath160 , it follows ( retaining the previous notation ) that @xmath161 where @xmath162 and @xmath36 are constants , @xmath163 and ( [ s(h ) ] ) still holds , satisfies @xmath164 as long as the parameters remain on the ellipsoid @xmath165 of particular interest here is the fact that in any subspace @xmath166 constant it follows from ( [ hformsix ] ) that @xmath167 only for @xmath168 . these subspaces are precisely the spaces considered in this paper .
as the foregoing discussion makes clear , solutions on the locus ( [ condition ] ) have quite distinct geometrical properties , and not all of these solutions can be considered equally valid in any physical approach .
for example , solutions in the quadrants 3 and 4 have the strongest possible geometrical singularities at the origin ( @xmath75 diverges at @xmath25 ) , and these singularities are visible throughout the associated spaces for the entire range in @xmath51 . in quadrant 1 for @xmath144 and quadrant 2 for @xmath169 the solutions
not only lack an origin , they have the strongest possible geometrical singularities . in contrast , quadrant 1 solutions with @xmath146 and quadrant 2 solutions with @xmath144 have visible but the weakest possible geometrical singularities at the origin ( @xmath170 ) .
in contrast , the exceptional solutions have quite distinct properties .
solutions @xmath90 and @xmath91 with @xmath146 and solutions @xmath93 and @xmath94 with @xmath144 have naked singularities at the origin but of intermediate strength ( @xmath75 remains finite @xcite ) . only for the solutions @xmath90 and @xmath91 with @xmath171 and the solutions @xmath93 and @xmath94 with @xmath146 are the coordinates used in ( [ hform ] ) incomplete . in these cases
singularities in the non - static extensions are obtained . because of the very distinct properties of the solutions as one covers the solution locus ( [ condition ] ) , the natural conclusion is that the solution ( [ hform ] ) is fundamentally unstable about these exceptional solutions .
that is , the properties of the exceptional black string and soliton solutions are unstable to any metric perturbation . in general ,
if one was to envisage a physical theory in which ( [ hform ] ) was the external geometry of an object collapsing to zero volume , quadrant 1 solutions with @xmath145 ( equivalently quadrant 2 solutions with @xmath144 ) would seem to be the natural choice since their properties are stable ( away from the exceptional solutions ) and have the weakest possible , albeit naked , geometrical singularities at the origin @xcite .
finally , it was pointed out how the spaces considered here can be thought of as subspaces of a higher dimensional ricci - flat manifold of similar type .
this work was supported by a grant from the natural sciences and engineering research council of canada and was made possible by use of _ grtensoriii _ @xcite .
in this appendix we prove that we can , without loss in generality , set @xmath172 in ( [ sform ] ) thus simplifying ( [ s(b ) ] ) to the form ( [ ss(b ) ] ) .
we include @xmath28 for completeness .
define @xmath173 so that @xmath174 next , define @xmath175 by @xmath176 with these definitions ( [ sform ] ) transforms as follows : @xmath177 where @xmath178 and @xmath179 where @xmath180 and finally @xmath181 thus , with the relabelling @xmath182 and the removal of the @xmath183 , we obtain a space equivalent to the form ( [ sform ] ) but with @xmath172 . [
sec : texbooks ] r. emparan and h. s. reall , phys .
88 * , 101101 ( 2002 ) ( hep - th/0110260 ) .
r. c. myers and m. j. perry , annals phys . * 172 * , 304 ( 1986 ) .
s. hwang , geometriae dedicata * 71 * , 5 ( 1998 ) , g. w. gibbons , d. ida and t. shiromizu , prog .
suppl . * 148 * , 284 ( 2003 ) arxiv : gr - qc/0203004 f. r. tangherlini , nuovo cimento * 27 * , 636 ( 1963 ) .
explicit regular coordinates that cover all of the tangherlini solutions are given by k. lake , jcap 0310 ( 2003 ) 007 arxiv : gr - qc/0306073 see , for example , j. m. overduin and p. s. wesson , phys .
* 283 * ( 1997 ) arxiv : gr - qc/9805018 . for further discussion of this property see j. ponce de leon , class .
quantum grav .
* 23 * 3043 ( 2006 ) arxiv : gr - qc/0512067 d. j. gross and m. j. perry , nucl
* b 226 , * 29 ( 1983 ) .
our use of the term
soliton " here is consistent with this work .
a. davidson and d. owen , phys .
* b 155 * , 247 ( 1985 ) .
r. s. millward arxiv : gr - qc/0603132 d. kramer , acta phys . polon . * b2 * , 807 ( 1970 ) .
the significance of this symmetry ( along with @xmath184 explained previously ) has sometimes been considered obscure .
see , for example , p. s. wesson , _ space - time - matter _
( world scientific , singapore , 1999 ) .
this invariant has been given previously in different coordinates ( see , for example , the references in @xcite ) , but to our knowledge the discussion given here is more complete than any given previously .
moreover , the evolution in @xmath185 must be considered along with the geometrical mass @xmath75 as defined below .
it is well known that the black string solutions are unstable to large - scale perturbations .
see r. gregory , class .
quantum grav .
, * 17 * , l125 ( 2000 ) arxiv : hep - th/0004101 in the quadrants 1 and 2 it might be argued that the @xmath131 singularities could be entirely to the future and therefore not visible .
however , since these degenerate to ( single ) null singularities in the radial direction , they are necessarily naked .
this is a special case of the geometrical mass in @xmath186 dimensions for spaces admitting a @xmath162-sphere : @xmath187 see k. lake arxiv : gr - qc/0507031 ( to be updated ) .
the mass defined in @xcite ( and in particular see p. s. wesson and j. ponce de leon , class . quantum grav . *
11 * , 1341 ( 1994 ) ) is not equivalent to the geometrical mass ( [ mass ] ) .
the mass used in @xcite is , in our notation , @xmath188 .
the fact that @xmath170 in quadrants 1 ans 2 but @xmath75 remains strictly non - zero in the exceptional solutions means that no limiting procedure can be invoked .
zero mass naked singularities arise in standard general relativity .
see k. lake , phys .
* 68 * , 3129 ( 1992 ) .
this is a package which runs within maple .
it is entirely distinct from packages distributed with maple and must be obtained independently .
the grtensorii software and documentation is distributed freely on the world - wide - web from the address _ http://grtensor.org _
grtensoriii software is in development . | we examine , in a purely geometrical way , static ricci - flat 5-manifolds admitting the 2-sphere and an additional hypersurface - orthogonal killing vector .
these are widely studied in the literature , from different physical approaches , and known variously as the kramer - gross - perry - davidson - owen solutions .
the 2-fold infinity of cases that result are studied by way of new coordinates ( which are in most cases global ) and the cases likely to be of interest in any physical approach are distinguished on the basis of the nakedness and geometrical mass of their associated singularities .
it is argued that the entire class of solutions has to be considered unstable about the exceptional solutions : the black string and soliton cases . any physical theory which admits the non - exceptional solutions as the external vacuua of a collapsing object has to accept the possibility of collapse to zero volume leaving behind the weakest possible , albeit naked , geometrical singularities at the origin .
finally , it is pointed out that these types of solutions generalize , in a straightforward way , to higher dimensions . |
the main goal in output regulation is to find a controller such that the output of a given plant asymptotically follows a given reference signal generated by an exosystem .
it is known that a regulating feedback controller contains a built - in copy of the exosystem @xcite .
robustness of regulation is needed in order to make the controller work despite some perturbations of the plant , e.g. , parameter uncertainties and modelling errors .
if the controller is required to tolerate all arbitrarily small perturbations of the plant , then the internal model principle due to francis and wonham @xcite and davison @xcite states that if the plant has @xmath0-dimensional output space , then every robustly regulating controller must contain a @xmath0-fold copy or in short @xmath0-copy of the exosystem . in this paper
we study the robust regulation problem in a situation where the controller is only required to tolerate uncertainties from a restricted class @xmath1 of perturbations .
such a situation can arise due to several different reasons . in the simplest situation
, @xmath1 contains only a finite number of plants , for example , if our controller is required to function after specific component failures @xcite . in the case of only one possible failure , the original plant @xmath2 changes to a new plant @xmath3 and @xmath4 . on the other hand , the class @xmath1 becomes infinite in a situation where the values of some specific parameters of the plant are not known accurately @xcite .
our example in section [ sec : example ] illustrates the latter case . in the situation where robustness is only required with respect to a given class @xmath1 of perturbations , it is natural to ask if the controller must contain full @xmath0-copy internal model of the exosystem .
this problem was studied by paunonen in @xcite using state space methods .
it was shown in @xcite that the @xmath0-copy internal model guaranteeing robustness with respect to all small perturbations can be relaxed in many situations , and this observation leads to design of controllers with so - called _ reduced order internal models_. in this paper , we introduce frequency domain conditions for a controller to achieve output regulation and robustness with respect to a given class @xmath1 of perturbations .
our results give a precise meaning for reduced order internal models in the frequency domain .
in addition , we present methods for constructing controllers with reduced order internal models .
our constructions result in minimal complexity requirements for the number of copies built into the robustly regulating controllers .
we choose the reference signal to be of the form @xmath5 the class of signals that can be presented in this form is fairly large as it contains reference signals that are linear combinations of sinusoids , and in particular finite approximations of uniformly continuous periodic signals . in the first part of this paper we present our theoretical results .
our first main result shows that a stabilizing controller @xmath6 where @xmath7 is analytic at @xmath8 , is robustly regulating for the class @xmath1 of perturbed plants if and only if @xmath9 for all @xmath10 .
this is the frequency domain analogue of the time domain condition that was presented in @xcite .
the condition leads to our second main results that gives a lower bound for the ranks of the matrices @xmath11 in the controller .
in particular , if the plants have the same number @xmath0 of inputs and outputs and @xmath12 are invertible at @xmath13 , then the lower bounds for the ranks of @xmath11 in the controller that is robust with respect to the class @xmath1 are @xmath14 the controller constructions presented shows that the bounds @xmath15 are optimal , as our method results in controllers satisfying @xmath16 for @xmath17 . in the frequency domain
the controller containing a full @xmath0-copy internal model of the exosystem satisfies @xmath18 for all @xmath19 @xcite . in light of results ,
the frequency domain interpretation is that the controller contains a _ reduced order internal model _ if @xmath20 for some @xmath19 . in a situation
where @xmath21 for some @xmath19 , e.g. , when @xmath22 and @xmath1 contains only two plants , robust output regulation can then be achieved without the full internal model of the exosystem . in the second part of the paper
we construct a controller to solve the robust output regulation problem for a given class of perturbations @xmath1 .
the design procedure is a two step process where we first stabilize the system and then design a robustly regulating controller for the stabilized plant .
the robust regulation of the stabilized plant is carried out using the controller @xmath23 where @xmath24 are chosen in such a way that they satisfy the regulation condition and have ranks @xmath15 defined in .
in addition , we choose @xmath11 such that the eigenvalues of @xmath25 have nonpositive real parts . we show that for such choices of parameters there exists @xmath26 such that for any @xmath27 controller stabilizes the closed - loop system .
controllers of this type have been used in robust output regulation with full internal models in @xcite . in the final part of the paper
we illustrate the results by designing a robustly regulating controller for a laboratory process with five water tanks . in the studied experimental setup
the restricted class of perturbations arises naturally from considering the unknown valve positions of the water tank system as parameters with uncertainty . by design the controller with a reduced order internal model achieves output regulation irregardless of the valve positions .
robust output regulation with a restricted class of perturbations has been studied previously using frequency domain techniques for stable systems in @xcite by the authors . in this paper
we extend the results of @xcite most notably by introducing a controller design procedure for unstable plants , by establishing the optimality of the presented lower bounds for @xmath28 , and by giving a new and simplified proof for the characterization of robust controllers .
our results also establish a new sufficient condition for the solvability of the robust output regulation problem for a class @xmath1 of perturbations . finally , we present a new real world example on construction of controllers with reduced order internal models .
locatelli and schiavoni studied a similar control problem in @xcite .
however , in @xcite the controller was required to be robustly regulating in a small neighborhood of a given finite set of plants , and consequently the controller required a full @xmath0-copy of the exosystem .
in this section we introduce the notation used in this paper and state the robust output regulation problem .
we denote the class of functions that are bounded and analytic in the right half plane @xmath29 by @xmath30 .
the set of all matrices of arbitrary size and of all @xmath31-matrices over a set @xmath32 are denoted by @xmath33 and @xmath34 , respectively .
we denote the rank , the range , the kernel , and the moore - penrose pseudoinverse of a matrix @xmath35 by @xmath36 , @xmath37 , @xmath38 , and @xmath39 , respectively .
we assume that the class @xmath1 of perturbations has the following properties throughout the rest of the paper . *
the nominal plant is in the class @xmath1 , i.e. , @xmath40 .
* every @xmath12 is analytic at the points @xmath41 .
in this paper we assume that the error feedback controller is of the form , @xmath42 where @xmath24 and where @xmath43 is analytic at @xmath13 for all @xmath17 .
this means in particular that the poles of the controller are located at the frequencies @xmath13 the reference signal and that their orders are at most one .
the plant and the controller form the closed loop depicted in fig .
[ fig : closedloop ] . here
@xmath44 is an external disturbance .
the closed loop transfer function from @xmath45 to @xmath46 is @xmath47 fig1.eps ( 1,13)@xmath48 ( 17,12.5)@xmath49 ( 32.5,9)@xmath50 ( 59,12.5)@xmath51 ( 53,23)@xmath44 ( 74,9)@xmath52 ( 91,13)@xmath53 given a class @xmath1 of perturbations , choose the parameters @xmath7 and @xmath54 of the controller in such a way that * the controller @xmath55 stabilizes the plant @xmath2 , i.e. @xmath56 . * if @xmath12 is such that @xmath55 stabilizes @xmath57 , then @xmath58 where @xmath59 with distinct real numbers @xmath60 and @xmath61 . if condition is satisfied , we say that @xmath55 _ regulates _ @xmath12 . in the time - domain , this corresponds to the output @xmath62 converging to @xmath63 of asymptotically with respect to time .
in this section we present a characterization for controllers that are robust with respect to a given class @xmath1 of perturbations .
the following theorem is the main result of this section .
[ thm : robchar2 ] assume the controller is of the form .
* if @xmath10 is such that @xmath55 stabilizes @xmath57 , then @xmath55 regulates @xmath57 if and only if @xmath64 for all @xmath65 . *
if @xmath55 stabilizes @xmath2 , then it solves the robust output regulation problem for the class @xmath1 of perturbations if and only if is satisfied for all @xmath10 that are stabilized by the controller @xmath55 .
the second part of the theorem [ thm : robchar2 ] is a direct consequence of the first part and the statement of the robust output regulation problem . in order to prove the first part ,
let @xmath10 be such that @xmath55 stabilizes @xmath57 and let @xmath65 be arbitrary . since @xmath66 is analytic at @xmath13 it has the taylor series expansion @xmath67 and since @xmath55 has a simple pole at @xmath13 it has the laurent series @xmath68 since @xmath69 is stable , it has the taylor series @xmath70 since @xmath71 , the coefficient of the term @xmath72 of the laurent series expansion on the left hand side is zero , i.e. @xmath73 similarly , since @xmath74 , comparing the constant terms we see that @xmath75 in order to show the sufficiency of , observe that implies that @xmath76 if the condition holds .
thus @xmath77 does not have a pole at @xmath13 .
consequently , if holds for all @xmath65 then @xmath55 is regulating .
next we show the necessity of .
if @xmath55 is regulating then @xmath77 is stable .
this implies that @xmath78 .
using , we see that @xmath79 . thus , the condition holds .
if the controller @xmath55 solves the robust output regulation problem for a class @xmath1 of perturbations , then theorem [ thm : robchar2 ] gives us the following optimal lower bounds for the ranks of the matrices @xmath11 .
we call the bound @xmath80 of the theorem _ the minimal order of the internal model related . _ [ thm : minimumrank ] let @xmath80 be the minimum dimension over all the subspaces @xmath81 of @xmath82 satisfying @xmath83 here @xmath84 is the preimage of @xmath85 for all @xmath12 .
a. if @xmath55 of the form stabilizes all the plants in @xmath1 and solves the robust output regulation problem , then @xmath86 b. if @xmath87 for all @xmath65 and @xmath2 is stabilizable , then there exists a robustly regulating controller satisfying @xmath88 for all @xmath65 .
the second item is justified by section [ sec : contdesign ] . in order to prove the first item ,
let @xmath89 be the rank of @xmath11 , and let @xmath90 be linearly independent columns of @xmath11 .
we set @xmath91 .
since @xmath55 is robustly regulating , theorem [ thm : robchar2 ] implies that for all @xmath12 there exists a vector @xmath92 such that @xmath93 thus , @xmath94 .
it follows that @xmath95 .
we want to point out that if @xmath96 is invertible for all @xmath12 , then @xmath97 since @xmath98 is a unique element .
the second item implies the following sufficient solvability condition .
[ cor : solvability ] the robust regulation problem is solvable if the plant @xmath2 is stabilizable , @xmath87 for all @xmath65 , and @xmath99 for all @xmath100 . the input to output effect of the internal model is characterized by transmission zeros . in @xcite
it was shown that the regulation condition implies that @xmath101 this simply means that there exists a transmission zero at @xmath8 blocking the pole of the reference signal .
the transmission zero must be in the appropriate direction , which is determined by the plant @xmath66 and the reference signal @xmath85 .
thus , the aim of the robust regulation is to find a controller that aligns the direction of the transmission zero with the reference signal for every plant in @xmath1 .
this is achieved by building an internal model of large enough order into the controller . a way of doing this is proposed in section [ sec : contdesign ] .
an internal model of order @xmath89 basically introduces @xmath89 overlapping transmission zeros . in the extreme case @xmath102
there is a blocking zero , meaning that the whole transfer function vanishes at @xmath8 .
this is what is required in the classical robust regulation .
if we consider more than one reference signals , then the condition is required for all of them .
for example , if there is an additional reference signal @xmath103 then is replaced by the condition @xmath104 this naturally increases the size of the internal model .
in this section , we propose a two stage robust controller design .
first stage is to find a stabilizing controller for the given nominal plant , and in the second stage a robust controller for the stabilized plant is found .
combining the two controllers results in a robustly regulating controller for the nominal plant .
we begin by showing that for stable plants there exists a simple robust controller , and then proceed to show how the design procedure is carried out for unstable plants . the next theorem is the main result of this subsection and it gives a design procedure of a robustly regulating controller .
the simple structure suffices because the nominal plant is stable .
additional structure is needed for example if the plant is unstable ( see section [ sec : unstable ] ) or there are some other design goals , e.g , .
optimization , which together with the minimal internal model leads to the controller .
[ thm : contdesignstab ] assume that @xmath105 for all @xmath65 and that @xmath2 is stable .
then the contoller @xmath106 solves the robust output regulation problem for a class @xmath1 of perturbations if the design parameters @xmath11 and @xmath107 are chosen in the following way : 1 .
find a subspace @xmath81 such that @xmath83 and @xmath108 .
2 . choose a basis @xmath109 of @xmath81 .
3 . define @xmath110.\end{aligned}\ ] ] 4 .
choose an invertible matrix @xmath111 so that the eigenvalues of @xmath112 are zero or have negative real parts , and that the jordan blocks related to the zero eigenvalue are trivial .
5 . set @xmath113 6 .
choose suitably small @xmath107 to guarantee closed - loop stability .
the proof of theorem [ thm : contdesignstab ] is divided into two theorems .
theorem [ thm : designparameterck ] shows that the proposed controller is regulating and theorem [ thm : designparametere ] shows that with the choices made there exists a sufficiently small @xmath107 guaranteeing stability of the closed loop . before proceeding
further we discuss the choice of @xmath81 in the design procedure .
the condition @xmath108 is not automatically satisfied if the rank of @xmath114 is less than @xmath0 .
this can happen if the plant has transmission zeros at @xmath8 or there are less inputs than outputs . in the classical robust regulation
it is well known that in order to stabilize the nominal plant with a controller containing a full internal model the plant must not have transmission zeros at the poles of the reference signal @xcite .
indeed , the condition is not satisfied if @xmath115 and @xmath116 . in our case
, @xmath2 may have transmission zeros or less inputs than outputs , since @xmath81 need not in general be @xmath82 .
the choice @xmath81 achieving the minimal order internal model exists if @xmath114 has full rank since the condition @xmath108 is then trivially satisfied .
a particular choice for @xmath81 would be @xmath117 with @xmath118 .
the condition @xmath119 or equivalently @xmath120 is then needed @xcite . in general , choosing @xmath121 is not optimal , since @xmath15 can be strictly greater than the minimal order of the internal model related to the pole @xmath13 . on the other hand ,
if @xmath96 are invertible for all @xmath12 , then item ( iii ) of theorem [ thm : minimumrank ] shows that @xmath122 is an optimal choice .
[ thm : designparameterck ] let @xmath2 be stable and assume that @xmath123 for all @xmath100 .
if @xmath11 of are as in , then the condition holds for every @xmath100 .
we show that holds for arbitrarily chosen @xmath12 .
since @xmath124 and @xmath83 there exists @xmath125 such that @xmath126 it follows that @xmath127 .
[ thm : designparametere ] let @xmath2 be stable and assume that @xmath123 for all @xmath100 .
if @xmath11 of are as in , then there exists @xmath128 such that @xmath55 of stabilizes @xmath2 for every @xmath129 $ ] .
[ lem : localbound ] let @xmath2 be stable and assume that @xmath123 for all @xmath100 . if @xmath11 of are as in , then @xmath130 is bounded above in @xmath131 by a bound independent of @xmath132 . by the choice of @xmath11 ,
we know that the jordan blocks of @xmath133 related to the eigenvalue 0 are trivial , and that the non - zero eigenvalues of @xmath133 have negative real parts .
this means that there exist a matrix @xmath32 and a matrix @xmath134 with eigenvalues that have negative real parts such that @xmath135 so @xmath136 since all the eigenvalues of @xmath137 have positive real parts , @xmath138 is analytic in @xmath131 . in addition , it approaches @xmath139 as @xmath140 .
thus , it is bounded in @xmath131 . since @xmath141 it follows that @xmath142 is bounded in @xmath131 by a bound independent of @xmath132 .
this shows the claim .
_ proof of theorem [ thm : designparametere ] .
_ first we show that @xmath143 is stable for all sufficiently small @xmath107 . to this end
, we choose @xmath144 and define the half discs @xmath145 our aim is to show the existence a constant @xmath146 such that @xmath143 is bounded in @xmath147 whenever @xmath148 , and of @xmath149 such that @xmath143 is bounded in @xmath111 whenever @xmath150 . then @xmath143 is stable for all @xmath129 $ ] where @xmath151 . by the stability of @xmath2 and the definition of @xmath55 ,
@xmath152 is bounded in @xmath147 .
thus , there exists small enough @xmath146 such that @xmath143 is bounded in @xmath147 whenever @xmath153 .
next we show the existence of suitable @xmath149 .
we decompose @xmath154 where @xmath155 by lemma [ lem : localbound ] , @xmath156 is bounded in @xmath111 by a bound independent of @xmath157 .
in addition , @xmath158 is bounded in @xmath111 since @xmath2 and @xmath159 are analytic in @xmath111 .
the decomposition implies that we can choose @xmath149 such that @xmath160 is bounded in @xmath111 for all @xmath161 $ ] .
this completes the proof of the stability of @xmath143 .
since @xmath2 is stable , it remains to show the stability of @xmath162 . by the stability of @xmath143 and the decomposition , we only need to show that @xmath163 is stable . by the above discussion @xmath164
can only have poles of order one .
thus , it has the representation @xmath165 where @xmath166 is the projection to @xmath167 along @xmath168 and @xmath169 is an analytic function @xcite .
since @xmath108 and @xmath124 we have that @xmath170 .
consequently , @xmath171 , and @xmath172 is analytic .
this completes the proof .
@xmath173 for unstable plants ,
the design procedure is given in the following theorem .
it is based on the two stage approach proposed in ( * ? ? ?
* section 5.3 ) .
[ thm : contdesignunstab ] if the steps of items 1 and 2 below can be carried out , then the controller of step 3 is robustly regulating . 1 .
stabilize the nominal plant @xmath2 using a stabilizing controller @xmath174 that does not possess poles at @xmath13 for @xmath175 .
2 . find a controller @xmath176 of form that stabilizes the plant @xmath177 and satisfies the condition for every @xmath100 .
a robustly regulating controller of @xmath2 is given by @xmath178 [ rem : contdesign ] in step 2 we may use the approach proposed for stable plants if we choose the matrices @xmath179 implied by @xmath1 as earlier , but in we replace @xmath2 by @xmath180 when choosing @xmath111 . in particular ,
if the plant @xmath2 is invertible at @xmath8 , then so is the stabilized plant @xmath180 , because @xmath174 does not posses poles at @xmath8 .
it follows that step 2 of the above design procedure can be carried out .
_ proof of theorem [ thm : contdesignunstab ] . _
theorem 5.3.6 of @xcite ( the generalization to the current case follows by ( * ? ? ?
* section 8) ) shows that @xmath55 stabilizes @xmath2 since @xmath174 stabilizes @xmath2 and @xmath176 stabilizes @xmath180 .
thus we only need to show that @xmath55 is regulating for @xmath100 .
since @xmath174 does not possess poles at @xmath13 and @xmath176 is of the form , it is obvious that @xmath55 is of the form as well with the same matrices @xmath11 .
the matrices @xmath11 satisfy the condition by assumption .
let us consider the laboratory process of fig .
[ fig : tanks ] with five water tanks .
there is a hole in the bottom of each water tank and the water from the tanks four and five flows to the tanks below them .
the three pumps with operating voltages @xmath181 , @xmath182 , induce a flow @xmath183 where @xmath184 is a constant . denote the deviation from the initial water levels of tank @xmath185 by @xmath186 , @xmath187 .
the aim of our control problem is to choose the inputs @xmath181 so that the outputs @xmath188 follow the reference signal @xmath189 i.e. the water level of tank 1 is changing in a periodic manner while kept one unit above the initial level in the other two bottom tanks . the parameters @xmath201 correspond to how the three valves are set prior to the experiment .
the flow induced by the first pump to tank 1 is @xmath202 and @xmath203 to tank 5 , and similarly for the other two valves .
the changes to the valve positions can be considered as perturbations to the system .
the transfer function of the system linearized at the initial water levels is @xmath204 where the parameters @xmath205 and @xmath206 depend on the tank cross - sections , the outlet hole cross - sections , constants of proportionality @xmath207 , and the initial water levels . for more details ,
see @xcite where a similar system with four tanks was considered .
here we choose the initial setup so that @xmath208 and @xmath209 for simplicity , i.e. we have @xmath210 let the initial positions of the valves be @xmath211 , i.e. the nominal plant is @xmath212 the frequency domain representation of @xmath213 is @xmath214 in order to find a robustly regulating controller , we define a basis of @xmath215 for @xmath216 and @xmath217 .
let @xmath218 be the @xmath185th natural basis vector of @xmath219 .
because of the upper triangular block structure of @xmath96 , it is easy to deduce that @xmath220 and @xmath221 . following the design procedure of theorem [ thm : designparameterck ]
, we choose @xmath222 and @xmath223 .
now the controller satisfies the regulation property .
it remains to choose invertible @xmath111 and small enough @xmath132 to guarantee stability .
we can choose @xmath224 since all the eigenvalues of @xmath225 have positive real parts for every @xmath217 .
if we choose @xmath226 , then we have @xmath227 in order to show that the controller is stabilizing for @xmath2 we note that the proof of theorem [ thm : designparametere ] shows that it is sufficient that @xmath143 is stable .
the stability follows now by observing that the zeros of @xmath228 have negative real parts , i.e. @xmath143 can not have poles in the closed right half plane @xmath131 .
we end this section by comparing the proposed design procedure to the classical one . here
we have used the knowledge that the first two inputs do not affect the third output , i.e. , we have structured perturbations , whereas the classical design procedure ignores this fact and the perturbations are taken to be totally arbitrary .
one should also note that small perturbations in the other parameters such as the hole or tank diameters do not affect the zero components of the plant transfer function , so actually we gain robustness for all small parameter changes . in our controller
the internal model is minimal in the sense that the ranks of the matrices @xmath11 for @xmath217 are minimal . since @xmath229 and @xmath230 have rank two instead of rank three , which would be the case
if classical approach is used , the order of the controller s realization is reduced by two .
k. kim , s. skogestad , m. morari , and r. d. braatz .
necessary and sufficient conditions for robust reliable control in thepresence of model uncertainties and system component failures .
, 70:6777 , 2014 .
p. laakkonen and l. paunonen . a simple controller with a reduced order internal model in the frequency domain . in _ proceedings of the 15th european control conference
aalborg , denmark , june 29 july 1 2016 . | the internal model principle states that all robustly regulating controllers must contain a suitably reduplicated internal model of the signal to be regulated . using frequency domain methods
, we show that the number of the copies may be reduced if the class of perturbations in the problem is restricted .
we present a two step design procedure for a simple controller containing a reduced order internal model achieving robust regulation .
the results are illustrated with an example of a five tank laboratory process where a restricted class of perturbations arises naturally . |
we begin by reviewing the geometric formulation of the problem of swimming at low reynolds number developed by shapere and wilczek @xcite . the instantaneous rigid motion ( translation and rotation ) of a swimmer
is determined by the condition that the swimmer not be able to exert a net force or torque on itself , and the condition that the fluid velocity vanishes at infinity .
we should first explain why the problem of swimming at low reynolds number can be formulated in a purely geometric way , independent of the mass of the swimmer or the speed of the swimming stroke ( assuming the speed of the stroke is still small enough so that there is no appreciable momentum transfer to the fluid ) .
recall that the reynolds number , which is associated with a viscous fluid and an object in motion in that fluid , is a ratio of the inertial and viscous forces on that object ( we are not yet considering systems with odd viscosity so in this sentence the word viscous " refers to the traditional dissipative ( even ) viscosity of the fluid ) .
if @xmath1 is the even viscosity coefficient , @xmath2 is a typical speed of the fluid flow , @xmath3 is a characteristic dimension of the swimming object , and @xmath4 is the density of the fluid , then the reynolds number can be expressed as = .
the low reynolds number regime can be interpreted as the regime where the momentum density of the fluid , @xmath5 , is negligible compared to the scale @xmath6 . at low reynolds number
the drag force on the swimmer is proportional to its velocity .
this means that if the swimmer stops its stroke and just coasts through the fluid , its speed will decay exponentially until it comes to a stop . in the low reynolds number regime
this exponential decay is so fast that the motion of the swimmer at any given time can be considered to be completely independent of what the swimmer was doing at all previous times @xcite .
the motion of the swimmer at time @xmath7 depends only on its shape and the velocity of its surface at time @xmath7 . with these remarks in mind
we can move on to discuss the geometric theory of swimming at low reynolds number . in two dimensions , for swimmers modeled as the interior of deformed circles
, we can represent the swimming stroke ( the motion of the boundary of the swimmer ) by a time - dependent complex function @xmath8 , @xmath9 , whose real and imaginary parts give the @xmath10 and @xmath11 positions of the point on the swimmer described by the parameter @xmath12 at the time @xmath7 .
when we want to emphasize the dependence of @xmath8 on the real parameter @xmath13 instead of the complex parameter @xmath14 ( as we do in appendix [ sec : areaproof ] ) we call it @xmath15 instead
. the function @xmath8 lives in a space of un - located " shapes , which can be obtained from the space of located " shapes by partitioning it into equivalence classes @xmath16 $ ] containing all shapes differing only by a rigid motion .
the location and orientation of the swimmer in real space is specified by a rigid motion @xmath17 acting on a representative of the equivalence class @xmath16 $ ] , the simplest choice being @xmath8 itself : s(,t ) = ( t)s_0(,t ) . to take an example
, @xmath8 might be the representative of @xmath16 $ ] with its centroid at the origin and a distinguishing feature of the shape aligned with the x - axis at time @xmath18 . to be concrete ,
let us encode the translation and rotation represented by @xmath17 into a @xmath19 matrix and let this matrix act on @xmath8 represented as a three - dimensional vector with third entry equal to one , ( t)s_0(,t ) = ( ) & ( ) & x + - ( ) & ( ) & y + 0 & 0 & 1 [ s_0(,t ) ] + [ s_0(,t ) ] + 1 , where @xmath20 and @xmath21 are the vector and angle representing the translation and rotation effected by @xmath22 . the matrix @xmath17 is determined by integrating the equation = ( t)(t ) , [ eq : diffeq ] where the matrix @xmath23 determines the infinitesimal rigid motion of the swimmer during a time @xmath24 in the sense that @xmath25 _ is _ the rigid motion of the swimmer during the interval @xmath24 .
the matrix @xmath23 is completely determined by the requirements that the net force and torque on the swimmer vanish and that the fluid velocity goes to zero at infinity . to determine the swimming path we need to find @xmath26 for a given swimming stroke and then integrate eq .
[ eq : diffeq ] . integrating this equation
gives the solution for the rigid motion @xmath17 , ( t ) = ( 0 ) |p , where @xmath27 denotes a reverse path - ordering operation .
explicitly , we have |p & = & i + _ 0^t ( t ) dt + & + & _ 0^t ( _ 0^t_1 ( t_2)(t_1 ) dt_2 ) dt_1 + & + & [ poi ] so the matrix @xmath28 with the latest time @xmath29 appears furthest to the right in each integral , which is the reverse of the usual path ordering operation where the latest time goes furthest to the left in each integral .
we show how this integration is carried out numerically in appendix [ app : computation ] . to see how the idea of a gauge theory of shapes enters we first note that the choice of a representative from the equivalence class @xmath16 $ ] is analogous to a choice of gauge , and the matrix @xmath23 plays the role of a gauge potential . if we choose a different representative @xmath30 , related to @xmath8 by a rigid motion @xmath31 ( we can choose a different representative at each time t ) , _ 0(,t ) = u(t ) s_0(,t ) , then the requirement that the rigid motion of the swimmer in real space remain unchanged leads to the transformation law for @xmath17 ( t ) (t ) = ( t)u^-1(t ) .
the fact that the transformed gauge potential must satisfy the new differential equation = (t)(t ) yields the familiar transformation law ( t ) (t ) = u(t)(t)u^-1(t ) + u(t ) , which shows that @xmath23 does indeed transform like a gauge potential .
we can also represent @xmath23 in the form of a @xmath19 matrix , ( t ) = 0 & & v_x + - & 0 & v_y + 0 & 0 & 0 [ gauge ] where @xmath32 and @xmath33 are the instantaneous linear and angular velocity of the swimmer ( so @xmath23 is in the lie algebra of rigid motions ) .
sometimes we will refer to the translational and rotational parts @xmath34 and @xmath35 of the gauge potential , defined by _ tr & = & v_x + iv_y + _ rot & = & .
the components of @xmath23 can be completely determined by solving the equations of motion for stokes flow of the viscous fluid surrounding the swimmer , subject to no - slip boundary conditions at the surface of the swimmer .
now that we have reviewed the geometric formulation of swimming we will introduce the concept of the odd / hall viscosity in time - reversal breaking fluids .
we now review the basic definition of odd viscosity and the derivation of the isotropic odd viscosity contribution to the fluid stress tensor in two dimensions . throughout this section
we follow the presentation of ref . where most of these details were first worked out .
the general linear relation between the fluid stress tensor @xmath36 and the rate of strain tensor @xmath37 ( @xmath38 are the components of the fluid velocity vector @xmath39 ) is of the form t_ij = _ ijkl v_kl .
the symmetry of the stress and rate of strain tensors imply the symmetry of the viscosity tensor @xmath40 under the exchanges @xmath41 and @xmath42 , but in general @xmath40 can contain terms which are symmetric or anti - symmetric under the exchange of the pair of indices @xmath43 with the pair of indices @xmath44 .
we can always split @xmath40 into parts which are even and odd under such an exchange by writing @xmath45 . to extract the isotropic contribution to @xmath46
it is convenient to use a simple basis for representing a real , 4th - rank tensor that is symmetric under exchange of its first two and second two indices .
one such basis is provided by the tensor products ^a ^b , a , b \{0,1,3 } of the pauli matrices @xmath47 , @xmath48 and the @xmath49 identity matrix @xmath50 , where we have been careful to only use the symmetric matrices .
we can expand the viscosity tensor as _
ijkl = _ a , b = 0 , 1,3 _ ab ^a_ij^b_kl and then identify the odd part as ^o_ijkl = _ a b ^o_ab ( ^a_ij^b_kl - ^b_ij^a_kl ) . in two dimensions the generator of spatial rotations is @xmath51 , where @xmath52 is the second pauli matrix . in an isotropic fluid the viscosity tensor must commute with @xmath53 to be rotationally invariant . using the familiar commutation and anti - commutation relations for the pauli matrices , and the fact that all matrices commute with the identity @xmath50 , we find that in an isotropic fluid the odd part of the viscosity tensor must have the form ^o_ijkl = ^o ( ^1_ij^3_kl - ^3_ij^1_kl ) , where the single constant @xmath0 is the coefficient of odd viscosity .
finally we can use the explicit expressions ^1_ij & = & _ i1_j2 + _ i2_j1 + ^3_ij & = & _ i1_j1 - _ i2_j2 for the elements of the pauli matrices @xmath47 and @xmath48 to write down the form of the odd viscosity contribution to the stress tensor t^o_ij & = & ^o_ijklv_kl + & = & - 2^o ( _ i1_j1 - _ i2_j2)v_12 + & + & ^o ( _ i1_j2 + _ i2_j1)(v_11 - v_22 ) which was first obtained in ref . .
for comparison we also display the much more familiar even viscosity part of the stress tensor ( for an incompressible fluid ) t^e_ij = 2^ev_ij , where @xmath1 is the coefficient of even viscosity .
we see that diagonal elements of @xmath54 are proportional to off - diagonal elements of @xmath55 and off - diagonal elements of @xmath54 are proportional to diagonal elements of @xmath55 .
this atypical relation between the elements of @xmath54 and @xmath55 has a number of non - intuitive consequences .
for example , a circular object rotating in a fluid with odd viscosity will feel a pressure , directed either radially inwards or outwards depending on the sense of the rotation ( see @xcite and fig .
[ fig : droplet ] ) .
this is quite different from what would happen in a fluid with even viscosity only , where a rotating circle would feel a torque that opposes the rotation .
the fact that the direction of the pressure force ( radially inwards or outwards ) on a circle rotating in an odd viscosity fluid depends on the sense of the rotation means that time - reversal symmetry is broken in systems with odd viscosity .
in classical fluids with both even and odd viscosity avron has shown ( see @xcite ) that the equations of motion for incompressible stokes flow ( viscous force - dominated flow ) are ( p - ^o ) & = & ^e ^2 + & = & 0 , where @xmath56 is the pressure , @xmath57 is the vorticity , and @xmath1 and @xmath0 are the coefficients of even and odd viscosity , respectively . if @xmath58 , taking the curl of the first equation shows that the vorticity is a harmonic function , i.e. , @xmath59 .
this means that the stream function @xmath60 ( which can be used here because the flow is incompressible ) , defined by @xmath61 , is a biharmonic function , ^2(^2 ) = 0 . in two dimensions we can package the velocity vector @xmath62 into a complex variable @xmath63
the solution for @xmath64 can then be expressed in the complex form ( see @xcite ) v = _
1(z ) - z + [ velocity ] where @xmath65 , the bar denotes complex conjugation and @xmath66 .
the functions @xmath67 and @xmath68 are analytic functions ( away from the point @xmath69 , which lies inside the swimmer ) with the laurent series expansions _ 1(z ) & = & _ k<0a_k z^k+1 + _
2(z ) & = & _ k<-1b_k z^k+1 . to solve for the coefficients @xmath70 and @xmath71
we impose no - slip boundary conditions at the surface of the swimmer .
solving for these coefficients can be very difficult for general swimming strokes , so we will focus our attention on a class of simple swimmers introduced in ref .
whose shapes are conformal maps of the circle of degree @xmath72 . in appendix
[ app : degree3 ] we extend our results to swimmers that are conformal maps of the circle of degree @xmath73 . to calculate the force and torque on the swimmer
we will need the stress tensor .
we have seen in section [ sec : oddvisc ] that in the presence of odd viscosity the stress tensor gets an extra contribution .
the full stress tensor is now t_ij & = & -p_ij + 2^ev_ij - 2^o ( _ i1_j1 - _
i2_j2)v_12 + & + & ^o ( _ i1_j2 + _ i2_j1)(v_11 - v_22 ) .
[ odd_stress ] the components of the odd - viscosity part of the stress tensor are t^o_11 & = & -^o(_2 v_1 + _ 1 v_2 ) + t^o_12 & = & ^o(_1 v_1 - _ 2 v_2 ) + t^o_21 & = & ^o(_1 v_1 - _ 2 v_2 ) + t^o_22 & = & ^o(_2 v_1 + _ 1 v_2 ) .
since the fluid is incompressible , an application of the divergence theorem shows that the force and torque on the surface of the swimmer are the same as the force and torque on the fluid at infinity . using this equivalence , the components of the force on the swimmer are f_i & = & _ r_0 ^ 2 ( t_ijr_j ) r dand the torque on the swimmer is n = _
r_0 ^ 2 ( _ ij r_i t_jk r_k ) r^2 d . in these formulas
@xmath74 are the components of the radial unit vector @xmath75 and the integral is taken over the circle at infinity . using these equations , and the components of the odd - viscosity part of the stress tensor
, we can derive expressions for the odd - viscosity contribution to the force and torque on the swimmer .
in complex form they are f^o = _ r -2^o _ ( _ |z v ) d|z and n^o = _ r -2^o \ { i _ z ( _ z|v ) dz } where @xmath76 is a circular contour of radius @xmath77 ( to be taken to infinity ) , and we have switched to a complex notation for the force , @xmath78 ( the torque , being a scalar in 2d , is real ) .
plugging in the velocity expansion into these formulas gives f^o & = & 0 + n^o & = & -4^o[b_-2 ] [ odd_torque ] . in the next subsection
we will show that the physical interpretation of this result is that the odd - viscosity contribution to the torque is proportional to the flux of the fluid at infinity ( see section [ sec : physint ] ) .
previously it has been shown @xcite that the even - viscosity contribution to the force and torque on the swimmer is given by f^e & = & 0 + n^e & = & 4^e[b_-2 ] [ even_torque ] .
the swimmer feels no net force ( a generic result for stokes flows in two dimensions @xcite ) and the total torque is n = 4 ( ^e [ b_-2 ] - ^o[b_-2 ] ) .
[ total_torque ] we can cancel the torque on the swimmer by having the swimmer rotate at a certain angular velocity @xmath33 . this uniquely determines the rotational part of the gauge potential . in dimensions @xmath79 the translational part of the gauge potential
can be determined by the condition that the net force on the swimmer vanish . in two dimensions , however , the net force vanishes identically @xcite and so one must instead determine the translational part of the gauge potential by requiring that the fluid velocity vanish at infinity @xcite .
we discuss this condition in more detail in section [ sec : motion ] .
the physical content of the formula for the net torque on the swimmer can be better understood by looking at the relation of the coefficient @xmath80 to the circulation and flux of the fluid at infinity , denoted by @xmath81 and @xmath82 , respectively .
we can express the circulation and flux of the fluid at infinity in the form of line integrals of the velocity around a large circle of radius @xmath77 , to be taken to infinity .
we have , ( ) = _
r_0 ^ 2 r dand ( ) = _
r_0 ^ 2 r d[flux_infty ] .
using the velocity expansion , we find ( ) & = & -2[b_-2 ] + ( ) & = & 2[b_-2 ] .
using these expressions , the net torque on the swimmer can be rewritten in the form n = -2^e ( ) -2^o ( ) .
the condition of vanishing torque in the different cases can then be interpreted in terms of zero circulation at infinity for even viscosity only , zero flux at infinity for odd viscosity only , or a proportionality between the flux and circulation at infinity when both types of viscosity are present .
following ref . , we will begin by considering nearly circular swimmers with swimming strokes of the form s_0(,t)= _ 0(t)+ _ -2(t)^-1 + _ -3(t)^-2 [ stroke ] , where the @xmath83 s are coefficients which determine the time evolution of the swimming stroke . this kind of stroke is just a conformal map of degree @xmath72 from the unit circle to the complex @xmath84-plane .
the absence of a term @xmath85 fixes the gauge " with respect to translations @xcite .
an important formula is the area of the swimmer at time @xmath7 , which is given by a(t ) & = & \ { ds_0(,t ) } + & = & \{_0 ^ 2 d } [ area ] , which gives a(t)= ( |_0|^2 - |_-2|^2 - 2|_-3|^2 ) for the simple stroke .
general swimmers represented by conformal maps of degree @xmath86 have the form s_0(,t)= _ 0(t)+ _ n=1^ _ -n(t ) ^-n [ gen - stroke ] and in appendix [ app : degree3 ] we extend the swimming motion formulae to swimmers with @xmath73 .
to determine the coefficients @xmath70 and @xmath71 in the velocity expansion we need to conformally map the flow field back to the @xmath87 plane @xcite . recall that the shape of the swimmer @xmath8 is a conformal map in the other direction , from the unit circle @xmath9 in the @xmath88-plane to the @xmath84-plane . for general swimmers of the form the conformal mappings between the @xmath88 and @xmath84 planes take the form @xcite , [ cm ] z & = & s_0 ( ) = _ 0(t)+ _ n=1^ _ -n(t ) ^-n [ ct1 ] + & = & s_0 ^ -1(z ) = - + .
we now introduce a star @xmath89 symbol to denote the pull - back of a function in the @xmath84-plane to the @xmath88-plane obtained by substituting for @xmath84 in that function .
the pull - backs of @xmath67 and @xmath68 are denoted by _ 1^ * ( ) & = & _ k<0a_k^ * ^k+1 + _ 2^ * ( ) & = & _ k<-1b_k^ * ^k+1 , where the @xmath90 and @xmath91 are a new set of coefficients related to the original @xmath70 and @xmath71 through the conformal mapping .
next we pull back the velocity field onto the unit circle @xmath14 in the @xmath88-plane so that we can apply the no - slip boundary conditions there and determine the pull - back coefficients @xmath90 and @xmath91 in terms of the @xmath83 . on the unit circle
@xmath14 the velocity expansion takes the form ( suppressing the @xmath7 dependence ) v^ * ( ) = _ 1^ * ( ) - + .
[ velocity_pb ] the only coefficients we need to determine the translational and rotational motion of the swimmer are @xmath92 and @xmath80 .
this is because @xmath92 gives the fluid flow at infinity , so it determines the translational motion of the swimmer , and @xmath80 is related to the torque on the swimmer , so it determines the rotational motion of the swimmer . using the conformal mapping ,
the coefficients @xmath92 and @xmath80 can be expressed in terms of the pulled - back coefficients @xmath90 and @xmath91 as a_-1 & = & a_-1^ * + b_-2 & = & _ 0b_-2^ * .
we can solve for the pulled - back coefficients @xmath90 and @xmath91 in terms of the parameters @xmath93 using , and then use the pulled - back coefficients to solve for @xmath92 and @xmath80 .
as in @xcite we find a_-1 & = & -|_0 ^
-1_-3_-2 + b_-2 & = & |_0_0 - _ -2_-2 - 2_-3_-3 . [ b-2 ] to determine the translational part of the gauge potential we note that the coefficient @xmath92 is a constant contribution to the velocity expansion , which means that the fluid velocity at infinity is uniform and non - zero . following section 7.5 of ref .
, we argue that a finite - size swimmer located near the origin should not be able to induce a non - zero fluid velocity at infinity , and so we make a galilean transformation to a frame in which the fluid is at rest at infinity and the swimmer moves with a velocity _
tr v_x + iv_y = -a_-1 , where @xmath34 denotes the translational part of the gauge potential . to determine the rotational part of the gauge potential we attempt to cancel the torque on the swimmer by having the swimmer rotate at an appropriately chosen angular velocity @xmath33 . in the parameterization of the swimming stroke ,
having the swimmer rotate at an angular velocity @xmath33 amounts to the replacement _ i _ i , rot = _
i e^it . under this replacement
we find b_-2 b_-2,rot = b_-2 + i ( |_0|^2 + |_-2|^2 + 2|_-3|^2 ) , so that the condition that the net torque on the swimmer vanish becomes ^e [ b_-2,rot ] - ^o[b_-2,rot ] = 0 .
solving this equation for @xmath33 yields the rotational part of the gauge potential _
[ a_rot ] this expression shows that in the presence of odd viscosity the rotational part of the gauge potential picks up a term proportional to @xmath94 $ ] . for the simple swimming stroke ,
one can verify by explicit computation that [ b_-2 ] = [ area_relation ] , which shows that the odd viscosity contribution to the angular velocity of the swimmer is proportional to the rate of change of the area of the swimmer .
this conclusion is not limited to swimming strokes which are conformal maps of degree @xmath72 , but holds for _ generic _ swimmers bounded by a closed curve without any self - intersections , as we prove in appendix [ sec : areaproof ] .
we can use this area relation and the relation @xmath95 $ ] for the circulation of the fluid at infinity to rewrite the angular velocity formula in a way which clearly shows the physical meaning of each term .
we find = . as the ratio of @xmath96 increases , the angular velocity grows without bound .
therefore we conclude that in a fluid in which the odd viscosity terms completely dominate the stress tensor ( i.e. @xmath97 ) , the condition that the swimmer experience zero net torque must be satisfied by taking @xmath98 , otherwise the angular velocity of the swimmer would have to be infinite .
so a swimmer in a fluid where odd viscosity effects are dominant must have constant area .
when discussing the limit @xmath97 in this context , we must always assume that @xmath1 is finite and large enough so that we can still neglect any inertial forces in the problem ( and so we can still take advantage of the geometric formulation of the problem of swimming at low reynolds number ) .
this is why we have been careful to say when the odd viscosity is dominant " and not when @xmath99 . "
here we present some simple examples of swimming strokes that clearly demonstrate the difference between swimming in a fluid with just even viscosity and swimming in a fluid with both even and odd viscosity .
the first example is a swimmer which starts out as a circle but grows into an ellipse by elongating one of its axes through a dipolar - like distortion .
we use the parameterization [ ellipse ] _ 0 & = & 1 + + _ -2 & = & + _ -3 & = & 0 for this swimmer .
the boundary of the swimmer is an ellipse with the lengths of the major and minor axes given by @xmath100 , @xmath101 . with only the conventional even viscosity
this stroke will not cause any motion other than an increase in the area .
we can also see this from the reflection symmetry about the x - axis , which is equivalent to the fact that all the coefficients are real . however , when there is also odd viscosity this swimmer will start to rotate because its area is growing and the torque has a term proportional to the odd viscosity and the rate of area change .
the motion for different values of the odd viscosity can be seen in fig .
[ fig : ellipse ] . to further test our results
we chose a swimmer with a more complicated quadrupolar distortion which also has a uniform area growth .
we used the parameterization [ dwave ] _ 0 & = & 1+t + _ -2 & = & 0 + _ -3 & = & 0 + _ -4 & = & .
this swimming parameterization represents a conformal map of degree @xmath102 to see how to extend the analysis of the previous section to swimmers which are conformal maps of the circle of degree @xmath73 ( i.e. how to include @xmath103 terms ) , see appendix [ app : degree3 ] . in fig .
[ fig : dwave ] we see very similar results to the dipolar case , e.g. , the motion of the swimmer is just a rotation proportional to the growth of the area .
this indicates , as we expected from the general result of appendix [ sec : areaproof ] , that the odd viscosity does not distinguish between different types of shape distortions , and only couples to changes in the total area of the interior of the swimmer .
the third example is a swimmer parameterized with the cyclic stroke [ wandering ] _ 0 & = & r_0 + _ -2 & = & -i_1 ( 2 t ) + _ -3 & = & -i_2 ( 2 t ) where @xmath104 , @xmath105 and @xmath106 are all real parameters .
we chose this particular stroke because in the case when only the even viscosity is present , the swimmer s centroid moves in a straight line through the fluid .
additionally , this stroke has a periodic time - dependent area a(t ) = , which implies that it will feel a cyclic stress from the odd viscosity term when present . in fig .
[ fig : wandering_trajectory ] one can clearly see the outcome as we show two trajectories , one with @xmath107 and one with @xmath108 in the case when @xmath0 vanishes , the swimmer travels in a straight line , however in the second case the swimmer oscillates transverse to the straight - line path . in the inset
we show that the amplitude of the transverse oscillation at a fixed time increases linearly with the slope @xmath109 as the swimmer continues it will wander further and further off of the straight - line course although on average it seems like it will still progress linearly at a similar rate to that of the swimmer in the fluid with vanishing odd viscosity .
the fourth example is a stroke which will nominally rotate when just even viscosity is present , but for which the variation of the area of the shape has been chosen carefully so that when odd viscosity is also present the shape will not rotate at all . in other words ,
the odd viscosity contribution to the angular velocity exactly cancels the even viscosity contribution for a given particular ratio @xmath96 which , for the sake of this example , we pick to be unity . a glance at eq .
shows that in order to produce this cancellation , we need the stroke to satisfy [ b_-2 ] = [ b_-2 ] . for swimmers which are conformal maps of the circle of degree @xmath72 , the coefficient @xmath80 is given by eq . .
we see from that equation that we can design such a stroke by taking @xmath110 and _ j(t ) = r_j(t ) e^i_j(t ) [ cancel - rot ] for @xmath111 , where the functions @xmath112 and @xmath113 are functions which are determined in the following way .
we would like to have _
j(t)_j(t ) = ( 1+i)_j(t ) [ cancel-2 ] where the @xmath114 are some real periodic functions of time ( to give a periodic swimming stroke ) , which we are essentially free to choose .
this choice will guarantee the cancellation of the even and odd viscosity contributions to the torque on the swimmer , since the real and imaginary parts of eq .
are equal at all times .
the reason for using the derivative of the functions @xmath114 in the above formula is purely for convenience in the formulas that follow .
plugging the form for the @xmath115 into this last equation and solving the two coupled ordinary differential equations for @xmath112 and @xmath113 gives the form of the stroke in terms of the functions @xmath114 , r_j(t ) & = & + _ j(t ) & = & -(f_j(t ) + c_j,1 ) + c_j,2 , where @xmath116 and @xmath117 are arbitrary constants ( although they must be chosen carefully along with the functions @xmath118 to keep the argument of the logarithm from ever equaling zero ) .
now any choice of the periodic functions @xmath114 will give a cyclic swimming stroke that will not rotate in a fluid with our chosen ratio @xmath119 .
this shows in principle that it is possible to construct a stroke for which the even and odd viscosity contributions to the angular velocity exactly cancel each other .
swimmers using this type of stroke might be able to more efficiently navigate odd - viscosity fluids since the particular choice of stroke cancels the rotation effects due to the odd - viscosity .
we have applied the geometric theory of swimming at low reynolds number developed by wilczek and shapere @xcite to the case where the fluid has a non - vanishing hall , or odd , viscosity .
the main effect of the hall viscosity is to introduce an additional torque on the swimmer , proportional to the rate of change of the area of the swimmer , independent of the other shape changes occurring in the stroke pattern .
this torque is the companion effect to the fact that a swimmer rotating in a fluid with odd viscosity feels an inwards or outwards pressure proportional to its angular velocity @xcite .
as we show in appendix [ sec : areaproof ] this conclusion applies to generic swimming shapes and is not limited to swimmers whose boundaries are simple conformal maps of the unit circle . as a consequence of this extra torque , a swimming stroke which would not cause the swimmer to rotate in a fluid with conventional viscosity can cause the swimmer to rotate in a fluid with hall viscosity if the area of the swimmer is changing .
it is even possible to design a stroke which will rotate the swimmer in an even viscosity fluid but not in a fluid with both even and odd viscosity , for a certain value of the ratio @xmath96 .
it is possible that swimmers placed in fluids with an odd viscosity would have to adapt their strokes to efficiently move in a straight line .
additionally , it would be interesting to see if swimmers could use the interplay between the even and odd viscosity to perform more interesting or efficient motion patterns .
we acknowledge useful discussions with j. e. avron and p. zhang .
tlh acknowledges support from doe qmn under grant number defg02 - 07er46453 .
we are thankful for the support of the institute for condensed matter theory at uiuc .
to calculate the matrix @xmath17 , which gives the rigid motion of the swimmer after a finite time @xmath7 , we need to evaluate the reverse path - ordered integral . in practice
we do this by slicing time into many small steps ( say n steps ) of size @xmath120
. we can write |pe^_0^t ( t ) dt = |pe^_i=1^n _ ( i-1)t^it ( t ) dt .
if the time steps are small enough then we can approximate this as |pe^_i=1^n _
( i-1)t^it ( t ) dt _ i=1^n |pe^_(i-1)t^it ( t ) dt , where on the right side we now have a product of reverse path - ordered integrals over many small time intervals of size @xmath120 and we should put the earliest times on the right so that we are applying the rigid motions in these small intervals in chronological order .
since these time intervals are very small we can make a further approximation by expanding the path - ordered integral over the time interval @xmath120 to first order and neglecting the higher order terms to find : |pe^_(i-1)t^it ( t ) dt i + _ ( i-1)t^it ( t ) dt .
finally we can make one further approximation for the integral of the matrix @xmath23 over the small time interval @xmath120 , _
t_i-1^t_i ( t ) dt ( t_i-1)t where @xmath121 ( and @xmath122 ) .
our final expression for the approximation of the full path - ordered integral is then |pe^_0^t ( t ) dt _ i=1^n ( i + ( t_i-1)t ) , where again the matrices for the earliest times must to be to the right so that the rigid motions are applied in the proper order .
we have also tried expanding the reverse path - ordered integrals over the time interval @xmath120 to second order , but it seems that this makes almost no visible correction to the swimming trajectory when the swimming deformations are not too large and the step size @xmath120 is small .
our analysis of the simple swimmer suggests a deeper connection between the area of the swimmer and the odd viscosity contribution to the torque on the swimmer . to explore this connection
further we now show that eq
. holds for any swimmer whose boundary is a smooth curve without self - intersections .
the boundary of the swimmer is just a smooth curve parameterized by @xmath123 which also depends on the time @xmath7 .
if we write the shape in terms of real components s_0(,t ) = x(,t ) + iy(,t ) and plug into the area formula we find a(t ) = _ 0 ^ 2 d , where the prime denotes a derivative with respect to @xmath123 .
next take a time derivative to get & = & _ 0 ^ 2 d .
we can integrate by parts on the terms with mixed partial derivatives and use the fact that the boundary terms vanish since @xmath124 , @xmath125 , @xmath126 and @xmath127 are @xmath128-periodic in @xmath13 to get = _ 0 ^ 2 d . because of the no - slip boundary conditions the vector @xmath129 is just the fluid velocity @xmath130 evaluated on the surface of the swimmer , ( ) | _ = ( , t ) + ( , t ) .
then we can write = _
ds = ( ) , where @xmath131 is a vector normal to the surface of the swimmer with magnitude @xmath132 .
this integral is just the flux of the fluid at the surface of the swimmer . by the divergence theorem we have ( ) - ( )
= _ dxdy and since the fluid is incompressible , @xmath133 , we get = ( ) . a comparison with eq . for the flux of the fluid at
infinity yields the final result [ b_-2 ] = , proving that this relation is valid for general swimming shapes in incompressible fluids .
it is known that an object which rotates in a fluid with odd viscosity will feel a pressure directed radially inwards or outwards depending on the direction of the rotation @xcite .
the relation is the companion to this statement .
it says that an object which tries to expand or contract in a fluid with odd viscosity will feel a torque whose direction ( @xmath134 ) depends on whether the area of the object is growing or shrinking .
a swimmer whose boundary is a degree 3 ( @xmath73 ) conformal map of the circle has the form s_0(,t)= _ 0(t)+ _ -2(t)^-1 + _ -3(t)^-2 + _ -4 ^ -3 , with area a(t)= ( |_0|^2 - |_-2|^2 - 2|_-3|^2 - 3|_-4|^2 ) . to solve for @xmath135 and @xmath136 we need the coefficients @xmath137 and @xmath138 .
equations for these coefficients can be obtained by plugging into the pulled - back velocity expansion .
we find that the equation for @xmath139 is really just a matrix equation for a two - component vector consisting of the real and imaginary parts of @xmath139 .
the solution is a_-2^ * = . in terms of this coefficient
we find that a_-1 = -(|_0)^-1(|a_-2^*_-3 + 2_-3_-4 ) and b_-2 & = & |_0_0 - |_-2_-2 - 2_-3_-3 + & - & 3_-4_-4 + |_-2a_-2^ * - _ -2|a_-2^ * .
note that the last two terms in @xmath80 are complex conjugates of each other and appear with the opposite sign so that they will cancel when we take the real part of @xmath80 .
this means that the relation @xmath94 = \frac{1}{2\pi}\frac{da(t)}{dt}$ ] still holds in this case , as we expect based on the general arguments presented in appendix [ sec : areaproof ] . to solve for the new form of the angular velocity necessary to cancel the torque on the swimmer
, we again send @xmath140 and solve the equation ^e [ b_-2,rot ] - ^o[b_-2,rot ] = 0 , where now b_-2,rot = b_-2 + ij with j & = & ( |_0|^2|_-2|^2 + [ _ 0_-4(|_-2)^2 ] ) + & + & |_0|^2 - |_-2|^2 + 2|_-3|^2 + 3|_-4|^2 . the new angular velocity needed to cancel the torque on the swimmer is then = ( -[b_-2 ] + [ b_-2 ] ) , so that the translational and rotational parts of the gauge potential are now given by _
tr & = & ( |_0)^-1(|a_-2^*_-3 + 2_-3_-4 ) + _ rot & = & ( -[b_-2 ] + [ b_-2 ] ) .
25ifxundefined [ 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 * * , ( ) in @noop _ _ , vol .
( ) p. @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( )
@noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop _ _ ( , ) | we apply the geometric theory of swimming at low reynolds number to the study of nearly circular swimmers in two - dimensional fluids with non - vanishing hall , or odd " , viscosity .
the hall viscosity gives an off - diagonal contribution to the fluid stress - tensor , which results in a number of striking effects .
in particular , we find that a swimmer whose area is changing will experience a torque proportional to the rate of change of the area , with the constant of proportionality given by the coefficient @xmath0 of odd viscosity . after working out the general theory of swimming in fluids with hall viscosity for a class of simple swimmers ,
we give a number of example swimming strokes which clearly demonstrate the differences between swimming in a fluid with conventional viscosity and a fluid which also has a hall viscosity . a number of more technical results , including a proof of the torque - area relation for swimmers of more general shape ,
are explained in a set of appendices .
= 1 the theory of swimming in classical fluids at low reynolds number @xcite is remarkable because of the connections it makes between seemingly disparate fields @xcite .
for example , the motion of swimmers with cyclic swimming strokes is determined purely from classical fluid dynamics , but it can be re - cast into an elegant geometric formulation reminiscent of berry s phase physics and gauge fields @xcite . in fact , the motion of tiny organisms in fluids with high viscosity can be captured by a gauge - theory " of shapes .
since the initial work on the geometric formulation of swimming there have been generalizations to swimmers in quantum fluids @xcite and even to swimmers in fluids on curved spaces @xcite .
the theory has also been successfully applied in practice to describe the swimming of robots @xcite and microbots @xcite . in this article
we focus on swimmers in 2d fluids with broken time - reversal symmetry , for example , fluids in magnetic fields or rotating fluids .
we are not interested in the specific source of time - reversal breaking , but instead just consider a classical fluid with a microscopic source of local angular momentum ( on a much smaller scale than the size of the swimmer ) that gives rise to a non - vanishing hall ( or odd ) viscosity coefficient @xcite in addition to the usual isotropic viscosity coefficients .
the hall viscosity is an off - diagonal viscosity term that is dissipationless and produces forces perpendicular to the direction of the fluid flow .
it can have a quantum mechanical origin in , for example , systems exhibiting the quantum hall effect @xcite , or a classical origin in plasmas at finite - temperature @xcite .
we will not focus on the microscopic origin of the hall viscosity coefficient , but only assume it to be non - vanishing in conjunction with the usual viscosity coefficients . from this assumption
we will determine the motion of swimmers at low reynolds number in the presence of hall viscosity .
specifically , we will consider the problem of swimmers with circular boundaries that move via deformations of their boundaries analogous to the nearly - circular swimmers in ref .
@xcite .
we find a general result that connects the torque on a swimmer to the rate of area change of the swimmer with a proportionality constant given by the hall viscosity .
we use our results to give examples of swimmer motion due to cyclic circular deformations and compare cases where the conventional and hall viscosities each dominate .
our paper is organized as follows : we first review the geometric formulation of swimming and the appearance of hall viscosity in 2d fluids with broken time - reversal symmetry .
we then go on to derive the general consequences of the hall viscosity on swimmers and then give explicit examples of model swimming strokes that illustrate some differences between fluids with vanishing and non - vanishing hall viscosity .
finally , we have some appendices which collect derivations of the more technical results .
note that we will use hall viscosity and odd viscosity interchangeably throughout the text . |
the double - helix structure of dna is the source of many complications in its _ in - vivo _ functioning during condensation / decondensation , replication , or transcription .
for example the necessary unzipping of the molecule during transcription induces torsional strain along dna .
the development of magnetic tweezers @xcite or the use of quartz cylinders in optical tweezers @xcite allow researchers to investigate the _ in - vitro _ response of dna molecules to torsional stress .
studies of the behaviour of this twist storing polymer are not just a game for physicists as it has been clearly established that , e.g. the assembly of reca could be stalled by torsional constraints @xcite , or that the rate of formation and the stability of the complex formed by promoter dna and rna polymerase depends on the torque present in the dna molecule @xcite . on the theoretical side ,
matters are made difficult by the nonlocality of the topological property that is associated with the torsional constraint : the link .
the two sugar - phosphate backbones of dna have opposite orientation and the ends of a double - stranded dna molecule can only be chemically bound in such a way that each strand joins itself , thereby yielding two interwound closed curves ( no mbius - like configuration can exist ) . for a circularly closed dna molecule ,
the link is the number of times one of the sugar - phosphate backbone winds around the other .
once the three dimensional shape of the molecule is projected onto a plane , the link is given by half the number of signed crossings between the two backbones .
this quantity is best seen as the number of turns put in an initially planar plasmid ( a piece of circularly closed double - stranded dna ) before closing it .
link has been shown to consist of two parts : the two sugar - phosphate backbones of a plasmid can be linked because @xmath0 the plasmid lies in a plane but the base - pairs are twisted around their centre line ( the curve joining the centroids of the base - pairs ) and/or @xmath1 the centre line itself follows a writhed path in space . in general the two possibilities coexist and the link @xmath2 of a dna molecule is the addition of the two quantities @xcite : @xmath3 the twist @xmath4 is a local quantity in the sense that it can be computed as the single integral of the twist rate @xmath5 : @xmath6 , where @xmath7 is the arclength along the centre line and @xmath8 the total contour length of the molecule . in elastic dna models , the twist rate @xmath5
is normally coupled to mechanical quantities ( e.g. the torque ) characterizing the molecule .
contrary to the twist , wr is a _ global _ property of the centre line of the molecule .
we first define the _ directional writhe _ of a closed curve @xmath9 .
consider a closed curve in 3d and project this curve , along a certain direction , on a plane .
the number of signed crossings seen in the plane is the directional writhe _ for that direction_. one could then reiterate the procedure with different directions of projection and compute the directional writhe for each direction .
the average value obtained for the directional writhes , when all directions are considered , is the writhe @xmath10 of the 3d curve @xmath9 @xcite .
the following double integral has been introduced by clugreanu @xcite and white @xcite : @xmath11 where @xmath12 is the position of points on @xmath9 and @xmath13 is the unit tangent to @xmath9 .
as soon as a 3d curve does not self - intersect , the writhe @xmath10 of the curve is given by eq .
( [ equa : writhe_def ] ) : @xmath14 if @xmath15 @xmath16 with @xmath17 . the computation of the double integral of eq .
( [ equa : writhe_def ] ) is analytically hard and numerically time consuming to perform .
an important result though enables one to reduce the double integral to a single integral , provided several hypotheses are fulfilled @xcite .
it is the main purpose of this paper to show that these hypotheses are not met in the case of plectonemic dna .
fuller s theorem @xcite states that the writhe of the curve @xmath9 can be computed by considering the writhe of a reference curve @xmath18 ( which should be known or easy to compute ) and the continuous deformation ( a homotopy ) morphing @xmath18 to @xmath9 : @xmath19 where @xmath20 is the unit tangent to @xmath18 and where @xmath21 $ ] is a common parametrization for both curves , though not necessarily the arclength along @xmath9 or @xmath18 .
the total contour length of @xmath9 is @xmath8 .
the curve @xmath18 is normally chosen to be a fairly simple curve ( e.g. planar with zero writhe ) so the computation of @xmath10 boils down to estimating the single integral in eq .
( [ equa : fuller_2nd_theo ] ) . before doing so
, one should verify whether the hypotheses of fuller s theorem are fulfilled ( this is the sense of the question mark in eq.([equa : fuller_2nd_theo ] ) ) .
the continuous deformation from @xmath18 to @xmath9 introduces a familly of curves @xmath22 with @xmath23 $ ] and @xmath24 .
the unit tangent to @xmath22 is @xmath25 .
the first hypothesis is that none of the curves @xmath22 self - intersects .
the second hypothesis is that , and this is the point we want to emphasize , there should be no point along any of the curves @xmath22 where @xmath26 .
for each value of @xmath27 , the ( unit ) tangent @xmath28 with @xmath21 $ ] defines a curve , called the tangent indicatrix , on the unit sphere .
if for a certain @xmath29 and a certain @xmath30 we have @xmath31 then , on the unit sphere , the point corresponding to @xmath32 is antipodal to the point corresponding to @xmath33 . for brevity we shall also call antipodal the point @xmath34 or
the curve @xmath35 itself . in models where self - intersection is prevented ( e.g. models using hard- or soft - wall potentials of the molecule on itself ) a continuous deformation free of self - intersection ( an isotopy ) may be easy to devise and
the first hypothesis would be verified .
however nothing in these models ensure that the second hypothesis ( that no antipodal points exist for all @xmath23 $ ] and all @xmath21 $ ] ) is met and consequently fuller s formula ( [ equa : fuller_2nd_theo ] ) can not be used to compute the writhe .
we nevertheless remark that even in the case of self - intersection or antipodal points fuller s formula is always correct _ modulo _ 2 : the integer , and most important , part of @xmath10 will not be correct , but the fractional part will .
each antipodal point present in the continuous deformation from @xmath18 to @xmath9 introduces a shift of two units between the actual writhe @xmath10 of the curve @xmath9 and the value @xmath36 @xcite .
consequently , if there are @xmath37 antipodal points along the deformation @xmath23 $ ] , the computed value @xmath38 could be as far as @xmath39 from the correct value @xmath40 , but the fractional part will be accurate .
this discrepancy between @xmath38 and the actual writhe @xmath40 has already been pinpointed in the case where dna is treated as a fluctuating chain under low twist ( i.e. without plectonemes ) @xcite .
nevertheless we have found a certain number of references where fuller s formula is used @xcite . in most cases the hypotheses of fuller s
theorem were not checked , if only mentioned . in
some works the formula is used in a scheme that provides an estimate for the torsional stiffness of the dna molecule @xcite and this has been shown to lead to incorrect results @xcite .
nevertheless , in some few papers @xcite , the formula was used to asses the writhe of dna configurations under high stretching force and low twist , in which cases antipodal points are absent and the formula is correct .
during a force - extension experiment on a single dna molecule thermal agitation deforms the molecule whose shape locally adopts random directions ( fig . [
fig : antipodal_wlc ] , left ) . in the absence of ( or under low ) twist ,
dna is modeled as a worm - like chain @xcite ; the path followed by its centre line in space looks like a ( directed ) random walk @xcite . in such configurations ,
the writhe is usually evaluated using fuller s formula ( [ equa : fuller_2nd_theo ] ) with the reference curve shown in fig .
[ fig : antipodal_wlc ] , left .
since the writhe is classically defined for a closed curve , the reference and actual curves are all closed by imaginary @xmath41-shaped curves ( dashed in fig .
[ fig : antipodal_wlc ] ) that connect the top to the base of the configurations .
( other choices of closures and interferences due to the closure in the computation of the writhe are discussed in @xcite . ) strictly speaking the curve @xmath9 for which we compute the writhe consists of two parts : the closure and the part corresponding to the molecule . as the closure remains unchanged in the continuous deformation
@xmath23 $ ] its contribution to fuller s integral is zero .
consequently we will forget about the closure and call @xmath9 the ( open ) curve described by the molecule only .
the reference curve @xmath18 is then the @xmath42 axis , and fuller s formula ( [ equa : fuller_2nd_theo ] ) becomes : @xmath43 if the tangent @xmath44 is parametrized by euler angles @xmath45 $ ] , and @xmath46 . as stated above , eq .
( [ equa : fuller_euler_angle ] ) is only valid if the curve @xmath9 can be deformed into the reference curve @xmath18 without passing through configurations @xmath22 having their tangent vector @xmath28 facing -@xmath47 : @xmath48 for all @xmath27 , @xmath49 .
we remark that in the present case where @xmath18 is the @xmath42 axis an antipodal point is a point where the curve @xmath28 passes through the south pole of the unit sphere : @xmath50 . , @xmath51 , @xmath41 , @xmath52 , and @xmath53 ( see fig . [
fig : diag_z_de_n ] ) obtained by numerical continuation @xcite . configurations @xmath51 and @xmath52 each have an antipodal point to the reference curve of fig .
[ fig : antipodal_wlc ] .
these antipodal points are located at the middle point of the end - loop of the plectonemic structure.,title="fig : " ] , @xmath51 , @xmath41 , @xmath52 , and @xmath53 ( see fig . [
fig : diag_z_de_n ] ) obtained by numerical continuation @xcite .
configurations @xmath51 and @xmath52 each have an antipodal point to the reference curve of fig .
[ fig : antipodal_wlc ] .
these antipodal points are located at the middle point of the end - loop of the plectonemic structure.,title="fig : " ] in the case where the actual configuration @xmath9 exhibits a single or multiple antipodal points it has been proposed in @xcite to evaluate the writhe from eq .
( [ equa : fuller_euler_angle ] ) with euler angles defined on a truncated unit sphere : @xmath54 would not be allowed to reach @xmath55 and hence antipodal points would be avoided ( the curve @xmath56 so defined would be very near the real curve @xmath9 , hence the writhes would almost be the same ) .
we stress that this process is not sufficient as even for a given curve with no antipodal points , it is not clear whether eq .
( [ equa : fuller_euler_angle ] ) is valid or not . in order for eq .
( [ equa : fuller_euler_angle ] ) to be valid , one has to exhibit a continuous deformation from the @xmath42 axis to the curve @xmath9 that is entirely free of antipodal points .
there are many cases where the actual curve @xmath9 is free of antipodal points , but where a deformation free of antipodal points does not exist . consequently in these cases eq .
( [ equa : fuller_euler_angle ] ) yields an incorrect result ( unless antipodal points of opposite signs cancel out ) .
nevertheless , it has been noted that under high stretching force ( e.g. @xmath57 pn ) and low torque the dna molecule is almost straight and no such antipodal points exist @xcite .
in such a case fuller s formula is correct provided no plectonemes are present , as we will see now . .writhe ( first line ) and link ( third line ) of the configurations of fig .
[ fig : antipodal_plectonems ] , computed by continuation or use of the double integral ( eq .
( [ equa : writhe_def ] ) ) .
the second line is computed from eq .
( [ equa : fuller_euler_angle ] ) and the last line is computed from eq .
( [ equa : link_phi_psi ] ) . formulas ( [ equa : fuller_euler_angle ] ) and ( [ equa : link_phi_psi ] ) are not applicable on configurations @xmath51 and @xmath52 that each have an antipodal point , and yield incorrect results for configurations @xmath41 and @xmath53 . [ cols="<,^,^,^,^,^",options="header " , ] , @xmath51 , @xmath41 , @xmath52 , and @xmath53 of fig . [
fig : antipodal_plectonems ] .
an antipodal point arises when the curve passes through the south pole of the sphere , i.e. for configurations @xmath51 and @xmath52.,title="fig : " ] + , @xmath51 , @xmath41 , @xmath52 , and @xmath53 of fig .
[ fig : antipodal_plectonems ] .
an antipodal point arises when the curve passes through the south pole of the sphere , i.e. for configurations @xmath51 and @xmath52.,title="fig : " ] in magnetic tweezer experiments , when a large amount of turns are put in ( by rotation of the magnetic bead around the @xmath42 axis ) , the dna molecule reacts by forming plectonemes @xcite .
the number of turns @xmath58 imposed on the magnetic bead is given by the link @xmath2 of the molecule .
we now show that the presence of plectonemes in the supercoiled configuration prevents the existence of a deformation , from @xmath18 ( i.e. the @xmath42 axis ) to @xmath9 , that is free of antipodal points , and consequently forbid the use of eq.([equa : fuller_euler_angle ] ) . in terms of the euler angles
the twist of the molecule can be computed as @xmath59 where @xmath60 is the third euler angle ( see e.g. @xcite ) .
using eqs .
( [ equa : fuller_euler_angle ] ) and ( [ equa : twist ] ) we obtain : @xmath61 this formula usually is the starting point for computations of the link of supercoiled configurations , see e.g. @xcite or @xcite .
we show here that it yields incorrect results when plectonemes are present .
we have performed computations to model the elastic response of a twist storing filament subject to tensile and torsional constraints and we quantitatively reproduced the plectonemic regime characterised by the linear decrease of the end - to - end distance of the filament as a function of the number @xmath58 of turns put in @xcite .
the plectonemic configurations were computed numerically using a continuation algorithm and @xmath62 was computed by continuity so that no integer number of turns is missed ( we also performed numeric integration of the double integral of eq .
( [ equa : writhe_def ] ) for the writhe of the configurations , with closures , and always obtained consistent results ) .
these shapes serve as an illustration for the computation of the writhe and consequently are used for their geometry only .
the fact that they are mechanical equilibria is not relevant here .
the continuous dark curve in fig . [ fig : diag_z_de_n ] shows an output of the numerics , drawn in the @xmath63 plane , where @xmath64 is the vertical extension of the molecule .
the curve starts at point @xmath65 which corresponds to a straight and twisted configuration .
the path is then monotonically decreasing in @xmath64 .
we have selected five configurations @xmath66 , @xmath51 , @xmath41 , @xmath52 , and @xmath53 which are drawn in fig .
[ fig : antipodal_plectonems ] , together with their corresponding tangent indicatrices ( see definition in section [ section : global_and_local ] ) drawn in fig .
[ fig : unit_spheres ] . configurations @xmath51 and @xmath52 each comprise an antipodal point located at the middle point of the end - loop of the plectonemic structure . this can be verified in fig .
[ fig : unit_spheres]-@xmath51 and @xmath52 where the tangent indicatrices pass through the south pole of the unit sphere . ) .
five configurations @xmath66 , @xmath51 , @xmath41 , @xmath52 , and @xmath53 are selected and plotted in fig .
[ fig : antipodal_plectonems ] together with their tangent indicatrices in fig .
[ fig : unit_spheres ] .
the inset shows a zoom around point @xmath51 , the first antipodal point , where the gray curve jumps by an amount of -2 units . from @xmath65
to @xmath51 the two curves coincide . ]
using the same geometric configurations , we compute @xmath67 from eq .
( [ equa : link_phi_psi ] ) and we plot the corresponding curve , in gray , on the same diagram . we see that each antipodal event introduces a shift of two units in the gray curve , which is consequently broken .
as expected this confirms that eq .
( [ equa : link_phi_psi ] ) is only valid _ modulo _ 2 and hence should not be used to estimate the link of plectonemic configurations .
the writhe and link of the five example configurations are given in table [ table : wr_lk_wrf_lkf ] where they are compared with @xmath38 and @xmath67 given by eqs .
( [ equa : fuller_euler_angle ] ) and ( [ equa : link_phi_psi ] ) . for configuration @xmath53 ,
which is separated from the reference curve by two antipodal events , we see that @xmath38 ( resp .
@xmath67 ) is 4 units away from the correct @xmath40 ( resp .
@xmath2 ) value .
this configuration @xmath53 is an illustration of the fact that fuller s formulas ( eq . ( [ equa : fuller_euler_angle ] ) or ( [ equa : link_phi_psi ] ) ) can be wrong even for a configuration that does not comprise any antipodal point , which is clearly illustrated in fig .
[ fig : unit_spheres]-@xmath53 where we see that the tangent indicatrix is nowhere near the south pole .
we first comment on the size of the gap between the two curves .
a plectonemic dna configuration @xmath9 that shows @xmath68 ( positive ) crossings on a lateral projection has @xmath69 @xcite . if we continuously deform this configuration to the reference curve @xmath18 of fig .
[ fig : antipodal_wlc ] left by unwinding the plectonemic region , an antipodal point arises each time the tangent at the apex of the terminal loop ( at the end of the plectonemic region ) is facing downward .
this happen @xmath70 times .
as proved in @xcite the presence of @xmath71 antipodal points leads to a discrepancy in fuller s formula of @xmath39 : @xmath72 .
since @xmath73 , we have that : @xmath74 , which corresponds to an error of up to 100% .
this is apparent in fig .
[ fig : diag_z_de_n ] where the broken gray curve stays near the vertical axis while the ( continuous ) dark curve monotonically increases in link .
second we note that the discrepancy between the two curves occurs shortly after self - contact has started in the filament , for @xmath75 , see inset of fig .
[ fig : diag_z_de_n ] .
plectonemic structure may exist for small number of turns @xmath58 , provided the pulling force is not too large .
on the other hand , a large pulling force does not rule out the occurence of plectonemes , provided that @xmath58 is large enough .
this leaves a small parameter regime ( large pulling force , low number of turns ) where plectonemes are absent . when the two sources of discrepancies are considered ( random walk antipodal points @xcite , and plectonemic antipodal points ) one sees that the use of eq .
( [ equa : fuller_euler_angle ] ) ( resp .
( [ equa : link_phi_psi ] ) ) to compute the writhe ( resp .
the link ) in a model for dna under tensile and/or torsional stress is to be avoided unless @xmath76 and the tensile force is large .
we summarize here few properties of the quantities @xmath77 , @xmath78 which do not always give the writhe of a curve @xmath9 .
the quantity @xmath77 is a function of the curve @xmath9 only , whereas the quantity @xmath78 also depends on the reference curve @xmath18 .
the quantity @xmath79 yields the correct value for the writhe @xmath40 of a closed curve as soon as the curve is not self - intersecting . along a continuous deformation @xmath22 with @xmath23 $ ] ( with @xmath24 ) , the quantity @xmath80 jumps by two units when the curve @xmath22 intersects itself ( say at @xmath81 ) .
since the writhe @xmath40 also jumps by two units , the quantity @xmath79 is equal to @xmath40 before and after the self - crossing event , i.e. @xmath82 , @xmath83 .
now the quantity @xmath84 has no such discontinuity : along a continuous deformation where the curve self - intersects , the writhe @xmath40 will jump by two units , but the quantity @xmath84 will stay continuous .
this means that the quantity @xmath84 no longer yields the correct value for the writhe _
after _ the self - crossing event , @xmath85 , @xmath86 but @xmath87 , @xmath88 . the same is true for antipodal points . along a continuous deformation @xmath22 with @xmath23 $ ] , the quantity @xmath89 jumps by two units when the curve @xmath22 has an antipodal point ( say at @xmath81 ) with regard to the reference curve @xmath18 . on the other hand
the writhe @xmath40 stays continuous .
this means that the quantity @xmath84 no longer yields the correct value for the writhe as soon as an antipodal event happens , @xmath85 , @xmath86 but @xmath87 , @xmath88 .
this is an important point and many authors seem to believe that the quantity @xmath89 only has problems for configurations actually comprising an antipodal point ( @xmath81 in the above example )
. therefore the quantity @xmath78 may not be equal to the writhe @xmath10 even for curves that _ do not _ comprise any antipodal point .
this make the use of @xmath78 uneasy , as one has to first verify the absence of antipodal events in the _ entire _ continuous deformation @xmath90 . on the contrary
the use of @xmath77 is much easier in the sense that one just has to check that the actual curve @xmath9 does not self - intersect . in this sense
the quantities @xmath77 and @xmath78 do not suffer from the same pathologies in the computation of the writhe , contrarily to what is claimed in @xcite .
another consequence is that , when sampling dna configurations to construct a statistical ensemble and compute writhe averages and fluctuations , it is not enough to introduce , as was done in @xcite , a small forbidden region around the south pole of the unit sphere to ensure that @xmath78 yields a correct value .
in fact many of these sampled configurations , even free of antipodal points , are configurations that suffer the same problems as the configurations with @xmath91 above : @xmath92 , as numerically verified in @xcite
. finally we want to point out the following property .
we saw that if a continuous deformation @xmath22 with @xmath23 $ ] contains an antipodal event , then @xmath78 does not yield the correct result after the antipodal event . now one can argue that yet another continuous deformation , free of antipodal events and self - crossings , may exist ( with the same initial and final curves ) and that in this case @xmath78 would yield the correct result .
this is not the case since , as we show in appendix b , as soon as a corrupted deformation exists in between two curves @xmath18 and @xmath9 , then _ all _ continuous deformations are corrupted and @xmath78 definitely yields an incorrect result .
this means that the reference curve @xmath18 can not be used .
one is bound to find another reference curve or to use the double integral @xmath79 .
it appears that an easy way to verify the applicability of fuller formula ( [ equa : fuller_2nd_theo ] ) between an actual curve and a reference curve is to consider any convenient deformation that avoids antipodal events .
this can always be done if one allows for self - crossings .
then formula ( [ equa : fuller_2nd_theo ] ) is applicable if and only if the net sum of self - crossings is zero . in the case of numerical computations ,
either dealing with continuation of mechanical equilibria or statistical ensembles of configurations , the writhe can be assessed in an accurate way by using both the double integral ( [ equa : writhe_def ] ) and fuller integral ( [ equa : fuller_euler_angle ] ) in a cooperative way .
the double integral being time consuming to evaluate , one can in a first step discretize it @xcite with a reduced number of elements so that it produces an approximate result that only has to be accurate up to @xmath93 ( one still has to estimate how many elements are needed to obtain such an accuracy @xcite ) . then in a second step , fuller integral is used to refine the result .
even if fuller integral may be off by several integers , its fractional part is correct . in this scheme , the double integral yields the integer part of the writhe and fuller integral yields the fractional part of the writhe .
moreover in cases where one knows the correct value of the writhe of a nearby configuration ( e.g. a predecessor configuration , one move away in a monte - carlo scheme ) this value can be used as the approximate result of the first step and one only has to compute fuller integral to obtain a correct and accurate value of the writhe .
( the usual assumption that there is no self - crossing between the predecessor and the actual configuration still holds . ) in conclusion we have shown , by producing counter - examples and explaining the underlying causes , that formulas ( [ equa : fuller_euler_angle ] ) and ( [ equa : link_phi_psi ] ) can not be used to compute the writhe and the link of supercoiled dna configurations encountered in magnetic tweezer experiments .
the work of e.s . on this article was supported by the uk s engineering and physical sciences research council under grant number gr / t22926/01 .
in this appendix we compare , for a given example curve , the value computed from the double integral @xmath79 to the value computed from the single integral @xmath38 .
we start by introducing the parametrization of a circularly closed supercoiled plasmid ( see fig . [
fig : clamped_ply ] ) .
the parametrization is divided in four parts , called @xmath66 , @xmath51 , @xmath41 , and @xmath52 .
part @xmath66 is a right - handed helix of radius unity , total arclength @xmath94 , and helical angle @xmath95 ( with @xmath96 ) : @xmath97 with @xmath98 $ ] and @xmath99 .
we note @xmath100 .
we have @xmath101 for all @xmath102 , and consequently @xmath102 is the arc length in part @xmath66 .
the end loop @xmath51 connects the two helices @xmath66 and @xmath41 : @xmath103 with @xmath104 $ ] , and @xmath105 . for this part @xmath51 , @xmath106 .
part @xmath41 is a right - handed helix of radius unity , total arclength @xmath107 and helical angle @xmath108 : @xmath109 with @xmath110 $ ] and @xmath111 .
we have @xmath112 for all @xmath113 , and consequently @xmath113 is the arclength in part @xmath41 . the end loop @xmath52 closes the curve : @xmath114 with @xmath115 $ ] , and @xmath116 . for this part @xmath52 , @xmath117 .
we call @xmath9 the union of the four parts : @xmath118 .
the curve has continuous derivatives , and finite jumps in its curvature . with @xmath119 and @xmath120 .
] the evaluation of the double integral of eq .
( [ equa : writhe_def ] ) in the limit @xmath121 yields @xcite : @xmath122 the minus sign is due to the fact that the helices @xmath66 and @xmath41 are right - handed .
we then compute @xmath123 , given by eq .
( [ equa : fuller_2nd_theo ] ) , using the reference curve @xmath18 of fig .
[ fig : two_reference_curves_plasmid ] , top . the continuous deformation between @xmath18 and @xmath9 is then parametrized by @xmath124 $ ] . the reference curve of fig .
[ fig : two_reference_curves_plasmid ] top comes to mind naturally and makes calculations easiest , but yields antipodal events and hence an incorrect result , as we shall see now . for the supercoiled plasmid of fig .
[ fig : clamped_ply ] . ]
the writhe of the reference curve @xmath18 is zero .
consequently we only have to compute the integral @xmath125 of eq .
( [ equa : fuller_2nd_theo ] ) over the four different parts @xmath66 , @xmath51 , @xmath41 , and @xmath52 .
clearly when the plasmid is long enough , i.e. when @xmath94 is large , the contribution to @xmath125 of the helical parts @xmath66 and @xmath41 become dominant
. indeed their contribution scales with @xmath94 while the contribution of the end loops @xmath51 and @xmath52 remains bounded .
consequently we focus on the contributions of the helical parts @xmath66 and @xmath41 , for large values of @xmath94 , that is we look at the limit @xmath126 when @xmath127 @xcite . for the helical part @xmath66 ,
the corresponding tangent of the reference curve is @xmath128 .
fuller integral for part @xmath66 is then : @xmath129 for the helical part @xmath41 , the corresponding tangent of the reference curve is @xmath130 . the fuller integral for part @xmath41 is then : @xmath131 neglecting the end loop contributions we arrive at : @xmath132 we first remark that @xmath133 although @xmath134 ( see eq .
( [ equa : double_integral_formula ] ) ) .
the discrepancy @xmath135 is due to the fact that in the continuous deformation @xmath124 $ ] from the curve @xmath18 of fig .
[ fig : two_reference_curves_plasmid ] top to the plasmid of fig .
[ fig : clamped_ply ] , there are antipodal events .
these events happen each time the middle point of the end loop @xmath51 is pointing towards @xmath136 and hence become antipodal to the corresponding tangent in @xmath137 .
as @xmath138 is increased , such an event happens once each time @xmath139 increases by an amount of @xmath140 . in the entire deformation
there will be @xmath141 antipodal events .
each event introduces a shift between @xmath77 and @xmath36 of @xmath142 , the difference @xmath143 should then be equal to @xmath144 , which can be verified by comparing eq .
( [ equa : double_integral_formula ] ) to eq .
( [ equa : wrong_fu_classic_reference_curve ] ) .
a natural way to try to avoid the presence of these antipodal points is to rotate the reference curve , e.g. choose @xmath145 of fig .
[ fig : two_reference_curves_plasmid ] bottom as reference curve . in this case @xmath146 and @xmath147 . here
again the sum @xmath148 dominates @xmath149 .
integration shows that @xmath150 , with @xmath151 , which is not the correct result .
antipodal events are in fact still present with this rotated reference curve @xmath152 and trying another rotation will not help , as explained in appendix b. in fig .
[ fig : clamped_ply ] , a right - handed circularly closed plasmid is drawn with @xmath119 , @xmath153 .
the viewpoint selected to draw the curve in fig . [
fig : clamped_ply ] is such that three ( negative ) crossings appear .
for some other viewpoints , only two crossings appear .
the writhe being the average number of signed crossings one sees from all possible viewpoints , it is relatively easy to convince oneself that its value for the curve of fig .
[ fig : clamped_ply ] lies in between -2 and -3 . a numeric discretization scheme @xcite of the double integral of eq .
( [ equa : writhe_def ] ) with 600 points yields @xmath154 . a numerical integration of @xmath155 yields @xmath156 .
the difference between @xmath77 and @xmath36 is -2 , which indicates that the net number of signed antipodal events is -1 , as shown in appendix b.
the ` stadium ' shaped curve @xmath157 is continuously deformed through curves @xmath22 , as @xmath27 increases from @xmath158 to @xmath159 , to the supercoiled plasmid @xmath160 .
the plasmid is then deformed back to the ` stadium ' shaped curve , through another path , as @xmath27 increases from @xmath159 to @xmath161 .
the values @xmath158 and @xmath161 refer to the same curve , the reference curve , which plays a special role .
this reference curve is chosen to be fairly simple , in particular @xmath162 .
all the curves in the continuous deformation share a parametrization with parameter @xmath7 ( not necessarily the arclength ) taking values from @xmath163 to @xmath164 .
we recall that an antipodal point is a point on a curve @xmath22 such that the tangent @xmath165 to @xmath22 at @xmath166 is aligned and in opposite direction with the tangent @xmath167 to the reference curve @xmath18 , i.e. an antipodal point is such that @xmath168 .
an antipodal point is always defined with regard to a certain reference curve . along a continuous deformation
there may be @xmath27 values where the curves @xmath22 have antipodal points to @xmath18 , and there may be @xmath27 values where the curves @xmath22 self - intersect . _
the closed circuit theorem _ claims that in any closed circuit the ( signed ) number of self - intersections is equal to the ( signed ) number of antipodal points .
for example , in the closed circuit of fig .
[ fig : closed_circuit ] there is one antipodal point and one self - crossing .
this theorem was proposed and proven in @xcite for a particular case of reference curve , a straight line . here
we give an outline of a proof which is valid for more general reference curves : formula ( [ equa : writhe_def ] ) has a @xmath169 discontinuity at each self - crossing event encountered in the closed circuit , but no discontinuity during antipodal events . on the contrary formula ( [ equa : fuller_2nd_theo ] ) has no discontinuity during self - crossing events but a @xmath169 discontinuity at each antipodal event @xcite .
formula ( [ equa : writhe_def ] ) and ( [ equa : fuller_2nd_theo ] ) by definition agree when applied on the reference curve , and furthermore they may only differ by an even integer : @xmath170 . if we now follow the closed circuit , each formula will have its own discontinuities but both must eventually agree again at the end of the circuit .
it follows that the number of discontinuities ( respectively events ) of one kind must be equal to the number of discontinuities ( respectively events ) of the other kind . the quantity @xmath77 is not defined on curves that are self - intersecting , e.g. curve @xmath41 in fig .
[ fig : closed_circuit ] . the quantity @xmath36 is not defined on curves @xmath9 that have an antipodal point to @xmath18 , e.g. curve @xmath66 in fig .
[ fig : closed_circuit ] .
these two properties , once considered with the fact that in any closed circuit the number of configurations with self - crossing is equal to the number of configurations with antipodal points , could lead one to believe that some parallel exists between @xmath77 and @xmath36 .
in fact in @xcite it is proposed that , in the computation of the mean writhe ( or link ) of a statistical ensemble of curves , avoiding self - crossings was equivalent to avoiding antipodal points .
moreover it was infered that for such statistical ensemble of curves , fuller s formula , eq .
( [ equa : fuller_2nd_theo ] ) , could be used safely provided that curves with antipodal point(s ) were discarded when generating the statistical ensemble .
we argue that this is not the case as fuller s formula does not yield the right result for the ( many ) curves that have no antipodal point : e.g. curve @xmath51 in fig .
[ fig : closed_circuit ] .
again , fuller s formula can only be used for a curve @xmath35 if one can devise a continuous deformation from @xmath35 to @xmath18 where _ none _ of the curves @xmath22 with @xmath171 has antipodal points . for curve @xmath51 in fig .
[ fig : closed_circuit ] this is not the case since the curve @xmath66 has an antipodal point .
accordingly fuller s formula does not yield the correct value for the writhe of curve @xmath51 , as explicitly computed in appendix a. in conclusion we stress that there is no parallel between the two formulas @xmath77 and @xmath36 : on the one hand @xmath77 yields the correct value of the writhe of a curve as soon as the curve does not self - intersect and on the other hand @xmath36 often yields an incorrect value for the writhe of a curve even if this curve has no antipodal point .
all what is exposed here for the curves in fig .
[ fig : closed_circuit ] is directly applicable to the configurations of fig .
[ fig : antipodal_plectonems ] and shows that the writhe ( resp . the link ) of the configurations with @xmath172 in fig .
[ fig : diag_z_de_n ] is not given by eq .
( [ equa : fuller_2nd_theo ] ) ( resp .
( [ equa : link_phi_psi ] ) ) . [ [ impossibility - to - use - the - stadium - shaped - curve - as - reference - curve ] ] impossibility to use the ` stadium ' shaped curve as reference curve ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ fuller s theorem @xcite states that as soon as there is a continuous deformation between @xmath9 and @xmath18 which is free of antipodal points and self - crossings ( we called ` good ' such a deformation ) , @xmath36 yields the correct value for the writhe of the closed curve @xmath9 . in fig . [
fig : closed_circuit ] the deformation from @xmath158 to @xmath159 is a ` bad ' deformation as it contains one antipodal event .
we then conclude that @xmath173 yields an incorrect value .
but one might say that another deformation between @xmath174 and @xmath18 , with neither self - crossing nor antipodal event , could exist .
we show here that this is not the case .
_ statement _ : no ` good ' deformation between @xmath174 and @xmath18 of fig .
[ fig : closed_circuit ] exists . _ proof _ : assume a ` good ' deformation exists .
then we can build a closed circuit by joining this proposed deformation and the ( reversed ) actual deformation @xmath23 $ ] in fig .
[ fig : closed_circuit ] . one
would then have a closed circuit with one antipodal point and no self - crossing , in contradiction with the result established in @xcite .
hence no such ` good ' deformation exists.@xmath175 this means that as soon as one ` bad ' deformation exists between a reference curve @xmath18 and a curve @xmath9 the formula @xmath36 does not yield the correct result and no ` good ' deformation can exist .
a consequence of this is that rotating the reference curve of appendix a ( fig .
[ fig : two_reference_curves_plasmid ] , top ) to introduce a new , and ` good ' , reference curve ( e.g. fig .
[ fig : two_reference_curves_plasmid ] , bottom ) is hopeless . to show this we first introduce a ` bad ' deformation between the rotated reference curve ( e.g. fig .
[ fig : two_reference_curves_plasmid ] , bottom ) and the supercoiled plasmid of fig .
[ fig : clamped_ply ] : we untangle the plasmid using a self - crossing , as in fig .
[ fig : closed_circuit]-c , to obtain a ` stadium ' shaped curve which we subsequently rotate to the ( rotated ) reference curve .
the existence of this ` bad ' deformation means that no ` good ' deformation can exist . | the linking and writhing numbers are key quantities when characterizing the structure of a piece of supercoiled dna . defined as double integrals over the shape of the double - helix , these numbers are not always straightforward to compute , though a simplified formula exists @xcite .
we examine the range of applicability of this widely - used simplified formula , and show that it can not be employed for plectonemic dna .
we show that inapplicability is due to a hypothesis of fuller theorem @xcite that is not met .
the hypothesis seems to have been overlooked in many works .
* pacs numbers * : 87.10.-e , 87.14.gk , 02.40.-k , 87.15.-v + |
the response of a conducting system to an alternating electric field is one of the central topics of research in solid - state electrodynamics . besides the linear response , where the induced electric current oscillates at the frequency of the electric field and its amplitude linearly scales with the field amplitude , ac electric field can give rise to a direct motion of charge carriers .
such an electric rectification ( or electronic ratchet effect ) naturally occurs in macroscopically inhomogeneous structures such as diodes or field - effect transistors,@xcite asymmetric lattices,@xcite or systems with asymmetric patterning.@xcite the ratchet transport of carriers arises also in macroscopically homogenous structures ( homogenous in all three dimensions for bulk materials or homogenous in the plane for two - dimensional systems ) provided the structures lack the center of space inversion.@xcite the ratchet effects are used for the study of spatial symmetry of semiconductor structures,@xcite details of the electron energy spectrum,@xcite and also underlie the operation of fast detectors of microwave and terahertz radiation.@xcite the efficiency of the generation of a dc current by an ac electric field can be considerably enhanced in an external magnetic field if the frequency of the ac field is close to the carrier cyclotron frequency . previously , such an enhancement of the electric response at the cyclotron resonance conditions was demonstrated for the spin currents in hgte quantum wells ( qws)@xcite and hgte - based three - dimensional topological insulators@xcite and for the electric currents of photon drag induced by teraherzt radiation.@xcite the orbital ratchet transport of electrons in macroscopically homogenous qws in the presence of a normal magnetic field has not been analyzed so far . here , we develop a quasi - classical microscopic theory of the orbital ratchet effect in qws subjected to a static magnetic field and show that the arising current drastically increases at the cyclotron resonance .
we consider the ratchet current caused by the qw structure inversion asymmetry which may originate from an asymmetry in the confinement potential or doping profile , or induced by a gate voltage .
two geometries , at which the ratchet current emerges and which can be experimentally realized , are analyzed : ( i ) the static magnetic field is normal to the qw plane and the ac electric field has both the in - plane and out - of - plane components and ( ii ) the magnetic field is tilted with respected to the qw normal and the ac electric field is polarized in the qw plane . for the latter geometry
we find that the dc electric current has an additional resonant contribution .
this resonance occurs when the frequency of the ac field matches the double cyclotron frequency , in the spectral range where the free - carrier absorption has no peculiarity .
we also calculate the spatial distribution of the electrostatic potential induced by the ratchet current in infinite and finite - size samples and show that the distribution depends on the magnetic field strength , the sample geometry , and the boundary conditions used .
we begin the study with the geometry of the static magnetic field @xmath0 pointing along the qw normal @xmath1 , see fig .
[ geometry_f ] . in this configuration ,
a dc electric current @xmath2 emerges if the ac electric field @xmath3 where @xmath4 and @xmath5 are the field amplitude and frequency , has both the in - plane @xmath6 and out - of - plane @xmath7 components .
figure [ geometry_f ] illustrates the microscopic mechanism of the dc current formation .
the in - plane component @xmath8 of the electric field together with the static magnetic field @xmath0 causes the motion of electrons in elliptical orbits in the qw plane at the field frequency @xmath5 .
synchronously with this in - plane motions , the electric field out - of - plane component @xmath9 pushes the electrons to the top or bottom interfaces of the qw depending on the field polarity .
the corresponding distributions of the electron density in the qw cross section for positive and negative @xmath10 , where @xmath11 is the electron charge , are sketched in the insets .
the shift of the electron density along the @xmath1 axis in the asymmetric qw results , in turn , in the modulation of the electron mobility at the field frequency @xmath5 . in the insets in fig .
[ geometry_f ] , the qw asymmetry is modeled by placing the @xmath12-layer of impurities ( black dots ) , which cause electron scattering and control the electron mobility , closer to the bottom interface . as a result , the oscillating electron motion in the qw plane driven by @xmath8 together with the mobility modulation at the same frequency induced by @xmath9 leads to the generation of a direct electric current @xmath2 . at the cyclotron resonance conditions , the amplitude of the oscillating electron motion in the qw plane increases and so does the efficiency of the dc current generation
microscopic model of the generation of a dc electric current @xmath2 by an ac electric field @xmath13 in an asymmetric qw subjected to a static magnetic field @xmath0 at the cyclotron resonance conditions .
the dc component of the electric current emerges due to the combined action of the in - plane component of the electric field , which induces the cyclotron motion of electrons in the qw plane ( blue ellipse ) , and the field out - of - plane component @xmath7 , which causes the electron mobility modulation at the same frequency .
insets show that the electric force @xmath10 shifts the electron density to the top ( or bottom ) interface thereby decreasing ( or increasing ) the electron scattering by impurities ( black dots ) . ]
a quasi - classical theory of the orbital ratchet effect described above can be developed in the framework of the boltzmann transport equation . in this approach , the ac electric field and the static magnetic field
are considered as the electric force and the lorentz force , respectively , acting upon the electrons .
the approach is relevant if the energy @xmath14 is much smaller than the mean kinetic energy of the electrons and the magnetic field is within the classical range .
the electron distribution in the momentum space is described by the time - dependent function @xmath15 which satisfies the boltzmann equation @xmath16 \right ) \dfrac{\partial f_{\bm p}}{\partial \bm{p } } = { \rm st } f_{\bm p } \:,\ ] ] where @xmath17 is the momentum , @xmath18 and @xmath19 are the electron velocity and energy , respectively , and @xmath20 is the collision integral . for elastic scattering ,
the collision integral has the form @xmath21 where @xmath22 is the rate of scattering .
the rate of electron scattering in an asymmetric qw to first order in @xmath7 can be presented in the form @xmath23 where @xmath24 is the scattering rate at zero electric field and @xmath25 is the field - induced correction . in the first born approximation
, the term @xmath24 is given by @xmath26 where @xmath27 is the matrix element of intrasubband scattering between the states with the momenta @xmath17 and @xmath28 and the angle brackets denote averaging over the positions of scattering centers . the second term in the right - hand side of eq
. originates from the admixture of excited - subband states to the ground - subband states by the out - of - plane electric field and is given by @xcite @xmath29 where @xmath30 is the index of the excited electron subbands , @xmath31 is the intersubband matrix element of the coordinate operator , @xmath32 is the energy separation between the subbands , and @xmath33 is the matrix element of intersubband scattering .
we note that the use of the collision integral with the time - dependent scattering rate is justified in the adiabatic regime , when @xmath14 is much smaller than the energy separation between the excited and ground electron subbands . to solve the boltzmann equation we decompose the distribution function @xmath15 in the fourier series of frequency and angular harmonics as follows @xmath34 where @xmath35 is the polar angle of @xmath17 .
accordingly , the collision integral for the scattering rate @xmath22 dependent of @xmath36 is rewritten in the form @xmath37 \nonumber \\ & \times \exp(i m \varphi_{\bm p } - i n \omega t ) \:.\end{aligned}\ ] ] here , @xmath38 is the relaxation time of the @xmath39-th angular harmonic of the electron distribution function , @xmath40 \:,\ ] ] and @xmath41 \:.\ ] ] in the harmonics representation , the boltzmann equation has the form of the set of linear equations @xmath42 where @xmath43 , @xmath44 is the cyclotron frequency , @xmath45 is the cyclotron mass , @xmath46 , @xmath47 and @xmath48 are the unit vectors along the @xmath49 and @xmath50 axes , respectively , and @xmath51 . in thermal equilibrium ,
when the ac electric field is absent , the distribution function contains only the harmonic @xmath52 which is given by the fermi - dirac distribution @xmath53 .
to first order in the electric field amplitude , solution of the equation set has the form @xmath54 } \frac{d f_0(\varepsilon_p)}{d p } \ : , \nonumber \\
f^{1,-1 } = - \frac{e \tau_1 \bm e_{\parallel } \cdot \bm o_+}{2 [ 1 - i(\omega - \omega_c)\tau_1 ] } \frac{d f_0(\varepsilon_p)}{d p } \:,\end{aligned}\ ] ] @xmath55 , and @xmath56 .
the dc electric current is determined by the time - independent asymmetric part of the distribution function described by the harmonics @xmath57 for which one obtains @xmath58 and @xmath59 .
the direct current density @xmath60 can be then readily found from the general expression @xmath61 \:,\ ] ] where the factor @xmath62 takes into account the spin degeneracy .
straightforward calculations show that the current density can be presented in the form @xmath63 where @xmath64 is the unit vector along the @xmath1 axis .
the coefficients @xmath65 , @xmath66 , @xmath67 , and @xmath68 for a degenerate electron gas are given by @xmath69 where @xmath70 is the electron density , @xmath71 is the fermi momentum , and @xmath72 , @xmath73 , and @xmath74 are taken at the fermi level . the coefficients @xmath65 and @xmath66 describe the electric current excited by a linearly polarized ac electric field ( linear ratchet current ) whereas @xmath67 and @xmath68 characterize the current contribution which is induced by an elliptically or circularly polarized field and has opposite directions for the right - handed and left - handed polarized radiation ( circular ratchet current ) .
in accordance with general symmetry arguments , @xmath67 and @xmath65 describing the current component along @xmath6 are even functions of the static magnetic field @xmath75 while @xmath68 and @xmath66 describing the perpendicular component of the current are odd functions of @xmath75 .
the dependences of the linear ratchet current and the circular ratchet current on the field frequency @xmath5 are shown in figs .
[ pge]a and [ pge]b , respectively . at zero frequency ,
only the linear ratchet current is generated .
its direction in the qw plane with respect to @xmath6 is determined by the ratio @xmath76 $ ] .
the deflection of the current direction from @xmath77 is caused by the lorentz force acting upon the electrons .
the circular ratchet effect emerges at finite frequencies of the ac electric field .
the magnitude of the circular ratchet current is proportional to @xmath5 at small @xmath5 , the current direction is determined by the ratio @xmath78 /[1 - 3(\omega_c \tau_1)^2]$ ] . at the cyclotron resonance conditions , both the linear and the circular ratchet currents drastically increase .
their directions in the qw plane are very sensitive to the frequency detuning @xmath79 , see figs .
[ pge]a and [ pge]b .
finally , far from the cyclotron resonance , the high - frequency asymptotic behavior of the ratchet currents is described by @xmath80 , @xmath81 , and @xmath82 .
the magnitude of the ratchet current can be estimated from eqs . and .
for the electric field amplitude @xmath83 kv / cm , the static magnetic field @xmath84 t , the momentum relaxation time @xmath85 s , the effective mass @xmath86 ( corresponding to gaas - based qws ) , the carrier density @xmath87 @xmath88 , the qw width @xmath89 nm , and the degree of qw structure inversion asymmetry @xmath90 , an estimation gives @xmath91 @xmath92a / cm for the ratchet current at the cyclotron resonance .
dependences of the projections of the linear ratchet current ( panel a ) and projections of the circular ratchet current ( panel b ) on the electric field frequency @xmath5 .
curves are calculated after eqs . and for @xmath93 .
now we consider the geometry of a tilted static magnetic field @xmath0 and ac electric field @xmath13 polarized in the qw plane . in this case ,
an asymmetry of the electron distribution in the momentum space and , hence , a dc electric current may emerge due to the asymmetry of electron scattering induced by the in - plane component @xmath94 of the magnetic field .
such a magnetic quantum ratchet effect in a purely in - plane magnetic field was theoretically studied in refs . and has been recently observed in graphene .
@xcite here , we develop a microscopic theory of the effect for a tilted magnetic field where the in - plane component @xmath94 induces the scattering asymmetry while the out - of - plane component @xmath75 causes the cyclotron motion of the electrons and leads to a resonant enhancement of the ratchet current . to first order in @xmath94 , the rate of elastic electron scattering can be generally presented in the form @xmath95 where @xmath24 is the scattering rate at zero field and @xmath25
is the field induced correction ; @xmath96 and @xmath97 due to time inversion symmetry .
microscopically , the scattering asymmetry stems from the quantum analogue of the lorentz force which pushes moving electrons to the top or bottom interfaces of the qw . for a qw structure with the simple parabolic energy spectrum , the term @xmath25 is given by @xcite @xmath98 \\ \sum\limits_{\nu \neq 1 } \dfrac{z_{\nu 1}}{\varepsilon _ { \nu 1 } } \langle \operatorname{re } v^*_{11}(\bm p , \bm p ' ) v_{1\nu } ( \bm p , \bm p ' ) \rangle \ , \delta ( \varepsilon_p -\varepsilon_{p ' } ) \,,\end{gathered}\ ] ] where @xmath99 is the effective mass . to calculate the dc current due to the magneto - induced ratchet effect we solve the boltzmann equation with the scattering rate . in the harmonics representation ,
the boltzmann equation assumes the form of the linear equation set @xmath100 where @xmath101 are defined by @xmath102 the harmonics @xmath101 satisfy the conditions @xmath103 and @xmath104 due to the reality of the scattering rate and time inversion symmetry , respectively .
it also follows from eq . that , for central scattering , the only nonzero harmonics are @xmath105 .
we seek a solution of the equation set in the form of a perturbation series in @xmath106 and @xmath94 . to first order in the electric field amplitude ,
one obtains the harmonics @xmath107 which are given by eqs . .
the second iteration in @xmath108 and @xmath109 yields @xmath110}{2 ( 1 - 2 i \omega_c \tau_2 ) } \nonumber \:,\end{aligned}\ ] ] and @xmath111 . the dc electric current
is determined by the harmonics @xmath112 ; they have the form @xmath113}{2 ( 1-i \omega_c \tau_1 ) } \\ & + \dfrac{\tau_1 e b_{\|}(u_{1,-2}-u_{0,-1})f^{0,2}}{1-i \omega_c \tau_1 } \ : \nonumber\end{aligned}\ ] ] and @xmath114 . finally , calculating the dc electric current @xmath2 following the general expression we obtain @xmath115 where @xmath116 , @xmath117 , and @xmath118 are the stokes parameters determining the polarization state of the ac electric field .
accordingly , the parameters @xmath119 , @xmath120 , and @xmath121 describe the magnitudes of the magneto - induced ratchet effect independent of the ac field polarization , the linear magneto - induced ratchet ( lmr ) effect , and the circular magneto - induced ratchet ( cmr ) effect , respectively . for a degenerate electron gas ,
the parameters are given by @xmath122 where @xmath123 is the coefficient describing the electron scattering anisotropy induced by the in - plane component of the magnetic field .
all the values in eq . are taken at the fermi level .
we note that an additional contribution to the polarization independent current may arise due to asymmetry of the energy relaxation of hot carriers in the magnetic field.@xcite this current depends on the details of election - phonon interaction and is out of the scope of the present paper .
figure [ mpge ] shows the frequency dependence of the lmr and cmr currents which are determined by @xmath124 and @xmath125 , respectively .
the curves are calculated for the case when the electron mobility is limited by coulomb impurities , @xmath126 , and the intersubband scattering is determined by short - range defects , i.e. , @xmath127 is independent of the electron energy .
the coefficient @xmath127 is assumed to be real which corresponds to the geometry @xmath128 .
one can see that both the lmr and cmr currents drastically increase at cyclotron resonance . however , the resonant contribution to the lmr current exceeds the resonant contribution to cmr current by the factor of @xmath129 .
moreover , the cmr effect emerges due to the energy dependence of the momentum relaxation time .
the cmr current has an additional resonance at @xmath130 which occurs in the spectral range where the drude absorbance has no peculiarity . in the quantum - mechanical picture
, this additional resonance corresponds to the transitions between the landau levels @xmath131 and @xmath132 .
dependences of the projections of the linear magneto - induced ratchet current ( panel a ) and projections of the circular magneto - induced ratchet current ( panel b ) on the electric field frequency @xmath5 .
curves are calculated after eqs . and for @xmath133 , @xmath126 , and real and energy independent @xmath127 .
in experiments , ratchet currents can be excited by the ac electric field of terahertz or microwave radiation focused on the sample @xcite .
the dc electric signal can be measured via the voltage drop across contacts in the open - circuit configuration where the net electric current in the circuit is vanishingly small .
the voltage drop is defined by the distribution of the electrostatic potential @xmath134 in the qw plane .
the latter is determined by the spatial distribution of the radiation - induced ratchet current @xmath135 , the drift current @xmath136 , which tends to compensate the ratchet current , and the boundary conditions at the sample edges .
the continuity equation requires @xmath137 the components of the drift current are given by @xmath138 where @xmath139 is the conductivity tensor in the magnetic field , @xmath140 @xmath141 is the conductivity at zero magnetic field . from eqs .
- we obtain the poisson equation @xmath142 where @xmath143 .
equation should be solved with the boundary conditions . for finite - size samples ,
we consider the boundary condition of zero total electric current flowing across the sample edges , @xmath144 .
solution of the eq . can be generally presented in the form @xmath145 where @xmath146 can be interpreted as the function of nonlocal resistance .
the explicit form of the function @xmath146 depends on the sample shape and , in general case , can be calculated numerically .
below we discuss the spatial distribution of the electrostatic potential in infinite , semi - infinite , and rectangular - shape structures and present some analytical results . for an infinite two - dimensional system , the solution of the two - dimensional poisson equation can be readily found by green s function method , which yields @xmath147 the magnetic field @xmath75 only scales the function @xmath146 .
however the field also affect the magnitude and direction of @xmath148 and , hence , the spatial distribution of the electrostatic potential . using the method of mirror images and the function @xmath146 for the infinite system , one can derive an analytical equation for @xmath146 in a semi - infinite system .
for the system @xmath149 with a single boundary at @xmath150 , we obtain @xmath151 } \biggr ] , \nonumber \\
\rho_y ( \bm{r},\bm{r } ' ) & = -\dfrac{1}{2\pi \sigma } \biggl[\dfrac{y - y'}{(x - x')^2+(y - y')^2 } \nonumber \\ & + \dfrac{(1-\omega_c^2\tau_1 ^ 2)(y - y ' ) - 2\omega_c\tau_1(x+x')}{(1+\omega_c^2\tau_1 ^ 2)[(x+x')^2+(y - y')^2 ] } \biggr ] .
\end{aligned}\ ] ] by adding three additional boundaries and considering a net consisting of rectangular cells , one can also obtain the function of nonlocal resistance for a finite - size rectangular structure with the edges at @xmath152 and @xmath153 .
the function is given by @xmath154 + \text{c.c . }
\:,\end{gathered}\ ] ] where @xmath155 , @xmath156 ; the function @xmath157 is obtained from @xmath158 by replacing @xmath49 , @xmath50 , @xmath159 @xmath160 , @xmath161 , and @xmath162 with @xmath50 , @xmath163 , @xmath160 , @xmath164 , @xmath162 , and @xmath161 , respectively .
spatial distributions of the electrostatic potential induced by a ratchet current generated in the center area of a square - shape structure in a magnetic field .
distributions are calculated for @xmath165 and the gaussian distribution of the ratchet current density given by eq . for @xmath166 ( figure a ) and @xmath167 ( figure b ) .
dashes circles sketch the areas of the ratchet current generation . ]
figure [ potential_distribution ] demonstrates the spatial distribution of the electrostatic potential @xmath134 in a square - shape structure when the ratchet current is generated in the central area of the structure .
the spatial distribution of the ratchet current density is taken in the form of the gaussian function @xmath168 where @xmath169 and @xmath161 is the structure size .
the potential @xmath170 reaches extremal values at the positions close to the border of the current generation spot while at the sample edges it decreases .
the magnetic field changes the potential amplitude and also twists the potential spatial distribution that can be seen from equipotential lines .
the calculation shows that the voltage drop between the points of the highest and lowest potential is about @xmath171 mv for the spot radius @xmath172 mm , the ratchet current magnitude @xmath173 @xmath92a / cm and the parameters of the structure presented in sec .
[ sec_perpb ] .
the calculations also reveal that experimentally measured voltage drops across contacts may drastically depend on the contact positions in the structure and the structure geometry .
we have developed a microscopic theory of the ratchet transport of electrons confined in a quantum well .
it has been shown that the magnitude of the direct electric current induced by an alternating electric field is increased in an external static magnetic field at the cyclotron resonance when the electric field frequency matches the cyclotron frequency .
the magnetic field gives rise also to an additional mechanism of the ratchet current generation which stems from an asymmetry of electron scattering in the magnetic field .
the magneto - induced ratchet effect has a resonant behavior both at the cyclotron resonance and its first subharmonic , with the current being sensitive to the electric field polarization and the mechanism of electron scattering .
we have also analyzed the spatial distribution of the electrostatic potential induced by the ratchet current and shown that the voltage drop between contacts is highly sensitive to the sample geometry and the contact positions .
the resonant behavior of the ratchet current can be used for the study of electron scattering mechanisms and the development of tunable fast detectors of microwave and terahertz radiation .
w. knap , m. dyakonov , d. coquillat , f. teppe , n. dyakonova , j. lusakowski , k. karpierz , m. sakowicz , g. valusis , d. seliuta , i. kasalynas , a. el fatimy , y. m. meziani , and t. otsuji , j. infrared millim , field effect transistors for terahertz detection : physics and first imaging applications , terahertz waves * 30 * , 1319 ( 2009 ) .
p. olbrich , e. l. ivchenko , r. ravash , t. feil , s. d. danilov , j. allerdings , d. weiss , d. schuh , w. wegscheider , and s. d. ganichev , ratchet effects induced by terahertz radiation in heterostructures with a lateral periodic potential , , 090603 ( 2009 ) .
s. sassine , yu .
krupko , j .- c .
portal , z. d. kvon , r. murali , k. p. martin , g. hill , and a. d. wieck , experimental investigation of the ratchet effect in a two - dimensional electron system with broken spatial inversion symmetry , , 045431 ( 2008 ) .
i. bisotto , e. s. kannan , s. sassine , r. murali , t. j. beck , l. jalabert , and j .- c .
portal , microwave based nanogenerator using the ratchet effect in si / sige heterostructures , nanotechnology * 22 * , 245401 ( 2011 ) . c. drexler , s.a .
tarasenko , p. olbrich , j. karch , m. hirmer , f. mller , m. gmitra , j. fabian , r. yakimova , s. lara - avila , s. kubatkin , m. wang , r. vajtai , p.m. ajayan , j. kono , and s.d .
ganichev , magnetic quantum ratchet effect in graphene , nature nanotechnol .
* 8 * , 104 ( 2013 ) .
v. v. belkov , p. olbrich , s. a. tarasenko , d. schuh , w. wegscheider , t. korn , ch .
schller , d. weiss , w. prettl , and s.d .
ganichev , symmetry and spin dephasing in ( 110)-grown quantum wells , phys .
100 * , 176806 ( 2008 ) .
p. olbrich , c. zoth , p. vierling , k .-
dantscher , g. v. budkin , s. a. tarasenko , v. v. belkov , d. a. kozlov , z. d. kvon , n. n. mikhailov , s. a. dvoretsky , and s. d. ganichev , giant photocurrents in a dirac fermion system at cyclotron resonance , , 235439 ( 2013 ) .
dantscher , d. a. kozlov , p. olbrich , c. zoth , p. faltermeier , m. lindner , g. v. budkin , s. a. tarasenko , v. v. belkov , z. d. kvon , n. n. mikhailov , s. a. dvoretsky , d. weiss , b. jenichen , s. d. ganichev , cyclotron resonance assisted photocurrents in surface states of a 3d topological insulator based on a strained high mobility hgte film , arxiv:1503.06951 .
s. d. ganichev , j. kiermaier , w. weber , s. n. danilov , d. schuh , ch .
gerl , w. wegscheider , w. prettl , d. bougeard , and g. abstreiter , subnanosecond ellipticity detector for laser radiation , appl .
phys . lett . *
91 * , 091101 ( 2007 ) .
a. p. dmitriev , s. a. emelyanov , s. v. ivanov , p. s. kopev , ya .
v. terentev , and i. d. yaroshetsky , drag photocurrent in a 2d electron gas near the cyclotron resonance and its first subharmonic , pisma zh .
* 54 * , 460 ( 1991 ) [ jetp lett .
* 54 * , 462 ( 1991 ) ] .
s. stachel , g. v. budkin , u. hagner , v. v. belkov , m. m. glazov , s. a. tarasenko , s. k. clowes , t. ashley , a. m. gilbertson , s. d. ganichev , cyclotron - resonance - assisted photon drag effect in insb / inalsb quantum wells excited by terahertz radiation , , 115435 ( 2014 ) . | the driving of charge carriers confined in a quantum well lacking the center of space inversion by an alternating electric field leads to the formation of a direct electric current .
we develop a microscopic theory of such a quantum ratchet effect for quantum wells subjected to a static magnetic field .
we show that the ratchet current emerges for a linearly polarized alternating electric field as well as a rotating electric field and drastically increases at the cyclotron resonance conditions . for the magnetic field
tilted with respect to the quantum well normal , the ratchet current contains an additional resonance at the first subharmonic of the cyclotron resonance . |
we investigate the stationary flow of a heat conducting compressible fluid in a cylindrical domain . the fluid is assumed to be newtonian
. then the flow is described by the stationary navier - stokes - fourier ( nsf ) system : @xmath0 where @xmath1 $ ] with @xmath2 smooth , @xmath3 is the velocity field of the fluid , @xmath4 is the density , @xmath5 is the absolute temperature .
we assume that the pressure @xmath6 is a twice continuously differentiable function on @xmath7 such that @xmath8 we consider the newtonian compressible fluid , i.e. the stress tensor has the form @xmath9 where the constant viscosities @xmath10 and @xmath11 fulfill @xmath12 and @xmath13 .
further , the friction coefficient @xmath14 . finally , @xmath15 is the heat flux .
note that we could treat the situation @xmath16 , @xmath17 and @xmath18 with suitable assumptions on these functions .
it would only lead to further complications , therefore we take them rather constant not to hide the main ideas by too many technicalities . the boundary of the domain is divided in a natural way into the inflow part @xmath19 , the outflow part @xmath20 and the impermeable wall @xmath21 .
more precisely , we define @xmath22 we consider a channel - like flow , i.e. we further assume that @xmath23 , @xmath24 and @xmath25 $ ] .
we set @xmath26 on @xmath27 and @xmath28 to be a given positive constant on the wall @xmath21 .
we set @xmath29 on @xmath21 but we admit @xmath30 on @xmath27 ( see fig.[rys1 ] ) . under such choice , on the inflow and outflow part
the boundary condition for the temperature reduces to the inhomogeneous neumann condition . from the point of view of modeling these parts of the boundary
are artificial and we can measure the parameters of the flow , what can be reflected in different values of @xmath31 .
the condition on @xmath21 means that the flux of the temperature through the wall is proportional to the difference between the outside temperature @xmath32 and the temperature of the fluid .
the level of thermal insulation of the wall is given by @xmath28 .
existence and uniqueness of the special solution @xmath33 under our choice of boundary conditions is discussed in the next chapter .
the total energy is given by @xmath34 , where the internal energy @xmath35 is given by @xmath36 with @xmath37 i.e. the internal energy fulfills the maxwell relation . without loss of generality
we set in @xmath38 .
equivalently we can consider system ( [ main_system ] ) with total energy balance ( [ main_system])@xmath39 replaced with the internal energy balance @xmath40 we also set @xmath41 * known results * the existence theory for the steady compressible navier
stokes equations in the framework of strong solutions was intensively studied in eighties , see e.g. @xcite , @xcite and many other papers . in the beginning of this century , inspired by the results of p.l .
lions ( see @xcite ) , the attention turned rather towards the weak solutions , i.e. towards the existence theory without any smallness assumption on the data . the best results in this direction so far can be found in @xcite ( slip boundary conditions ) and @xcite ( space periodic situation ) .
note , however , that the theory is not able to treat the situation with nonzero inflow / outflow conditions .
strong solutions with inhomogeneous boundary data , which are in our scope of interest in this paper , have been considered for the first time in @xcite in hilbert spaces and later in @xmath42 framework ( @xcite , @xcite , @xcite ) . in
@xcite a constant flow in the direction of the axis of the cylinder is investigated and existence of a strong solutions in its vicinity is shown in the barotropic case . concerning the nsf system ,
the existence of weak solutions was studied much later , see @xcite , @xcite . the best result in this direction
can be found in @xcite .
note , however , that similarly to the barotropic case , all these results do not allow to treat the inflow / outflow problems .
the theory of strong solutions for the nsf system describing the thermal effects is much less developed , to our knowledge the only result in @xmath42 framework is the paper by beirao da veiga @xcite where existence of strong solutions in a vicinity of a constant flow with zero velocity is shown .
* special solution . * here we are interested in showing existence of strong solutions , i.e. we are looking for @xmath43 , solving system ( [ main_system ] ) so that the solution is close to a given special solution
. hence we arrive at a problem with some smallness of the data .
even though smallness obviously facilitates the analysis , the results of this kind are still missing in the theory of compressible flow , especially for heat - conducting fluids ( see the overview above ) .
this is mostly due to high complexity of the system ( [ main_system])@xmath44 , in particular due to its mixed , elliptic ( or parabolic in time dependent case ) and hyperbolic character .
investigation of stationary solutions close to given laminar flows is of particular importance if we want to investigate stability of special solutions , which is quite well developed for weak solutions but so far almost not investigated in strong solutions framework ( see @xcite for an overview of results ) .
the simplest example of a special solution to perturb is a solution with zero velocity and constant density and temperature .
such flow has been investigated in @xcite with homogeneous dirichlet boundary conditions . from the point of view of applications
it is important to investigate flows with large velocity , which leads in a natural way to inhomogeneous boundary data . in a cylindrical domain
a natural example of such solution is a constant flow in the direction of the axis of the cylinder : @xmath45 , \quad \bar \rho = 1.\ ] ] in other words , we can state that such a flow is natural to be investigated as a special solution to the system ( [ main_system ] ) in our domain @xmath46 as for the temperature of the basic solution the simplest choice would be to assume it constant .
it seems however quite artificial with the flow @xmath47 defined above .
it is more interesting from physical point of view to consider a flow with nontrivial temperature function @xmath33 which corresponds to the solution @xmath47 as a solution to corresponding energy balance equation .
on the boundary we assume the temperature of the form @xmath48 with constant , possibly large @xmath49 and small variation @xmath50 . substituting @xmath47 to the internal energy balance equation
we see that @xmath33 should satisfy the system @xmath51 now we write @xmath52 , where @xmath53 solves @xmath54 \kappa \frac{\partial \overline { \theta}_0}{\partial n}|_{\gamma } = g & \mbox { on } \gamma,\\[2pt ] \frac{1}{|\omega|}\int_{\omega } \overline { \theta}_0 \,dx = t_0 , \end{array}\ ] ] and @xmath55 the existence of a solution @xmath56 to ( [ eqn_tilde_theta ] ) is a classical elliptic result , we only need to apply symmetry to deal with corner singularities .
the details are given in lemma [ lem_neumann ] .
in particular @xmath53 satisfies @xmath57 \leq c \|g\|_{w^{1 - 1/p}_p(\gamma)}.\ ] ] now we see that @xmath58 satisfies @xmath59 the existence and uniqueness of solution to the above system is discussed in subsection [ sec_theta1 ] .
our goal is to show the existence of a solution to ( [ main_system ] ) close to @xmath60 .
hence it is convenient to introduce the following quantity to measure the distance of the data from the special solution : @xmath61 we are now in a position to formulate our main result .
[ main_thm ] assume that @xmath62 defined in ( [ d0 ] ) is small enough , @xmath63 is large enough , @xmath28 is large enough on @xmath21 , @xmath64 large enough on @xmath19 and @xmath65 .
then there exists a solution @xmath66 to system ( [ main_system ] ) such that @xmath67 this solution is unique in the class of solutions satisfying ( [ est_main ] )
. * notation .
* we use standard notation for the sobolev spaces , i.e. @xmath68 . for the spaces defined on @xmath46 we will skip the domain ; for example we write @xmath69 instead of @xmath70 . a generic constant that is controlled ( but not necessarily small ) will be denoted by @xmath71 , whereas @xmath72 will denote a small constant ( i.e. which can be arbitrarily small for the data small enough ) .
we will also use the space @xmath73 , for simplicity we denote it by @xmath74 .
in this section we collect some standard results which we use throughout the paper .
we start with the sobolev imbedding theorem , let us state here the particular case which is used throughout the paper ( the general version with the proof can be found in ( * ? ? ?
* theorem 5.4 ) ) let @xmath46 be defined as above and let @xmath75 with @xmath65
. then @xmath76 and there exist @xmath77 s.t .
@xmath78 next we recall the poincar inequality . among its different versions
we will need the following two .
let @xmath79 and @xmath80 denote two nontrivial parts of the boundary @xmath81 and @xmath82
. then @xmath83\ ] ] and @xmath84,\ ] ] where @xmath85 .
( interpolation inequality ) : [ lem_int ] + let @xmath86 .
then for all @xmath87 such that @xmath88 : @xmath89 the interpolation inequality in the lebesgue spaces and the imbedding @xmath90 with @xmath91 for @xmath65 , @xmath92 , arbitrary for @xmath93 and @xmath94 for @xmath95 yield @xmath96 now application of the young inequality leads to ( [ int1 ] ) .
another result we use is the following version of the korn inequality [ lem_korn ] let @xmath97 , @xmath12 and if @xmath98 then @xmath46 is not axially symmetric .
then @xmath99 it is a straightforward generalization of ( * ? ? ?
* lemma 4 ) .
we also apply the following well - known result in finite dimensional hilbert spaces ( the proof can be found in @xcite ) : [ lem_p ] let @xmath100 be a finite dimensional hilbert space and let @xmath101 : @xmath102 be a continuous operator satisfying @xmath103 then @xmath104 our proofs will be based on the @xmath42 regularity of elliptic problems with neumann and slip boundary conditions . these are classical results , in case of cylindrical domain we only have to deal with the singularities of the boundary , for the sake of completeness we give the proofs . the first result concerns
the neumann problem .
[ lem_neumann ] let @xmath105 , @xmath106 , @xmath107 and @xmath108 .
let @xmath109 .
then there exists a solution @xmath110 to the problem @xmath111 \frac{\partial u}{\partial n } = d & \mbox { on } & \gamma , \\[2pt ] \frac{1}{|\omega|}\int_\omega u \ , dx & = c_0 , \end{array}\ ] ] and @xmath112.\ ] ] moreover , this solution is unique in the given regularity class .
in fact , the only difficulty we have to overcome are the singularities of the boundary . if @xmath113 , then we can simply extend the weak solution using symmetric reflection that preserves the homogeneous neumann condition . in the neighbourhood of @xmath19 it will be defined as @xmath114 where @xmath115 and @xmath116 . for @xmath117
we obtain a neumann problem in a domain with smooth boundary that we solve using classical elliptic theory .
if @xmath118 does not vanish on @xmath27 , we reduce the problem to the previous case .
let us consider @xmath19 , on @xmath20 we proceed the same way .
introducing an extension @xmath119 such that @xmath120 and @xmath121 , we consider @xmath122 that satisfies @xmath123 where @xmath124 $ ] . the second result concerns the lam system with slip boundary conditions : @xmath125 the following lemma gives existence of a solution and the maximal regularity elliptic estimate that we apply in our proofs .
[ lem_lame ] under the assumptions of theorem [ main_thm ] there exists a unique @xmath110 solving ( [ lam ] ) with @xmath126.\ ] ] under the assumptions on @xmath10 and @xmath11 ( [ lam ] ) is elliptic so we easily get a weak solution . the only problem we encounter showing regularity of the weak solution are the singularities of the boundary on the junctions of @xmath21 with @xmath19 and @xmath20 .
these singularities can again be dealt using the symmetry arguments .
hence we define the operator @xmath127 extending a vector field defined for @xmath128 on the whole space as @xmath129 where @xmath130 is as above and @xmath131 $ ]
. then we have @xmath132 and on the plane @xmath133 the extension @xmath127 preserves the slip boundary conditions , i.e. @xmath134 and @xmath135 .
now we can show higher regularity of @xmath136 , extension of the weak solution to ( [ lam ] ) , and since @xmath127 preserves the boundary conditions we conclude that @xmath137 is a strong solution to ( [ lam ] ) satisfying the boundary conditions on @xmath19 .
application of analogous antisymmetric extension on @xmath20 completes the proof .
our next aim is to construct solutions to .
recall that @xmath138 where @xmath139 solves .
therefore by lemma [ lem_neumann ] we have @xmath140 as well as by the lax - milgram theorem @xmath141 next for @xmath142 we have problem ( [ eqn_theta1 ] ) .
our aim is to show existence of a weak solution to this problem as well as its regularity .
as the problem is linear , the existence and uniqueness of a weak solution is direct consequence of corresponding a priori estimates .
first we show estimates in @xmath143 , then in @xmath144 .
since we will require @xmath63 and @xmath28 large , we have to control how all constants depend on them . for the sake of simplicity
we assume that @xmath63 and @xmath28 are approximately of the same size . multiplying ( [ eqn_theta1 ] ) by @xmath145 and integrating over @xmath46 we get @xmath146 all terms on the l.h.s . are nonnegative . applying the poincar inequality to the first term on the r.h.s
the young inequality to the other two ones we get @xmath147 therefore we get , for @xmath63 and @xmath28 sufficiently large with respect to @xmath148 , @xmath149 hence , using also @xmath150 where @xmath71 is a fixed constant independent of @xmath63 and @xmath28 , provided @xmath151 .
uniqueness is a direct consequence of ; if the data are zero , then the solution is zero as well .
finally , proceeding similarly as above we will deduce the existence of strong solutions together with the estimates of the corresponding norms .
first we rewrite ( [ eqn_theta1 ] ) to the form @xmath152
\kappa \frac{\partial \overline{\theta}_1}{\partial
n } = -l(\overline{\theta}_1 + \overline{\theta}_0 - t_0 -t_1 ) & \mbox { on } \gamma , \end{array}\ ] ] and using lemma [ lem_neumann ] @xmath153 now , applying the poincar inequality in the form @xmath154 and using we end up with [ l7 ] let @xmath155 , @xmath156 , @xmath65 and let @xmath63 , @xmath28 be sufficiently large . then there exist a unique weak solution @xmath145 to ( [ eqn_theta1 ] ) such that @xmath157.\ ] ] moreover , the solution is strong , i.e. @xmath158 and @xmath159.\ ] ] here , the constants @xmath71 are independent of @xmath63 and @xmath28 .
as we are interested in small perturbations of @xmath60 , it is natural to introduce the perturbations as unknown functions . for technical reasons it is better to get rid of inhomogeneity on the boundary .
hence we construct @xmath160 such that @xmath161 .
more precisely , we set @xmath162 , where @xmath163 solves @xmath164 next we introduce the perturbations @xmath165 replacing the total energy balance ( [ main_system])@xmath39 with the internal energy balance ( [ int_en_bal ] ) we can rewrite ( [ main_system ] ) as @xmath166 where @xmath167 g(u,\sigma ) = -(\sigma + 1 ) \ , { \rm div } \ , u_0 - \sigma \ , { \rm div } \,u , \\ [ 5pt ]
h(u,\sigma,\eta ) = { \bf s}(\nabla(u+u_0 ) ) \cdot \nabla ( u+u_0 ) - t_0 \partial_\rho \pi(1+\sigma , \eta + \overline{\theta}_0 + \overline{\theta}_1 ) { \rm div}\,u_0 \\
-(1+\sigma ) ( u+u_0 ) \cdot \nabla ( \overline{\theta}_0 + \overline{\theta}_1 + \eta ) + t_0 { \rm div } \,u \big(p_2- \partial_\theta \pi ( 1+\sigma , \overline{\theta}_0 + \overline{\theta_1 } + \eta)\big ) \\ -(\eta + ( \overline{\theta}_0 -t_0 ) + \overline{\theta}_1 ) \partial_\theta \pi(1+\sigma , \overline{\theta}_0 + \overline{\theta_1 } + \eta ) { \rm div}\ , ( u+u_0 ) \\ + \partial_{x_1 } ( \overline{\theta}_0 + \overline{\theta}_1 + \eta ) \big(e_2- \partial_\theta e ( 1+\sigma , \overline{\theta}_0 + \overline{\theta_1 } + \eta)\big ) , \\[5pt ] b_i = b_i - \alpha \tau_i^{(1 ) } - ( { \bf s}(\nabla u_0 ) n ) \cdot \tau_i -\alpha u_0 \cdot \tau_i . \end{array}\ ] ] the above expressions result directly from substitution of ( [ pert ] ) to ( [ main_system ] ) .
it is only worth to notice that to derive the expression for @xmath168 we have applied the continuity equation which yields @xmath169 now in ( [ main_system])@xmath39 the first term on the r.h.s . of ( [ div_id ] ) can be combined with the @xmath170 , what finally gives the above form of @xmath171 .
let us state at this early stage the following estimates on the quantities @xmath172 defined in ( [ fgh ] ) : @xmath173 , \\ \\
\smallskip \|g(u,\sigma)\|_{w^1_p } & \leq c \big[(\|u\|_{w^2_p } + \|\sigma\|_{w^1_p})^2 + d_0(1+d_0)\big],\\ \\
\smallskip \|h(u,\sigma,\eta)\|_{l_p } & \leq c \big [ ( \|u\|_{w^2_p } + \|\sigma\|_{w^1_p } + \|\eta\|_{w^2_p})^2 \\ & + ( \|u\|_{w^2_p } + \|\sigma\|_{w^1_p}+ \|\eta\|_{w^2_p})^3 + d_0(1+d_0)\big ] , \\
\end{array}\ ] ] where @xmath62 is defined in ( [ d0 ] ) .
the bounds ( [ est_fgh ] ) are straightforward , it is enough to observe that @xmath174 and @xmath168 are composed of terms which are either quadratic or cubic with respect to the functions @xmath175 or linear with respect to them multiplied by something which can be bounded by @xmath62 .
note that in the estimate of terms contained differences of the derivatives of @xmath176 or @xmath35 we use the fact fact that these functions are twice continuously differentiable and hence we may use the taylor expansion theorem .
we sketch briefly the idea of the proof of theorem 1 , and hence the structure of the remainder of the paper . in subsection 2.4
we explain how to deal with the linearization of the continuity equation , i.e. equation @xmath177 . to show existence of solutions we can apply a method of successive approximations and look for the solution as a limit of a sequence @xmath178 in section 3 we deal with the linear system ( [ system_lin ] ) showing the a priori estimates .
the first step is the energy estimate which is derived in quite a standard way .
the energy estimate is then applied to show the maximal regularity @xmath179 estimates , hence estimates in the space where we look for the solution ( lemma [ lem_est_lin ] ) .
it can be considered as the main result of section 3 . in the end of section 3
we solve the linear system , hence we show that the sequence ( [ system_seq ] ) is well defined . in section 4
we deal with the sequence of approximations .
using the estimates for the linear system combined with ( [ fgh ] ) we show it is bounded in @xmath180 and has a contraction property in a weaker space @xmath181 .
an analogous estimate gives uniqueness in the class of solutions satisfying ( [ est_main ] ) .
in other words we apply a generalization of the banach fixed point theorem to show the convergence of the sequence of approximations to the solution of ( [ main_system ] ) . in this section
we show the well posedness of the operator @xmath182 defined as @xmath183 where @xmath184 is a weak solution to in the sense as introduced in .
the solvability of the above problem , to which we refer as a steady transport equation , will be needed to show the energy estimate for the density in section 3 and to solve the continuity equation in section 4 .
the result is given in the following [ lem_s ] let @xmath185 and @xmath186 .
let @xmath187 be such that @xmath188 is small enough and @xmath189 .
then there exists a unique solution @xmath190 and @xmath191.\ ] ] the proof follows @xcite . for the sake of completeness we recall here the most important steps , referring to @xcite for the missing details .
the idea is to get rid of the term @xmath192
introducing a change of variables @xmath193 satisfying the identity @xmath194 we construct the mapping @xmath195 in the following [ lem_change ] let @xmath188 be small enough . then there exists a set @xmath196 and a diffeomorphism @xmath197 defined on @xmath198 such that @xmath199 and ( [ change_id ] ) holds .
moreover , if @xmath200 and @xmath201 then @xmath202 , where @xmath203 is the outward normal to @xmath198 .
before we sketch the proof let us make one remark .
the last condition states that the first component of the normal to @xmath204 vanishes , but since @xmath195 is defined only on @xmath198 we formulate this condition using the limits .
it means simply that the image @xmath205 is also a cylinder with a flat wall .
_ sketch of the proof of lemma [ lem_change ] . _
the identity ( [ change_id ] ) means that @xmath195 must satisfy @xmath206 a natural condition is that @xmath207 . thus we can search for @xmath208 , where for all @xmath209 such that @xmath210 the function @xmath211 is a solution to a system of ode : @xmath212 using the regularity and smallness of @xmath213 we show existence for ( [ ode ] ) , hence we have @xmath195 defined on some @xmath198 such that @xmath199 .
the smallness of @xmath213 implies also that @xmath214 , where @xmath215 is small and we conclude that @xmath195 is a diffeomorphism .
let us denote @xmath216 .
now it is natural to define the subsets of @xmath217 as @xmath218 where @xmath219 , @xmath220 and @xmath221 . to show that @xmath222 for @xmath223 it is enough to observe that @xmath224 ) = [ 1+u^1(x),u^2(x),u^3(x)],\ ] ] where @xmath197 .
but for @xmath225 the vector on the r.h.s is tangent to @xmath21 since @xmath226 .
we can conclude that on @xmath227 the image in @xmath195 of a straight line @xmath228 is a curve tangent to @xmath21 , and thus @xmath227 is a sum of such lines and so we have @xmath222 .
the proof of lemma [ lem_change ] is completed .
@xmath229 _ proof of lemma [ lem_s ] .
_ first we define @xmath230 for a continuous function @xmath231 as @xmath232 the condition @xmath233 on @xmath234 guarantees that a straight line @xmath235 has a picture in @xmath46 and thus we integrate along a curve contained in @xmath46 .
it means that @xmath236 is well defined for continuous functions defined on @xmath46 and the construction of @xmath195 clearly ensures that @xmath236 satisfies ( [ def_s ] ) .
next we have to extend @xmath236 on @xmath70 . to this end
we show the estimate ( [ est_s ] ) for continuous @xmath231 .
let @xmath237 denote an @xmath238-cut of @xmath46 and let @xmath239 .
then by ( [ s ] ) we have @xmath240 ^ 2 \ , d \bar x \\ \leq 2 \|w_{in}\|_{l_2(\gamma_{in})}^2 + c \int_{\omega_{x_1 } } \int_0^{\phi_1(x ) } h^2(\psi(s,\phi_2(x),\phi_3(x ) ) ) \,ds \,d\bar x \leq c \ , \big [ \|w_{in}\|_{l_2(\gamma_{in})}^2 + \|h\|_{l_2(\omega)}^2 \big ] .
\end{array}\ ] ] the above estimate holds for every @xmath241 what implies ( [ est_s ] ) .
now we can define @xmath242 for @xmath243 using a standard density argument , the details are given in ( @xcite ) .
to show the uniqueness of the solution we can rewrite the r.h.s of ( [ def_s ] ) as @xmath244 where the equation is to be understood in the sense of distribution , and , treating @xmath238 as a `` time '' variable , adapt di perna - lions theory of transport equation ( @xcite ) that implies the uniqueness of solution to ( [ def_s_1a ] ) in the class @xmath245 .
the proof is thus complete .
in this section we deal with the linear system @xmath246 with given functions @xmath247 such that @xmath248 note that this system actually represents , where we only redefined a few constants .
the main difficulty lies in deriving appropriate estimates for ( [ system_lin ] ) .
we start with the energy estimates , then applying the properties of slip boundary conditions , helmholtz decomposition of the velocity and classical elliptic theory we derive the @xmath42 estimates .
apart from solving the system ( [ system_lin ] ) , the estimates derived in this section are used later , combined with ( [ est_fgh ] ) , to show the convergence of the sequence of approximations ( [ system_seq ] ) .
[ l 8 ] let @xmath249 solve system ( [ system_lin ] ) with @xmath250 with @xmath188 small enough , @xmath251 on @xmath252 fulfilling conditions above , where @xmath253 is the dual space to @xmath254 assume further that @xmath255 and @xmath64 fulfill the assumptions of theorem [ main_thm ] .
then the following estimate holds : @xmath256 .
\end{array}\ ] ] we multiply ( [ system_lin])@xmath257 by @xmath137 and integrate applying the following identity : @xmath258 with application of the korn inequality and the boundary condition ( [ system_lin])@xmath259 we get @xmath260 next we divide ( [ system_lin])@xmath39 by @xmath261 ( recall it is constant ) , multiply by @xmath262 and integrate .
it yields @xmath263 hence @xmath264 to get rid of the boundary term on the r.h.s we can apply the poincar inequality ( [ poin1 ] ) and rewrite ( [ ene2 ] ) as @xmath265 where @xmath266 is the constant from ( [ poin1 ] ) .
for @xmath63 large enough and @xmath28 large enough on @xmath21 the first two terms on the l.h.s .
will be positive .
now we can combine ( [ ene1 ] ) and ( [ ene2 ] ) obtaining @xmath267 \leq \int_{\omega } h\eta \,dx + \int_{\omega } f \cdot u \,dx \\
+ \int_{\omega } g\sigma \,dx + \frac{1}{2 } \int_{\omega } \sigma^2 { \rm div } \
, u \,dx + \frac{1}{2 } \int_{\gamma_{in } } \sigma_{in}^2 ( 1-u\cdot n ) \,ds
+ \sum_{i=1}^2 \int_\gamma b_i u \cdot \tau_i \ , ds .
\end{array}\ ] ] using ( [ poin2 ] ) , the imbedding @xmath268 and hlder s inequality we derive @xmath269 recall that e denotes a small constant and so to complete the proof we have to find the bound on @xmath270 .
but this follows from the existence result in the previous subsection , see .
hence we have @xmath271 combining ( [ ene3 ] ) and ( [ ene4 ] ) we conclude ( [ ene ] ) .
the main result of this section is [ lem_est_lin ] let @xmath272 solve system ( [ system_lin ] ) with @xmath273 , where @xmath188 is small enough and @xmath274 satisfy the assumptions of theorem [ main_thm ] .
then @xmath275 : = d_{lin}. \end{array}\ ] ] the proof of ( [ est_lin ] ) will be performed in several consecutive lemmas .
we show the bound on @xmath276
. then we proceed with the bound on @xmath277 which is the most demanding .
first we show bound on the vorticity which , together with helmholtz decomposition of the velocity , makes possible to eliminate the term @xmath278 from the continuity equation , leading to ( [ trans ] ) . using this equation we show the bound on the density .
then ( [ est_lin ] ) easily follows from the classical elliptic estimate on @xmath279 .
we start with the bound on @xmath262 ; to show it let us rewrite ( [ system_lin])@xmath280 as @xmath281 the classical elliptic estimate ( @xcite , @xcite ) for the above system yields @xmath282.\ ] ] applying the trace theorem to the boundary term and then the interpolation inequality ( [ int1 ] ) we get @xmath283 + \epsilon \|u\|_{w^2_p}.\ ] ] by the energy estimate ( [ ene ] ) we conclude @xmath284 now we proceed towards the bound on @xmath277 . as mentioned before , we start with the estimate on the vorticity . taking the curl of ( [ system_lin])@xmath257 we get @xmath285 & \mbox{in } & \omega , \\ \alpha \cdot \tau_2 = ( 2 \chi_1 - \frac{\alpha}{\mu } ) u \cdot \tau_1 + \frac{b_1}{\mu } & \mbox{on } & \gamma , \\ \alpha \cdot \tau_1 = ( \frac{\alpha}{\mu } - 2 \chi_2 ) u \cdot \tau_2 - \frac{b_2}{\mu } & \mbox{on } & \gamma , \\ { \rm div}\,\alpha = 0 & \mbox{on } & \gamma , \end{array}\ ] ] where @xmath286 denote the curvatures of the curves generated by tangent vectors @xmath287 . in order to show the boundary relations ( [ system_rot])@xmath288 it is enough to differentiate ( [ system_lin])@xmath259 with respect to the tangential directions and apply ( [ system_lin])@xmath39 . a rigorous proof is given in @xcite or @xcite .
the condition @xmath289 in @xmath46 results simply from the fact that @xmath290 .
we introduce this relation as a boundary condition ( [ system_rot])@xmath259 , that completes the conditions on the tangential parts of the vorticity .
note that the boundary conditions ( [ system_rot])@xmath288 give the tangential parts of the vorticity on the boundary the regularity of the velocity itself and the data . we will now use this feature of slip boundary conditions to show the higher estimate on the vorticity ( see @xcite,@xcite , @xcite ) .
for the above system we have ( see @xcite , theorem 10.4 ) : @xmath291.\ ] ] applying the interpolation inequality ( [ int1 ] ) and then ( [ ene ] ) we arrive at @xmath292 for any @xmath293 , where @xmath294 is defined in ( [ est_lin ] ) .
now we introduce the helmholtz decomposition of the velocity ( see e.g. @xcite ) @xmath295 where @xmath296 and @xmath297 .
we see that the field @xmath298 satisfies the following system @xmath299 for this system we have ( see @xcite ) : @xmath300 what by ( [ alpha_w1p_1 ] ) can be rewritten as @xmath301 + \epsilon \|u\|_{w^2_p}\ ] ] for any @xmath293 .
now we substitute the helmholtz decomposition to ( [ system_lin])@xmath257 .
we get @xmath302 = f - \partial_{x_1 } a + \mu \delta a - \partial_{x_1 } \nabla \phi - p_2\nabla \eta,\ ] ] but @xmath303 .
we denote @xmath304 combining the last equation with ( [ system_lin])@xmath305 we arrive at @xmath306 where @xmath307 and @xmath308 equation ( [ trans ] ) makes possible to estimate the @xmath309-norm of the density in terms of @xmath309 - norm of @xmath310 , which in turn will be controlled by ( [ nablak ] ) using interpolation and the energy estimate ( [ ene1 ] ) .
first we estimate @xmath277 in terms of @xmath310 .
the result is stated in the following lemma assume that @xmath311 satisfies equation ( [ trans ] ) with @xmath312
. then @xmath313.\ ] ] in order to find a bound on @xmath314 we multiply ( [ trans ] ) by @xmath315 and integrate over @xmath46 .
integrating by parts and next using the boundary conditions we get @xmath316 with the smallness of @xmath317 , the above implies @xmath318 and so @xmath319 .\ ] ] now we estimate the derivatives . in order to find a bound on @xmath320 we differentiate ( [ trans ] ) with respect to @xmath321 . note that @xmath322 is not defined in general , however , for @xmath323\ ] ] and @xmath324 we may set @xmath325 .
hence we can differentiate ( [ trans ] ) with respect to @xmath321 , multiply by @xmath326 and integrate . since @xmath327
, we have @xmath328 next , since @xmath329 , we can write @xmath330 = - \frac{1}{p } \int_{\omega } |\partial_{x_i}\sigma|^p \ , { \rm div}\ , \tilde u \,dx + \frac{1}{p } \int_{\gamma } |\partial_{x_i}\sigma|^p \ , \tilde u \cdot n \ , ds \\[5pt ] = - \frac{1}{p } \int_{\omega } |\partial_{x_i}\sigma|^p \ , { \rm div}\ , u \,dx - \frac{1}{p } \int_{\gamma_{in } } |\partial_{x_i}\sigma_{in}|^p \ , ( 1+u^1 ) \,ds + \frac{1}{p } \int_{\gamma_{out } } |\partial_{x_i}\sigma|^p \ , ( 1+u^1 ) \,ds . \end{array}\ ] ] for @xmath331 we have @xmath332 and hence the above defines the trace of @xmath333 on @xmath20 .
we arrive at @xmath334 for @xmath331 ( [ wxi ] ) gives straightforward bound on @xmath335 . in order to estimate @xmath336
we also differentiate ( [ trans ] ) with respect to @xmath238 and multiply by @xmath337 .
the difference is that @xmath336 is not given on @xmath19 . to overcome this difficulty we can observe that on @xmath19 equation ( [ trans ] ) reduces to @xmath338 \,\partial_{x_1}\sigma = k,\ ] ] what can be rewritten as @xmath339,\ ] ] where the lower index @xmath340 denotes the tangential component .
thus we have @xmath341.\ ] ] using this bound in ( [ wxi ] ) , @xmath342 , we arrive at the estimate @xmath343.\ ] ] applying the trace theorem to the term @xmath344 and then combining ( [ wxi ] ) ( for @xmath345 and @xmath346 ) with ( [ wx1 ] ) we get @xmath347.\ ] ] the term @xmath348 can be put on the l.h.s . due to the smallness assumption and
thus by young s inequality @xmath349,\ ] ] what combined with ( [ w ] ) yields @xmath350.\ ] ] the lemma is proved .
now we estimate @xmath310 in terms of the data .
the result is [ lemk ] let @xmath310 be defined in ( [ k ] ) . then @xmath351 we have @xmath352 where @xmath294 is defined in ( [ est_lin ] ) .
applying first ( [ int1 ] ) and then ( [ ene1 ] ) we get @xmath353.\ ] ] hence , it is enough to find the bound on @xmath354 . by ( [ nablak ] )
we have @xmath355,\ ] ] where @xmath356 is the helmholtz decomposition of the velocity . applying ( [ aw2p ] ) , ( [ eta_w2p ] ) , trace theorem and ( [ int1 ] ) we arrive at @xmath357 \\
+ \delta_1 \|u\|_{w^2_p } + c(\delta_1 ) \big[\|f\|_{l_2 } + \|g\|_{l_2 } + \|b\|_{l_2(\gamma ) } \big ] .
\end{array}\ ] ] combining ( [ klp ] ) and ( [ nablahlp ] ) we get ( [ lemk_teza ] ) .
the bound on @xmath311 now follows directly . substituting ( [ lemk_teza ] ) to ( [ w_w1p ] )
we get @xmath358 where @xmath294 is given by ( [ est_lin ] ) .
now the only missing piece to complete the proof of ( [ est_lin ] ) is the bound on @xmath279 . to show it note that the velocity satisfies the lam system @xmath359 the classical theory of elliptic equations ( @xcite,@xcite ) yields @xmath360.\ ] ]
applying ( [ int1 ] ) to the term @xmath361 and then ( [ ene1 ] ) we get @xmath362 _ proof of lemma [ lem_est_lin ] . _
we combine ( [ eta_w2p ] ) , ( [ u_w2p ] ) and ( [ w_w1p_ost ] ) . in ( [ w_w1p_ost ] )
we choose for example @xmath363 where @xmath71 is the constant from ( [ u_w2p ] ) . now
( [ est_lin ] ) follows immediately .
in this section we solve the linear system ( [ system_lin ] ) and hence show that the sequence ( [ system_seq ] ) is well defined .
we start with showing the existence of an appropriately defined weak solution applying the galerkin method modified in a way to deal with the continuity equation .
then we show the regularity of the solutions for the given regularity of the data . for simplicity
let us denote @xmath364.\ ] ] in order to define a weak solution to ( [ system_lin ] ) we recall the definition of the space @xmath365 ( [ def_v ] ) .
a natural definition of a weak solution to the system ( [ system_lin ] ) is @xmath366 such that * 1 . *
the identity @xmath367 is satisfied @xmath368 ; * 2 . *
( [ system_lin])@xmath305 is satisfied in the following sense : @xmath369 @xmath370 , where @xmath371 is defined in ( [ tilde_u ] ) , and * 3 . *
the equation @xmath372 holds for all @xmath373 .
we introduce an orthonormal basis of @xmath365 formed by @xmath374 .
we consider finite dimensional spaces : . the sequence of approximations to the velocity will be searched for as @xmath375 due to the equation ( [ system_lin])@xmath305 we have to define the approximations to the density in an appropriate way .
namely , we set @xmath376 , where @xmath377 is defined in ( [ def_s ] ) , with @xmath378 . finally , to define the approximations of the temperature we introduce @xmath379 as a weak solution operator to equation ( [ system_lin])@xmath280 , i.e. we set @xmath380 the well - posedness of the operator @xmath381 is direct as it is just a solution operator for a standard elliptic equation . to show it
is well defined it is enough to recall the estimate from the proof of ( [ ene ] ) . in the same way
we show the estimate for ( [ weak3 ] ) which gives existence of @xmath262 satisfying ( [ weak3 ] ) for given @xmath382 , hence @xmath381 is well defined .
in particular , we have @xmath383.\ ] ] with the operators @xmath236 and @xmath381 well defined we can proceed with the galerkin method . taking @xmath384 , @xmath385 , @xmath386 and @xmath387 in ( [ weak1 ] ) , @xmath388 ,
we arrive at a system of @xmath389 equations @xmath390 where @xmath391 is defined as @xmath392 \ , ( v^n \cdot \tau_j ) \,ds - \int_{\omega } f \cdot v^n \,dx .
\end{array}\ ] ] now , if @xmath393 satisfies ( [ system_aprox ] ) for @xmath394 and @xmath395 and @xmath396 , defined as above , then the triple @xmath397 satisfies ( [ weak1])([weak3 ] ) for @xmath398 , @xmath399 .
we will call such a triple an approximate solution to ( [ weak1])([weak3 ] ) .
the following lemma gives existence of a solution to system ( [ system_aprox ] ) : let @xmath400 , @xmath401 , @xmath402 , @xmath403 .
assume that @xmath188 and @xmath63 , @xmath28 , @xmath64 fulfill the assumptions of theorem [ main_thm ] .
then there exists @xmath404 satisfying ( [ system_aprox ] ) for @xmath386 .
moreover , @xmath405 we will apply a well - known tool of the galerkin method , lemma [ lem_p ] . thus we define @xmath406 as @xmath407 we have to show that @xmath408 on some sphere in @xmath409 . as @xmath410 is linear in the second variable , we have @xmath411 by the korn inequality we have @xmath412 for @xmath64 large enough . to deal with @xmath413
let us denote for a moment @xmath414 . then setting @xmath415 in ( [ weak3 ] ) ( with @xmath416 instead of @xmath137 ) we have @xmath417\ ] ] and repeating the reasoning from the proof of ( [ ene ] ) we infer @xmath418 we have to find a bound on @xmath419 . denoting @xmath420 we have @xmath421 using ( [ est_s ] ) we get @xmath422 the remaining part of ( [ i3_1 ] ) is also not very difficult .
with the first integral on the r.h.s we have @xmath423 \,dz .
% \end{array } = \frac 12 \int_\omega \big(\partial_{x_1 } |\sigma^n|^2 + u \cdot \nabla
|\sigma^n|^2\big ) \ , dx \\[2pt ] = -\frac12 \int_{\gamma_{in } } |\sigma^n|^2 \ , ds + \frac 12 \int_{\gamma_{out } } |\sigma^n|^2 \ , ds + \frac 12 \int_{\gamma } u\cdot n |\sigma^n|^2 \ , ds - \frac 12 \int_{\omega } |\sigma^n|^2 { \rm div}\ , u \ , dx \\[2pt ] \geq - \frac 12 \int_{\gamma_{in } } \sigma_{in}^2 \ , ds + \frac 12 \int_{\gamma } u\cdot n |\sigma_{in}|^2 \ , ds - \frac 12 \|{\rm div}\ , u\|_{l_\infty } \|\sigma^n\|_{l_2}^2 \\[2pt ] \geq -c \|\sigma_{in}\|_{l_2}^2 -e \|\sigma^n\|_{l_2}^2 \geq -c
-e \big(\|g\|_{l_2}^2 + \|\sigma^n\|_{w^1_2}^2 + \|\sigma_{in}\|_{l_2}^2\big ) . \end{array}\ ] ] hence we have @xmath424 combining ( [ est_p_1]),([est_p_2 ] ) and ( [ est_p_3 ] ) we conclude @xmath425,\ ] ] where @xmath426 . thus there exists @xmath427 such that @xmath428 , and applying lemma [ lem_p ] we conclude that @xmath429 . by the definition of @xmath430
, @xmath431 is a solution to ( [ system_aprox ] ) . now showing the existence of weak solution is straightforward .
the result is [ lem_weak ] assume that @xmath432 , @xmath401 , @xmath402 , @xmath403 .
let @xmath188 be small enough and assume @xmath433 satisfy the assumptions of theorem [ main_thm ] .
then there exists @xmath366 , that is a weak solution to system ( [ system_lin ] ) .
moreover , the weak solution satisfies estimate ( [ ene ] ) .
let us set @xmath434 and @xmath435 , where @xmath393 is the solution to ( [ system_aprox ] ) .
estimates ( [ est_s ] ) , ( [ est_r ] ) and ( [ est_un ] ) imply that @xmath436 .
thus , at least for a chosen subsequence ( denoted however in the same way ) @xmath437 for some @xmath438 . passing to the limit in ( [ weak1])([weak3 ] ) for @xmath397
we conclude that @xmath272 satisfies ( [ weak1])([weak3 ] ) , thus we have the weak solution . to show the boundary condition on the density we can rewrite the r.h.s of ( [ system_lin])@xmath439 as @xmath440 and , treating @xmath238 as a `` time '' variable , adapt di perna - lions theory of transport equation ( @xcite ) that implies the uniqueness of solution to ( [ def_s_1 ] ) in the class @xmath245 .
the proof is thus complete . with the estimate
the only problem is to tackle the singularities at the junction of @xmath21 with @xmath19 and @xmath20 . to this end
we reflect the weak solution in a way which preserves the boundary conditions and then apply the classical elliptic theory to the extended solution of the lam system .
the details are given in lemma [ lem_lame ] . to show the regularity of the temperature we can apply lemma [ lem_neumann ] since on @xmath27 the boundary condition on the density reduces to the neumann condition .
in this section we will show the bounds on the sequence ( [ system_seq ] ) . due to the term @xmath441 in the continuity equation we are not able to show directly the convergence of this sequence in @xmath180 to the strong solution of ( [ system ] ) .
we can show however its boundedness in @xmath180 .
next , using this bound we derive the cauchy condition in @xmath181 , and thus show the convergence in this space to some @xmath272 . on the other hand ,
the boundedness implies weak convergence in @xmath180 , and the limit must be @xmath272 ; hence the solution is strong .
the following lemma gives the boundedness of @xmath442 in @xmath180 .
[ lem_seq_bound ] let @xmath443 be a sequence of solutions to ( [ system_seq ] ) starting from @xmath444,0,0)$ ] .
then @xmath445 where @xmath446 can be arbitrarily small provided that @xmath62 defined in , quantities @xmath447 , @xmath403 , @xmath448 and @xmath188 are small enough and @xmath64 , @xmath28 , @xmath63 fulfill the assumptions of theorem [ main_thm ] .
estimate ( [ est_lin ] ) for system ( [ system_seq ] ) reads @xmath449 .
\end{array}\ ] ] denoting @xmath450 , from ( [ est_fgh ] ) and ( [ unwn_1 ] ) we get @xmath451 we aim at showing that @xmath452 thus @xmath452 is bounded by a constant that can be arbitrarily small provided that @xmath453 and @xmath62 are small enough .
indeed , let us fix @xmath454 .
( note that the constant @xmath71 can be without loss of generality taken larger than @xmath455 . )
assume that @xmath456 .
then ( [ an1 ] ) entails an implication @xmath457 and we can conclude that @xmath458 the next lemma almost completes the proof of the cauchy condition in @xmath181 for the iterating scheme .
[ lem_cauchy1 ] let the assumptions of lemma [ lem_seq_bound ] hold
. then we have @xmath459 where @xmath446 is the constant from ( [ est_seq_bound ] ) and @xmath460 can be taken arbitrarily small . subtracting ( [ system_seq])@xmath461 from ( [ system_seq])@xmath462 we arrive at @xmath463 @xmath464 @xmath465 @xmath466 estimate ( [ ene ] ) applied to this system yields @xmath467 in order to derive ( [ est_cauchy1 ] ) from the above inequality we have to examine the r.h.s .
the differences in @xmath468 and @xmath168 are bounded in a straightforward way : using the imbedding @xmath469 and hlder s inequality we derive @xmath470 the difference in @xmath471 must be investigated more carefully .
a direct calculation yields @xmath472 where in @xmath473 we include all the terms which do not contain @xmath474 as well as other terms containing the gradient of density or its difference .
direct calculation using the imbedding @xmath475 , @xmath469 and hlder s inequality yields @xmath476 next , @xmath477 we have to compute @xmath253 norm of @xmath478 , hence we multiply by @xmath479 and integrate . with the first term we have @xmath480 hence @xmath481 we have used the fact that @xmath482 and the imbedding @xmath483 .
the other term @xmath484 can be estimated similarly , only without integration by parts . combining the estimates on @xmath473,@xmath485 and @xmath484 we conclude @xmath486 the only term that remains to estimate is @xmath487 .
we emphasize that this is the term which makes it impossible to show the convergence in @xmath488 directly .
namely , if we would like to apply the estimate ( [ est_lin ] ) to the system for the difference then we would have to estimate @xmath489 what can not be done as we do not have any knowledge about @xmath490 .
fortunately we only need the @xmath69-norm of this awkward term , which can be bounded in a direct way as @xmath491 since @xmath492 for @xmath95 .
we have thus completed the proof of ( [ est_cauchy1 ] ) .
now , lemma [ lem_seq_bound ] implies that the constant @xmath493 provided that the data is small enough and the starting point @xmath494,0,0)$ ] .
it completes the proof of the cauchy condition in @xmath495 for the sequence @xmath442 .
* lemmas [ lem_seq_bound ] and [ lem_cauchy1 ] hold for any starting point @xmath496 small enough in @xmath180 , but we can start the iteration from @xmath497,0,0)$ ] without loss of generality .
in this section we prove our main result , theorem [ main_thm ] .
first we show existence of the solution passing to the limit with the sequence @xmath442 and next we show that this solution is unique in the class of solutions satisfying ( [ est_main ] ) . *
existence of the solution*. since we have the cauchy condition on the sequence @xmath442 only in the space @xmath181 , first we have to show the convergence in the weak formulation of the problem ( [ system ] ) .
the sequence @xmath442 satisfies in particular the following weak formulation of ( [ system_seq ] ) : now using the convergence in @xmath181 combined with the bound ( [ est_seq_bound ] ) in @xmath180 we can pass to the limit in ( [ weak1_seq])([weak3_seq ] ) . the convergence of the l.h.s
. of ( [ weak1_seq])([weak3_seq ] ) is obvious . recalling the definition ( [ fgh ] ) of @xmath504 and @xmath171
we verify easily that @xmath505 and @xmath506 and the only step in showing the convergence of @xmath471 which requires more attention is to show that @xmath507 however , since @xmath508 and @xmath509 in @xmath510 as well as @xmath511 in @xmath512 , we easily verify that holds true .
we conclude that @xmath272 satisfies ( [ weak1_seq])([weak3_seq ] ) .
now we need to show that this implies the strong formulation ( [ system ] ) , what can be done in a standard way , just integrating by parts in the weak formulation .
now we set @xmath513 and we see that @xmath66 solves ( [ main_system ] ) .
we clearly have @xmath514 where @xmath62 is defined in ( [ d0 ] ) , hence estimate ( [ est_main ] ) holds
. * uniqueness .
* the uniqueness in the class of solutions satisfying ( [ est_main ] ) actually results directly from the method of the proof , more precisely from the proof of ( [ est_cauchy1 ] ) .
namely , we can show uniqueness on the level of perturbations , i.e. uniqueness for ( [ system ] ) .
consider two solutions with the same data , denote it by @xmath515 and @xmath516 .
their difference then satisfy @xmath517 @xmath518 @xmath519 @xmath520 now recall that to show ( [ est_cauchy1 ] ) we applied only ( [ dif_seq1])([dif_seq4 ] ) and ( [ est_seq_bound ] ) , hence from the above equations we conclude @xmath521 provided the data are small enough we have @xmath522 and so @xmath523 which completes the proof of uniqueness and hence of theorem [ main_thm ] .
99 r.adams , _ sobolev spaces _ , pure and applied mathematics , vol . 65 . academic press , new york - london , 1975
s.agmon , a.douglis , l.nirenberg , _ estimates near the boundary for solutions of elliptic partial differential equations satisfying general boundary conditions i _ , comm.pure appl.math .
12 ( 1959 ) , 623727 .
s.agmon , a.douglis , l.nirenberg , _ estimates near the boundary for solutions of elliptic partial differential equations satisfying general boundary conditions ii _ , comm.pure appl.math
. 17 ( 1964 ) , 3592 .
da veiga , _
an @xmath179-theory for the n - dimensional , stationary , compressible navier - stokes equations , and the incompressible limit for compressible fluids .
the equilibrium solutions _
, comm.math.phys . 109
( 1987 ) , 229248 .
r.j.diperna , p.l.lions , _ ordinary differential equations , transport theory and sobolev spaces _ , invent.math .
98 ( 1989 ) , 511547 .
e.feireisl , _ dynamics of viscous compressible fluids _ , oxford lecture series in mathematics and its applications , 26 .
oxford university press , oxford , 2004 .
g.p.galdi , _ an introduction to the mathematical theory of the navier - stokes equations _ , vol.i , springer - verlag , new york , 1994 .
d.jessl , a.novotn , _ existence of renormalized weak solutions to the steady equations describing compressible fluids in barotropic regimes _ , to appear in annales ihp analyse nonlinaire .
d.jessl , a.novotn , m.pokorn , _ steady navier
stokes fourier system with slip boundary conditions _ , submitted .
s.jiang , c.zhou , _ existence of weak solutions to the three dimensional steady compressible navier
stokes equations _ , annales ihp analyse nonlinaire 28 ( 2011 ) , 485498 .
r.b.kellogg , j.r.kweon , _ compressible navier - stokes equations in a bounded domain with inflow boundary condition _ , siam j.math.anal .
28,1(1997 ) , 94108 .
r.b.kellogg , j.r.kweon , _ smooth solution of the compressible navier - stokes equations in an unbounded domain with inflow boundary condition _ , j.math.anal . and app . 220 ( 1998 ) , 657675 .
p.l.lions , _ mathematical topics in fluid mechanics .
2 . compressible models _ , oxford lecture series in mathematics and its applications , 10 .
oxford science publications . the clarendon press , oxford university press , new york , 1998 .
p.b.mucha , _ on cylindrical symmetric flows through pipe - like domains _ , j.differential equations 201 ( 2004 ) , 304323 .
p.b.mucha , _ on navier - stokes equations with slip boundary conditions in an infinite pipe _
, acta applicandae mathematicae 76 ( 2003 ) , 115 .
p.b.mucha , m.pokorn , _ on a new approach to the issue of existence and regularity for the steady compressible navier - stokes equations _ ,
nonlinearity 19 ( 2006 ) , 17471768 .
p.b.mucha , m.pokorn , _ on the steady compressible navier stokes fourier system _ , comm .
( 2009 ) , 349377 .
p.b.mucha , m.pokorn , _ weak solutions to equations of steady compressible heat conducting fluids _ , mathematical models and methods in applied sciences 20,5 ( 2010 ) , 785813 .
p.b.mucha , r.rautmann , _ convergence of rothe s scheme for the navier - stokes equations with slip conditions in 2d domains _ , zamm z. angew .
86,9 ( 2006 ) , 691701 .
a.novotn , m. padula , _ @xmath179-approach to steady flows of viscous compressible fluids in exterior domains _ , arch .
rational mech .
126,3 ( 1994 ) , 243297 .
a.novotn , k.pileckas _ steady compressible navier - stokes equations with large potential forces via a method of decomposition _ , math .
methods appl .
21,8 ( 1998 ) , 665684 .
a.novotn , i.straskraba , _ an introduction to the mathematical theory of compressible flows _ ,
oxford science publications , oxford 2004 .
t.piasecki , _ on an inhomogeneous slip - inflow boundary value problem for a steady flow of a viscous compressible fluid in a cylindrical domain _ , journal of differential equations 248 ( 2010 ) , 21712198 .
p.i.plotnikov , e.v.ruban , j.sokolowski , _ inhomogeneous boundary value problems for compressible navier - stokes equations : well - posedness and sensitivity analysis _ ,
siam j.math.anal . 40,3 ( 2008 ) , 11521200 .
p.i.plotnikov , j.sokolowski , _ on compactness , domain dependence and existence of steady state solutions to compressible isothermal navier - stokes equations _ , j.math.fluid.mech . 7
( 2005 ) , 529573 .
m.pokorn , p.b.mucha , _
3d steady compressible navier - stokes equations _ ,
discrete and continuous dynamical systems s , 1(1 ) ( 2008 ) , 151163 .
v.a.solonnikov , _ overdetermined elliptic boundary value problems _ , zap.nauch.sem.lomi 21 ( 1971 ) , 112158 .
r.temam , _ navier stokes equations _ , north - holland , amsterdam , 1977 .
a.valli , w.m.zajaczkowski , _ navier stokes equations for compressible fluids : global existence and qualitative properties of the solutions in the general case _ , comm . math . phys . 103,2 ( 1986 ) , 259296 .
w.m.zajaczkowski , _ existence and regularity of solutions of some elliptic systems in domains with edges _ , dissertationes math . , 274(1989 ) | we consider a system of partial differential equations describing the steady flow of a compressible heat conducting newtonian fluid in a three - dimensional channel with inflow and outflow part .
we show the existence of a strong solution provided the data are close to a constant , but nontrivial flow with sufficiently large dissipation in the energy equation . |
it is a fundamental question in standard quantum mechanics ( sqm ) of what type of restrictions the schrdinger evolution imposes on the behavior in time of basic objects . in particular , it is of interest to ask if sqm allows for self - adjoint operators having the so - called lyapunov property , that is , monotonicity of the expectation value irrespective of the initial state of the system .
clearly , such an operator would indicate the direction of time .
a natural candidate for a self - adjoint lyapunov operator is a ( self - adjoint ) time operator @xmath0 canonically conjugate to the hamiltonian @xmath1 , such that @xmath0 and @xmath1 form an imprimitivity system @xcite ( implying that each generates a translation on the spectrum of the other ) .
however , a well known theorem of pauli tells us that this is impossible @xcite .
recently , galapon attempted to bypass pauli s arguments and found pairs of @xmath0 and @xmath1 satisfying the canonical commutation relations , but do not constitute an imprimitivity system @xcite .
it can be shown that the @xmath0 operator obtained in this way does not have the lyapunov property .
other authors do not insist on the conjugacy of @xmath0 and @xmath1 . in this context , unruh and wald s proof that a ` monotonically perfect clock ' does not exist @xcite should be noted , as well as misra , prigogine , and courbage s no - go theorem @xcite .
still another solution , advocated by some authors , is to do away with the requirement of self - adjointness @xcite .
a lypaunov self - adjoint operator acting within the framework of standard quantum mechanics was recently introduced in @xcite . in this paper
we study some of its properties .
in particular , we derive its spectrum and generalized eigenstates , and treat the example of the free particle .
let @xmath2 be a separable hilbert space and let @xmath1 be a self - adjoint operator generating a unitary evolution group @xmath3 , with @xmath4 , on @xmath2 .
we take @xmath2 to represent the hilbert space corresponding to some given quantum system and @xmath1 its hamiltonian . for an initial state @xmath5
@xmath6 denotes the state of the system at time @xmath7 and @xmath8 its trajectory .
let @xmath9 be the space of bounded linear operators on @xmath2 .
a ( forward ) lyapunov operator is defined as follows @xcite : let @xmath10 be a self - adjoint operator on @xmath2 .
let @xmath11 be the trajectory corresponding to an initial state @xmath12 .
let @xmath13 be the set of expectation values of @xmath14 in states in @xmath11 .
then @xmath14 is a forward lyapunov operator if the mapping @xmath15 defined by @xmath16 is one to one and monotonically decreasing for all non - recurring trajectories.@xmath17 let @xmath18 be a separable hilbert space and @xmath19 the hilbert space of lebesgue square - integrable @xmath18 valued functions defined on @xmath20 , let @xmath1 be the operator of multiplication by the independent variable on @xmath19 , and let @xmath3 be the continuous one - parameter unitary evolution group generated by @xmath1 , i.e. , @xmath21(e)=[e^{-iht}f](e)=e^{-iet}f(e),\quad f\in l^2(\mathbb r^+;\,\mathcal k),\ \ e\in\mathbb r^+.\ ] ] denote by @xmath22 the hardy spaces of @xmath18 valued functions analytic in @xmath23 .
the hilbert spaces @xmath24 consisting of non - tangential boundary values on @xmath25 in @xmath22 are , respectively , isomorphic to @xmath26 . the spaces @xmath24
are orthogonal subspaces of @xmath27 with @xmath28 .
we denote the orthogonal projections in @xmath27 on @xmath29 and @xmath30 , respectively , by @xmath31 and @xmath32 .
let @xmath33 be the subspaces of @xmath27 consisting of square - integrable functions supported on @xmath34 .
there exists another orthogonal decomposition @xmath35 with projections on @xmath36 and @xmath37 .
the following theorem is proved in @xcite : [ thm : l_2_lyapunov_op ] let @xmath38 be the operator defined by @xmath39 then @xmath14 is a positive , contractive and injective operator on @xmath19 , such that @xmath40 is dense in @xmath19 and @xmath14 is a forward lyapunov operator for the evolution on @xmath19 defined in eq .
( [ eq : l_2_schrodinger_evol ] ) , i.e. , for every @xmath41 we have @xmath42 and moreover @xmath43 . @xmath17 the following corollary to theorem [ thm : l_2_lyapunov_op ] casts eq .
( [ eq : l_2_m ] ) in a more useful form : [ cor : l_2_lyapunov_op ] let @xmath14 be the lyapunov operator given in eq .
( [ eq : l_2_m ] ) .
then , for any function @xmath44 we have @xmath45 @xmath17 * proof : * for every function @xmath46 one has @xcite @xmath47 where @xmath48 is the boundary value function of @xmath49 on @xmath25 . hence , the orthogonal projection @xmath31 of @xmath19 on @xmath29 is explicitly given by @xmath50 where eq .
( [ eq : p_plus_explicit ] ) hold a.e . for @xmath51 . from eqns .
( [ eq : l_2_m ] ) , ( [ eq : p_plus_explicit ] ) we then immediately obtain eq .
( [ eq : m_explicit]).@xmath52 [ thm : m_spectrum ] the spectrum of @xmath14 satisfies @xmath53 $ ] . in @xmath19 choose an orthogonal basis @xmath54 with @xmath55 an appropriate index set such that for every @xmath56 the set @xmath57 is a basis of @xmath18 and we have @xmath58 , where
@xmath59 stands for the kronecker delta for discrete indices and the dirac delta for continuous indices .
then for @xmath60 the function defined by @xmath61 is a generalized eigenfunction of @xmath14 satisfying @xmath62 .
these eigenfunctions are normalized in such a way that @xmath63 and we have the eigenfunction expansion @xmath64 @xmath17 * proof : * consider the eigenvalue equation @xmath65 . in the basis @xmath54 the kernel of @xmath14 is given by @xmath66 .
hence , the eigenvalue equation for @xmath14 assumes the form @xmath67 where @xmath68 .
assume that @xmath60 .
to solve eq .
( [ eq : eigenvalue_equation ] ) we assume that the function @xmath69 is the boundary value on @xmath20 from above of an analytic function @xmath70 defined in @xmath71 .
so that @xmath72 .
this allows us to analytically continue eq .
( [ eq : eigenvalue_equation ] ) into the cut complex plane @xmath73 taking in eq .
( [ eq : complex_eigenvalue_eqn ] ) the difference between the limits from above and below @xmath20 we get @xmath74 the function @xmath75 can now be continued to a second riemann sheet by making use of the branch cut along @xmath76 in eq .
( [ eq : complex_eigenvalue_eqn ] ) . denoting this two sheeted function again by @xmath77 , eq . ( [ eq : complex_eigenvalue_eqn ] ) , reduces to @xmath78 eq .
( [ eq : riemann_surface_eigenvalue_eqn ] ) admits solutions of the form @xmath79 with @xmath80 , @xmath81 , and @xmath82 a normalization factor dependent on @xmath83 .
setting @xmath84 and @xmath85 the solutions of the eigenvalue equation , eq .
( [ eq : eigenvalue_equation ] ) , satisfy @xmath86 . with this choice of @xmath87 and
@xmath88 the generalized eigenfunctions of @xmath14 are given by eq .
( [ eq : m_eigenfunctions ] ) . we proceed to show by construction that the set of generalized eigenfunctions , eq .
( [ eq : m_eigenfunctions ] ) , forms a complete set for the operator @xmath14 .
consider the following integral @xmath89 where @xmath90 is a lebesgue measure on @xmath91 $ ] .
the right hand side of eq .
( [ eq : m_spectral_representation ] ) is readily obtained from that of eq .
( [ eq : m_eigenfunctions ] ) . recall the definition of the euler beta function @xmath92 .
@xmath93 is well defined for @xmath94 and @xmath95 , and can be analytically continued to other parts of the complex @xmath96 and @xmath97 planes .
however , it is not well defined for @xmath98 and @xmath99 , i.e. whenever @xmath100 in eq .
( [ eq : m_spectral_representation ] ) . to avoid this problem we shift @xmath101 ( or @xmath102 ) away from the real axis .
hence , in the integration over @xmath83 on the right hand side of eq .
( [ eq : m_spectral_representation ] ) we take the limit @xmath103 where in the last line in eq .
( [ eq : m_spectral_representation_ii ] ) we used the identity @xmath104 .
thus , the intgral on the spectrum of @xmath14 on the left hand side of eq .
( [ eq : m_spectral_representation_ii ] ) reconstructs the kernel of @xmath14 in eq .
( [ eq : m_explicit ] ) , i.e. , we have reconstructed the operator @xmath14 using the set of generalized eigenfunctions @xmath105 .
taken together with eq .
( [ eq : m_spectral_representation ] ) , eq .
( [ eq : m_spectral_representation_ii ] ) also shows that @xmath14 has no point spectrum at @xmath106 and @xmath107 and that @xmath53 $ ] .
@xmath52 consider a quantum mechanical problem for which the hamiltonian @xmath1 , defined on an appropriate separable hilbert space @xmath2 , satisfies the conditions : ( i ) its ac spectrum is @xmath108 .
( ii ) the multiplicity of @xmath109 is uniform on @xmath20 .
let @xmath110 be the subspace of @xmath2 corresponding to the ac spectrum of @xmath1
. then @xmath110 has a representation in terms of a function space @xmath19 , where @xmath18 is a hilbert space whose dimension correpsonds to the multiplicity of @xmath109 and the evolution generated by @xmath1 is represented by @xmath3 defined in eq .
( [ eq : l_2_schrodinger_evol ] ) . for such models we can apply the results discussed above . using the standard dirac notation ,
( [ eq : l_2_schrodinger_evol ] ) for @xmath14 can be rewritten as @xmath111 where the summation over @xmath112 stands for summation over discrete degeneracy indices and appropriate integration over continuous degeneracy indices .
t as an example we take a one - dimensional gaussian wave - packet representing the propagation free particle of mass @xmath113 along @xmath20 @xmath114 @xmath115 and @xmath116 are the location and width of the wave - packet in momentum space at @xmath117 . after obtaining the generalized eigenfunctions of the free hamiltonian , constructing the lyapunov operator @xmath14 according to eqs .
( [ dirac representation ] ) and ( [ eq : m_spectral_representation ] ) , and applying it to the wave - packet in eq .
( [ 1d gaussian wave - packet ] ) we obtain fig .
1 , showing the time evolution of the expectation value of @xmath14 , and fig .
2 , showing the time evolution of the associated probability density function in the position basis and that of the eigenstates of @xmath14 .
we see that if the sequence of time frames is shown in reverse order one is unable to tell whether time is running backwards or whether one is observing a gaussian wave - packet propagating along the @xmath118 ( with time running forward ) .
however , if to all frames we attach the expectation value of @xmath14 , then it possible to distinguish between these two scenarios .
the example plainly illustrates the time - ordering of states introduced into the hilbert space by the existence of a lyapunov operator . in a forthcoming publication
it will be shown that the existence of the operator @xmath14 leads to an irreversible representation of the quantum dynamics , and to a novel representation of the scattering process , in which the contribution of the resonance is singled out .
+ _ addendum _ this paper presents a mathematically rigorous reworking of some of the results appearing in @xcite , in particular , the derivation of the operator s spectrum and generalized eigenfunctions .
the operator has since been generalized to a larger class of self - adjoint lyapunov operators @xcite .
y.s . acknowledges the support of the israeli science foundation grant no . 1169/06 , and the center for advanced studies in mathematics at ben - gurion university , where part of this research was conducted .
j.s . and s.m .
acknowledge the support of the israeli science foundation grant no .
784/06 .
galapon , pauli s theorem and quantum canonical pairs : the consistency of a bounded , self adjoint time operator canonically conjugate to a hamiltonian with nonempty point spectrum , proc .
a ( 2002 ) 451 - 472 .
see , for example , h.m .
evans , perturbation theory and atomic resonances since schrdinger s time , in : f. gesztes , p. deift , c. galvez , p. perry , w. schlag ( eds . ) , spectral theory and mathematical physics : a festschrift in honor of barry simon s 60th birthday , american mathematical society , 2007 . | * in [ j. math .
phys .
51 ( 2010 ) 022104 ] a self - adjoint operator was introduced that has the property that it indicates the direction of time within the framework of standard quantum mechanics , in the sense that as a function of time its expectation value decreases monotonically for any initial state . in this paper
we study some of this operator s properties .
in particular , we derive its spectrum and generalized eigenstates , and treat the example of the free particle . * |
in this work we study dynamics of two coupled oscillators , one of which is driven by an external periodic force .
equations governing dynamics of such system are of form:@xmath0where @xmath1 , @xmath2 and @xmath3 , @xmath4 are nonlinear elastic restoring force and nonlinear force of internal friction for mass @xmath5 and mass @xmath6 , respectively .
dynamic vibration absorber , consisting of a ( generally small ) mass @xmath7 , attached to the primary vibrating system of ( typically larger ) mass @xmath5 is a generic mechanical model described by ( [ model1 ] ) @xcite .
we shall consider a special case:@xmath8 dynamics of coupled periodically driven oscillators is very complicated szemplinska1990,awrejcewicz1991,kozlowski1995,janicki1995,kuznetsov2009,awrejcewicz2012 .
starting from equations ( [ model1 ] ) , ( [ simplified ] ) we derived the exact fourth - order nonlinear equation for internal motion as well as approximate second - order effective equation @xcite ( this approximation performs well for @xmath9 ) . applying the krylov - bogoliubov - mitropolsky ( kbm ) method to the effective equation we have computed and studied the corresponding nonlinear resonances .
more exactly , we investigated the amplitude profiles ( resonance curves ) @xmath10 , i.e. dependence of the amplitude on the frequency @xmath11 , given implicitly by the kbm method .
metamorphoses of the resonance curves @xmath12 induced by changes of the control parameters , leading to new nonlinear phenomena , have been studied within the theory of algebraic curves
they occur in the neighbourhoods of singular points of @xmath12 @xcite . in the present paper
we study the exact fourth - order equation for internal motion .
it turns out that the kbm method can be applied to yield equation defining the amplitude profile @xmath13 implicitly , where @xmath14 are some parameters .
this resonance curve is more complicated then in the case of effective equation and hence more complicated metamorphoses are possible .
the aim of the present paper is to explore these possibilities .
the paper is organized as follows . in the next section the exact 4th - order equation for the internal motion in non - dimensional form
is presented . in section 3 equation for the resonance curves @xmath15 is derived from the exact fourth - order equation for internal motion via the krylov - bogoliubov - mitropolsky approach . in section 4 the theory of algebraic curves
is used to compute singular points on the exact equation amplitude profiles - metamorphoses of amplitude profiles occur in neighbourhoods of such points . in section 5
examples of analytical and numerical computations are presented for the 4th - order equation equation .
our results are summarized in the last section .
in new variables , @xmath16 , @xmath17 , equations ( model1 ) , ( [ simplified ] ) can be written as : @xmath18 where @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 .
adding equations ( [ model2 ] ) we obtain important relation between variables @xmath25 and @xmath26 : @xmath27 where @xmath28 .
we can eliminate variable @xmath25 in ( [ model2 ] ) to obtain the following exact equation for relative motion : @xmath29where @xmath30 , @xmath31 and @xmath32 is a nondimensional parameter @xcite , see also ref .
@xcite where separation of variables for a more general system of coupled equations was described . equations ( [ 4th - a ] ) , ( [ relation ] ) are equivalent to the initial equations ( [ model1 ] ) , ( [ simplified ] )
. for small @xmath33 we can reject the term proportional to @xmath33 to obtain the approximate ( effective ) equation which can be integrated partly to yield the effective equation:@xmath34 in what follows we shall assume@xmath35this model was also investigated in @xcite where limiting phase trajectories approach was used . in this work
we shall investigate the exact equation ( [ 4th - a ] ) .
we write eqns .
( [ 4th - a ] ) , ( [ re - ve ] ) in nondimensional form . introducing nondimensional time @xmath36 and rescaling variable @xmath26 : @xmath37 where:@xmath38we get:@xmath39where @xmath40 is a linear operator:@xmath41and nondimensional constants are given by:@xmath42
we apply the krylov - bogoliubov - mitropolsky ( kbm ) perturbation approach nayfeh1981 to the exact nonlinear fourth - order equation ( [ 4th - b ] ) describing internal motion of the small mass .
the equation ( [ 4th - b ] ) is written in the following form:@xmath43where @xmath40 is defined in ( [ l ] ) and @xmath44 is given by:@xmath45and@xmath46equation ( [ prepared ] ) was prepared in such way that for @xmath47 the general solution , @xmath48 @xmath49 , @xmath50 , with constant and arbitrary @xmath51 , @xmath52 , @xmath53 , @xmath54 and , moreover , the solution for @xmath55 does not contain secular terms and @xmath56 for @xmath57 we shall now look for @xmath58 resonance using the kbm method . for small nonzero @xmath59 the solution of eqns.([prepared ] ) ( [ omega ] ) and ( [ re - ve ] ) is sought in form:@xmath60with slowly varying amplitude and phase:@xmath61 computing now derivatives of @xmath62 from eqns.([sol1:1 ] ) , ( [ a ] ) , ( phi ) and substituting to eqns.([prepared ] ) ( [ omega ] ) , ( re - ve ) and eliminating secular terms and demanding @xmath63 , @xmath64 we obtain the following equations for the amplitude and phase of steady states : [ steady ] @xmath65where @xmath66 , @xmath67 , @xmath68 . solving the system of equations ( [ steady ] ) we get the implicit expressions for the amplitude @xmath69 and the phase @xmath70 : [ af ] @xmath71equation for the correcting term @xmath72 is of form : @xmath73 where @xmath74 .
solving eqn.([z1 ] ) and substituting to ( [ sol1:1 ] ) we get finally:@xmath75where @xmath76 are given by eqns.([af ] ) .
after introducing new variables , @xmath77 , @xmath78 , the equation ( [ aomega ] ) defining the amplitude profile reads@xmath79where , as before , @xmath66 , @xmath67 , @xmath80 .
a new parameter @xmath81 is a renormalized @xmath82 , @xmath83 . to obtain the corresponding expression for the effective equation ( [ eff ] )
one can put @xmath84 so that @xmath85 and then @xmath86 .
singular points of @xmath87 are computed from equations wall2004 : [ sing1 ] @xmath88we can eliminate @xmath81 from eqns .
( [ sing1a ] ) , ( [ sing1b ] ) computing @xmath89 where @xmath90 to obtain simplified equations:@xmath91from which @xmath92 , @xmath93 can be computed as functions of parameters @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 and , finally , @xmath81 can be computed from the last equation @xmath99 equations ( [ sing2a ] ) , ( [ sing2b ] ) , ( [ sing2c ] ) are still very complicated making analytical investigation virtually impossible .
we shall thus solve these equations numerically .
in the present section singular points of amplitude profiles solutions of eqns .
( [ sing2a ] ) , ( [ sing2b ] ) , ( [ sing2c ] ) are studied .
more exactly , resonance curves with one singular point , two singular points on one curve , and with degenerate singular point are presented and metamorphoses of bifurcation diagrams are shown .
we have computed singular points for the following values of control parameters : @xmath100 , @xmath101 , @xmath102 , @xmath103 , @xmath104 obtaining four physical solutions ( i.e. with @xmath105 , @xmath106 , @xmath107 ) : @xmath92 & @xmath93 & @xmath81 & @xmath108 + @xmath109 & @xmath110 & @xmath111 & @xmath112 + @xmath113 & @xmath114 & @xmath115 & @xmath116 + @xmath117 & @xmath118 & @xmath119 & @xmath120 + @xmath121 & @xmath122 & @xmath123 & @xmath124 + the first two solutions correspond to self - intersections , see fig .
[ f1 ] , while the second pair represents isolated points .
metamorphoses of bifurcation diagrams which occur in the neighbourhood of self - intersections for the exact fourth - order equation are , for small @xmath125 , qualitatively similar to those studied for the case of @xmath58 resonance in the effective equation in @xcite and are not shown here . ,
@xmath126 , @xmath102 , @xmath103 , @xmath104 , @xmath127 ( left self - intersection , red curve , @xmath128 in table @xmath112 ) , @xmath129 ( right self - intersection , red curve , @xmath130 in table @xmath112 ) and neighbouring curves ( blue and green lines).,width=377,height=302 ] it is possible , tuning the parameters properly , to obtain amplitude profile with two singular points .
let , as before , @xmath100 , @xmath101 , @xmath103 , @xmath104 , @xmath97 being arbitrary .
we can compute , for some @xmath97 , from ( [ sing2a ] ) , ( [ sing2b ] ) @xmath131 , @xmath132 , then from eqn .
( [ sing2c ] ) we get @xmath133 and @xmath134 corresponding to two curves with one intersection each . the condition for a curve with two intersections is @xmath135 for some @xmath97 . to find this value of @xmath97 we compute @xmath136 , @xmath137 for two values of @xmath97 , @xmath138 @xmath139 and use linear extrapolation to compute @xmath140 such that @xmath141 . in one step of this procedure we compute new value of @xmath142 from known @xmath143 , @xmath144 , @xmath145 and @xmath146 , @xmath147 , @xmath148 solving linear system of equations for @xmath149 , @xmath150@xmath151where @xmath152
then the next value of @xmath153 is computed as @xmath154 .
the convergence is quite fast , see tables @xmath155 , @xmath120 .
@xmath156 & @xmath157 & @xmath158 & @xmath143 & @xmath159 + @xmath109 & @xmath110 & @xmath160 & @xmath161 & @xmath162 + @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 + @xmath168 & @xmath169 & @xmath170 & @xmath171 & @xmath116 + @xmath172 & @xmath173 & @xmath174 & @xmath175 & @xmath120 + @xmath176 & @xmath177 & @xmath178 & @xmath179 & @xmath124 + @xmath180 & @xmath181 & @xmath182 & @xmath183 & @xmath184 + @xmath185 & @xmath186 & @xmath187 & @xmath143 & @xmath159 + @xmath113 & @xmath114 & @xmath188 & @xmath161 & @xmath162 + @xmath189 & @xmath190 & @xmath191 & @xmath166 & @xmath167 + @xmath192 & @xmath193 & @xmath194 & @xmath171 & @xmath116 + @xmath195 & @xmath196 & @xmath197 & @xmath175 & @xmath120 + @xmath198 & @xmath199 & @xmath200 & @xmath201 & @xmath124 + @xmath202 & @xmath203 & @xmath182 & @xmath183 & @xmath184 + figures below show convergence of two curves with one singular point to one curve with two singular points .
( left figure , green curves ) , and from table @xmath120 ( right figure , blue).,title="fig:",width=226,height=226 ] ( left figure , green curves ) , and from table @xmath120 ( right figure , blue).,title="fig:",width=226,height=226 ] bifurcation diagrams computed for parameters in the neighbourhood of such resonance curve display presence of two singular points , see figs .
[ f3 ] . in figures [ f3 ]
the parameters are @xmath100 , @xmath101 , @xmath103 , @xmath204 in both cases and @xmath205 , @xmath206 for fig . @xmath207 and @xmath208 , @xmath209 for fig . @xmath210 .
these diagrams correspond to amplitude profiles shown in fig .
[ f4 ] .
it is possible , by smooth change of the parameters , to merge two singular points lying on the red curve in figs .
@xmath116 , @xmath120 .
the resulting singular point is degenerate , i.e. fulfills the following set of equations wall2004:@xmath211where @xmath87 is given by ( [ lxy ] ) .
solving eqns .
( [ deg1 ] ) for @xmath100 , @xmath212 ( these two parameters correspond to the parameters of the critical red curve with two singular points ) we get @xmath213 , @xmath214 , @xmath215 , @xmath216 , @xmath217 , @xmath218 , see fig .
[ f5 ] .
bifurcation diagrams computed in the neighbourhood of the degenerate singular point depend sensitively on small changes of parameters , fig . f6 .
in this work we have studied dynamics of two coupled periodically driven oscillators .
the inner motion of this system has been described by the exact fourth - order equation ( [ 4th - a ] ) ( or ( [ 4th - b ] ) in nondimensional form ) . applying the kbm method we have computed approximate resonance curves ( amplitude profiles ) @xmath69 .
although the kbm method is basically used for the second - order equations we managed to apply it to the fourth - order equation since it was possible to eliminate secular terms and impose steady - state conditions .
dependence of the amplitude @xmath51 on the forcing frequency @xmath219 is complex since @xmath69 is defined implicitly as an algebraic curve , @xmath220 , see eqn.([lxy ] ) , with polynomial function @xmath221 depending on variables @xmath222 , @xmath223 and control parameters @xmath94 , @xmath95 , @xmath96
, @xmath97 , @xmath98 , @xmath81 in a complicated manner .
in our previous paper we stressed that near singular points of algebraic curves , defining amplitude profiles , metamorphoses of bifurcation diagrams ( and hence of dynamics ) take place . in the present paper
we have studied three cases of singular points of the resonance curves defined by eqn.(lxy ) : i ) the case of one singular point ( section [ 1s ] ) , ii ) the case of two singular points on one resonance curve ( section ( [ 2s ] ) ) , iii ) the case of degenerate singular point ( section ( [ 1deg ] ) ) .
indeed , dynamics of the system ( [ 4th - b ] ) changes significantly in the neighbourhood of singular points of resonance curve @xmath220 .
singular points described in section [ comp ] are just the tip of the iceberg and thus we are going to study multitude of singular points of amplitude profiles ( lxy ) in our future work .
r. starosta , j. awrejcewicz , l. manevitch , in : dynamical systems .
analytical / numerical methods , stability , bifurcation and chaos , j. awrejcewicz , m. kamierczak , p. olejnik , j. mrozowski , editors , the university of d publishing house , d 2011 ; pp . 7984 . | we study dynamics of two coupled periodically driven oscillators .
the internal motion is separated off exactly to yield a nonlinear fourth - order equation describing inner dynamics .
periodic steady - state solutions of the fourth - order equation are determined within the krylov - bogoliubov - mitropolsky approach we compute the amplitude profiles , which from mathematical point of view are algebraic curves . in the present paper
we investigate metamorphoses of amplitude profiles induced by changes of control parameters near singular points of these curves .
it follows that dynamics changes qualitatively in the neighbourhood of a singular point . |
the strong forward and reverse shocks in young supernova remnants ( snrs ) heat the ambient interstellar medium ( ism ) and supernova ejecta material to x - ray emitting temperatures .
this x - ray emission contains information on the supernova ( e.g. , type , explosion energy , mass and composition of ejecta ) and on the nature of the ambient medium ( e.g. , uniform , stellar wind ) .
in addition to heating the plasma , the forward and reverse shocks are thought to accelerate some fraction of the shocked material to cosmic ray energies .
a large body of evidence suggests that this acceleration can be quite efficient with as much as 50% of the ram kinetic energy of the ejecta being placed in relativistic particles and removed from the thermal plasma ( blandford & eichler @xcite ; jones & ellison @xcite ; kang & jones @xcite ; berezhko , ksenofontov , & petukhov @xcite ) , and some early work investigated its observational impact ( heavens @xcite , boulares & cox @xcite , dorfi & bhringer @xcite ) .
however , despite the expected efficiency of shock acceleration , the modeling and interpretation of x - ray observation from snrs are generally done assuming that the shocks _ do not _ place a significant fraction of their energy in cosmic rays . the recent discovery of nonthermal x - ray emission in shell - like snrs like sn1006 ( koyama et al .
@xcite ) , cas a ( allen et al .
@xcite ) , g347.3 - 0.5 ( slane et al .
@xcite ) and rcw86 ( borkowski @xcite ) has shown that snr shocks do accelerate electrons to tev energies and supports the case for efficient proton acceleration and the necessity of investigating its effects in the modeling of thermal x - ray emission .
besides providing a more self - consistent snr model , x - ray models of cosmic ray modified snrs provide a test of the fundamental assumption that snrs are the primary sources of galactic cosmic ray ions .
since shocks put more energy into accelerated ions than electrons , effects seen in x - ray emission will be evidence for the efficient shock acceleration of ions as well . here ,
we investigate the effects of efficient particle acceleration on the hydrodynamics and thermal x - ray emission of young snrs .
we have developed a model which couples self - similar hydrodynamics ( chevalier @xcite ) to diffusive shock acceleration ( berezhko & ellison @xcite ; berezhko @xcite ; ellison , berezhko , & baring @xcite ) , and x - ray emission including non - equilibrium ionization effects ( decourchelle & ballet @xcite ) .
we show that a modified self - similar approach is relevant in most cases and compare it to the standard unmodified case .
there is convincing direct and indirect evidence from spacecraft observations and plasma simulations that collisionless shocks are efficient accelerators .
consequently , they self - generate magnetic turbulence and develop a smooth shock precursor ( drury @xcite ; blandford & eichler @xcite ; jones & ellison @xcite ; achterberg , blandford , & reynolds @xcite ; terasawa @xcite ; ellison @xcite ) .
when the acceleration is , the shock compression ratio increases and the shocked temperature decreases compared to values . the entire proton spectrum must be accounted for self - consistently with a non - thermal tail connecting the quasi - thermal population to the energetic one ; any change in the relativistic population must impact the thermal population .
in contrast , the slope and normalization of test - particle power laws can be changed without changing the temperature and density of the shock heated gas .
a description of effects and the model used here is given in berezhko & ellison ( @xcite ) , ellison ( @xcite ) .
the evolution of young snrs can be described by self - similar solutions ( chevalier @xcite ; nadezhin @xcite ) provided that the initial density profiles in the ejecta and in the ambient medium have power law density distributions , respectively @xmath0 ( with @xmath1 ) and @xmath2 ( with @xmath3 ) , where @xmath4 is the radius , @xmath5 the snr age and @xmath6 and @xmath7 are constant .
@xmath6 can be expressed in terms of the explosion energy and ejected mass ( decourchelle & ballet @xcite ) .
the effects of cosmic ray pressure on snr dynamics were first considered by chevalier ( @xcite ) using a two - fluid , self - similar solution with a thermal gas ( adiabatic index @xmath8 ) and a relativistic gas ( @xmath9 ) having an arbitrary constant fraction of the total energy . here , we use chevalier s ( @xcite ) solutions for driven waves but with boundary conditions calculated from the shock acceleration model from berezhko & ellison ( 1999 ) . as a function of shock velocity , @xmath10 , ambient density , @xmath11 , unshocked magnetic field , @xmath12 , and ambient temperature @xmath13 , as well as the maximum particle energy , @xmath14
is determined from the shock radius , age , and diffusive mean free path ( e.g. , ellison , berezhko , & baring @xcite ) . ] and injection efficiency , @xmath15 ( the fraction of protons with superthermal energies ) , the model yields downstream values for the density and pressures in the thermal and relativistic gases , at the forward and reverse shocks .
wave heating in the precursor is used , reducing the efficiency compared to adiabatic heating . starting with test - particle solutions at a given age , we iterate until the modified solutions and nonlinear boundary conditions are satisfied .
our solutions neglect cosmic ray diffusion ( assuming they are spatially coupled to the gas an excellent approximation for all but the highest energy particles ) , and assume that both the compression ratio and the relativistic to total pressure ratio are independent of time at either shock , which is required for self - similarity to be valid . unless explicitly stated , we take @xmath16 k and @xmath17 at both shocks . for the snr parameters ,
we take : @xmath18 cgs , corresponding to @xmath19 erg , and @xmath20 amu/@xmath21 .
we can _ a posteriori _ determine the validity of the self - similar assumption , by examining the variation versus time of the compression ratio and relativistic to total pressure ratio , at the reverse shock ( rs ) and at the forward shock ( fs ) . as seen in figure 1 , these ratios are nearly constant for injection parameters @xmath22 and negligible for @xmath23 , so that the self - similar solutions are well justified in these cases . for values of @xmath24 , unmodified solutions can occur at very high sonic mach numbers @xmath25 ( i.e. , early ages ) which show a rapid transition to solutions as @xmath25 drops ( see berezhko & ellison @xcite ) .
the self - similar approach is valid if this rapid transition occurs early or late compared to the age of the snr .
figure 2 shows the density and temperature as a function of radius for various values of @xmath15 . as @xmath15 increases
, the compression ratio becomes higher ( up to 15 here instead of 4 ) , while the shocked temperature drops ( nearly a factor of 10 between the result and @xmath26 ) .
such trends were expected from chevalier ( @xcite ) , but the actual effects lead to larger modifications because of the influence of particle escape : the maximum compression ratio is no longer limited to 7 , but can rise , in principle , to infinite values .
= 0.98 notably , these modifications occur for @xmath27 , considerably lower than that deduced for the earth s bow shock and from plasma simulations ( @xmath28 ; ellison @xcite ) .
table 1 gives the relative fractions of kinetic , @xmath29 , thermal , @xmath30 , cosmic ray , @xmath31 , and escaped energy , @xmath32 , summed over the shocked ejecta and the shocked ambient medium .
an important fraction of the energy can be carried off by escaping particles .
note that in young snrs , the total energy in the shocked region increases with time as more and more ejecta is being shocked .
cosmic ray modified dynamics lead to a thinner interaction region between the forward and reverse shocks and much larger density and pressure gradients .
this suggests that the growth of hydrodynamic instabilities in the shocked material near the contact discontinuity will be faster than otherwise . with a forward shock closer to the contact discontinuity
, this may help these turbulent effects extend to the forward shock and beyond .
equally important , the higher densities and lower temperatures will lead to more rapid coulomb equilibration than expected in a test - particle case .
= 0.98 table 1 .
temperature - velocity relation and energy fractions [ cols="<,<,<,<,<,<,<",options="header " , ] the increase in shocked density and decrease in shocked temperature from effects , strongly impact the thermal x - ray emission . compared to test - particle models , we predict that regions in young snrs undergoing efficient acceleration will have a lower temperature , more intense thermal x - ray emission ( due to the larger emission integral @xmath33 and lower temperature ) , and a faster ionization timescale . in older remnants
, the temperature should drop below x - ray emitting values earlier .
we have coupled our modified hydrodynamics model to the calculation of the non - equilibrium ionization and thermal x - ray emission .
the ionization and recombination rates have been taken from arnaud & rothenflug ( @xcite ) and arnaud & raymond ( @xcite ) .
the x - ray emission code is from mewe , gronenschild , & van den oord ( @xcite ) and mewe , lemen , & van den oord ( @xcite ) .
the abundances in the interstellar medium are taken to be solar ( meyer @xcite ) .
we assume full equipartition between the electrons and ions ( @xmath34 ) and neglect any contribution from the superthermal electrons to the ionization or line excitation . for efficient shock acceleration , this may be important and we are investigating these effects ( porquet et al . 2000 ) .
= 0.98 as an indication of how the interpretation of x - ray data depends on acceleration efficiency , we model the x - ray observations of kepler s snr obtained over a broad energy range with _
( kinugasa & tsunemi 1999 ) and _ rxte _ ( decourchelle & petre 1999 ; decourchelle et al . 2000 ) . for several values of @xmath15 , we adjust the snr parameters to fit the observed spectrum , which is dominated by the emission from the shocked ejecta as shown in figure 3 .
all values of @xmath15 yield similar values for the reduced chi - squared ( @xmath35 ) , the ionization timescale , the reverse shock temperature @xmath36 kev , the hydrogen column density @xmath37 @xmath38 , and oversolar abundances of heavy elements in the shocked ejecta ( o , ne , mg , si , s , ar , ca , and fe ) .
the derived values of forward and reverse shock compression ratios ( @xmath39 and @xmath40 ) , fs and rs speeds ( @xmath41 and @xmath42 ) , ambient density , distance , and the energy - mass relation are , however , very different as shown in table 2 . to get a temperature high enough to produce the iron k - line and the high energy continuum ,
all fits require larger shock velocities and lower ambient densities .
these imply more kinetic energy in the ejecta and a larger inferred distance , @xmath43 , than the case . to obtain a distance consistent with observations
( 4.8 @xmath44 1.4 kpc , reynoso and goss @xcite ) , a low efficiency at the reverse shock is required .
this can be done in three ways : with @xmath45at the reverse shock large enough to dampen acceleration , with @xmath46 and @xmath47 to allow high mach number , unmodified solutions , or with no reverse shock acceleration ( @xmath48 ) .
a low magnetic field is expected from the expansion of the ejecta , but magnetic field amplification could increase it . in all cases ,
a good fit is obtained for efficient acceleration at the forward shock and test - particle conditions at the reverse shock .
figure 3 and table 2 ( last row ) show the case with @xmath49 .
spatially resolved spectroscopy is required to confirm both a low efficiency of the acceleration at the reverse shock and our prediction of efficient acceleration at the forward shock , which is expected for standard values of the upstream density and magnetic field .
the arbitrary power law component required to fit the broad - band x - ray continuum ( decourchelle & petre @xcite ) is likely related to the forward shock . in future work
, this component will be replaced by including self - consistently the nonthermal continuum , and using additional constraints on the acceleration parameters from the radio data .
nonlinear effects from efficient particle acceleration strongly influence the x - ray emission in snrs , and may serve as a tracer of cosmic ray acceleration . here , we have focused on the consequences of shock acceleration on the thermal x - ray emission and have presented the first , non - equilibrium ionization modeling of the x - ray emission from a specific snr . while our self - similar approach has limitations which must be carefully considered , we believe it contains the essential physics and allows the exploration of a large range of parameters . we find that typical source parameters of young snrs should result in acceleration at the forward shock , but yield near test - particle conditions at the reverse shock if the magnetic field is as low as expected from the ejecta expansion .
when cosmic ray production is efficient , shock _ heating _ is linked to particle _ acceleration _ and thus to the broad - band photon emission from energetic electrons and ions .
the high energy electrons produce both radio and x - ray synchrotron and this will be included in future work .
in addition to a possible nonthermal synchrotron component , regions undergoing strong acceleration should exhibit stronger and lower temperature thermal emission , while regions with little acceleration should be fainter and at a higher temperature .
also , other complications ( e.g. , the variation of the ambient magnetic field orientation ( reynolds @xcite ) , shock interactions with clouds and dense knots , etc . ) influence the full modeling of snrs . the spatially resolved spectra from the _ chandra _ and _ xmm - newton _
satellites should greatly aid the investigation of these and other effects . | we have developed a simple model to investigate the modifications of the hydrodynamics and non - equilibrium ionization x - ray emission in young supernova remnants due to nonlinear particle acceleration . in nonlinear , diffusive shock acceleration ,
the heating of the gas to x - ray emitting temperatures is strongly coupled to the acceleration of cosmic ray ions .
if the acceleration is efficient and a significant fraction of the shock ram energy ends up in cosmic rays , compression ratios will be higher and the shocked temperature lower than test - particle , rankine - hugoniot relations predict .
we illustrate how particle acceleration impacts the interpretation of x - ray data using the x - ray spectra of kepler s remnant , observed by _
asca _ and _ rxte_. the thermal x - ray emission provides important constraints on the efficiency of particle acceleration , in complement to nonthermal emission .
x - ray data from _ chandra _ and _ xmm newton _ , plus radio observations , will be essential to quantify nonlinear effects .
= 3000 # 1 old_#1_old # 1#1 # 1 comment#1comment = cmr5 = cmbx5 = cmr7 = cmbx7 = cmr8 = cmmi8 = cmsy8 = cmbx8 = cmtt8 = cmti8 = cmsl8 = cmr6 = cmmi6 = cmsy6 = cmbx6 # 1 # 110^#1 = 11 #
1#2 # 1#2#3 19#1 , a&a , # 2 , # 3 # 1#2#3 19#1 , a&as , # 2 , # 3 # 1#2#3 19#1 , aj , # 2 , # 3 # 1#2#3 19#1 , ann .
geophys . , # 2 , # 3 # 1#2#3 19#1 , ann .
geophysicae , # 2 , # 3 # 1#2#3 19#1 , ann .
rev .
astr .
ap . , # 2 , # 3 # 1#2#3 19#1 , apj , # 2 ,
# 3 # 1#2#3 19#1 , apj(letts ) , # 2 , # 3 # 1#2#3 19#1 , apjs , # 2 , # 3 # 1 19#1 , apj , submitted .
2dce#1 20#1 , apj , submitted . # 1#2#3 19#1 , ara&a , # 2 , # 3 # 1#2#3 19#1 , astr .
sp .
sci . , # 2 , # 3 # 1#2 1999 , proc .
26th int . cosmic ray conf .
( salt lake city ) , # 1 , # 2 .
# 1#2 19#1 , proc .
26th int
. cosmic ray conf .
( salt lake city ) , paper # 2 .
# 1#2#3 19#1 , j.g.r . , # 2 , # 3 # 1#2#3 19#1 , m.n.r.a.s . ,
# 2 , # 3 # 1#2#3 19#1 , nature , # 2 , # 3 # 1#2#3 19#1 , phys . repts . , # 2 , # 3 # 1#2#3 19#1 , rev .
geophys and sp .
phys . , # 2 , # 3 # 1#2#3 19#1 , rep .
prog .
phys . ,
# 2 , # 3 # 1#2#3 19#1 , space sci .
rev . ,
# 2 , # 3 -40pt |
recent studies show that about 75% of all spiral galaxies have bars , and out of those , some 40% have at least one inner ( second ) bar ( see e.g. laine et al .
2002 , erwin & sparke 2002 ) .
the inner bar appears randomly oriented with respect to the outer , main , bar ( e.g. buta & crocker 1993 ) , indicating that the two bars may rotate independently .
such systems of nested bars may funnel galactic gas towards an active galactic nucleus ( agn ) , as proposed by shlosman , frank & begelman ( 1989 ) . however , shlosman et al .
considered only gaseous inner bars resulting from the instability in the nuclear disk .
detection of nuclear bars in the near infrared ( e.g. jungwiert , combes & axon 1997 ) indicates that their predominant component may be stellar .
friedli & martinet ( 1993 ) investigated the stability of stellar double bars , and found that such systems can survive for a few rotations after the two bars decouple .
recently , rautiainen & salo ( 1999 ) showed that various modes , such as spiral arms and multiple bars , can coexist for several giga - years .
maciejewski & sparke ( 2000 ) found orbital families that can support double bars , and that are related to the @xmath9 and @xmath10 orbits in single barred galaxies .
they predicted that gas flows in dynamically possible stellar nested bars lack straight shocks in the inner bar , and do not involve strong gas inflow ; they therefore differ significantly from the instability scenario proposed by shlosman et al . given this discrepancy , it is important to observe the molecular gas in double barred galaxies , so that its morphology and kinematics can be confronted with the diverging dynamical scenarios .
detailed comparison of individual galaxies with high - resolution dynamical models allows us to explore the mechanisms of gas transport on large and small scales .
gas condensations in lanes running along the leading side of a large - scale bar in the models of athanassoula ( 1992 ) are observed as dust lanes in a number of galaxies .
recent dynamical models by englmaier & shlosman ( 2000 ) for the gas flow in the inner kiloparsec of a barred galaxy showed that an induced gas density wave could form a grand - design structure .
a recent hst survey of the nuclear regions of nearby seyfert galaxies shows that spiral structures ( either grand - design or flocculent ) are common in the central kiloparsec ( pogge & martini 2002 ) . despite the progress in modeling , few observations of the gas flow in double barred galaxies exist to - date which allow one to carefully test the validity of the theoretical assumptions .
an almost face - on inclination is favored for comparisons to dynamical models , since deprojection effects are minimized .
furthermore , the bars should not lie close to the kinematic major or minor axes in order that non - circular motion can be more easily studied . for our study
, we selected the nearly face - on galaxy ngc 4303 ( m 61 ) .
ngc 4303 is probably associated with the virgo cluster ( binggeli , sandage & tammann 1985 ) . a large - scale bar of about 40 ( laine et al .
2002 ) lies inside the outer spiral arms .
in addition to the large - scale bar , ngc 4303 hosts a second , inner bar of about 4 length which is surrounded by a circumnuclear starburst ring / spiral of @xmath11 ( @xmath12 0.5 kpc ) diameter ( colina & wada 2000 , perez - ramirez et al .
2000 , colina et al .
a high - angular resolution hst uv image shows several compact star forming regions within this ring .
the starburst ring dominates the integrated uv output of ngc 4303 ( colina et al .
the nucleus is classified as a liner / seyfert 2 type ( ho , filippenko & sargent 1997 ) .
for consistency with the work by colina et al .
( 1997 , 1999 , 2000 ) , we adopt the distance of m 100 , the brightest spiral in the virgo cluster ( 16.1 mpc , ferrarese et al . 1996 ; @xmath13pc ) .
the paper is organized as follows : after a brief description of the observations in section [ sec : observations ] , the kinematic and dynamical quantities directly derivable from our mm - interferometric data are presented in section [ sec : mmobs ] . in section [ sec : mmstar ] , the molecular gas distribution is compared to the stellar properties in the inner kiloparsec , and the bar properties as well as the dynamical resonances are derived . in sections
[ sec : geomdyn ] and [ subsec : gashst ] we interpret the observed gas morphology and kinematics in the framework of available dynamical models . implications and discussion are presented in section [ sec : bmods ] , followed by the summary and conclusions in section [ sec : conc ] .
ngc 4303 was observed in its @xmath0co @xmath1 line between 1999 december and 2000 march using the owens valley radio observatory ( ovro ) millimeter interferometer with six 10.4 m telescopes in its e , h and u configurations .
the resulting baselines range from 30 m to 480 m providing a spatial resolution of @xmath14 ( 150 pc ) with natural weighting ( beam size : @xmath15 with @xmath16 of @xmath17 ) .
two spectrometer modules were utilized resulting in 120 channels with a spectral resolution of 5.2kms@xmath2 per channel .
the quasar 3c273 served as a passband and phase calibrator and was observed every 20 minutes .
the average single sideband temperature was around 800k at the observed line frequency of 114.67ghz .
the resulting noise per channel is @xmath18mjybeam@xmath2 in the combined data of the 3 tracks .
( the two channels at velocities of 1596kms@xmath2 and 1574kms@xmath2 have a slightly higher r.m.s . due to the reduced sensitivity at the edge of one spectrometer module . )
the data were calibrated using the ovro software package mma ( scoville et al .
the final @xmath0co cube was mapped and deconvolved using the clean algorithm within the software package miriad ( sault et al .
for the deconvolution we used defined clean regions , 10000 iterations , a gain of 0.0075 and a 1@xmath19 cutoff level of 25 mjybeam@xmath2 , thus allowing for a proper reduction of the beam effects in the dirty cube .
side lobe artifacts might dilute the line emission seen in the straight gas lanes running in north - south direction , as the dirty beam has side lobes at the 35% level about @xmath11 north and south of the center due to the low declination of the source ( @xmath20 ; fig .
[ fig : coall ] ) .
however , since we detected no line emission above the 5@xmath19 level in the channel maps ( fig . [
fig : cochan ] ) on the opposite sites of the gas lanes , we are confident that all prominent side lobes artifacts have been removed . further analysis ( moment maps , flux measurements , rotation curve fitting ) was performed using the radio software package gipsy ( van der hulst et al .
the data cubes were corrected for the response of the primary beam , since the outer emission peaks coincide with the half power beam width ( hpbw ) of the ovro dishes .
the ( biased ) moment maps were made using a clipping level of @xmath21 with the restriction that emission above the clipping level is at least present in two adjacent channels in the primary beam corrected data cube . through the stsci hst archive we obtained high resolution imaging data of the nuclear region of ngc 4303 .
the datasets are summarized in table [ tab : hstlog ] , and include hst / wfpc2 data in the f606w filter , hst / nicmos ( camera 2 ) imaging in the f160w filter ( see also colina & wada 2000 ) , and hst / stis imaging with the stis / nuv - mama detector ( p.i .
we used standard reduction procedures .
the optical image was resampled to the pixel scale of nir image .
all images were registered to match the location of the bright central source with the h band peak emission .
the registration of all images was checked on the dusty regions and the stellar clusters , and showed good agreement ( @xmath22 ) .
the molecular gas emission is concentrated in two straight gas lanes running north and south of the nucleus ( fig .
[ fig : coiband ] and [ fig : coall ] ) .
these gas lanes show a lateral offset from the nucleus of about 7@xmath23 and curl towards the nucleus inside a radius of @xmath24 .
the distribution of the molecular gas emission is not smooth , showing local contrasts of more than a factor of four within the gas lanes .
in addition , two peaks of emission about @xmath25 ( 2.0kpc ) north and south of the nucleus are present ( see fig . [
fig : cofet ] ) .
comparison of spectra extracted for these peaks with the nuclear region rule out the possibility that these peaks could arise from side lobe artifacts due to the low declination of the source ( fig .
[ fig : cobar ] ) .
those peaks clearly show a velocity shift of the line centers .
the geometry of the molecular gas as observed in the @xmath0co @xmath1 line emission is reminiscent of the dust lane structures seen in simulations of the gas response in a barred potential ( e.g. fig . 2 of athanassoula 1992 ; hereafter a92 ; see section [ sec : geomdyn ] ) .
the @xmath0co @xmath1 velocity field in the inner 12@xmath23 appears regular in most locations , consistent with the `` spider '' diagram expected for a differently rotating disk ( fig .
[ fig : coall ] and fig .
[ fig : cochan ] ) .
the line of nodes changes at a radius of @xmath26 between the inner region and the gas lanes indicating the presence of non - circular motions .
this can also be seen in the channel maps ( fig .
[ fig : cochan ] ) where the molecular gas in the lanes moves almost in north - south direction whereas in the inner @xmath27 the gas emission moves from south - east to the north - west .
the distribution of the velocity dispersion in the 2nd moment map is relatively smooth with an average value of @xmath12 10kms@xmath2throughout the observed region ( fig .
[ fig : coall ] ) .
the position - velocity ( @xmath28 ) diagram along the kinematic major axis shows a steep rise in velocity to a distance of @xmath29 .
there is good agreement with the h@xmath30 @xmath28 diagram for the same position angle ( rubin et al .
1999 ; also fig .
[ fig : pv ] ) .
the deviations are probably due to the slightly different angular resolution and the higher spectral resolution of the @xmath0co data .
comparison with stellar absorption line data along a position angle of pa@xmath31 ( heraudeau et al .
1998 ) shows that the velocity of the stars is much lower .
it is unlikely that this is due to non - circular gas motions in the inner @xmath32 , as the @xmath0co velocity field ( fig .
[ fig : coall ] ) as well as the h@xmath33 velocity field ( colina & arribas 1999 ) are similar to a smoothly rotating disk with radial motions of less than 10kms@xmath2 ( colina & arribas 1999 ) . the lower velocity of the stars in the central region very likely reflects the fact that the stars have a different 3-dimensional distribution , and therefore different kinematics .
this is consistent with large velocity dispersions of the stellar absorption lines ( @xmath12 80kms@xmath2 ; heraudeau et al .
1998 ) compared to those of @xmath0co gas emission lines ( @xmath34kms@xmath2 ; fig .
[ fig : conuc ] ) in the nuclear region .
a similar behavior is seen in ngc 1068 where the derived stellar rotation curve has lower velocities than the rotation curve derived from the @xmath0co data and the h@xmath30 line emission ( schinnerer et al .
2000a ) . before a rotation curve can be fit to the @xmath0co data ,
the major axis position angle and inclination , the dynamical center , and the systemic velocity need to be determined .
we used an iterative approach to obtain these parameters by fitting tilted annuli to the observed @xmath0covelocity field .
for the fit we weighted points closer to the major kinematic axis higher and neglected regions that lie within 15@xmath35 of the minor axis .
we used the routine ` rotcur ' in gipsy , in which only the rotation velocity and one parameter were fit at a time , keeping all other parameters fixed .
we estimate a position angle of ( 317@xmath365)@xmath35 for the inner @xmath37 , where the @xmath0co velocity field should not be disturbed by the bar potential .
this agrees with guhathakurta et al .
( 1988 ) , who obtained a position angle of ( 318@xmath365)@xmath35 from a fit to the large scale hi velocity field , and with colina & arribas ( 1999 ) , who find a position angle of @xmath38 from modeling the h@xmath33 velocity field in the inner @xmath27 .
a robust determination of the inclination is difficult because of the almost face - on aspect of ngc 4303 and the optical disk disturbed due to tidal interaction with two nearby companions ( ngc 4292 and ngc 4303a ; binggeli et al .
from the optical images of frei et al .
( 1996 ) using the lowest undisturbed contours ( excluding the regions of the spiral arms ) , we derive an inclination of @xmath12 25@xmath35 by excluding radii that are already affected by the bar which has a position angle of @xmath12 10@xmath35 .
a fit to the @xmath0co velocity field in the inner @xmath39 gives an inclination of @xmath40 .
cayatte et al .
( 1993 ) find a similar value , and ma , peng & gu ( 1998 ) obtain 22@xmath35 from a fit to the spiral structure .
note that colina & arribas ( 1999 ) used an inclination of @xmath41 assuming that the ionized gas structure of the inner @xmath27 is circular .
however , as it seems unlikely that the inner @xmath27 disk has a different inclination , we adopt the value for the outer disk of 25@xmath35 , which is consistent with the inclination derived from the @xmath0co velocity field , and we use a position angle of 318@xmath35 to fit the rotation curve .
direct measurements of the dynamical center from the @xmath0co line data is complicated by the fact that the line emission is not smoothly distributed in the inner @xmath27 .
nevertheless , since the gas distribution appears point - symmetric , the bar potential producing this pattern is plausibly expected to be point - symmetric as well with respect to the dynamical center . if we assume that the two gas lanes are symmetrically distributed about the nucleus , we find that the dynamical center of the galaxy is located within @xmath42 of the central peak in the nicmos f160w image .
( we verified this by aligning the dust lanes seen in the hst f606w image and the gas lanes . )
thus the dynamical center is identical with the peak of the old stellar population as seen in the f160w image .
we therefore adopt 12:21:54.99 + 04:28:25.6 ( j2000.0 ) as the dynamical center for fitting of the rotation curve .
we find a systemic velocity of @xmath43kms@xmath2 using tilted ring fitting .
this translates into @xmath44kms@xmath2@xmath45kms@xmath2 for ngc 4303 .
previous measurements in the hi line gave @xmath46kms@xmath2 with the wsrt ( westerbork synthesis radio telescope ) at an angular resolution of @xmath37 ( warmels 1988 ) , and ( 1566@xmath365)kms@xmath2 with the vla ( very large array ) at an angular resolution of @xmath47 ( guhathakurta et al . 1988 ) .
our value is in very good agreement with the hi systemic velocity .
the lower value of @xmath48kms@xmath2 from the modeling of the h@xmath33 velocity field ( colina & arribas 1999 ) might be due to the lower spectral resolution of the h@xmath33 data .
we used a value of @xmath49kms@xmath2 to fit the rotation curve .
since the molecular line emission in ngc 4303 is distributed very unevenly throughout the disk , it is very difficult to obtain an optimally averaged rotation curve .
in addition , from the gas morphology we expect strong non - circular motions in the straight gas lanes .
therefore the rotation curve that we derive here is strongly biased towards regions with higher emission , and is more representative of the streaming motion outside the inner @xmath37 . on the other hand ,
the coverage of the inner @xmath37 is complete enough to give a good estimate of the averaged rotation curve representing the underlying gravitational potential .
two methods were used to obtain the rotation curve ( fig .
[ fig : rot ] ) : a least square fit to the velocity field using inclined rings ( routine ` rotcur ' ) and a direct fit to @xmath28 diagrams along eight different position angles separated by steps of 22.5@xmath35 for a good coverage of the galaxy disk ( routine ` inspector ' ) .
the latter method allowed us to examine the effect of beam smearing . in this procedure ,
a rotation curve is obtained by fitting a tilted ring model to the whole data cube by comparing the model with the @xmath28 diagrams .
this method is more robust against multiple line components or additional non - circular components .
it also allows us to fit the rotation curve to the two outer @xmath0coemission peaks at a radius of @xmath50 .
as discussed above , the dynamical center , the systemic velocity ( @xmath51kms@xmath2 ) , the position angle ( pa @xmath52 ) and the inclination ( @xmath53 ) were kept fixed . in the case of ngc 4303
the two rotation curves agree very well with each other for the inner @xmath54 ( fig .
[ fig : rot ] ) .
the only discrepancy is the peak velocity at a radius of @xmath29 which is about 20 kms@xmath2 higher for the rotation curve derived by ` inspector . '
this is expected , since this method is less affected by beam smearing . the derived velocity value at @xmath29 from the h@xmath33 velocity field ( colina & arribas 1999 ) is in very good agreement with the @xmath0co rotation curve when corrected for the different assumed inclinations of the disk ( see section [ subsec : dyn ] ) .
the drop of the rotation velocity from @xmath55kms@xmath2 to @xmath56kms@xmath2 is sub - keplerian ( fig .
[ fig : rot ] ) .
thus , the derived rotation velocity can only be regarded as a lower limit .
this behavior confirms the presence of strong non - circular motion for radii between @xmath57 .
therefore , we assume a constant rotation velocity of 150kms@xmath2 between @xmath58 . for larger radii the @xmath0co , h@xmath30 ( rubin et al .
1999 ) and hi ( cayatte et al . 1990 ) rotation velocities are consistent .
we constructed a model velocity field from the derived rotation curve .
the residual velocity field of the observed minus the model field ( fig .
[ fig : resvel ] ) shows differences of @xmath59kms@xmath2at the outer @xmath0co peaks , and @xmath60kms@xmath2 in the inner @xmath61 where the @xmath0co emission starts to form a spiral - like structure .
therefore , the @xmath0co line emission is well approximated by a rotating disk with the derived rotation curve in the inner @xmath61 .
outside that region , along the gas lanes , a maximum value for the residual velocities of about @xmath62kms@xmath2 is found at a radius of @xmath63 ( @xmath64 pc ) which is about the position of the inner lindblad resonance ( ilr ; see section [ sec : geomdyn ] ) . after correcting for inclination ,
these values translate into @xmath12 60kms@xmath2 which is the difference between the @xmath0co rotation velocity and the adopted rotation curve .
if the spiral arms are trailing , ngc 4303 is rotating clockwise and the north - eastern side is the near side of the galaxy disk . as negative residuals are found in the north - western gas lane and positive residuals in the south - eastern gas lane this is consistent with strong inflowing motions , as predicted by numerical simulations of gas flow in a barred potential ( e.g. a92 ; piner et al .
1995 ; for a detailed explanation see also regan , sheth & vogel 1999 ) .
this situation is similar to the findings of regan et al .
( 1999 ) for ngc 1530 .
the dynamical mass @xmath65 can be derived for the inner @xmath27 where the measured rotation curve agrees with solid body rotation using @xmath66 where @xmath67 is the measured circular velocity ( @xmath68 is the gravitational constant ) . for @xmath69kms@xmath2
we get a dynamical mass @xmath70m@xmath71 .
the @xmath0co @xmath1 line fluxes @xmath72 for several regions ( nucleus , inner disk , inner spiral arms , inner and outer gas lanes , and the outer peaks as indicated in fig .
[ fig : cofet ] ) are summarized in table [ tab : cofluxes ] .
the total line flux of @xmath73jykms@xmath2 detected by the interferometer can be compared to the single dish flux of @xmath74kkms@xmath2 detected at the center with the fcrao 14 m ( kenney & young 1988 ) . without any assumptions and corrections for the size of the molecular gas disk , this translates into @xmath75jykms@xmath2 .
therefore the interferometer detects only about 40% of the single dish flux .
kenney & young ( 1988 ) give a total @xmath0co @xmath1 line flux of @xmath76jykms@xmath2 for the whole galaxy using several single dish pointings , so that the molecular gas emission mapped by the interferometer comprises only about 10% of the total molecular gas present .
a recent bima song ( survey of nearby galaxies ) map of ngc 4303 with several pointings shows a large fraction of the molecular gas along the outer spiral arms ( sheth 2001 ) .
this indicates that a large - scale disk of molecular gas is present , and that the molecular gas in the spiral disk of ngc 4303 is only partially redistributed towards the nucleus .
we used the @xmath77 conversion factor of @xmath78 @xmath79(kkms@xmath2)@xmath2 from strong et al .
( 1987 ) to derive molecular gas masses .
note , that the value of the @xmath77 conversion factor has an uncertainty of a factor of 2 - 3 ( see e.g. meier , turner & hurt 2000 ; wei et al .
the molecular gas in the inner @xmath27 ( @xmath12 630 pc ) has a mass of about @xmath80 m@xmath71 ( excluding the uncertainty of the conversion factor ) , which is about @xmath12 4% of the dynamical mass of @xmath81 m@xmath71 .
close to the nucleus about @xmath82 m@xmath71 of molecular gas are present .
the amount of molecular gas mass in the two prominent parts of the straight gas lanes is about twice the amount in the inner 630 pc .
there are various ways to estimate the velocity dispersion in the inner @xmath27 of the molecular gas disk .
the velocity dispersion is about @xmath83 kms@xmath2 in the 2nd moment map which can be regarded a lower limit due to the 3@xmath19 clipping which was used to derive the map .
the spectra in the nuclear region ( fig .
[ fig : conuc ] ) show average line widths of @xmath84 kms@xmath2 which are a better measure of the observed velocity dispersion .
if the inner molecular gas ` disk ' is in hydrostatic equilibrium , the velocity dispersion of the molecular gas , after correcting for the rotational contribution due to the finite beam size , can be used to derive a disk scale height .
the observable velocity gradient is @xmath85kms@xmath2 arcsec@xmath2 in the inner @xmath27 ( using the observed velocity of @xmath86kms@xmath2 ) .
the true velocity dispersion is then @xmath87kms@xmath2 using quadratic deconvolution to correct for the contribution from the rotation velocity . several theoretical relations between the velocity dispersion and the disk height can be found in the literature .
quillen et al . (
1992 ; their equation 3.1 ) and combes & becquaert ( 1997 ; third equation in their introduction ) have derived such relations for gas disks in potentials similar to the ones of elliptical galaxies or galaxy bulges . with the true velocity dispersion of @xmath87kms@xmath2 , and a rotational velocity @xmath88kms@xmath2 as well as the associated molecular and dynamical masses
, we find a disk thickness of about @xmath62pc at a radius of @xmath89pc .
this is similar to the low molecular gas disk scale heights found in the central few 10 pc in the two nearby seyfert galaxies ngc 1068 and ngc 3227 ( schinnerer et al .
2000a , b ) .
in the inner kiloparsec , two or three dust lanes running from east to north are present in the @xmath6 color map ( fig .
[ fig : cohst ] ) .
the eastern end of the outermost dust lane coincides with the northern gas spiral arm seen in @xmath0co at our resolution of 2 .
directly north of the nucleus , the dust lane splits into two : a fainter part follows the @xmath0co pattern , but its much clearer extension continues into the ring - like blue feature .
this feature appears also in the uv continuum as a string of young stellar clusters forming the starburst ring or spiral at @xmath90 ( @xmath91pc ) ( fig .
[ fig : cohst ] ; see also fig . 1 of colina et al .
this continuation suggests that the ring might be more extincted towards the northern part , or that the star formation is weaker in the north of the ring , and ceases at the dust lane ( see also section [ subsec : gashst ] ) . in a second scenario ,
the northern molecular spiral continues inwards , and is seen as a dust lane , with star formation starting at still smaller radii .
it is interesting to note that if the uv morphology is interpreted as a spiral , it is a _ single - arm _ spiral originating from the northern molecular spiral arm .
the young stellar clusters are offset from the molecular gas distribution next to the disturbed , southern molecular spiral arm .
this arm discontinues in @xmath0co at pa@xmath92 , and finds no continuation in other features , aside possibly for the second eastern dust lane , which appears almost straight in the v - h color map .
it appears that apparently , the massive star formation next to the southern molecular arm has affected the gas distribution in that region ( see section [ subsec : sfmods ] ) .
colina & wada ( 2000 ) derive an excess extinction of about 1.7@xmath93 for the northern dust lane relative to values for face - on spiral galaxies , similar to values observed in the circumnuclear regions of other spiral galaxies ( e.g. regan & mulchaey 1999 ; martini & pogge 1999 ) .
this translates into an extinction of about 20@xmath93 in the uv band using the effective extinction law derived for the stellar continuum in starburst galaxies ( calzetti , kinney & storchi - bergmann 1994 ) .
this high extinction in the northern dust lane would be enough to obscure any uv cluster in this region , thus it seems possible that the star formation might actually occur in the ring traced by the uv clusters and the prominent ( northern ) dust lane .
comparison with the optical line , and the optical and nir continuum emission suggests that the stellar clusters are between @xmath94 myr old and have stellar masses of about @xmath95m@xmath71 ( colina & wada 2000 ) .
colina & wada ( 2000 ) noted that the uv knots east of the inner bar are older ( @xmath96myr ) and more massive ( up to @xmath97 m@xmath71 ) than the ones west of the nir bar ( @xmath98myr ) , suggesting an age trend within the star forming ring .
the molecular spiral arms can be divided into about six individual clumps with molecular gas masses of @xmath99 m@xmath71 and sizes between @xmath100 and @xmath101 ( @xmath102pc ) , typical for gmc complexes ( the 0th moment map is sensitive to cloud masses above 10@xmath103 m@xmath71 ) .
the star formation efficiency ( sfe ) for a molecular cloud lies typically between 3% ( for the total cloud including low - mass star formation ) and 24% ( for the hot cores ; see review by elmegreen et al .
if one allows for an elevated sfe of about 10% in the circumnuclear region , which seems plausible given the different physical properties compared to the disk of galaxies , then a single massive gmc complex similar to those observed in ngc 4303 could be a sufficient to form a dozen stellar clusters with an average stellar mass of @xmath104 m@xmath71 ( colina & wada 2000 ) as inferred from the uv continuum ( see section [ subsec : sfmods ] ) . colina & arribas ( 1999 ) isolated a second velocity component in the [ o iii ] line which has a velocity field almost perpendicular to that of the molecular gas . they interpret this as the signature of an ionization cone which is observed in a number of agns ( e.g. schmitt & kinney 1996 ) . since the north - east side is the near side of the galaxy , the ionization cone is in front of the south - western part of the galaxy disk .
the obvious asymmetry in the extinction in the inner @xmath27 could be then partly due to the presence of an ionization cone most affecting the region south - west of the nucleus .
this suggests that the circumnuclear colors in ngc 4303 are ( a ) strongly modified by the varying colors of the different stellar populations ( the clusters are obvious in the @xmath6 map ) , and that ( b ) an ionization cone might affect the overall distribution of the extincting material .
this cautions the interpretation of dust features seen in hst color maps as being only due to molecular gas condensations .
the large - scale bar in ngc 4303 runs almost north - south ( pa @xmath105 ) inside the spiral arms .
martin ( 1995 ) and chapelon et al . ( 1999 ) find its semi - major axis length to be about @xmath106 .
laine et al . ( 2002 ) revised this result and obtained a deprojected semi - major axis length of @xmath107 ( @xmath108kpc ) and a deprojected axial ratio of @xmath109 .
because of the large discrepancy between these results , we fit ellipses to the @xmath110 images of frei et al .
( 1996 ) and find a semi - major axis length between @xmath111 and @xmath112 .
a more exact determination of the scale length is hampered by the resolution of the frei et al .
data and the fact that the spiral arms start at a radius of @xmath113 , which must be close to the end of the bar .
since our value for the semi - major axis appears to be more consistent with the dynamical models where the peak gas density is close to the end of the bar ( see section [ sec : geomdyn ] ) , we assume a deprojected semi - major axis length of @xmath114 . with its axial ratio the bar
is considered a strong bar , as defined by friedli & martinet ( 1993 ) .
the large - scale spiral arms are sites of massive star formation as can be seen in the h@xmath30 line emission ( banfi et al .
1993 , koopmann , kenney & young 2001 ) .
however , no prominent h@xmath30 line emission is found along the bar .
this is in agreement with model predictions that the shear along _ strong _ bars can be too high for stars to form .
the secondary bar with a deprojected semi - major axis length of @xmath115 ( @xmath116pc ) is visible in the hst @xmath117 band image inside the stellar clusters ( colina & wada 2000 ; also fig .
[ fig : cohst ] ) .
the deprojected axial ratio of the inner bar of @xmath118 ( perez - ramirez et al .
2000 ) is at the high end of observed ratios for secondary bars ( jungwiert , combes & axon 1997 ) . if the observed axial ratio is not diluted by the underlying bulge component it indicates a weaker bar potential for the secondary bar relative to the primary bar .
a string of faint star - forming clusters can be seen in the uv image north of the nucleus ( e.g. fig .
[ fig : cohst ] ; colina & arribas 1999 ) .
it coincides with the leading side of the secondary nir bar assuming it has the same rotation direction as the outer / primary bar .
the ratio between the semi - major axes of the two bars is about 15 , which is at the upper end of the observed range ( jungwiert , combes & axon 1997 ; maciejewski & sparke 2000 ; laine et al . 2002 ) .
the deprojected position angle of the secondary bar seen in the hst @xmath117 band image relative to the kinematic major axis is pa@xmath119 .
therefore the two stellar bars are oriented at about @xmath120 with respect to each other in the plane of the host galaxy .
assuming the standard relation of @xmath121 ( e.g. a92 ) between the bar semi - major axis length @xmath122 and corotation radius @xmath123 where the angular velocity @xmath124 is equal to the bar pattern speed @xmath125 , we find that the corotation radius lies between @xmath126 and @xmath127 .
thus the primary bar rotates with a pattern speed of about @xmath128kms@xmath2kpc@xmath2 . for this pattern speed , and the rotation curve derived in section [ subsec : rotcurve ]
( see fig [ fig : rot ] ) , the large - scale bar has two ilrs . the inner ilr ( iilr ) , when localized directly from the data , lies at @xmath14 radius , and coincides with the semi - major axis radius of the nuclear bar ( @xmath129 ; perez - ramirez et al .
this puts the corotation resonance of the inner bar between the two ilrs of the outer bar , roughly in agreement with theoretical expectations ( e.g. friedli & martinet 1993 , maciejewski & sparke 1997 ) .
however , the dynamically preferred scenario ( maciejewski & sparke 2000 ) requires that the inner bar ends far outside the iilr of the main bar , so there is a sufficient set of orbits that can support it .
we discuss this discrepancy with the observed near - overlapping of one bar s end with the others iilr in section [ subsec : gashst ] .
the position of the outer ilr ( oilr ) can be determined using the adopted flat rotation curve for this radial distance as an upper measured limit ( see section [ subsec : dyn ] ) .
we derive the oilr to be at @xmath130 radius .
the gas response strongly depends on the gravitational potential , and especially on the properties of the stellar bar component ( a92 ) .
a large variety of possible gas morphologies is explored by a92 .
the overall distribution of the @xmath0co line emission in the inner arcminute of ngc 4303 is very reminiscent of the standard model 001 by a92 .
the molecular gas is concentrated towards the nucleus in two narrow straight gas lanes , and in additional emission peaks close to each end of the bar . near the nucleus
the straight lanes start to curl inward resembling a structure similar to a ` nuclear spiral ' or ` nuclear ring ' . as pointed out by a92 such behavior
can only be observed if an inner lindblad resonance ( ilr ) is present .
the observed structure can be explained by the gas moving gradually from @xmath9 orbits ( the family of stable periodic orbits elongated along the bar major axis ) to @xmath10 orbits ( the family of stable periodic orbits elongated along the minor bar axis ; in the notation of contopoulos 1981 ) , which only exist in the presence of an ilr .
the amazing similarity between the a92 standard model and the @xmath0codistribution in ngc 4303 suggests a bar strength consistent with the model axial ratio of about 0.4 .
this roughly agrees with the deprojected ratio of @xmath109 of laine et al .
note though that the observed value is integrated over the total light that includes the bulge , while the models give the axial ratio of the bar only .
thus the observations suggest that the axial ratio of the bar may be considerably lower than 0.4 . also , the observed angle between the straight part of the gas lanes and the bar major axis is @xmath131 in ngc 4303 which is half the value found for model 001 ( a92 ) .
this reinforces the supposition that the actual ratio of the bar might be even lower than 0.4 .
if the northern and southern emission peaks at a distance of @xmath132 are due to the dynamics imposed by the large - scale bar , they should lie inside the bar potential , roughly at the position of the 4:1 resonance ( englmaier & gerhard 1997 ; see also fig .
2(d ) of a92 ) .
for a flat rotation curve , the 4:1 resonance occurs at @xmath133 of the corotation radius , which places the corotation at @xmath134 in this interpretation .
this is consistent with the value that we assumed , and after dividing by @xmath135 it gives the bar s semi - major axis of @xmath126 . the qualitative comparison between the a92 standard model 001 and the distribution of the @xmath0co emission in ngc 4303 shows very good agreement .
the fact that some h@xmath30 emission is associated with the two outermost emission peaks ( e.g fig . 3.3 in sheth 2001 ,
5 in koopmann , kenney & young 2001 ) is also in agreement with the model that has a factor of @xmath136 lower shear in these regions .
this is sufficiently low to allow star formation to occur ( a92 ) .
the higher - resolution models of maciejewski et al .
( 2002 ) clearly show that the gas density peaks correspond to an almost shear - free region along the bar at about the position of the 4:1 resonance .
the resolution of the a92 models is , unfortunately , too coarse to allow for a direct comparison of the dynamics in the ( circum)nuclear region itself . to date
, only a few high - resolution dynamical calculations have been done to investigate the gas flow in the inner kiloparsec of barred galaxies .
englmaier & shlosman ( 2000 ) studied gas response in nuclear regions of a barred potential similar to that in the standard model of a92 .
they found that the velocity dispersion in the gas ( which is assumed to be isothermal ) together with the potential shape determines whether the gas settles on a nuclear ring , disk or spiral .
in particular , high - velocity - dispersion gas forms a nuclear grand design spiral inside the ilr .
this structure can be explained in terms of gas - density - wave theory as a response of non - selfgravitating gas to the torque of the bar .
a gas density wave can exist inside the ilr , in a region where the torque is not too strong ( the @xmath10 orbits are almost circular ) , and where no shocks are present .
the pitch angle of the nuclear spiral has a strong dependence on the gas velocity dispersion , and is higher for higher sound speed , which is the measure of velocity dispersion .
maciejewski et al . (
2002 ) noticed that although weak waves inside the ilr can be described by linear - wave theory , sometimes the strong straight shock in the large - scale bar can propagate all the way towards the center as a spiral shock , rendering the linear wave approximation invalid .
strong streaming motions seen along the gas spirals in ngc 5248 ( jogee et al .
2002 ) may be indicative of such a spiral shock there .
since the nuclear molecular gas emission in ngc 4303 shows a spiral - like pattern in the inner @xmath27 , a detailed comparison between the @xmath0co data and the predictions of the dynamical models by englmaier & shlosman ( 2000 ) was performed in order to assess how far inwards the outer bar dominates the gas dynamics .
the agreement between model predictions and the data is surprisingly good given the fact that no tailored models with the exact galaxy parameters are used . the azimuthal velocity of the molecular gas decreases in the region where the shock is strongest , i.e. close to the ilr , creating a net inflow .
it is very reminiscent of fig .
8 of englmaier & shlosman ( 2000 ) .
although that figure displays velocities at the maximal gas condensation , our derived @xmath0co rotation curve of ngc 4303 is also strongly biased towards bright emission , and thus can be explained by inflow in a strong bar in the context of the models . comparing our curve to the model one
, we find that inside a radius of @xmath29 where the rotation curve is rising , the @xmath0corotation curve can be used as a tracer for the dynamical mass . according to the model ,
the ( outer ) ilr coincides with the largest difference between the rotation curve and the azimuthal velocity .
since there are no other measurements of the rotation velocity in ngc 4303 with sufficient angular resolution , the oilr can be placed at a radius between @xmath37 and @xmath137 where our measured @xmath0co rotation velocity is lowest .
this location agrees well with the estimate in section [ subsec : barprop ] , for which we assumed a constant rotation speed .
the models predict that the arm - interarm contrast in the nuclear spiral is much smaller than the contrast between the straight shock and the surrounding disk ( see fig . 4 of englmaier & shlosman 2000 ) .
a similar behavior is observed in ngc 4303 ( fig .
[ fig : damp ] ) .
the arm - interarm density contrast _ inside _ the transition radius @xmath138 is about two in ngc 4303 , as is found for the model . below this radius
the gas density wave is driving the gas structure and forms a grand - design spiral structure . ] .
however , the contrast between the straight outer shock and the outer disk in ngc 4303 is about three times higher than in the model .
as noted by englmaier & shlosman ( 2000 ) their plot shows only a lower limit as the outer shocks are unresolved in their model .
the gas lanes in ngc 4303 are unresolved in our @xmath0co data , implying that they must be significantly narrower than @xmath139pc ( @xmath140 ) .
the dust feature in the hst @xmath141 band image suggests a width of about @xmath142 ( @xmath143pc ) .
this implies that the contrast should be even higher . on the other hand , for deriving the arm - interarm contrast we assumed that there is no low - level smooth large - scale emission present between the gas lanes .
since we detect only about 40% of the single dish flux , this assumption is not completely correct .
for example , the effect of such a smooth component at the 1@xmath19 level would be to lower the contrast by about a factor of three for larger radii .
therefore , the measured arm - interarm contrast is very likely a good approximation .
a factor of 1.5 higher velocity dispersion is found for the density peaks within the gas lanes compared to the average dispersion of @xmath144kms@xmath2 in the rest of the gas lanes ( fig .
[ fig : coall ] ) , which might be due to the high shear in these regions .
this may imply that the velocity dispersion is higher in the denser regions past the shock , and may put constraints on the post - shock star formation , e.g. may limit cluster sizes .
in contrast to other barred galaxies almost no h@xmath30 emission is seen along the bar ( e.g. sheth 2001 ) , despite the high molecular gas density .
the behavior of the pitch angle in a spiral arm can be a clear diagnostic for the presence of a gas density wave ( englmaier & shlosman 2000 ) . in our case ,
the gas pitch angle drops for radii between the ilr and the transition radius @xmath145 , presumably due to the transition from @xmath9 to @xmath10 orbits ( fig .
[ fig : pitch ] ) . inside the transition
radius the pitch angle rises again , indicating that another mechanism a gas density wave is responsible for the spiral structure . measuring the pitch angle is complicated by the fact that the spiral structure in the @xmath0co line emission is not continuous .
in addition the southern nuclear spiral shows distortions which are probably related to the ongoing star formation in the uv ring ( see section [ sec : mmstar ] ) . in order to measure the pitch angle ,
the @xmath146 moment map was deconvolved using the lucy algorithm with 1000 iterations and deprojected afterwards .
the pitch angle was measured by eye fitting a tangent to the spiral structure at several radii . to verify the result , the spiral structure seen in the @xmath0co emission
is locally fitted by various logarithmic spirals with a constant pitch angle ( @xmath147 and @xmath148 ) in fig .
[ fig : logsp ] .
the change of the pitch angle close to the transition radius @xmath138 is quite obvious .
there is an indication that the pitch angle is rising again after its drop at the transition radius , as expected from the model of a nuclear gas density wave .
note that in fig .
[ fig : pitch ] we show the pitch angle of the northern spiral only , and its morphology strongly deviates from the dust and uv continuum morphology inside of the transition radius ( see section [ sec : innerkpc ] ) .
it bends almost directly towards the nucleus , while the dust and uv continuum appear to form a ring .
we will return to this discrepancy in section [ subsec : gashst ] .
assuming that the indicated rise is real , the measured value of the pitch angle is higher than in the model by englmaier & shlosman ( 2000 )
. it may be due to the fact that the inner spiral is beyond the linear regime explored in that model .
the primary bar of ngc 4303 is stronger than the one in the model , and it may generate a stronger large - scale shock , which even in the central parts can not be approximated by the linear wave theory , as in the spiral - shock model by maciejewski et al .
as shown in the previous section , englmaier & shlosman s ( 2000 ) models of gas flow in a single bar are in a good agreement with the observed molecular gas behavior in the inner kiloparsec of ngc 4303 down to the 2-arcsec beam size of our data .
the dynamics in this region can therefore be already well described by a model of a single bar with an ilr . in particular
, we do not detect any strong inflowing motions throughout the whole inner bar , which would be characteristic for a _ gaseous _ inner bar as postulated by shlosman et al .
however , a secondary or inner , likely stellar , bar of @xmath8 length is clearly present in the hst @xmath117 band image ( e.g fig . [ fig : cohst ] ) . due to its small size the use of our @xmath0co data to search for inflow inside the inner bar is limited .
nevertheless , we can use the morphological information of the high resolution hst images ( @xmath149 ) to shed more light on the smaller - scale gas morphology in the presence of the secondary bar . in the inner 4
the @xmath0co morphology diverges from that seen in uv and traced by dust lanes ( see section [ sec : innerkpc ] ) . to examine the behavior of the three tracers ( co , uv , and dust ) , we plot the distance from the nucleus as a function of the position angle in the inner kiloparsec for all of them ( fig . [
fig : pitch ] ) . for pa
= 330@xmath150 and 345@xmath150 the parameters of the two dust lanes ( described in section [ sec : innerkpc ] ) , are given .
the weaker dust lane follows the @xmath0co pattern , while the more prominent lane joins the nuclear uv ring at pa = 0@xmath150 .
the uv continuum forms almost a ring , as it is evident from the constant radius .
its geometry only deviates from a ring for position angles of @xmath151 , where the string of uv - emitting regions seems to curl towards the center .
the derived uv pitch angle oscillates around 20@xmath150 , and does not show the rapid rise inwards as observed for the pitch angle of the @xmath0co spiral .
two interpretations of the uv images and the @xmath6 color maps seem plausible , and are discussed in detail below : ( a ) the uv continuum forms a nuclear ring which is extincted north of the nucleus . this ring
could then be related to gas dynamics expected in a double barred system .
( b ) the uv continuum forms a single - arm spiral . in this scenario ,
the inner bar might not at all affect the gas dynamics . in order to distinguish between the two scenarios , kinematic data of higher angular resolution are crucial .
the outer uv continuum sources form a ring , onto which the @xmath0co spiral converges , as seen in the transition from @xmath0co ( pa @xmath152 ) through dust ( @xmath153 pa @xmath154 ) to uv ( pa @xmath155 ) in fig .
[ fig : pitch ] . here
we show that the ring morphology is consistent with gas flow around a dynamically possible _ stellar _ secondary bar .
a fixed relative position of two stellar bars can not be stable for a long time , unless they remain parallel or orthogonal to each other ( see e.g. friedli 1996 ) .
the deprojected relative angle between the two bars in ngc 4303 is @xmath156 , and thus the bars most likely do rotate with respect to each other . in numerical simulations of friedli & martinet ( 1993 ) and rautiainen & salo ( 1999 )
the two bars decouple dynamically , and may survive for long times , so that they can be treated as stable , long - lived systems .
the presence of a moderate ilr of the primary bar is essential for the decoupling .
maciejewski & sparke ( 2000 ) performed a search for orbits supporting bars within bars , and found that orbits supporting the inner bar originate from the @xmath10 orbits of the outer bar , and therefore the inner bar should form inside the ilr of the outer one .
maciejewski & sparke ( 2000 ) were the first ones who postulated that straight shocks or dust lanes in the secondary bar are unlikely , because the orbital structure forces the secondary bar to end well inside its corotation .
in such a bar , oval rings are expected instead of straight shocks .
this finding was confirmed by hydrodynamic modeling by maciejewski et al .
( 2002 ) , which shows that the main effect of the secondary bar is to widen the nuclear ring created by the single bar .
this ring evolves into an ellipse which rotates with the inner bar .
the distance between the straight gas lanes in the primary bar of ngc 4303 is much larger than the size of the inner bar , which differs from the modeled setup ( maciejewski et al .
this allows the molecular gas spiral generated by the large - scale shock in the main bar to propagate considerably inwards before entering the region dominated by the secondary bar .
contrary to the model , the straight gas lanes in the main bar reach directly to the region dominated by the secondary bar .
this difference may explain why the molecular gas in ngc 4303 forms a spiral outside the inner bar .
the innermost morphology , where the uv continuum forms a ring , is consistent with the model that predicts no straight shocks along the inner bar . however , one observed feature contradicts the model : a string of uv regions , which lies almost north of the nucleus ( indicated by the arrow in fig .
[ fig : cohst ] ; see also fig . 2 of colina & wada 2000 ) , coincides with the leading edge of the northern half of the secondary bar .
these uv regions may indicate a shock along the inner bar that is not prohibitive to star formation .
thus , this secondary bar which is smaller than the one in the models of maciejewski et al . ( 2002 ) can not only form a ring around itself , but it may also generate gas flows similar to those seen in the main bar , and , therefore , it might enhance gas inflow to the galactic center along the string of uv regions .
if this is so , it is baffling to see no counterpart of this string on the other side of the bar , especially since the uv regions in the ring are more profound south of the nucleus . in maciejewski & sparke s ( 2000 ) model
, the inner bar can not form inside the inner ilr of the main bar , since the @xmath10 orbits which are vital for its support do not extend there . in ngc 4303 ,
the secondary bar is relatively smaller , and a direct calculation of gradients from measured points of the inner rotation curve gives a position of the iilr almost coinciding with the inner bar s semi - major axis .
if this is true , the secondary bar in this galaxy is _ not _ made out of equivalents of the @xmath10 orbits in the outer bar , and gas morphology and dynamics around it can not be compared to the model above .
however , the measured inner rotation curve should be treated with caution , as we are not sensitive to variations in velocity gradient of the inner 4 due to the resolution of the @xmath0co data . as a result ,
one gets a rotation curve with almost linear growth for the innermost measured points .
a linear rotation curve generates an angular frequency curve @xmath157 equal identically to zero , and unavoidably creates an artificial inner ilr .
based on our molecular data , a linear inner rotation curve can not be distinguished from a non - linear rotation curve , which has an @xmath157 frequency diverging monotonically to infinity at small radii , and therefore no inner ilr . thus ngc
4303 may not have an iilr at all , which justifies comparing it to the models of maciejewski et al .
the morphology of the uv sources can also be interpreted as lined up along a single - arm spiral extending all the way to the galactic center . in this scenario
the effect of the inner bar onto the gas flow would be negligible . as shown in section [ sec : innerkpc ] , the morphology of the star - forming uv clusters in ngc 4303 can be interpreted as a continuation of the northern @xmath0co spiral arm . as there is no obvious spiral counterpart originating from the southern @xmath0co arm ,
the uv clusters appear to be arranged into an @xmath7 mode propagating to the galaxy center .
such lopsidedness is not only detected on large scales but also in the central parts of spiral galaxies such as ngc 1808 ( emsellem et al .
2001 ) and ngc 3504 ( emsellem 2001 ) .
moreover , single - arm nuclear spirals appear to be quite common in recent high - contrast observations ( martini 2001 , pogge & martini 2002 ) .
no prominent deviations in the residual @xmath0co velocity field in the inner @xmath27 of ngc 4303 ( fig .
[ fig : resvel ] ) can be associated with the uv structure , in contrast to the cases of ngc 1808 and ngc 3504 , where the @xmath7 modes are associated with clear kinematic signatures ( emsellem 2001 ) . in ngc 4303
, the appearance of the southern @xmath0cospiral arm is distorted , but this asymmetry in the gas is probably too weak to display a kinematic signature . if this @xmath7 mode is real , it underlines the importance of self - gravity for the gas component in the central region of ngc 4303 : the induced and non - self - gravitating @xmath7 mode can not propagate in nuclear regions ( maciejewski et al .
2002 ) .
alternatively , the @xmath7 mode may be only a transient feature created by a single sheared gmc . in section [ sec : innerkpc ]
, we showed that one of the more massive gmcs detected in @xmath0co might already be able to form all the uv clusters .
consider a gmc originating from the northern arm : it gets sheared while spiraling into the galactic center , with star formation being triggered in separated parts at separate times .
the dynamical time at a radius of 2.5 ( @xmath12 200 pc ) is 10 myr .
the deprojected angle between the northern spiral arm and the western uv knots is about 120@xmath150 , which corresponds to a time of @xmath12 4 myr , consistent with the ages derived by colina & wada ( 2000 ) .
the ages of @xmath12 10 myr for the eastern knots are similar to the travel time of @xmath12 8 myr , if one assumes that the onset of star formation was delayed . as pointed out in section [ subsec : kin ] , the stellar and molecular gas kinematics might not be the same , and , therefore , the stellar angular velocity might be lower , resulting in higher ages . however , in order to test this hypothesis good knowledge of the stellar kinematics in the inner 6 as well as age - dating of the individual clusters are essential . in ngc 4303 molecular gas
is found ( within our spatial resolution ) up to the very nucleus suggesting the presence of a mechanism transporting the gas down to small radii . however , the nature of this mechanism is not clear , as the combination of the @xmath0co data with the hst data is consistent with two very different scenarios for the stellar and the gas component as discussed in the previous sections .
the possible interpretations of our data leave the question about the importance of the inner bar for the gas flow towards the center unanswered . if the molecular gas forms a self - gravitating @xmath7 mode or the observed spiral arm is a transient phenomenon , the inner bar is not important for the fueling of the nucleus . _ gaseous _ inner bars as postulated by shlosman et al .
( 1989 ) are characterized by strong inflowing motions throughout the bar , and the @xmath0co emission being aligned with the bar . since neither of these
are observed , this scenario has to be rejected .
we also find some inconsistency with the predicted gas flow in stellar double bars : some uv continuum emission coincides with the leading side of the inner bar on the scales of the hst resolution .
this suggests that on this smaller scale , the inner bar in ngc 4303 may enable gas inflow , contrary to the results of the hydrodynamic model by maciejewski et al .
compared to the modeled bars , the primary bar in ngc 4303 has a stronger potential which should offset the interaction between the two bars .
also the ratio of the bar lengths is at the upper end of that observed for double barred galaxies .
this might be expected , since the stronger primary bar may allow only smaller secondary bars to form within it .
these changes to the potential are likely to alter the gas flow .
dynamical models exploring a variety of double barred galaxies as well as more high - resolution observations of double barred galaxies are essential to better understand the fueling process on scales of a few 10 pc .
the gas leaving the straight shock in hydrodynamic models at the region where the shock starts to curl enters the diverging spray flow towards the other shock ( e.g. fig .
8 and 9 of regan et al .
this behavior reinforces the bisymmetry of the flow , and smears out effects of global density variations . as discussed in section [ sec : mmobs ] , the morphology as well as the intensity distribution of the @xmath0co line emission is not completely point - symmetric .
the western gas lane is about a factor of 1.8 brighter and therefore more massive than the eastern gas lane . since we are only detecting about 40% of the total molecular gas in the inner arcminute ( see section [ sec : mmobs ] )
, one possible explanation might be that the asymmetry in intensity is reflecting global density variations in the now - disturbed molecular gas disk .
the mass / intensity ratio between the northern and southern gas spiral arm is about 1.6 if the nuclear emission is added to the northern spiral ( assuming this emission is a continuation of the northern spiral , see section [ sec : geomdyn ] ) . the clump about @xmath101 to the south - east of the nucleus , between the eastern gas lane and the southern spiral arm , could then be a result of that asymmetry : dense material from the western gas lane enters the ` spray ' region there , and is more visible than its counterpart from the eastern lane .
the time scale for molecular gas to travel from the western gas lane to the eastern one at the radius of @xmath158 ( @xmath159pc ) can be estimated by assuming that the rotation velocity is a good measure of the net circular velocity in this region .
we find that it takes only about @xmath160myr .
this would imply that the global density variations were either very strong in the beginning and the bar potential is still trying to smooth them out , or that the torque of the bar onto the molecular gas disk has only started , i.e. the bar has only formed recently . because of the large amount of missing flux we favor the second possibility .
most dynamical models do not include star formation and its effects like supernova explosions and stellar winds .
therefore they might miss a critical component for the gas dynamics in the inner kiloparsec of barred galaxies ( e.g. sheth et al .
ngc 4303 offers the possibility to explore the importance of these effects , as it hosts young stellar clusters in addition to molecular gas .
we neglect the consumption of molecular gas during star formation , since this effect is smaller than changes in kinematics resulting from outflows from young stars .
the molecular gas geometry in the inner @xmath27 of ngc 4303 is disturbed , and shows no tight spatial correlation with the star formation .
this is in contrast to observations of other starburst rings such as the one in ngc 4314 ( benedict et al .
although ngc 4303 is in a tidal interaction with two nearby companions ( ngc 4292 and ngc 4303a , binggeli et al .
1985 ) , it is very likely that the gravitational potential of ngc 4303 dominates the inner kiloparsec . as outlined in section [ sec : gasasym ] , there is no obvious reason for the asymmetry from a dynamical point of view .
one possibility might be the impact of massive star formation via stellar winds and supernovae explosions which will be discussed below . as shown in section [ sec : innerkpc ] the uv continuum lies in the region of the molecular gas spiral arms , and between the inner and outer ilr ( see section [ sec : ilr ] ) .
elmegreen ( 1994 ) used a linear instability analysis to estimate the critical density for star formation within the inner lindblad resonance ( ilr ) .
since it is possible to measure the gas density in ngc 4303 in the presence of star formation , we can compare the observations to theoretical predictions .
the critical density @xmath161 can be approximated with @xmath162 , where @xmath163 is the epicyclic frequency . using our values for @xmath164kms@xmath2kpc@xmath2 at the distance of the uv ring of @xmath165pc
, we get a critical density @xmath166m@xmath71pc@xmath167 .
the observed molecular hydrogen mass is @xmath12 3.5@xmath16810@xmath169 m@xmath71 . assuming a ring with a width of about @xmath42 ( 78pc ) at a radius of 225pc and a gas disk scale height of 30pc ( see section [ sec : mmobs ] )
, we find a gas density of @xmath170m@xmath71pc@xmath167 for the position of the uv ring .
this is only about a factor of 15 lower than the critical density , similar to the high critical density found for the starburst ring in m 82 ( wild et al .
however , since the width of the ring is not resolved in our @xmath0co data , the gas density has to be regarded as a lower limit .
making the ring thinner by a factor of two , similar to the dust features seen in the hst @xmath6 color map , and correcting the molecular gas mass for the helium contribution ( 36% of the molecular gas mass ) brings the gas density close enough to the critical density to explain the presence of the uv starburst ring . as pointed out in section [ sec : innerkpc ] the young stellar clusters might have affected the distribution of the molecular gas by the impact of stellar winds and supernovae explosions .
a simple energy estimate can be used to test if such a large redistribution of the interstellar medium due to stellar winds and sne can be responsible for the disturbed @xmath0co distribution in the central kiloparsec of ngc 4303 .
the mechanical luminosity ( of stellar winds and supernovae ) of a 10@xmath103 m@xmath71 stellar cluster is about 10@xmath171 ergs@xmath2 during the first few myr regardless if the star formation is instantaneous or continuous ( e.g. starburst99 ; leitherer et al .
therefore the average stellar cluster in ngc 4303 with a stellar mass of @xmath104 m@xmath71 has released a mechanical energy of about @xmath172erg ( @xmath173erg ) into the interstellar medium after 6myr ( 17myr ) .
this is equivalent to 38 ( 110 ) type ii sne per stellar cluster .
if the observed molecular gas did coincide with the sites of the stellar clusters at a radius of @xmath174 ( 230 pc ) at a former time , then in order to push it to its present location at a radius of @xmath175 ( 310 pc ) its potential and kinetic energy have to be increased . under the assumption that the gravitational potential can be approximated by a spherically symmetric mass distribution ,
the gain in potential energy is @xmath176 an average cloud of molecular gas in the southern spiral arm has a mass of @xmath177 m@xmath71 ( see section [ sec : innerkpc ] ) , the velocity at the position of the uv ring and the southern gmcs are @xmath178 kms@xmath2 and @xmath179 kms@xmath2 , respectively .
integration over the circular velocities @xmath180 taken directly from the measured rotation curve ( fig.8 ) results in the acquired potential energy of @xmath181 erg . at the same time
, this cloud of molecular gas will have to gain the kinetic energy @xmath182 thus the total energy needed to move the molecular gas cloud outwards from @xmath183 to @xmath184 is the sum of the increment in potential energy @xmath185 needed to get out of the potential well , and the increase in kinetic energy @xmath185 on the rising part of the rotation curve . together these amount to at least @xmath186 erg .
this value is already considerably larger than the mechanical energy released by a single stellar cluster during its lifetime .
in addition , most of the mechanical energy heats up the gas , and we did not take into account these radiative losses .
thus the realistic estimate of energy needed to push a single molecular cloud out from 3 to 4 is even much higher than @xmath186 erg ; it therefore seems very unlikely that the uv clusters seen south of the nucleus have altered the southern co spiral arm .
however , a different effect might be important for changing the apparent gas distribution .
wei et al .
( 2001 ) have shown in m 82 that the excitation conditions for the molecular lines can differ in the presence of massive star formation : the molecular gas appears brighter than the surrounding gas which has lower temperatures , thus suggesting a different geometry . a multi - transition study of the molecular gas is needed to test if a similar scenario with varying excitation conditions can explain the observed asymmetry in ngc 4303 .
this highlights the fact that it is essential to better understand the physical interplay between the molecular gas properties and massive star formation occurring in the inner kiloparsec of double - barred galaxies .
the striking difference in the molecular gas morphology of ngc 4303 compared to other double barred galaxies such as ngc 4314 ( benedict et al .
1996 ) or ngc 1068 ( schinnerer et al .
2000a ) is the fact that the bulk of the molecular gas is located in the gas lanes , and not in the star - forming ring .
this might indicate that the circumnuclear region in ngc 4303 is in an early state of evolution .
another difference is the appearance of the ring : ngc 4314 and ngc 1068 both show almost circular tightly wound spirals resembling a ring whereas the molecular gas in ngc 4303 forms a grand design spiral pattern .
in addition , the off - nuclear star formation coincides with the gas rings in ngc 4314 and ngc 1068 suggesting that the stars have formed within the gas rings .
this is in agreement with the assumption that these rings might be older than the one in ngc 4303 .
therefore , the spiral structure seen in the uv continuum in ngc 4303 might evolve into a ring as well , similar to those observed in ngc 4314 and ngc 1068 . in this case
, on - going star formation is expected to last for a longer time than a few million years within the ring .
this would be in agreement with the finding of maoz et al .
( 2001 ) who argued for a constant star formation of about 200myr in the two starburst rings they analyzed .
ngc 4303 is one of the first double barred galaxies in which the molecular gas properties can be studied in detail at high angular and spectral resolution .
the primary bar in ngc 4303 has a typical length of about 6 kpc whereas the inner bar is considerably smaller , with a length of about 340 pc .
there is a surprisingly good agreement between the observed overall gas geometry and dynamical models for the gas flow in barred galaxies .
we detect only about 40% of the single - dish flux in our interferometric map , indicating that there is still a large smooth reservoir of molecular gas in the inner arcminute of ngc 4303 .
the gas forms two straight lanes where deviations of up to 90kms@xmath2 from the mean rotational velocity are observed . in the context of hydrodynamic models
this behavior can be explained as inflowing motions in the large - scale bar .
the velocity field of the inner @xmath27 is consistent with a circular rotating disk and shows deviations of only @xmath187kms@xmath2 .
the bulk of the molecular gas is found in the gas lanes and not in the nuclear region .
this may indicate that the circumnuclear molecular gas is still settling into the typical ring - like configuration , and that the star forming ring in ngc 4303 is still young . the star - forming ring analyzed by colina et al .
( 1997 , 1999 ) has no direct counterpart in the molecular gas distribution .
the critical density for star formation in the uv ring region ( region of the ilr ) is only about @xmath188 times higher than the estimated gas density .
therefore it seems plausible that the circumnuclear star formation was started due to the gas inflow triggered by the primary bar .
the disturbed geometry of the southern gas spiral arm suggests that the massive star formation has affected the molecular gas distribution .
however , a simple energy estimate suggests that the mechanical energy produced by the stellar clusters is too low to overcome the gravitational potential .
there is no strong correlation between the extinction as seen in the hst @xmath6 map and the distribution of the molecular gas .
possible reasons are that the light is dominated by different stellar populations as well as that an ionization cone might have affected the distribution of extincting material .
the overall morphology of the molecular gas resembles closely the predicted gas distribution in dynamical models of strongly barred galaxies . in the inner kiloparsec
the gas lanes start to curl and form a spiral structure which is distorted in the southern spiral arm .
detailed comparison between high resolution models of englmaier & shlosman ( 2000 ) and the @xmath0co data of ngc 4303 for the inner kiloparsec shows that the observed spiral structure can be explained by the density wave generated in a large - scale bar .
no strong inflowing motions characteristic for a secondary gaseous bar have been detected in our @xmath0co data .
after including the hst data in the analysis of the central 400 pc ( @xmath189 ) two scenarios seem plausible : 1 .
the innermost morphology observed with hst shows a nuclear ring inside of the nuclear @xmath0co spiral : this is uncommon in a single bar , but consistent with the gas flows in a dynamically possible double bar model ( maciejewski et al .
the large ratio ( @xmath188 ) between the two bar lengths in ngc 4303 is much larger than in this model , therefore there is room for the @xmath0co spiral structure to propagate inwards before it enters the region dominated by the secondary bar .
the faint uv emission present at the leading side of the inner bar might indicate inflow of gas along the inner bar .
2 . the uv continuum traces an @xmath7 spiral mode which is self - gravitating , and propagates independently of the inner bar .
it is imperative to study more double barred systems at high angular resolution to investigate the fueling of the central few parsecs . also
, dynamical models of double barred galaxies covering a wide range of different properties such as primary bar strength , and varying bar length ratios are important in order to better understand the gas flow in these systems .
this study has shown that the overall gas flow in ngc 4303 is dominated by the large - scale bar , and very reminiscent of current dynamical models .
however , the observed asymmetry in the gas distribution in ngc 4303 can not be explained in detail with current models . also , the impact of star formation and feedback on the ism inside the ilr may be quite significant , and
needs further consideration .
ngc 4303 obviously offers an important example against which to test elaborate models , though it will be interesting to see with comparable observations of other double barred galaxies whether it is unique , or whether it is representative of a broader class of galaxies .
it is a pleasure to thank p. englmaier , p. erwin and e. emsellem for stimulating discussions .
es acknowledges support by national sciene foundation grant ast 96 - 13717 , and lam was supported by the pparc rolling grant ppa / g / o/1999/00193 at the university of oxford .
lam is also thankful for partial support by nsf grant ast 96 - 13717 to visit the california institute of technology as part of this work .
support of this work was also provided by a grant from the k. t. and e. l. norris foundation .
research with the owens valley radio telescope , operated by california institute of technology , is supported by nsf grant ast 96 - 13717 .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
we have made use of the leda database ( http://leda.univ-lyon1.fr ) .
athanassoula , e. , 1992 , , 259 , 345 ; ( a92 ) banfi , m. , rampazzo , r. , chincarini , g. , henry , r.b.c . , 1993 , , 280 , 373 benedict , g.f . ,
smith , b.j . ,
kenney , j.d.p .
, 1996 , , 112 , 1318 binggeli , b. , sandage , a. , tammann , g.a . , 1985 , , 90 , 1681 buta , r. , crocker , d.a . , 1993 , 105 , 1344 calzetti , d. , kinney , a. l. , & storchi - bergmann , t. 1994 , , 429 , 582 cayatte , v. , van gorkom , j.h . ,
balkowski , c. , kotanyi , c. , 1990 , , 100 , 604 chapelon , s. , contini , t. , davoust , e. , 1999 , , 345 , 81 contopoulos , g. , 1981 , , 102 , 265 colina , l. , arribas , s. , 1999 , , 514 , 637 colina , l. , garcia vargas , m.l . , mas - hesse , j.m . ,
alberdi , a. , krabbe , a. , 1997 , , 484 , l41 colina , l. , wada , k. , 2000 , , 529 , 845 combes , f. , becquaert , j - f . , 1997 , , 326 , 554 elmegreen , b.g . , 1994 , , 425 , l73 elmegreen , b.g . ,
efremov , y. , pudritz , r.e .
, zinnecker , h. 1999 , in protostars and planets iv , univ . of arizona press , p. 179
emsellem , e. , 2001 , in the central kpc of starbursts and agn , ed .
knapen , j.e .
beckman , i. shlosman , & t.j .
mahoney , asp conf .
ser . , 249 , 91 emsellem , e. , greusard , d. , combes , f. , friedli , d. , leon , s. , pcontal , e. , wozniak , h. 2001 , , 368 , 52 englmaier , p. , gerhard , o. , 1997 , , 287 , 57 englmaier , p. , shlosman , i. , 2000 , , 528 , 677 erwin , p. , sparke , l.s .
2002 , , in press , astro - ph/0203514 ferrarese , l. , et al .
1996 , apj , 464 , 568 frei , z. , guhathakurta , p. , gunn , j.e . ,
tyson , j.a .
, 1996 , , 111 , 174 friedli , d. 1996 , , 312 , 761 friedli , d. , martinet , l. , 1993 , , 277 , 27 garcia , a.m. , 1993 , suppl . , 100 , 47 genzel , r. , weitzel , l. , tacconi - garman , blietz , m. , krabbe , a. , lutz , d. , sternberg , a. , 1995 , , 444 , 129 guhathakurta , p. , van gorkom , j.h . , kotanyi , c.g .
, balkowski , c. , 1988 , , 96 , 851 heraudeau , ph . ,
simien , f. , 1998 , suppl . , 133 , 317 ho , l.c . , filippenko , a.v . , sargent , w.l.w . , 1997 , , 487 , 591 van der hulst , j.m . , terlouw , j.p . , begeman , k. , zwitser , w. , roelfsema , p.r . , in `` astronomical data analysis software and systems i '' , eds .
d. m. worall , c. biemesderfer and j. barnes , asp conf .
series no .
25 , p. 131 jogee , s. , shlosman , i. , laine , s. , englmaier , p. , knapen , j.h . ,
scoville , n.z . ,
wilson , c.d .
, 2002 , , sub .
, astro - ph/0202270 jungwiert , b. , combes , f. , axon , d.j .
1997 , suppl . , 125 , 479 kenney , j.d .
, young , j.s . , 1988 , , 66 , 261 koopmann , r.a . , kenney , j.d.p . , young , j. , 2001 , , 135 , 125 laine , s. , knapen , j.h . ,
p ' erez - ram ' irez , d. , englmaier , p. , matthias , m. 2001 , , 324 , 891 laine , s. , shlosman , i. , knapen , j.h . , peletier , r.f .
2002 , , 567 , 97 leitherer , c. et al .
1999 , , 123 , 3 ma , j. , peng , q .- h .
, gu , q .- s . , 1998 ,
suppl . , 130 , 449 maciejewski , w. , sparke , l.s . , 1997 , letters , 484 , l117 maciejewski , w. , sparke , l.s . ,
2000 , , 313 , 745 maciejewski , w. , teuben , p.j . ,
sparke , l.s . , stone , j.m .
2002 , , 329 , 502 maoz , d. , barth , a.j . , ho , l.c . ,
sternberg , a. , & filippenko , a.v .
2001 , , 121 , 3048 martin , p. , 1995
, , 109 , 2428 martini , p. , 2001
, in the central kpc of starbursts and agn , ed .
knapen , j.e .
beckman , i. shlosman , & t.j .
mahoney , asp conf .
ser . , 249 , 98 martini , p. , pogge , r.w .
1999 , , 118 , 2646 meier , d.s . ,
turner , j.l . , hurt , r.l .
2000 , , 531 , 200 perez - ramirez , d. , knapen , j.h . ,
peletier , r.f . ,
laine , s. , doyon , r. , nadeau , d. , 2000 , , 317 , 234 piner , b.g . ,
stone , j.m . ,
teuben , p.j . , 1995 , , 449 , 508 pogge , r.w . , martini , p. , 2002
, , in press , astro - ph/0201185 quillen , a.c .
, de zeeuw , p.t . ,
phinney , e.s . ,
phillips , t.g . , 1992 , , 391 , 121 rautiainen , p. , salo , h. 1999 , , 348 , 737 regan , m.w . , mulchaey , j.s . , 1999 , , 117 , 2676 regan , m.w . ,
sheth , k. , vogel , s.n .
, 1999 , , 526 , 97 rubin , v.c . ,
waterman , a.h . , kenney , j.d.p .
, 1999 , , 118 , 236 sault , r.j . , teuben , p.j . , wright , m.c.h .
, 1995 , in `` astronomical data analysis software and systems iv '' , ed .
r. shaw , h.e .
payne , j.j.e .
hayes , asp conf .
ser . , 77 , 433 schinnerer , e. , eckart , a. , tacconi , l.j . , 2000b , , 533 , 850 schinnerer , e. , eckart , a. , tacconi , l.j . ,
genzel , r. , downes , d. 2000a , , 533 , 850 schmitt , h.r . ,
kinney , a.i .
, 1996 , , 463 , 498 scoville , n.z . ,
carlstrom , j.e . ,
chandler , c.j . ,
phillips , j.a .
, scott , s.l .
, tilanus , r.p.j . ,
wang , z. , 1993 , , 105 , 1482 sheth , k. , 2001 , ph.d .
thesis , univ . of maryland sheth , k. , regan , m. w. , vogel , s. n. , & teuben , p. j. 2000 , , 532 , 221 shlosman , i. , frank , j. , begelman , m.c .
1989 , natur , 338 , 45 strong , a. w. , et al .
1987 , proc .
20th intern .
cosmic ray conf .
, i , 125 telesco .
, becklin , e. , wynn - williams , c. , harper .
d. , 1984 , , 282 , 427 wada , k. , norman , c.a .
, 1999 , , 516 , l13 warmels , r.h . , 1988 , suppl . , 72 , 57 wei , a. , neininger , httemeister , s. , klein , u. , 2001 , , 365 , 571 wei , a. , walter , f. , neininger , n. , klein , u. , 1999 , , 345 , l23 wild , w. , harris , a.i . ,
eckart , a. , genzel , r. , graf , u.u . ,
jackson , j.m . ,
russell , a.p.g . ,
stutzki , j. 1992 , , 265 , 447 young , j.s .
, et al . , 1995 , , 98 , 219 the @xmath0co @xmath1 line fluxes and molecular gas masses for various components of the molecular gas distribution .
the components are indicated in fig .
[ fig : cofet ] . note , that the molecular gas masses have an uncertainty of a factor of about 2 - 3 due to the uncertainty in the @xmath77conversion factor . | we report on the molecular gas properties in the central kiloparsec of the almost face - on double barred galaxy ngc 4303 ( m 61 ) , using the @xmath0co @xmath1 line emission observed with the owens valley radio observatory ( ovro ) millimeter interferometer .
the bulk of the molecular line emission comes from two straight gas lanes which run north - south along the leading sides of the large - scale primary bar .
velocity deviations of up to 90 kms@xmath2 from the mean rotational velocity are associated with these gas lanes . inside a radius of @xmath3 ( @xmath4pc )
the molecular gas forms a spiral pattern which , for the northern arm , can be traced to the nucleus .
the high angular resolution of our ovro data ( @xmath5 150 pc ) , together with archival hst data , allows for a comparison with dynamical models of gas flow in the inner kiloparsec of single- and double - barred galaxies .
we find that the observed global properties of the molecular gas are in agreement with models for the gas flow in a strong , large - scale bar , and the two - arm spiral structure seen in @xmath0coin the inner kiloparsec can already be explained by a density wave initiated by the potential of that bar . only a weak correlation between the molecular gas distribution and the extinction seen in the hst @xmath6 map is found in the inner 400 pc of ngc 4303 : the innermost part of one arm of the nuclear @xmath0co spiral correlates with a weak dust filament in the color map , while the overall dust distribution follows a ring or single - arm spiral pattern well correlated with the uv continuum .
this complicated nuclear geometry of the stellar and gaseous components allows for two scenarios : ( a ) a self - gravitating @xmath7 mode is present forming the spiral structure seen in the uv continuum . in this case
the gas kinematics would be unaffected by the small ( @xmath8 ) inner bar .
( b ) the uv continuum traces a complete ring which is heavily extincted north of the nucleus .
such a ring forms in hydrodynamic models of double bars , but the models can not account for the uv emission observed on the leading side of the inner bar .
comparison with other starburst ring galaxies where the molecular gas emission and the star forming clusters form a ring or tightly wound spiral structure suggests that the starburst ring in ngc 4303 is in an early stage of formation . |
we would like to thank juan rojo for providing us with fig . [ pseudo ] and for useful discussions . | we present a global fit to the structure function @xmath0 measured in lepton - proton experiments at small values of bjorken-@xmath1 , @xmath2 , for all experimentally available values of @xmath3 , @xmath4 , using the balitsky - kovchegov equation including running coupling corrections . using our fits to f@xmath5
, we reproduce available data for f@xmath6 and perform predictions , parameter - free and completely driven by small - x evolution , to the kinematic range relevant for the lhec . deep inelastic lepton - hadron scattering ( dis ) experiments constitute one of the main sources of information about the partonic structure of hadrons , as well as a testing ground of pqcd techniques .
the dis program carried out in hera has explored an unprecedented kinematic regime in the dis variables @xmath3 and bjorken-@xmath1 . at small values of @xmath1 , the emission of additional soft gluons in the hadron wavefunction
is enhanced by factors of @xmath7 , resulting in a fast growth of the gluon density of the hadron with decreasing bjorken-@xmath1 .
such growth has been observed in data and , at small enough values of @xmath1 , or , equivalently , for large gluon densities , it is expected to be tamed by repulsive gluon - gluon self - interactions .
the latter mechanism is known as _ saturation _ of the gluon distribution .
it is non - linear phenomenon , closely related to unitarity of the theory , which is not included in the more standard linear approaches to dis based on the dglap or bfkl pqcd evolution equations . here
we present an analysis of available dis data at small-@xmath1 based on the non - linear balitsky - kovchegov ( bk ) equation including the recently calculated running coupling corrections to the evolution kernel @xcite .
the starting point of our analysis is the dipole model of dis , which stems from the @xmath8-factorization theorems in the high - energy limit .
it states that , at small-@xmath1 , the virtual photon - proton cross section can be written as a convolution of the light - cone wavefunction for a virtual photon to split into a quark - antiquark dipole , @xmath9 , times the cross section for dipole - proton interaction . for a hadron homogeneous in the transverse plane ,
one has : @xmath10 where the subscripts t , l stand for the polarization of the virtual photon ( transverse or longitudinal ) , @xmath11 is the dipole transverse size and @xmath3 the photon virtuality .
explicit expressions for @xmath12 in the case of any number of active flavors can be found in @xcite . in our analysis
we consider only the three flavor case , i.e , with no charm quark .
all the information about the strong interactions , and also all the @xmath1 dependence of the absorption cross section , is encoded in the imaginary part of the dipole scattering amplitude , @xmath13 ( related to the dipole cross section through the optical theorem ) .
the normalization @xmath14 in originates from integration over the transverse plane , trivial in the case of a _ cylindrical _ proton considered here , and will be one of the free parameters in the fit .
the inclusive dis structure function is then given by @xmath15 whereas the longitudinal structure function @xmath16 is obtained by just considering the longitudinal component .
general arguments in qcd indicate that the dipole scattering amplitude is small for small dipole sizes @xmath11 ( _ color transparency _ ) , whereas large dipoles suffer a large absorption .
phenomenological works in the context of saturation parametrize the transition between these two regimes , _ dilute _ and _ saturated _ , in terms of a @xmath1-dependent saturation scale @xmath17 , such that @xmath18 for @xmath19 , whereas @xmath20 for @xmath21 .
the details of the parametrization depend on the physical input considered in each case .
the leading @xmath1-dependence of the structure functions is thus encoded in that of the saturation scale , commonly assumed to have a power - law behavior : @xmath22 gev@xmath23 , with the _ evolution speed _
@xmath24 being a free parameter fitted to experimental data . the results from different groups , albeit using slightly different parametrizations of the dipole amplitude , report @xmath25 . in this work , rather than resorting to phenomenological models
, we calculate the @xmath1-dependence of the dipole amplitude in terms of the pqcd non - linear bk evolution equation including the recently calculated running coupling corrections .
importantly , the original , leading - log ( ll ) in @xmath26 bk equation yields an evolution speed @xmath27 , with @xmath28 the bare coupling @xcite .
such value is too large to attempt a good description of data .
importantly , the running coupling corrections to the ll kernel obtained through resummation of @xmath29 terms bring the evolution speed down to values compatible with those extracted from data , among other interesting dynamical effects @xcite .
the bk equation reads @xmath30\ , , \label{bklo}\end{aligned}\ ] ] where @xmath31 and @xmath32 is the starting point for the evolution . in our case
@xmath33 is the largest experimental value of @xmath1 included in the fit .
the precise shape of the running coupling kernel @xmath34 depends on the scheme chosen to single out the ultra - violet divergent contributions from the finite ones that originate after the resummation of quark loops .
it was found in @xcite that the prescription suggested by balitsky in @xcite minimizes the role of the subleading contributions , and is therefore better suited for phenomenological applications .
the corresponding kernel is @xmath35\ , .
\label{kbal}\ ] ] .values of the fitting parameters from the fit to available @xmath36 data with @xmath37 and for all available values of @xmath3 , @xmath38 . [ cols="^,^,^,^,^,^",options="header " , ] [ restab ] -0.2 cm details about the numerical method used to solve the bk equation can be found in @xcite .
we evaluate the coupling according to the standard one - loop qcd expression . in order to regularize it in the infrared , we freeze it to a constant value for @xmath39 , with @xmath40 .
thus @xmath41 and @xmath42 for @xmath39 .
the parameter @xmath43 in is a free parameter in the fit .
we take @xmath44 gev . .
solid red lines correspond to gbw i.c . , and
dotted blue ones to mv i.c.,width=8 ] a final ingredient in our analyses is the initial condition for the evolution , i.e. the specific form of the dipole amplitude at @xmath45 .
we considered two different initial conditions , _ gbw _ : @xmath46}\ , , \label{gbw}\ ] ] and _ mv _ : @xmath47}\ , , \label{mv}\ ] ] where @xmath48 is the initial saturation scale and @xmath49 is another free parameter which controls the small @xmath11 behavior of the dipole amplitude . for the fit with _
gbw _ i.c .
@xmath49 is kept fixed to 1 .
all in all , there are four ( three when @xmath49 is fixed ) free parameters in the fit : the initial saturation scale @xmath50 , the overall normalization @xmath14 , the infrared parameter @xmath43 and the anomalous dimension @xmath49 . with this setup
we obtain an excellent fit to available experimental data for the inclusive dis structure function @xmath36 with @xmath51 and all existing values for @xmath3 , @xmath52 gev@xmath23 , which adds up to 847 data points . in order to smoothly go to photoproduction ,
we have used the following redefinition of the bjorken variable : @xmath53 , with @xmath54 gev for the three light flavors considered in this work .
the experimental data included in the fit has been obtained by the h1 and zeus collaborations ( hera ) and by the e665 ( fermilab ) and nmc ( cern ) collaborations . the parameters from the fit are presented in table 1 .
we have checked that the quality of the fit and the values of the parameters are stable under the restriction of the data range to the region @xmath55 gev@xmath23 ( which leaves 703 data points for the fit ) .
a comparison between a selection of experimental data and the results from the fit for @xmath36 is shown in fig . [ figf2 ] . the two initial conditions _ gbw _ and _ mv _ offer equally good fits to data . with the parameters values yielded by the fit
, we obtain a good description of the longitudinal structure function , as shown in fig .
[ figfl ] .
-0.2 cm .
the theoretical results have been computed at the same @xmath56 as the experimental ones.,title="fig : " ] .
the theoretical results have been computed at the same @xmath56 as the experimental ones.,title="fig : " ] the analyses of available data , including this work , offer compelling evidence for the presence of gluon saturation effects .
however , a clear distinction from alternative ( linear ) approaches can not yet be established unambiguously .
indeed , a good description of all data for @xmath57 can be achieved using dglap fits , even though the very application of dglap at such low @xmath3 is questionable . on the other hand , the observables like @xmath58 which are more sensitive to the gluon distribution are poorly determined experimentally .
the proposed lhec @xcite may have the potential to disentangle different physical scenarios through a deeper exploration of the small-@xmath1 kinematic region .
the question rises of which observable is best suited to pin down the saturation of the proton gluon distribution .
a partial answer is given in fig .
[ pseudo ] , courtesy of juan rojo @xcite .
it presents a nnpdf1.2 dglap fit @xcite to pseudodata for proton f@xmath5 and f@xmath6 in the kinematic regime relevant for the lhec ( down to @xmath59 approx . ) .
the pseudodata are generated by extrapolation of the fit to f@xmath5 presented here using bk evolution , and thus include saturation effects . while @xmath0 can be well fitted by dglap ,
the best dglap fit to pseudotata underestimate @xmath58 at small @xmath3 and overshoots it at the largest @xmath3 considered , yielding a large @xmath60d.o.f .. one concludes that a precise experimental determination of @xmath58 at the lhec over a large enough @xmath3 range might suffice to pin down the kinematic region where departure between dglap and non - linear ( or , maybe , linear resummed small-@xmath1 evolution ) takes place , while @xmath0 does not offer such discrimination power . -1 cm ( left plot ) and f@xmath6 ( right plot ) generated by extrapolation of our fits down to @xmath59 for @xmath61 and @xmath62 gev@xmath23.,title="fig : " ] ( left plot ) and f@xmath6 ( right plot ) generated by extrapolation of our fits down to @xmath59 for @xmath61 and @xmath62 gev@xmath23.,title="fig : " ] |
many social , technological , and biological networks belong to a common scale - free ( sf ) structure @xcite which consists of many low degree nodes and a few high degree nodes called as hubs .
the degree distribution follows a power - law , therefore an sf network has an extreme vulnerability against hub attacks @xcite .
in addition , these real networks are classified into assotative and disassortative networks @xcite . for examples , typical social networks , e.g. coauthorships and actor collaborations ,
are assotative , while typical technological and biological networks , e.g. internet , world - wide - web , protein - interaction , and food webs , are disassortative . in assotative networks ,
nodes with similar degrees tend to be connected , and thus positive degree - degree correlations appear . in disassotative networks
, nodes with different degrees : low and high degrees tend to be connected , and thus negative degree - degree correlations appear . through the above findings in network science , superior network theories and efficient algorithms have been developed for analyzing network topology and dynamics @xcite .
however , studies for the cases with degree - degree correlations are not clear enough for successful analysises of topological structures and epidemics on networks except some percolation analysises . recently
, it has been numerically and theoretically found that an onion - like structure with positive degree - degree correlations gives the nearly optimal robustness against hub attacks in an sf network @xcite .
on the other hand , the average behavior of stochastically generated network models or empirical data samples of real networks is discussed in many applications .
usually , some characteristic quantities such as degree distribution or clustering coefficient are investigated for a network , then their quantities are averaged over the samples of networks in which the existence of a generation rule ( mechanism ) of the networks is assumed . in this paper , we focus on the beforehand averaged network structure over samples , and calculate the degree distribution for several models of growing network with or without degree - degree correlations . this representation will give a general framework for numerically investigating the characteristic topological quantities in growing networks . since our framework is supported by the interesting property of ordering that older nodes tend to have higher degrees in a randomly growing network @xcite , a wide range of application to growing networks can be expected .
we consider a set of growing networks in which a new node is added with probabilistic links to existing nodes in a network at each time step .
to study the average behavior of the stochastic processes in many samples , we use the ensemble average of adjacency matrix defined as follows .
@xmath0 here , without loss of generality , we set connected two nodes as the minimum initial configuration : @xmath1 , @xmath2 , and the degree @xmath3
. note that at each time step the matrix is expanding with the inverse @xmath4 shape of elements @xmath5 , @xmath6 , @xmath7 at the right - bottom corner .
the diagonal element @xmath8 is always @xmath6 due to no self - loop at each node @xmath9 .
other elements are @xmath10 , @xmath11 , as the average number of links from @xmath9 to @xmath12 over the samples of networks .
the value of each element is defined in order according to the passage of time .
we assume that there are no adding links between existing nodes @xmath13 at any time @xmath14 .
only links between selected old nodes and new node @xmath15 are added in a growing network . in the sample - based description ,
the ensemble average of adjacency matrix is @xmath16 where @xmath17 denotes an adjacency matrix of the @xmath18-th sample whose elements are @xmath19 or @xmath6 corresponded to the connection or disconnection from @xmath9 to @xmath12 , but @xmath20 or undefined for @xmath21 because of the size @xmath22 at time @xmath23 .
@xmath24 denotes the number of samples .
we remark that an adjacency matrix is fundamental and important because it includes the necessary and sufficient information about a network structure and is useful for a mathematical treatment . in this explicit representation of general framework , for each @xmath9-th node , @xmath25 , the ( out-)degree is updated from time @xmath26 to @xmath23 , @xmath27 the ( out-)degree of @xmath15-th node added at time @xmath14 is defined by the sum of links from node @xmath15 to nodes @xmath9 , @xmath28 we should remark that the iterative calculations of eqs .
( [ eq_de_general_ki])([eq_de_general_kn ] ) are equivalent to the averaged values over the samples after calculating the degrees for each sample of the networks at time @xmath23 . with this equivalence in mind , we investigate the asymptotic behavior of @xmath29 for a large @xmath23 .
we note that @xmath29 is a monotone non - decreasing function of time @xmath23 because of @xmath30 from eq .
( [ eq_de_general_ki ] ) in growing networks .
as examples , we apply the ensemble average of adjacency matrix to some network models . however , our approach is applicable to other growing networks especially in a wide class , e.g. with approximately power - law or exponential degree distribution . in the following , we assume that each link is undirected : @xmath31 . since the continuous - time approximation of eq.([eq_de_general_ki ] )
is generally @xmath32 our approach is applicable to the babarsi and albert ( ba ) model @xcite as follow .
preferential attachment : @xmath33 uniform attachment : @xmath34 where @xmath35 denotes the initial number of nodes , and @xmath36 denotes the number of adding links at each time step .
for the two cases of preferential and uniform attachments , @xmath37 and @xmath38 have been derived with the corresponding degree distributions @xmath39 and @xmath40 , respectively @xcite . the analysis in ba model
is based on the invariant ordering property of degrees in which older nodes get more links averagely at any time in the growth of network . under the invariant ordering property ,
our approach can be regarded as an extension of mathematical treatment in the ba model through the representation by the ensemble average of adjacency matrix over samples of growing networks .
we preliminary introduce a duplication - divergence ( d - d ) model @xcite without mutations of random links between existing nodes , whose generation mechanism is known as fundamental in biological protein - protein interaction networks . in the d - d model , at each time step , a new node is added and links to neighbor nodes of a uniformly randomly chosen node ( see figure [ fig_basic_process](a ) ) .
some duplication links are deleted with probability @xmath41 . here ,
no mutations are to simplify the discussion and to connect to the next subsection .
although the degree distribution in the d - d model can be approximately analyzed by the approach of mean - field - rate equation @xcite , we show the applicability of our approach to the d - d model in order to extend it to more general networks .
moreover , in the next section , we reveal that older nodes tend to have higher degrees @xcite in the d - d model , whose ordering of degree @xmath29 for node index @xmath9 was not found from the above approach .
\(a ) d - d model \(b ) copying model since the @xmath15-th new node links to the neighbor node @xmath9 of a chosen node @xmath42 from existing nodes @xmath43 in a network of d - d model , we have @xmath44 where we use the uniform selection probability @xmath45 of each node @xmath12 and the no - deletion rate @xmath46 for linking to the neighbor nodes . from eqs .
( [ eq_de_general_ki])([eq_de_general_kn ] ) ( [ eq_a_ni_d - d ] ) , we obtain @xmath47 @xmath48 by applying eq .
( [ eq_de_d - d_ki ] ) recursively , we derive @xmath49 \(a ) @xmath50 \(b ) @xmath51
\(c ) @xmath52 in addition , the continuous - time approximation of eq .
( [ eq_de_d - d_ki ] ) is @xmath53 from the separation of variables method , we obtain the solution @xmath54 when we denote the initial degree @xmath55 at the inserted time @xmath56 for a node @xmath9 , the above solution is rewritten as @xmath57 from the existence of parallel curves shown in fig .
[ fig_para_d - d ] , the ordering of degrees @xmath58 is not changed . in other word
, older nodes get more links averagely .
thus , we obtain @xmath59 @xmath60 where @xmath61 is the number of nodes at time @xmath23 , and @xmath62 denotes the initial number of nodes .
in the tail of degree distribution , the exponent of power - law is @xmath63 asymptotically .
note that the slightly different exponent @xmath63 to the conventional approximation @xcite is not strange , since the mutations are necessary for their d - d models .
figure [ fig_d - d](a)-(c ) shows the time - course of @xmath64 in the case of @xmath65 , @xmath66 , and @xmath67 , respectively , averaged over @xmath68 samples .
the black , orange , and magenta lines are the numerical results of eqs .
( [ eq_de_d - d_ki])([eq_de_d - d_kn ] ) for the node @xmath69 , @xmath70 , and @xmath68 .
the cyan line guides the estimated slope of @xmath46 in log - log plot . in fig .
[ fig_d - d](d ) , the red , green , and blue lines show the degree distributions for @xmath65 , @xmath66 , and @xmath67 , respectively , at the size @xmath71 .
the magenta , cyan , and gray dashed lines guides the corresponding slopes of @xmath63 for these @xmath72 .
\(a ) @xmath50 \(b ) @xmath51
\(c ) @xmath52 \(d ) @xmath73 a modification @xcite of the d - d model @xcite by adding a mutual link between a new node and a randomly chosen node has been proposed .
the mutual link contributes to avoid the singularity called as non - self - averaging even for @xmath74 without mutations @xcite .
the growing network is constructed as shown in fig . [ fig_basic_process](b ) .
it is referred to as copying model . at each time step , a new node is added .
the new node links to a uniformly randomly chosen node , and to its neighbor nodes with probability @xmath46 . in the copying model ,
we have @xmath75 since the @xmath15-th new node links to a uniformly randomly chosen node @xmath9 and to the neighbor node @xmath9 when other node @xmath12 is chosen from existing nodes @xmath43 in the network .
these effects are in the first term @xmath45 and the second term @xmath76 in eq .
( [ eq_a_ni_copying ] ) . from eqs .
( [ eq_de_general_ki])([eq_de_general_kn ] ) ( [ eq_a_ni_copying ] ) , we obtain @xmath77 @xmath78 by applying eq .
( [ eq_de_mut_ki ] ) recursively , we derive @xmath79 where we use @xmath80 and the staring formula @xmath81 . in particular , by the mathematical induction , we confirm that the case of @xmath82 generates a sequence of the complete graphs with @xmath83 links at every node of @xmath84 .
first , @xmath85 and @xmath86 are obvious .
next , we assume @xmath87 , from eqs.([eq_de_mut_ki])([eq_de_mut_kn ] ) we derive @xmath88 @xmath89 \(a ) @xmath50
\(b ) @xmath51 \(c ) @xmath52 on the other hand , the continuous - time approximation of eq.([eq_de_mut_ki ] ) is @xmath90 since this form is a 1st order linear differential equation @xmath91 , by applying the solution @xmath92 , we obtain @xmath93 where @xmath94 and @xmath95 are constants of integration , and we use @xmath96 and @xmath97 .
note that the solution @xmath98 is only different by @xmath99 to eq.([eq_de_sol ] ) , and can be ignored for a large @xmath23 . as similar to the d - d model in subsection 3.2 , from eq .
( [ eq_approx_pk ] ) under the invariant ordering ( [ eq_ordering ] ) in the parallel curves shown as fig .
[ fig_para_copying ] , the degree distribution asymptotically follows a power - law with the exponent @xmath63 .
figure [ fig_copying](a)-(c ) shows the time - course of @xmath64 in the cases of @xmath65 , @xmath66 , and @xmath67 , respectively , averaged over @xmath68 samples .
the black , orange , and magenta lines are the numerical results of eqs .
( [ eq_de_mut_ki])([eq_de_mut_kn ] ) for the node @xmath69 , @xmath70 , and @xmath68 .
the cyan line guides the estimated slope of @xmath46 in log - log plot .
it fits to the lines of @xmath29 for a large @xmath23 .
moreover , as shown in fig .
[ fig_copying](d ) , eq .
( [ eq_approx_pk ] ) gives a good approximation at the size @xmath71 .
the red , green , and blue lines show the degree distributions for @xmath65 , @xmath66 , and @xmath67 , respectively .
the magenta , cyan , and gray dashed lines guides the corresponding slopes of @xmath63 for these @xmath72 in the fitting to the tails .
\(a ) @xmath50 \(b ) @xmath51 \(c ) @xmath52 \(d ) @xmath73 in this subsection , we emphasize that our approach is effective through the numerical estimation , even when an analytic derivation is intractable .
we consider a copying model with positive degree - degree correlations based on a cooperative generation mechanism by linking homophily , in which densely connected cores among high degree nodes emerge @xcite . in more detail ,
the difference to the previously mentioned copying model is that the @xmath15-th new node links to the neighbor nodes @xmath12 of a randomly chosen node @xmath9 with a probability @xmath100 from existing nodes in the network .
such a function @xmath101 is necessary to enhance the degree - degree correlations , and @xmath102 is a parameter @xcite . since the degree of new node is unknown in advance due to the stochastic process , it is temporary set as @xmath103 .
thus , instead of eq .
( [ eq_a_ni_copying ] ) , we substitute @xmath104 for eqs .
( [ eq_de_general_ki])([eq_de_general_kn ] ) .
\(a ) @xmath50 \(b ) @xmath51
\(c ) @xmath52 although the theoretical analysis of eqs .
( [ eq_de_general_ki])([eq_de_general_kn ] ) for eq .
( [ eq_a_ni_correl ] ) is difficult , the iterative calculations are possible numerically .
when we assume @xmath105 , we derive an exponential distribution as follows . from @xmath106
, we obtain @xmath107 @xmath108 @xmath109 under the invariant ordering ( [ eq_ordering ] ) in the parallel curves shown as fig .
[ fig_para_correl ] .
figure [ fig_correl](a)-(c ) shows that @xmath29 denoted by black , orange , and magenta lines for @xmath69 , @xmath70 , and @xmath68 is approximated by @xmath110 in the copying model with degree - degree correlations .
the cyan lines guide the estimated slopes @xmath111 , @xmath112 , and @xmath113 for @xmath65 , @xmath66 , and @xmath67 , respectively , in the numerical fittings for the iterative calculations of eqs .
( [ eq_de_general_ki])([eq_de_general_kn])([eq_a_ni_correl ] ) .
thus , as shown in fig .
[ fig_correl](d ) , the tails in @xmath73 denoted by red , green , and blue lines for @xmath65 , @xmath66 , and @xmath67 are approximated by @xmath114 shown as magenta , cyan , and gray dashed lines at the size @xmath115 .
note that @xmath73 is only slightly deviated but the exponential part is remained by adding shortcut links in order to self - organize a robust onion - like structure in the incrementally growing network @xcite .
\(a ) @xmath50 \(b ) @xmath51 \(c ) @xmath52
\(d ) @xmath73 we consider the asymptotic behavior of the node degrees for a general case in growing networks .
when the time - course @xmath29 of degree follows a monotone increasing function @xmath116 of time @xmath23 , there exists the inverse function @xmath117 .
it is possible that the time - course @xmath29 is an average of observed real data for node @xmath9 , e.g. born at time @xmath118 .
from @xmath119 we have @xmath120 then , on the assumption of the invariant ordering ( [ eq_ordering ] ) in parallel curves of @xmath121 , we derive @xmath122 where @xmath123 denote the derivative of @xmath124 by the variable @xmath18
. we should remark that various degree distributions of non - power - law may appear depending on the shape of monotone increasing function @xmath29 according to what type of generation in growing networks .
we derive the invariant ordering ( [ eq_ordering ] ) at any time @xmath23 within parallel curves of monotone increasing functions for the d - d and copying models discussed in subsections 3.2 and 3.3 . in the following ,
we use double mathematical induction for node index @xmath9 and time @xmath23 .
once @xmath125 is satisfied for @xmath126 or @xmath127 at @xmath128 , @xmath129 is obtained from eqs .
( [ eq_de_general_ki])([eq_de_d - d_ki])([eq_de_mut_ki ] ) . here , from eqs .
( [ eq_de_d - d_ki])([eq_de_d - d_kn ] ) at @xmath130 with the initial condition @xmath3 , we obtain @xmath131 from eqs .
( [ eq_de_mut_ki])([eq_de_mut_kn ] ) at @xmath130 with the same initial condition , we also obtain @xmath132 since the assumption in eq .
( [ eq_assump ] ) is satisfied for @xmath133 or @xmath134 and @xmath135 at @xmath136 , we have @xmath137 by applying eq . ( [ eq_assump ] ) recursively for @xmath138 . on the other hand , from eqs .
( [ eq_de_d - d_ki])([eq_de_d - d_kn ] ) , we rewrite @xmath139 @xmath140 @xmath141 at @xmath14 , and @xmath142 for @xmath143 .
then , we have @xmath144 similarly , from eqs .
( [ eq_de_mut_ki])([eq_de_mut_kn ] ) , we rewrite @xmath145 @xmath146 @xmath147 at @xmath14 , and @xmath148 for @xmath143 . then , we also have @xmath149 by applying ( [ eq_assump ] ) recursively for @xmath150 after substituting eq .
( [ eq_delta_kn1 ] ) or ( [ eq_delta_kn2 ] ) to the right - hand side of eq .
( [ eq_assump ] ) , we obtain @xmath151 from eqs .
( [ eq_induct1])([eq_induct2 ] ) , we obtain the ordering ( [ eq_ordering ] ) after all .
next , we consider the existing condition of the ordering ( [ eq_ordering ] ) within parallel curves of @xmath121 for a general case of growing networks discussed in subsection 3.5 . from eq .
( [ eq_de_general_ki ] ) for @xmath126 or @xmath127 at @xmath152 , we have @xmath153 if older nodes tend to get more links , @xmath154 hold in the ensemble average of adjacency matrix . then , we remark @xmath155 from eq .
( [ eq_cond_general ] ) and @xmath156 or @xmath157 with the initial values @xmath158 and @xmath159 . on the other hand , from eqs .
( [ eq_de_general_ki])([eq_de_general_kn ] ) , we derive @xmath160 if eq .
( [ eq_cond_general ] ) hold with @xmath31 , for @xmath161 we have @xmath162 therefore , on the condition ( [ eq_cond_general ] ) , we obtain the ordering ( [ eq_ordering ] ) from eqs .
( [ eq_general_ij])([eq_general_n ] ) .
we have proposed the explicit representation by the ensemble average of adjacency matrix over samples of growing networks .
the important point is that the adjacency matrix is averaged in advance before calculating a characteristic quantity about the topological structure for each sample .
the ensemble average has been applied to some network models : ba @xcite , d - d @xcite , and copying @xcite models for investigating the degree distributions in the asymptotic behavior by using the theoretical and numerical analysises for difference equations and the corresponding continuous - time approximation of differential equations with variables @xmath23 and @xmath29 .
we have derived @xmath64 and @xmath163 for the d - d and copying models under the invariant ordering of degrees which is supported in randomly grown networks @xcite . moreover , for the copying model with positive degree - degree correlations , we have shown that the numerical calculations of the difference equation give a good approximation of an estimated exponential distribution , even when an analytic derivation is intractable . the copying model with positive degree - degree correlations
is related to the self - organization @xcite of robust onion - like networks @xcite
. our approach may be also applicable to data analysis for social networks , when the observed time - course of degree is a monotone increasing function like power - law or logarithm in the average over samples by ignoring short - time fluctuations , and a node index @xmath9 or @xmath12 represents an ordering of its birth time .
this expectation is supported as follows .
it is helpful for grasping a trend to study the average behavior of many users ( nodes ) added at a same ( sampling interval ) time into a network community .
random growth in a social network probably corresponds to encountered chances among people .
moreover , as time goes by , the number of his / her friends for a member of social networks is usually increasing . it is natural that the connections to friends are maintained .
however , we must consider the effect of rewirings between old members in the definition of adjacency matrix . also from a practical viewpoint ,
@xmath164 space to store an adjacency matrix may cause a problem for dig data . from the conventional analysis for special network models to a general framework , the representation by the ensemble average will open a door for investigating the characteristic quantities , e.g. node degrees in growing networks .
in particular , the time - course of a quantity depending on the birth time of node is considered as a key point .
the discussion about other quantities such as clustering coefficient or the number of paths of a given length requires further studies of how to analyze the average behavior related to the transitivity .
the author would like to thank anonymous reviewers for their valuable comments .
this research is supported in part by a grant - in - aid for scientific research in japan , no .
25330100 .
62 albert , r. , jeong , h. and babarsi , a .-
l . error and attack tolerance of complex networks .
_ nature _ 406 : pp.3644 , ( 2000 ) .
babarsi , a .-
albert , r. emergence of scaling in random networks .
_ science _ 286 : pp.509512 , ( 1999 ) .
babarsi , a .-
l . , albert , r. and jeong , h. mean - field theory for scale - free random networks . _ physica a _ 272 : pp.173187 , ( 1999 ) .
callaway , d.s . , hopcroft , j.e . , kleinberg , j.m . , newman , m.e.j . and strogatz , s.h .
are randomly grown graphs really random ? .
_ physical review e _ 64 : pp.041902 , ( 2001 ) .
hayashi , y. growing self - organized design of efficient and robust complex networks .
_ ieee xplore digital library _ http://dx.doi.org/10.1109/saso2014.17 , proc . of 2014 ieee 8th int
on saso : self - adaptive and self - organizing systems 2014 , pp.5059 .
arxiv : physics/1411.7719 , ( 2014 ) .
herrmann , h.j . ,
schneider , c.m . , moreira , a.a .
, andrade jr .
j.s . , and havlin , s. onion - like network topology enhances robustness against malicious attacks .
_ journal of statistical mechanics _
p01027 , ( 2011 ) .
kim , j. , krapivsky , p.l . and redner , s. infinite - order percolation and giant fluctuations in a protein interaction networks .
_ physical review e _ 66 : pp.055101(r ) , ( 2002 ) .
newman , m.e.j .
assortative mixing in networks .
_ physical review letters _ 89(20 ) : pp.208701 , ( 2003 ) .
newman , m.e.j .
networks -an introduction .
oxford university press , ( 2010 ) .
pastor - satorras , r. , smith , e. and sole , r.v . evolving protein interaction networks through gene duplication . _
journal of theoretical biology _
222(2 ) : pp.199210 , ( 2003 ) .
schneider , c.m . , moreira , a.a .
, andrade jr .
havlin , s. and herrmann , h.j . mitigation of malicious attacks on networks .
_ proceedings of the national academy of sciences of the united states of america _
810(10 ) : pp.38383841 , ( 2011 ) .
sole , r.v .
, pastor - satorras , r. , smith , e. and kepler , t.b . a model of large - scale proteome evolution .
_ advances in complex systems _
5(1 ) : pp.4354 , ( 2002 ) .
tanizawa , t. , havlin , s. and stanley , h.e .
robustness of onionlike correlated networks against targeted attacks .
_ physical review e _ 85 : pp.046109 , ( 2012 ) .
wu , z .- x . and holme , p. onion structure and network robustness .
_ physical review e _ 81 : pp.026116 , ( 2011 ) . | various important and useful quantities or measures that characterize the topological network structure are usually investigated for a network , then they are averaged over the samples . in this paper
, we propose an explicit representation by the beforehand averaged adjacency matrix over samples of growing networks as a new general framework for investigating the characteristic quantities .
it is applied to some network models , and shows a good approximation of degree distribution asymptotically .
in particular , our approach will be applicable through the numerical calculations instead of intractable theoretical analysises , when the time - course of degree is a monotone increasing function like power - law or logarithm . |
low excitation forbidden lines of [ o i ] and [ s ii ] are some of the defining spectroscopic characteristics of the low - mass , pre - main sequence stars known as t tauri stars ( tts , * ? ? ?
their broad , blueshifted emission profiles were originally interpreted as arising in winds with receding flows occulted by the circumstellar disk @xcite , and the correlation between their luminosity and the luminosity of infrared emission from the disk demonstrated that the forbidden line emission was powered by accretion @xcite . studies of large samples of tts conducted by @xcite , @xcite ( hereafter heg ) , and @xcite showed that forbidden lines in these stars are characterized by two distinct components : a high velocity component ( hvc ) and a low velocity component ( lvc ) . the hvc forbidden emission , typically blueshifted by 30 to 150 km / s , was demonstrated to be formed in microjets , small - scale analogs to the parsec - long collimated jets emerging from more embedded class i young stellar objects ( ysos ) @xcite .
a correlation between mass loss rates , derived from the luminosity of the hvc in the strongest [ o i ] 6300 forbidden line and the accretion luminosity and/or disk accretion rate has been the foundation of the accretion - outflow connection in young accreting stars ( heg , @xcite ) .
the origin of the outflows traced by the hvc is not yet fully understood , but is likely tied to mass and angular momentum loss in the accretion disk and/or the accreting star through magnetized magnetohydrodynamic ( mhd ) winds @xcite .
the origin of the lvc forbidden emission is even less well understood .
heg found the lvc to be present in all tts with near infrared ( nir ) excess at k - l ( class ii sources , e.g. @xcite ) , typically with small blueshifts @xmath1 5 km / s .
they considered the possibility the lvc might arise in a slow wind on the surface of a disk in keplerian rotation with the lvc surface brightness decreasing as @xmath2 .
this possibility was investigated more thoroughly by @xcite who used line luminosities and line ratios to evaluate the physical conditions in the wind , and estimated disk wind mass loss rates as @xmath3/yr .
the possibility that the lvc emission might trace thermally driven disk winds powered by photoevaporative heating due to extreme ultraviolet ( euv ) radiation from the central accreting star was investigated by @xcite .
however , the euv heated flows produced very little neutral oxygen and thus could not account for the lvc observations .
more recently , the growing realization that photoevaporative flows are likely an important means of disk dispersal during the era of planet formation has led to renewed interest in additional sources of heating for photoevaporation , i.e. x - ray and far ultraviolet ( fuv ) radiation .
these have the potential of producing significant amounts of neutral oxygen in the flow and accounting for the lvc forbidden emission @xcite .
another new development is the acquisition of high resolution spectra of [ ne ii ] at 12.8 . to date , 24 high - resolution profiles of [ ne ii ] in tts have been acquired , showing , like the optical [ o i ] profiles , a mixture of high velocity and low velocity components . low velocity components of [ ne ii ] , blueshifted from -2 to -12 km / s , are seen in 13/24 of the high resolution spectra , and have been interpreted as direct tracers of photoevaporative flows ( see the recent review by @xcite and references therein ) . although the ionization required to produce substantial [ ne ii ] emission in a photoevaporative flow could arise either from euv or x - ray heating , @xcite compared the measured [ ne ii ] luminosities with upper limits on the euv radiation reaching the disk and demonstrated that , at least in three systems , this emission probes the x - ray rather than euv - ionized surface . due to the fact that x - rays penetrate deep in the disk and drive flows that are mostly neutral , blueshifted [ ne ii ] emission signals mass loss rates that are considerably higher than if the heating is from euv , possibly as high as @xmath4m@xmath5/yr . such mass loss rates are comparable to those estimated by @xcite from the lvc optical lines .
if these mass loss rates are characteristic of photoevaporative flows , their similarity to tts disk accretion rates ( e.g. @xcite ) , suggest that photoevaporation may play a major role early in the evolution and dispersal of protoplanetary material .
photoevaporation can drastically change the disk surface density by opening gaps in planet - forming regions that widen with time , thus setting the timescale over which [ giant ] planet formation must occur .
the implication this would have on the final architecture of planetary systems , the chemistry of the disk , planet interactions ( such as the delivery of volatiles to planets located in the inner solar system ) , and the final mass and location of giant planets specifically , would be critical to enhancing our understanding of planet formation more generally . a better understanding of the empirical properties of the forbidden lvc in tts is needed to assess whether it arises in photoevaporative flows .
two recent works , @xcite , ( hereafter r13 ) and @xcite , ( hereafter n14 ) have begun this process and provide the motivation for the present study .
r13 introduced two changes in interpreting the lvc .
firstly , r13 revisited the forbidden line data from heg with modern estimates for the accretion luminosity .
@xcite demonstrated that the mass accretion rates in heg were too large by nearly an order of magnitude due to uncertain bolometric corrections to the continuum excess measured in the r band , and from the simplistic assumption that the accretion luminosity was produced in a boundary layer at the stellar surface .
r13 re - derived the accretion luminosity for 30 of the heg stars from h@xmath6 emission lines , observed simultaneously with the forbidden lines , which appeared in @xcite .
they found correlations between the lvc luminosity from heg with the improved accretion luminosities as well as with published values for fuv and stellar luminosities , but not with literature values for the x - ray luminosity .
this led to the suggestion that _ if _ the lvc is from a photoevaporative flow , fuv heating may be more important than x - ray radiation in generating [ o i ] emission , although if [ ne ii ] emission were also present then x - rays would need to be a contributor as well .
a further implication would be that photoevaporative flows might be prevalent throughout the t tauri phase , decreasing in proportion to the disk accretion rate . secondly , r13 demonstrated , with a small set of new high resolution spectra , that the lvc itself may have two kinematic components .
focusing on only two [ o i ] lvc that were well separated from any hvc emission , the profiles were decomposed into a broad and a narrow contribution , with the suggestion that the broad feature may be formed in bound material in the disk and rotationally broadened , while the narrow feature may be associated with material from further out in the disk , possibly in a photoevaporative flow .
the study of n14 provided further refinement of the definition of the lvc for 44 tts in lupus and @xmath7 ori . in heg
the lvc was defined very simply , assigning any emission within 60 km / s of line center to the lvc and outside of that to the hvc , using profiles from echelle spectra with a velocity resolution of 12 km / s .
n14 improved on this by applying gaussian fitting techniques to separate hvc from lvc emission , although their low spectral resolution of 35 km / s meant that a clear separation between these two components was not always possible , nor could they resolve the lvc into a broad and narrow component as done in r13 .
nevertheless , they found good correlations between the total lvc luminosity and both the stellar luminosity and the accretion luminosity , but not the x - ray luminosity .
they interpret the lvc as arising in a slow wind ( @xmath8 km / s ) that is dense ( n@xmath9@xmath10 ) , warm ( t @xmath11k ) , and mostly neutral . in this paper
, we continue the investigation of the empirical properties of the lvc forbidden lines in tts in order to elucidate their origin .
while the possibility that they may provide a direct tracer of photoevaporative flows that are responsible for protoplanetary disk clearing is intriguing , other potential contributors are the base of mhd centrifugal winds and heated , bound gas in the disk itself .
our study differs from previous ones in two major ways .
first , our sample of tts spans a larger range of evolutionary stages when compared to the sample of heg ( sects .
[ sect : sample ] and [ sect : accr ] ) . secondly , using keck / hires we reach a spectral resolution that is more than five times higher than n14 and about two times higher than heg ( see section [ sect : obs ] ) .
this high spectral resolution enables us to define the kinematic structure of the lvc , where about half the detected sources show two kinematic components and the remaining lvc are separated into either broad or narrow profiles ( sects .
[ sect : profiles ] and [ sect : res_lvc ] ) .
we discuss the relation of these components to mhd and photoevaporative winds in section [ sect : discussion ] .
for this project , we focus on the kinematic properties of several forbidden lines in t tauri stars : [ o i ] 6300.304 , [ o i ] 5577.339 , and [ s ii ] 6730.810
. we also searched for [ o ii ] 7329.670 but did not detect it in any source .
in addition , we use h@xmath6 as a tracer of the accretion luminosity and disk accretion rate @xcite . in the following subsections
we describe our sample ( [ sect : sample ] ) , the observations ( [ sect : obs ] ) , the methodology to create forbidden line profiles free from telluric and photospheric absorption ( [ sect : line_profiles ] ) , and the evaluation of line equivalent widths and upper limits ( [ sect : ews ] ) .
all of our science targets , with the exception of tw hya , belong to the well - characterized star - forming region of taurus - auriga ( age @xmath11 myr and average distance 140 pc ; @xcite ) .
our sample represents the spread of disk properties and disk accretion rates , but not the statistical distribution of mass , age , or other properties in this star formation region .
the required high s / n precluded observing late m dwarfs and brown dwarfs , and disk inclination was not one of the criteria used to select our targets .
our sample is presented in table [ tab : source_properties ] and consists of 33 mostly single and bright ( b@xmath1216 ) t tauri stars ( tts ) with disk spectral energy distributions encompassing : 26 class ii , full disks with optically thick inner regions ; 5 transition disks , with absent or low nir to mid - infrared ( mir ) excess emission from the inner disk but large far - infrared ( fir ) emission ; and , 2 class ii / iii evolved disks with weak nir to fir excess . additional information on this sample can be found in @xcite , who analyzed the na d and k profiles ( at 5889.95 and 7698.96 , respectively ) from the same spectra presented here .
the spectral types ( spty ) , extinctions ( @xmath13 ) , stellar bolometric luminosities ( @xmath14 ) , and stellar masses ( @xmath15 ) in table [ tab : source_properties ] are all taken from @xcite who derived them in a homogeneous way from spectrophotometric data , while taking into account excess continuum emission ( veiling ) and extinction when deriving the stellar properties .
is taken from @xcite . ]
table [ tab : source_properties ] also gives disk inclinations for 22 stars taken from the literature based on spatially resolved disk images , shown by @xcite to be the most reliable means of determining system orientations .
for most sources uncertainties are reported to be @xmath1 10% . however , for one source , dr tau , nominally reliable techniques for estimating inclination range from close to face - on to almost edge - on and suggest that the inner and outer disk ( @xmath1610au ) have different orientations @xcite . resolved millimeter continuum images point to a highly inclined outer disk ( @xmath1@xmath17 , @xcite and @xmath1@xmath18 , @xcite ) .
however , mir interferometric visibilities coupled with spectral energy distribution ( sed ) fitting , as well as modeling of the spectroastrometic signal in the co ro - vibrational line , suggest a much smaller inclination for the inner disk ( @xmath19 , @xcite and @xmath20 , @xcite ) . additional evidence for a small inclination to the inner disk comes from exceptionally deep and broad sub - continuum blueshifted absorption at both he i 10830 and h@xmath6 @xcite , requiring absorption along a line of sight through a wind that emerges radially from the stellar poles . from these considerations ,
we adopt a disk inclination of @xmath21 for dr tau , since , as will be shown in section [ sect : inclination ] , the lvc forbidden line emission is likely to arise within just @xmath15au of the star . for an additional 10 targets with no reliable inclination reported in the literature , we calculate potential disk inclinations based on forbidden line widths , as described in section [
sect : inclination ] and noted in table [ tab : source_properties ] with a dagger .
we observed all targets using the keck / hires spectrograph @xcite with the c5 decker and a 1.1x 7slit , which covers a wavelength range of 4,800 - 9,000 at a nominal resolution of 37,500 .
@xcite independently calculated the spectral resolution achieved by hires in this setting and found a slightly better resolution of @xmath145,000 corresponding to 6.6 km / s .
the targets were observed in two campaigns with the same instrumental setting , one in 2006 and another in 2012 .
two of the targets , ux tau a and ip tau , were observed in both campaigns .
in addition to the science targets , we observed 5 late type stars that are used as photospheric standards and a set of o / b type stars which are used as telluric standards .
spectra were acquired in the standard mode which places the slit along the parallactic angle to minimize slit losses .
slit position angles with respect to disk position angles are discussed in appendix [ pa_appendix ] and the table therein .
the data reduction was carried out using the highly automated mauna kea echelle extraction ( makee ) pipeline written by tom barlow .
in addition to bias - subtraction , flat - fielding , spectral extraction and wavelength calibration , the pipeline automatically subtracts the sky .
an example of a spectrum before and after sky subtraction is discussed in appendix [ pa_appendix ] and the figure therein .
further details about the data reduction and source exposure times are given in @xcite . in order to identify weak intrinsic emission in the forbidden lines
we first remove any telluric and/or photospheric absorption contaminating the spectral order of interest .
telluric absorption is prevalent in the orders containing [ o i ] 6300 and [ o ii ] 7330 , but minimal for [ o i ] 5577 and [ s ii ] 6731 .
we remove the atmospheric features by matching the telluric lines in an o / b standard star to those in the target spectrum and then dividing the target spectrum by the telluric standard . in order to remove photospheric lines ,
we follow a procedure motivated by the approach of @xcite .
first , we select a photospheric standard with a spectral type that closely matches the spectral type of the target star . if need be , we broaden the absorption lines of the photospheric standard to match the width of the lines present in the target spectrum .
we then apply a cross - correlation technique to shift the photospheric standard to align with the photospheric lines in the target spectrum , and if necessary , we add a flat continuum to the photospheric standard to match veiled photospheric lines in the young accreting stars .
the veiling ( @xmath22 ) is defined as the ratio of the excess to the photospheric flux , and @xmath23 is included in table [ tab : oi_5577_6300 ] , with values ranging from 5.6 ( dr tau ) to 0 ( 5 sources ) .
the final corrected profile is created by dividing the target spectrum by the veiled photospheric spectrum of the standard .
this method for correcting the line profiles is illustrated in figure [ fig : resid_example ] for the [ o i ] emission features at 6300 and 5577 for two sources .
one of them , ds tau , has a moderate veiling , while the other , aa tau , has zero veiling .
no corrections are made for h@xmath6 , which has no telluric absorption and is a strong emission feature in all sources .
line profiles for two sources : aa tau and ds tau .
the original spectrum ( in black ) includes the emission feature plus telluric and photospheric absorption lines that have not yet been subtracted . once the telluric lines are removed the spectrum in red is produced .
the spectrum in blue depicts the final corrected line profile after the telluric and photospheric lines have been removed . for the [ o i ] 5577 feature ,
no telluric correction is required.,scaledwidth=50.0% ] four of the five photospheric standards are luminosity class v and one is the weak lined t tauri star ( wtts ) v819 tau , with no infrared excess from a disk . as discussed elsewhere ( see @xcite ) , wtts are ideal candidates for matching photospheric features in tts , and v819 tau was used whenever it was a good match to the spectral type of the target .
the five photospheric standards and their spectral types are : hr 8832 ( k3 ) , hbc 427 ( k6 ) , v819 tau ( k8 ) , gl 15a ( m2 ) , v1321 tau ( m2 ) .
the standard applied to each target and line of interest is included in table [ tab : oi_5577_6300 ] .
final corrected line profiles are presented in figures [ fig : hvc_6300_panel1 ] and [ fig : hvc_6300_panel2 ] for [ o i ] 6300 for all 30 stars , including the 3 non - detections ( see next section ) .
final corrected profiles for the other two forbidden lines are shown only for the detections , in figure [ fig : hvc_5577 ] for [ o i ] 5577 and in figure [ fig : hvc_sii ] for [ s ii ] 6731 .
they are plotted as corrected flux above the continuum versus radial velocity .
the velocity is relative to the stellocentric frame , as determined from photospheric line centroids .
the stellar radial velocities of our sources are given in tables 5 and 6 of @xcite and have a 1@xmath7 uncertainty of @xmath11km / s .
6731 profiles for the 8 sources with detections .
the lvc is only detected in two sources ( cw tau and hn tau ) , and is shaded in purple.,scaledwidth=50.0% ] a number of our sources have weak or absent forbidden lines .
the procedure used to identify a detection is to calculate the standard deviation of the residuals ( rms ) from a linear continuum fit to a region @xmath1 3 outside the line of interest .
any emission near the wavelength of interest , with a peak higher than 3 times the rms and equal or broader in width than the 6.6 km / s of an unresolved line , is deemed a detection . based on this method we find detections in the [ o i ] 6300 transition for 30/33 tts .
non detections are dn tau , vy tau and v710 tau , all with measured veilings very near or at zero . both detections and non - detections
are included in figures [ fig : hvc_6300_panel1 ] and [ fig : hvc_6300_panel2 ] for [ o i ] 6300 .
additionally , we detect [ o i ] 5577 for 16 stars and [ s ii ] 6731 for 8 stars .
detections are shown for these lines in figures [ fig : hvc_5577 ] and [ fig : hvc_sii ] .
the [ o ii ] transition at 7330 is never detected . for detected lines we calculate the line equivalent width ( ew ) by integrating over the wavelength range where there is emission above the continuum .
in addition to measuring the line ew , we compute its uncertainty using a monte carlo ( mc ) approach ( e.g. @xcite ) .
we do this by adding a normally distributed noise to each spectrum with the noise being the rms on the continuum next to the line of interest .
we compute the ews of 1,000 individual spectra generated by the mc approach and assign as the uncertainty the standard deviation of the distribution of ews . for lines we determine to be non - detections , we compute a 3@xmath24 upper limit ( ul ) on the ew assuming a gaussian unresolved profile , @xmath25 where the rms is calculated from the standard deviation of the residuals on a @xmath1 3 continuum , @xmath26 in is defined as @xmath27 where @xmath28 is the wavelength of the line of interest
, @xmath29 is the fwhm of an unresolved line ( 6.6 km / s ) and @xmath30 is the speed of light in km / s .
table [ tab : oi_5577_6300 ] presents the ew for the detected forbidden lines and the upper limits for the non - detections for our 33 sources .
in accreting tts , most of the uv and optical excess continuum emission derives from energy released through accretion , attributed to magnetospheric accretion shocks on the stellar surface @xcite .
ideally accretion luminosities are determined from flux calibrated spectra that include the balmer discontinuity , where the spectrum of the continuum uv - excess can be modeled with accretion shocks @xcite . since our echelle spectra
are not flux - calibrated , accretion luminosities can be estimated using well calibrated relationships relating line and accretion luminosities first demonstrated by @xcite and recently summarized in @xcite .
this approach is superior to the older technique of using the emission excess ( veiling ) at one wavelength in the paschen continuum , which requires applying an uncertain bolometric correction .
here we will use the calibration on the h@xmath6 line luminosity from @xcite because it is based on a large sample of low - mass stars and simultaneous uv - excess measurements of the accretion luminosity . with our echelle spectra , emission line equivalent widths ( @xmath31 )
can be converted into a line luminosity using the following relation : @xmath32 where @xmath33 is the distance to the science target , @xmath34 is the photospheric continuum flux density near the line of interest , and the factor ( 1 + @xmath22 ) converts the observed equivalent width to one that is veiling corrected , i.e. measured in units of the stellar continuum . for this work , we chose h@xmath6 for an accretion diagnostic since it is detected in all of our targets .
we adopt distances from @xcite , and use their published extinction corrected continuum flux density at 7510 , in conjunction with the ratio of the flux densities at 6600 and 7510 in the pickles atlas for stars matched in spectral type to each science target @xcite to determine the stellar flux density near h@xmath6 .
the veiling @xmath22 at 6300 listed in table [ tab : oi_5577_6300 ] is very close to that at h@xmath6 and was adopted here . while this approach assumes the stellar continuum has not varied , it does account for any variability in veiling between our observations and that of @xcite in setting the continuum flux adjacent to the line .
the h@xmath6 line luminosities ( l@xmath35 ) , are then converted to accretion luminosities , l@xmath36 , using the relation derived by @xcite : @xmath37 the calibration based on l@xmath35 can be compromised in sources with large blueshifted and redshifted absorption , masking what otherwise would be a larger emission ew . however , since it is observed in all our targets , we preferred it to other indicators such as he i. finally , the accretion luminosities are converted into mass accretion rates using the magnetospheric model developed by @xcite and the following equation : @xmath38 where @xmath39 and @xmath15 are the radius and mass of the star , @xmath40 is newton s gravitational constant , and @xmath41 is the inner truncation radius of the disk .
@xmath41 is generally unknown , but it is usually assumed to be @xmath42 , the co - rotation radius ( e.g. @xcite ) .
for all of the sources , we use the stellar masses and radii from @xcite .
all accretion parameters , including ew@xmath35 , @xmath43 , @xmath44 and @xmath45 are listed in table [ tab : acc_properties ] .
the table also includes columns for the line luminosity of [ o i ] 6300 , first for the whole line , then for the lvc and then for the nc of the lvc , all with @xmath34 determined from the pickles atlas near 6300 .
the definition of the latter two will be described in the next section .
we estimate a typical uncertainty for @xmath44 as follows .
sixteen of our targets have multi - epoch r - band photometry measurements @xcite where the standard deviation of the photometric points over the mean r - magnitude is at most 0.1 , which indicates that variability in the continuum is probably not a major source of uncertainty for most stars .
the extinctions derived by @xcite also have a small uncertainty , only 0.15 dex .
therefore , the major uncertainty in @xmath44 will be the calibration of @xmath43 and @xmath44 .
errors in the slope and y intercept of this relation are included in equation 2 . adding these uncertainties in quadrature gives a final uncertainty on @xmath44 of @xmath1 0.3 dex .
variations in tts h@xmath6 ew and veiling , attributed to variations in accretion luminosity , are typically less than a factor of two .
two of our targets , ux tau a and ip tau , were observed in both 2006 and 2012 with changes in h@xmath6 ew by factors 1.2 and 1.5 , respectively , with negligible changes in veiling .
this level of variability corresponds to a variation of @xmath1 0.2 dex in @xmath44 .
the range in h@xmath6 ew for our sample runs from a low of 1.2 to a high of 230 , translating into a span of more than three orders of magnitude in @xmath44 and @xmath45 , with accretion luminosities from @xmath46 to @xmath47 @xmath48 , and mass accretion rates from @xmath49 to @xmath50 @xmath51/year .
@xcite demonstrated that for @xmath44 @xmath52 there is a possibility for chromospheric emission to dominate the line luminosity .
since the continuum veiling is a diagnostic of accretion , deriving from excess emission in the accretion shock on the stellar surface , the relation between @xmath23 and @xmath44 , shown graphically in figure [ fig : lacc_veiling ] , offers additional insight on whether a star is accreting . in general these quantities
are correlated , in the sense that there is a well defined lower boundary in @xmath44 at a given veiling with considerable scatter above that boundary = 5.6 and 3.5 respectively , do not fall into the relation between veiling and l@xmath36 defined by the other stars .
exceptionally high veilings for these sources were also found by heg . either the relation between veiling and accretion luminosity breaks down at high veilings @xcite and/or the accretion luminosities based on h@xmath6 are severely underestimated due to strong wind absorption features in sources with high disk accretion rates and high veilings . ] .
for example , the objects with low veiling ( @xmath23 of 0 or 0.1 ) have inferred accretion luminosities spanning two orders of magnitude . we conclude from this that it is not possible to assign an unambiguous accreting / non - accreting status based on the level of the h@xmath6 ew or absence of veiling , sometimes used as thresholds for defining classical ( accreting ) or weak ( non - accreting ) tts .
since all of our targets have disks in various evolutionary states , we will treat them all identically in converting h@xmath6 into accretion luminosities .
we note that three sources have h@xmath6 based accretion luminosities near or below the threshold where accretion can not be distinguished from chromospheric activity ( coku tau 4 , vy tau , and v710 tau ) .
two of these three also have no detection of even the strongest forbidden line , [ o i ] 6300 .
the third non - detection at [ o i ] 6300 is dn tau , with h@xmath6 ew @xmath53 13.5 , @xmath23 @xmath53 0 , and an inferred log @xmath44 of -1.93 .
as discussed in section [ sect : heg_compare ] , this appears to the only example of a star that is accreting but does not show forbidden emission . with increasing veiling , defined by a clear lower boundary for @xmath44 at a given veiling with considerable scatter above that boundary . the extreme veiling sources dr tau and fm tau , with @xmath23 = 5.6 and 3.5 , respectively , exceed the plot boundary .
symbols with radial spokes denote sources with a high velocity component at [ o i ] 6300 .
, scaledwidth=50.0% ]
the aim of this paper is to better understand the lvc of forbidden line emission in tts . in this section
we describe the fitting technique used to define and separate hvc and lvc emission , and then describe the kinematic properties of the lvc emission .
the focus of this paper is the behavior of the lvc of the forbidden line profiles .
although in some instances the hvc and lvc are well resolved ( e.g. cw tau , do tau ) in most cases they are blended ( e.g. dk tau , hn tau ) . in the earlier heg study the separation between hvc and lvc
was made simply by assigning emission further than 60 km / s from the stellar velocity to the hvc . with almost a factor of two higher spectral resolution
, we attempt to separate blended hvc and lvc using gaussian fitting . to this end
, we fit the profiles interactively using the data analysis and visualization environment ( dave ) that runs as a graphical user interface ( gui ) in idl .
this program was developed by the national institute of standards and technology ( nist ) center for neutron research @xcite . in order to find the best fit parameters for each emission feature
, we identify the minimum number of gaussians required to describe each profile , specifying an initial estimate of the centroid velocity ( @xmath54 ) and full width at half maximum ( fwhm ) for each component using the peak analysis ( pan ) feature of dave .
pan then performs many iterations to minimize the reduced chi - squared and outputs the best fit centroid velocities , the fwhm , and the areas under the gaussian fits .
errors in centroid velocities , measured relative to the stellar photosphere , can not be less than @xmath551km / s , which is the uncertainty in the stellar radial velocity ( see @xcite for more details ) .
the number of gaussians required to fit each line profile depends on the profile s shape and an rms estimate of the goodness of fit .
the fitted components , individually and summed , are superposed on all 30 detected [ o i ] 6300 profiles in figures [ fig : full_fit1 ] and [ fig : full_fit2 ] .
eleven profiles are well fit with a single gaussian component and another ten with two gaussians .
the remaining nine profiles required 3 components ( 5 stars ) , 4 components ( 3 stars ) and in once case , fz tau , five components .
the method we adopt to attribute a component to hvc or lvc emission begins with an examination of the distribution of centroid velocities for all individual components across all stars , shown in figure [ fig : vc_hist ] .
the component centroid velocities range from -144 km / s ( dg tau ) to + 130 ( fz tau ) , with blueshifts far more common than redshifts and a high concentration at low velocities . based on this distribution
, we adopt a centroid velocity of @xmath55 30 km / s as the threshold between hvc versus lvc emission . with this method , although all detected [ o i ] lines show lvc emission , hvc emission ( highlighted in green in figures [ fig : full_fit1 ] and [ fig : full_fit2 ] ) is seen in only 13 sources at [ o i ] 6300 and only 3 sources at [ o i ] 5577 .
in contrast all 8 [ s ii ] detections show hvc emission but only 2 ( cw tau and hn tau ) show weak lvc emission .
the hvc fit parameters of centroid velocity and fwhm are listed in table [ tab : hvc_params ] for all 3 lines , the lvc fit parameters for the two [ o i ] lines in table [ tab:2_gaus_params ] and the lvc fit parameters for [ s ii ] in table [ tab : sii_properties ] .
6300 profiles .
we adopt @xmath55 30 km / s ( dashed lines ) as the threshold between hvc ( green ) and lvc ( black ) emission.,scaledwidth=50.0% ] we find this approach to be reinforced when comparing component fits for the cases when both [ o i ] lines are present .
we illustrate four examples in figure [ fig : pan ] including two of the three stars with hvc emission detected at 5577 ( cw tau and do tau ) , plus two cases where hvc emission is seen at [ o i ] 6300 but not 5577 ( aa tau and df tau ) . in all cases
we find that although the hvc components differ between [ o i ] 6300 and 5577 , their lvc component(s ) are similar .
the tendency for the lvc components to be comparable in the two [ o i ] lines , once the hvc is accounted for , gives us confidence that our method to characterize the lvc emission is generally robust .
a close comparison of the lvc in the two [ o i ] lines will follow in the next section .
however , one of the two stars with spectra from both observing epochs , ip tau , indicates an exception to the finding that lvc emission is always present .
we illustrate the change in the morphology of the [ o i ] 6300 profile between the two epochs in figure [ fig : iptau ] . in 2006
the profile is asymmetric , requiring two gaussians to describe it , but in 2012 it is symmetric and fit by one component . using our definition of hvc and lvc ,
the single component seen in 2012 is classified as hvc emission , and it is identical in velocity centroid and fwhm to one of the two components describing the 2006 profile .
in contrast the second component in 2006 meets the criteria of lvc emission . throughout this work
we will use the 2006 spectrum for ip tau so its lvc profile can be compared with the other stars .
we note that the issue of variability of the lvc will be addressed further in section 5.1 , where we find that the lvc is typically constant when compared to profiles found in the literature .
thus cases like ip tau would be interesting to monitor for further variability . lines are present , the lvc components for 6300 ( left ) and 5577 ( right ) are very similar .
the narrow component of the lvc is shaded in blue , and the broad component of the lvc is outlined in red .
areas shaded in green meet the criteria for hvc and the purple line shows the sum of all fits . , scaledwidth=50.0% ] 6300 profiles for two observing epochs for ip tau . in 2006 two gaussian components
describe the profile , but in 2012 only one is required .
the area shaded in green is the hvc , identical in both epochs .
the red line denotes a broad lvc , seen only in 2006 .
, scaledwidth=50.0% ] in this section we subdivide the lvc into two components , one broad ( bc ) and one narrow ( nc ) , based on examination of residual lvc profiles generated by subtracting the hvc component fits from the observed profile .
the motivation for this further subdivision is that although [ o i ] lvc residual profiles from 17 stars are described by one gaussian component , 13 have lvc profiles that require a composite of two gaussians , with one component significantly broader than the other .
the residual lvc [ o i ] 6300 profiles , with component fits superposed , are presented in two figures , in figure [ fig : lvc_1gaus ] for the 17 single component fits and in figure [ fig : lvc_2gaus ] for the 13 two component fits .
the combination of a broad and narrow component in the 13 composite lvc profiles is shown quantitatively by the distribution of their fwhm , presented in the upper panel of figure [ fig : histogram ] . in the figure ,
the lvc with two - component fits are highlighted with a darker shading than those with one - component fits . among the components with the composite profiles ,
the narrower component , with fwhm from 12 to 39 km / s , a median of 27 km / s and a standard deviation 9 km / s , can be contrasted with the broader component , with fwhm from 49 to 140 km / s , a median of 97 km / s and a standard deviation of 29km / s .
based on this fwhm separation between the two components , we designate nc lvc as those with fwhm @xmath56 40 km / s , and bc lvc as those with fwhm @xmath16 40 km / s , and color - code them in the figure with red for bc and blue for nc . with this subdivision , we can further classify the single - component lvc fits ( lighter shading in figure [ fig : histogram ] ) , into 12 bc lvc and 5 nc lvc . in sum , of the 30 tts with [ o i ] 6300 lvc emission , 18 have nc and 25 have bc , with 13 stars showing both components .
the lower panel of figure [ fig : histogram ] presents the distribution of centroid velocities ( all by definition less than 30 km / s ) of the nc and bc lvc .
their centroid velocities overlap , together spanning a range from -27 km / s to + 6 km / s with both groups peaking at blueshifts of a few km / s .
a k - s test between the nc and the bc centroid velocities gives a @xmath1 9% probability that they are drawn from the same parent population , in the sense that the nc and bc velocity distributions are statistically indistinguishable . however , there are a few considerations that suggest otherwise .
although they have similar average centroid velocities , -2.5 km / s for the nc and -3.7 km / s for the bc , the standard deviation for the nc is almost a factor of three smaller than for the bc , 3.6 km / s versus 9.7 km / s . we will explore possible differences in the centroid velocities of the nc and bc in the next section .
6300 lvc .
the darker shading highlights sources with 2-component fits , with narrower fwhm ( blue ) clearly separated from broader fwhm ( red ) components .
this leads to the adoption of 40 km / s as the boundary between nc and bc lvc emission .
lighter shading with the same color scheme shows the distribution of the sources with a single kinematic component .
_ lower _ : histogram of the peak centroid values for [ o i ] 6300 for the components found in the [ o i ] 6300 lvc .
again , darker shading is for sources with 2-component fits .
bin sizes are 2 km / s .
, scaledwidth=50.0% ] a comparison of the residual lvc profiles between the two [ o i ] lines can made for the 16 stars with [ o i ] 5577 , where only 3 stars ( cw tau , dg tau , do tau ) had hvc that needed to be subtracted .
this is illustrated in figure [ fig:5577_scaled_6300 ] where both residual lvc are normalized to their respective peaks so the profile structure can be compared .
we see that for most stars the lvc structure is very similar between the two lines .
this similarity is strengthened when their lvc fit parameters , listed in table [ tab:2_gaus_params ] , are compared .
for example in aa tau and bp tau the centroids and fwhm of both the bc and nc between the two lines is identical and the difference in the superposed profiles is due to a different ratio of bc to nc .
this similarity again suggests that the process of subdividing the lvc into bc and nc is robust .
there are 3 cases where there are differences in the lvc components between the two [ o i ] lines .
the most extreme case is df tau , where the [ o i ] 6300 bc lvc extends further to the blue than 5577 , with centroids of -26 km / s and -14 km / s , respectively .
for this star it is possible that there is uncorrected hvc emission at 6300 that is blended with the lvc ( see figure [ fig : pan ] ) .
reassuringly , this is the only case with such an extreme difference . for the other two cases , cw tau and do tau ,
the lvc centroid velocities are again more blueshifted at 6300 compared to 5577 but by only a few km / s .
we will discuss these differences further in the next section .
however we reiterate that for the majority of the stars the centroids and fwhm of the lvc components are essentially identical , within the errors , for both [ o i ] lines , suggesting that this approach is robust .
6300 ( black ) , [ o i ] 5577 ( purple ) and [ s ii ] 6731 ( red ) for the 4 sources that show different velocity centroids in the lvc components ( see section [ sect : lvcshifts ] ) . in the cases of cw and df tau ( top two )
the [ o i ] lvc had both a bc and a nc , but in this figure the bc is removed . when detected
the [ s ii ] emission is more blueshifted than the [ o i ] 6300 which , in turn , is more blueshifted than the [ o i ] 5577.,scaledwidth=50.0% ] although most stars with both [ o i ] lines show similar lvc kinematic properties , 4 sources , all with significant hvc emission , show lvc velocity centroids with velocity shifts among different lines , in the sense that the [ o i ] 5577 is less blueshifted than the [ o i ] 6300 , which in turn is less blueshifted than the [ s ii ] 6731 .
these small velocity offsets are illustrated in figure [ fig : s2lvc ] , where the forbidden line lvc are superposed and plotted on an expanded velocity scale . in the case of cw tau and df tau ,
the velocity differences are seen in the nc of the lvc and in do and hn tau the velocity shifts are seen in the bc of the lvc .
to best illustrate the nc shifts for cw tau and df tau , where both [ o i ] lines require a two - component fit to their lvc , in figure [ fig : s2lvc ] the bc of the lvc has been subtracted , so only the nc are shown .
the velocity centroids for the 3 nc lvc for cw tau are -5.5 km / s for the [ s ii ] , -2.63 km / s for the [ o i ] 6300 , and 0.92 km / s for the [ o i ] 5577 . for df tau they are -1.3km
/ s and 0.9km / s for the 6300 and the 5577 lines , respectively .
for the two stars with shifts in the bc , there are no nc contributions to the lvc . for do tau , the centroid velocity for [ o i ]
6300 is -25 km / s compared to -17 km / s at 5577 .
again , for hn tau , the centroid for [ s ii ] is -9.4 km / s and -1.2 km for [ o i ] 6300 , although in this source the major differences in the two lines is a more extended blue wing at [ s ii ] .
a similar effect of more blueshifted lvc in lines of lower critical density was also found by heg , suggesting that the lower critical density lines reflect acceleration in a slow wind in a few stars ( see also section [ sect : lineratios ] ) .
as shown in the previous section , lvc emission , seen in all 30 tts with detected forbidden lines , can be further subdivided into nc and bc kinematic features .
we do not find any commonality in the line profiles of the 10 multiple systems in our sample and see no trend with the companion separation ( see tables 3 and 4 in pascucci et al .
2015 for stellar separations ) . as such we will not discuss the possible effect of binaries on the lvc
instead , we will examine how the emission relates to accretion luminosities , assess the role of disk inclination in determining their fwhm , explore disk surface brightness distributions that can account for the observed profiles , and look at line ratios among the different forbidden lines .
we begin by comparing relevant aspects of our results to the previous results .
since both this study and that of heg focus on tts in taurus , there are 20 objects in common whose forbidden line properties can be examined for variability . in figure
[ fig : heg_profile_scaled ] , we overlay the [ o i ] 6300 profiles for these 20 sources , with the older profiles rescaled in peak intensity to match the current spectra , providing a comparison of the kinematic structure of the profiles over several decades .
recalling that there is almost a factor of 2 higher spectral resolution in the hires spectra ( 6.6 km / s versus 12 km / s ) , two things are apparent . _
( 1 ) the structure of the lvc is identical in most instances_. one dramatic exception is dn tau , where the lvc has disappeared between 1995 and 2006 .
the dg tau lvc profile is also different , where in 2006 there is a nc peak at -12 km / sec that is not seen in heg . in the earlier spectrum
the lowest velocity peak is at -50 km / s and does not qualify as `` low velocity '' emission .
however the red side of the profiles in both epochs are very similar , suggesting the more recent lvc peak may have been present , but much weaker , a few decades ago .
( 2 ) _ the hvc is quite different in 4 sources _ ( cw tau , df tau , dg tau , and dr tau ) , and in the case of dr tau it has vanished between 1995 and 2006 . since the hvc arises in spatially extended microjets that form time variable knots of shocked gas , such changes are not surprising , especially as differences in slit length and orientation coupled with differences in the width of the stellar point spread functions ( psf ) between the two studies can also yield different profiles for extended emission .
however , we conclude that the velocity structure of the lvc is generally stable over a timescale of decades in most stars
. we can also examine whether the strength of the lvc emission is comparable between the 2 studies .
we do this in figure [ fig : ew_veiling_heg95 ] by comparing the lvc equivalent widths normalized to the photospheric continuum , ( ew @xmath57 ( @xmath58 ) ) , with the caveats that the heg definition of the lvc was any emission within 60 km / s of the stellar velocity , in contrast to our approach of isolating kinematic components by gaussian fitting , and the difference in wavelength for which the veiling is reported , @xmath59 in heg versus @xmath23 here .
this comparison of veiling corrected ew shows comparable emission strength of the lvc for most sources , with the exception of dn tau , which , as seen in figure [ fig : heg_profile_scaled ] , has disappeared since 1995 . in the other sources ,
the differences in the veiling corrected equivalent width between the two epochs could be attributed to differences in the definition of the lvc , variation in the stellar continuum and/or variability in the line luminosity itself , or possibly to extended emission in some lvc observed with different stellar psf or slit orientations , as hinted at in @xcite .
the complete disappearance of [ o i ] 6300 emission in dn tau is surprising , making it the only source known to date that is accreting but has no [ o i ] 6300 emission .
the [ o i ] 6300 equivalent width found by heg is 0.43 while our upper limit @xmath1 0.007 , about two orders of magnitude lower than in 1995 .
the corresponding decrease in the h@xmath6 equivalent width is only @xmath1 30% ( see @xcite ) , and differences in veiling ( @xmath59 = 0.1 versus @xmath23 = 0 ) are small , consistent with the estimated uncertainty .
we checked for variability in the continuum with the @xcite catalogue .
it was rather stable at r mag @xmath1 11.5 between 1980 and 1986 , and brightened by @xmath1 0.5 mag through 1995 , the last year of recorded data .
the profile for dn tau seen in heg qualifies as solely lvc , which combined with our current two epochs for ip tau ( see figure [ fig : iptau ] ) , gives two examples where the lvc has vanished , although in most stars it appears stable over timescales of decades .
additional observations of both dn tau and ip tau would be of interest to see if their lvc returns .
another interesting comparison is the implication for the velocity of the lvc , noted to be typically blueshifted by @xmath1 5 km / s by heg . although the kinematic structure in the lvc is unchanged in the two studies ( figure [ fig : heg_profile_scaled ] ) , in the present study of lvc emission we find that only 12/25 of the bc and 9/18 of the nc are blueshifted relative to the stellar photosphere ( table [ tab:2_gaus_params ] ) .
the process of defining and separating full profiles into hvc , nc , and bc and then subtracting a hvc that is blended with a lvc ( e.g. aa tau ) and/or separating a blueshifted bc lvc from a nc lvc ( e.g. fz tau ) , means the resultant bc or nc may no longer be identified as blueshifted , although in the full profile the lvc looks to be predominantly blueward of the stellar velocity . however , we find roughly half the lvc components are blueshifted , and as seen in heg and discussed here in section [ sect : lvcshifts ] , in four stars these blueshifts show velocity gradients among different forbidden lines , with higher blueshifts for lines of lower critical density .
we note that these four stars , cw tau , df tau , do tau and hn tau , have some of the highest veiling corrected [ o i ] 6300 lvc equivalent widths , as shown in figure [ fig : ew_veiling_heg95 ] .
6300 lvc in common with this work and heg .
the definition of the lvc is different in the two studies , where heg includes any hvc emission within 60 km / s of the stellar velocity .
the dashed line is for a one - to - one correspondence.,scaledwidth=50.0% ] the two recent studies of forbidden lvc emission , r13 and n14 , found correlations between the accretion luminosity with both the luminosity of the [ o i ] 6300 lvc and the stellar luminosity , but not with the x - ray luminosity , for samples primarily from taurus and lupus .
additionally , a relation between @xmath44 and @xmath14 was found by @xcite for a large sample of objects in various star forming regions .
although the correlation in r13 was based mostly on the heg sample , with which we strongly overlap , we look here at these relations for our data , since r13 used the original 1995 assessment of the lvc luminosity in contrast to our component fitting and here we have a more consistent and reliable conversion of ew of both h@xmath6 and [ o i ] to line luminosities .
we use the astronomy survival package ( asurv ) developed by @xcite because it includes upper limits in the linear regression and correlation tests . in the lower panel of figure [ fig : lacc ] , we compare @xmath60 for both the complete lvc and only for the nc of the lvc to @xmath44 . for the former ,
the kendall @xmath61 test lends to a strong correlation , with a probability of 0.07% that the variables are uncorrelated .
( dn tau , v710 tau , and vy tau ) . ]
the best fit linear regression is : @xmath62~lvc}= 0.65(\pm 0.13 ) \times log l_{acc } - 3.84 ( \pm 0.23)\ ] ] when both luminosities are measured in @xmath63 .
this fit is the same as that in r13 and n14 .
the reason we have a larger uncertainty on the slope and intercept of our linear regression is because our targets cover a narrower range of @xmath44 ( @xmath64 between -3.7 and -0.6 ) than the samples of r13 ( @xmath64 between -3.2 and 1.8 ) and n14 ( @xmath64 from -4.8 to -0.3 ) .
the lower panel of figure [ fig : lacc ] also shows the relation between @xmath60 for the nc of the lvc and @xmath44 .
again , for the 18/33 sources for which nc lvc is detected there is only a @xmath1 2% probability that the nc lvc and accretion luminosity are uncorrelated .
of note in this figure is the fact that the nc of the lvc is found over the full range of @xmath44 , and when present , increases proportionately with the accretion luminosity .
in contrast to the correlation with @xmath60 , @xmath44 shows no correlation with @xmath14 .
although such a correlation has been found in other samples , its absence here is probably because our sample covers a factor of @xmath1100 in @xmath14 while that of @xcite which includes brown dwarfs , spans 8 orders of magnitude .
6300 , both for the full lvc ( asterisks ) and for the nc of the lvc ( blue circles ) , when detected.,scaledwidth=50.0% ] for the comparison with the x - ray luminosity , 22 of our sources have @xmath65 values as reported in table [ tab : source_properties ] . as in the earlier studies
, we find no correlation with @xmath60 of the lvc , with a kendall @xmath61 probability not low enough ( 10.8% ) to indicate that the @xmath65 and @xmath60 lvc are correlated .
( dn tau and v710 tau ) . ]
thirteen out of 30 sources with [ o i ] 6300 emission have lvc comprised of a combination of bc and nc emission .
the remaining sources show lvc that are either entirely bc ( 12/30 ) or entirely nc ( 5/30 ) .
the proportion of the lvc that is bc ( or nc ) emission is not strongly dependent on the accretion luminosity , as shown in figure [ fig : bcfraction ] .
for example , the 12/30 sources that show only bc emission cover a wide range of disk accretion rates . the same is true for the 5/30 sources that show only nc emission .
however , this figure also shows distinctive behavior for transition disk sources , all 5 of which have lvc emission dominated by the nc , and 4 of those 5 have no bc emission . combining the relations in figure [ fig : lacc ] and figure [ fig : bcfraction ] we find that while nc and bc emission are seen over a wide range of accretion luminosities , the luminosity of each component correlates with the accretion , but not the stellar , luminosity .
the histogram of the velocity centroids for the two components of the [ o i ] 6300 lvc shows considerable overlap in velocity distributions ( see figure [ fig : histogram ] ) . here
we add another dimension in the comparison of the centroid velocities , with figure [ fig : lacc_vc ] showing the relationship between @xmath44 and @xmath54 for the bc and nc .
we conservatively use a @xmath551.5km / s from the stellar velocity as the area ( depicted in gray in figure [ fig : lacc_vc ] ) within which velocities can not be distinguished from being at rest with respect to the star . and
@xmath54 for the 6300 nc of the lvc ( upper ) and bc of the lvc ( lower ) .
sources with hvc removed are marked with spokes .
the gray area around the stellar velocity , @xmath55 1.5 km / sec , marks the uncertainty in centering the profiles.,scaledwidth=50.0% ] for the bc centroids there is a tendency for the stars with higher accretion rates to have blueshifted @xmath54 .
specifically , 7/12 stars with log @xmath66 @xmath67 -1.5 have @xmath54 blueshifted from -9 to -26 km / s , while with one exception ( ip tau ) all the stars with lower accretion rates have @xmath54 within 6 km
/ s of the stellar velocity .
a kendall @xmath61 test gives a probability of only 2% that the bc centroids and @xmath44 are uncorrelated .
interestingly , 6 sources have bc that are redshifted between + 3.8 to + 6.2 km / s ( aa tau , bp tau , fm tau , gi tau , go tau , v773 tau ) . while blueshifted centroids
are readily associated with winds ( 12 sources ) , and unshifted centroids with bound disk gas ( 7 sources ) , the small redshifts seen in 6 sources can also be consistent with a wind with certain disk geometries . for moderate disk inclinations or strong
flaring of the surface , the extended disk height at large radii may obscure ( from the observer s perspective ) the approaching part of a wind with a wide opening angle from the inner disk while the receding gas from the far side of the disk remains unobstructed ( gorti et al . in preparation ) . among the 18 nc detections
, there is no trend between @xmath54 and accretion luminosity and the centroid velocities either have small blueshifts or are consistent with the stellar velocity .
a kendall @xmath61 test lends to a 65% chance that the nc centroids and @xmath44 are uncorrelated .
blueshifted nc centroids ( @xmath54 from -1.75 to -13.8 km / s ) indicative of slow winds are seen in 9 sources ( coku tau 4 , cw tau , dg tau , dh tau , dk tau , ds tau , fm tau , gi tau , gk tau ) while the remaining 9 are consistent with bound gas ( aa tau , bp tau , df tau , dm tau , dr tau , fz tau , gm aur , ux tau a and tw hya ) .
there are no redshifted nc . from these comparisons ,
although the k - s test comparing the distribution of nc and bc centroids indicated a 9% chance they were drawn from the same parent population ( section [ sect : the_lvc ] ) , we conclude that there are significant differences in the behavior of the @xmath54 in these two components . the two - component nature of the lvc forbidden line emission in tts is reminiscent of kinematic behavior found in a series of papers looking at high resolution vlt - crires spectra of the 4.6 @xmath68 m co ro - vibrational fundamental lines in class i and ii sources ( @xcite ) .
the co @xmath691 - 0 line sometimes shows a nc / bc structure , while the @xmath692 - 1 transition predominantly traces the bc , which allows it to be isolated when both transitions are observed .
the finding that the fwhm of the broad and narrow co features may derive from the system inclination leads us to explore this possibility for the fwhm of the two components of the [ o i ] 6300 lvc .
first , we look at the relation between the fwhm of the nc and bc in those 13 sources where both components are seen at [ o i ] 6300 lvc . each of these components shows a range in line widths ( figure [ fig : histogram ] ) , and if both were rotationally broadened we would expect to see a correlation between them in sources where both are present .
we show this relation in figure [ fig : fwhm ] , where the fwhm of the nc , ranging from 12 to 39 km / sec , is reasonably well correlated with the fwhm of the bc , ranging from 44 to 140 km / sec , in all but one source ( dk tau ) .
the kendall @xmath61 probability that the two quantities are uncorrelated is indeed only 0.9% if we exclude dk tau , but it increases to 7.5% when dk tau is included .
6300 lvc in the 13 sources where both are present.,scaledwidth=50.0% ] we next explore the relationship between the width of the [ o i ] 6300 lvc nc and bc with the published values for disk inclination in 22 sources .
to directly compare our results with those already published for the co ro - vibrational line , we follow @xcite and plot the fwhm divided by the square root of the stellar mass versus the inclination from spatially resolved disks , rather than the fwhm vs the sine of disk inclination , which would be a better description of the relation expected for keplerian broadening ( left panel of figure [ fig : inclination_plot ] ) .
the figure shows that the two quantities are positively correlated , both for the bc and the nc components .
there are two outliers in the bc relation , dk tau and fm tau . dk tau is the source that did not show the expected behavior between nc and bc fwhm in figure [ fig : fwhm ] and here we see that the nc is in line with other sources at similar inclinations .
the very low mass source fm tau ( 0.1 m@xmath70 ) lies off the scale of the plot , but as we will show in section [ sect : radialbrightness ] , this can be explained if the emitting region is closer in than for stars of higher mass .
again , following @xcite we will use the positive linear correlation to infer disk inclinations for the remaining sources . in order to include uncertainties in this process
we assume an uncertainty of @xmath1 10% in the disk inclination , as reported in the references in table [ tab : source_properties ] .
for the fwhm , we adopt a monte carlo approach similar to that discussed in section [ sect : ews ] in relation to the uncertainties on the measured ews . using the 17 single component lvc sources
, we find an average uncertainty that is @xmath1 13% of the measured fwhms . as for the uncertainties on stellar masses , we assume 10@xmath71 for stars with masses
@xmath67 1 m@xmath70 and 30@xmath71 for stars with masses @xmath72 1 m@xmath70 ( see @xcite ) . finally ,
in order to calculate the total uncertainty on the y - axis , we propagate the error on the stellar mass and fwhm , assuming they are independent . adopting these uncertainties and using a linear relationship between fwhm / @xmath73 and disk inclination , we find the following best fits from the 22 sources with measured inclinations : @xmath74 @xmath75 where fwhm is measured in km / s , m@xmath76 is measured in m@xmath70 and inclination is measured in degrees . these fits
demonstrate that although the nc and bc both correlate with disk inclination , they have different slopes , suggesting that they trace different regions .
assuming that the broadening is due solely to keplerian rotation , we compute the [ o i ] disk radii from the velocity at the hwhm . the black solid lines in figure [ fig : inclination_plot ] show that the nc probes disk radii between 0.5 and 5 au while the bc traces gas much closer in , between 0.05 and 0.5 au .
we can compare these results to those of @xcite for co , where our linear fits are shown as red lines in figure [ fig : inclination_plot ] and those for co are shown in green .
we find similar slopes for the bc of both co and [ o i ] but slightly more extended disk radii for the [ o i ] than for the co ( the bc co disk radii range from 0.04 to 0.3au ) .
the slopes for the nc of co and [ o i ] are not in agreement , and the inferred formation region for [ o i ] , from 0.5 to 5au , is again slightly further from the star than that inferred for the nc of co ( 0.2 - 3au ) .
we can then use the linear fits presented above to derive disk inclinations for the sources with no disk - based inclination values in the literature ( right panel of figure [ fig : inclination_plot ] ) . for the sources where the lvc has only one component
, we calculate the inclination using equations 5 or 6 depending on whether it is classified as nc or bc . for the sources where the lvc has both a broad and a narrow component , we calculate the inclination of each component individually (
both are shown in figure [ fig : inclination_plot ] ) and take a weighted mean to derive the source inclination .
inclinations derived from these fits to the bc and nc relations are reported in table [ tab : source_properties ] and are marked with a dagger . in the previous section
, we found that the nc of the lvc has a line width consistent with keplerian broadening at a distance between 0.5 and 5au from the star . here
we will explore the range of radii in the disk required to account for the observed nc profile assuming a simple power law surface brightness fall off .
of the 18 sources with nc in their lvc , 9 have velocity centroids consistent with bound gas meaning that their profiles can be modeled with a simple keplerian disk . of these 9 sources 4 are transition disks ( tds ) , so we can also test whether the distribution of the gas in disks with dust cavities extending from a few to tens of au @xcite , is different from the nc of class ii sources .
our modeling uses a power law distribution for the line surface brightness of the form @xmath77 , where @xmath78 is the radial distance from the star and @xmath6 is varied between 0 and 2.5 ( see e.g. @xcite ) .
the radial profile is converted into a velocity profile assuming keplerian rotation , with the stellar mass and disk inclination in table [ tab : source_properties ] as additional input parameters .
the model line is convolved with a velocity width @xmath79 where @xmath80 is
the instrumental broadening ( 6.6km / s ) and @xmath81 is the thermal broadening .
we assume a temperature of 5,000k to compute @xmath81 because the [ o i ] emission might trace hot collisionally excited gas ( @xcite ) .
we then use the _ mpfitfun _ idl routinecraigm / idl / idl.html ] to find the best fit to the observed line profiles where an uncertainty equal to the rms on the continuum is adopted for each flux measurement .
the parameters that are varied in the fitting procedure are the inner and outer radii of the emitting gas and @xmath6 .
our best fits to the nc [ o i ] 6300 profiles of three representative sources with very similar inclinations ( @xmath82 ) but different disk types and mass accretion rates are shown in figure [ fig : td_models ] .
our simple model does a good job in reproducing the observed nc profiles and shows that the radial extent of the gas emitting the nc lvc is from within 1au , to explain the relatively large fwhms , out to @xmath110au , to explain the lack of double peaks in the profiles ( see table [ tab : model_results ] for a summary of inferred parameters ) .
we see a trend between the inner and outer radii and the stellar mass , with the smallest [ o i ] emitting region located around the lowest mass star dm tau and the largest around the most massive @xmath83 stars in our sample dr tau and ux tau a. the power law index of the surface brightness ranges from flat @xmath6= 0 ( for df tau ) to @xmath6= 2.2 ( for ux tau a and tw hya ) . using the best fit surface brightness we also compute the radius within which 80% of the nc lvc emission arises ( r@xmath84 in figure [ fig : td_models ] and table [ tab : model_results ] ) and find it to be within @xmath85au , in agreement with the [ o i ] disk radii estimated from the nc hwhm ( see figure [ fig : inclination_plot ] ) .
our steepest power law indexes of @xmath12.2 are very similar to those inferred for cw tau and dq tau by heg after re - centering their blueshifted [ o i ] profiles in the stellocentric frame and assuming that the entire line broadening is due to keplerian rotation .
the extent of the [ o i ] emitting region for these two sources is inferred to be between 0.1 out 20au , similar to the ranges we find .
this hints that keplerian broadening may also dominate the profile of wind sources .
we plan to model wind profiles in a future paper by adding an unbound component with a prescribed velocity field to the bound keplerian disk .
6300 profiles in 3 sources with the same inclination .
two ( dm tau and ux tau a ) are transition disks and one ( bp tau ) is a class ii source . values for the inner and outer disk radii , the radius within which 80% of the emission arises , and the surface brightness power law are indicated.,scaledwidth=50.0% ] in summary , our modeling of lvc nc profiles consistent with bound gas shows no difference in the radial distribution of the gas for tds and class ii sources .
given that the sub - mm dust cavity of ux tau a and dm tau are @xmath1 25au ( @xcite ) and @xmath1 19au ( @xcite ) respectively , our results also imply that the [ o i ] 6300 nc traces gas inside the dust cavity .
we find similar results for the other 2 transition disks with nc centroids consistent with bound gas , see table [ tab : model_results ] .
thus , we conclude that the [ o i ] 6300 nc profiles of tds can be explained by radially extended [ o i ] emission , most of which is confined within their dust cavities ( see @xcite for the dust cavities of tds ) .
ratios of forbidden line equivalent widths can provide insight into the temperature and density of the lvc emitting line region . for example , the ratio of [ o i ] 5577/6300 for one of our sources , the transition disk tw hya , with a value of @xmath11/7 @xcite , has been interpreted by @xcite as tracing the warm ( @xmath11,000k ) and bound molecular layer where oh molecules are photodissociated by fuv photons .
the lvc [ o i ] 5577/6300 ratio for the heg stars was examined by r13 who found values between @xmath11 and 1/8 for all sources except two with microjets that display smaller ratios ( @xmath10.07 ) .
r13 attributed the larger ratios of 1 to 1/8 also to fuv photodissociation of oh molecules while n14 prefer the alternative possibility of thermal emission from a very hot ( @xmath15,000 - 10,000k ) and dense ( n@xmath86@xmath10 ) gas to explain similar ratios in a different sample of stars with disks .
the mean ew [ o i ] 5577/6300 ratio for the full lvc for our sample stars is 0.25 , similar to those found previously .
however , with the decomposition of our high resolution spectra into bc and nc contributions to the lvc we can look to see if these two lvc components have the same ratio .
there are only 4 stars where we have sufficient signal to noise in both of these lines to evaluate the [ o i ] 5577/6300 for each component : aa tau , bp tau , cw tau , gi tau . in each case
the bc ratio is a factor of a few higher than the nc ratio of [ o i ] 5577/6300 .
figure [ fig : oiline_ratios ] compares the observed [ o i ] 5577/6300 values for these 4 stars ( bc in red and nc in blue ) with those predicted by a homogeneous and isothermal gas where the excitation is due solely to electron collisions , see appendix [ appendix ] for details .
this figure shows that the bc emitting region is about a factor of 3 denser than the nc region if these two components arise from gas at the same temperature .
alternatively , if they trace similarly dense ( @xmath87@xmath10 ) gas , the bc gas is @xmath125% hotter than the nc gas . in either scenario ,
gas needs to be hot ( @xmath125,000k ) to explain the observed ratios if the lines are thermally excited . for gas at 8,000k the observed ratios can be explained by @xmath88 ranging from @xmath89 to @xmath90@xmath10 implying gas densities of a few @xmath91@xmath10 for an ionization fraction of 0.33 , close to the value expected in the [ o i ] emitting region in some photoevaporative wind models ( see figure 2 in @xcite ) . by 5,000k electron densities
become very high ( @xmath92@xmath10 ) and , as discussed in the context of tw hya @xcite , are unlikely to be present in the hot surface of protoplanetary disks . an alternative interpretation for those sources with no definite blueshift
is that the [ o i ] emission is not thermal but results from the photodissociation of oh molecules in a cooler ( @xmath11,000k ) , bound , and mostly neutral layer of the disk @xcite .
although the higher ionization [ o ii ] 7330 line is not detected in any of our sources , we can use the upper limits to further constrain the gas temperature and density discussed above . figure [
fig : line_ratios ] shows the predicted [ o ii ] 7330/ [ o i ] 6300 ratios versus the [ o i ] 5577/6300 ratios for the full lvc . as expected , since the critical electron densities for both the [ o ii ] and [ o i ] lines are similar ( @xmath93 @xmath10 ) , the [ o ii]/[o i ] ratio is most sensitive to the gas temperature , given an ionization fraction which again we take to be 0.33 . models with the same electron density ( green lines ) , but different temperatures , run diagonally in figure [ fig : line_ratios ] .
this figure shows that the gas temperature must be lower than 6,500k if a single temperature is to explain all of our [ o ii ] 7330/ [ o i ] 6300 upper limits , which would imply high electron densities ( @xmath94@xmath10 ) based on the [ o i ] 5577/6300 ratios .
higher temperatures are possible only in combination with an ionization fraction lower than 0.33 .
as discussed earlier , it is also possible that the [ o i ] emission is not thermal but results from the dissociation of oh molecules @xcite .
a way to discriminate between thermal vs non - thermal emission would be to obtain high spectral resolution observations of the [ s ii ] line at 4068.6 because this line has a critical density of @xmath95@xmath10 , very similar to that of the [ o i ] line at 6300 ( see e.g. @xcite ) .
similar line profiles for the [ s ii ] and [ o i ] lines would clearly point to thermal emission in a hot dense gas .
the major contribution of this work is demonstrating that the low velocity forbidden emission in tts has kinematic properties that can be described as coming from a combination of a broad and a narrow line formation region .
most of our analysis is based on characterizing the behavior of the bc and nc as though they are two physically distinct emission regions . however , based on the current data we can not assess whether they both arise from the same phenomenon from different radial ranges in the disk or whether they arise in two different formation scenarios .
the most likely formation scenarios for the lvc are mass loss in the outer part of a centrifugally driven disk wind , mass loss in a photoevaporative flow , and bound gas in the disk .
we first review the highlights of our findings here and then look at the merits of attributing the lvc to these scenarios .
the bc of the lvc is very common , seen in 25/30 stars spanning the full range of disk accretion rates . in contrast , the nc of the lvc is somewhat less common , detected in only 18/30 stars , but again is found over the full range of disk accretion rates .
of the 5 stars which show only nc emission , 4 are transition disks and the 5th transition disk ( gm aur ) , has the highest percentage of nc emission in the lvc among the sources with bc emission . if the bc comes primarily from within 0.05 to 0.5 au , as suggested by the relation between its fhwm and disk inclination , then its absence in transition disks is likely due to a paucity of significant disk gas in this region .
this conclusion is in line with what has been proposed for tw hya from detailed modeling of emission lines covering a large range of radial distances @xcite .
four of our sources ( bp tau , df tau , fz tau , and gi tau ) have a redshifted hvc that might also hint to depletion in the inner disk , in this case of the dust component , which is the main source of opacity . however , their infrared indices , as reported in furlan et al .
( 2011 ) , place them in the full disk portion .
also their lvc are not reduced with respect to sources with no redshifted hvc ( compare bp tau to aa tau ) , thus viewing through a disk hole can not explain the redshifted hvc emission . while the range of observed fwhm for the bc and the nc
can be explained as a result of keplerian broadening from different radii in the disk ( between 0.05 to 0.5au for the bc and 0.5 to 5au for the nc ) , the different behaviors of their centroid velocities are the most useful in trying to understand their connection to disk winds .
a basic expectation of any wind model is that not only the fwhm but also the centroid @xmath54 of a line formed in the wind will correlate ( oppositely ) with disk inclination . in a close to face - on configuration keplerian broadening is minimal and the vertical component of the wind dominates , giving rise to an asymmetric blueshifted profile . on the opposite extreme , for a close to edge - on configuration , keplerian broadening dominates and because the wind is emerging in the plane of the sky , the resulting profile is symmetric and centered at the stellar velocity ( see e.g. @xcite ) .
we see the expected relations with fwhm and inclination for both the bc and nc lvc but the situation is not so clear when their centroid velocities are also considered .
we illustrate the relation between all 3 quantities in figure [ fig : eo_models ] , plotting the observed [ o i ] 6300 fwhm vs @xmath54 for the lvc bc ( upper panel ) and nc ( lower panel ) with datapoints color - coded by disk inclination ( see section [ sect : inclination ] ) . in the case of the bc ,
we see larger fwhm associated with larger disk inclination ( vertical color gradient ) coupled with , for most of the sources , larger blueshifts for narrower lines and lower disk inclination ( horizontal color gradient ) . these are the expected behaviors for a wind .
a kendall @xmath61 test gives a probability of only @xmath1 1% that the bc fwhm and @xmath54 are uncorrelated .
moreover , the sources with small bc redshifts are mostly in disks seen at high inclination , where , as described in section 5.4 , an extended disk height at large radii may obscure part of the wind from the inner disk while the receding gas ( from the observer s perspective ) remains unobstructed ( gorti et al . in preparation).as
the bc has the general characteristics expected for a simple wind and seems to arise in the innermost disk with maximum velocities of 10 to 27 km / s , we find it likely that the bc is the base of an mhd disk wind , since thermal speeds can not reach the necessary escape speeds this deep in the stellar gravitational potential field and photoevaporation can not occur . in the case of the nc , plotted over a smaller range of fwhm and @xmath54 than the bc ,
again we see fwhm associated with larger disk inclination ( vertical color gradient ) , however there is no relation between the @xmath54 and either the fwhm or the inclination .
a kendall @xmath61 test lends to a probability of @xmath1 21% that fwhm and @xmath54 are uncorrelated , consistent with no correlation .
two sources ( ds tau and dg tau ) stand out with much higher blueshifts than the rest of the nc , but one is close to face on and the other close to edge on . thus , although half of the nc have blueshifted @xmath54 , they do not behave in the manner expected for a simple wind .
although figures [ fig : inclination_plot ] and figure [ fig : eo_models ] show a relation between fwhm and inclination in the bc and nc there is considerable scatter and a few outliers .
however , we have not taken into account the fact that there will be an additional source of broadening in both the bc and nc if they are formed in outflowing gas . in the class
ii sources , comprising most of our sample , a dusty disk is assumed to occult the receding flow , and the observed fwhm would come both from keplerian rotation and expanding wind streamlines , producing anomalously high line widths for certain wind geometries ( e.g. bc of dk tau and fz tau ) . in the case of the transition disks , dominated by nc lvc emission
, the large dust - free cavities appear to have bound gas , although the gas is not as close in to the star to produce bc emission ( section 5 ) .
however , as there would presumably be no opacity source to occult a receding flow , the nc could include a contribution from flowing gas but still be centered on the stellar velocity and the line width would be enhanced by both approaching and receding gas ( see the case of tw hya discussed in @xcite and modeling by @xcite ) .
the past decade has seen significant advancements in the theory of thermal photoevaporative flows and sophisticated models , accounting for heating by stellar x - rays , euv , and fuv photons , are now available ( e.g. @xcite for a review ) .
we anticipate that these models will be soon providing more rigorous tests of whether the lvc might arise in thermal winds . theoretical models for magnetically driven centrifugal disk winds have also grown increasingly sophisticated and recently the relevance of mri - driven accretion over much of the disk has been challenged ( e.g. @xcite ) putting mhd disk winds back in the spotlight for extracting disk angular momentum to enable accretion onto the star ( e.g. @xcite ) .
global simulations of these winds have been recently presented by @xcite and thermal - chemical models have been investigated by @xcite , but predictables that can be directly compared with observations of tts forbidden lines are still lacking . we began this study to investigate the possibility that the lvc in tts might arise in photoevaporative flows .
it seems unlikely this could be the case for the bc of the lvc , which we attribute to the base of an mhd disk wind .
however , despite the lack of a convincing trend between centroid velocity and fwhm / disk inclination for the nc of the lvc we note that the _ range _ of fwhm and @xmath54 for the nc are consistent with those predicted for photoevaporative flows by @xcite . using the radiation - hydrodynamic code of @xcite , which includes x - ray and euv irradiation ,
@xcite find that the [ o i ] 6300 line is produced by collisional excitation and extends to @xmath135au above the disk , where euv photons and soft x - rays are absorbed and the gas is hot .
predicted fwhm range from 8 to 32km / s while peak centroids go from 0 to -7km / s , with the largest blueshifts occurring for intermediate disk inclinations ( @xmath96 ) , contrary to a simple wind geometry .
we do not see this trend either .
although hydrodynamical models that can predict line profiles of flows driven by fuv photons have not been developed , the expectation is that these flows , being cooler ( @xmath11,000k ) and mostly neutral , will have smaller velocities than those driven by x - rays .
the sound speed for such cool flows is only @xmath12km / s and the critical radius beyond which the gas would be unbound is @xmath110au around a solar - mass star . in this scenario ,
the [ o i ] emission will not be thermal and will arise from the dissociation of oh molecules in mostly bound gas inside of 10 au @xcite , which could explain the very small or absent blueshifts we see in many nc lines . in summary , at the present time the connection of the forbidden line nc of the lvc in tts to photoevaporative flows remains ambiguous , and we can not exclude that the nc is also part of a mhd disk wind .
we have analyzed optical high - resolution ( @xmath17km / s ) spectra from a sample of 33 tts whose disks span a range of evolutionary stages to clarify the origin of the lvc from oxygen and sulfur forbidden emission lines .
we detect the [ o i ] 6300 line in 30/33 tts , the [ o i ] 5577 line in 16/33 , and the [ s ii ] 6730 in only 8/33 tts . after isolating the forbidden lvc emission by removing any hvc contributions ,
if present , we draw the following conclusions about the residual lvc component : * all tts with [ o i ] detections show lvc emission .
thirteen out of 30 sources with [ o i ] 6300 emission have lvc emission that can be described as the combination of a broad ( bc ) and a narrow ( nc ) line emitting region .
the remaining sources show lvc emission that is either only bc ( 12/30 ) or only nc ( 5/30 ) . *
the bc of the lvc is very common , seen in 25/30 tts .
the nc of the lvc is less common , seen in 18/25 tts .
both components are found over the full range of accretion luminosities / disk accretion rates and their luminosities , combined or individually , correlate with the accretion luminosity .
lvc that are solely or predominantly nc are usually transition disk sources .
* comparison with spectra from heg shows that in most stars the lvc is stable over timescales of decades .
however , we do find evidence for variations , with the lvc disappearing entirely in one star and only the nc of the lvc disappearing in another star . *
the bc shows a relation between the fwhm and either observed or derived disk inclination suggesting it is broadened by keplerian rotation at disk radii between 0.05 to 0.5au .
also , a significant number of bc have blueshifts in excess of 5 km / s .
these larger blueshifts are associated with narrower lines and lower disk inclinations , as expected if the bc includes emission from a wind , in addition to keplerian broadening .
the bc with larger blueshifts also tend to be found in sources with higher accretion luminosity and hvc emission from microjets .
since the emission likely arises from 0.05 to 0.5 au , where the gravity of the star and the disk is strong , it is unlikely to trace a photoevaporative flow but rather the slower moving portion of an mhd disk wind .
* the nc also shows a relation between the fwhm and either observed or derived disk inclination suggesting it is broadened by keplerian rotation at disk radii between 0.5 to 5au .
half of the nc features are blueshifted between -2 to -5 km / s and the other half have centroids consistent with the stellar velocity .
although the expected relation for a simple wind between disk inclination and centroid velocity is not found , we can not exclude the possibility that the nc of the lvc arises in photoevaporative flows . *
regardless of the disk evolutionary stage , nc profiles consistent with bound gas can be reproduced by gas in keplerian motion with a surface brightness decreasing as a power law between 0.1au and @xmath110au , but with 80% of the emission arising within 5au .
the implication for transition disks is that the nc arises from gas inside the dust cavity . *
if forbidden emission lines are produced by collisional excitation with electrons , the [ o i ] 5577/6300 ratios suggest high temperatures ( @xmath165,000k ) and large electron densities ( @xmath97@xmath10 ) . without the additional constraints on density and temperature that would be provided by high resolution spectra of the [ s ii ] 4069 line
, the possibility remains that the excitation of [ o i ] is not thermal .
dissociation of oh molecules in a cool ( @xmath11,000k ) , bound , mostly neutral disk layer could be the source of the [ o i ] emission in objects with very small and absent shifts in the nc lvc .
disk winds , both mhd and photoevaporative , deplete material from several scale heights above the midplane . as dust grains
grow they settle toward the midplane .
the implication is that disk winds mostly deplete the protoplanetary disk of gas , which consequently increases the dust - to - gas ratio with time @xcite .
this increase can directly impact the formation of planetesimals , terrestrial planets , and the cores of giant planets . since disk winds play a significant role in disk dispersal and planet formation ,
both models and expanded observational data sets need to pursue the origin of the lvc nc , and constrain the rate at which material is lost via disk winds .
i. pascucci , u. gorti , and d. hollenbach acknowledge support from a collaborative nsf astronomy & astrophysics research grant ( ids : 1312962 and 1313003 ) .
s. edwards acknowledges support from nasa grant nng506ge47 g issued through the office of space science .
the data presented herein were obtained at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w.m .
keck foundation .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
_ facilities : _
as mentioned in section [ sect : obs ] the makee pipeline performs an automatic sky subtraction .
we show in figure [ fig : sky_sub ] the spectrum of fm tau before ( black ) and after sky subtraction ( red ) to highlight that even the strong terrestrial [ o i ] emission line at 6300 is well removed by the pipeline .
6300 before ( black ) and after ( red ) sky subtraction .
note that the strong [ o i ] terrestrial line is well removed by the makee pipeline.,width=576 ] the keck spectra presented here were acquired in the standard mode which places the slit along the parallactic angle in order to minimize potential slit losses .
this approach was taken because our main interest was to study the lvc , which was known to be compact , rather than the jet emission , which , most likely , extends beyond the slit width . for completeness , we provide in table [ tab : pa_table ] slit position angles and disk position angles ( which should be close to 90@xmath98 of the jet position angles ) .
go tau , ux tau a , and ds tau have the slit most closely aligned of a possible jet , within @xmath1 10@xmath98 , yet none of them show a jet signature in the [ o i ] spectra .
the collisional excitation model described here is used to gain physical insight in the temperature and electron density of the region traced by oxygen forbidden lines .
we assume a homogeneous and isothermal slab of gas , where the excitation is due solely to electron collisions .
we considered a ground state and four additional excited states ( 5-level atom ) both for the neutral and ionized oxygen .
we computed the relative populations of the levels as a function of gas temperature and density by including the processes of collisional excitation , collisional de - excitation , and spontaneous radiative decay .
einstein coefficients for radiative decay were taken from the nist database ] and electron collision strengths from @xcite .
we have not included collisions with neutral hydrogen because the de - excitation cross section of the level @xmath99s@xmath100 is not known ( see e.g. discussion in @xcite ) .
however , neutral collisions should be negligible when the electron abundance is larger than @xmath101 , as suggested by the same disk models , because the electron rate coefficients ( @xmath102@xmath103/s ) are much larger than those for h ( @xmath104@xmath103/s ) . because of the very similar ionization potential of h and o we have taken the ratio of h@xmath105/h to be equal to o@xmath105/o , equal to 0.5 for an ionization fraction of 0.33 , close to the value expected in the [ o i ] emitting region in some photoevaporative wind models ( see figure 2 in @xcite ) .
akeson , r. l. & jensen , e. l. n. 2014 , apj , 784 , 62 alcal , j. m. , natta , a. , manara , c. f. , et al .
2014 , a&a , 561a , 2 alexander , r. d. 2008 , mrnas , 391 , l64 alexander , r. d. & pascucci , i. 2012 , mnras , 422l , 82 alexander , r. , pascucci , i. , andrews , s. , et al .
2014 , protostars and planets vi , 914 , 475 andrews , s. m. , wilner , d. j. , espaillat , c. , et al .
2011 , apj , 732 , 42 andrews , s. m. & williams , j. p. 2007 ,
apj , 659 , 705 appenzeller , i. , oestreicher , r. , jankovics , i. 1984 , a&a , 141 , 108 appenzeller , i. & bertout , c. 2013 , a&a , 558 , 83 azuah , r. t. , kneller , l. r. , qiu , y. 2009 , j. res .
technol . , 114 , 341 bai , xue - ning , ye . j. , goodman , j. , yuan , f. 2016 , apj , 818 , 152 .
banzatti , a. & pontoppidan , k. m. 2015 , apj , 809 , 167 bast , j. e. , brown , j. m. , herczeg , g. j. , et al .
2011 , a&a , 527 , a119 beristain , g. , edwards , s. , kwan , j. 2001 , apj , 551 , 1037 brown , j. m. , pontoppidan , k. m. , van dishoeck , e. f. , et al .
2013 , apj , 770 , 94 cabrit , s. , edwards , s. , strom , s. e. , strom , k. m. 1990 , apj , 354 , 687 cabrit , s. 2007 , iaus , 243 , 203 calvet , n. , & gullbring , e. 1998 , , 509 , 802 cox , a. w. , grady c. a. , hammel , h. b. , et al .
2013 , apj , 762 , 40 draine , b. t. 2011 , physics of the interstellar and intergalactic medium edwards , s. , cabrit , s. , strom , s. e. , et al .
1987 , apj , 321 , 473 edwards , s. , fischer , w. , kwan , j. , hillenbrand , l. , dupree , a. k. 2003 , apj , 599 , l41 ercolano , b. & owen , j. e. 2010 , mnras , 406 , 1553 ercolano , b. & owen , j. e. 2016 , mrnas , 2016arxiv160504066e espaillat , c. , muzerolle , j. , najita , j. , andrews , s. , zhu , z. , calvet , n. , kraus , s. , hashimoto , j. , kraus , a. , dalessio , p. 2014 , in protostars and planets vi , henrik beuther , ralf s. klessen , cornelis p. dullemond , and thomas henning ( eds . ) , university of arizona press , tucson , 914 pp .
, p.497 - 520 fedele , d. , pascucci , i. , brittain , s. , et al .
2011 , apj , 732 , 106 ferreira , j. , dougados , c. , cabrit , s. 2006 , a&a , 453 , 785 fischer , w. , edwards , s. , hillenbrand , l. , kwan , j. 2011 , apj , 730 , 73 font , a. s. , mccarthy , i. g. , johnstone , d. , ballantyne , d. r. 2006 , apj , 607 , 890 gahm , g. f. , walter , f. m. , stempels , h. c. , et al .
2008 , a&a , 482 , l35 gorti , u. , hollenbach , d. , najita , j. , pascucci , i. 2011 , apj , 735 , 90 gorti , u. , hollenbach , d. , dullemond , c. p. 2015 , , 804 , 29 gressel , o. , turner , n. j. , nelson , r. p. , & mcnally , c. p. 2015 , apj , 801 , 84 gdel m. , briggs k. r. , arzner k. , et al .
2007 , a&a , 468 , 353 guilloteau , s. , dutrey , a. , pitu , v. , boehler , y. 2011 , a&a , 529 , a105 gullbring , e. , hartmann , l. , briceno , c. , calvert , n. 1998 , apj , 492 , 323 hamann , f. 1994 .
apj , 93 , 485 hartmann , l. , calvet , n. , gullbring , e. , & dalessio , p. 1998
, , 495 , 385 hartigan , p. , hartmann , l. , kenyon , s. , hewett , r. , stauffer , j. 1989 , apjs , 70 , 899 hartigan , p. , morse , j. a. , raymond , j. 1994 , apj , 436 , 125 hartigan , p. , edwards , s. , ghandour , l. 1995 , apj , 452 , 736 ( heg ) herbig , g. h. 1962 , adv .
, 1 , 47 herbst , w. , herbst , d. k. , grossman , e. j. , weinstein , d. 1994 , aj , 108 , 1906 herczeg , g. j. , & hillenbrand , l. a. 2008 , , 681 , 594 herczeg , g. j. & hillenbrand , l. a. 2014 , apj , 786 , 97 hirth , g. a. , mundt , r. , solf , j. 1997 , a&a , 126 , 437 hollenbach , d. & gorti , u. 2009 , , 703 , 1203 ingleby , l. , calvet , n. , herczeg , g. , et al . 2013 , apj , 767 , 112 isella , a. , carpenter , j. m. , sargent , a. i. 2009 , apj , 701 , 260 keane , j. t. , pascucci , i. , espaillat , c. , et al .
2014 , apj , 787 , 153 kenyon , s. j. , gmez , m. , whitney , b. a. 2008 , handbook of star forming regions , volume i : the northern sky asp monograph publications , vol .
4 . edited by bo reipurth , p. 405
kraus , a. l. , ireland , m. j. , martinache , f. , hillenbrand , l. a. 2011 , apj , 731 , 8 kwan , j. & tademaru , e. 1995 , apj , 454 , 382 lada , c. j. & wilking , b. a. 1984 , apj , 287 , 610 lavalley , m. , isobe , t. , feigelson , e. 1992 , in asp conf . ser .
25 , astronomical data analysis software and systems i , ed .
d. m. worrall , c. biemesderfer , & j. barnes ( san francisco , ca : asp ) , 245 manara , c. f. , testi , l. , rigliaco , e. , et al .
2013 , a&a , 551 , a107 mendiguta , i. , oudmaijer , r. d. , rigliaco , e. , et al . 2015 , mrnas , 452 , 2837 muzerolle , j. , hartmann , l. , & calvet , n. 1998 , , 116 , 2965 nagel , e. , dalessio , p. , calvert , n. , et al .
2010 , apj , 708 , 38 natta , a. , testi , l. , alcala , j. m. , rigliaco , e. , covino , e. , stelzer , b. , delia , v. 2014 , a&a , 569a , 5 owen , j. e. , ercolano , b. , clarke , c. j. 2011 , mrnas , 412 , 13 panoglou , d. , cabrit , s. , pineau des forts , g. , et al .
2012 , , 538 , a2 pascucci , i. , apai , d. , hardegree - ullman , e. e. , et al .
2008 , apj , 673 , 477 pascucci , i. , sterzik , m. , alexander , r. d. , et al .
2011 , 736 , 13 pascucci , i. , ricci , l. , gorti , u. , et al .
2014 , apj , 795 , 1 pascucci , i. , edwards , s. , heyer , m. , et al .
2015 , apj , 814 , 14 pickles , a. j. 1998 , pasp , 110 , 863 pitu , v. , guilloteau s. , di folco , e. , dutrey , a. , boehler , y. 2014 , a&a , 564 , a95 pontoppidan , k. m. , blake , g. a. , smette , a. 2011 , apj , 733 , 84 pontoppidan , k. m. , blake , g. a. , van dishoeck , e. , et al .
2008 , apj , 684 , 1321 ray , t. , dougados , c. , bacciotti , f. , eislffel , j. , & chrysostomou , a. 2007 , protostars and planets v , 231 rigliaco , e. , pascucci , i. , gorti , u. , edwards , s. , hollenbach , d. 2013 , apj , 772 , 60 rosenfeld , k. a. , qi , c. , andrews , s. m. , et al .
2012 , apj , 757 , 129 schegerer , a. a. , wolf , s. , hummel , c. a. , quanz , s. p. , richichi , a. 2009 , a&a , 502 , 367 shu , f. , najita , j. , ostriker , e. , et al .
1994 , apj , 429 , 781 stassun , k. g. , scholz , a. , dupuy , t. j. , kratter , k. 2014 , apj , 796 , 119 stelzer , b. , & schmitt , j. h. h. m 2004 , a&a , 418 , 687 turner , n. j. , fromang , s. , gammie , c. , klahr , h. , lesur , g. , wardle , m. , bai , x .-
2014 in protostars and planets vi , 411 , 432 vogt , s.s .
1994 , proc .
spie , 2198 , 362 lcccccccc + source & spty & sed & @xmath13 & log @xmath14 & @xmath15 & @xmath106 & ref & log @xmath65 + & & & & [ l@xmath70 ] & [ m@xmath70 ] & [ deg ] & ( @xmath106 ) & [ l@xmath70 ] + + + aa tau & m0.6 & ii & 0.40 & -0.35 & 0.57 & 71 & c13 & -3.49 + bp tau & m0.5 & ii & 0.45 & -0.38 & 0.62 & 39 & g11 & -3.45 + ci tau & k5.5 & ii & 1.90 & -0.20 & 0.90 & 44 & g11 & -4.30 + coku tau 4 & m1.1 & t & 1.75 & -0.50 & 0.54 & 46@xmath108 & & + cw tau & k3 & ii & 1.80 & -0.35 & 1.01 & 65 & p14 & -3.13 + cy tau & m2.3 & ii & 0.35 & -0.58 & 0.41 & 34 & g11 & -4.37 + df tau & m2.7 & ii & 0.10 & -0.04 & 0.32 & 52@xmath108 & & + dg tau & k7 & ii & 1.60 & -0.29 & 0.76 & 38 & g11 & -4.18@xmath109 + dh tau & m2.3 & ii & 0.65 & -0.66 & 0.41 & @xmath190@xmath108 & & -2.66 + dk tau & k8.5 & ii & 0.70 & -0.27 & 0.68 & 41 & aj14 & -3.62 + dl tau & k5.5 & ii & 1.80 & -0.30 & 0.92 & 38 & g11 & + dm tau & m3 & t & 0.10 & -0.89 & 0.35 & 35 & an11 & -4.33@xmath109 + dn tau & m0.3 & ii & 0.55 & -0.08 & 0.55 & 39 & i09 & -3.52 + do tau & m0.3 & ii & 0.75 & -0.64 & 0.70 & 37@xmath108 & & + dr tau & k6 & ii & 0.45 & -0.49 & 0.90 & 20 & s09 & + ds tau & m0.4 & ii & 0.25 & -0.72 & 0.69 & 65 & aj14 & + fm tau & m4.5 & ii & 0.35 & -1.15 & 0.15 & @xmath170 & p14 & -3.86 + fz tau & m0.5 & ii & 3.5 & -0.48 & 0.63 & 69@xmath108 & & -3.78 + gh tau & m2.3 & ii & 0.40 & -0.19 & 0.36 & 80@xmath108 & & -4.55 + gi tau & m0.4 & ii & 2.55 & -0.25 & 0.58 & 83@xmath108 & & -3.66 + gk tau & k6.5 & ii & 1.50 & -0.03 & 0.69 & 73 & aj14 & -3.42 + gm aur & k6 & t & 0.30 & -0.31 & 0.88 & 50 & g11 & + go tau & m2.3 & ii & 1.5 & -0.70 & 0.42 & 66 & aw07 & -4.19 + hn tau & k3 & ii & 1.15 & -0.77 & 0.70 & 52 & aj14 & -4.40@xmath109 + hq tau & k2 & ii & 2.6 & 0.65 & 1.53 & 20@xmath108 & & -2.86 + ip tau & m0.6 & ii & 0.75 & -0.41 & 0.59 & 33@xmath108 & & + it tau & k6 & ii & 3.1 & -0.01 & 0.76 & 42 & aj14 & -2.77 + tw hya & m0.5 & t & 0.00 & -0.72 & 0.69 & 6 & r12 & -3.85 + ux tau a & k0 & t & 0.65 & 0.22 & 1.51 & 35 & an11 & + vy tau & m1.5 & ii / iii & 0.6 & -0.41 & 0.47 & & & + v710 tau & m3.3 & ii & 0.8 & -0.43 & 0.30 & 44 & aj14 & -3.45 + v773 tau & k4 & ii / iii & 0.95 & 0.48 & 0.98 & 34@xmath108 & & -2.61 + v836 tau & m0.8 & ii & 0.6 & -0.52 & 0.58 & 48 & p14 & + + lccccc + source & photospheric standard & @xmath23 & ew ( [ o i ] 6300 ) & ew ( [ o i ] 5577 ) & ew ( [ s ii ] 6731 ) + & & & [ ]&[]&[ ] + + + aa tau & v819 tau & 0 & 0.74 @xmath55 0.03 & 0.21 @xmath55 0.03 & @xmath720.005 + bp tau & v819 tau & 0.6 & 0.35 @xmath55 0.05 & 0.20 @xmath55 0.04 & @xmath720.005 + ci tau & hbc 427 & 0.6 & 0.16 @xmath55 0.06 & @xmath720.016 & @xmath720.130 + coku tau 4 & v1321 tau & 0 & 0.13 @xmath55 0.02 & @xmath720.006 & @xmath720.004 + cw tau & hbc 427 & 1.5 & 1.70 @xmath55 0.02 & 0.35 @xmath55 0.05 & 0.18 @xmath55 0.07 + cy tau & gl 15a & 0.6 & 0.42 @xmath55 0.06 & 0.18 @xmath55 0.07 & @xmath720.006 + df tau & v819 tau & 1.6 & 1.51 @xmath55 0.12 & 0.15 @xmath55 0.03 & @xmath720.004 + dg tau & v819 tau & 1.0 & 9.83 @xmath55 0.19 & 0.62 @xmath55 0.04 & 1.62 @xmath55 0.04 + dh tau & v1321 tau & 0.5 & 0.75 @xmath55 0.04 & 0.22 @xmath55 0.02 & @xmath720.007 + dk tau & v819 tau & 0.4 & 1.33 @xmath55 0.09 & 0.31 @xmath55 0.05 & 0.10 @xmath55 0.06 + dl tau & hbc 427 & 1.0 & 1.11 @xmath55 0.06 & @xmath720.018 & 0.16 @xmath55 0.06 + dm tau & v1321 tau & 0.1 & 0.56 @xmath55 0.09 & @xmath720.017 & @xmath720.026 + dn tau & v819 tau & 0 & @xmath720.007 & @xmath720.007 & @xmath720.007 + do tau & v819 tau & 1.5 & 5.24 @xmath550.06 & 0.49 @xmath55 0.02 & 0.49 @xmath55 0.04 + dr tau & v819 tau , no correction & 5.6 & 0.22 @xmath55 0.05 & @xmath720.009 & @xmath720.008 + ds tau & v819 tau & 0.6 & 0.20 @xmath55 0.05 & 0.11 @xmath55 0.02 & @xmath720.005 + fm tau & v1321 tau , no correction & 3.5 & 0.70 @xmath55 0.04 & 0.28 @xmath55 0.04 & @xmath720.005 + fz tau & v1321 tau & 1.6 & 0.78 @xmath55 0.04 & 0.12 @xmath55 0.03 & 0.15 @xmath55 0.02 + gh tau & v1321 tau & 0.1 & 0.33 @xmath55 0.03 & @xmath720.004 & @xmath720.004 + gi tau & v819 tau & 0.5 & 1.30 @xmath55 0.07 & 0.36 @xmath55 0.07 & 0.19 @xmath55 0.03 + gk tau & v819 tau & 0.1 & 0.28 @xmath55 0.02 & @xmath720.004 & @xmath720.005 + gm aur & hbc 427 & 0.6 & 0.35 @xmath55 0.03 & @xmath720.009 & @xmath720.011 + go tau & v1321 tau & 0.3 & 0.16 @xmath55 0.02 & 0.11 @xmath55 0.04 & @xmath720.005 + hn tau & v819 tau , no correction & 1.3 & 5.77 @xmath55 0.08 & @xmath720.025 & 1.28 @xmath55 0.30 + hq tau & hbc 427 & 0.7 & 0.37 @xmath55 0.05 & @xmath720.007 & @xmath720.006 + ip tau & v819 tau & 0.4 , 0.3 & 0.69 @xmath55 0.07 , 0.52 @xmath55 0.03 & @xmath720.008 , @xmath720.005 & @xmath720.010 , @xmath720.009 + it tau & hbc 427 & 0 & 0.30 @xmath55 0.06 & @xmath720.004 & @xmath720.004 + tw hya & v819 tau & 0.5 & 0.50 @xmath55 0.04 & 0.09 @xmath55 0.02 & @xmath720.012 + ux tau a & hr 8832 , hbc 427 & 0.2 , 0.2 & 0.22 @xmath55 0.02 , 0.09 @xmath55 0.02 & @xmath720.006 , @xmath720.006 & @xmath720.003 , @xmath720.003 + vy tau & v1321 tau & 0 & @xmath720.010 & @xmath720.008 & @xmath720.008 + v710 tau & v1321 tau & 0.1 & @xmath720.008 & @xmath720.007 & @xmath720.011 + v773 tau & hbc 427 & 0.1 & 0.54 @xmath55 0.03 & @xmath720.003 & @xmath720.003 + v836 tau & v819 tau & 0.2 & 0.58 @xmath55 0.06 & 0.22 @xmath55 0.09 & @xmath720.007 + + lccccccc source & ew ( h@xmath6 ) & log @xmath43 & log @xmath44 ( h@xmath6 ) & log @xmath45 ( h@xmath6 ) & log @xmath60 ( total ) & log @xmath60 ( total lvc ) & log @xmath60 ( lvc nc ) + & [ ] & [ l@xmath70 ] & [ l@xmath70 ] & [ m@xmath70/year ] & [ l@xmath70 ] & [ l@xmath70 ] & [ l@xmath70 ] + + aa tau & 10.5 & -3.48 & -2.40 & -9.34 & -4.74 & -4.82 & -5.33 + bp tau & 97.0 & -2.34 & -1.12 & -8.17 & -4.89 & -5.00 & -5.83 + ci tau & 84.5 & -2.11 & -0.87 & -8.03 & -4.87 & -4.90 & + coku tau 4 & 1.16 & -4.62 & -3.67 & -10.7 & -5.68 & -5.72 & -5.72 + cw tau & 110 & -1.91 & -0.64 & -8.01 & -3.74 & -3.91 & -4.46 + cy tau & 114 & -2.52 & -1.33 & -8.20 & -5.09 & -5.09 & + df tau & 46.7 & -2.22 & -0.98 & -7.46 & -3.94 & -4.30 & -5.27 + dg tau & 63.5 & -2.26 & -1.03 & -8.12 & -3.16 & -4.05 & -4.05 + dh tau & 34.5 & -3.14 & -2.02 & -8.93 & -4.99 & -5.01 & -5.50 + dk tau & 33.5 & -2.67 & -1.50 & -8.53 & -4.17 & -4.28 & -5.69 + dl tau & 92.6 & -2.10 & -0.85 & -8.06 & -4.06 & -4.64 & + dm tau & 104 & -3.14 & -2.02 & -8.94 & -5.65 & -5.74 & -5.74 + dn tau & 13.5 & -3.06 & -1.93 & -8.76 & @xmath72 -5.87 & @xmath72 -5.87 & + do tau & 136 & -2.23 & -1.00 & -8.21 & -3.75 & -4.37 & + dr tau & 43.9 & -2.10 & -0.85 & -8.10 & -4.45 & -4.56 & -5.01 + ds tau & 49.4 & -2.97 & -1.82 & -9.03 & -5.47 & -5.47 & -6.17 + fm tau & 78.3 & -3.19 & -2.07 & -8.66 & -5.54 & -5.56 & -5.92 + fz tau & 176 & -1.95 & -0.68 & -7.77 & -4.41 & -4.70 & -5.22 + gh tau & 11.8 & -3.28 & -2.18 & -8.80 & -5.02 & -5.04 & + gi tau & 22.2 & -2.85 & -1.69 & -8.62 & -4.19 & -4.38 & -4.87 + gk tau & 14.9 & -2.87 & -1.71 & -8.58 & -4.66 & -4.77 & -5.15 + gm aur & 92.9 & -2.19 & -0.95 & -8.15 & -4.66 & -4.65 & -4.88 + go tau & 45.9 & -3.12 & -2.00 & -8.92 & -5.76 & -5.73 & + hn tau & 89.2 & -2.47 & -1.27 & -8.69 & -3.67 & -4.33 & + hq tau & 2.22 & -2.76 & -1.60 & -8.68 & -3.54 & -3.71 & + ip tau & 10.4 & -3.39 & -2.29 & -9.30 & -4.70 & -5.50 & + it tau & 16.9 & -2.83 & -1.67 & -8.63 & -4.63 & -4.62 & + tw hya & 230 & -2.32 & -1.10 & -8.34 & -5.09 & -5.13 & -5.13 + ux tau a & 10.2 & -2.70 & -1.52 & -8.84 & -4.75 & -4.70 & -4.70 + vy tau & 4.24 & -3.96 & -2.93 & -9.81 & @xmath72 -6.13 & @xmath72 -6.13 & + v710 tau & 1.87 & -4.49 & -3.52 & -10.1 & @xmath72 -6.52 & @xmath72 -6.52 & + v773 tau & 3.02 & -3.02 & -1.88 & -8.79 & -3.79 & -3.86 & + v836 tau & 10.4 & -3.58 & -2.51 & -9.56 & -4.94 & -4.91 & + + + aa tau & [ o i ] @xmath28 6300 & -33 & 25 & & & & & & + + bp tau & [ o i ] @xmath28 6300 & & & & & 124 & 46 & & + + cw tau & [ o i ] @xmath28 6300 & -112 & 89 & & & & & & + & [ o i ] @xmath28 5577 & -112 & 89 & & & & & & + & [ s ii ] @xmath28 6731 & -116 & 38 & & & & & & + + df tau & [ o i ] @xmath28 6300 & -114 & 56 & & & 82 & 102 & & + + dg tau & [ o i ] @xmath28 6300 & -144 & 118 & -38 & 68 & & & & + & [ o i ] @xmath28 5577 & -144 & 118 & -38 & 68 & & & & + & [ s ii ] @xmath28 6731 & -103 & 171 & -34 & 46 & & & & + + dk tau & [ o i ] @xmath28 6300 & -126 & 37 & -42 & 51 & & & & + & [ s ii ] @xmath28 6731 & -56 & 38 & - & & & & & + + dl tau & [ o i ] @xmath28 6300 & -138 & 115 & & & & & & + & [ s ii ] @xmath28 6731 & -133 & 70 & & & & & & + + do tau & [ o i ] @xmath28 6300 & -97 & 51 & -95 & 19 & & & & + & [ o i ] @xmath28 5577 & -85 & 74 & & & & & & + & [ s ii ] @xmath28 6731 & -93 & 27 & & & & & & + + fz tau & [ o i ] @xmath28 6300 & -124 & 32 & & & 76 & 70 & 125 & 29 + & [ s ii ] @xmath28 6731 & & & & & 130 & 23 & & + + gi tau & [ o i ] @xmath28 6300 & -71 & 49 & & & 40 & 23 & & + & [ s ii ] @xmath28 6731 & -61 & 31 & & & 39 & 22 & & + + hn tau & [ o i ] @xmath28 6300 & -66 & 130 & & & & & & + & [ s ii ] @xmath28 6731 & -84 & 113 & & & & & & + + hq tau & [ o i ] @xmath28 6300 & -41 & 27 & & & & & & + + ip tau & [ o i ] @xmath28 6300 & -38 & 124 & & & & & & + + aa tau & 27 & -0.19 & 0.19 & 111 & 4.67 & 0.42 & 27 & -0.19 & 0.02 & 111 & 4.67 & 0.18 + bp tau & 16 & 0.11 & 0.04 & 87 & 5.65 & 0.23 & 16 & 0.11 & 0.01 & 87 & 5.65 & 0.15 + ci tau & & & & 66 & -8.70 & 0.15 & & & & & & + coku tau 4 & 18 & -2.90 & 0.12 & & & & & & & & & + cw tau & 24 & -2.63 & 0.32 & 76 & -2.70 & 0.83 & 23 & 0.92 & 0.05 & 73 & 1.61 & 0.25 + cy tau & & & & 48 & -1.02 & 0.47 & & & & 61 & -0.98 & 0.22 + df tau & 14 & -1.34 & 0.07 & 57 & -25.7 & 0.58 & 14 & 0.93 & 0.03 & 49 & -14.0 & 0.09 + dg tau & 18 & -12.3 & 1.25 & & & & 19 & -10.6 & 0.26 & & & + dh tau & 27 & -1.75 & 0.23 & 119 & -1.31 & 0.48 & 27 & -1.75 & 0.08 & 119 & -1.31 & 0.15 + dk tau & 12 & -1.90 & 0.04 & 137 & -11.8 & 0.98 & 12 & -1.90 & 0 & 137 & -11.8 & 0.30 + dl tau & & & & 44 & -11.3 & 0.29 & & & & & & + dm tau & 23 & 1.10 & 0.45 & & & & & & & & & + do tau & & & & 54 & -24.9 & 1.24 & & & & 58 & -17.2 & 0.31 + dr tau & 12 & -0.11 & 0.06 & 49 & -11.6 & 0.11 & & & & & & + ds tau & 39 & -13.8 & 0.04 & 115 & 1.10 & 0.16 & 39 & -13.8 & 0.02 & 115 & 1.10 & 0.09 + fm tau & 30 & -1.82 & 0.29 & 140 & 6.15 & 0.38 & 30 & -1.82 & 0.12 & 140 & 6.15 & 0.13 + fz tau & 26 & -1.28 & 0.12 & 95 & -14.6 & 0.28 & 26 & -1.28 & 0.06 & 95 & -14.6 & 0.09 + gh tau & & & & 84 & -0.66 & 0.31 & & & & & & + gi tau & 33 & -2.85 & 0.27 & 97 & 5.61 & 0.57 & 33 & -2.85 & 0.09 & 97 & 5.61 & 0.23 + gk tau & 39 & -3.50 & 0.09 & 116 & 0.40 & 0.13 & & & & & & + gm aur & 28 & -1.26 & 0.21 & 73 & -4.11 & 0.15 & & & & & & + go tau & & & & 106 & 3.84 & 0.17 & & & & 106 & 3.84 & 0.11 + hn tau & & & & 48 & -1.18 & 1.27 & & & & & & + hq tau & & & & 44 & -3.47 & 0.25 & & & & & & + ip tau & & & & 44 & -26.9 & 0.11 & & & & & & + it tau & & & & 71 & -1.54 & 0.31 & & & & & & + tw hya & 12 & -0.78 & 0.46 & & & & 11 & 0.37 & 0.07 & & & + ux tau a & 29 & 1.50 & 0.10 & & & & & & & & & + v773 tau & & & & 59 & 3.91 & 0.46 & & & & & & + v836 tau & & & & 67 & -5.23 & 0.63 & & & & 67 & -5.23 & 0.12 + + + ux tau a & 1.51 & 0.50 & 18.0 & 1.83 & 2.22 + + gm aur & 0.88 & 0.07 & 10.9 & 0.40 & 1.89 + + tw hya & 0.69 & 0.05 & 11.1 & 0.18 & 2.23 + + dm tau & 0.35 & 0.02 & 3.00 & 0.35 & 1.47 + + + + + dr tau & 0.90 & 0.46 & 18.2 & 3.39 & 1.65 + + bp tau & 0.62 & 0.08 & 9.20 & 3.90 & 0.94 + + fz tau & 0.63 & 0.17 & 19.7 & 1.50 & 1.66 + + aa tau & 0.57 & 0.22 & 11.4 & 2.04 & 1.56 + + df tau & 0.32 & 0.20 & 8.61 & 6.92 & 0.00 + lccccc + source & slit pa & disk pa & ref + & [ @xmath98]&[@xmath98 ] & ( disk pa ) + + + aa tau & 105 & 97 & c13 + bp tau & 246 & 107 & g11 + ci tau & 84.8 & 285 & g11 + coku tau 4 & 223 & & + cw tau & 260 & 332 & p14 + cy tau & 251 & 63 & g11 + df tau & 102 & & + dg tau & 102 & 43 & g11 + dk tau & 190 & 15 & aj14 + dl tau & 84.1 & 141 & g11 + dm tau & 95.6 & 155 & an11 + dn tau & 236 & 86 & i09 + do tau & 270 & & + dr tau & 78.8 & 108 & i09 + ds tau & 252 & 165 & aj14 + fm tau & 260 & 83 & p14 + fz tau & 256 & & + gh tau & 91.7 & & + gi tau & 260 & & + gk tau & 263 & 93 & aj14 + gm aur & 92.1 & 144 & g11 + go tau & 99.2 & 0 & aw07 + hn tau & 79.5 & 65 & aj14 + hq tau & 273 & & + ip tau & 112 & & + it tau & 254 & 106 & aj14 + tw hya & 337 & 332 & pd08 + ux tau a & 278 & 176 & an11 + vy tau & 274 & & + v710 tau & 96.4 & 82 & aj14 + v773 tau & 270 & & + v836 tau & 92.5 & -122 & p14 + + | using keck / hires spectra ( @xmath0 7 km / s ) we analyze forbidden lines of [ o i ] 6300 , [ o i ] 5577 and [ s ii ] 6731 from 33 t tauri stars covering a range of disk evolutionary stages . after removing a high velocity component ( hvc ) associated with microjets ,
we study the properties of the low velocity component ( lvc ) .
the lvc can be attributed to slow disk winds that could be magnetically ( mhd ) or thermally ( photoevaporative ) driven .
both of these winds play an important role in the evolution and dispersal of protoplanetary material .
lvc emission is seen in all 30 stars with detected [ o i ] but only in 2 out of eight with detected [ s ii ] , so our analysis is largely based on the properties of the [ o i ] lvc . the lvc itself is resolved into broad ( bc ) and narrow ( nc ) kinematic components . both components are found over a wide range of accretion rates and their luminosity is correlated with the accretion luminosity , but the nc is proportionately stronger than the bc in transition disks . the fwhm of both the bc and nc correlates with disk inclination , consistent with keplerian broadening from radii of 0.05 to 0.5 au and 0.5 to 5au , respectively .
the velocity centroids of the bc suggest formation in an mhd disk wind , with the largest blueshifts found in sources with closer to face - on orientations .
the velocity centroids of the nc however , show no dependence on disk inclination .
the origin of this component is less clear and the evidence for photoevaporation is not conclusive . |
the dark energy is now well established as a dominant constituent of the present day universe .
its existence is directly inferred from the acceleration in cosmic expansion as indicated by snia observation @xcite and indirectly from cmb anisotropy measurements @xcite and integrated sachs - wolfe effect @xcite .
but , the nature of dark energy still remains enigmatic whether it arises from the cosmological constant , quintessence or phantom fields with constant equation of state parameter or varying @xmath0 .
the most commonly used method to explore the dark energy models in the literature is to assume an adhoc equation of state @xmath0@xmath5@xmath6 for dark energy and parametrize @xmath0 or dimensionless dark energy function @xmath7@xmath5@xmath8 by one , two or more free parameters and then constrain these parameters by fitting to the observed snia , cmb and lss data . using this method
wang and tegmark @xcite have discussed four viable parametrizations and put tight constraints over @xmath9 in a rather model independent way .
we have adopted a different approach in this paper to constrain @xmath0 using dynamical age of the universe as the criterion to test the physical viability of different parametrizations .
first , we argue that the age of the universe known observationally from various independent methods is found to be reasonably consistent and it can be used as an effective constraint on the evolution of the dark energy equation of state @xmath0 because the dynamical age of the universe depends on the hubble expansion rate , which essentially involves @xmath10 .
the possibility of using the age of the universe as a constraint has been discussed in recent papers [ 3,6 - 12 ] .
the oldest globular clusters yield ages of about 12.5 gyr with an uncertainty of 1.5 gyr @xcite . assuming the genesis of dark energy from the cosmological constant ( @xmath11 ) , the sn ia data yield the product of age and hubble constant @xmath12 @xcite . taking @xmath13 mpc@xmath14 km s@xmath15
, it gives @xmath16 gyr .
padmanabhan @xcite has given the maximum likelihood for @xmath17 based on the analysis of sn ia observations .
this can be consistent with @xmath18 models only if @xmath19 and @xmath20 .
it is noteworthy that the sne observations constrain the combination @xmath21 better than the individual parameters . with the implicit assumption of the cosmological constant ( @xmath11 ) , the wmap data @xcite yields @xmath22 gyr .
relaxation of this constraint would lead to variable estimates of @xmath23 depending upon the prior choice of @xmath1 , @xmath2 and @xmath24 . on the other hand , knox et al @xcite
have given a method for precise determination of the dynamical age of the universe from cmb anisotropy measurements .
the best estimate of the dynamical age of the universe , coming from cmb data , is @xmath25 gyr .
although this method also presumes the cosmological constant , it is shown that the variation of the equation of state away from @xmath11 at fixed @xmath26 ( the angle subtended by the acoustic horizon on the last scattering surface ) has very little effect on the age of the universe . moreover
, the cmb method of age determination does not involve the observational parameters @xmath1 and @xmath27 and is therefore free from observational uncertainties involved in the measurement of these parameters .
these diverse observations leading to mutually agreeable estimates of the age of the universe reinforce our confidence in using the dynamical age of the universe as an effective tool to constrain the equation of state and the dark energy density parameter @xmath28 .
we would take @xmath25 gyr ( based on cmb anisotropy observations @xcite ) as the most reliable estimate of the present age of the universe since it is not sensitive to variation in the values of the hubble constant @xmath1 and the fractional energy density @xmath27 of the non - relativistic matter in the universe .
we would use it as the standard criterion for the dynamical age of the universe for comparison with a variety of kinessence models @xcite whose equation of state @xmath0 is assumed to have different parametrizations as discussed in sec .
iv of the paper .
although , the present dark energy density @xmath29 @xmath30 is precisely known @xcite and the cosmological density parameter @xmath31 from wmap data @xcite , we have no definite estimate of @xmath32 and @xmath33 at small @xmath34 or large @xmath34 in the past . as such , we do not know whether the dark energy evolves with cosmic time or not and if it does then what is the mode of its evolution . however , it is possible to find @xmath33 precisely in terms of @xmath0 at the point of transition @xmath35 from the decelerating phase to accelerating expansion phase by taking @xmath36 in eq .
we have discussed in our previous paper @xcite , how the transition red shift @xmath35 may be used to compute the age of the universe @xmath37 up to the transition epoch and also the dynamical age of the universe @xmath23 up to the present epoch .
we have used this technique in sec .
iii of this paper to investigate the range of admissible values of @xmath3 for quiessence model , which are compatible with the range of dynamical age of the universe .
the same constraint is applied in sec .
iv to test the physically viable parametrizations of @xmath0 assuming the slow roll down condition for the scalar field in kinessence models .
during the matter dominated era onwards , the contribution of cmb photons and neutrinos to the cosmic energy density is trivial and the major energy constituents are non - relativistic matter ( baryonic matter and dark matter ) and the dark energy .
the friedmann equations for a spatially flat ( @xmath38 ) universe are @xmath39 \nonumber \\ & = & \;\;h_{o}^{2}\left [ \omega _ { m}^{o}(1+z)^{3}+\omega _ { x}^{o}f(z)\right ] \end{aligned}\ ] ] and @xmath40 \nonumber \\ & = & -\frac{h^{2}}{2}\left [ \omega _ { m}+\omega _ { x}\left ( 1 + 3w\right ) \right ] \end{aligned}\ ] ] where @xmath41 is the hubble parameter .
the energy density of the non - relativistic matter @xmath42 is given by @xmath43 and the dark energy density @xmath44 is written as @xmath45 with @xmath46 \label{dade}\ ] ] in particular for quiessence models ( @xmath47 ) @xmath48 this condition holds good during ` tracking ' wherein slow varying equation of state ( @xmath49 ) is a pre - requisite . in case of the cosmological constant ( @xmath11 ) , @xmath50 and @xmath51 . in all other cases , the dark energy density @xmath52 evolves with the redshift both for varying and non - varying @xmath0 . logarithmic differentiation of eq .
( 4 ) yields @xmath53 using eq . ( 6 ) , a suitable parametrization for the dark energy density can be assumed to find the @xmath0 which conforms to observational constraints .
using the cosmic energy density parametric relation @xmath54 @xmath55 and for a spatially flat universe , eq .
( 6 ) may be written in the form @xmath56 where @xmath57 it is a significant relation as it reveals how the dark energy density varies with the evolution of the equation of state parameter @xmath0 . with the help of eq .
( 7 ) , the friedmann eq . ( 2 ) can be recast in the form @xmath58 finally , the dynamical age of the universe is given by @xmath59 ^{1/2 } } \label{age}\end{aligned}\ ] ] the impact of the evolution of dark energy on the dynamical age of the universe can be clearly seen from the fact that for a given functional form of @xmath7 , the contribution of @xmath44 to @xmath41 in eq .
( 1 ) goes on decreasing with more negative values of @xmath0 .
consequently , the expansion age of the universe given by eq.(9 ) increases with evolution of @xmath0 to more negative values . at the same time
, the cosmic expansion accelerates according to eq .
( 2 ) for more negative values of @xmath0 .
thus , the expansion dynamics of the universe @xcite revolves around the equation of state parameter @xmath0 .
quiessence models @xcite of dark energy ( @xmath47 ) find wide application in tracker field theory @xcite wherein slow roll down condition of scalar fields demands @xmath60 .
melchiorri et al .
@xcite have combined constraints from cmb observations ( including latest wmap data ) , large scale structure , luminosity measurement of sn type ia and hubble space telescope measurements to find the bounds of dark energy equation of state parameter to be @xmath61 on the basis of wmap data for the dynamical age of the universe , johri @xcite has shown that @xmath3 lies in a thin strip around @xmath11 , viz @xmath62 taking @xmath63 gyr . at the transition epoch ( @xmath36 ) , eq .
( 8) simplifies to the differential equation @xmath64 where @xmath65 is the redshift at the transition from decceleration to accelerating expansion phase corresponding to equation of state parameter @xmath3 .
a particular integral of eq .
( 11 ) yields @xmath66 ^{-1/3w}\ ] ] in fact , eq . ( 12 ) follows directly from eq .
( 2 ) for quiessence models . for a prior choice of the ratio @xmath67 , it is interesting to plot the variation of @xmath35 versus @xmath3 according to the eq .
( 12 ) as shown in our previous paper @xcite .
11 ) gives the gradient of the @xmath35 @xmath68 @xmath3 curve .
this equation can be alternatively written as @xmath69 and @xmath70 in table 1 , we have given values of transition redshift @xmath35 for various quiessence models with dark energy equation of state parameter @xmath71 the dark energy density parameter @xmath72 and corresponding age of the universe @xmath23 taking @xmath73 gyr .
we have also plotted the variation of @xmath35 and @xmath23 with respect to @xmath3 in fig .
it is noticed that there exists an inverse correlation between the variation of @xmath74 and @xmath23 with respect to @xmath3 in quiessence models of dark energy .
jimenez et al .
@xcite have detected a similar age - redshift correlation between the ages of the oldest galaxies and their redshifts .
in case of kinessence models @xcite with varying equation of state @xmath75 , there are various competing dark energy models with different parametrizations of @xmath0 such as one index parametrizations by gong et al .
@xcite , two index parametrizations by huterer et al .
@xcite , weller et al .
@xcite , chevallier et al .
@xcite , linder @xcite , jassal et al .
@xcite , upadhye et al .
@xcite as well as wetterich @xcite and four index parametrizations by hannestad et al . @xcite and lee @xcite
. the question is how well the dark energy equation of state and the cosmological density parameter @xmath28 in these models behave with increasing redshift @xmath34 .
we have discussed the critical evaluation of some of these models in our earlier work @xcite . in the following section
, we examine the cosmic age predictions of these models given in table 2 and compare them with the observational estimates of the dynamical age of the universe @xmath25 gyr derived from cmb anisotropy measurements @xcite . *
gong - zhang 1st parametrization * the one index dark energy equation of state parameter @xmath0 is given by gong et al @xcite as @xmath76 the best fit values to sn ia ` gold set ' , sdss and wmap data are @xmath77 and @xmath78 .
hence , the parameters favor dark energy of phantom origin . combining eqs .
( [ gong1 ] ) and ( [ dade ] ) , the dark energy density is given by @xmath79\ ] ] and the dark energy parameter @xmath80 can be written as @xmath81 ^{-1}\ ] ] we calculate the transition redshift @xmath82 for this model by taking @xmath83 in eq.(8 ) and inserting for @xmath3 and @xmath84 ( the best fit values of the parameters @xmath85 and @xmath2 ) from eqs .
( 15 ) and ( 17 ) respectively . knowing the transition redshift @xmath35
, the total dynamical age of the universe ( worked out using the technique of @xcite ) turns out to be @xmath86 gyr .
since it lies outside the observational range given by knox et al @xcite , this parametrization is found to be incompatible with the age constraint .
following the same procedure , we have found the dynamical age of the universe for various parametrizations discussed in this section and have tested their physical viability on the basis of the age constraint .
the results are summarized in table 2 . *
gong - zhang 2nd parametrization * the second one index dark energy equation of state parameter @xmath0 is given by @xcite @xmath87 the best fit values to sn ia ` gold set ' , sdss and wmap data are @xmath88 and @xmath89 and the parameters favor dark energy of quintessence origin . inserting for @xmath0 from eq .
( [ gong2 ] ) into eq .
( [ dade ] ) , one gets @xmath90 and the dark energy parameter @xmath80 turns out to be @xmath91 ^{-1}\ ] ] for this parametrization , the transition red shift @xmath92 , we get @xmath93 . the age of the universe @xmath94 gyr .
linear - redshift parametrization * the dark energy equation of state parameter @xmath0 is given by huterer et al .
@xcite and weller et al .
@xcite as @xmath95 it has been used by riess et al .
@xcite for probing sn ia observations at @xmath96 .
the best fit values to sn ia ` gold set ' data @xcite are @xmath97 , @xmath98 and @xmath99 .
hence , the parameters favor dark energy of phantom origin . inserting eqs .
( [ hute ] ) into eq.([dade ] ) , one gets @xmath100 and @xmath101 ^{-1}\ ] ] in this case , the transition red shift @xmath92 , the corresponding dynamical age of the universe is @xmath102 gyr which lies beyond . the observational estimates . *
chevallier - polarski - linder parametrization * the dark energy equation of state parameter @xmath0 is given as @xcite @xmath103 the best fit values of @xmath104 @xmath105 and @xmath106 to sn ia ` gold set ' data @xcite are @xmath107 , @xmath108 and @xmath109 respectively and the parameters suggest that the dark energy is of phantom origin .
the dark energy density is given by @xmath110 and the dark energy parameter @xmath111 ^{-1}\ ] ] the transition red shift @xmath112 and the calculated dynamical age of the universe @xmath113 gyr lies beyond the observational estimates .
jassal - bagla - padmanabhan parametrization * jassal _ et al .
_ have parametrized @xmath0 as @xcite @xmath114 the best fit values to sn ia ` gold set ' data are @xmath115 , @xmath116 and @xmath99 .
one has @xmath117 at @xmath118 .
the parameters suggest that the dark energy is of phantom origin . combining eq .
( [ jasl ] ) and eq .
( [ dade ] ) one obtains @xmath119\ ] ] and @xmath120 ^{-1}\ ] ] the transition redshift @xmath121 and the corresponding age of the universe turns out to be @xmath122 gyr for the parametrization of jassal et al.@xcite . *
upadhye - ishak - steinhardt parametrization * upadhye et al .
have parametrized @xmath0 as @xcite @xmath123 the best fit values of parameters for sn ia ` gold set ' , galaxy power spectrum and cmb power spectrum data are @xmath124 , @xmath125 and @xmath126 . hence
, the parameters suggest that the dark energy has phantom origin .
the dark energy density is given by @xmath127 and the dark energy density parameter is written as @xmath128 ^{-1 } & & if & & z<1 \end{array } \\
\begin{array}{lllll } \left [ 1+\frac{\omega _ { m}^{0}}{\omega _ { x}^{0}}% ( 1+z)^{-3(w_{0}+w_{1})}e^{-3w_{1}\left ( 1 - 2\ln 2\right ) } \right ] ^{-1 } & & if & & z\geq 1 \end{array } \end{array } \right.\ ] ] the transition red - shift @xmath35 turns out to be @xmath129 and corresponding dynamical age of the universe turns out to be @xmath130 gyr . *
* * wetterich parametrization
* wetterich has parametrized the dark energy equation of state @xmath0 as @xcite @xmath131 ^{2 } } \label{wett}\ ] ] the best fit values to sn ia ` gold set ' data are @xmath132 , @xmath133 and @xmath134 .
one has @xmath135 at @xmath118 .
the parameters suggest that the dark energy is of phantom origin . inserting eq.([wett ] ) for @xmath0 in eq .
( [ dade ] ) , one gets @xmath136 and @xmath137 ^{-1}\ ] ] for @xmath138 , the calculated age of the universe @xmath139 gyr in comparison to the age @xmath140 gyr derived from cmb anisotropies by knox et al @xcite . *
hannestad - mrtsell parametrization * let us now consider hannestad parametrization @xcite which involves 4-parameters .
@xmath141 where @xmath85 and @xmath105 are the asymptotic values of @xmath0 in the distant future ( @xmath142 ) and in the distant past ( @xmath143 ) respectively .
the @xmath144 and @xmath145 are the scale factor at the change over and the duration of the change over in @xmath3 respectively . taking the best fit values for the combined cmb , lss and sn ia data @xcite , @xmath146 @xmath147 @xmath148 @xmath149 and @xmath150 with a prior @xmath151 ,
one has @xmath152 at @xmath153 these parameters suggest that the dark energy is of phantom origin .
further , the dark energy density @xmath154 is given by @xmath155 ^{\frac{3\left ( w_{0-}w_{1}\right ) } { p}}\ ] ] and the dark energy parameter is written as @xmath156 ^{-1}\ ] ] at transition redshift @xmath92 , the dark energy parameter @xmath157 . the age of the universe @xmath139 gyr in comparison to @xmath25 gyr derived by knox et al @xcite from cmb observations . *
2 . lee parametrization * the dark energy equation of state parameter @xmath0 is parametrized by lee as @xcite @xmath158 where @xmath159 and the symbols @xmath160 and @xmath161 have the usual meaning as in @xcite .
the parameters @xmath160 is chosen to be @xmath162 using the tracking condition and @xmath85 is taken as @xmath163 .
the other parameters are obtained by analyzing the separation of cmb peaks and the time variation of the fine structure constant . for @xmath164 and @xmath165 , the range of @xmath145
is taken as @xmath166 . at @xmath118 ,
one has @xmath167 . hence ,
the parameters suggest that the dark energy is of quintessence origin . inserting eq .
( [ lee ] ) into eq .
( [ dade ] ) , one gets @xmath168 the dark energy parameter @xmath169 is given by @xmath170 ^{-1}\ ] ] for @xmath171 and @xmath172 , the dark energy parameter @xmath28 is 0.353 and 0.333 at the transition red shift @xmath173 respectively . in comparison to the observational estimate of the age derived by knox et al @xcite @xmath25 gyr , the age of the universe for this parametrization turns out to be @xmath174 and @xmath175 gyr for @xmath171 and @xmath172 respectively .
in the case of kinessence models with equation of state @xmath176 , @xmath44 and @xmath28 are expressible in terms of @xmath7 which depends upon the choice of parametrization of @xmath0 .
so , it is not possible to predict the variation of @xmath28 @xmath177 precisely in the past without knowing the realistic form of @xmath7 .
various parametrizations as discussed in the sec .
4.1 and 4.2 give only the tentative variation of @xmath44 and @xmath178 with the redshift .
however , we can put an upper bound on the variation of @xmath169 in the quintessence models . according to the table 1 , the dynamical age of the universe given by knox et al .
@xcite constrains @xmath179 to lie within the range @xmath180 which is consistent with wmap upper bound @xmath181 @xcite and satisfies the dark energy condition @xmath182 .
applying this condition to the integral in eq .
( 4 ) leads to an upper bound on the dimensionless dark energy function @xmath7 given by @xmath183 combining @xmath184 with @xmath185 we get the upper bound on @xmath52 and @xmath28@xmath186 and @xmath187 as shown in the fig . 2 , to check the validity of eq . ( 45 ) ,
let us choose a quintessence model with @xmath188constant .
the corresponding @xmath189 from table 1 at the transition redshift @xmath190 whereas eq .
( 45 ) yields the upper bound @xmath191 for @xmath190 , @xmath192 and @xmath31 .
( 45 ) can also be applied , in principle , to dark energy models with varying equation of state parameter @xmath0 using well - behaved parametrizations to yield the upper bound of the fractional dark energy density @xmath28 .
a new approach to the exploration of dark energy parameters is discussed based on the impact of the evolution of dark energy equation of state @xmath0 on the dynamical age of the universe .
the dynamical age depends upon the hubble expansion of the universe of which dark energy is presently a dominant constituent .
therefore , the age of the universe carries the signature of the dark energy and can be used as an effective constraint to check the physical viability of the various quiessence and kinessence models as discussed in this paper . unlike the conventional procedure of assuming a parametrization for @xmath0 and constraining it by fitting it to the observational data
, we have calculated the dynamical age of the universe assuming different parametrizations and tested the physical viability of these parametrizations with the cosmic age constraint derived from the cmb data @xcite .
direct comparison of the theoretically calculated age of the universe ( table 1 and 2 ) with the observational range of the cosmic age as laid down by cmb anisotropy measurements reveals that lee parametrizations @xcite satisfy both the astrophysical @xcite and the cosmic age constraint @xcite . of course
, the quintessence models in the range @xmath193 , the cosmological constant and a wide class of phantom models also satisfy the cosmic age constraint .
recently jassal , bagla and padmanabhan @xcite have carried out a detailed analysis of constraints on cosmological parameters from different observations with particular reference to equation of state parameter @xmath0 . according to their analysis , the sn ia observations alone favor phantom models ( @xmath194 ) with large @xmath195 whereas the wmap observations favor models with @xmath196 if dark energy perturbations are included . in the absence of any definite evidence of variation of dark energy density with the redshift ,
the cosmological constant still remains a favorite candidate for dark energy @xcite as it remains compatible with astrophysical observations @xcite and satisfies the cosmic age constraint as laid down in this paper as well as the recently published wmap three years data @xcite estimating @xmath197 @xmath198 and @xmath199 mpc@xmath14 km s@xmath14 .
the authors gratefully acknowledge the facilities and financial assistance for carrying out the research work done under the dst project at university of lucknow , lucknow , india .
the authors also thank the referee for the constructive suggestions for the improvement of paper . | the dynamical age of the universe depends upon the rate of the expansion of the universe , which explicitly involves the dark energy equation of state parameter @xmath0 .
consequently , the evolution of @xmath0 has a direct imprint on the age of the universe .
we have shown that the dynamical age of the universe as derived from cmb data can be used as an authentic criterion , being independent of the priors like the present value of the hubble constant @xmath1 and the cosmological density parameter @xmath2 , to constrain the range of admissible values of @xmath3 for quiessence models and to test the physically viable parametrizations of the equation of state @xmath4 in kinessence models .
an upper bound on variation of dark energy density is derived and a relation between cosmological density parameters and the transition redshift is established . |
the top quark has been experimentally observed at the fermilab tevatron by the cdf and d0 collaborations @xcite with a measured mass around 175 gev . because it has a mass of the order of the fermi scale , the top quark couples quite strongly to the electroweak symmetry - breaking sector . in the standard model ( sm ) ,
the electroweak symmetry - breaking sector consists of a complex fundamental higgs scalar , but `` triviality '' @xcite and `` naturalness '' @xcite of the scalar sector suggest that the higgs sector may not be so simple .
it is therefore plausible to assume that the mass generation mechanism is more complicated and the higgs sector of the sm is just an effective theory , and that new physics phenomena may be manifested through effective interactions of the top quark @xcite .
if anomalous top - quark couplings beyond the sm exist , they will affect top - quark production and decay processes at hadron and @xmath4 colliders @xcite .
furthermore , such couplings would certainly affect certain low - energy quantities which are measured with high precision .
one such possibility is the partial width ratio @xmath5 measured at lep - i , which is about 3@xmath6 higher than the standard model expectation @xcite and could be an indication of new physics associated with the heavy top quark . in ref .
@xcite , the experimental constraints on an anomalous top - quark coupling @xmath7 and the experimental observability of the induced rare decay mode @xmath8 , at the fermilab tevatron and the lhc at cern , have been investigated in detail . in this paper
, we examine the possible anomalous top - quark coupling @xmath9 and its implications in both low and high energy processes . in our analysis
we will also allow for the possibility of a @xmath10 coupling because it has a similar form to the @xmath9 coupling . since the standard model prediction of @xmath11 is unobservably small @xcite , any experimental evidence for the @xmath9 coupling will be an unequivocal indication of new physics beyond the standard model .
this paper is organized as follows . in section
ii , we examine the constraints on the anomalous @xmath12 couplings from @xmath13 and from the limit on non - standard top decays at the tevatron . in section iii , we study the possibility of detecting the @xmath9 coupling at the tevatron and the lhc , paying particular attention to the kinematical characteristics to extract the signal from the possible background .
the cuts are adjusted to the different physics requirements at the two accelerators .
finally , in section iv , we summarize our results .
following ref .
@xcite , we introduce an effective lagrangian involving the anomalous top - quark couplings , @xmath14 where @xmath15 is the standard model lagrangian and @xmath16 includes the anomalous top - quark couplings . for the purpose of this paper ,
we consider only the lowest dimension , @xmath17-conserving operators which give rise to anomalous @xmath18 vertices
. then @xmath19 + { \it h.c . } \ , \label{leff}\end{aligned}\ ] ] where @xmath20 ( @xmath21 ) is the @xmath22 @xmath23 field strength tensor ; @xmath24 ( @xmath25 ) is the corresponding coupling constant ; and @xmath26 is the cutoff of the effective theory , which we take to be the order of 1 tev . the parameters @xmath27 and @xmath28 can be interpreted as the strengths of the anomalous interactions , or , alternatively , @xmath29 is the approximate scale at which new physics in the top - quark sector occurs . the measurement of the inclusive branching ratio for the process @xmath30 @xcite puts severe constraints on various extensions of the standard model @xcite . here
we study its constraints on anomalous top - quark couplings @xmath9 and @xmath10 .
anomalous couplings of various types associated with the top quark and their constraints obtained from low - energy processes have been considered in the literature @xcite .
the effective hamiltonian for @xmath13 is given by @xcite , @xmath31 where @xmath32 runs from 1 to 8 , and @xmath33 denotes the energy scale at which @xmath34 is applied .
the operators set @xmath35 consist of six four - quark operators @xmath36 , the electromagnetic dipole operator @xmath37 , and the chromo - magnetic dipole operator @xmath38 . at the weak scale ,
the only contributing operator is @xmath37 .
however , when evolving down to the low energy scale @xmath39 , @xmath37 will mix with operator @xmath38 and others . here , @xmath40 are the elements of the cabbibo - kobayashi - maskawa mixing matrix . the partial decay width for @xmath13 is , neglecting the strange - quark mass , @xmath41 normalized by the inclusive semileptonic branching ratio , the @xmath42 branching ratio can be written as @xmath43 where @xmath44 , with @xmath45 . in the standard model , the complete leading logarithmic approximation gives @xcite @xmath46 with @xmath47 .
the coefficients @xmath48 , @xmath49 and @xmath50 can be found in @xcite .
the anomalous top - quark couplings @xmath9 and @xmath10 modify the coefficients of operators @xmath37 and @xmath38 .
the coefficients of these operators at the electroweak scale can be written as @xmath51 @xmath52 where @xmath53 and @xmath54 are the standard model contributions . using @xmath55 and the recent cleo measurement @xcite @xmath56 we plot the allowed region for @xmath27 and @xmath28 in fig . 1 , for @xmath57 gev , @xmath58 gev , @xmath59 and @xmath60 @xcite . throughout this paper , we take @xmath61 tev . in the figure
there are two allowed bands : the right band corresponds to a small anomalous correction to the sm @xmath62 coefficient which obeys the limit in eq .
( [ eq : bsglim ] ) , while the left band corresponds to a large correction which is approximately twice the magnitude of the sm contribution but of the opposite sign , so that the magnitude of @xmath62 is still consistent with eq .
( [ eq : bsglim ] ) . a further restriction on @xmath27 and @xmath28 can be obtained from the fact that @xmath63 and @xmath64 decays are not the most significant top decay modes as shown by the recent cdf data @xcite . a straightforward calculation of partial widths yields @xmath65 where the masses of the @xmath66 and @xmath1 quarks have been ignored . for the case of @xmath67 , @xmath68 is replaced by @xmath69 in the above equation . the cdf data on the branching ratio of top decaying to @xmath1 @xcite , @xmath70 places the limit @xmath71 at the one standard deviation level , where the insignificant contribution of @xmath72 has been ignored .
this constraint gives the limit @xmath73 and is shown in fig . 1 as the dashed lines .
one can see that @xmath74 combined with @xmath75 has constrained @xmath27 very strongly .
the branching ratio for @xmath76 is shown in fig .
2(a ) versus @xmath27 for the maximum and minimum allowed magnitudes of @xmath28 .
specifically , we find that @xmath77 when accounting for the contribution of the @xmath63 decay to the total width . a similar constraint for the case with @xmath78 has been obtained in ref .
the branching ratio for @xmath79 is shown in fig .
2(b ) versus @xmath28 .
at the fermilab tevatron , the cross section for @xmath80 production is about 5 pb at @xmath81 tev . a detailed study of top - quark decays is possible with the expected 1 fb@xmath82/yr integrated luminosity of the main injector , or the 10 fb@xmath82/yr anticipated at the tevatron upgrade @xcite .
top - quark production is much greater at the lhc , with a cross section being of order 1000 pb , due to the larger center of mass energy ( 14 tev ) and higher integrated luminosity ( 100 fb@xmath82/yr ) .
therefore it is feasible to search for evidence of anomalous couplings such as @xmath83 at hadron colliders . the decay @xmath64 has already been discussed in the previous section .
to obtain the signal event rate , we calculate the top - quark production via @xmath84 with the lowest order matrix elements , but normalize the total cross sections to values which include order @xmath85 corrections @xcite .
we thus find that the constant @xmath86 factors are @xmath87 for the parton distributions
we use the recent parameterization mrs set - a @xcite . due to the enormous qcd backgrounds at hadron colliders ,
it is very difficult , if not impossible , to search for the signal if both the top and antitop decay purely hadronically .
we therefore look for events with one top - quark decay leptonically @xmath88 ( or @xmath89 ) , where @xmath90 . the cross section for such events is calculated using the exact matrix elements for an on - shell @xmath91 . we have ignored the spin correlations for the decaying top quarks since the top - quark production mechanisms we consider give insignificant top - quark polarization @xcite .
figure 3 shows the calculated total cross section plotted versus @xmath27 , for two values of @xmath28 , for the process @xmath92 , where @xmath90 and @xmath93 represents a jet from either a @xmath1 or @xmath66 quark .
the curves stop at the maximum allowed value of @xmath27 in each case .
figure 3(a ) is for the fermilab tevatron ( @xmath94 collisions with @xmath95 tev ) , and fig .
3(b ) for the lhc ( @xmath96 collisions with @xmath97 tev ) .
we see that for the allowed values of @xmath27 , the signal cross section at the tevatron could be as large as 4 fb , which is rather small but still might be observable at the tevatron upgrade . at the lhc ,
the maximal signal cross section is nearly 600 fb , which means that there could be a large number of events even for non - maximal couplings .
we should remark that the top - quark production cross section is somewhat sensitive to the top - quark mass . for @xmath98 gev , for example
, the cross section is reduced by slightly more or less than 50% at the tevatron and the lhc , respectively .
the only irreducible backgrounds to the signal are electroweak ( ew ) @xmath99 production plus two qcd jets @xmath100 there are also possible `` fake '' events @xmath101 where one of the qcd jets fakes a photon .
these backgrounds have substantial production rates .
the signal is however distinctive : besides the isolated charged lepton and large missing transverse energy ( @xmath102 ) from @xmath103 decay , there is a highly energetic photon , and two hard ( @xmath104-quark ) jets with which @xmath105 and @xmath106 both reconstruct a top ( or @xmath107 ) . without requiring @xmath1-tagging ,
our signal selection procedure is as follows : we first examine the two values of the invariant mass @xmath108 and @xmath109 and identify the jet which gives an @xmath110 value closer to @xmath111 as the @xmath66-quark jet .
naturally , the other jet will be identified as @xmath1-quark jet . due to the missing neutrino from the @xmath91 decay ,
the @xmath91 momentum ca nt be reconstructed unambiguously due to the lack of knowledge of the parton c.m
this is known as the two - fold ambiguity in constructing the neutrino momentum along the beam direction @xcite . taking the @xmath91 mass as an input and for massless leptons , using the measured charged lepton momentum ( @xmath112 ) and the _ transverse _ momentum ( with respect to the beam direction ) of the neutrino ( @xmath113 )
, the two solutions for the _ longitudinal _ momentum of the neutrino can be expressed by @xmath114^{1/2 } \biggr\ } \>. \label{plnu}\end{aligned}\ ] ] there are therefore two solutions for @xmath115 , correspondingly .
we again take the one that gives an @xmath116 value closer to @xmath111 . with this procedure ,
both top - quark momenta are experimentally identifiable . to make the calculation more realistic
, we simulate the detector effects by assuming a gaussian energy smearing for the electromagnetic and hadronic calorimetry as follows : @xmath117 where the @xmath118 indicates that the @xmath119-dependent and @xmath119-independent errors are to be added in quadrature , and @xmath119 is to be measured in gev . to quantify the experimental sensitivity to the anomalous couplings
, we first impose acceptance cuts on the transverse momentum ( @xmath120 ) , pseudo - rapidity ( @xmath121 ) , and the separation in the azimuthal angle - pseudo rapidity plane ( @xmath122 ) for the charged lepton and photon from the jets .
we choose the following `` basic '' acceptance cuts : @xmath123 the high transverse momentum thresholds for the jets and the photon are motivated by the hard @xmath120 spectrum from the heavy top decay . with these basic cuts , the signal rate in fig .
3(a ) is reduced by about 40% ( with a maximal value of about 2.7 fb ) , while the ew irreducible background of eq .
( [ eq : ew ] ) is about 30 fb , and the qcd process for @xmath124 production in eq .
( [ eq : qcd ] ) is about 3000 fb .
it has been shown that the fake rate for a jet to a photon ( @xmath125 ) at the tevatron experiments @xcite can be controlled down to a level of @xmath126 for @xmath127 gev , making it insignificant .
therefore , the ew background dominates and we will consequently concentrate on that .
figure 4(a ) shows the reconstructed distributions for the top - antitop invariant mass . obviously , @xmath128 for the signal has a kinematical lower limit ( @xmath129 if no energy smearing is applied ) ; while the lower limit is significantly smaller for the background , near the @xmath130 threshold . if we only accept events with @xmath131 we expect to improve the signal - to - background ratio ( @xmath132 ) .
also , due to the nature of top - quark two - body decay , the final state jets and the photon have transverse momenta typically the order of @xmath133 gev , while all the jets and photon in the background events tend to be soft . to further discriminate the signal from the background , we define two scalar sums of the transverse momenta : @xmath134 figure 4(b )
shows the distributions for these variables .
we see indeed that the signal spectra are much harder at the low end . on the other hand , they are limited by the physical scale @xmath129 for the signal at the high end , while they extend further for the background . with these in mind , our choices of the cuts are @xmath135 with the additional cuts of eqs .
( [ eq : mtt ] ) and ( [ eq : ptsum ] ) , the ew background is reduced to about 10 fb , while the signal may be as large as 2.4 fb . in fig .
5 we show the reconstructed top - quark mass distributions @xmath136 and @xmath116 , after making the additional cuts in eqs .
( [ eq : mtt ] ) and ( [ eq : ptsum ] )
. we see from fig .
5(a ) that , despite the improvement in @xmath132 , the continuum background is still above the @xmath136 signal peak at @xmath111 .
further improvement can be made if we only study the events with @xmath137 the short - dashed histogram in fig .
5(b ) shows how the background in the @xmath116 distribution is reduced by the cut eq .
( [ eq : mcg ] ) .
we find that the signal peak is nearly unaffected by this cut and is now above the background , so that statistically significant effects may be observed near @xmath111 in the @xmath116 spectrum . if we examine the events in the mass range @xmath138 the signal observability should be nearly optimized .
after the cut in eq .
( [ eq : mbw ] ) we are left with a maximal signal of 2.0 fb and an ew background 1.8 fb . at the lhc energy , due to the more complicated hadronic backgrounds , we need to increase the transverse momentum thresholds , especially for jets , direct photons and missing @xmath139 .
we adopt the `` basic '' acceptance cuts as @xmath140 the signal rates given in fig .
3(b ) are reduced by about 60% with the cuts of eq .
( [ eq : basiclhc ] ) ( with a maximal value of about 0.22 pb ) , while the ew background is about 0.5 pb , and the `` faked '' qcd background is about 70 pb . assuming that we can effectively reject the qcd background by photon identification to a level of @xmath141 for @xmath142 @xcite , then the dominant background is again the electroweak process .
we see that after the basic cuts of eq .
( [ eq : basiclhc ] ) the s / b ratio is much closer to unity than in the tevatron case .
similar to the tevatron study , we have constructed distributions for the top - antitop invariant mass and scalar sums of transverse momenta ( see fig . 6 ) .
we see that further background suppression can be achieved , without much loss of signal , by the cuts of eqs .
( [ eq : mtt ] ) and ( [ eq : ptsum ] ) .
the signal integrated cross section after these cuts may be as large as 200 fb , while the ew background rate is about 300 fb .
as at the tevatron , reconstructed top - quark mass distributions are also crucial to non - ambiguously identify the signal . figure 7 shows the top - quark mass distributions reconstructed from the @xmath105 and @xmath106 final states after making the cuts in eqs .
( [ eq : mtt ] ) and ( [ eq : ptsum ] ) . once the final cuts in eqs .
( [ eq : mcg ] ) and ( [ eq : mbw ] ) are made , the signal integrated cross section has a maximal value of 175 fb and the ew background is 33 fb .
we see that the s / b ratio is significantly better than at the tevatron , primarily due to the huge enhancement for the @xmath143 production at higher c.m .
energies .
the cdf collaboration were able to achieve about 50% @xmath1-tagging efficiency @xcite and one hopes to reach about the same efficiency for the lhc experiments @xcite .
the impurity from the light quarks / gluons is assumed to be 1% .
we therefore can naively expect to further suppress the backgrounds by a factor of @xmath144 ( where @xmath145 is the number of jets ) at a cost of signal reduction of 50% .
one has to be cautious in making this statement since there is direct production of @xmath146 among the other light quark / gluon jets in the background events .
we have explicitly calculated the @xmath147 production rate and found the cross section is about 0.12 fb at the tevatron and 1.1 fb at the lhc with the basic cuts plus eqs .
( [ eq : mtt ] ) and ( [ eq : ptsum ] ) .
this implies that the misidentified @xmath1 from light quarks / gluons is indeed the major background source .
consequently , we will not carry out the calculation for the @xmath146 fraction in @xmath148 events , since this channel is always smaller to begin with .
table [ t : one ] lists the signal rate ( for @xmath149 ) , ew and qcd backgrounds at the tevatron and the lhc .
comparison is made for the results with only cuts of eq .
( [ eq : basic ] ) for tevatron or eq .
( [ eq : basiclhc ] ) for lhc and eqs .
( [ eq : mtt])and ( [ eq : ptsum ] ) , and those plus @xmath1-tagging
. we indeed obtain significant improvement for the signal / background ratios , although the signal rate at the tevatron is very limited .
to estimate the sensitivity to the anomalous couplings for a given accumulated luminosity , we first establish the approximate cross section formula for the @xmath150 case in terms of the anomalous couplings : @xmath151 where @xmath152 is calculated for the appropriate set of cuts , and the couplings are subject to the constraints in fig . 1 .
the cross section @xmath152 is 80 fb and 6.8 pb at the tevatron and lhc , respectively , after the full set of cuts is made .
we start with poisson statistics since the signal rate is often small , especially at the tevatron energy .
figure 8 presents the anomalous coupling @xmath27 versus the luminosity needed to probe @xmath27 at 95% confidence level ( cl ) . in gaussian statistics when the number of events is large . ]
the solid curves presented in each panel are for @xmath78 and the dashed for @xmath153 .
the upper curves for each case are those with kinematical cuts only , while the lower ones are that including @xmath1-tagging improvement .
we see that at the tevatron energy , minimal luminosity of about 5 fb@xmath82 is needed in order to probe @xmath27 near the current low energy constraints at 95% cl . at the lhc with 100 fb@xmath82
, one expects to improve the sensitivity to @xmath27 by more than one order of magnitude .
for instance , for @xmath78 and with @xmath1-tagging , the sensitivity at the tevatron reaches @xmath154 and at the lhc @xmath155 alternatively , setting @xmath156 and letting @xmath26 vary ( see eq .
( [ leff ] ) ) , we can view this as a probe of the scale of new physics up to @xmath157 tev at the tevatron and about 100 tev at the lhc .
if @xmath158 , there of course is also the possibility of @xmath159 decay , which is only mildly constrained by current tevatron data .
although the hadronic nature of the decay might make the detection of this mode difficult , the jets resulting from this decay will have large transverse momentum , and may allow one to distinguish the signal from the background .
there is also the possibility of single top production in association with a charm quark @xcite .
the phenomenology of the @xmath10 coupling at hadron colliders is currently under study @xcite .
we would like to thank liangxin li for collaboration in the calculation of the @xmath160 limits during an early stage of this work .
we thank roberto peccei for discussions .
bly would like to thank hai - yang cheng and shih - chang lee for warm hospitality extended to him during his sabbatical leave at the institute of physics , academia sinica , taipei , when part of the work was performed .
he would also like to thank the national science council for financial support .
xz thanks g. buchalla for discussions .
this work was supported in part by the u.s .
department of energy under contracts de - fg03 - 91er40674 ( t. han ) and de - fg02 - 94er40817 ( k.whisnant , b .-
young and x.zhang ) .
.cross sections in units of fb for the @xmath161 signal ( @xmath149 ) , and ew and qcd backgrounds at the tevatron and the lhc .
comparison is made for the results with only cuts eqs .
( [ eq : basic ] or [ eq : basiclhc ] ) and ( [ eq : mtt]-[eq : ptsum ] ) and those plus @xmath1-tagging .
50% tagging efficiency and 1% impurity are assumed @xcite . [ cols="^,^,^,^",options="header " , ] for a recent review , see , _ e. g. _ , h. neuberger , in _ proceedings of the xxvi international conference on high energy physics _ , dallas , texas , 1992 , edited by j. sanford , aip conf .
272 ( aip , new york , 1992 ) , vol .
ii , p. 1360 .
g. t hooft , in _ recent developments in gauge theories _ ,
proceedings of the cargese summer institute , cargese , france , 1979 , edited by g. t hooft et al .
, nato advanced study institute series b ; physics vol .
59 ( plenum , new york , 1980 ) .
b. grzadkowski , j.f .
gunion and p. krawczyk , phys .
b268 * , 106 ( 1991 ) ; g. eilam , j.l . hewett and a. soni ,
phys . rev . * d44 * , 1473 ( 1991 ) ; m. luke and m.j .
savage , phys .
b307 * , 387 ( 1993 ) ; g. couture , c. hamzaoui and h. knig , phys
. rev . * d52 * , 1713 ( 1995 ) . j. stroughair and c. bilchak , z. phys .
* c26 * , 415 ( 1984 ) ; j. gunion , z. kunszt , and m. soldate , phys . lett . *
b163 * , 389 ( 1985 ) ; j. gunion and m. soldate , phys . rev . *
d34 * , 826 ( 1986 ) ; w. j. stirling _ et al .
b163 * , 261 ( 1985 ) ; j. cortes , k. hagiwara , and f. herzog , nucl . phys . *
b278 * , 26 ( 1986 ) .
e. g. _ , h. aihara , in proceedings of _ international symposium on vector boson self - interactions _
_ , los anglese , ca , feb .
1995 , p. 72 .
fig . 1 constraints on anomalous couplings @xmath28 and @xmath27 from @xmath163 ( the two diagonal bands in solid lines ) and @xmath164 ( dashed horizontal lines ) . @xmath61
tev is assumed throughout the paper .
2 ( a ) branching ratios for ( a ) @xmath76 versus @xmath165 for @xmath166 ( solid line ) and @xmath167 ( dashed line ) , and ( b ) @xmath79 versus @xmath168 .
the curves stop at the maximum allowed value of the abscissa in each case .
3 production cross section for @xmath169 at ( a ) the tevatron ( @xmath170 ) with @xmath81 tev , and ( b ) the lhc ( @xmath96 ) with @xmath171 tev .
the top - quark mass @xmath172 gev is used throughout the paper .
the cross sections are plotted versus @xmath165 for @xmath166 ( solid line ) and @xmath153 ( dashed line ) .
the curves stop at the maximum allowed value of @xmath165 in each case .
4 ( a ) top - antitop invariant mass distributions for the signal @xmath173 ( solid histogram ) and background @xmath174 ( dashed histogram ) at the tevatron with @xmath175 tev , assuming the basic acceptance cuts defined in eq .
( [ eq : basic ] ) in the text .
( b ) distributions for the transverse momenta scalar sums @xmath176 and @xmath177 for the signal ( solid histograms ) and background ( dashed histograms ) .
the signal histograms assume the maximal allowed value of @xmath27 with @xmath78 .
fig . 5 reconstructed top - quark mass distributions for ( a ) @xmath136 and ( b ) @xmath116 at the tevatron with @xmath175 tev . in ( a )
, the solid histogram shows the signal @xmath173 and the short dashed histogram shows the background @xmath174 after applying basic cuts and cuts on the top - antitop invariant mass and the transverse momenta scalar sums @xmath176 and @xmath177 , while in ( b ) the same curves are used to show the signal and background after making the further cut on @xmath136 . the long dashed histogram in ( b ) shows the @xmath116 distribution of the background before @xmath136 cut is made ( the effect of the @xmath136 cut on the signal is minimal ) .
the signal histograms assume the maximal allowed value of @xmath27 with @xmath78 .
fig . 6 ( a ) top - antitop invariant mass distributions for the signal @xmath173 ( solid histogram ) and background @xmath174 ( dashed histogram ) at the lhc with @xmath178 tev , assuming the basic acceptance cuts defined in eq .
( [ eq : basiclhc ] ) in the text .
( b ) distributions for the transverse momenta scalar sums @xmath176 and @xmath177 for the signal ( solid histograms ) and background ( dashed histograms ) .
the signal histograms assume the maximal allowed value of @xmath27 with @xmath78 .
7 reconstructed top - quark mass distributions for ( a ) @xmath136 and ( b ) @xmath116 at the lhc with @xmath178 tev . in ( a )
, the solid histogram shows the signal @xmath173 and the short dashed histogram shows the background @xmath174 after applying basic cuts and cuts on the top - antitop invariant mass and the transverse momenta scalar sums @xmath176 and @xmath177 , while in ( b ) the same curves are used to show the signal and background after making the further cut on @xmath136 .
the long dashed histogram in ( b ) shows the @xmath116 distribution of the background before @xmath136 cut is made ( the effect of the @xmath136 cut on the signal is minimal ) .
the signal histograms assume the maximal allowed value of @xmath27 with @xmath78 .
8 95% cl sensitivity to @xmath27 vs. integrated luminosity for ( a ) the tevatron with @xmath81 tev and ( b ) the lhc with @xmath171 tev .
we consider the limits for @xmath78 ( solid lines ) and @xmath153 ( dashed lines ) .
the upper curves in each case correspond to the limit obtained when comparing the signal and background after making all the cuts , while the lower curves correspond to the limits obtained when using @xmath1-tagging with the cuts in eqs .
( [ eq : basic ] or [ eq : basiclhc ] ) and ( [ eq : mtt]-[eq : ptsum ] ) .
the curves cut off at the maximal allowed value of @xmath27 in each case . | we determine the constraints on the anomalous top - quark coupling associated with the flavor - changing neutral current vertex @xmath0 from the limits on the @xmath1-quark rare decay @xmath2 and non - standard top - quark decays . based on these constraints
, we discuss the experimental observability of the rare decay mode @xmath3 , both at the fermilab tevatron with a luminosity - upgrade and at the lhc . |
in order to fix ideas , it is useful to give an explicit `` neural network '' interpretation to the theory that will be developed .
the model will consist of 2 layers of nodes .
the input layer has a `` pattern of activity '' that represents the components of the input vector @xmath0 , and the output layer has a pattern of activity that is the collection of activities of each output node .
the activities in the output layer depend only on the activities in the input layer .
if an input vector @xmath0 is presented to this network , then each output node `` fires '' discretely at a rate that corresponds to its activity .
after @xmath1 nodes have fired the probabilistic description of the relationship between the input and output of the network is given by @xmath2 , where @xmath3 is the location in the output layer ( assumed to be on a rectangular lattice of size @xmath4 ) of the @xmath5 node that fires .
in this paper it will be assumed that the order in which the @xmath1 nodes fire is not observed , in which case @xmath2 is a sum of probabilities over all @xmath6 permutations of @xmath7 , which is a symmetric function of the @xmath3 , by construction .
the theory that is introduced in section [ subsect : probencodedecode ] concerns the special case @xmath8 . in the @xmath8 case
the probabilistic description @xmath9 is proportional to the firing rate of node @xmath10 in response to input @xmath0 .
when @xmath11 there is an indirect relationship between the probabilistic description @xmath2 and the firing rate of node @xmath10 , which is given by the marginal probability @xmath12 it is important to maintain this distinction between events that are observed ( i.e. @xmath7 given @xmath0 ) and the probabilistic description of the events that are observed ( i.e. @xmath2 ) .
the only possible exception is in the @xmath13 limit , where @xmath2 has all of its probability concentrated in the vicinity of those @xmath7 that are consistent with the observed long - term average firing rate of each node .
it is essential to consider the @xmath11 case to obtain the results that are described in this paper .
a theory of self - organising networks based on an analysis of a probabilistic encoder / decoder was presented in @xcite .
it deals with the @xmath8 case referred to in section [ subsect : nninterp ] .
the objective function that needs to be minimised in order to optimise a network in this theory is the euclidean distortion @xmath14 defined as @xmath15 where @xmath0 is an input vector , @xmath10 is a coded version of @xmath0 ( a vector index on a @xmath16-dimensional rectangular lattice of size @xmath4 ) , @xmath17 is a reconstructed version of @xmath0 from @xmath10 , @xmath18 is the probability density of input vectors , @xmath19 is a probabilistic encoder , and @xmath20 is a probabilistic decoder which is specified by bayes theorem as @xmath21 @xmath14 can be rearranged into the form @xcite @xmath22 where the reference vectors @xmath23 are defined as @xmath24 although equation [ eq : objective ] is symmetric with respect to interchanging the encoder and decoder , equation [ eq : objectivesimplified ] is not .
this is because bayes theorem has made explicit the dependence of @xmath25 on @xmath19 . from a neural network viewpoint @xmath19
describes the feed - forward transformation from the input layer to the output layer , and @xmath23 describes the feed - back transformation that is implied from the output layer to the input layer .
the feed - back transformation is necessary to implement the objective function that has been chosen here .
minimisation of @xmath14 with respect to all free parameters leads to an optimal encoder / decoder . in equation
[ eq : objectivesimplified ] the @xmath19 are the only free parameters , because @xmath23 is fixed by equation [ eq : refvect ] .
however , in practice , both @xmath19 and @xmath23 may be treated as free parameters @xcite , because @xmath23 satisfy equation [ eq : refvect ] at stationary points of @xmath14 with respect to variation of @xmath23 .
the probabilistic encoder / decoder requires an explicit functional form for the posterior probability @xmath19 .
a convenient expression is @xmath26 where @xmath27 can be regarded as a node `` activity '' , and @xmath28 .
any non - negative function can be used for @xmath29 , such as a sigmoid ( which satisfies @xmath30 ) @xmath31 where @xmath32 and @xmath33 are a weight vector and bias , respectively .
a drawback to the use of equation [ eq : postprob ] is that it does not permit it to scale well to input vectors that have a large dimensionality .
this problem arises from the restricted functional form allowed for @xmath29 .
a solution was presented in @xcite @xmath34 where @xmath35 , and @xmath36 is a set of lattice points that are deemed to be `` in the neighbourhood of '' the lattice point @xmath10 , and @xmath37 is the inverse neighbourhood defined as the set of lattice points that have lattice point @xmath10 in their neighbourhood .
this expression for @xmath19 satisfies @xmath38 ( see appendix [ app : postprobpmdnorm ] ) .
it is convenient to define @xmath39 which is another posterior probability , by construction .
it includes the effect of the output nodes that are in the neighbourhood of node @xmath40 only .
@xmath41 is thus a localised posterior probability derived from a localised subset of the node activities .
this allows equation [ eq : postprobpmd ] to be written as @xmath42 , so @xmath19 is the average of the posterior probabilities at node @xmath10 arising from each of the localised subsets that happens to include node @xmath10 .
the model may be extended to the case where @xmath1 output nodes fire .
@xmath19 is then replaced by @xmath43 , which is the probability that @xmath44 are the first @xmath1 nodes to fire ( in that order ) . with this modification
, @xmath14 becomes @xmath45 where the reference vectors @xmath46 are defined as @xmath47 the dependence of @xmath43 and @xmath46 on @xmath1 output node locations complicates this result .
assume that @xmath43 is a symmetric function of its @xmath44 arguments , which corresponds to ignoring the order in which the first @xmath1 nodes choose to fire ( i.e. @xmath43 is a sum over all permutations of @xmath44 ) .
for simplicity , assume that the nodes fire independently so that @xmath48 ( see appendix [ app : upperbound ] for the general case where @xmath49 does not factorise ) .
@xmath14 may be shown to satisfy the inequality @xmath50 ( see appendix [ app : upperbound ] ) , where @xmath51 @xmath52 and @xmath53 are both non - negative .
@xmath54 as @xmath13 , and @xmath55 when @xmath56 , so the @xmath52 term is the sole contribution to the upper bound when @xmath56 , and the @xmath53 term provides the dominant contribution as @xmath13 .
the difference between the @xmath52 and the @xmath53 terms is the location of the @xmath57 average : in the @xmath53 term it averages a vector quantity , whereas in the @xmath52 term it averages a euclidean distance .
the @xmath53 term will therefore exhibit interference effects , whereas the @xmath52 term will not .
the model may be further extended to the case where the probability that a node fires is a weighted average of the underlying probabilities that the nodes in its vicinity fire .
thus @xmath19 becomes @xmath58 where @xmath59 is the conditional probability that node @xmath10 fires given that node @xmath40 would have liked to fire . in a sense , @xmath59 describes a `` leakage '' of probability from node @xmath40 that onto node @xmath10 .
@xmath59 then plays the role of a soft `` neighbourhood function '' for node @xmath40 .
this expression for @xmath19 can be used wherever a plain @xmath19 has been used before .
the main purpose of introducing leakage is to encourage neighbouring nodes to perform a similar function .
this occurs because the effect of leakage is to soften the posterior probability @xmath19 , and thus reduce the ability to reconstruct @xmath0 accurately from knowledge of @xmath10 , which thus increases the average euclidean distortion @xmath14 . to reduce the damage that leakage causes
, the optimisation must ensure that nodes that leak probability onto each other have similar properties , so that it does not matter much that they leak .
the focus of this paper is on minimisation of the upper bound @xmath60 ( see equation [ eq : upperboundpieces ] ) to @xmath14 in the multiple firing model , using a scalable posterior probability @xmath19 ( see equation [ eq : postprobpmd ] ) , with the effect of activity leakage @xmath59 taken into account ( see equation [ eq : leakage ] ) .
gathering all of these pieces together yields @xmath61 where @xmath62 . in order to ensure that the model is truly scalable
, it is necessary to restrict the dimensionality of the reference vectors . in equation [ eq : objectivemodel ] @xmath63 , which is not acceptable in a scalable network . in practice
, it will be assumed any properties of node @xmath10 that are vectors in input space will be limited to occupy an `` input window '' of restricted size that is centred on node @xmath10 .
this restriction applies to the node reference vector @xmath23 , which prevents @xmath60 from being fully minimised , because @xmath23 is allowed to move only in a subspace of the full - dimensional input space .
however , useful results can nevertheless be obtained , so this restriction is acceptable .
optimisation is achieved by minimising @xmath60 with respect to its free parameters .
thus the derivatives with respect to @xmath23 are given by @xmath64 and the variations with respect to @xmath29 are given by @xmath65 the functions @xmath66 , @xmath67 , @xmath68 , and @xmath69 are derived in appendix [ app : derivatives ] . inserting a sigmoidal function @xmath70 then yields the derivatives with respect to @xmath32 and @xmath33 as @xmath71 because all of the properties of node @xmath10 that are vectors in input space ( i.e. @xmath23 and @xmath32 ) are assumed to be restricted to an input window centred on node @xmath10 , the eventual result of evaluating the right hand sides of the above equations must be similarly restricted to the same input window .
the expressions for @xmath52 and @xmath53 , and especially their derivatives , are fairly complicated , so an intuitive interpretation will now be presented .
when @xmath60 is stationary with respect to variations of @xmath23 it may be written as ( see appendix [ app : d1d2refvect ] ) . @xmath72
the @xmath73 and @xmath74 factors do not appear in this expression because @xmath19 is normalised to sum to unity . the first term ( which derives from @xmath52 )
is an incoherent sum ( i.e. a sum of euclidean distances ) , whereas the second term ( which derives from @xmath53 ) is a coherent sum ( i.e. a sum of vectors ) .
the first term contributes for all values of @xmath1 , whereas the second term contributes only for @xmath75 , and dominates for @xmath76 . in order to minimise the first term the @xmath77 like to be as large as possible for those nodes that have a large @xmath19 .
since @xmath23 is the centroid of the probability density @xmath78 , this implies that node @xmath10 prefers to encode a region of input space that is as far as possible from the origin .
this is a consequence of using a euclidean distortion measure @xmath79 , which has the dimensions of @xmath80 , in the original definition of the distortion in equation [ eq : objective ] . in order to minimise the second term the superposition of @xmath23 weighted by
the @xmath19 likes to have as large a euclidean norm as possible .
thus the nodes co - operate amongst themselves to ensure that the nodes that have a large @xmath19 also have a large @xmath81 .
the purpose of this section is to work through a case study in order to demonstrate the various properties that emerge when @xmath60 is minimised .
it convenient to begin by ignoring the effects of leakage @xmath82 , and to concentrate on a simple ( non - scaling ) version of the posterior probability model ( as in equation [ eq : postprob ] ) @xmath83 , where the @xmath84 are threshold functions of @xmath0@xmath85 it is also convenient to imagine that a hypothetical infinite - sized training set is available , so it may be described by a probability density @xmath18 .
this is a `` frequentist '' , rather than a `` bayesian '' , use of the @xmath18 notation , but the distinction is not important in the context of this paper .
assume that @xmath86 is drawn from a training set , that has 2 statistically independent subspaces , so that @xmath87 furthermore , assume that @xmath88 and @xmath89 each have the form @xmath90 i.e. @xmath91 is a loop ( parameterised by a phase angle @xmath92 ) of probability density that sits in @xmath93-space . in order to make it easy to deduce the optimum reference vectors ,
choose @xmath94 so that the following 2 conditions are satisfied for @xmath95@xmath96 this type of training set can be visualised topologically .
each training vector @xmath97 consists of 2 subvectors , each of which is parameterised by a phase angle , and which therefore lives in a subspace that has the topology of a circle , which is denoted as @xmath98 . because of the independence assumption in equation [ eq : independentpdf ] , the pair @xmath97 lives on the surface of a 2-torus , which is denoted as @xmath99 .
the minimisation of @xmath60 thus reduces to finding the optimum way of designing an encoder / decoder for input vectors that live on a 2-torus , with the proviso that their probability density is uniform ( this follows from equation [ eq : parametricpdf ] and equation [ eq : parametricpdfconstraint ] ) . in order to derive the reference vectors
@xmath23 , the solution(s ) of the stationarity condition @xmath100 must be computed .
the stationarity condition reduces to ( see appendix [ app : d1d2refvect ] ) @xmath101 it is useful to use the simple diagrammatic notation shown in figure [ fig : s1s1 ] .
[ fig : s1s1 ] topology with a threshold @xmath102 superimposed.,title="fig:",width=377 ] each circle in figure [ fig : s1s1 ] represents one of the @xmath98 subspaces , so the two circles together represent the product @xmath99 .
the constraints in equation [ eq : parametricpdfconstraint ] are represented by each circle being centred on the origin of its subspace ( @xmath103 is constant ) , and the probability density around each circle being constant ( @xmath104 is constant ) .
a single threshold function @xmath84 is represented by a chord cutting through each circle ( with 0 and 1 indicating on which side of the chord the threshold is triggered ) .
the @xmath93 that lie above threshold in each subspace are highlighted . both @xmath105 and
@xmath106 must lie above threshold in order to ensure @xmath107 , i.e. they must both lie within regions that are highlighted in figure [ fig : s1s1 ] . in this case
node @xmath10 will be said to be `` attached '' to both subspace 1 and subspace 2 .
a special case arises when the chord in one of the subspaces ( say it is @xmath106)does not intersect the circle at all , and the circle lies on the side of the chord where the threshold is triggered . in this case
@xmath84 does not depend on @xmath106 , so that @xmath108 , in which case node @xmath10 will be said to be `` attached '' to subspace 1 but `` detached '' from subspace 2 .
the typical ways in which a node becomes attached to the 2-torus are shown in figure [ fig : torus ] .
[ fig : torus ] topology as a torus with the effect of 3 different types of threshold @xmath102 shown.,title="fig:",width=377 ] in figure [ fig : torus](a ) the node is attached to one of the @xmath98 subspaces and detached from the other . in figure
[ fig : torus](b ) the attached and detached subspaces are interchanged with respect to figure [ fig : torus](a ) . in figure
[ fig : torus](c ) the node is attached to both subspaces .
consider the configuration of threshold functions shown in figure [ fig : attachone ] .
this is equivalent to all of the nodes being attached to loops to cover the 2-torus , with a typical node being as shown in figure [ fig : torus](a ) ( or , equivalently , figure [ fig : torus](b ) ) .
[ fig : attachone ] when @xmath60 is minimised , it is assumed that the 4 nodes are symmetrically disposed in subspace 1 , as shown .
each is triggered if and only if @xmath105 lies within its quadrant , and one such quadrant is highlighted in figure [ fig : attachone ] .
this implies that only 1 node is triggered at a time .
the assumed form of the threshold functions implies @xmath109 , so equation [ eq : stationaryrefvect ] reduces to @xmath110 whence @xmath111 consider the configuration of threshold functions shown in figure [ fig : attachboth ] .
this is equivalent to all of the nodes being attached to patches to cover the 2-torus , with a typical node being as shown in figure [ fig : torus](c ) .
[ fig : attachboth ] in this case , when @xmath60 is minimised , it is assumed that each subspace is split into 2 halves .
this requires a total of 4 nodes , each of which is triggered if , and only if , both @xmath105 and @xmath106 lie on the corresponding half - circles .
this implies that only 1 node is triggered at a time .
the assumed form of the threshold functions implies that the stationarity condition becomes @xmath112 whence @xmath113 consider the configuration of threshold functions shown in figure [ fig : attacheither ] .
this is equivalent to half of the nodes being attached to loops to cover the 2-torus , with a typical node being as shown in figure [ fig : torus](a ) .
the other half of the nodes would then be attached in an analogous way , but as shown in figure [ fig : torus](b ) .
thus the 2-torus is covered twice over .
[ fig : attacheither ] in this case , when @xmath60 is minimised , it is assumed that each subspace is split into 2 halves .
this requires a total of 4 nodes , each of which is triggered if @xmath105 ( or @xmath106 ) lies on the half - circle in the subspace to which the node is attached .
thus exactly 2 nodes @xmath114 and @xmath115 are triggered at a time , so that @xmath116 for simplicity , assume that node @xmath117 is attached to subspace 1 , then @xmath118 and the stationarity condition becomes @xmath119 this may be simplified to yield @xmath120 write the 2 subspaces separately ( remember that node @xmath117 is assumed to be attached to subspace 1 ) @xmath121 if this result is simultaneously solved with the analogous result for node @xmath117 attached to subspace 2 , then the @xmath122 terms vanish to yield @xmath123 consider the left hand side of figure [ fig : attachone ] for the case of @xmath73 nodes , when the @xmath73 threshold functions form a regular @xmath73-ogon .
@xmath124 then denotes the part of the circle that is associated with node @xmath117 , whose radius of gyration squared is given by ( assuming that the circle has unit radius ) @xmath125 gather the results for @xmath126 in equations [ eq : refvectone ] ( referred to as type 1 ) , [ eq : refvectboth ] ( referred to as type 2 ) , and [ eq : refvecteither ] ( referred to as type 3 ) together and insert them into @xmath60 in equation [ eq : d1d2refvect ] to obtain ( see appendix [ app : d1d2compare ] ) @xmath127 in figure [ fig : plotn1 ] the 3 solutions are plotted for the case @xmath8 .
[ fig : plotn1 ] for @xmath8 for each of the 3 types of optimum.,title="fig:",width=377 ] for @xmath8 the type 3 solution is never optimal , the type 1 solution is optimal for @xmath128 , and the type 2 solution is optimal for @xmath129 this behaviour is intuitively sensible , because a larger number of nodes is required to cover a 2-torus as shown in figure [ fig : torus](c ) than as shown in figure [ fig : torus](a ) ( or figure [ fig : torus](b ) ) . in figure [
fig : plotn2 ] the 3 solutions are plotted for the case @xmath130 . [
fig : plotn2 ] for @xmath130 for each of the 3 types of optimum.,title="fig:",width=377 ] for @xmath130 the type 1 solution is optimal for @xmath131 , and the type 2 solution is optimal for large @xmath132 , but there is now an intermediate region @xmath133 ( type 1 and type 3 have an equal @xmath60 at @xmath134 ) where the @xmath1-dependence of the type 3 solution has now made it optimal .
again , this behaviour is intuitively reasonable , because the type 3 solution requires at least 2 observations in order to be able to yield a small euclidean resonstruction error in each of the 2 subspaces , i.e. for @xmath130 the 2 nodes that fire must be attached to different subspaces .
note that in the type 3 solution the nodes that fire are not guaranteed to be attached to different subspaces . in the type 3 solution
there is a probability @xmath135 that @xmath136 ( where @xmath137 ) nodes are attached to subspace @xmath138 , so the trend is for the type 3 solution to become more favoured as @xmath1 is increased . in figure
[ fig : plotnasymptotic ] the 3 solutions are plotted for the case @xmath13 .
[ fig : plotnasymptotic ] for @xmath13 for each of the 3 types of optimum.,title="fig:",width=377 ] for @xmath139 the type 2 solution is never optimal , the type 1 solution is optimal for @xmath140 , and the type 3 solution is optimal for @xmath141 .
the type 2 solution approaches the type 3 solution from below asymptotically as @xmath142 . in figure
[ fig : phasediagram ] a phase diagram is given which shows how the relative stability of the 3 types of solution for different @xmath73 and @xmath1 , where the type 3 solution is seen to be optimal over a large part of the @xmath143 plane .
[ fig : phasediagram ] thus the most interesting , and commonly occurring , solution is the one in which half the nodes are attached to one subspace and half to the other subspace ( i.e. solution type 3 ) .
although this result has been derived using the non - scaling version of the posterior probability model @xmath9 ( as in equation [ eq : postprob ] ) , it may also be used for scaling posterior probabilities ( as given in equation [ eq : postprobpmd ] ) in certain limiting cases , and also for cases where the effect of leakage @xmath82 is small .
the effect of leakage will not be analysed in detail here . however , its effect may readily be discussed phenomenologically , because the optimisation acts to minimise the damaging effect of leakage on the posterior probability by ensuring that the properties of nodes that are connected by leakage are similar .
this has the most dramatic effect on the type 3 solution , where the way in which the nodes are partitioned into 2 halves must be very carefully chosen in order to minimise the damage due to leakage .
if the leakage is presumed to be a local function , so that @xmath144 , which is a localised `` blob''-shaped function , then the properties of adjacent node are similar ( after optimisation ) . since nodes that are attached to 2 different subspaces necessarily have very different properties , whereas nodes that are attached to the same subspace can have similar properties , it follows that the nodes must split into 2 continguous halves , where nodes @xmath145 are attached to subspace 1 and nodes @xmath146 are attached to subspace 2 , or vice versa . the effect of leakage is thereby minimised , with the worst effect occurring at the boundary between the 2 halves of nodes .
the above analysis has focussed on the non - scaling version of the posterior probability , in which all @xmath73 nodes act together as a unit .
the more general scaling case where the @xmath73 nodes are split up by the effect of the neighbourhood function @xmath147 will not be analysed in detail , because many of its properties are essentially the same as in the non - scaling case . for simplicity
assume that the neighbourhood function @xmath147 is a `` top - hat '' with width @xmath148 ( an odd integer ) centred on @xmath117 .
impose periodic boundary conditions so that the inverse neighbourhood function @xmath149 is also a top - hat , @xmath150 . in this case
an optimum solution in the non - scaling case ( with @xmath151 ) can be directly related to a corresponding optimum solution in the scaling case by simply repeating the node properties periodically every @xmath148 nodes . strictly speaking
, higher order periodicities can also occur in the scaling case ( and can be favoured under certain conditions ) , where the period is @xmath152 ( @xmath153 is an integer ) , but these will not be discussed here .
the effect of the periodic replication of node properties is interesting . the type 3 solution ( with leakage and with @xmath154 splits the nodes into 2 halves , where nodes @xmath155 are attached to subspace 1 and nodes @xmath156 are attached to subspace 2 , or vice versa .
when this is replicated periodically every @xmath148 nodes it produces an alternating structure of node properties , where @xmath157 nodes are attached to subspace 1 , then the next @xmath157 nodes are attached to subspace 2 , and thenthe next @xmath157 nodes are attached to subspace 1 , and so on .
this behaviour is reminiscent of the so - called `` dominance stripes '' that are observed in the mammalian visual cortex .
the purpose of this section is to demonstrate the emergence of the dominance stripes in numerical simulations .
the main body of the software is concerned with evaluating the derivatives of @xmath60 , and the main difficulty is choosing an appropriate form for the leakage ( this has not yet been automated ) .
the parameters that are required for a simulation are as follows : 1 .
@xmath158 : size of 2d rectangular array of nodes .
@xmath160 : size of 2d rectangular input window for each node ( odd integers ) .
ensure that the input window is not too many input data `` correlation areas '' in size , otherwise dominance stripes may not emerge .
dominance stripes require that the correlation _ within _ an input window are substantially stronger than the correlations _ between _ input windows that are attached to different subspaces .
@xmath161 : size of 2d rectangular neighbourhood window for each node ( odd integers ) .
the neighbourhood function @xmath162 is a rectangular top - hat centred on @xmath163 .
the size of the neighbourhood window has to lie within a limited range to ensure that dominance stripes are produced .
this corresponds to ensuring that @xmath73 lies in the type 3 region of the phase diagram in figure [ fig : phasediagram ] .
it is also preferable for the size of the neighbourhood window to be substantially smaller than the input window , otherwise different parts of a neighbourhood window will see different parts of the input data , which will make the network behaviour more difficult to interpret .
4 . @xmath164 : size of 2d rectangular leakage window for each node ( odd integers ) . for simplicity the leakage @xmath82
is assumed to be given by @xmath165 , where @xmath166 is a `` top - hat '' function of @xmath167 which covers a rectangular region of size @xmath164 centred on @xmath168**. * * the size of the leakage window must be large enough to correlate the parameters of adjacent nodes , but not so large that it enforces such strong correlations between the node parameters that it destroys dominance stripes . 5 .
@xmath169 : additive noise level used to corrupt each member of the training set
@xmath170 : wavenumber of sinusoids used in the training set . in describing
the training sets the index @xmath117 will be used to denote position in input space , thus position @xmath117 in input space lies directly `` under '' node @xmath117 of the network . in 1d simulations
each training vector is a sinusoid of the form @xmath171 , where @xmath172 is a random phase angle , and @xmath173 is a random number sampled uniformly from the interval @xmath174 $ ] - this generates an @xmath98 topology training set ( i.e. parameterised by 1 random angle ) . in 2d simulations
each training vector is a sinusoid of the form @xmath175 , where the additional angle @xmath176 is a random azimuthal orientation for the sine wave - this generates an @xmath99 topology training set ( i.e. parameterised by 2 independent random angles ) .
note that @xmath177 and @xmath178 must be an integer multiple of @xmath179 in order to ensure that the probability density around the @xmath98 subspace generated by @xmath172 has uniform density ( in effect , the @xmath98 then becomes a circular lissajous figure , which therefore has uniform probability density , unlike non - circular lissajous figures ) , and thus to ensure that there are no artefacts induced by the periodicity of the training data that might mimic the effect of dominance stripes .
if @xmath177 and @xmath178 are much greater than @xmath179 then it is not necessary to fix them to be integer multiples of @xmath179 - because the fluctuations in the probability density are then negligible .
note that this restriction on the value of @xmath170 would not have been necessary had complex exponentials been used rather than sinusoids .
@xmath180 : number of subspaces .
this fixes the number of statistically independent subspaces in the training set .
when @xmath181 the training set is generated exactly as above .
when @xmath182 the training set is split up as follows .
the 1d case has even @xmath117 in one subspace , and odd @xmath117 in the other subspace , thus successive components of each training vector alternate between the 2 subspaces .
the 2d case has even @xmath183 in one subspace , and odd @xmath183 in the other subspace , thus each training vector is split up into a chessboard pattern of interlocking subspaces .
this strategy readily generalises for @xmath184 , although this is not used here . within each subspace
the training vector is generated as above , and the subspaces are generated so that they are statistically independent .
@xmath185 : update parameter used in gradient descent .
this is used to update parameters thus @xmath186 there are 3 internally generated update parameter , which control the update of the 3 different types of parameter , i.e. the biases , the weights , and the reference vectors .
this is necessary because these parameters all have different dimensionalities , and by inspection of equation [ equpdateprescription ] the dimensionality of an update parameter is the dimensionality of the parameter it updates ( squared ) divided by the dimensionality of the euclidean distortion .
these 3 internal parameters are automatically adjusted to ensure that the average change in absolute value of each of the 3 types of parameter is equal to @xmath185 times the typical diameter of the region of parameter space populated by the parameters .
this adjustment is made anew as each training vector is presented .
the size of @xmath185 determines the `` memory time '' of the node parameters .
this memory time determines the effective number of training vectors that the nodes are being optimised against , and thus must be sufficiently long ( i.e. @xmath185 sufficiently small ) that if @xmath187 it is possible to discern that the subspaces are indeed statistically independent .
this is crucially important , for dominance stripes can not be obtained if the subspaces are not sufficiently statistically independent .
so @xmath185 must be small , which unfortunately leads to correspondingly long training times .
the training set is globally translated and scaled so that the components of all of its training vectors lie in the interval @xmath188 $ ] .
there are 3 parameter types to initialise .
the weights were all initialised to random numbers sampled from a uniform distribution in the interval @xmath189 $ ] , whereas the biasses and the reference vector components were all initialised to 0 .
because the 2d simulations took a very long time to run , they were periodically interrupted and the state of all the variables written to an output file .
the simulation could then be continued by reading this output file in again and simply continuing where the simulation left off .
alternatively , some of the variables might have their values changed before continuing .
in particular , the random number generator could thus be manipulated to simulate the effect of a finite sized training set ( i.e. use the _ same _ random number seed at the start of each part of the simulation ) , or an infinite - sized training set ( i.e. use a _ different _ random number seed at the start of each part of the simulation ) .
the size of the @xmath185 parameter could also thus be manipulated should a large value be required initially , and reduced to a small value later on , as required in order to guarantee that when @xmath75 the input subspaces are seen to be statistically independent , and dominance stripes may emerge .
there are many ways to choose the boundary conditions . in the numerical simulations
periodic boundary conditions will be avoided , because they can lead to artefacts in which the node parameters become topologically trapped .
for instance , in a 2d simulation , periodic boundary conditions imply that the nodes sit on a 2-torus .
leakage implies that the node parameter values are similar for adjacent nodes , which limits the freedom for the parameters to adjust their values on the surface of the 2-torus .
for instance , any acceptable set of parameters that sits on the 2-torus can be converted into another acceptable set by mapping the 2-torus to itself , so that each of its @xmath98 `` coils up '' an integer number of times onto itself . such a multiply wrapped parameter configuration
is topologically trapped , and can not be perturbed to its original form .
this problem does not arise with non - periodic boundary conditions .
there are several different problems that arise at the boundaries of the array of nodes : 1 .
the neighbourhood function @xmath162 can not be assumed to be a rectangular top - hat centred on @xmath163 .
instead , it will simply be truncated so that it does not fall off the edge of array of nodes , i.e. @xmath190 for those @xmath163 that lie outside the array .
2 . the leakage function @xmath191 will be similarly truncated . however , in this case @xmath191 must normalise to unity when summed over @xmath163 , so the effect of the truncation must be compensated by scaling the remaining elements of @xmath191 .
the input window for each node implies that the input array must be larger than the node array in order that the input windows never fall off the edge of the input array .
the most important result is the emergence of dominance stripes . for @xmath130
there are thus 2 numbers that need to be displayed for each node : the `` degree of attachment '' to subspace 1 , and similarly for subspace 2 .
there are many ways to measure degree of attachment , for instance the probability density @xmath25 gives a direct measurement of how strongly node @xmath10 depends on the input vector @xmath0 , so its `` width '' or `` volume '' in each of the subspaces could be used to measure degree of attachment . however , in the simulations presented here ( i.e. sinusoidal training vectors ) the degree of attachment is measured as the average of the absolute values of the components of the reference vector in the subspace concerned .
this measure tends to zero for complete detachment . for 1d simulations
2 dominance plots can be overlaid to show the dominance of subspaces 1 and 2 for each node . for 2d simulations it is simplest to present only 1 of these plots as a 2d array of grey - scale pixels , where the grey level indicates the dominance of subspace 1 ( or , alternatively , subspace 2 ) .
the parameter values used were : @xmath192 , @xmath193 , @xmath194 , @xmath195 , @xmath196 , @xmath197 , @xmath182 , @xmath198 , @xmath199 .
this value of @xmath170 implies @xmath200 , so @xmath170 is approximately an integer multiple of @xmath179 , as required for an artefact - free simulation . in figure
[ fig : dominance1d ] a plot of the 2 dominance curves obtained after 3200 training updates is shown .
[ fig : dominance1d ] this dominance plot clearly shows alternating regions where subspace 1 dominates and subspace 2 dominates .
the width of the neighbourhood function is 21 , which is the same the period of the variations in the dominance plots , i.e. within each set of adjacent 21 nodes half the nodes are attached to subspace 1 and half to subspace 2 .
there are boundary effects , but these are unimportant .
the normalisation of the expression for @xmath19 in equation [ eq : postprobpmd ] may be demonstrated as follows : @xmath201 in the first step the order of the @xmath10 and the @xmath40 summations is interchanged using @xmath202 , in the second step the numerator and denominator of the summand cancel out .
it is possible to simplify equation [ eq : objectivemultifire ] by using the following identity @xmath203 note that this holds for all choices of @xmath204 .
this allows the euclidean distance to be expanded thus @xmath205 each term of this expansion can be inserted into equation [ eq : objectivemultifire ] to yield @xmath206 @xmath43 has been assumed to be a symmetric function of @xmath44 in the first two results , and the definition of @xmath46 in equation [ eq : refvectmultifire ] has been used to obtain the third result .
these results allow @xmath14 in equation [ eq : objectivemultifire ] to be expanded as @xmath207 , where @xmath208 by noting that @xmath209 , an upper bound for @xmath14 in the form @xmath50 follows immediately from these results .
note that @xmath210 whereas @xmath53 can have either sign . in the special case where @xmath211 ( i.e. @xmath212 and @xmath213 are independent of each other given that @xmath214 is known )
@xmath53 reduces to @xmath215 which is manifestly positive .
this is the form of @xmath53 that is used throughout this paper .
@xmath52 and @xmath53 are as given in equation [ eq : upperboundpieces ] , i.e. it is assumed that @xmath216 and @xmath19 has the scalable form given in equation [ eq : postprobpmd ] . define a compact matrix notation as follows @xmath217 using this matrix notation , the functions @xmath66 , @xmath67 , @xmath68 , and @xmath69 may be defined as @xmath218 the variation of @xmath41 is then given by @xmath219 in order to rearrange the expressions to ensure that only a single dummy index is required at every stage of evaluation of the sums it will be necessary to use the result @xmath220 the derivative is given by @xmath222 use matrix notation to write this as @xmath223 finally remove the explicit summations to obtain the required result @xmath224 the derivative is given by @xmath226 use matrix notation to write this as @xmath227 finally remove the explicit summations to obtain the required result @xmath228 the differential is given by @xmath230 use matrix notation to write this as @xmath231 reorder the summations to obtain @xmath232 relabel the indices and evaluate the sum over the kronecker delta to obtain @xmath233 finally remove the explicit summations to obtain the required result @xmath234 the differential is given by @xmath236 use matrix notation to write this as @xmath237 reorder the summations to obtain @xmath238 relabel the indices and evaluate the sum over the kronecker delta to obtain @xmath239 finally remove the explicit summations to obtain the required result @xmath240
from equation [ eq : upperboundpieces ] @xmath60 can be written as @xmath242 where the constant terms do not depend on @xmath23 . however , from equation [ eq : upperboundpieces ] the derivative @xmath243 can be written as @xmath244 using bayes theorem the stationarity condition @xmath100 yields a matrix equation for the @xmath23@xmath245 which may then be used to replace all instances of @xmath0 in equation [ eq : upperboundrefvectpieces ] .
this yields the result @xmath246
in order to compare the value of @xmath60 that is obtained when different types of supposedly optimum configurations of the threshold functions @xmath84 are tried , the @xmath23 that solves @xmath100 ( see appendix [ app : d1d2refvect ] ) must be inserted into the expression for @xmath60 . in the following derivations
the constant term is omitted , and the definition @xmath247 ( see equation [ eq : radiussquared ] ) has been used . [ [ type-3-optimumhalf - the - nodes - are - attached - one - subspace - and - half - are - attached - to - the - other ] ] type 3 optimum : half the nodes are attached one subspace and half are attached to the other ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | this paper shows how a folded markov chain network can be applied to the problem of processing data from multiple sensors , with an emphasis on the special case of 2 sensors .
it is necessary to design the network so that it can transform a high dimensional input vector into a posterior probability , for which purpose the partitioned mixture distribution network is ideally suited .
the underlying theory is presented in detail , and a simple numerical simulation is given that shows the emergence of ocular dominance stripes . |
after the advent of relativistic heavy ion collider ( rhic ) , much attention has been paid for the properties of matter near and above the critical temperature of qcd phase transition , @xmath1 . to understand the structure of the matter in this region , it is desirable to identify the basic degrees of freedom of the system and their excitation properties . in the present study
, we analyze the spectral properties of quarks , which are one of the fundamental degrees of freedom of qcd , at finite temperature above and below @xmath1 using quenched lattice qcd with clover improved fermions @xcite . at asymptotically high temperatures
, one can calculate the quark propagator using perturbative techniques .
it is known that the collective excitations of quarks in this limit develop a mass gap ( thermal mass ) that is proportional to @xmath2 , where @xmath3 and @xmath0 denote the gauge coupling and temperature , respectively @xcite . moreover , in this limit the number of poles in the quark propagator is doubled ; in addition to the normal modes , which reduce to poles in the free particle propagator , plasmino modes appear . in order to see the properties of quark spectral function @xmath4 in this limit ,
let us first consider the spectral function at zero momentum .
in this case , the dirac structure of @xmath4 can be decomposed using the projection operators @xmath5 as @xmath6 while @xmath7 for free quarks are given by @xmath8 , in the high temperature limit one obtains @xmath9 /2 $ ] , where @xmath10 is the thermal mass .
the two poles at @xmath11 correspond to the normal and plasmino modes .
the dirac structure of @xmath4 in the high temperature limit is also decomposed as @xmath12 with the projection operators @xmath13 and @xmath14 .
the spectral functions @xmath15 in the high temperature limit read @xmath16 where @xmath17 represents the contribution of the continuum taking non - zero values in the space - like region , and poles at @xmath18 and @xmath19 corresponds to the normal and plasmino modes , respectively .
the dispersion relation of the plasmino has a minimum at nonzero @xmath20 @xcite .
in this section , we analyze the quark spectral function for zero momentum but with finite bare quark mass on the lattice .
we use gauge field ensembles which have been generated and used previously by the bielefeld group to study screening masses and spectral functions @xcite .
the quark spectral function in this case is decomposed into @xmath7 as in eq .
( [ eq : rho^m ] ) . in order to extract the spectral function @xmath21 from the lattice correlation function
, we assume a simple ansatz for the shape of @xmath21 including few fitting parameters . we found that the two - pole ansatz for @xmath21 , @xmath22 reproduces the lattice correlator quite well , where @xmath23 , and @xmath24 are fitting parameters determined from the correlated fit . the pole at @xmath25 in eq .
( [ eq:2pole ] ) corresponds to the plasmino mode .
the success of two - pole ansatz for the quark correlation functions suggests that the positions of the poles of the quark propagator would be near the real axis at @xmath26 and @xmath27 with small imaginary parts .
provided that the positions of poles of the propagator are gauge independent , this also indicates that our results on the fitting parameters @xmath28 and @xmath29 have small gauge dependence . and the relative strength of the plasmino mode , @xmath30 , at @xmath31 , @xmath32 and @xmath33 obtained from calculations on lattice of size @xmath34 . ] in fig .
[ fig : ze ] , we show the dependence of @xmath28 , @xmath29 and @xmath30 on the bare quark mass , @xmath35 , for @xmath31 , @xmath32 , and @xmath33 obtained from calculations on lattices of size @xmath34 .
the figure shows that the ratio @xmath36 becomes larger with decreasing @xmath35 and eventually reaches @xmath37 irrespective of @xmath0 .
the numerical result for each @xmath0 shows that @xmath38 is satisfied within statistical errors there .
these results mean that the quark propagator is chirally symmetric at this point @xcite . moreover , @xmath21 at this point has the same form as the spectral function in the high temperature limit .
we therefore define the thermal mass of the quark on the lattice as @xmath39 at this point .
one finds that the ratio @xmath40 is insensitive to @xmath0 in the range analyzed in this work , although it becomes slightly larger with decreasing @xmath0 , which would be in accordance with the expected parametric form at high temperature , @xmath41 .
figure [ fig : ze ] also shows that the relative strength of the plasmino pole , @xmath42 , decreases with increasing values of the bare mass , @xmath35 .
the spectral function @xmath21 thus will eventually be dominated by a single - pole .
this result agrees with the perturbative result in yukawa models that @xmath21 approaches the spectral function of free quarks as the bare quarks mass becomes larger @xcite . for @xmath43
, we found that the lattice correlator is concave in the log - scale plot .
this behavior indicates that the positivity condition of @xmath21 is violated below @xmath1 .
in fact , we have checked that the two - pole ansatz eq .
( [ eq:2pole ] ) gives unacceptably large @xmath44 below @xmath1 .
it is also found that the quark correlator does not approach the chirally symmetric one even in the chiral limit , in contrast to the previous result for @xmath45 .
these results for @xmath43 seem consistent with a nave picture that quark excitations are confined and the chiral symmetry is spontaneously broken below @xmath1 .
and @xmath29 and the ratio @xmath42 on the lattice momentum @xmath46 for @xmath47 and @xmath33 . ]
next let us analyze the quark spectral function at finite momentum on lattices with size @xmath34 for @xmath47 and @xmath33 . throughout this section
we consider the quark propagator in the chiral limit .
the quark propagator in the chiral limit is decomposed into @xmath48 according to eq .
( [ eq : rho^p ] ) .
following the same approach used in the previous section , we adopt the two - pole ansatz @xmath49 and determine four parameters from a correlated fit .
the @xmath50-functions at @xmath26 and @xmath27 correspond to the normal and plasmino modes , respectively .
we found that @xmath51 with this ansatz is always smaller than @xmath32 for all momenta analyzed in this study .
this result means that the two - pole ansatz again reproduces the lattice correlation function well . in fig .
[ fig : p>0 ] , we show the momentum dependence of the fitting parameters @xmath28 , @xmath29 , and @xmath42 for @xmath47 and @xmath33 .
the horizontal axis represents the momentum on the lattice @xmath52 .
the figure shows that for large momentum @xmath42 rapidly decreases and @xmath28 approaches the light cone .
the spectral function at large momentum therefore approaches that of a free quark , consistent with the perturbative result .
one also finds that @xmath29 is always smaller than @xmath28 , in contrast to the results in the previous section .
this behavior qualitatively agrees with the behavior of poles in the high @xmath0 limit @xcite .
one also observes from fig .
[ fig : p>0 ] that @xmath29 enters the space - like region at high momentum .
an interesting property of the quark propagator in the high temperature limit eq .
( [ eq : rho^p_htl ] ) is that the dispersion relation of the plasmino has a minimum at finite momentum . in fig .
[ fig : p>0 ] , one sees that the value of @xmath29 at lowest non - zero momentum on our lattice , @xmath53 , is slightly larger than that at zero momentum , and the existence of such a minimum is suggested but not yet confirmed in the present analysis . in summary
, we analyzed the dependence of the quark spectral function on temperature @xmath0 , bare quark mass @xmath54 , and momentum @xmath20 in quenched lattice qcd with landau gauge fixing .
above @xmath1 , we found that the two - pole approximations for the spectral functions in the projected channels , @xmath7 and @xmath48 , can well reproduce the lattice correlation functions .
although further studies on the volume dependence is needed , this result indicates that the excitations of quarks have small decay width even near @xmath1 . below @xmath1 ,
on the other hand , the two - pole ansatz fails completely .
f. karsch _ et al .
_ , phys .
b * 530 * , 147 ( 2002 ) ; s. datta , f. karsch , p. petreczky and i. wetzorke , phys .
d * 69 * , 094507 ( 2004 ) .
g. baym , j. p. blaizot and b. svetitsky , phys .
d * 46 * , 4043 ( 1992 ) . | we analyze the quark spectral function above and below the critical temperature for deconfinement and at finite momentum in quenched lattice qcd .
it is found that the temporal quark correlation function in the deconfined phase near the critical temperature is well reproduced by a two - pole ansatz for the spectral function .
the bare quark mass and momentum dependences of the spectral function are analyzed with this ansatz . in the chiral limit
we find that even near the critical temperature the quark spectral function has two collective modes corresponding to the normal and plasmino excitations in the high temperature ( @xmath0 ) limit .
the pole mass of these modes at zero momentum , which should be identified to be the thermal mass of the quark , is approximately proportional to @xmath0 in a rather wide range of @xmath0 in the deconfined phase . |
observing from the cape of good hope , herschel ( 1847 ) was the first to note the peculiar nature of ngc 5128 : _ ` a nebula consisting of two lateral portions and no doubt of a small streak of nebula along the middle of the slit or interval between them , having a star at its extremity ' _ and _ ` ... a very problematical object , and must be regarded as forming a genus apart , since it evidently differs from mere ` double nebulae ' ... ' _ .
although hubble ( 1922 ) included the galaxy in a list of ( mostly galactic ) nebulae with line emission , it was mostly ignored until the advent of radio astronomy about 50 years ago ( but see paraskevoulos 1935 ) . shortly after world war
ii , bolton ( 1948 ) announced the discovery of six discrete sources of 100 mhz radio emission in addition to the already known case of cygnus a. precisely a year later , using their remarkable ` cliff - interferometry ' technique , bolton , stanley @xmath1 slee ( 1949 ) identified the sources taurus a , virgo a and centaurus a with the nebulous objects m1 ( crab nebula ) , ngc 4486 ( m 87 ) and ngc 5128 respectively .
although they were actually convinced of the extragalactic nature of the latter two , their published discussion was noncommittal , reflecting referee objections to this notion ( r.d .
ekers , private communication ) .
more precise measurements by mills ( 1952 ) confirmed the identification of centaurus a with ngc 5128 . discussing the nature of optical counterparts to the newly discovered radio sources , baade @xmath1 minkowski ( 1954 ) established ngc 5128 as an unusual extragalactic system , located in space well ` beyond m 31 ' .
they described it as _ ` ... an unresolved e0 nebula with an unusually strong and wide central dark band , a combination highly anomalous for a spherical nebula ' _ and interpreted it as the interaction or merger of an elliptical and a spiral galaxy .
nevertheless , the merger hypothesis then receded into the background until it was revived some 25 years later by the work of graham ( 1979 ) , dufour et al .
( 1979 ) , tubbs ( 1980 ) and malin , quinn @xmath1 graham ( 1983 ) .
it is now the favoured explanation for the unusual characteristics of ngc 5128 .
other radio measurements soon established the nonthermal nature of the radio emission ( haddock , mayer @xmath1 sloanaker 1954 ) .
subsequent radio observations with increased sensitivity yielded ever - larger dimensions for the object until the low - frequency studies by sheridan ( 1958 ) and shain ( 1958 ) first established an overall size of 8@xmath2 @xmath3 4@xmath2 for centaurus a much larger than the 20@xmath4 size of the optical parent galaxy ngc 5128 .
a multi - frequency radio study by cooper , price @xmath1 cole ( 1965 ) showed that at decimetre wavelengths the source consists of bright central emission , representing 20@xmath5 of the total flux density and much larger lower - brightness lobes to the north and south , with 40@xmath5 of the radio flux each . the advent of interferometric techniques , in particular aperture synthesis radio telescopes with their superior angular resolution , has since led to the discovery of many more similar radio galaxies ( _ cf .
_ allen , hanbury brown @xmath1 palmer 1963 ; see review by miley 1980 ) .
it has also become clear that many other elliptical galaxies have more or less pronounced dust lanes ( hawarden et al .
1981 ; sadler @xmath1 gerhard 1985 ; van dokkum @xmath1 franx 1995 ) , including the other major southern radio galaxy ngc 1316 ( fornax a ) . both
its striking optical appearance ( fig . 1 ) , and
its associated giant radio source have made ngc 5128 humphreys ( 1982 ) and zickgraf et al .
( 1990 ) , reaching down to visual magnitudes of 16.6 , 18.2 and 22.7 magnitudes respectively .
astrometric coordinates of field stars near the centre of ngc 5128 can be found in griffin ( 1963 ) .
very accurate field star positions and magnitudes may be obtained from http://astro.estec.esa.nl/sa-general/projects/hipparcos . ] into one of the most extensively studied galaxies in the southern hemisphere . however , in their comprehensive review ebneter @xmath1 balick ( 1983 ) point out that ngc 5128 if placed at similar large distances would look very much like other radio elliptical galaxies , in spite of its apparent peculiarity . by its fortuitous proximity ,
centaurus a uniquely allows detailed studies aimed at determining the nature of the galaxy and in particular the origin of the giant radio source it is hosting .
a probably incomplete search of the literature published since ebneter @xmath1 balick s ( 1983 ) review reveals over a hundred refereed papers mentioning centaurus a in the title , whereas in the same period the number of papers referring to centaurus a was almost 700 .
this review focusses on the observed properties of centaurus a / ngc 5128 and their immediate interpretation . for a more complete review of the early work up to 1983
, the reader is referred to ebneter @xmath1 balick ( 1983 ) , and for more general information to recent proceedings such as ` the second stromlo symposium the nature of elliptical galaxies ' ( eds .
n. arnoboldi , g.s .
da costa , p. saha , 1997 pasp conference series vol .
116 ) , ` energy transport in radio galaxies and quasars ' ( eds .
hardee , a.h .
bridle , j.a .
zensus , 1996 pasp conference series vol .
100 ) and ` extragalactic radio sources ' ( eds .
r. ekers , c. fanti , l. padrielli , 1996 iau symposium 175 ) . with its great apparent brightness and galactocentric velocity of about @xmath6 = + 325 @xmath7 ,
ngc 5128 is clearly a nearby galaxy .
early estimates of its distance ranged from 2.1 mpc ( sersic 1958 ) to 8.5 mpc ( sandage @xmath1 tammann 1974 ) and a ` compromise ' distance of 5 mpc has frequently been used in the past .
many of the earlier distance values were derived by indirect means : group membership , redshift , comparison to virgo cluster distance ( assuming the latter to be known ) etc .
summaries of the various distance determinations prior to 1993 can be found in shopbell , bland - hawthorn @xmath1 malin ( 1993 ) and especially de vaucouleurs ( 1993 ) .
attempts to use the occurrence of the snia supernova 1986 g in ngc 5128 to derive its distance ( frogel et al .
1987 ; phillips et al . 1987 ;
ruiz - lapente et al .
1991 ) were frustrated by uncertainties in the internal reddening of ngc 5128 and in the absolute magnitudes of the supernova ( see sect .
it appears more useful to _ assume a distance _ to determine the supernova properties than the other way around .
other methods have been more succesful , consistently yielding a relatively nearby distance .
the accuracies of these techniques were reviewed by jacoby et al .
tonry @xmath1 schechter ( 1990 ) interpret the globular cluster counts by harris et al .
( 1984a , 1986 ) to imply a distance modulus @xmath8 = [email protected] mag .
planetary nebula counts by hui et al .
( 1993a ) yield @xmath8 = [email protected] .
the lower value @xmath8 = [email protected] obtained by tonry @xmath1 schechter ( 1990 ) from the surface brightness fluctuation of globular clusters must be revised to @xmath8 = [email protected] ( tonry 1991 ) .
finally , soria et al . ( 1996 ) estimate @xmath8 = [email protected] from halo red giant branch stars observed with hst / wfpc2 .
as there is no a priori preference for any of these methods , and as the weighted and unweighted averages of the various derived distance moduli are practically identical , the best value is thus @xmath8 = [email protected] , corresponding to a distance of @xmath10 = [email protected] mpc .
note that at this distance , 1@xmath4 on the sky corresponds almost precisely to 1 kpc .
= 3.4 mpc so that they may be different from the published values as given in the associated reference . ]
the systemic velocity of ngc 5128 is most frequently derived from measurements of components of the disk ( sect .
an overview of estimates up to 1984 can be found in hesser et al .
( 1984 ) ; table 1 lists all estimates * a. * later than 1975 ; * b. * having a quoted accuracy of 10 @xmath7 or better ; and * c. * _ not _ based on hi or molecular line absorption ( see sect . 7 ) .
the estimates range from 536 to 551 @xmath7 , and the mean is @xmath11 = 543@xmath92 @xmath7 .
lll systemic velocity & obtained from : & reference + @xmath11 ( @xmath7 ) & & + 548@xmath95 & nebular emission lines & graham ( 1979 ) + 551.4 & h@xmath12 emission & whiteoak @xmath1 gardner ( 1979 ) + 541@xmath98 & nebular emission lines & rodgers @xmath1 harding ( 1980 ) + 545@xmath95 & h@xmath12 emission & marcelin et al .
( 1982 ) + 538@xmath910 & stellar absorption lines & wilkinson et al .
( 1986 ) + 536@xmath95 & h@xmath12 emission & bland et al .
( 1987 ) + 542@xmath97 & hi emission & van gorkom et al ( 1992 ) + 541@xmath95 & co modelling & quillen et al .
( 1992 ) + 541@xmath97 & pn emission lines & hui et al .
( 1995 ) + based on position and radial velocity , ngc 5128 is part of a group of 25 galaxies extending over about 25@xmath2 on the sky ( de vaucouleurs 1975 ; webster et al . 1979 ; hesser et al . 1984 ; ct 1995 , 1997 ) .
in addition to ngc 5128 , major members of the group are ngc 4945 and ngc 5236 ( m 83 ) as well as the lesser ngc 5102 , ngc 5253 and some twenty dwarf galaxies ( fig .
ngc 5128 is the only massive elliptical in the group .
the group galactocentric velocity is very close to the individual velocities
@xmath6 = 320330 @xmath7 of the major members ngc 4945 , ngc 5128 and ngc 5236 ( hesser et al .
the members have an average projected radial distance to the centre of mass of 0.72 mpc .
the group has a crossing time of 5 gigayear and is probably still collapsing ( ct 1995 , 1997 ) .
assuming a distance of 3.4 mpc , the group mass estimate by hesser et al .
( 1984 ) becomes 517 @xmath3 10@xmath13 @xmath14 .
it is quite remarkable that the other two major members , ngc 4945 and ngc 5236 , as well as ngc 5253 , exhibit signs of unusually vigorous star formation , while ngc 4945 also has a nuclear outflow .
in addition , the dwarfs in the group have relatively high central surface brightnesses and exhibit clear star formation activity ( ct 1995 ) .
ngc 5128 , itself a strong x - ray source ( sects .
2.4 ; 5.5 ) , is surrounded by several x - ray point sources of a nature still unknown , although the majority of them seems to be associated with the galaxy ( arp 1994 ; dbereiner et al .
owing to its proximity , centaurus a ( pks 1322 - 427 ) is the largest extragalactic radio source in the sky .
it extends predominantly in a north - south direction between declinations @xmath15 , and between right ascensions 13@xmath1615@xmath17 13@xmath1632@xmath18 ( fig .
its overall angular dimensions are 8@xmath19 4@xmath20 ( cooper et al .
1965 ; haynes et al . 1983 ;
junkes et al .
1993 ; combi @xmath1 romero 1997 ) .
this translates into a linear size of 500@xmath3 250 kpc .
although larger radio galaxies have been found ( _ cf .
_ miley 1980 ) , centaurus a is still , in an absolute sense , one of the largest known . its moderate radio luminosity places it in class i of the classification by fanaroff @xmath1 riley ( 1974 ) , suggesting only moderate relavistic beaming effects ( jones et al . 1993 ) .
for a recent study of the properties distinguishing fr - i and fr - ii radio galaxies , see zirbel ( 1996 ) , and references therein .
the radio source is very complex ; it shows significant structure ranging over a factor of 10@xmath21 in size from the largest scales down to less than a milli - arcsecond .
major components of the radio source ( see fig
. 3 ) are the giant outer lobes extending to 250 kpc , the northern middle lobe ( no southern counterpart ) extending to about 30 kpc , the inner lobes and central jets extending to about 5 and 1.35 kpc respectively and the compact core with associated nuclear jets extending over 1 pc ( see fig .
table 2 provides a guide to radio continuum observations of centaurus a. part a lists radio continuum maps covering the _ whole _ source ; integrated flux - densities are not always provided by the authors .
total source flux - densities between 85 mhz and 4750 mhz define an overall spectral index @xmath12 = @xmath22 is defined by @xmath23 .
part b refers to aperture synthesis radio maps of the bright inner part which comprises the inner lobes , jets and core , and single - dish maps at very high frequencies .
references to other early single - dish maps and flux - densities at various frequencies can be found in slee et al .
( 1983 ) and tateyama @xmath1 strauss ( 1994 ) .
part c , finally , summarizes radio images and flux - densities of the nuclear source .
very high resolution ( vlbi ) observations have shown a complex structure even for the inner few parsec which is further discussed in sect . 5.4 . here ,
we draw attention to the high - frequency variability of the source which is clear from the table and which is further discussed in sect . 5.6 . lrcr reference & frequency & resolution & total flux + & ( mhz ) & & ( jy ) + + shain ( 1958 ) & 20 & & 28000 + sheridan ( 1958 ) & 85 & & @xmath247400 + cooper et al .
( 1965 ) & 406 & 48@xmath4 & 2710 + haslam et al .
( 1981 ) & 408 & 48@xmath4 & + bolton @xmath1 clark ( 1960 ) & 960 & 20@xmath4 & @xmath241675 + cooper et al .
( 1965 ) & 1410 & 14@xmath4 & 1330 + combi @xmath1 romero ( 1997 ) & 1435 & 30@xmath4 & + cooper et al .
( 1965 ) & 2650 & 7.4@xmath4 & + junkes et al .
( 1993 ) & 4750 & 4.3@xmath4 & 681 + haynes et al .
( 1983 ) & 5000 & 4.1@xmath4 & + + slee et al .
( 1983 ) & 327 & 55@xmath25 & 734 + slee et al .
( 1983 ) & 843 & 40@xmath2660@xmath25 & 392 + christiansen et al .
( 1977 ) & 1415 & 50@xmath2645@xmath25 & 215 + clarke et al .
( 1992 ) & 1446 & [email protected]@xmath25 & 78.3 + clarke et al . ( 1992 ) & 1634 & [email protected]@xmath25 & 63.5 + clarke et al .
( 1992 ) & 4866 & [email protected]@xmath25 & 37.5 + tateyama @xmath1 strauss ( 1992 ) & 22000 & 258@xmath25 & 51.2 + tateyama @xmath1 strauss ( 1992 ) & 43000 & 132@xmath25 & 31.2 + + slee et al .
( 1983 ) & 327 & 55@xmath25 & 2 : + slee et al .
( 1983 ) & 843 & 40@xmath2660@xmath25 & 3 : + schreier et al .
( 1981 ) & 1407 & [email protected]@xmath25 & 3.4 + clarke et al .
( 1992 ) & 1446 & [email protected]@xmath25 & 3.7 + clarke et al . ( 1992 ) & 1634 & [email protected]@xmath25 & 4.9 + jones et al . ( 1994 ) & 2300 & 0.1@xmath25 & 3.4 + schreier et al . ( 1981 ) & 4866 & [email protected]@xmath25 & 5.1 + clarke et al . ( 1992 ) & 4866 & [email protected]@xmath25 & 6.9 + jones et al .
( 1994 ) & 8400 & 0.03@xmath25 & 5.9 + clarke et al .
( 1992 ) & 14984 & [email protected]@xmath25 & 9.4 + botti @xmath1 abraham ( 1993 ) & 22000 & 252@xmath25 & @xmath271632 + fogarty @xmath1 schuch ( 1975 ) & 22000 & 252@xmath25 & 21.5 + kellerman ( 1973 ) & 31500 & 225@xmath25 & 24 + tateyama @xmath1 strauss ( 1992 ) & 43000 & 132@xmath25 & 10.8 + botti @xmath1 abraham ( 1993 ) & 43000 & 132@xmath25 & @xmath27720 + kellerman ( 1973 ) & 89000 & 80@xmath25 & 18 + israel ( unpublished ) & 89000 & 57@xmath25 & @xmath288 + israel ( unpublished ) & 110000 & 47@xmath25 & @xmath287 + israel ( unpublished ) & 145000 & 36@xmath25 & @xmath286 + notes : a. as revised by cooper et al .
( 1965 ) ; b. range of variation over period 1979 1992 ; c. measurements in september 1996 .
+ the whole radio source has been mapped with single - dish telescopes only , at linear resolutions no better than 4 kpc .
the radio - bright inner lobes have been mapped with aperture synthesis telescopes at hundredfold higher resolutions of about 40 pc and the inner jets and core have been mapped at ( sub)parsec resolutions in vlbi experiments .
aperture synthesis mosaicing of the outer features at substantially higher linear resolutions should be undertaken .
although a daunting task in view of the large angular sizes and low surface brightnesses , it is of great importance as no other radio galaxy allows its lobe structures to be studied in the detail possible in the case of centaurus a. the overall distribution of radio emission is roughly @xmath29-shaped .
the distribution of radio emission across the _ northern and southern outer lobes _ is very asymmetric .
the brightest part of the giant northern lobe is usually called the northern middle lobe ; there is no such feature in the giant southern lobe for reasons that are not understood .
one of the dwarf companion galaxies , uks 1324 - 412 = eso 324-g024 , is seen in projection against the centre of the giant northern lobe .
the isolated radio peak at @xmath12 = @xmath30 , @xmath31 = @xmath32 ( fig .
4 ) is not part of the southern lobe , but is due to a 15 mag background elliptical radio galaxy ( cooper et al . 1975 ; haynes et al . 1983 ) .
this galaxy could be used as a tracer for conditions in the lobe ( junkes et al .
another unrelated dwarf galaxy , somewhat misleadingly known as the fourcade - figueroa shred , is located at the tip of the southern main lobe ( dottori @xmath1 fourcade 1973 ) .
while radio surface brightnesses steadily decline away from the core in the northern lobe , they reach a maximum in the southern lobe at 1.8@xmath20 from the core .
the position angle of the northern lobe is close to 0@xmath20 , while the southern outer lobe is displaced to the west and has a position angle of roughly 135@xmath20 .
the two lobes are connected by a low surface brightness ` bridge ' and the northern lobe is probably closer to us than the southern lobe ( junkes et al .
most of the emission from the lobes has a spectral index @xmath33 , characterized by a ratio of random to uniform magnetic field strength @xmath34 = 0.6 ( combi @xmath1 romero 1997 ) .
the outer lobes may be associated with hard x - ray limb brightening ( arp 1994 ) . the _ northern middle lobe _ is a major feature of the radio source : at 5 ghz it contributes almost 45% to the total radio emission from centaurus a. its overall orientation is north - south at @xmath35 and it fades into the outer lobe at about @xmath31 = @xmath36 .
its radio peak is at a projected distance of 20 kpc to the nucleus , at position angle 36@xmath20 ( haynes et al .
whereas its radio intensities drop rather gently towards the north and northwest , its southeastern edge is relatively sharply delineated .
the middle lobe is associated with soft x - ray emission , with a luminosity of about 5 @xmath3 10@xmath37 erg s@xmath38 ( feigelson et al .
although inverse compton scattering of the microwave background a priori is expected to occur in extended radio lobes , this explanation does not seem to fit the x - ray emission from this lobe ( feigelson et al .
1981 ; marshall @xmath1 clark 1981 ; morini et al .
high - resolution radio observations not yet existing might enable a choice to be made between the remaining explanations of thermal and synchrotron emission .
the polarization e - vectors ( perpendicular to the magnetic field direction ) are well - aligned , but change their direction with increasing right ascension across the middle lobe from position angle @xmath39 15@xmath20 to 90@xmath20 , emphasizing its concave southeastern boundary ( fig . 4 ; junkes et al .
polarized intensities are between 20% and 40% , and the polarization from the middle lobe continues smoothly into that of the northern outer lobe where peak polarized intensities reach somewhat higher levels of up to 50% . the southern outer lobe has polarized intensity levels similar to those in the north , but south of @xmath31 = @xmath40 its polarization structure abruptly becomes very chaotic .
this could be due either to turbulence in the southern lobe , or to the presence of galactic foreground polarization .
junkes et al .
( 1993 ) entertained the possibility that the giant northern lobe is bending into our line of sight , while the southern lobe may be turning away from the plane of the sky in the opposite direction . in both lobes ,
steeper spectra are found where the radio surface brightness declines to low levels ( see fig .
these steep - spectrum regions appear to be unpolarized , but this may merely reflect a lack of sensitivity ( combi @xmath1 romero 1997 ; junkes et al .
1993 ) . the two inner lobes ( _ cf . _ fig .
3 ) together contribute almost 30% to the total 5 ghz flux - density of centaurus a. the radio emission from the northern inner lobe is about 40% higher than that from the southern inner lobe .
slee et al .
( 1983 ) show that both lobes have identical nonthermal spectral indices @xmath12 = @xmath41 , and estimate from this an age of 6 @xmath3 10@xmath21 years . with a maximum angular size of 12@xmath4 , the inner lobes are usually not , or poorly , resolved by single - dish telescopes . aperture synthesis maps ( fig . 5 ) reveal pronounced filamentary structure especially in the southern lobe , probably generated by velocity shear in the turbulent flow of the lobe rather than by shocks ( clarke et al .
the northern lobe has a fairly sharply delineated southeastern edge . at the same position angle as the inner lobes , arp ( 1994 )
has found a jet of diffuse x - ray emission , extending , however , much beyond northern radio lobe limits .
its nature and origin are not clear .
no soft x - ray emission is seen towards the inner lobes ( feigelson et al .
( 1981 ) . in the high - resolution maps shown by clarke et al .
( 1992 ) , the inner lobes are revealed as subsonic plumes emanating from a supersonic central jet at a position angle of 51@xmath20 .
the jet lobe transition occurs at the location of the innermost optical shell identified by malin et al .
( 1983 ) and marked no .
9 in fig .
this transition has been interpreted as evidence for shock disruption of the jet at the interface of the interstellar and intergalactic medium of the galaxy ( norman et al .
1988 ; gopal - krishna @xmath1 saripalli 1984 ) .
the northern jet is quite pronounced and shows several knots , whereas the southern ( counter)jet was first surmised from the detection of a few weak knots in the southern lobe ( clarke et al . 1986 ; 1992 ) ; it is now confirmed by vlbi images ( sect .
the direction of polarization in the inner lobes is dramatically different from that in the middle and outer lobes . beyond the sharply delineated northern plume perimeter , at a projected distance of 6.3 kpc to the nucleus , it changes abruptly by 90@xmath20 , so that the northern inner lobe is quite distinct from the middle lobe . e - vector alignments indicate a magnetic field emanating radially from the nucleus along the lobe axis ; at the edges of both lobes , the magnetic field is parallel to the plume perimeter ( clarke et al .
1992 ) . at 5 ghz , both lobes are @xmath39 35% polarized with some regions having polarizations as low as 15% or as high as 50% .
the southern lobe is more strongly depolarized , most likely because it is located behind the clumpy interstellar medium of the galaxy s disk component ( sect . 4 ) .
it exhibits a much less ordered polarization pattern ( clarke et al .
1992 ; junkes et al .
1993 ) .
broad - line high excitation gas appears to be associated with the inner lobes , but its nature is still unclear ( phillips et al . 1984 and references therein ) .
the morphologies of the inner radio jet ( schreier et al .
1981 ) and the x - ray jet ( schreier et al . 1979 ; feigelson et al . 1981 ; dbereiner et al . 1996
; turner et al .
1997 ) are very similar .
they fill the gap between the nucleus and the optical filaments reported by dufour @xmath1 van den bergh ( 1978 sect .
seven major concentrations ( usually labelled a through g following feigelson et al .
1981 ) can be discerned with rather identical radio / x - ray flux ratios .
the knots are brightest nearest the nucleus and are separated by typically 30@xmath25 .
they have 0.44.5 kev x - ray luminosities of 0.21.0 @xmath3 10@xmath37 erg s@xmath38 .
once again , it appears that inverse compton radiation can not explain their nature , although both thermal and synchrotron models are possible ( feigelson et al . 1981 ; see also turner et al .
the kinetic energy of the jet averaged over time exceeds the radiative energy of the nucleus .
it has been suggested that both radio and x - ray emission are produced by a single population of relativistic electrons with lorentz factors @xmath42 8 @xmath3 10@xmath0 in 60 microgauss magnetic fields ( schreier et al .
1981 ; burns et al .
as such electrons have lifetimes of 50 years or less , in any event much shorter than light travel times _ even across the knots _ , this requires in situ acceleration ( schreier et al .
1981 ; feigelson et al .
1981 ; burns et al . 1983 ; turner et al .
there is no evidence for x - ray flux variations of the _ inner jet _
( dbereiner et al . 1996 ; turner et al .
1997 ) , in contrast to the _ nuclear jet _
( see sect .
high resolution radio images ( 18@xmath35 pc ) by clarke et al .
( 1986 ; 1992 ) illustrate the extremely inhomogeneous nature of the main ( northern ) jet , with a hierarchy of knots reaching down to sizes less than 2.5 pc .
the jet is not a linear structure , but exhibits side - to - side limb brightening , perhaps due to a combination of external and internal shocks ( clarke et al .
as most of the jet is inside the main body of the galaxy , some degree of interaction with the gaseous medium of the galaxy is to be expected .
indeed , burns et al .
( 1983 ) suggested that the inner jet is confined by external pressure . in the knots , the spectral index decreases from @xmath43 at the limb - brightened side to @xmath44 at the diffuse side , while polarized fractions drop from 40% to less than 20% .
filaments or streamers connect knot complexes to ones downstream ; they have steep spectra @xmath44 .
if the knots are shock fronts reaccelerating electrons , the streamers will represent material flowing down the jet away from these fronts at speeds of 5000 @xmath7 ; the spectral steepening reflects increasing importance of synchrotron energy losses ( burns et al .
1983 ; clarke et al . 1986 ) . intense optical emission from the knots in the brightest part of the x - ray / radio jet , at [email protected]@xmath4 , and also from in between the knots ,
was discovered by brodie et al .
the emission is due to gas at densities of a few hundred per cc in the knots , and from gas at less than 10 per cc in the interknot region .
at least one of the optical features corresponds precisely to radio / x - ray structure suggesting shock excitation of clumps carried along in the jet flow .
the continuum appears to be the optical counterpart of the synchrotron x - ray / radio emission ( brodie @xmath1 bowyer 1985 ) .
somewhat farther out , a long chain of filaments , dust clouds , h@xmath12 knots and blue stellar objects emanates from the centre of the galaxy more or less along the direction of the jet .
this chain starts at about 2@xmath4 from the nucleus and extends out to 28@xmath4 ( see the review by danziger 1981 and references therein ) .
it breaks up into three distinct groups of features .
the socalled _ inner jet structures _ of this chain are found between 2@xmath4 and 6@xmath4 ( dufour @xmath1 van den bergh 1978 ) ; the features closest to the nucleus are aligned with but offset from the radiojet , just where this starts to develop in the plume of the northern inner lobe .
the outermost features of this group continue at the same position angle , although the radio plume expands away to the north . in their fig .
1 , morganti et al . ( 1991 ) provided high - quality images as well as a very useful diagram of the relative positions of optical and radio structures .
the next group out ( commonly called the _ inner filaments _ ) occurs between 8@xmath4 and 10@xmath4 from the nucleus at the same position angle as the jet ( @xmath45 = 55@xmath20 ) , although the outermost may start ` veering off course ' ( see morganti et al .
they are easily visible on long - exposure images of ngc 5128 ( _ cf . _
figs . 6 and 7 ) .
these filaments are in the northern middle radio lobe , but have no known specific radio counterpart .
this may reflect the lack of high - resolution radio images of the middle lobe more than anything else and provides another rationale for obtaining such .
the filaments have very high [ oiii]/h@xmath46 ratios of about 15 and are blueshifted by up to 400 @xmath7 from the systemic velocity of the galaxy ( osmer 1978 ; morganti et al .
they consist of turbulent low - density gas and coincide with the soft x - ray emission associated with the northern middle lobe ( feigelson et al .
1981 ; graham @xmath1 price 1981 ; graham 1983 ) .
the _ outer filaments _ ( at 17.5@xmath4 ) are also in the middle radio lobe , again without obvious radio counterpart at least at the low resolutions presently available .
their position angle , @xmath45 = 43@xmath20 is markedly different from that of the other filaments ( fig .
in addition to the brightest main filaments , there are weaker features to the north and west . graham ( 1983 )
measured some rather high blueshifts up to 1000 @xmath7 , but morganti et al .
( 1991 ) found blueshifts of only 70215 @xmath7 for the main filaments .
the origin and excitation of the filaments has been a matter of some debate .
although some of the features have spectra similar to those of ` normal ' hii regions , several ( notably the inner filaments ) have spectra suggesting photo - ionization by a ( nuclear ) power - law spectrum or by shock excitation .
metallicities appear to rule out a primordial origin of the gas .
the filaments may be formed by the expanding radio lobes compressing ambient gas to temperatures of a few million k followed by rapid cooling on a timescale of 10 million years ( feigelson et al .
1981 ; gouveia dal pino @xmath1 opher 1989 ) .
alternatively , the jet may consist of material ejected from the nuclear region ( graham @xmath1 price 1981 ) , although it is questionable whether this would survive long enough to have reached the presently observed distance to the nucleus ( graham 1983 ) . in a rather extensive study of the filaments , morganti et al .
( 1991 , 1992 ) concluded that they are predominantly photoionized by the radiation from the nucleus itself , extinguished in our line of sight but travelling unimpededly along the jet .
they proposed that a bl lac nucleus lies at the core of centaurus a. this conclusion was tentatively supported by schiminovich et al .
( 1994 ) who found that the apparently counterclockwise rotating hi clouds near these filaments have velocities very similar to those found optically .
however , viegas @xmath1 prieto ( 1992 ) argued that additional heating mechanisms are required , and suggested the presence of shocks .
they also concluded that a mixture of optically thin and optically thick gas is necessary for photoionization to be a viable explanation .
in contrast , sutherland et al .
( 1993 ) proposed that instead of photo - excitation , the mechanical flux of a mildly supersonic low - density jet interacting with dense ambient clouds is sufficient to energize shock waves with velocities @xmath39 200450 @xmath7 .
their model reproduces several observed features of the filaments , and removes the need for a narrow ionizing beam as proposed by morganti et al .
( 1991 , 1992 ) . as the latter model predicts strong uv line fluxes from the filaments , whereas the former does not , such an observational test would be interesting .
finally , we note that the inner filaments are at the same position angle as the inner jets , the more distant outer filaments are at the same position angle as the bright nl-2 component ( clarke et al .
1992 ; see fig . 5 ) in the plume , and the even more distant faint galaxy extensions in fig .
8 line up with the nl-4 component in the plume ( see fig . 5 ) .
whether this is accidental , or provides clues to their origin is unclear .
with @xmath47 = 6.98 mag ( de vaucouleurs et al . , 1976 rc2 ) , ngc 5128 is the fifth brightest galaxy in the sky , immediately after the local group members m 31 , m 33 , lmc and smc .
images of relatively short exposure , limited to a surface brightness in b of about 22 mag per arcsec@xmath48 , show an almost circular appearance , which has led to the classification s0p or e0p ( fig . 1 ) .
however , at lower surface - brightness levels ( fig . 8) , the shape of the galaxy becomes increasingly noncircular . at about 25 mag per arcsec@xmath48 ,
the axial ratio has increased to 1.3 , after which elliptical symmetry is lost ( _ cf . _
haynes et al .
a more appropriate classification appears to be e2 ( dufour et al .
1979 ; mcelroy @xmath1 humphreys 1982 ; haynes et al . 1983 ; ebneter @xmath1 balick 1983 ) , with the photometric major axis of the galaxy at position angle 35@xmath20 .
the light distribution closely follows the @xmath49 de vaucouleurs law ( van den bergh 1976 ) characteristic of elliptical galaxies .
the inner parts of ngc 5128 have a roughly constant total mass - to - light ratio inccreasing with radius , probably due to the presence of dark matter ( sect .
6.2 ) . of particular interest
are the deep and specially processed optical images presented by malin et al .
( 1983 see fig . 6 ) and haynes et al .
( 1983 see fig
the former reveals an extensive system of shells of old disk stars within the extended elliptical galaxy .
the system is most regular on the northeastern side , and more fragmented on the southwestern side .
the outermost shell is found at 18@xmath4 from the nucleus .
atomic neutral hydrogen emission with a total mass of 1.5@xmath310@xmath21 @xmath14 is present just outside several of the outer shells ( schiminovich et al .
1994 ) , with a rotation axis at position angle 285@xmath20 , somewhat offset from the position angle of the minor axis of the e2 galaxy and the dust band ( fig . 7 ) .
the image presented by haynes et al ( 1983 ; see also cannon 1981 ) shows diffuse extensions emanating from the elliptical galaxy at position angle @xmath39 30@xmath20 almost perpendicular to the major axis of the dusty disk in the centre ( fig .
the northeastern extension is narrower , longer and better - defined than the one in the southwest .
it is just west from the northern middle radio lobe .
the faint extensions most likely consist of stars , or possibly of dust reflecting light from stars further in ( cannon 1981 ; haynes et al .
1983 ) .
the appearance of ngc 5128 departs strikingly from that of a normal elliptical galaxy because of its broad and patchy equatorial dark band .
bisecting the almost circular bright central part of the galaxy , it is oriented along the _ minor axis _ of the elongated shape seen in deep images ( fig .
consequently , ngc 5128 is sometimes also classified as a polar ring galaxy although it fails to share several of the characteristics common to these ( see richter et al .
the dark band is associated with young stellar objects ( dufour et al .
1979 see also fig .
modelling has shown that it is in fact a _ thin _ , strongly warped disk embedded in the host galaxy that creates the superficial appearance of a broad band ( sects . 4 and 6 ) .
kinematically , the galaxy and the dark band represent different entities .
the elliptical system and its globular cluster retinue have low rotational velocities , whereas the dust disk exhibits much higher rotational velocities ( sect .
6.1 ) .
the galactic foreground confusion caused by the low galactic latitude ( @xmath50 = + 19@xmath2 ) of ngc 5128 for a long time impeded attempts to identify its globular clusters , notwithstanding its proximity . however , after the first identifications were finally made by graham @xmath1 phillips ( 1980 ) and van den bergh , hesser @xmath1 harris ( 1981 ) , the number of confirmed globular clusters rose steadily from 20 ( hesser et al . 1984 ) to 35 ( hesser , harris @xmath1 harris 1986 ) to 87 ( harris et al . 1992 ) while minitti et al . (
1996 ) added another 26 globular clusters in the inner 3 kpc of the galaxy .
analyzing image counts , harris et al .
( 1984a ) have estimated a total cluster population of 1550@xmath9350 .
globular cluster studies have been used primarily to extract information on the galaxy s metallicity and dynamics , as well as its distance ( sect .
1.2 ) . in terms of both numbers and metallicities ,
the globular cluster system of ngc 5128 appears to be normal for large elliptical galaxies ( harris et al .
, 1984b , 1992 ) or indeed for spheroidal components of galaxies in general ( jablonka et al .
1996 ) .
at least at radii beyond 4@xmath4 , the surface density of globular clusters appears to follow the same r@xmath51 law as found by van den bergh ( 1976 ) for the spheroidal halo light .
in addition , the distribution of globular clusters on the sky hints at a preferential orientation , with a major axis aligned with the major axis of the outer isophotes of ngc 5128 ( hesser et al .
although the first determinations of very high metallicity by frogel ( 1984 ) turned out to be overestimates , the clusters in ngc 5128 do seem to have somewhat greater metallicities ( @xmath52 } \leq$ ] + 0.1 ) than their milky way counterparts which they otherwise resemble ( harris et al .
1992 ; jablonka et al .
1996 ; minniti et al . 1996 ; alonso @xmath1 minniti 1996 ) .
rather similar metallicities were derived for red giant branch stars by soria et al .
the innermost clusters are on average more metal - rich , implying the presence of a metallicity gradient @xmath53}/\delta r$ ] = @xmath54 dex kpc@xmath38 ; this gradient is only apparent , and in fact caused by _
different concentrations _ of metal - poor and metal - rich clusters ( alonso @xmath1 minniti 1996 , but see jablonka et al .
1996 ) .
half a dozen of the innermost clusters have magnitudes and colours ( but undetermined metallicities ) suggesting that they are intermediate - age clusters such as found in the magellanic clouds ( minniti et al . 1996
; alonso @xmath1 minniti 1996 ) .
hui et al .
( 1995 ) showed that the metal - poor globular cluster ensemble lacks significant rotation whereas the metal - rich ensemble rotates more rapidly .
a puzzling , but still uncertain result was obtained by hesser et al .
they found that the metal - rich innermost cluster ensemble has a mean velocity about a 100 @xmath7 ( 4@xmath55 ) higher than the systemic velocity of ngc 5128 , in contrast to the outermost clusters that conform to the systemic velocity .
the bimodal globular cluster population of ngc 5128 is of great interest in view of the proposed merger - nature of the galaxy , particularly as a bimodal metallicity distribution with metal - rich clusters more concentrated than metal - poor clusters has been identified by zepf @xmath1 ashman ( 1993 ) as the natural consequence of galaxy mergers . a survey of [ oiii ] @xmath565007 emission from planetary nebulae in ngc 5128 by hui et al .
( 1993a ) , extending over 20 kpc along the major axis and fully covering the central 10 kpc , yielded 785 detections ( hui et al .
the high - luminosity cut - off of the resulting pn luminosity function was used to derive the galaxy distance in sect .
hui et al .
( 1995 ) measured radial velocities for 433 pn s , which were used to study the kinematics and dynamics of the galaxy ( sect . 6.2 ) . a v - i colour - magnitude diagram of 10000 red giant branch ( rgb ) stars in the halo of ngc 5128 was constructed by soria et al .
( 1996 ) from hst / wfpc2 images , the first time that individual stars were resolved in a spheroid system beyond the local group .
analysis of the results yielded the distance and metallicity values already mentioned ( sects .
1.2 and 3.2 ) .
the i - luminosity function of these stars is very similar to that of the rgb stars in the m 31 dwarf elliptical companion ngc 185 .
about 200 stars were found to be brighter than the tip of the rgb ; most of them are probably upper asymptotic giant branch ( agb ) stars although confusion with unresolved multiple stars is still a problem .
the luminosity functions of these agb stars suggest the presence of an intermediate - age population of about 5 gigayear , but making up at most 10 @xmath5 of the total halo stellar population . over the five - year period 19851989 , ciardullo et al .
( 1990 ) monitored ngc 5128 for the appearance of novae .
they detected 16 novae , twelve of which formed a statistically complete and homogeneous sample unaffected by the dust lane .
normalized to the near - infrared k - band luminosity , the derived nova rate of 4.2 yr@xmath38 per 10@xmath57 @xmath58 is virtually identical to that derived for the bulge of m 31 .
they further argued that the normalized nova - rate of galaxies is largely independent of their luminosity , colour or hubble type .
the galaxy was the host of the type ia supernova 1986 g ( evans 1986 ) , which reached a maximum at @xmath59 = [email protected] ( cristiani et al .
1992 ) . with @xmath60 = 4.4 mag and the currently best value for the distance modulus ,
this translates into an absolute magnitude @xmath61 = @xmath62 .
however , the reddening of sn 1986 g , mostly caused by dust internal to ngc 5128 , is variously given as @xmath63 = 0.9 ( phillips et al . 1987 ) , 1.1 ( cristiani et at .
1992 ) , 0.6 ( phillips 1993 ) and 1.6 ( hough et al . 1987 ) .
the latter also argued from their polarization measurements that near sn 1986 g the ratio of total to selective extinction r is 2.4 , rather than 3.1 .
the resulting @xmath64 values thus range from 1.9 to 3.8 mag , so that the extinction correction remains a major source of uncertainty .
this could be resolved by assuming an snia mean peak absolute magnitude .
unfortunately , estimated values range from @xmath61 = @xmath65 ( cristiani et al .
1992 ) , @xmath66 ( branch et al .
1996 ) to @xmath67 ( phillips 1993 ) , the major uncertainty this time being galaxy distance moduli . a further discussion of the use of snia s as standard candles is outside the scope of this review ; for recent contributions see e.g. riess , press @xmath1 kirshner ( 1996 ) and branch , romanishin @xmath1 baron ( 1996 ) .
the integrated colour of ngc 5128 as given in the rc2 is @xmath68 = + 0.98 .
this is identical to the colour van den bergh ( 1976 ) finds at distances of 150@xmath25 to the dark band centerline , where also @xmath69 = + 0.49 .
the foreground reddening is @xmath63 = [email protected] ( newell et al . 1969 ; frogel 1984 ; harris et al .
1992 ) , so that the intrinsic colours become @xmath70 = 0.88 and @xmath71 = 0.41 , bluer by 0.08 mag resp .
0.15 mag than normal for elliptical galaxies ( van den bergh 1976 ) .
the photometry by van den bergh shows _ increasing blueness _ approaching the dark band ; at its edge , 40@xmath25 from the centerline , the colours abruptly spread over a great range + 0.4 @xmath72 + 1.2 and @xmath73 + 0.5 , whereas in the dark band only _ red colours _ are found ( + 1.0 @xmath72 + 1.6 ; + 0.3 @xmath74 + 0.9 ) .
the excess blue emission has been explained as due to optical synchrotron emission from the inner lobes ( dufour et al .
1979 ) , a metal - poor population , or a young population of hot , blue stars .
the first explanation is unlikely ( ebneter @xmath1 balick 1983 ) , as is the second in view of the metallicity results given in sect . 3.2 , particularly those by soria et al .
the last explanation is supported by the presence of a substantial number of hii regions in the dust band itself ( dufour et al .
1979 ; hodge @xmath1 kennicutt 1983 ) , and even more so by colour images of the galaxy that show substantial very blue stellar populations at the northwestern and southeastern edges of the dust band ( _ cf .
_ fig . 1 ) . a much less sharply delineated and less extreme increase of blue colours occurs farther out , in the inner halo , and probably reflects the increasing presence of metal - poor objects ( van den bergh 1976 ) .
dufour et al . ( 1979 ) first established that the dark band crossing the elliptical galaxy is in fact the image of a highly inclined _ disk component _ consisting of a metal - rich population of stars , nebulae and dust clouds ( fig . 1 ) .
metallicities are close to those in the solar neighbourhood ( dufour et al .
1979 ; phillips 1981 ; eckart et al . 1990a ; viegas @xmath1 prieto 1992 ) .
the disk is at position angle 122@xmath2 @xmath9 4@xmath2 and star formation is rampant .
the present burst of star formation apparently started 50 million years ago and created at least a hundred hii regions embedded in the disk ( dufour et al .
1979 ; hodge @xmath1 kennicutt 1983 ) .
remarkable concentrations of luminous and very blue stars can be seen at the northwestern and southeastern edges of the dark band ; they must represent very large ob associations ( fig . 1 ) .
recent hst observations allowed schreier et al .
( 1996 ) to also find a large number of point - like sources _ embedded in _ the dust band with colours likewise suggestive of ob associations ( see also alonso @xmath1 minniti 1996 ) .
however , since the review by ebneter @xmath1 balick ( 1983 ) , little quantitative progress has been made on the issue of star formation .
this may occur at rates ten times higher than in the milky way ( telesco 1978 ) although disk average uv energy densities close to solar neighbourhood values ( eckart et al .
1990a ) suggest more moderate rates . as ngc 5128
is no more distant than e.g. m 82 , its brightest hii regions and supernova remnants should be detectable at centimetre radio wavelengths .
most of the published high - resolution radio maps lack the dynamic range to reach the low flux - density levels expected for hii regions or snr s in the disk . the 1425 mhz map published by condon et al .
( 1996 ) does , however , show an extension with peaks of 125 mjy in an 18@xmath25 ( 295 pc ) beam coinciding with the eastern half of the dark band , while the 43 ghz map by tateyama @xmath1 strauss ( 1992 ) likewise seems to show weak emission from the eastern dark band .
further high - resolution observations of the disk at centimetre wavelengths are desirable as they provide one of the few means studying the star formation history of the disk .
as defined by its ob stars and hii regions , the disk extends out to a radius of 4 kpc . molecular line and infrared continuum emission , further discussed in sects .
4.2 and 4.3 , is concentrated within 40@xmath5 of this radius ( joy et al . 1988 ; eckart et al . 1990a ; quillen et al . 1992 ) .
hi emission , however , extends much farther out , to radii of 7 kpc ( _ cf . _
van gorkom et al . 1990 ; schiminovich et al . 1994 ) ) .
the outer parts of the disk , as traced by the dark band and the hi emission , show a pronounced warp to position angle 90@xmath2 ( _ cf . _
the disk is in rapid rotation ( graham 1979 ; van gorkom et al .
1990 ; quillen et al .
the tilted - ring modelling by nicholson et al .
( 1992 ) shows that in spite of appearances , the distribution of dust in ngc 5128 is that of a warped _ thin disk _ of about 200 pc thickness ( sect .
6.1 ) along the minor axis .
deep images of ngc 5128 show the disk to be well inside the elliptical galaxy .
its inclination is a function of radius , but remains generally high with respect to the plane of the sky .
the hii regions are distributed throughout the warped disk and embedded in diffuse ionized gas .
nicholson et al . ( 1992 ) showed that their warped disk model also quite naturally explains the various caii and nai velocity components seen in absorption against supernova 1986 g by dodorico et al .
in addition to the seven components associated with ngc 5128 , these observationsy also showed three components with galactic foreground gas , and two intermediate velocity components of unknown origin .
the inner part of ngc 5128 is associated with diffuse x - ray emission in the form of ridges along the dark band edges but also in more isotropically distributed form ( feigelson et al . 1981 ; turner et al .
although the origin of this diffuse emission is not established unequivocally , among the most reasonable explanations for its existence are gas ejected from late - type stars dynamically heated to the required temperatures , x - ray binaries associated with the young stellar population , stellar winds in hii regions or combinations thereof ( feigelson , 1981 ; turner et al .
in any case , ngc 5128 is underluminous in diffuse x - ray emission as compared to other early - type galaxies ( dbereiner et al .
1996 ) . the hi observations by van gorkom et al .
( 1990 ) and schiminovich et al .
( 1994 ) show the atomic hydrogen to follow the dust lane , including the warp ( fig .
it could , however , not be traced over the central 2.5 kpc because of strong absorption against the centre .
van gorkom et al . ( 1990 ) found a total hi amount of about 3.3 @xmath7510@xmath21 @xmath14 , but cautioned that they might have missed a significant amount because of limited sensitivity ; nor does this estimate include the hi in the shells found by schiminovich et al .
1994 ( fig . 7 ) .
indeed , richter , sackett @xmath1 sparke ( 1994 ) find within the 21@xmath4 ( 21 kpc ) beam of the green bank 140 ft telescope a higher mass of [email protected] @xmath3 10@xmath21 @xmath14 , still uncertain because of the strong central absorption ( sect .
in the central part of the disk , molecular line emission from co and its isotopes is found out to radii of about 2 kpc , but most of it is concentrated within a radius of 1 kpc ( phillips et al . 1987 ; eckart et al .
1990a ; quillen et al .
1992 ; rydbeck et al .
1993 ) . within @xmath76 = 1 kpc ,
the area filling factor of the disk is of the order of 312@xmath5 , its thickness is less than 35 pc and the velocity dispersion is about 10 @xmath7 ( quillen et al .
the @xmath77=21/@xmath77=10 temperature ratios of about 0.9 for both @xmath13co and @xmath78co as well as the isotopic emission ratios @xmath13co/@xmath78co = 11 and @xmath13co / c@xmath79o = 75 are comparable to those of milky way giant molecular cloud complexes ( wild , eckart @xmath1 wiklind 1997 ) . modelling the co observations as tracer for the much more abundant h@xmath80 molecule , eckart et al .
( 1990a ) and wild et al .
( 1997 ) estimate molecular hydrogen temperatures @xmath81
= 1015 k and densities of a few times 10@xmath82 .
emission from other molecular species has also been detected in the disk ( whiteoak , gardner @xmath1 hglund 1980 ; seaquist @xmath1 bell 1988 ; dodorico et al .
1989 ; israel 1992 ; paglione , jackson @xmath1 ishizuki 1997 ) .
total molecular hydrogen masses are probably about 4@xmath3 10@xmath21 @xmath14 , but may be a factor of two higher depending on the co - to - h@xmath80 conversion factor favoured .
the vibrationally excited warm h@xmath80 ( @xmath81 @xmath39 1000 k ) detected by israel et al .
( 1990 ) represents only a minute fraction of all molecular hydrogen and is associated with the circumnuclear disk ( sect .
the total gaseous mass of the disk , including helium , is thus of the order of 1.3 @xmath3 10@xmath83 @xmath14 , only about 2% of the dynamical mass contained in the elliptical component within the radius of the disk ( @xmath76 = 7 kpc ) .
however , because of the pronounced concentration of interstellar gas at smaller radii , that fraction increases to about 8% at @xmath76 = 2 kpc . at far - infrared wavelengths , the disk of ngc 5128 stands out by its emission from warm dust .
dust temperatures are 3040 k depending on the assumed dust emissivity @xmath84 @xmath85 @xmath86 or @xmath87 respectively ( joy et al .
1988 ; marston @xmath1 dickens 1988 ; eckart et al .
the overall distribution of far - infrared emission is very similar to that of the carbon monoxide .
the present far - infrared information still leaves considerable room for improvement .
the kao scans presented by joy et al .
( 1988 ) do not fully sample the galaxy , but show that 10@xmath5 of the total far - infrared luminosity arises in central source which may be identified with the circumnuclear disk ( see sect .
iras survey observations cover the whole galaxy , but the resolution is limited .
several of the published fluxes refer to poorly calibrated data or underestimate the total flux from the extended galaxy .
best fluxes are probably the colour - corrected values @xmath88 = 26.4 jy , @xmath89 = 25.7 jy , @xmath90 = 236 jy and @xmath91 = 520 jy given by rice et al . (
use of iras non - survey data ( dsd maps : marston @xmath1 dickens 1988 ; cpc - maps : eckart et al .
1990a ; marston 1992 ) provided some improvement in resolution , but at the cost of photometric accuracy .
because of unsolveable calibration problems ( _ cf
. _ van driel et al . 1993 ) , the image - sharpened cpc maps discussed by eckart et al . (
1990a ) and by marston ( 1992 ) must be considered as unreliable .
the most recent image - sharpened iras maps ( resolution 12@xmath4 ) incorporating all survey - instrument data are shown in fig .
9 ; they still lack the resolution to bring forth the full detail of the disk .
good mid - infrared imaging of the dust disk itself is still lacking .
the mean 12@xmath92 m surface brightness of the disk as derived from iras data is about 25 mjy sr@xmath38 .
in addition to the nucleus ( sect .
5.3 ) , telesco ( 1978 ) detected 10@xmath92 m emission weaker by a factor of about five from a number of disk hii regions .
clearly , much more remains to be done . as much as 50@xmath5 of the 100@xmath92 m emission may be due to ` cirrus ' ( marston @xmath1 dickens 1988 ; eckart et al .
the various far - infrared data indicate the presence of small amounts of dust outside the disk in the main elliptical galaxy as well as large amounts in the disk out to a radius of 3 kpc ( eckart et al . 1990a ; marston 1992 ) . the total dust mass can be estimated from iras photometry as @xmath93 = 12 @xmath3 10@xmath94 @xmath14 ( _ cf .
_ hunter et al .
1989 ) with a luminosity @xmath95 = 2 @xmath3 10@xmath57 @xmath96 ( joy et al . 1988 ; eckart et al .
with considerable uncertainty , the _ total _ gas - to - dust ratio within a radius of 2 kpc is @xmath97 = 450 .
this is an upper limit if significant amounts of cold dust are present , imperfectly sampled by iras .
marston @xmath1 dickens ( 1988 ) explicitly modelled the iras emission in terms of large , cool grains and warm , small grains , and arrived at a ( distance - corrected ) dust mass an order of magnitude higher , implying a gas - to - dust ratio of 45 , which seems rather low . at optical and infrared wavelengths ,
the disk is significantly polarized up to 6@xmath5 parallel to the dust band ( elvius @xmath1 hall 1964 ; hough et al .
1987 ; scarrott et al .
1996 ) . the observations by the latter show higher levels of polarization at the dust band extremities , with directions perpendicular to the dust band
hough et al .
( 1987 ) concluded that the dust grains sampled are about 20@xmath5 smaller than those in the milky way , implying an extinction law differing from that in the solar neighbourhood , with a total to selective extinction ratio of 2.4 .
the hst r- and i - band imaging polarimetry presented by schreier et al .
( 1996 ) shows the polarization to reach a peak at a knot close to the nucleus , shining by reflected light . like hough et al .
( 1987 ) , they assumed scattering to be negligible elsewhere , but that is inconsistent with the apparent importance of scattering in the central region and is unlikely in view of the conclusions reached by packham et al . (
1996 see section 5.3 ) .
the observations by scarrott et al .
( 1996 ) can only be explained by assuming simultaneous operation of both dichroic extinction and scattering , with the latter dominating at the dust band extremities .
this suggests that the very blue colours observed by van den bergh ( 1976 ) just at the _ northern _ dark band edge are caused by the light of intrinsically blue objects ( _ cf .
_ alonso @xmath1 minniti 1996 and references therein ) enhanced by scattered light and suffering relatively little extinction .
this is supported by the optical continuum observations of the central region by storchi - bergmann et al .
( 1997 ) who find that the major contribution comes from a metal - rich old bulge , but that there are also significant contributions from young stars and from scattered light especially at the dark band edges .
the optical colours suggests significant extinction in the dust band itself ( @xmath98 0.5 mag van den bergh 1976 ; dufour et al .
1979 ) . indeed , the near - infrared estimates by harding et al .
( 1981 ) yield @xmath64 = 36 mag , while hst observations indicate v - band extinctions ranging from 0.5 to 7 mag in the dark band ( schreier et al .
1996 ) and even reaching values in excess of 30 mag ( @xmath99 3 mag ) just south of the optically invisible nucleus ( alonso @xmath1 minniti 1996 ) .
the r - band and i - band images by schreier et al .
( 1996 ) clearly show how the dust band is seen nearly edge - on , but slightly tilted with the near side south of the centre so that we are looking from above .
the glow of the nuclear region ( but not the nucleus itself ) on the north - side of the dust disk is strikingly apparent even through the high extinction it suffers .
the central region of the galaxy , within a few hundred parsec from the nucleus , is very complex .
the nucleus itself , hidden behind thick dust clouds , is visible at infrared ( sect .
5.3 ) , ( sub)millimetre and centimetre wavelengths ( sect . 5.4 ) and again at high energies ( sect .
5.5 ) , but generally requires very high resolutions to separate it from its surroundings . at centimetre wavelengths and at high energies ,
confusion is caused by the emission from the inner jet and even more so from the milliarcsec nuclear jets ( meier et al .
1989 ; jones et al . 1996 ; tingay et al .
1998 ) . with increasing observing frequency
, the steep nonthermal spectrum of the jet features reduces them to insignificance , so that shortwards of 1 cm ( @xmath100 30 ghz ) only the nucleus shines brightly .
infrared and millimetre observations have established the presence of a compact circumnuclear disk ( israel et al . 1990 , 1991 ; rydbeck 1993 ; hawarden et al .
1993 ; sect .
5.2 ) . at wavelengths shorter than 1 mm ( @xmath100 300 ghz ) , thermal emission from this compact circumnuclear disk becomes a new and important source of confusion ( cunningham et al .
1984 ; hawarden et al . 1993 ) .
although at the wavelengths and resolutions at which the nucleus itself can be seen , absolute flux - density determinations suffer from various calibration problems , it is clear that it is strongly time - variable ( sect .
the highly collimated nuclear radiojets likewise vary their structure and intensity with time ( jauncey et al .
1995 ; tingay et al .
1998 ) . analyzing co measurements and infrared data from the literature , israel et al .
( 1990 ; 1991 ) discovered a circumnuclear disk at the core of ngc 5128 , with an estimated gas mass 10@xmath101 .
its outer radius is 110280 pc and it contains a central region of radius 40 pc , devoid of co. such a 100 pc - scale disk appears to be a common feature of active galaxies ( _ cf .
_ maiolino @xmath1 rieke 1995 ) .
the co gas in the disk has excitation temperatures of the order of 25 k and is significantly warmer than the co gas in the dust band ( @xmath102 = 1015 k ( sect .
the kinematic signature of this rapidly rotating circumnuclear disk is evident in central co profiles ( israel et al .
1991 ) , as well as in major - axis position - velocity diagrams ( _ cf .
_ fig . 2 of quillen et al .
the total dynamical mass within the disk area is estimated at about 10@xmath83 @xmath14 , i.e. a hundred times higher than the molecular gas mass ( israel et al . 1991 ; rydbeck et al . 1993 ) .
_ emission _ from other molecular species in the circumnuclear disk , such as hco@xmath103 , hcn , hnc and probably also c@xmath104h@xmath80 , hnco and h@xmath80co , has also been detected ( israel 1992 ) .
although the disk may have been revealed by its near - infrared extinction ( meadows @xmath1 allen 1992 ) , it is the observations at 450@xmath92 and 800@xmath92 m by hawarden et al .
( 1993 ) that provided its first direct image .
its major axis is at position angle @xmath105 ( hawarden et al . 1993 ; rydbeck et al .
1993 ) , quite different from that of the dust band , but perpendicular to the position angle of the jets ( sects . 2.4 and 5.4 ) .
the unresolved source seen in limited - resolution far - infrared ( joy et al . 1988 ) and mid - infared observations ( fig .
9 ) is due to thermal emission from this disk , as the flat - spectrum nonthermal nucleus is expected to contribute no more than about 25%30% to the total observed emission at 100@xmath92 m .
the disk appears to be larger in the @xmath77=10 transition than in the @xmath77=21 transition ( _ cf . _
rydbeck et al . 1993 ) . at submillimetre wavelengths it is larger again than in co or at 100@xmath92 m continuum .
this suggests a temperature gradient across the disk , in addition to the density gradient postulated by israel et al .
such a temperature gradient is to be expected if the excitation of the circumnuclear disk is caused primarily by winds and high - energy radiation from the power - law nucleus impinging on the cavity walls .
the very red near - infrared colours , with their implicit suggestion of elevated dust temperatures in the inner region ( sect .
5.3 ) in addition to high extinction , as well as the compactness of [ feii ] and h@xmath80 emission from the centre ( israel et al . 1990 ; meadows @xmath1 allen 1992 ) support this picture . such near - infrared line emission is commonly interpreted as evidence either for shocks or for strong x - ray / uv irradiation , also required by x - ray observations ( feigelson et al . 1981 ; turner et al .
1997 ; see sect .
mouri ( 1994 ) has suggested that the h@xmath80 emission , with a temperature of about 1000 k , is excited by uv radiation from the nucleus , implying local densities in excess of 10@xmath106 ( sternberg @xmath1 dalgarno 1989 ) .
this is supported by the detection of x - ray line emission and absorption indicating the presence of variably ionized absorbing clouds vigorously interacting with the nuclear continuum radiation ( morini et al .
1989 ; sugizaki et al . 1997 ; turner et al .
1997 ) .
unpublished observations by p. van der werf ( private communication ) show the outer radius of hot , vibrationally excited h@xmath80 emission to be about 40 pc where it is rotating with 215 @xmath7 ( fig .
this implies a dynamical mass of 4.4 @xmath3 10@xmath21 @xmath14 within 40 pc , or an almost twentifold increase of mass per unit volume going from @xmath76 = 150 pc to @xmath76 = 40 pc . assuming the outer limit of the extent of excited h@xmath80 emission to mark the cavity radius ( koornneef @xmath1 israel 1996 )
, it is doubtful whether the nucleus of ngc 5128 is powerful enough to provide the required excitation by radiation alone ( israel et al .
unless disk material extends much further inwards , it can not control the collimation of the nuclear jets either , because this occurs on significantly smaller scales ( see clarke et al .
1992 ; jones et al . 1996 ) . whether or not there is another , much smaller and thinner accretion disk of the sort proposed by e.g. krolik @xmath1 lepp ( 1989 )
, it should be emphasized that the agent collimating the jets must be closely , if possibly indirectly , connected with the circumnuclear disk discussed here in order to explain the excellent alignment between its rotation axis and the observed jet flow .
optically , the core of ngc 5128 is completely obscured . because dust extinction in the near - infrared is much less than at optical wavelengths ( e.g. @xmath107 ) ,
the core has been frequently observed ( becklin et al .
1971 ; kunkel @xmath1 bradt 1971 ; grasdalen @xmath1 joyce 1976 ; lpine , braz @xmath1 epchtein 1984 ) as well as mapped ( harding , jones @xmath1 rodgers 1981 ; quillen , graham @xmath1 frogel 1993 ; adams , adamson @xmath1 giles 1983 ; giles 1986 ; joy et al . 1991 ; meadows @xmath1 allen 1992 ; turner et al . 1992 ; golombek , quoted in schreier et al . 1996 ) in the 15@xmath92 m wavelength range .
near - infrared images longwards of 2@xmath92 m , usually show a compact source close to the central position , but it is questionable whether this represents direct emission from the nucleus ( see e.g. antonucci @xmath1 barvainas 1990 ) . the relatively large - scale k - band maps by quillen et al .
( 1993 ) and adams et al .
( 1983 ) show a central 2.2@xmath92 m peak embedded in an extended emission component in addition to several foreground stars .
the extended emission is mostly due to the innermost stars of the elliptical galaxy , although there is also some emission from stars and hii regions in an inner dust lane of outer radius of 80@xmath25 and projected thickness of 15@xmath25 ( harding et al .
1981 ; quillen et al . 1993 ; adams et al . 1983 ; meadows @xmath1 allen 1992 ) .
these k - band images show the ` blossoming ' of emission northeast of a dust absorption band , visible even at this wavelength , that extends across the nucleus . below 2@xmath92 m
, the central feature is cometary in shape , points to the southwest in the same position angle as the x - ray / radio jet ( sect .
2.4 ) and is less compact than at longer wavelengths . the northeastern part of this feature corresponds to the ubvri ` hot spot ' discovered by kunkel @xmath1 bradt ( who prematurely identified it with the nucleus ) .
joy et al .
( 1991 ) speculated that this cometary j / h - band ` blue ' source is the near - infrared counterpart of the x - ray / radio jet discussed in sect .
however , its ` blue ' infrared colours and high degree of polarization are consistent with reflected light from a stellar or active nucleus in a clear line of sight ( _ cf . _ turner et al .
1992 ; packham et al .
1996 ; schreier et al .
1996 ) , whereas the opening angle of the ` blue ' component is much wider than that of the radio / x - ray jets .
the position of peak emission moves to the southwest with increasing wavelength ( _ cf .
_ schreier et al .
the accuracy of positional measurements is insufficient to determine whether or not the k - band peak coincides with the nucleus ( for the best position determinations see giles 1986 ) , but data tabulated by schreier et al .
( 1996 ) suggest it is still somewhat off .
near - infrared polarimetry of the visually obscured centre of ngc 5128 has shown that , with increasing wavelength as well as with decreasing aperture , the position angle of the polarized vectors steadily increases from @xmath45 @xmath39 120@xmath2 to @xmath45 = 145@xmath2 ( bailey et al .
1986 ; packham et al .
( 1996 ) , i.e. from parallel to the dust band to parallel to the circumnuclear disk and perpendicular to the radio jet . at the ` nuclear source ' ,
the intrinsic polarization is 17@xmath5 .
schreier et al . ( 1996 ) further showed that this polarization peak coincides with a compact knot close to both the extinction and k - band emission peaks .
the knot appears to be a heavily obscured interstellar cloud , scattering the optical and near - infrared radiation of the invisible nearby nucleus toward our line of sight .
the absence of significant polarization at _ millimetre _ wavelengths led packham et al .
( 1996 ) to reject bailey et al.s ( 1986 ) notion of an intrinsically highly polarized nucleus , emitting near - infrared synchrotron radiation and instead led them to conclude that the nucleus itself is obscured even at near - infrared wavelengths with @xmath64 = 16 mag and that the ` nuclear ' polarization is produced by scattering and not by synchrotron emission or dichroic extinction .
thus , the true nucleus is only fully revealed at wavelengths longer than 2.2@xmath92 m , and the compact k - band source is mostly due to reflected light .
indeed , the extinction of the _ nucleus itself _ is variously estimated to be @xmath108 , centering on @xmath64 = 2530 ( see lpine et al . 1984 ; giles 1986 ; turner et al .
1992 ; meadows @xmath1 allen 1992 , and references therein ) . as this is much higher than the extinction caused by the dust band ( sect .
4.4 ) , the nucleus must be obscured primarily by the circumnuclear disk .
apart from the extinction problem , it is extremely difficult to determine reliable flux - densities , colours or even the very reality of weak unresolved k - band sources in galaxy centres , because of the practical difficulties in separating such sources from the central cusp of the surrounding galaxy ( simpson 1994 ; see also fig . 8 by turner et al .
1992 ) .
the nucleus of ngc 5128 does appear as an infrared point source at 3.3 @xmath92 m ( turner et al .
1992 ) and at 10@xmath92 m ( size @xmath109 1@xmath25 i.e. @xmath109 16 pc ; flux 1.4 jy
p. van der werf , private communication ) ; it also is variable at least at 3@xmath92 m ( lpine et al . 1984 ; turner et al . 1992
; see sect .
observations at longer infrared wavelengths so far have lacked the resolution to accurately separate the nucleus from the circumnuclear disk emission .
infrared spectra centered on the nucleus show emission lines due to molecular hydrogen ( h@xmath80 ) , brackett-@xmath110 and [ feii ] ( israel et al . 1990 ; meadows @xmath1 allen 1992 ) , but lack the noticeable co - band absorption ( characteristic for late - type stellar populations ) which is seen off the nucleus where , in turn , the emission lines are absent ( meadows @xmath1 allen , 1992 ) . the near - infrared colours of the central area , in contrast to those of the ` blue feature ' mentioned above , are extremely red , going from @xmath111 = 1.5 , @xmath112 = 0.9 in a 2.5@xmath25 aperture ( turner et al .
1992 ) to @xmath111 = 1.35 , @xmath112 = 1.75 in a 1@xmath25 aperture ( giles 1986 ) .
the very red colours are confined to a region within 12@xmath25 from the nucleus ( giles 1986 ) .
they are far too red be caused by extinction .
a rather similar situation has been found for ngc 3079 by israel et al .
1998 who interpret such colours as due to a combination of extinction and emission from hot ( 1000 k ) dust grains .
reddened synchrotron emission , also considered by turner et al .
( 1992 ) appears to be ruled out by the work of packham et al .
( 1996 ) discussed above . at the core of centaurus
a lies a compact radio nucleus for which ( kellerman , zensus @xmath1 cohen ( 1997 ) have measured a size of [email protected] milli - arcsec corresponding to linear dimensions of only 0.008 pc or 1700 au ) . as there may be unresolved fine - scale structure even at this resolution
, actual nuclear dimensions could be even smaller , the _ lower limit _ being about 10@xmath113 cm or 700 au ( grindlay 1975 ; mushotzky et al .
1978 ; jourdain et al .
1993 ) .
the nuclear source has a strongly inverted spectrum ( @xmath114 4 ; jones et al 1996 ) .
it is all but invisible below 5 ghz , but shows up in vlbi maps at 8.4 ghz with significantly time - variable flux - densities ranging over an order of magnitude ( preston et al .
1983 ; meier et al . 1989 ; jauncey et al . 1995
; jones et al . 1996 ; tingay et al . 1998 ) .
below about 2030 ghz the radio emission from the nucleus probably suffers from both synchrotron self - absorption and free - free absorption in a circumnuclear ionized gas ( jones et al .
1996 ) . above the turnover frequency , the spectrum appears to remain flat ( flux - density about 810 jy up to at least 1000 ghz ( _ cf . _
kellerman et al .
1997 ) .
very high resolution ( vlbi ) radio measurements are needed to separate the extremely compact nucleus from its surroundings . over some 65 milli - arcsec ( projected linear distance 1 pc ) a bright linear jet
can be traced emanating from the nucleus at a position angle of 51@xmath20 and with a width of a few milli - arcsec ( preston et al .
1983 ; jones et al .
1996 ) . the vlbi maps by jones et al .
( 1996 ) and tingay et al .
( 1998 ) show a weak counterjet to the bright northeastern nuclear jet .
the jet components have very similar spectral indices @xmath115 and , like the nucleus , are variable in intensity ( meier et al . 1989 ) .
monitoring the jet over several years , tingay et al .
( 1998 ) found significant structural ( c.q .
intensity ) changes of the knots in the jet ( fig .
11 ) , implying internal evolution as well as a subluminal projected motion @xmath116 @xmath39 0.1c of the knots .
this _ observed motion _ may in fact represent slow patterns superposed on a significantly faster relativistic jet flow ( tingay et al .
the large difference in jet and counterjet brightnesses can be explained by relativistic doppler beaming , enhancing the radiation of the jet approaching us . from the observed brightness ratio of 48
, it appears that the northeastern jet is approaching and the southwestern jet receding at an angle of 50@xmath2080@xmath20 to the line of sight at moderately relativistic speeds @xmath116 @xmath100 0.45c ( jones et al .
1996 ; tingay et al . 1998 ; see also bao @xmath1 wiita 1997 ) .
the two brightest knots c1 and c2 separated recently from the nucleus , in 1983 and in 1989 respectively ( tingay et al .
1998 ) .
there is still a significant gap in radio coverage of the nuclear region between aperture synthesis maps ( 5 ghz _ resolution _
0.3@xmath25 : clarke et al . 1992 ) and vlbi maps ( 8.4 ghz _
0.1@xmath25 : jauncey et al .
1995 ; tingay et al .
1998 ) so that a full description of the nuclear jets is still lacking . since the nuclear jet velocities are much higher than those estimated for the inner jet ( sect . 2.4 )
, significant deceleration must take place just at the missing size range so that any possibility of filling the observational gap should be pursued . the hard x - ray / soft @xmath110-ray emission
is thought to arise from compton up - scattering of lower energy photons by relativistic electrons ; the photons may be supplied by the relativistic electrons themselves via the synchrotron self - compton process , or by accretion heating .
major outbursts , correlated at x - ray and radio wavelengths , would then represent injection of fresh relativistic electrons , or re - acceleration of compton - cooled electrons .
they should decline on timescales determined by synchrotron loss rates , i.e. , a few years .
faster variations probably reflect source expansions ( grindlay 1975 ; mushotzky et al . 1978 ; baity et al . ,
1981 ; skibo et al .
1994 ; _ cf . _ bond et al .
1996 ) . unlike the nucleus ,
at least the outer jets radiate primarily at soft x - ray energies ( feigelson et al . 1981 ; turner et al .
1997 ) the best available x - ray resolution , 6@xmath25 provided by the rosat hri , is insufficient to separate the nucleus and the nuclear jets , and even the innermost part of the inner jet .
the energy distribution of the nucleus has been characterized by power - law spectra ( ubertini et al .
1993 jourdain et al .
1993 ; miyazaki et al .
1996 ; sugizaki et al . 1997 ; turner et al . 1997 ) with exponential cutoffs at 200300 kev during periods of increased activity and 700 kev during quiescent periods ( kinzer et al .
centaurus a is difficult to detect above energies of 1 mev ( see kinzer et al .
1995 and references therein ) where the bl - lac object ms1312 - 4122 ,
2@xmath20 west of centaurus a is also a potentially serious source of confusion ( fichtel et al .
1994 ; kinzer et al .
1995 ; thompson et al .
in particular , detections in the 120 mev range by von ballmoos et al .
( 1987 ) were not confirmed by oneill et al .
( 1989 ) , although the latter observed the nucleus in a more intense state .
likewise , an early detection at very high energies ( grindlay et al . 1975 ) could not be repeated ( carramiana et al . 1990 ; _ cf . _
allen et al . 1993 )
. however , observations with comptel have now provided good detections in the 130 mev range ( steinle et al .
1998 ) , which together with a 200 mev egret detection ( thompson et al . 1995 ) allow the spectrum to be fitted with sets of broken power laws , with spectral indices steepening from @xmath117 to @xmath118 to @xmath119 for the low ( @xmath120 = @xmath121 erg s@xmath38 ) intensity state and from @xmath117 to @xmath122 to @xmath123 for the intermediate ( @xmath120 = @xmath124 erg s@xmath38 ) intensity state .
turner et al .
( 1997 ) concluded that nuclear emission dominates the x - ray flux above 2 kev and estimated that 40% of the source suffers extinction by a column @xmath125 = 4@xmath310@xmath126 , 59% by a column @xmath125 = 1@xmath310@xmath126 and the remaining 1% by a column two orders of magnitude less .
it is tempting to ascribe the first fraction to the nucleus itself , and the second to the nuclear jets , but there is no real evidence supporting this speculation .
it is clear , in any case , that both components are obscured by the circumnuclear disk discussed in sect .
5.2 whose estimated parameters fit the x - ray observations rather well .
x - ray k - edge absorption and fluorescent emission from fe at 6.4 kev ( mushotzky et al . 1978 ; wang et al .
1986 morini et al .
1989 ) , but also mg , si and s ( sugizaki et al .
1997 ) can be explained by irradiation and reprocessing of clouds in the circumnuclear disk by the nuclear x - ray source ( _ cf .
_ morini et al . 1989 ; turner et al . 1997 ) and thus provide independent evidence for the fierce interaction of the nucleus with the inner parts of the circumnuclear disk . both the nucleus and the jet structures within at least 0.1@xmath4 are variable . as they do not seem to vary in tandem ( tingay et al .
1998 ) and have different spectra , accurate determinations of the spectral index and its time variation are extremely difficult to make especially with relatively large single - dish beams . at 1.4 ghz , considerable variability is seen on timescales of a hundred days or longer ( romero , benaglia @xmath1 combi 1997 ) .
as the amplitude of variation is large , and the nucleus is strongly ( self)absorbed at this frequency ( sect .
5.4 ) , it can not reflect activity in the nucleus itself .
instead , shocks interacting with density inhomogeneities in the inner jet appear responsible
. well - established variability at higher frequencies is harder to interpret .
large - beam monitoring by botti @xmath1 abraham ( 1993 ) has yielded large flux ( 22 ghz : 1632 jy ; 43 ghz : 620 jy ) and spectral index variations ( @xmath127 = 2 ) , while kellerman et al .
( 1997 ) noted a 60% drop in 43 ghz intensity over only 3 months .
qualitatively , this may reflect expansion of the nuclear source and jets , initially opaque and then becoming ( partially ) optically thin .
however , the large beams used include significant non - nuclear emission from the nuclear and inner jets , confusing the issue .
nevertheless , the radio variability appears to be correlated with that at hard x - ray wavelengths ( botti @xmath1 abraham 1993 ; jourdain et al . 1993 ; turner et al .
1997 ) which must be associated with the nucleus .
the 312 kev light curve presented by turner et al .
( 1997 ) for a 26-year period ( based largely on the vela 5b light - curve from terrell 1986 , and sparsely sampled after 1980 ) shows major outbursts in 19721976 and in 1979 ( see also feigelson et al .
1981 ; baity et al .
1981 ; gehrels et al . 1984 ; bond et al .
1996 ) when the observed x - ray fluxes increased by an order of magnitude .
these outbursts were separated by a low - intensity state with fluctuations by a factor of two . at millimetre wavelengths , likewise sampling nuclear behaviour , the 19721976 maximum is evident in the 90 ghz measurements by kellermann ( 1974 ) and beall et al .
( 1978 ) . the 90800 ghz measurements by cunningham et al .
( 1984 ; see also hwarden et al .
1993 ) show that the outburst starting in 1979 was still going on in 1981 , which is also confirmed at 100 kev ( gehrels et al .
1984 ; jourdain et al .
a poorly covered outburst occurred in 1985 ( jourdain et al .
1993 ; bond et al . 1996 ; turner et al .
1997 ) ; this outburst may be associated with the expulsion of radio knot c1 ( tingay et al .
the near - infrared 3.5@xmath92 m variability noted by lpine et al .
( 1984 ) and turner et al .
( 1992 ) follows a pattern similar to that of the radio and x - ray emission ( low in in 1971 and 1987 , high in 1975 and 1981 ) suggesting that the emission at this wavelength arises from a similar or identical mechanism .
the observed 0.12 kev luminosity of the nuclear region increased by a factor of two between 1990 and mid-1993 .
a probably related rise was seen in the 43 ghz radio flux occurring at the end of 1992 lasting into early 1993 ( abraham 1996 ) .
however , no flux increase was seen at hard x - ray energies ( turner et al .
1997 ; bond et al .
1996 ) whereas 100/230 ghz monitoring by tornikoski et al .
( 1996 ) reveals only modest ( @xmath128 ) flux increases over relatively short times in 1992/1993 and 1994/1995 .
as both hard x - rays and millimetre wave frequencies sample the nucleus rather than the jets , this suggests that the soft x - ray/43 ghz increase is related to the jet expansion observed by tingay et al .
( 1998 ) and not to nuclear activity .
the hard x - ray flares of july / october 1991 ( jourdain et al .
1993 ; bond et al .
1996 ) were unfortunately not covered by the millimetre monitoring programme .
the october 1991 flare is evident , however , in @xmath110-ray observations by kinzer et al .
( 1995 ) ; see also @xmath110-ray fluxes listed by steinle et al .
( 1998 ) for the period 19911995 , and their graph of the hard x - ray variability over the same period showing significant activity in mid-1994 .
minor x - ray flares in the low - intensity state have doubling times of about two days and last for one or two weeks ; they appear to occur several times a year ( terrell 1986 ; beall et al .
1987 ; jourdain et al . 1993 ; turner et al .
1997 ) and involve energy transfers of order 10@xmath129 erg s@xmath38 ( terrell 1986 ; bond et al .
the nucleus seems to have been in the quiescent , low - intensity state for most of the time since 1989 , as also indicated by the 8.4 ghz vlbi core monitoring by tingay et al .
in addition to these variations , there is at least at x - ray and soft @xmath110-ray wavelengths a considerable small - amplitude ( 1020@xmath5 ) variation at timescales of minutes and hours ( morini et al .
1989 ; kinzer et al . 1995 ; miyazaki et al . 1995 ; turner et al .
1997 , but see also jourdain et al .
very fast flux variations reported by wang et al .
( 1986 ) have not been confirmed and are now generally doubted .
over the last two decades , several major kinematical studies have been carried out of both the ellipsoidal and the disk component of nc 5128 .
the kinematics of the ellipsoidal component have been traced by studies of the motions of stars ( wilkinson et al . 1986 ) , globular clusters ( hesser , harris @xmath1 hesser 1986 ; harris , harris @xmath1 hesser 1988 ) and planetary nebulae ( hui et al . 1985 ) .
stellar motions were fully mapped by wilkinson et al .
( 1986 ) over 100@xmath25 and supplemented by major axis data out to 400@xmath25 ( 6.6 kpc ) .
the ellipsoidal stellar component is rotating rather slowly at velocities not surpassing 40 @xmath7 in the line of sight around an axis at position angle 135@xmath20 , i.e. rotating only roughly perpendicular to the dust lane .
the velocity dispersions range from 95 to 150 @xmath7 .
the planetary nebula system , studied by hui et al .
( 1995 ) out to 20@xmath4 , has a similar mean velocity dispersion of 110 @xmath7 .
major axis rotation increases to 100 @xmath7 at 7 kpc and after that remains constant to at least 22 kpc .
minor axis rotation is less than 50 @xmath7 out to the last observed radius of 10 kpc .
the projected rotation axis is at position angle 165@xmath20 , i.e. displaced from the photometric minor axis ( and disk orientation ) by about 40@xmath20 ; the line of zero rotation is not orthogonal to the line of maximum rotation .
such observations suggest that ngc 5128 has a triaxial potential .
the kinematics of the globular cluster system again appear to be similar to those of the spheroidal stellar system ( hesser et al .
1986 ; harris et al .
in particular , the metal - rich globular clusters rotate very much like the planetary nebulae , whereas the metal - poor clusters appear to lack significant rotation ( hui et al .
1995 ) .
the rich gas content of the dust lane ( sects .
4.1 and 4.2 ) has allowed detailed kinematical studies of the disk component to be performed in lines of ionized hydrogen ( bland , taylor @xmath1 atherton 1987 ) , neutral atomic hydrogen ( van gorkom et al .
1990 ; schiminovich et al .
1994 ) and co ( eckart et al . 1990a ; quillen et al . 1992 ; wild et al .
optical and radio observations ( graham 1979 ; van gorkom et al .
1990 ; quillen et al .
1992 ) show the disk to be in rapid rotation with a rotation gradient of about 150 @xmath7 kpc@xmath38 close to the nucleus , in sharp contrast to the rather modest rotation of the elliptical component .
graham ( 1979 ) estimated the plane of rotation to be tilted by 73@xmath2 with respect to the plane of the sky .
bland et al .
( 1987 ) mapped the ionized hydrogen gas at arcsec resolution over 7@xmath1305@xmath4 , with a moderate velocity resolution of 36 @xmath7 similar to that obtained by wilkinson et al .
( 1986 ) ; the neutral hydrogen observations have comparable velocity resolutions . although the co observations have high velocity resolutions of about 1 @xmath7 , that advantage is largely undone by their spatial resolution of 20@xmath2540@xmath25 causing significant beam - smearing which can only be partially undone by modelling .
the limited resolutions of the hi and co data allow only relatively simple modelling of circular orbits in warped disks approximated by sets of variously tilted rings .
virtually all attempts to dynamically model the observed kinematics of the disk therefore rely largely or completely on the optical observations by bland et al .
( 1987 ) . assuming circular motion consistent with an r@xmath51 law mass distribution , nicholson , bland - hawthorn @xmath1 taylor ( 1992 ) confirmed graham s ( 1979 ) conclusion that the gas and dust are in a highly inclined rotating disk , with an amplitude @xmath131 = 250 @xmath7 at @xmath76 = 2.0 kpc and a rotation gradient similar to the one derived by graham ( 1979 ) , eastern edge approaching .
projected onto the sky , the hii regions are confined to an envelope similar tothe form of a hysteresis loop , characteristic of a warped disk rather than a ring .
they convincingly showed that the kinematical data can be represented by a class of models involving a thin warped disk geometry with a spatially averaged optical depth sufficiently low to allow emission from all positions of the disk to be seen .
double - peaked velocity profiles mark folds in the warped disk seen in projection .
quillen et al .
( 1992 ; 1993 ) confirmed that such an interpretation also applies to the dusty molecular disk .
kinematical information is restricted to the distribution on the sky of radial velocities only .
the large number of free parameters in combination with observational limitations such as finite resolution and extinction , renders the outcome of dynamical modelling very sensitive to initial assumptions ( _ cf . _
kormendy @xmath1 djorgovski 1989 ) .
it is therefore not surprising that the general agreement on kinematical ( tilted ring ) models describing the motion of the disk component is not matched by a similar convergence on a unique dynamical model . for a review of the methods and problems of determining structure and dynamics of elliptical galaxies in general
the reader is referred to de zeeuw @xmath1 franx ( 1991 ) .
wilkinson et al . (
1986 ) found a ratio of maximum rotational velocity to peak nuclear velocity dispersion characteristic of a rotationally supported oblate system seen almost edge - on ( davies et al .
assuming the radio jet to mark the long axis , an effectively stationary rotating nearly oblate triaxial model with axis ratios 1:0.98:0.55 provided their best fit ; our line of sight is then 30@xmath2040@xmath20 from the plane defined by the two long axes . dropping the requirement of identical jet direction and stellar symmetry axis
, they also found a satisfactory fit to an effectively stationary prolate spheroidal model . following earlier work by tubbs ( 1980 ) , a prolate model with clouds moving in circular orbits in a warped disk
was favoured by quillen et al .
( 1992 ) on the basis of their co observations and the ionized gas kinematics determined by bland et al .
quillen et al . (
1993 ) presented a revised version of this model to reproduce the morphology of their near - infrared maps .
assuming the planetary nebulae to form a relaxed system ( implying a triaxial potential ) , a stationary galaxy figure , and the gas disk to be in the preferred plane defined by the long axis , hui et al .
( 1995 ) derived a model with axial ratios of 1:0.92:0.79 where the short axis lies in the plane of the sky , and the intermediate axis is along the line of sight .
mathieu , dejonghe @xmath1 hui ( 1996 ) used the planetary nebula system to construct triaxial models , assuming a spherical potential obtained by inverting the major axis photometry .
failing to obtain satisfactory fits with constant m / l ratios , they concluded to a significant presence of dark matter , causing a significant increase of the mass - to - light ratio with radius .
they consider the kinematics and photometry of ngc 5128 to be well described by two kinematically distinct subsystems supporting the merger hypothesis . on the whole
, there is substantial agreement that the inner parts of ngc 5128 have a roughly constant mass - to - light ratio @xmath132 @xmath39 4 increasing with radius to the extent that the _ global _ ratio of ngc 5128 is @xmath132 = 10 ( hesser et al .
1984 ; van gorkom et al . 1990 ; hui et al . 1995 ; mathieu et al . 1996 ) .
the total mass of the galaxy is @xmath133 = 4@xmath91 @xmath3 10@xmath134 m@xmath135 , with typically half of it due to dark matter ( mathieu et al .
the galaxy mass is thus between 2@xmath5 and 10@xmath5 of the ngc 5128/ngc 5236 group mass . interpreting the dust band as a precessing warped structure , nicholson et al .
( 1992 ) ruled out prolate models and concluded that its geometry is consistent with a near - polar gas disk in an oblate , almost spherical triaxial potential with its short axis likewise in the plane of the sky . in their models , the dust band is separated into an inner detached disk ( @xmath1361.7 kpc ) and an outer extended annulus ( 1.7 kpc @xmath1376.8 kpc ) .
sparke ( 1996 ) showed that , at least qualitatively , such models can explain not only the the observed disk features but also the broken hi ring observed by schiminovich et al .
an important factor is the variation in oblateness of the galaxy , which is taken to be nearly spherical at small radii and more flattened farther out .
sparke s ( 1986 ) model resembles that of wilkinson et al .
( 1986 ) but is in contradiction to that of hui et al .
( 1995 ) . the discrepancy may be resolved if , in fact , the planetary nebulae do not represent a relaxed and well - mixed system ( sparke 1996 ) . of particular interest
is the shell system referred to in sect .
4.1 and shown in figs . 6 and 7 .
malin et al . ( 1983 ) interpreted the shells as the signature of the collision between a dynamically cold system and a rigid potential well ( quinn , 1984 ; hernquist @xmath1 quinn 1988 , 1989 ) under the condition that the infalling system had a mass significantly less than that of the elliptical galaxy .
the shells arise from phase - wrapping of the disk after it has been disrupted by the tidal field of the massive elliptical galaxy .
in fact , this work , as well as that by tubbs ( 1980 ) served to underpin the galactic - encounter interpretation of ngc 5128 first suggested by baade @xmath1 minkowski ( 1954 ) and revived by graham ( 1979 ) .
malin et al . ( 1983 ) suggested that a late - type galaxy of mass a few times 10@xmath57 @xmath14 , similar to e.g. m 33 , merged with ngc 5128 a few hundred million years ago .
specific , more detailed merger scenarios were presented by tubbs ( 1980 ) and quillen et al .
( 1993 ) .
if the gas dust disk would define a preferred plane in a triaxial system , its rotation should be retrograde with respect to the tumbling motion of the stellar ellipsoid ( van albada , kotanyi @xmath1 schwarzschild 1982 ) . however , wilkinson et al .
( 1986 ) found that the disk has in fact prograde rotation , and suggested the warped disk to be due to incomplete settling of material into a symmetry plane of the potential .
this is consistent with the conclusion by van dokkum @xmath1 franx ( 1995 ) that dust disks in early - type galaxies , with a radius exceeding 250 pc , are generally not settled .
the disk would thus be a transient phenomenon , unless it were stabilized by self - gravity ( nicholson et al . 1992 ; see also sparke 1996 ) .
a stable disk is required if the merger and dust band formation occurred some 10@xmath83 years ago ( graham 1979 ; nicholson et al .
quillen et al.s ( 1992 ; 1993 ) ` short ' timescales of 12 @xmath3 10@xmath21 years were criticized by sparke ( 1996 ) who estimated three quarters of a gigayear as the time elapsed since capture of a companion galaxy from a polar orbit .
the presence of an intermediate age stellar population in the halo of ngc 5128 prompted soria et al .
( 1996 ) to suggest an even longer timescale of several gigayear , while intermediate scales of 28 @xmath3 10@xmath21 years ( and incomplete settling ) have been suggested by tubbs ( 1980 ) and malin et al .
all these timescales are , however , much longer than the age of the current burst of star formation in the disk which is typically a few times 10@xmath0 years ( van den bergh 1976 ; dufour et al . 1979 )
hi line absorption against the centre of cen a over the velocity range of 500600 @xmath7 was first recognized in single - dish observations by roberts ( 1970 ) and whiteoak @xmath1 gardner ( 1971 ; 1976b ) .
it was studied at high spatial ( 2.6@xmath2611@xmath25 ) and moderate velocity ( 6 @xmath7 ) resolutions by van der hulst , golisch @xmath1 haschick ( 1983 ) .
subsequently , centimetre wavelength absorption by various molecular species ( oh , h@xmath80co , c@xmath104h@xmath80 , nh@xmath104 ) was observed by gardner @xmath1 whiteoak 1976a , b ; 1979 ; seaquist @xmath1 bell 1986 ; 1988 ; 1990 ) .
the hi observations by van der hulst et al .
( 1983 ) show the inner jets ( 20@xmath25 and 30@xmath25 northeast of the nucleus ) as well as the nuclear source in absorption .
the inner jets are also seen in h@xmath80co absorption ( seaquist @xmath1 bell 1990 ) , while h@xmath80co absorption is likewise found in the direction of the dust band , 4@xmath4 southeast from the nucleus ( gardner @xmath1 whiteoak 1976b ) . the flat and strong nuclear continuum spectrum at millimetre wavelengths ( sect . 5.4 )
provides the opportunity to extend absorption studies to wavelengths where molecular line transitions are most abundant .
high velocity resolution ( @xmath138 2.6 @xmath7 ) absorption spectra of the lower transitions of @xmath13co and @xmath78co were obtained by israel et al .
( 1990 ; 1991 ) and eckart et al .
( 1990b ) .
various other molecular species , such as hco@xmath103 , h@xmath78co@xmath103 , hcn , hnc , cs , c@xmath80h , cn , c@xmath104h@xmath80 and h@xmath80co have likewise been detected in absorption ( eckart et al .
1990b ; israel et al .
1991 ; wiklind @xmath1 combes 1997 ; israel et al .
unpublished ) .
centaurus a is unique in this sense : as of yet , no other active galactic nucleus is known to exhibit absorption by a molecular disk ( drinkwater , combes @xmath1 wiklind 1996 )
. below 10 ghz , emission from the inner and nuclear jet dominates the continuum , whereas above 30 ghz the continuum emission arises mostly or wholly from the nucleus .
thus , centimetre - wavelength single - dish measurements sample sightlines as far as 500 pc away from the nucleus while aperture synthesis observations , after elimination of the inner jet contribution , preferentially sample sightlines missing the nucleus itself by about 1.5 parsec , i.e. 200 times the diameter of the nuclear source . only at ( sub)millimetre wavelengths
does the absorption sample a narrow line of sight ( 0.5 milliarcsec ) directly to the nucleus . a typical molecular line spectrum of the centaurus a nucleus consists of a relatively strong emission line arising from the molecular disk ( sect .
4 ) as well as weaker and broader emission representing the circumnuclear disk ( sect . 4.2 ) superposed on the nuclear continuum ( _ cf .
_ fig . 1 in israel , 1992 ) .
as absorption occurs over a significant fraction of the total emission line spectrum , the intrinsic shape of the latter is hard to determine but of crucial importance in determining the pure absorption spectrum .
the absorption spectra show numerous peaks at various velocities ( fig .
12 ) , relatively strong between @xmath139 , i.e. within 10 @xmath7 from the systemic velocity .
it is not clear which of these lines , if any , marks the systemic velocity .
there are at least five ( blended ) components in this velocity range but only two of these have discernible hi counterparts in vla spectra ( van der hulst et al .
1983 ; j.m .
van der hulst , private communication ) .
as these two hi features are also seen against the inner jets at 30@xmath25 ( 500 pc ) above the circumnuclear disk , they originate most likely in the more distant molecular disk associated with the dust band structure ( _ cf . _ van der hulst et al .
this is consistent with the apparent low excitation of the corresponding molecular absorption lines , which suggest the material to be located at considerable distance to the nucleus ( gardner @xmath1 whiteoak 1979 ; seaquist @xmath1 bell 1990 ; eckart et al .
the velocity range is consistent with the velocity dispersion in the molecular disk ( quillen et al .
1992 ) . of particular interest is a second absorption line system , extending from @xmath11 = 560 @xmath7 to 650 @xmath7 .
it is seen only against the nuclear source , and not against the inner jets ( disregarding the marginal h@xmath80co feature seen by seaquist @xmath1 bell 1990 ) .
this high - velocity , redshifted absorption system takes the shape of continous molecular absorption with a few individual peaks of relatively low optical depth ( israel et al .
1991 ; wiklind @xmath1 combes 1997 ) .
the redshifted absorption system is also prominent in hi ( gardner @xmath1 whiteoak 1976b ; van der hulst et al .
with respect to e.g. hco@xmath103 , the hi absorption is relatively strong at 560 @xmath140 600 @xmath7 , and weak at higher velocities .
this redshifted system is generally interpreted as due to infalling clouds ( gardner @xmath1 whiteoak 1976b ; van der hulst 1983 ; seaquist @xmath1 bell 1990 ; israel et al .
it implies accretion rates sufficiently high to explain the overall radio luminosity of centaurus a ( van gorkom et al . 1989 ) .
as the absorption optical depth ratios of the various species appear to be a function of ( infall ) velocity , this opens the interesting possibility of modelling the processing of material falling into the nucleus of centaurus a. as the optical depths and consequently signal - to - noise ratios are low , this is not an easy task . at velocities @xmath141
there is a hint of a very low optical depth blueshifted hco@xmath103 absorption wing ( _ cf . _
israel 1992 ; wiklind @xmath1 combes 1997 ) .
its nature is not clear , and it is not seen in hi ( j.m .
van der hulst , private communication ) .
high resolution oh maser absorption spectra against the nuclear source in the four transitions at 18 cm wavelength show exactly conjugate behaviour at all velocities in the satellite lines : the sum of the two transitions is zero ( van langevelde et al . 1995 ) .
this effect is caused by the competition of the two transitions for the same pumping photons and allows a direct determination of the oh column density @xmath142 6 @xmath3 10@xmath143 .
ngc 5128 is a massive elliptical galaxy at the heart of a moderately rich group of galaxies ( sect .
1.4 ) ; basic data are summarized in table 3 .
although most of its properties are fairly normal for a luminous triaxial elliptical galaxy , it is remarkable in two aspects : it hosts a very large radio source ( sect . 2 ) and
its inner parts harbour a relatively massive disk of dust , gas and young stars ( sect . 4 ) .
both have been proposed as the consequence of past merger activity .
the location of the galaxy amidst several dwarf galaxies lends plausibility to such a suggestion . indeed , although no _ direct _ evidence of a merger has been found , the appearance of the galaxy and in particular the properties of its embedded disk , such as gas mass , kinematics , warp and polar orientation along the photometric minor axis ( _ cf .
_ bertola et al .
1988 ) as well as the bimodality of its globular cluster system ( zepf @xmath1 ashman 1993 ; perelmuter 1995 ) , the existence of luminous optical and and hi shells as well as the outcome of various dynamical models and scenarios ( sect .
6.2 ) all provide strong indirect evidence that at least one major merger event occurred some 10@xmath2110@xmath83 years ago .
the gas mass of the dusty disk , a few times 10@xmath83 @xmath14 , point at capture of a fairly sized late - type spiral galaxy rather than a small irregular .
the shell structures in particular suggest that ngc 5128 experienced more than just one merger ( weil @xmath1 hernquist 1996 ) .
the radio source centaurus a , associated with ngc 5128 , is a very near example of a large class of radio galaxies of moderate luminosity known as fr - i galaxies ( fanaroff @xmath1 riley 1974 ) .
radio sources of this class are generally presumed to have moderately active nuclei with relativistic outflows on a subparsec scale not aligned with our line of sight .
the observations of centaurus a at radio and x - ray/@xmath110 ray wavelengths are consistent with this interpretation .
ngc 5128 contains a very compact nucleus of size 1200@xmath9500 a.u .
from which subluminal relativistic jets emanate ( sect .
5.4 ) that become subrelativistic within 1.5 pc .
the jets appear to propagate at a large angle to our line of sight .
the nucleus itself is strongly obscured by a small ( radius @xmath39 150 pc ) circumnuclear disk ( sect .
5.2 ) and is quite variable at radio and x - ray wavelengths ( sect .
the polarization of the central region , the ionization of the optical filaments and the apparent similarity of the high - energy spectrum ( but not the luminosity ) of centaurus a ( in particular at @xmath110-ray energies ) to that of blazars and quasars such as 3c273 , have been used to argue that the galaxy indeed harbours a misaligned bl lac / blazar nucleus ( bailey et al .
1986 ; morganti et al .
1991 ; dermer @xmath1 schlickeiser 1993 ; kinzer et al . 1995
; see also steinle et al .
1998 , and references therein ) .
the substantially lower luminosity of centaurus a is then explained by our viewing the galaxy from the side , and not down the jet axis .
however , some caution to this conclusion has been expressed by antonucci @xmath1 barvainis ( 1990 ) and kellerman et al .
( 1997 ) .
lrll & value & units & reference + @xmath12(1950)@xmath144 & 13:22:[email protected] & & giles 1986 + @xmath31(1950)@xmath144 & -42:45:[email protected] & & giles 1986 + galactic longitude @xmath145 & 309.5 & degrees & + galactic latitude @xmath50 & + 19.4 & degrees & + systemic velocity @xmath11 & 543@xmath92 & @xmath7 & table 1 + galaxy size d@xmath146 & 18@xmath314 & arcmin & rc2 + radio source size & 8@xmath34 & degrees & section 2.1 + distance & [email protected] & mpc & section 1.2 + apparent magnitude @xmath59 & 7.96 & mag & rc2 + colour @xmath68 & 0.98 & mag & rc2 + foreground reddening @xmath63 & 0.11 & mag & section 3.4 + total galaxy mass & 4@xmath91 @xmath3 10@xmath134 & @xmath147 & mathieu et al .
1996 + total hi mass & [email protected]@xmath3 10@xmath21 & @xmath147 & section 4.2 + gas mass dusty disk & [email protected]@xmath3 10@xmath83 & @xmath147 & section 4.2 + gas mass circumnuclear disk & [email protected]@xmath3 10@xmath0 & @xmath147 & section 5.2 + linear sizes : & & & + outer radio lobe & 250 & kpc & section 2.1 + middle radio lobe & 30 & kpc & section 2.2 + inner radio lobe & 5 & kpc & section 2.3 + inner radio jet & 1.35 & kpc & section 2.3 + relativistic nuclear jet & 1.65 & pc & jones et al .
1996 + radio core & 0.008 & pc & kellerman et al .
1997 + radius dusty disk & 7 & kpc & section 4.1 + radius circumnuclear disk & 150(-40 , + 130 ) & pc & section 5.2 + nuclear activity must have been going on for a considerable amount of time , given the size of the outer radio lobes .
the bulk speeds of 5000 @xmath7 estimated for the inner jets ( sect .
2.4 ) and the outer radius of 250 kpc of the giant lobes of radio emission suggest _ a lower limit _ of 50 million years . as the inner jets appear to dissolve into plumes ( ` inner lobes ' ) at about 5 kpc from the nucleus ( sect .
2.3 ) , and as the position angle of the outer radio features is much different from that of the inner features , it is reasonable to conclude that significant deceleration occurs over most of the radio source , leading to a substantially higher age .
indeed , the age of the inner lobes alone was already estimated at 6 @xmath3 10@xmath21 years ( slee et al .
1983 ) , although this may be too high .
the jets appear to lose much of their energy within a few parsec from the nucleus , presumably by interaction with ambient material .
the peculiar radio brightness evolution of component c1 in the nuclear jet may provide a clue to this process ( tingay et al . 1998 ) underscoring the need for further vlbi monitoring of centaurus a , as well as the desirability of filling the resolution gap in the 0.10.3@xmath25 range .
the inner jets dissolve in the more extended inner lobe plumes , which exhibit a profound clockwise bending ( decreasing position angle ) .
again , ambient material and its movement in the galaxy , may explain the observed morphology ( see e.g. sparke 1982 ; gopal - krishna @xmath1 saripalli 1984 ; heckman et al .
1985 ) , but hard evidence is lacking .
moreover , this is unlikely to also explain the similarly profound clockwise bending of the giant outer lobes , well outside the optical galaxy . noting a continuous decrease of position angle ( i.e. clockwise bending ) of various features at increasing distance to the nucleus , haynes et al .
( 1983 ) have proposed that the central collimating source precesses at a rate of the order of 10@xmath148 degrees per year .
the discovery of a circumnuclear disk perpendicular to the nuclear and inner jet , yet inclined to the minor axis of the elliptical galaxy , supports the idea of precession .
if the rate of procession is correctly estimated , the structure of the radio source should exhibit the effect of several precessional periods
. it would be interesting to see whether the run of position angles with radius can indeed be modelled by such a precession of the collimating agent .
alternatively , a combination of precession and ambient gas dynamics may be required , while the structure of the outer lobes , in addition , may be influenced by tumbling and orbital motion of the galaxy as a whole ( burns et al .
1983 ) . the putative age of the merger ( sect .
6.3 ) suggests a link to the origin of the radio source .
although the presence of an active nuclear source predating a merger can not be excluded , it is tempting to associate its origin with the accumulation of matter in the centre caused by transfer of angular momentum through viscous damping after such an event .
an intriguing indication that the origin of the radio source is connected to merger activity is provided by a morphological argument . the bended appearance of the giant radio lobes ( fig . 4 ) is very similar to that of the tilted rings forming the dust band ( _ cf . _ fig .
11b in nicholson et al .
1992 ) rotated by 90@xmath20 .
for instance , the position angle of 0@xmath20 characterizing the outer radio contours corresponds to the position angle of 90@xmath20 of the outer rings . because the dynamical time scale of the outer rings is much longer than that of the more strongly tilted inner rings , their _ present
_ position angle should more closely resemble the _ original _ orientation of the inner disk structure at the time that the matter now forming the outer lobes was ejected . as the dust disk originated in a merger event , and the morphology of the radio lobes appear to follow its subsequent evolution , it seems likely that the activity creating these lobes is also a consequence of the merger event .
if the nuclear source is in fact a black hole , its estimated bolometric luminosity ( half of it at high energies ) of about 10@xmath149 erg s@xmath38 implies a _ lower limit _ to the black hole mass of about 5@xmath310@xmath150 whereas the total luminosity of the radio source suggests a mass @xmath151 ( terrell 1986 and references therein ; kinzer et al .
1995 ) . as the dynamical mass within 40 pc is about 4@xmath310@xmath21 @xmath14 , and
there is no obvious sign of keplerian rotation ( see fig .
10 ) , its _ upper limit _ must be a few times 10@xmath0 @xmath14 .
although the high obscuration of the centre of ngc 5128 and the lack of h@xmath80o masers ( braatz et al . 1996 ) precludes , at the moment , a more accurate mass determination , the actual mass of the putative black hole nevertheless is fairly well - constrained and is comparable to that of the circumnuclear disk ( sect .
this mass is not very high and infalling molecular clouds , especially dense cores , may penetrate deeply before being tidally disrupted .
the variability of the nucleus may represent the accretion of individual stellar or cloud remnants onto the black hole triggering renewed jet activity ( sect .
5.4 through 5.6 ) and fueling the radio source .
details of these processes are not clear yet , but careful and frequent monitoring of centaurus a at radio , x - ray and @xmath110-ray wavelengths may provide important information .
for instance , how does the nucleus drive the nuclear jets , and how are the relativistic nuclear jets transformed into the nonrelativistic inner jets ?
the circumnuclear disk ( sect .
5.2 ) does not seem capable of controlling the collimation of the nuclear jets , but its orientation exactly perpendicular to these jets , suggests that it is somehow connected with the collimating agent .
comparison of centaurus a features with very - high resolution observations ( hst , vlba ) of other active elliptical galaxies suffering less nuclear extinction , such as the ten times more distant ngc 4261 ( e.g. jones @xmath1 wehrle 1997 , and references therein ) may prove particularly fruitful .
it is a pleasure to thank david malin , do kester , norbert junkes , thijs van der hulst , stphanie ct , steven tingay , jack burns and paul van der werf for kindly supplying the illustrations in this review .
i also would like to thank paul van der werf , hans bloemen , george miley and in particular tim de zeeuw for critical comments on an earlier version of this work .
the burden of literature searches was greatly relieved by the use of the nasa astrophysics data system ( ads ) astronomy abstract service abraham z. , 1996 , in : ` extragalactic radio sources ' , iau symposium 175 , eds .
r. ekers , c. fanti , l. padrielli , ( dordrecht : kluwer ) , p. 25
adams d.j . ,
adamson a.j . , giles a.b .
, 1983 202 , 241 allen l.r . , hanbury brown r. , palmer h.p . , 1963 125 , 57 allen w.h .
, bond i.a . , budding e. , et al .
1993 405 , 554 alonso m.v .
, minniti d. , 1997 109 , 397 antonucci r. , barvainas r. , 1990 363 , l17 arp h. , 1994 288 , 738 baade w. , minkowski r. , 1954 119 , 215 bailey j. , sparks w.b . , hough j.h .
, axon d.j . ,
1986 nature 32 , 150 baity w.a .
, rothschild r.e . ,
lingenfelter r.e .
, et al . , 198 244 , 429 bao g. , wiita p.j . , 1997 485 , 136 beall j.h . , rose w.k . , graf w. , et al .
1978 219 , 836 becklin e.e .
, frogel j.a . ,
kleinmann d.e .
, et al . , 1971 170 , l15 bell m.b . , seaquist e.r . , 1988 329 , l17 bertola f. , galletta g. , kotanyi c. , zeilinger w.w .
, 198 234 , 733 bland j. , taylor k. , atherton p.d . , 1987 228 , 595 bolton j.g . , 1948 nature 162 , 141 bolton j.g . ,
stanley g.j . ,
slee o.b . , 1949
nature 164 , 101 bond i.a . , ballet j. , denis m. , et al .
1996 307 , 708 botti l.c.l , abraham z. , 1993 264 , 807 braatz j.a . ,
wilson a.s .
, henkel c. , 1996 106 , 51 branch d. , romanishin w. , baron e. , 1996 465 , 73 branch d. , fisher a. , baron e. , nugent p. , 1996
470 , l7 brodie j. , knigl a. , bowyer s. , 1983 272 , 154 brodie j. , bowyer s. , 1985 292 , 447 burns j.o .
, feigelson e.d . ,
schreier e.j . , 1983 273 , 128 cannon r.d . , 1981 in ` optical jets in galaxies ' ,
esa sp 162 p. 45
carramiana a. , chadwick p.m. , dipper n.a . , et al . , 1990 228 , 327 christiansen w.n . ,
frater r.h .
, watkinson a. , osullivan j.d . ,
lockhart i.a . , 1977 181 , 183 ciardullo r. , ford h.c .
, williams r.e .
, tamblyn p. , jacoby g.h .
, 1990 99 , 1079 clarke d.a . , burns j.o .
, feigelson e.d .
, 1986 300 , l41 clarke d.a . , burns j.o , norman m.l .
, 1992 395 , 444 combi j.a . ,
romero g.e . , 1997 121 , 11 condon j.j . , helou g. , sanders d.b . , soifer b.t . , 1996 103 , 81 cooper b.f.c . , price r.m
, cole d.j .
, 1965 aust .
18 , 589 ct s. , 1995 , ph .
d. thesis mt .
stromlo and siding spring observatories , australian national university ct s. , freeman k.c .
, carignan c. , quinn p.j . , 199 114 , 1313 cristiani s. , cappellaro e. , turatto m. , et al .
1992 259 , 63 cunningham c.t . ,
ade p.a.r . ,
robson e.i . ,
radostitz j.v . , 1984 211 , 543 danziger i.j . , 1981 in ` optical jets in galaxies ' , esa sp 162 , p. 143
davies r.l . ,
efstathiou g. , fall m. , illingworth g. , schecter p.l . , 1983 266 ,
41 de vaucouleurs g. , 1975 , in _ stars and stellar systems _ , vol . 9 , _ galaxies and the universe _ , ed .
a. sandage , m. sandage @xmath1 j. kristian ( chicago : univ .
chicago press ) , p. 557
de vaucouleurs g. , de vaucouleurs a. , corwin h.g . , 1976 , _ second reference catalogue of bright galaxies _ , ( austin : univ .
of texas press),(rc2 ) de vaucouleurs g. , 1993 415 , 10 de zeeuw p.t .
, franx m. , 1991 29 , 239 dermer c.d . , schlickeiser r. , 1993 416 , 458 dbereiner s. , junkes n. , wagner s.j .
, et al . 1996 470 , l15 dodorico s. , di serego alighieri s. , pettini m. , et al . ,
1989 215 , 21 dottori h.a .
, fourcade c.r . , 1973 23 , 405 drinkwater m.j .
, combes f. , wiklind t. , 1996 312 , 771 dufour r.j . ,
van den bergh s. , 1978 226 , l73 dufour r.j . ,
van den bergh s. , harvel c.a .
, et al . 1979 84 , 284 ebneter k. , balick b. , 1983 95 , 675 eckart a. , cameron m. , rothermel h. , et al . , 1990a 363 , 451 eckart a. , cameron m. , genzel r. , et al . , 1990b 354 , 522 elvius a. , hall j.s . , 1964 bull .
lowell obs .
vi , 123 evans r. , 1986 iau circular no .
4208 fanaroff b.l .
, riley j.m . , 1974 167 , 31p feigelson e.d . ,
schreier e.j .
, delvaille j.p . , et al . , 1981 251 , 31 fichtel c.e . , bertsch d.l . , chiang j. , et al . 1994 94 , 551 fogarty w.g . , schuch n.j . , 1975 nature 254 , 124 frogel j.a . , 1984 278 , 119 frogel j.a . , gregory b. , kawara k. , et al . , 1987 315 , l129 gardner f.f . , whiteoak j.b . , 1976a 175 , 9p gardner f.f . , whiteoak j.b . , 1976b proc .
a.s.a . 3 , 63 gardner f.f . , whiteoak j.b . , 1979 189 , 51p gehrels n. , cline t.l . ,
teegarden b.j . ,
1984 278 , 112 giles a.b . , 1986 218 , 615 1986 gopal - krishna , saripalli l. , 1984 141 , 61 gouveia dal pino e.m . , opher r. , 1989 342 , 686 graham j.a . , 1979 232 , 60 graham j.a . , 1981 93 , 291 graham j.a .
, 1983 269 , 440 graham j.a . , phillips m.m . , 1980 239 , l97 graham j.a . , price r.m . , 1981 247 , 813 grasdalen g.l . , joyce r.r . , 1976 208 , 317 griffin r.e . , 1963 68 , 615 grindlay j.e . , 1975 199 , 49 grindlay j.e . , helmken h.f . , hanbury brown r. , davis j. , allen l.r . , 1975 197 , l9 haddock f.t . , mayer c.h . ,
sloanaker r.m . , 1954 119 , 456 harding p. , jones t.j . , rodgers a.w . , 1981 251 , 530 harris g.l.h . , hesser j.e . , harris h.c . ,
curry p.j . , 1984a 287 , 175 harris h.c . ,
harris g.l.h .
, hesser j.e . ,
mcgillivray h.t .
, 1984b 287 , 185 harris h.c . ,
harris g.l.h .
, hesser j.e . ,
1986 in _ the harlow shapley symposium on on globular cluster systems in galaxies _ , iau symp .
126 , eds .
grindlay @xmath1 a.g.d .
phillips ( dordrecht : reidel ) , p. 205
harris g.l.h . , geisler d. , harris h.c . , hesser j.e . , 1992 104 , 613 haslam c.g.t . , klein u. , salter c.j . , et al
1981 , 100 , 209 hawarden t.g .
, elson r.a.w . , longmore a.j .
, tritton s.b .
, corwin h.g . , 1981 196 , 747 hawarden t.g . ,
sandell g. , matthews h.e .
, friberg p. , watt g.d .
, smith p.a . , 1993 260 , 844 haynes r.f . , cannon r.d . , ekers r.d . , 1983 proc .
asa 5 , 241 heckman t.m . ,
illingworth g.d . ,
miley g.k . , van breugel w.j.m . , 1985 299 , 41 hernquist l. , quinn p.j . , 1988 331 , 682 hernquist l. , quinn p.j . , 1989 342 , 1 herschel j.f.w . , 1847 ,
_ results of astronomical observations at the cape of good hope _
( london : smith , elder ) , pages 105 and 20 respectively hesser j.e . ,
harris h.c .
, van den bergh s. , harris g.l.h .
1984 276 , 491 hesser j.e . ,
harris h.c . ,
harris g.l.h . , 1996 303 , l51 hough j.h . ,
bailey j.a .
, rouse m.f . , whittet d.c.b .
, 1987 227 , 1p hubble , e. , 1922 56 , 162 hui x. , ford h.c . , ciardullo r. , jacoby g.h . , 1993a 414 , 463 hui x. , ford h.c . , ciardullo r. , jacoby g.h .
, 1993b 88 , 423 hui x. , ford h.c . ,
freeman k.c . ,
dopita m.a . , 1995 449 , 592 hunter d.a . , gallagher j. , rice w.l . ,
gillett f.c . , 1989 336 , 152 hodge p.w . ,
kennicutt r.c . , 1983 88 , 296 israel f.p . , 1992 265 , 487 israel f.p . , van dishoeck e.f . , baas f. , et al .
1990 227 , 342 israel f.p .
, van dishoeck e.f . , baas f. , de graauw t. , phillips t.g . , 1991 245 , l13 israel f.p .
, van der werf p.p . ,
hawarden t.g . , aspin c. , 1998 submitted jablonka p. , bica e. , pelat d. , alloin d. , 1996 307 , 385 jacoby g.h . , branch d. , ciardullo r. , et al .
1992 104 , 599 jauncey d.l .
, tingay s.j . ,
preston r.a . , et al . , 1995 proc .
usa 92 , 11368 joy m. , lester d.f . , harvey p.m. , ellis h.b .
, 1988 326 , 662 joy m. , harvey p.m. , tollestrup e.v . , et al . , 1991 366 , 82 jones p.a . , mcadam w.b .
, reynolds j.e .
1994 268 , 602 jones d.l . ,
tingay s.j . ,
murphy d.w . , et al . , 1996 466 , l63 jones d.l . , wehrle a.e . , 1997 484 ,
186 jourdain e. , bassani l. , roques j .-
p . , 1993 412 , 586 junkes n. , haynes r.f .
, harnett j.i . , jauncey d.l . , 1993 269 , 29 kellerman , k.i . ,
1974 194 , l135 kellerman k.i .
, zensus j.a . , cohen m.h . ,
1997 475 , l93 kinzer r.l . , johnson w.n . ,
dermer c.d . , et al . , 1995 449 , 105 koornneef j. , israel f.p . , 1996 new astr . 1 , 271 kormendy j. , djorgovski s. , 1989 27 , 235 krolik j.h . , lepp s. , 1989 347 , 179 kunkel w.e . ,
bradt h.v . , 1971 170 , l7 lpine j.r.d . , braz m.a . , epchtein n. , 1984 131 , 72 malin d.f . , quinn p.j
. , graham j.a . , 1983 272 , l5 marcelin m. , boulesteix j. , courts g. , milliard b. , 1982 nature 297 , 38 malin d.f . , quinn p.j . , graham , j.a . , 1983 272 ,
l5 marshall f.j .
, clark g.w . , 1981 245 , 840 )
marston a.p .
, dickens r.j . , 1988 193 , 27 marston a.p . , 1992 in _ relationships between galactic nuclei and starburst galaxies _ , asp conference series 31 , p. 123
mathieu a. , dejonghe h. , hui x. , 1996 309 , 30 maiolino r. , rieke g.h . , 1995 454 , 95 mcelroy d.b . ,
humphreys r.m . , 1982 94 , 828 meadows v. , allen d.a .
, 1992 proc .
10 , 104 meier d.l .
, jauncey d.l . ,
preston r.a . , et al . , 1989 98 , 27 miley g.k . , 1980 18 , 165 mills b.y . , 1952 austral
. j. phys . 5 , 456 minniti d. , alonso m.v . ,
goudfrooij p. , jablonka p. , meylan g. , 1996 467 , 221 miyazaki s. , takahashi t. , gunji s. , et al . , 1996 48 , 801 morganti r. , robinson a. , fosbury r.a.e .
, di serego alighieri s. , tadhunter c.n .
, malin d.f . , 1991 249 , 91 morganti r. , fosbury r.a.e . ,
hook r.n . , robinson a. , tsvetanov z. , 1992 256 , 1p morini m. , anselmo f. , molteni d. , 1989 347 , 750 mouri h. , 1994 427 , 777 mushotzky r.f .
, serlemitsos p.j . ,
becker r.h . ,
boldt e.a . ,
holt s.s . , 1978 220 , 790 newell e.b . , rodgers a.w . , searle l.t . ,
1969 158 , 699 nicholson r.a . ,
bland - hawthorn j. , taylor k. , 1992 387 , 503 norman m.l . , burns j.o , sulkanen m.s .
1988 nature 335 , 146 oneill t. , tumer o.t .
, zych a. , white r.s . , 1989 339 , 78 osmer p.s . , 1978 226 , l79 packham c. , hough j.h . , young s. , et al . , 1996 278 , 406 paglione t.a.d .
, jackson j.m . , ishizuki s. , 1997 484 , 656 paraskevoulos , j.s . ,
1935 harvard bulletin 890 , 1 perelmuter j .- m .
, 1995 454 , 762 phillips m.m . , 1993 413 , l105 phillips m.m . , taylor k. , axon d.j . , atherton p.d . ,
hook r.n . , 1984
nature 310 , 554 phillips m.m . , phillips a.c . , heathcote s.r . , 1987 99 , 592 phillips t.g . , ellison b.n .
, keene j.b . , et al .
1987 322 , l73 preston r.a . ,
wehrle a.e . ,
morabito d.d . , et al . ,
1983 266 , l93 quillen a.c . , de zeeuw p.t . ,
phinney e.s . ,
phillips t.g . , 1992 391 , 121 quillen a.c . , graham j.r .
, frogel j.a . , 1993 412 , 550 rice w. , lonsdale c.j . , soifer b.t .
1988 68 , 91 richter o .- g . , sackett p.d .
, sparke l.s . , 1994 107 , 99 riess a.g . , press w.h .
, kirshner r.p . , 1996 473 ,
88 rodgers a.w .
, harding a.w . , 1980 , 236 , l17 romero g.e . ,
benaglia p. , combi j.a .
, 1997 124 , 307 ruiz - lapuente p. , lucy l.b .
, danziger i.j . , 1992 mem .
astr . it .
63 , 233 rydbeck g. , wiklind t. , cameron m. , et al .
1993 270 , l13 sadler e.m . , gerhard o.e . , 1985 214 , 177 sandage a.r .
, tammann g.a .
1974 194 , 551 scarrott s.m .
, foley n.b . ,
gledhill t.m . ,
wolstencroft r.d . , 1996 282 , 252 schreier e.j . , feigelson e. , delvaille j. , et al . , 1979 234 , l39 schreier e.j .
, burns j.o , feigelson e.d . , 1981 251 , 523 schreier e.j . , capetti a. , macchetto f. , sparks w.b .
, ford h.j . , 1996 459 , 535 schiminovich d. , van gorkom j.h . , van der hulst j.m . , kasow s. , 1994 423 , l101 seaquist e.r . , bell m.b .
, 1986 303 , l67 seaquist e.r . , bell m.b . ,
1990 364 , 94 shain c.a .
, 1958 aust .
j. phys 11 , 517 sheridan k.v .
, 1958 aust .
11 , 400 shopbell p.l . , bland - hawthorn j. , malin d.f . , 1993 106 , 1344 simpson c. , 1994 271 , 247 skibo j.g . , dermer c.d .
, kinzer r.l . , 1994 , 426 , l23 slee o.b . ,
sheridan k.v . ,
, little a.g . , 1983 proc .
asa 5 , 247 soria r. , mould j.r . , watson a.m. , et al .
1996 apj 465 , 79 sparke l.s . , 1982 254 , 456 sparke l.s . , 1996 473 , 810 steinle h. , bennett k. , bloemen h. , et al .
1998 330 , 97 storchi - bergmann t. , bica e. , kinney a.l . , bonatto c. , 290 , 231 sugizaki m. , inoue h. , sonobe t. , takahashi t. , yamamoto y. , 1997 49 , 59 sutherland r.s . , bicknell g.v .
, dopita m.a . , 1992 414 , 510 tateyama c.e . , strauss f.m . , 1992 256 , 8 telesco c.m . , 1978 226 , l125 terrell j. , 1986 300 , 669 thompson d.j . , bertsch d.l . ,
dingus b.l . , et al . , 1995 101 , 259 tingay s.j . , jauncey d.l . , reynolds j.e . , et al . , 1998 in press tonry j.l
. , schecter p.l . , 1990 100 , 1794 tonry j.l . , 1991 373 , l1 tornikoski m. , valtaoja e. , tersranta h. , et al . 1996
116 , 157 tubbs a.d . ,
1980 241 , 969 turner p.c .
, forrest w.j . ,
pipher j.l . ,
shure m.a . , 1992 393 , 648 turner t.j . , george i.m . , mushotzky r.f . ,
nandra k. , 1997 475 , 118 ubertini p. , bazzano a. , cocchi m. , la padula c. , sood r. , 1003 97 , 105 van albada t.s . , kotanyi c.g . , schwarzschild m. , 1982 198 , 303 van den bergh s. , 1976 208 , 673 van den bergh s. , hesser j.e . , harris g.l.h . , 1981 86 , 24 van der hulst j.m . , golisch w.f . , haschick a.d .
, 1983 264 , l37 van dokkum p.g . , franx m. , 1995 110 , 2027 van driel w. , de graauw th . , de jong t. , wesselius p.r . , 1993 101 , 207 van gorkom j.h . , knapp g.r . , ekers r.d . , et al . , 1989 97 , 708 van gorkom j.h . , van der hulst j.m . , haschick a.d
, tubbs a.d . , 1990 99 , 1781 van langevelde h.j . , van dishoeck e.f . ,
sevenster m.n . , israel f.p . , 1995 448 , l123 viegas s.m . , prieto m.a
. , 1992 258 , 483 von ballmoos p. , diehl r. , schoenfelder v. , 1987 312 , 134 wang b. , inoue h. , koyama k. , et al .
1986 38 , 685 webster b.l .
, goss w.m . , hawarden t.g . ,
longmore a.j . , mebold u. , 1979 186 , 31 weil m.l . , hernquist l. , 1996 460 , 101 whiteoak j.b . , gardner f.f . , 1979 ,
a.s.a . 3 , 319 whiteoak j.b . , gardner f.f . , hglund b. , 1979 , 190 , 17p wiklind t. , combes f. , 1997 324 , 51 wild w. , eckart a. , wiklind t. , 1997 322 , 419 wilkinson a. , sharples r.m .
, fosbury a.e . ,
wallace p.t . , 1986 218 , 297 quinn p.f . , 1984 279 , 596 zepf s.e . , ashman k.m . , 1993 264 ,
611 zickgraf f .- j . , humphreys r.m . , graham j.a
, phillips a. , 1990 102 , 920 zirbel e.l . , 1996 473 , 713 | at a distance of 3.4 mpc , ngc 5128 ( centaurus a ) is by far the nearest active radio galaxy .
it is often considered to be the prototype fanaroff - riley class i ` low - luminosity ' radio galaxy , and as such it plays an important role in our understanding of a major class of active galaxies .
its proximity has spawned numerous detailed investigations of its properties , yielding unrivalled but still incomplete knowledge of its structure and dynamics .
the massive elliptical host galaxy is moderately triaxial and contains a thin , strongly warped disk rich in dust , atomic and molecular gas and luminous young stars .
its globular cluster ensemble has a bimodal distribution of metallicities .
deep optical images reveal faint major axis extensions as well as a system of filaments and shells . these and other characteristics are generally regarded as strong evidence that ngc 5128 has experienced a major merging events at least once in its past .
the galaxy has a very compact , subparsec nucleus exhibiting noticeable intensity variations at radio and x - ray wavelengths , probably powered by accretion events .
the central object may be a black hole of moderate mass . towards the nucleus ,
rich absorption spectra of atomic hydrogen and various molecular species suggest the presence of significant amounts of material falling into the nucleus , presumably ` feeding the monster ' . emanating from the nucleus
are linear radio / x - ray jets , becoming subrelativistic at a few parsec from the nucleus . at about 5 kpc from the nucleus
, the jets expand into plumes .
huge radio lobes extend beyond the plumes out to to 250 kpc .
a compact circumnuclear disk with a central cavity surrounds the nucleus .
its plane , although at an angle to the minor axis of the galaxy , is perpendicular to the inner jets .
the jet - collimating mechanism , probably connected to the circumnuclear disk , appears to precess on timescales of order a few times 10@xmath0 years .
this review summarizes the present state of knowledge of ngc 5128 and its associated radio source centaurus a. underlying physical processes are outside its scope : they are briefly referred to , but not discussed .
3cm^-3 2cm^-2 2h_2 2_n(h2 ) _ |
the dark matter problem is one of the most fascinating and intriguing issue in the astroparticle physics .
this subject requires theoretical concepts and experimental techniques coming from cosmology , astrophysics and fundamental particle physics . by definition dark matter is a kind of non luminous matter .
its existence is required by a bunch of striking experimental evidences as well as by some compelling theoretical motivations .
one of most interesting experimental evidences is given by the observations of the spiral galaxy rotation curves @xcite .
they can be determined well outside the luminous core of the galaxy , measuring the circular velocities of the neutral hydrogen clouds using the @xmath1 cm lines emissions .
the result when expressed as a function of the galactocentric distance @xmath2 is that @xmath3 for large @xmath2 ( in particular for @xmath4 ) rather than the expected @xmath5 . if we suppose that the first newton s law still holds at the galactic scale we must conclude that @xmath6 where @xmath7 is the total mass inside a shell of radius @xmath2 . hence there must be a `` dark '' contribution to the galactic mass @xmath8 .
recent measurements @xcite of the anisotropy of the cosmic microwave background ( cmb ) strongly indicates that the total matter contribution is @xmath9 where @xmath10 is the hubble constant in units of 100 km s@xmath11 mpc@xmath11 .
hence @xmath12 is much larger than the baryonic term @xmath13 .
necessarily the dominant matter component must be of a non baryonic form .
it turns out that the best motivated candidates are some kind of weak interacting massive particles ( wimps ) .
usually these type of candidates are stable particles that appear in various extension of the standard model ( sm ) of particle physics .
from the theoretical point of view the most interesting extension of the sm involves supersymmetry . here
we will focus in particular on @xmath14 supersymmetric extension of the sm with soft supersymmetry breaking terms .
this is the so called minimal supersymmetric standard model ( mssm ) @xcite .
usually the lightest supersymmetric particle ( lsp ) that appear in the mssm mass spectrum is a good cdm candidate @xcite .
it turns out that the lsp is often the lightest neutralino .
the four neutralino gauge eigenstates are linear combination of the gaugino and higgsino fields : @xmath15 where @xmath16 .
the corresponding mass eigenstates can be obtained by diagonalizing the mass matrix : @xmath17 that depends by various mssm parameters , like the gaugino masses @xmath18 and @xmath19 ( that are associated , in the lagrangian , to explicit supersymmetry breaking terms ) or the @xmath20 parameter that appears in the higgs sector @xcite . the lowest eigenvalue of the mass matrix in equation [ neutmassmatrix ] is exactly the lsp , until the so called r - parity is conserved @xcite .
the neutralino is a majorana fermion , so it coincides with its own antiparticle , and it is a weak interacting particle .
the r - parity conservation implies that the lightest neutralino can not decay in sm particles and so it is a stable particle . for these reasons
the neutralino is a good cdm candidate .
moreover a neutralino pair can self annihilate into various sm particles that we can detect .
in the previous section we have seen that the neutralino is a good cdm candidate .
it is possible to study its properties through indirect detection of cosmic rays coming from pair annihilations in the dark galactic halo . in particular here we focus on the cosmic @xmath0-rays from the galactic center ( gc ) .
the egret data from the gc @xcite give a strong indication of an excess with respect of the `` standard '' model production of @xmath0-rays @xcite . in the standard picture , the diffuse @xmath0-ray background is due to the following processes : * @xmath21 production that promptly decays in @xmath22 @xmath23 where @xmath24 and @xmath25 are primary protons and heliums and @xmath26 is the interstellar hydrogen or helium . *
bremsstrahlung * inverse compton we are only interested in the energy range above @xmath27 gev where the dominant component is by far the @xmath21 production .
in the subsequent analysis we consider only this component for the diffuse background .
we want to explain the gc excess adding a neutralino induced component in the @xmath21 production .
the leading intermediate annihilation channels for the @xmath21 production , through fragmentation and/or decay processes , are often @xmath28 , @xmath29 , @xmath30 and @xmath31 ( see figure [ feynman - diag ] ) .
this is valid not only for the neutralino but also for a generic majorana fermion wimp , as for such particles the s - wave annihilation rate into lighter fermions is suppressed by a factor @xmath32 , where @xmath33 is the final state fermion mass and @xmath34 is the wimp mass . we can write the total diffuse @xmath0-ray flux as a sum of two components : @xmath35 where the background flux @xmath36 is splitted into two factors : @xmath37 where @xmath38 ( measured in @xmath39 ) is the local emissivity per hydrogen atom . the neutralino induced component @xmath40 is given by @xcite : @xmath41 where @xmath42 is the detector angular acceptance and where we have introduced the dimensionless function j that contains the dependence on the dark halo density profile : @xmath43 we have introduced the unknown normalization factor @xmath44 in order to take into account our ignorance of the exact hydrogen column density , _
i.e. _ the interstellar medium , while the @xmath45 factor parametrize our ignorance of the dark matter halo model .
results from n - body simulations @xcite@xcite give for for the dark matter density : @xmath46^{\left(\beta -\gamma\right)/\alpha}\ ] ] where @xmath47 , @xmath48 and @xmath0 are parameters that describe the halo profile ( isothermal sphere , nfw , moore , .. ) , @xmath49 is the wimp density measured here and @xmath50 is the galactocentric distance .
having introduced the neutralino induced component the fit of the gc egret data greatly improves . in the context of a simplified _ toy model _
@xcite , valid for a generic majorana fermion wimp of mass @xmath34 and with a fixed dominant intermediate annihilation channel , the result is shown in figure [ egretdata - fit ] .
gev ) induced component with @xmath30 dominant annihilation channel .
the best fit parameters @xmath44 and @xmath45 are indicated in the upper right corner . ]
our analysis shows that the egret data fit improves for small neutralino masses . in order to extract much more information about the gc egret excess we have to wait for the upcoming generation of telescope satellite , like glast @xcite . with respect to egret
, glast has a wider energy range , increased effective area and better energy and angular resolution .
hence we have studied what kind of data glast would collect from the gc , under the hypotesis that the excess , as mapped by egret , is due to wimp annihilations . relying on a simplified picture of the glast detector performances ( energy resolution of about @xmath51 , angular resolution of @xmath52 sr and peak effective area of @xmath53 ) , we have computed the gc data set which will be obtained by glast in 2 years . we have superimposed the error bars associated to the statistical errors only for the chosen energy binning and with an angular acceptance of @xmath54 .
the result , in this case , is shown in figure [ egretdata - glast ] .
all the previous results are valid in the case of a generic wimp . at the end
we want to identify the neutralino as our wimp candidate .
to do so we must consider a `` realistic '' supersymmetric model , _
i.e. _ the mssm .
in order to reduce the number of free parameters of the mssm it is possible different supersymmetry breaking scenarios in which the supersymmetry breaking terms derive from an underlying high energy theory .
one of the most widely studied frameworks is the so - called minimal supergravity ( msugra ) or constrained mssm ( cmssm ) @xcite . in the msugra model there is an important assumption about the _ universality _ of all the coupling constants at the grand unification scale ( gut ) . with these constraints
the number of free parameters is only five : @xmath55 where @xmath56 is the common scalar mass , @xmath57 is the common gaugino mass and @xmath58 is the proportionality factor between the supersymmetry breaking trilinear couplings and the yukawa couplings .
@xmath59 denotes the ratio of the vacuum expectation values of the two neutral components of the su(2 ) higgs doublet , while the higgs mixing @xmath20 is determined ( up to a sign ) by imposing the electro - weak symmetry breaking ( ewsb ) conditions at the weak scale . in this context
the mssm can be regarded as an effective low energy theory .
the parameters at the weak energy scale are determined by the evolution of those at the unification scale , according to the renormalization group equations ( rges ) .
fixing @xmath59 , @xmath58 and @xmath60 we have performed an accurate scan in the @xmath61 msugra parameter space , looking for the minimum normalization factor @xmath45 needed to be able to single out the neutralino annihilation signal with glast . in this case
we have taken advantage of the better glast angular resolution ( @xmath62 ) with respect to egret . in this deeper analysis
we have also computed the neutralino relic density : @xmath63 solving the boltzmann equation that describes the time evolution of the neutralino number density @xmath64 : @xmath65\ ] ] this equation can be easily solved numerically including resonances , threshold effects and all possible coannihilation processes @xcite .
the results , for two possible choices of the parameters @xmath59 , @xmath58 and @xmath60 , are shown in figure [ msugra - plots ] .
plane , for the value of the normalization factor @xmath45 , that allows the detection of the neutralino @xmath0 ray signal , with glast .
the light shaded region corresponds to the cosmologically favoured region where @xmath66 , while the dark shaded one corresponds to models that are excluded either by incorrect ewsb , lep bounds violations or because the neutralino is not the lsp.,title="fig : " ] plane , for the value of the normalization factor @xmath45 , that allows the detection of the neutralino @xmath0 ray signal , with glast .
the light shaded region corresponds to the cosmologically favoured region where @xmath66 , while the dark shaded one corresponds to models that are excluded either by incorrect ewsb , lep bounds violations or because the neutralino is not the lsp.,title="fig : " ] the contour plots , in the msugra @xmath61 plane , show the normalization factor @xmath45 needed for glast detection , at @xmath67 level , of the neutralino induced @xmath0 ray signal .
we have performed an analysis of the possibility of an indirect detection of supersymmetric dark matter with cosmic @xmath0-rays .
the most important indication is that there is indeed room for supersymmetric dark matter in the already available @xmath0-ray data from egret .
moreover the computation of the expected @xmath0-ray flux coming from dark matter annihilation with upcoming detector glast shows that it will be possible to study in more details a supersymmetric signal .
this work has been performed in collaboration with alessandro cesarini , francesco fucito and aldo morselli in roma tor vergata university and with piero ullio at sissa in trieste .
i would like to thank the glast dark matter working group , in particular the slac people , for many interesting discussions on the argument .
spergel _ et al _ , astro - ph/0302209 .
h. e. haber _ et al _ , phys . rept .
* 117 * , 75 ( 1985 ) .
k. griest _ et al _ , phys .
rept . * 333 * , 167 ( 2000 ) .
s. p. martin , arxiv : hep - ph/9709356 .
h. mayer - hasselwander _ et al _ , astron
. astrophys . * 335 * , 161 ( 1998 ) .
a. w. strong _ et al _ , astrophys .
j. * 537 * , 763 ( 2000 ) [ erratum - ibid . *
541 * , 1109 ( 2000 ) ] .
a. cesarini _ et al _ , arxiv : astro - ph/0305075 .
proposal for the gamma - ray large area space telescope , slac - r-522 ( 1998 ) ; glast proposal to nasa a0 - 99 - 055 - 03 ( 1999 ) .
j.f . navarro _
et al _ , astrophys . j. * 462 * , 563 ( 1996 ) .
s. ghigna _ et al _ , astrophys .
j. * 544 * , 616 ( 2000 ) . l. j. hall _ et al _ , phys .
d * 27 * , 2359 ( 1983 ) . s. p.
martin _ et al _ , phys .
d * 50 * , 2282 ( 1994 ) [ arxiv : hep - ph/9311340 ] .
j. edsjo _ et al _ , jcap * 0304 * , 001 ( 2003 ) [ arxiv : hep - ph/0301106 ] . | we present a short review of the status of the dark matter problem .
in particular we show that one of the best motivated candidate for the dark matter is the neutralino , a supersymmetric particle .
finally we study the possibility to detect @xmath0-rays coming from neutralino pair annihilations that take place in our galactic center with the upcoming satellite detector glast .
= 11.6pt |
magnetic reconnection is arguably one of the most important processes in plasma physics , which provides a mechanism to release the energy stored in magnetic field and convert it to thermal energy or bulk plasma kinetic energy .
it is generally believed to be the underlying mechanism that powers explosive events such as solar flares , magnetospheric substorms , and sawtooth crashes in fusion plasmas.@xcite the key challenge of magnetic reconnection theory applied to these events is how energy can be released explosively in time scales that are very short compared with the characteristic resistive diffusion time scale . traditionally , it was widely accepted that magnetic reconnection in the resistive magnetohydrodynamics ( mhd ) model is described by the classical sweet - parker theory.@xcite sweet - parker theory predicts that reconnection rate scales as @xmath10 , where @xmath11 is the lundquist number ( here @xmath3 is the upstream alfvn speed , @xmath2 is the reconnection layer length , and @xmath12 is the resistivity ) . because the lundquist number @xmath13 is usual very high ( e.g. in solar corona @xmath14 , assuming the classical spitzer resistivity ) , the sweet - parker reconnection rate is too slow to account for energy release events . for this reason ,
research on fast reconnection in the past two decades has mostly focused on collisionless reconnection , which can yield reconnection rates as fast as @xmath15 ( here @xmath16 is the upstream magnetic field).@xcite in order to trigger collisionless reconnection , the current sheet width has to first go down to kinetic scales such as the ion skin depth or ion thermal gyroradius.@xcite the super - alfvnic plasmoid instability,@xcite which is a secondary tearing instability acting on a sweet - parker current sheet , has drawn considerable interest in recent years .
while the scaling features of this linear instability are surprising in their own right,@xcite what is more important is that the instability leads to a nonlinear regime in which the reconnection rate becomes nearly independent of @xmath13,@xcite in striking departure from the prediction of sweet - parker theory .
furthermore , the plasmoid instability causes fragmentation of the sweet - parker current sheet to smaller secondary current sheets , which allows collisionless reconnection to be triggered earlier than previously thought possible.@xcite over the past few years , the linear analysis of the plasmoid instability mediated by resistivity has been extended by various authors to include hall,@xcite three dimensional ( 3d),@xcite and shear flow effects.@xcite nonlinear evolution of the instability has also been extensively studied in two dimensional ( 2d ) systems.@xcite the resistive plasmoid instability in 2d is now relatively well understood . however , the resistive tearing mode is not the only mechanism that causes plasmoid formation .
in fact , plasmoid formation has been found to be ubiquitous in large scale reconnection simulations , regardless of the underlying physical models .
@xcite in this work , we explore the consequences of hyper - resistivity on magnetic reconnection mediated by the plasmoid instability . in this model
, the ohm s law assumes the form @xmath17 , where @xmath4 is the hyper - resistivity .
the origin of hyper - resistivity has been attributed to anomalous electron viscosity due to micro - scale field line stochasticity as well as tearing - mode turbulence.@xcite this paper is organized as follows .
the linear theory of the instability is established in sec .
[ sec : linear - theory ] , where the scaling of the linear growth rate is derived .
[ sec : numerical - simulations ] presents the results from nonlinear simulations .
first we verify the scaling of the linear growth rate , then move on to study the effects on magnetic reconnection when the instability has evolved into fully nonlinear regime .
we focus on scalings of reconnection rate , the number of plasmoids , and the sizes of secondary current sheets .
we give a heuristic justification for the scalings .
finally , we examine the statistical distribution of the magnetic flux contained in plasmoids , which is a topic of considerable interest in recent years .
the results from both linear and nonlinear studies are summarized and discussed in sec .
[ sec : summary - and - discussion ] , and comparisons are made with the plasmoid instability in resistive mhd .
the linear analysis of the plasmoid instability in resistive mhd was first carried out by loureiro _ _
et al.__@xcite the analysis shows that the maximum growth rate @xmath18 scales as @xmath19 , and the number of plasmoids scales as @xmath20 .
subsequently , it was shown that these scalings emerge directly from the classic tearing mode dispersion relation,@xcite by taking into account the property that the width of the sweet - parker current sheet @xmath21 scales as @xmath22.@xcite here we follow the latter approach for the hyper - resistive plasmoid instability . to do that we need two ingredients : a generalization of the sweet - parker theory and a linear tearing mode theory with hyper - resistivity in place of resistivity .
the generalization of sweet - parker theory@xcite with hyper - resistivity is straightforward .
let @xmath23 and @xmath24 be the inflow speed and the outflow speed , respectively .
the conditions for conservation of mass and energy remains unchanged , which give @xmath25 @xmath26 here @xmath27 is the plasma density , @xmath16 is the magnetic field , and we have neglected numerical factors of @xmath28 .
the only deviation from the resistive sweet - parker theory comes from the ohm s law , which is now @xmath17 . under quasi - steady condition ,
the out - of - plane electric field is spatially uniform , which gives the following condition @xmath29 where we have made use of the relations @xmath30 and @xmath31 . from eqs .
( [ eq : mass - conservation ] ) ( [ eq : induction ] ) , the following scaling relations are obtained : @xmath32 @xmath33 and @xmath34 where @xmath5 is he hyper - resistive lundquist number defined as @xmath35 the linear dispersion relation of hyper - resistive tearing mode has been derived by aydemir.@xcite here we outline the analysis to keep this paper self - contained . for simplicity , the plasma is assumed to be incompressible with a uniform density . in two - dimensional ( 2d ) cartesian
coordinates @xmath36 , the plasma flow @xmath37 and the magnetic field * @xmath38 * can be expressed in terms of the stream function @xmath39 and the flux function @xmath8 as @xmath40 and @xmath41 , and the system can be described by the well - known reduced mhd equations.@xcite consider an equilibrium magnetic field @xmath42 , with @xmath43 being the corresponding alfvn speed profile . assuming linear perturbations of the form @xmath44 and @xmath45 , where @xmath46 is the growth rate and @xmath47 is the wavenumber along the @xmath48 direction , the linearized reduced mhd equations with
hyper - resistivity are : @xmath49 @xmath50 here primes denote @xmath51 and the operator @xmath52 .
we further assume that @xmath53 is the resonant surface , where @xmath54 .
away from @xmath53 , the effect of hyper - resistivity is negligible .
therefore , in the outer region @xmath55 , which can be used in eq .
( [ eq : phi ] ) to eliminate @xmath56 .
it follows that if we assume @xmath57 , the plasma inertia ( left hand side of eq .
( [ eq : phi ] ) ) is negligible .
hence , in the outer region , the perturbed flux function @xmath58 is governed by @xmath59 the solution of eq .
( [ eq : outer ] ) is subject to appropriate outer boundary condition , e.g. @xmath60 at infinity if the global length scale of the domain along the @xmath61 direction under consideration is much larger than the current sheet width . in general , the solutions of @xmath58 from both regions @xmath62 and @xmath63 will not match smoothly at @xmath53 .
the mismatch in the slope of @xmath58 is characterized by the tearing stability index @xcite @xmath64 which is completely determined by the equilibrium profile and the wave number @xmath47 . because the outer solutions do not match smoothly at @xmath53 , a boundary layer exists around @xmath53 . in the inner region around @xmath53 , a separate set of boundary layer equations @xmath65 @xmath66
are solved .
asymptotic matching of the inner and outer solutions give the linear growth rate .
interested readers are referred to ref .
@xcite for a detailed asymptotic analysis .
it is instructive , however , to employ a simple heuristic argument that gives correct scalings of the linear growth rate , as follows .
let @xmath67 be the width of the equilibrium current sheet and @xmath68 be the width of the boundary layer .
tearing modes are often classified into the so - called constant-@xmath8 and nonconstant-@xmath8 regimes , depending on whether @xmath69 is approximately constant or not within the boundary layer . in the constant-@xmath8 regime ,
the variation of @xmath70 is approximately @xmath71 across the boundary layer .
hence , we may estimate @xmath72 and @xmath73 .
also @xmath74 may be estimated as @xmath75 , and @xmath76 .
balancing terms in eqs .
( [ eq : inner1 ] ) and ( [ eq : inner2 ] ) yields @xmath77 @xmath78 solving eqs .
( [ eq : const - psi1 ] ) and ( [ eq : const - psi2 ] ) yields the scalings of @xmath68 and @xmath46 .
the results are @xmath79 and @xmath80 where @xmath81 is the hyper - resistive lundquist based on the length scale of the equilibrium current sheet width @xmath67 . for the commonly employed harris sheet profile with @xmath82 , the tearing stability index is given by @xmath83 using eq .
( [ eq : deltaprime - harris ] ) in eqs .
( [ eq : nonconst - psi - delta ] ) and ( [ eq : nonconst - psi - gamma ] ) yields @xmath84 and @xmath85 in the constant-@xmath8 regime . more precisely , the @xmath28 numerical factor in eq .
( [ eq : gamma - harris ] ) can be determined by an asymptotic matching calculation,@xcite which gives @xmath86 in the nonconstant-@xmath8 regime , @xmath69 varies significantly within the boundary layer , and we may estimate @xmath87 and @xmath88 . following the same procedure of balancing terms in eqs .
( [ eq : inner1 ] ) and ( [ eq : inner2 ] ) yields @xmath89 and @xmath90 the transition wavenumber from the constant-@xmath8 regime to the nonconstant-@xmath8 regime may be estimated from the self - consistency of the constant-@xmath8 assumption , as follows .
the variation of @xmath69 within the boundary layer may be estimated as @xmath91 .
therefore , @xmath69 being approximately constant requires @xmath92 , i.e. @xmath93 . from eqs .
( [ eq : deltaprime - harris ] ) and ( [ eq : nonconst - psi - delta ] ) , the self - consistency criterion @xmath93 requires @xmath94 assuming @xmath95 . therefore , we expect that the transition occurs when @xmath96
. note that @xmath46 is a monotonically decreasing function of @xmath97 in the constant-@xmath8 regime , and a monotonically increasing function of @xmath97 in the nonconstant-@xmath8 regime . at the transition wavenumber
the growth rates from two branches coincide , which gives the peak growth rate @xmath98 as a function of @xmath97 , for a harris sheet profile with @xmath99 .
markers denote values obtained by numerically solving the eigenvalue problem .
dashed lines are theoretical predictions for the constant-@xmath8 and nonconstant-@xmath8 branches , eqs .
( [ eq : gamma - harris1 ] ) and ( [ eq : nonconst - psi - gamma ] ) .
[ fig : gamma - vs - ka ] ] with @xmath100 , for a harris sheet profile .
markers denote values obtained by numerically solving the eigenvalue problem .
dashed line is the theoretical prediction @xmath101 .
[ fig : gamma - vs - sha ] ] these scalings are verified by numerically solving the eigenvalue problem , eqs .
( [ eq : phi ] ) and ( [ eq : psi ] ) .
figure [ fig : gamma - vs - ka ] shows the dimensionless growth rate @xmath102 as a function of @xmath97 , for a harris sheet profile with @xmath99 .
the numerical values agree with the analytic predictions remarkably well .
likewise , figure [ fig : gamma - vs - sha ] shows the scaling of the maximum dimensionless growth rate @xmath103 with @xmath100 .
the numerical values agree well with the theoretical scaling relation @xmath101 .
now we have all the ingredients for the linear theory of hyper - resistive plasmoid instability . substituting the hyper - resistive sweet - parker current sheet width @xmath104 for the current sheet width @xmath67 in the linear tearing mode theory
, we obtain the following relation between @xmath5 and @xmath100 @xmath105 also the maximum growth rate scales as @xmath106 as such , the instability growth rate increases for higher @xmath5 , similar to the resistive counterpart .
the transition from constant-@xmath8 regime to nonconstant-@xmath8 regime occurs at @xmath107 . unlike the case of the resistive plasmoid instability ,
we are not able to obtain a precise scaling for the number of plasmoids in the hyper - resistive case , for the following reason . in the resistive plasmoid instability
, the scaling of the number of plasmoids can be inferred from the wavenumber of the fastest growing mode , which coincides with the transition wavenumber from the constant-@xmath8 regime to the nonconstant-@xmath8 regime.@xcite that is not the case with hyper - resistivity . for hyper - resistive tearing modes ,
the growth rate is approximately constant within the range @xmath108 ( see , for example , figure [ fig : gamma - vs - ka ] ) , or equivalently within the range @xmath109 , and the notion of fastest growing wavenumber loses its significance .
there are some subtleties in making qualitative comparisons between the resistive and hyper - resistive plasmoid instabilities because the mechanisms that break the frozen - in condition are quite different in the two cases .
a meaningful comparison may be made by rewriting the scaling laws in terms of the aspect ratio @xmath110 of the primary sweet - parker current sheet , which is a common feature for both models .
for the resistive case , the maximum growth rate scales as @xmath111 and the fastest growing mode has a dimensionless wavenumber @xmath112 on the other hand , for the hyper - resistive plasmoid instability , the maximum growth rate scales as @xmath113 and the growth rate peaks when @xmath114 is within the range @xmath115 from these scaling laws we may conclude that the hyper - resistive plasmoid instability has a higher growth rate and prefers shorter wavelengths as compared to the resistive one .
therefore , the hyper - resistive plasmoid instability is even more explosive , and more efficient in generating copious plasmoids , than the resistive plasmoid instability .
the resistive plasmoid instability is of great interest because it leads to a nonlinear regime where magnetic reconnection is drastically different from the sweet - parker model .
a pivotal question is how the hyper - resistive plasmoid instability affects reconnection in the nonlinear regime . to address this question ,
we employ the same simulation setup of two coalescing magnetic islands as in ref . @xcite .
the governing equations are identical to the ones before , except that resistivity is now replaced by hyper - resistivity .
an isothermal equation of state is assumed for simplicity . in normalized units ,
the simulation box is a square in the domain @xmath116\times[-1/2,1/2]$ ] .
the initial magnetic field @xmath117 .
the parameter @xmath118 , which is set to @xmath119 for all simulations , determines the initial current layer width .
the initial plasma density @xmath27 is approximately @xmath120 , and the plasma temperature @xmath121 is @xmath122 .
the density profile has a weak nonuniformity such that the initial condition is approximately force - balanced .
the initial peak magnetic field and alfvn speed are both approximately unity .
therefore , the hyper - resistive lundquist number @xmath123 is simply @xmath124 .
the plasma beta @xmath125 is greater than @xmath126 everywhere . perfectly conducting and free slipping boundary conditions are imposed along both @xmath48 and @xmath61 directions .
specifically , the highest resolution is 16000 grid points along the @xmath48 direction and 1000 grid points along the @xmath61 direction , with the smallest grid size @xmath127 .
figure [ fig : initial_condition ] shows the initial current density distribution , overlaid with magnetic field lines . as the simulation proceeds , the current layer first thins down and forms the primary hyper - resistive sweet - parker layer .
subsequently , the primary current layer may become unstable to the plasmoid instability if @xmath5 is above a threshold @xmath128 .
we show snapshots of a simulation with @xmath129 in fig .
[ fig : snapshots ] . ]
dashed lines represent separatrices separating the two primary merging islands that drive the reconnection .
the top panel shows the instability in the early stage , and the bottom panel shows the fully developed nonlinear stage .
[ fig : snapshots ] ] before we study magnetic reconnection in fully nonlinear regime for this system , we first verify the prediction @xmath130 from the linear analysis .
one difficulty in measuring the linear growth rate is due to the fact that we do not start the simulation with a sweet - parker current sheet .
rather , the sweet - parker current sheet is established self - consistently during the current sheet thinning phase .
therefore , the standard technique of adding a small perturbation to an initial equilibrium and measuring the growth rate as the perturbation grows does not apply here .
the problem is that most physical quantities evolve quite substantially during the current sheet thinning phase before onset of the plasmoid instability , and it is difficult to filter out the variations that are not due to the growth of the plasmoid instability .
this difficulty is overcome by looking at the component @xmath131 along the central part of the reconnection layer @xmath53 , where * @xmath131 * is identically zero initially and remains small before onset of the plasmoid instability . after the onset , the component @xmath131 develops fluctuations which rapidly grow as time proceeds . to obtain the linear growth rate ,
we integrate @xmath132 at the central part of the current sheet along @xmath53 , from @xmath133 to @xmath134 at each time .
the magnitude of @xmath135 remains small before onset of the plasmoid instability , and increases abruptly after the onset .
the linear growth rate @xmath46 can be obtained by fitting @xmath136 to a linear function @xmath137 during the period the plasmoid instability exhibits approximately linear growth .
this procedure is illustrated in figure [ fig : growthrate - measurement ] for the case @xmath129 .
we measure the linear growth rates for cases with @xmath5 ranging from @xmath138 to @xmath139 .
figure [ fig : scaing - of - growthrate ] shows the scaling of @xmath46 with respect to @xmath5 .
the results are in good agreement with the prediction @xmath130 .
.[fig : growthrate - measurement ] ] .
the dashed line is the prediction from the linear theory .
[ fig : scaing - of - growthrate ] ] to @xmath5 . ] $ ] with respect to @xmath5.[fig : scaling - no - plasmoids ] ] and width @xmath68 of secondary current sheets with respect to @xmath5.[fig : scaling - length ] ] with respect to @xmath5.[fig : jy - scaling ] ] in and ( b ) probability distribution function @xmath140 from a @xmath129 simulation .
the vertical dotted line denotes where @xmath141 , indicating where the dominant loss mechanism switches from coalescence to advection.[fig : pdf ] ] seeking statistical descriptions of plasmoids has been a topic of considerable interest in recent years , @xcite partly due to the possible link between plasmoids and energetic particles.@xcite in our recent work with resistive mhd , it was found numerically that the distribution function @xmath140 of magnetic flux @xmath8 inside plasmoids exhibits a @xmath142 power - law distribution over an extended range , followed by an exponential tail for large plasmoids .
a theoretical model was proposed that yields results consistent with the numerical simulations.@xcite it has been clarified that the transition from the power - law distribution to the exponential tail is due to the dominant plasmoid loss mechanism switching from coalescence to advection .
this transition typically occurs when the cumulative distribution function @xmath143 obeys the approximately inequality @xmath144 , i.e. for the very largest plasmoids in each snapshot .
because the theoretical model only relies on the key assumption that secondary current sheets between plasmoids are close to marginal stability and a few general assumptions regarding coalescence and advection of plasmoids , the model can be readily adapted to the case of hyper - resistive plasmoid instability .
therefore , we expect plasmoids in hyper - resistive mhd model to follow a similar distribution . that indeed appears to be the case . figure [ fig : pdf ] shows the cumulative distribution function @xmath145 and the distribution function @xmath140 from a @xmath129 simulation .
the data set comprises 13486 plasmoids collected from 352 snapshots during the period of reconnecting 25% of the initial magnetic flux .
the distribution function @xmath140 clearly exhibits an extended @xmath142 power - law regime in the range between @xmath146 and @xmath147 .
above @xmath147 the distribution makes a transition to a more rapid falloff . and
this transition approximately coincides the vertical dotted line , which denotes where @xmath141 , indicating a switch of the dominant loss mechanism from coalescence to advection .
these features are qualitatively similar to the ones with resistive plasmoid instability .
.comparison between resistive and hyper - resistive plasmoid instabilities .
the scaling laws are expressed in terms of the aspect ratio @xmath148 . here
@xmath18 is the peak linear growth rate ; @xmath149 is the fastest growing wave number ; @xmath150 is the number of plasmoids in nonlinear regime ; @xmath68 and @xmath151 are the thickness and length of secondary current sheets ; and @xmath152 is the current density .
[ tab : comparison - of - scaling ] [ cols="^,^,^",options="header " , ] in summary , we have carried out a linear instability analysis and nonlinear simulations of the plasmoid instability when hyper - resistivity is the mechanism of breaking field lines .
we have found that the hyper - resistive plasmoid instability is qualitatively similar to the resistive plasmoid instability , although they follow different scaling laws both linearly and nonlinearly . in the plasmoid - unstable regime , the reconnection rate
is found to be nearly independent of the hyper - resistive lundquist number @xmath5 instead of following the predicted @xmath153 scaling obtained by assuming a stable , extended current layer .
the reconnection rate in high-@xmath5 regime is approximately @xmath154 , which is similar to the value obtained with the resistive plasmoid instability .
the scaling laws in the nonlinear regime can be heuristically derived by assuming secondary current sheets between plasmoids are close to marginally stable , even though that assumption is clearly oversimplified .
the distribution of plasmoid magnetic flux is found to obey a @xmath142 power law over an extended range , followed by a rapid falloff for large plasmoids similar to the result obtained for the resistive plasmoid instability .
table [ tab : comparison - of - scaling ] summarizes the comparison between resistive and hyper - resistive plasmoid instabilities . here
the scaling laws are expressed in terms of the aspect ratio @xmath148 , which is a common feature of both models .
the aspect ratio scales with respect to the resistive and hyper - resistive lundquist numbers as @xmath155 and @xmath156 , respectively . from these scaling relations
, we can see that for the same aspect ratio @xmath157 , the hyper - resistive plasmoid instability has a higher peak linear growth rate , and shorter wavelengths .
therefore , the hyper - resistive plasmoid instability will set in more rapidly , with more plasmoids at the early stage , compared to the resistive case .
however , after the plasmoid instability has developed into fully nonlinear regime , more plasmoids will be present in the resistive case .
the reason is that resistivity is less effective in smoothing out small - scale structure , which allows current sheet fragmentation to cascade down to deeper level .
the results in this paper may be relevant to the solar atmosphere , where hyper - resistivity has been proposed as a possible mechanism for corona heating.@xcite in addition , some recent studies have found that current sheets formed after coronal mass ejection ( cme ) events have thicknesses far broader than classical or anomalous resistivity would predict , and it was suggested that hyper - resistivity may be the cause.@xcite recently , a comparison of plasmoid distributions in post - cme current sheets obtained from both solar observation and resistive mhd simulation has been made.@xcite however , because plasmoid distributions obtained from both resistive mhd and hyper - resistive mhd models are essentially identical , statistical study of plasmoid distribution alone will not be able to distinguish the two models . this conclusion calls for other measures that can better tell apart different models .
the various scaling relations obtained in this paper may be able to provide other insights on how this can be done .
60ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo
secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1146/annurev - astro-082708 - 101726 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.82.603 [ * * , ( ) ] link:\doibase 10.1007/bf02962520 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1029/2001ja009180 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1063/1.3264103 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.105004 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.3420208 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.3703318 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1063/1.3606363 [ * * , ( ) ] link:\doibase 10.1063/1.3633473 [ * * , ( ) ] link:\doibase 10.1063/1.3678211 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.3086867 [ * * , ( ) ] link:\doibase 10.1088/0004 - 637x/737/1/24 [ * * , ( ) ] link:\doibase 10.1088/0004 - 637x/737/1/14 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1111/j.1365 - 2966.2012.21625.x [ * * , ( ) ] link:\doibase 10.1063/1.4752744 [ * * , ( ) ] link:\doibase 10.1029/2006gl025957 [ * * , ( ) ] @noop * * , ( ) in @noop _ _ , , vol . ,
( ) pp .
link:\doibase 10.1038/nphys1965 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.1694467 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1063/1.865798 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1038/nature10091 [ * * , ( ) ] @noop _ _ ( , ) @noop * * , ( ) link:\doibase 10.1063/1.4802941 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1029/2010ja016271 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.265002 [ * * , ( ) ] , link:\doibase 10.1038/nature05116 [ * * , ( ) ] link:\doibase 10.1038/nphys777 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1088/2041 - 8205/771/1/l14 [ * * , ( ) ] | magnetic reconnection mediated by the hyper - resistive plasmoid instability is studied with both linear analysis and nonlinear simulations .
the linear growth rate is found to scale as @xmath0 with respect to the hyper - resistive lundquist number @xmath1 , where @xmath2 is the system size , @xmath3 is the alfvn velocity , and @xmath4 is the hyper - resistivity . in the nonlinear regime ,
reconnection rate becomes nearly independent of @xmath5 , the number of plasmoids scales as @xmath6 , and the secondary current sheet length and width both scale as @xmath7 .
these scalings are consistent with a heuristic argument assuming secondary current sheets are close to marginal stability .
the distribution of plasmoids as a function of the enclosed flux @xmath8 is found to obey a @xmath9 power law over an extended range , followed by a rapid fall off for large plasmoids .
these results are compared with those from resistive magnetohydrodynamic studies . |
gallager s capacity theorem ( * ? ? ? * theorem 8.5.1 ) is literally reproduced on p. 1 of the poster along with the figure ( * ? ? ?
* figure 8.5.1 ) plus some explanations taken from gallager s book @xcite . on p. 2 , the so - called gallager channel ( now called gaussian waveform channel ) , i.e. , a gaussian filter with additive white gaussian noise ( awgn ) , is treated as a special instance of gallager s theorem .
on p. 3 , the heat channel , a linear time - varying ( ltv ) filter with awgn , is depicted followed by a characterization of its capacity by water - filling in the time - frequency plane ( * ? ? ?
* theorem 2 ) . finally , on p. 4 , an example of a signal transmission is given aiming at underpinning the apparent contradiction to gallager s theorem in figs . 3 and 4 of the poster .
the curves for the heat channel in the latter two figures are incorrect ; the correct curves are provided in the present publication
. actually tends to infinity ( see the proof of the theorem in @xcite ) ]
evaluation of the double integrals occurring in the water - filling theorem ( * ? ? ? * theorem 2 ) ( cf .
p. 3 of @xcite ) and subsequent elimination of the parameter @xmath0 results in the closed - form representation of the capacity ( in nats per transmission ) of the heat channel , namely @xmath1 ^ 2+o(\alpha\beta)\,(\alpha\beta\rightarrow\infty ) , \label{eq1}\ ] ] where @xmath2 is the inverse function of @xmath3 , and @xmath4 is the standard landau little - o symbol .
( [ eq1 ] ) coincides with the capacity formula ( * ? ? ?
* eq . ( 9 ) ) ( as it should ) .
by reason of the gaussian time window ( see fig .
[ figure_1 ] ) at the input and output side of the ltv filter ( see ( * ? ? ?
* fig . 5 ) and ( * ? ? ?
( 4 ) ) , resp . ) , capacity achieving input / output signals for the heat channel will have the approximate duration @xmath5 ( at least when the input energy is sufficiently high [ but not too high ! ] ) . therefore ,
transition to capacity as a rate should be done as follows . put for the average input energy @xmath6 , where @xmath7 is average power .
then , form the time average @xmath8 . because of eq .
( [ eq1 ] ) it asssumes the value @xmath9 ^ 2\log_2{\mathrm{e}}\;\;\mbox{(bit / s ) } , \label{eq2}\ ] ] where @xmath10 , @xmath11 is the approximate bandwidth in positive frequencies measured in hertz ( cf .
2 ) ) , and @xmath12 is the one - sided noise power spectral density of the awgn .
similarly , for @xmath13 as a function of @xmath14 we obtain the parametric representation @xmath15 ^ 2 \log_2{\mathrm{e}}}\label{eq3a}\\ \frac{\bar{c}(\mathrm{snr})}{w}&=&\frac{1}{2\pi}[w_0(4\pi\mathrm{snr})]^2 \log_2{\mathrm{e}}\;\;(\mbox{bit / s / hz})\label{eq3b}.\end{aligned}\ ] ] in fig .
[ figure_2 ] , @xmath13 is plotted against snr ; here , eq .
( [ eq2 ] ) is used . in fig .
[ figure_3 ] , the curve for @xmath13 as a function of @xmath14 is given parametrically by eqs .
( [ eq3a ] ) , ( [ eq3b ] ) . in both figures , in case of the heat channel the label @xmath17 stands for @xmath13 .
figs . 3 and 4 of the poster @xcite should be replaced by figs . 2 and 3 , resp . , of the present publication .
moreover , the caption in ( * ? ? ?
7 ) is to be replaced by
example : @xmath18 , @xmath19 , @xmath20 bits per transmission ( exact value ) .
" the exact value , by the way , has been computed numerically
. 1 r. g. gallager , _ information theory and reliable communication_. new york , ny : wiley , 1968 .
e. hammerich , see version 1 of this publication ( arxiv:1207.4707v1 ) .
e. hammerich , `` on the heat channel and its capacity , '' _ proc .
information theory _ ,
seoul , korea , 2009 , pp .
e. hammerich , `` on the heat channel and its capacity , '' available online at arxiv:1101.0287v3 . | we correct an alleged contradiction to gallager s capacity theorem for waveform channels as presented in a poster at the 2012 ieee international symposium on information theory . |
after decades of intensive research the exact mechanism of core - collapse supernovae ( sne ) still remains a mystery .
it is widely believed that the explosion is driven by neutrinos emitted by the proto - neutron star ( pns ) formed as the result of the collapse @xcite .
however , the attempts to reproduce core - collapse sne in computer simulations have encountered severe problems .
the reason for the failures becomes more or less clear when one compares the total energy radiated in the form of neutrinos , @xmath7erg , with the typical energy of supernovae , @xmath8erg .
this tells us that in order to obtain a reliable answer the computational error in the energy budget has to be below @xmath9 , which a is very demanding condition .
the neutrino transport , of six different species , has to be treated with great accuracy including the processes effecting heating and cooling rates .
the alternative magnetic mechanism of core - collapse sne has been around for a while , first proposed by bisnovatyi - kogan and leblanc & wilson , whose numerical simulations were miles ahead of their time . for three decades
this mechanism was not taken seriously and only occasionally efforts were made to develop it further @xcite .
now it is experiencing `` renaissance '' @xcite , thanks to the slow and difficult progress of the neutrino models of sne , the development of robust numerical methods for mhd , the accumulated evidence for asphericity of supernovae @xcite , the high collimation of flows associated with grbs ( e.g. piran 2005a , b ) and the increasing popularity of the magnetic mechanism for the origin of other astrophysical jets .
the results show that magnetic field can facilitate explosions of standard power and even more powerful explosions provided the magnetic field is sufficiently strong .
the energy of high - velocity supernovae or hypernovae ( hne ) exceeds that of normal core - collapse supernovae by about ten times ( e.g. nomoto et al . 2004 ) .
the few sne that have been reliably identified with grbs are all hne .
although , the statistics is very poor and is subject to observational bias towards most powerful and hence brightest events , the connection between hne and grbs is widely accepted .
the physical mechanism of hne is not established and there is no shortage of competing theoretical models .
the most popular one , the collapsar model , relates hne with collapse of massive rapidly rotating stars @xcite . in this model , the stellar core collapses straight into a black hole ( which used to be considered as an indication of `` failed '' supernova explosion ) but the stellar envelope does not and forms a hyper - accreting disk around the hole .
the energy released in the disk can be very large , sufficient to drive hn and grb , but the way it is transferred to the supernova ejecta and the grb jets remains to be determined .
it is widely assumed that grb jets are powered by the energy released in the funnel region of the disk via annihilation of neutrinos @xcite .
however , taking into account opacity in the inner regions of the disk significantly reduces the efficiency of this mechanism @xcite . alternatively , as it has been suggested by many authors , the grb - jets in the collapsar scenarion can be powered via magnetic mechanism tapping the rotational energy of the black hole and/or the accretion disk ( e.g. lyutikov & blandford 2003 , vlahakis & konigl 2003 , proga et al . , 2003 ) .
the energy of hne ejecta is of the same magnitude as the rotational energy of a neutron star ( ns ) with very short period of rotation , 1 - 2 milliseconds .
this energy can be released via magnetic braking within a very short period of time , from few seconds to few hundreds of a second , provided the magnetic field is very strong , of order @xmath10 g @xcite .
this value is three orders of magnitude above the typical strength of magnetic field of normal neutron stars and would be considered a huge over - stretch not so long time ago .
however , the identification of soft gamma ray repeaters and x - ray anomalous pulsars with magnetars , neutron stars with magnetic field @xmath11 g @xcite , shows that the magnetar model may be a viable alternative to the collapsar model of hne as well as grbs @xcite .
the origin of magnetar s magnetic field is still debated .
one possibility is the generation of unusually strong magnetic field in convective cores of some pre - supernova stars . during the core collapse the poloidal field
is amplified like @xmath12 and in order to reach the required @xmath13 g after the collapse the original field in the core , on the scale of @xmath14 km , should be of order @xmath15 g .
the collapse of magnetized rotating stellar cores has been studied by many groups .
the general conclusion from the latest parameter studies is that only for @xmath16 g the magnetic field is strong enough to influence the dynamics of supernovae explosions @xcite .
the key feature of these explosions is strong differential rotation and generation of toroidal magnetic field whose pressure actually drives the explosion .
the hoop stress of this field also ensures anisotropy of the explosion - it is more powerful along the rotational axis . on the other hand ,
the super - strong magnetic field of magnetars could be generated via @xmath17-@xmath18-dynamo in the convective pns @xcite . in this theory
the strength of saturated field strongly depends on the rotational period - shorter period leading to stronger field . in order to generate the dipolar field of @xmath19 g the period of pns
should be very short , not much higher than few milliseconds @xcite .
thus , this theory unites both basic conditions required to produce a hypernova in the magnetar model , rapid rotation and super - strong magnetic field , in one . on the downside ,
the pns is no longer magnetically coupled to the envelope and the transfer of its rotational energy to the stellar envelope is more problematic .
turbulence required for the magnetic dynamo action can also be generated via the magneto - rotational instability ( mri , balbus & hawley,1991 ) .
calculations based on the linear theory show that in the supernova context strong saturation field can be reached very quickly on the time scale of only several tens of rotational periods @xcite .
thompson et al . proposed a model of magnetically driven hn explosions and grb where the usual delayed neutrino - driven mechanism plays the role of bomb detonator .
the neutrino - driven blast creates a rarefaction around the newly - born magnetar , its magnetosphere expands and eventually develops a magnetically driven pulsar wind ( pw ) .
this wind acts as a piston on the expanding supernova envelope , that is the layer of compressed stellar material behind the supernova shock , and energizes it beyond the level of normal sn .
metzger et al . generalized the 1d model of equatorial magnetically - driven wind by weber & davis to fit the conditions of pns .
axisymmetric magnetized winds of magnetars with various degree of mass loading have been studied in bucciantini et al . via 2d numerical simulations .
they concluded that during the initial phase ( kelvin - helmholtz timescale ) when the wind is already magnetically - driven but still non - relativistic the rotational energy loss rate is @xmath20erg / s , where @xmath21 is the magnetic field of the pns in @xmath22 g and @xmath23 is its period in milliseconds .
bucciantini et al .
applied the model of pulsar wind nebula ( pwn ) by begelman & li to study the late phases of hypernova explosion , @xmath24s .
they used the thin - shell approximation to describe the dynamics of the supernova shock driven by the anisotropic pressure of the magnetized bubble , `` baby pulsar wind nebula '' , created inside the exploding star by the magnetar wind .
they concluded that when the magnetic energy in the bubble exceeds @xmath25 of the thermal energy the shell expansion proceeds in a highly anisotropic fashion and eventually an axial channel is created inside the stellar envelope .
they proposed that later this channel collimates the ultra - relativistic wind from cooled magnetars into a pair of grb jets . in the follow up paper @xcite they studied the interaction of the super - fast magnetar wind with the supernova envelope via fully relativistic mhd simulations and arrived to similar conclusions . for the wind model to become applicable
the size of the `` central cavity '' must exceed , and likely many times , that of the fast surface as found in the unconfined case .
prior to this one would expect a somewhat different dynamics .
the most pessimistic possibility is a rigidly rotating magnetosphere in which case no extraction of rotational energy takes place .
however , the theory of relativity places a upper limit on the size of such a magnetosphere .
indeed , the magnetosphere whose size exceeds the light cylinder ( lc ) radius can not be rigidly rotating as this would imply superluminal motion of magnetospheric plasma and should become differentially rotating .
this limit is particularly relevant for the magnetospheres of millisecond pulsars as the radius of lc , @xmath26 km , is only few times the radius of the star itself .
once the differential rotation sets up it amplifies the toroidal field , the magnetic pressure grows and promotes further expansion of the magnetosphere .
uzdensky & macfadyen suggested that this factor along can explain supernova explosions but they did not elaborate ways to achieve the initial expansion beyond lc .
moreover , they assumed low mass loading of magnetic field lines , more suitable for ns than hot proto - ns , and concluded that the magnetospheric plasma will exhibit ultra - relativistic rotation in the equatorial plane . in any case , this argument suggests that the magnetic braking should begin to operate at the very early stages of the `` detonated '' magnetically - driven explosions , well before the central cavity expands beyond the fast surface of unconfined magnetar wind and this may to have important implications for their later development . in this paper
we describe an attempt to explore the early phase ( @xmath27s ) of such explosions via axisymmetric general relativistic mhd simulations .
our main goals are to study the basic dynamics of this phase , to determine the spin - down power , and to assess the implications for the later evolution .
we use an upwind conservative scheme that based on a linear riemann solver and uses the constrained transport method to evolve the magnetic field .
the details of this numerical method and various tests are described in komissarov . in sec.2
we describe the details of numerical experiments , including initial and boundary conditions , sec.3 describes the results of simulations , and in sec.4 we discuss their implications for the model of magnetar - driven supernovae explosions and grb jets .
our conclusions are listed in sec.5 .
the ultimate hypernovae / supernovae simulations should trace all phases the stellar explosions , the onset of core collapse , core bounce , delayed explosion , generation of magnetic field , formation of magnetic cavity , and break out of magnetically - driven jets .
unfortunately this is still beyond our current capabilities and we have to make some simplifications .
here we start our simulations from the point when strong magnetic field has just appeared from surface of already formed pns . to account for the previous stages of the explosion
we assume that pns is surrounded by a `` radiation bubble '' with a large amount of energy deposited by neutrino in form of heat .
the bubble is surrounded by the collapsing stellar envelope .
the gravitational attraction of pns is introduced via schwarzschild metric in boyer - lindquist coordinates , @xmath28 .
the two - dimensional computational domain is @xmath29 , where @xmath30 corresponds to the radius of pns and @xmath31 km .
the total mass within the domain is small compared to the mass of pns that allows us to ignore its self - gravity .
the grid is uniform in @xmath32 where it has 200 cells and almost uniform in @xmath33 where it has 430 cells , the linear cell size being the same in both directions .
the evolution of a non - rotating proto - neutron star of mass @xmath34 have been studied in details by burrows & lattimer . according to this study the deleptonization time - scale is rather long , it takes around 5 seconds before half the leptons inside @xmath35 are lost .
however , the contraction time - scale is much shorter - it takes only @xmath36ms for the star radius to decrease down to @xmath37 km ( see fig.5 in burrows & lattimer , 1986 ) .
using these results we fix the mass and radius of pns to @xmath34 and @xmath38 km respectively , and we choose the rotation period to be @xmath39ms . to set the boundary conditions on the pns surface we utilize the results of high resolution 1d numerical models of pns winds @xcite .
the gas temperature of `` ghost cells '' is to @xmath40 mev , which is typical for a hot newly - born pns , and their density is @xmath41 .
this density corresponds to the interface between the very thin exponential atmosphere of pns and its wind .
the initial magnetic field is purely poloidal with the azimuthal component of vector potential @xmath42 where @xmath43 @xmath44 g and @xmath45 . for @xmath46 equation [ pot ]
describes magnetic dipole of strength @xmath47 at @xmath48 , @xmath49 . the masking function ( [ mask ] ) `` moves '' the magnetic field lines inside the sphere of radius @xmath50 without changing the magnetic flux distribution over the pns surface .
this distribution is preserved for the whole duration of simulations .
the radial velocity of ghost cells is set to zero and the mass outflow through the boundary is found via solving the riemann problem at the boundary interface .
we assume that neutrino luminosity in each species is given by the black body formula @xmath51 and that the mean neutrino energy is @xmath52 mev .
this oversimplification is unlikely to have a strong effect on the largely magnetically driven outflows from pns .
here we adopt the simple free - fall model by bethe . according to this model
the radial velocity is @xmath53 and the density is @xmath54 where @xmath55 is a coefficient between 1 and 10 @xcite and @xmath56 is the time since the onset of collapse .
the corresponding accretion rate and ram pressure are @xmath57 @xmath58 respectively .
since in the simulations we fix the core mass to @xmath34 the model is fully determined by the combination @xmath59 . for a core of radius @xmath60 cm and mass 1.4@xmath61 the collapse duration can be estimated as @xmath62s .
the time - scale of the delayed explosion is likely to be around few tens of a second after bounce @xcite .
thus , the value of @xmath56 is relatively well constrained and we fixed it to be @xmath63s . since grbs are currently associated with more massive progenitors we consider @xmath64 . to account for rotation of the pre - supernova star
the accreting mass is attributed with specific angular @xmath65 where @xmath66 .
this is slightly higher then the equatorial value of @xmath67 on the pns surface , @xmath68 but still low enough to prevent the development of accretion disk around the magnetar .
in fact , the rotation does not seem to have a strong effect on the solution .
eqs.([b1],[b2],[b5 ] ) are also used to set the flow variables in the ghost cells of the outer boundary .
the radiation bubble is assumed to extend up to @xmath69 km or @xmath70 km in different models .
the radial and polar velocity components of matter in the bubble are set to zero and the azimuthal component is calculated using eq.([b5 ] ) .
the prescribed density and pressure distributions are @xmath71 @xmath72 + p_b . \label{c2}\ ] ] for @xmath73 these distributions correspond to a hydrostatic equilibrium . in the simulations we use @xmath74 and vary @xmath75 to get the desired amount of thermal energy stored in the bubble , @xmath76erg .
the parameters of four models described in this paper are summarized in table 1 .
.parameters of the models used for test simulations . [ cols="^,<,<,<,<",options="header " , ] we use realistic equation of state ( eos ) that takes into account the contributions from radiation , lepton gas including pair plasma , and non - degenerate nuclei - we have incorporated the eos code helm kindly provided by frank timmes for free download ( http://www.cococubed.com/code_pages/eos.shtml ) . in this code
the contribution from the electron - positron plasma is computed via table interpolation of the helmholtz free energy @xcite .
the neutrino transport is treated in the optically thin regime .
the neutrino cooling is computed using the interpolation formulas given in ivanova et al.@xcite , for the urca - process , and in schinder at al .
, for pair annihilation , photo - production , and plasma emission .
the neutrino heating rates are computed using the prescriptions of thompson et al . which take into account gravitational redshift and geodesic bending .
we ignore the doppler effect due to plasma motion as its speed relative to the grid never becomes highly relativistic ( see sec.[results ] ) .
photo - disintegration of nuclei is included via modifying eos ( e.g. ardeljan et al.2005 ) .
the equation for mass fraction of free nucleons is adopted from woosley & baron .
figure [ f1 ] shows the global structure of model a at time @xmath77 ms when the simulations were terminated because the shock wave has reached the outer boundary of the computational domain .
the low density `` column '' along the symmetry axis reveals the volume occupied by the collimated outflows from the magnetar .
one can see that these jets have already `` drilled '' holes in the supernova envelope , that is the layer of stellar material compressed by the supernova shock , and are beginning to propagate directly through the collapsing star .
the unit length in this and other figures is @xmath78 km .
thus , the propagation speed of the jets is about @xmath79 or @xmath80 . at this speed
the jets would travel across the star of radius @xmath81 km in about 4 s. the right panel of fig.[f1 ] shows that the flow speed of the jets can be as high as @xmath82 .
the plots show no signs of the jet termination shock - this is because the jet flow is sub - fast and can decelerate smoothly when it reaches the jet head .
the knotty structure of jet is not due to internal conical shocks or the pinch instability but due to the non - stationary nature of the central engine .
the dead zone of magnetar s magnetosphere opens up in a quasi - periodic fashion with a characteristic time of 20ms .
this causes significant restructuring of the open magnetosphere and variability of the outflow .
this is also the origin of waves clearly visible in the velocity field within the circular region of 3000 km radius ( right panel of fig.[f1 ] ) .
one can also see weaker outflow along the equatorial current sheet of magnetars magnetosphere .
the colour image shows @xmath83 and the contours show the magnetic field lines .
the unit length in all figures in this paper is @xmath84 km .
, width=321 ] figure [ f2 ] shows the global structure of model b at the same time .
the most apparent difference is in the size of the quasi - spherical supernova shock .
it propagates noticeable slower because of the lower energy deposited in the radiation bubble .
however , the propagation speed of the jets is almost the same as in model a suggesting that their power must be similar as well .
one can also see that inside the supernova envelope the jets propagate inside a relatively thin channel and outside of the envelope they propagate through a low density cocoon inflated during the previous evolution .
a similar cocoon is beginning to form in model a ( see fig.[f1 ] ) suggesting that at later times models a and b could look even more alike . figure [ f2a ] shows the global structure of model c at time @xmath85ms .
since the ram pressure of the infalling material is higher than in models a and b the jet propagation speed is somewhat lower .
the same factor explains lower propagation speed of the supernova shock in this model compared to model a. in other respects the solution is not much different .
the evolution of model d starts in very much the same fashion as for other models but at later times we observe a substantial fall - back of the shocked stellar material on the magnetar .
the neutrino - cooled accreting matter eventually blocks the jet channel and squashes the magnetosphere against the star surface ( see fig.[f2b ] ) . soon after the code crashes as the conditions near the surface become too extreme .
figure [ f3 ] shows the total flux of `` free energy '' , the total energy at infinity minus the rest mass energy , as a function of radius for models a , b , and c. the `` spikes '' at @xmath86 for model b , between @xmath87 and @xmath88 for model a , and between @xmath87 and @xmath89 for model c correspond to the blast wave of normal supernova explosion .
the `` plateaus '' to the left and to the right of the spikes correspond to the jets .
thus , in all models the spin - down power released in the jets is indeed more or less the same , @xmath90erg / s .
the corresponding spin - down time is @xmath91s suggesting that most of the rotational energy will be released after the jets break out from the progenitor star .
while this is a promising result for the magnetar model of grbs it also suggests that only a modest fraction of the total rotational energy of the magnetar can be transferred to the supernova ejecta . in all three models
the ratio of poynting flux to the rest mass energy flux is @xmath92 .
this is approximately 5 times higher than the value predicted by the theory of unconfined equatorial magnetar wind ( see eq.(27 ) in metzger et al . ) .
figure [ f4 ] shows model a at the early times , @xmath93 ms .
one can see that the magnetar is already producing an outflow which has blown a cavity in the shape of a column inside the collapsing star .
the radius of this column is close to the radius of magnetar s light cylinder , @xmath94 km , which is shown in the right panel of fig.[f4 ] by the dashed line .
we note , however , that for @xmath95 the column is still inside the light cylinder .
the total pressure of the outflow is dominated by the magnetic field , the ratio of magnetic to gas pressure varying between 10 and 300 .
one can see strong axial compression of the outflow due to the hoop stress of azimuthal magnetic field .
the distribution of total pressure over the boundary of the volume occupied by the outflow has a clear maximum at the point of intersection with the polar axis which is obviously related to the axial compression . this distribution explains the faster expansion of the cavity blown by the outflow in the axial direction . in the middle panel of fig.[f4 ] one can see the bow shock , at @xmath96 , driven by the ouflow into the surrounding .
there are no indications of the termination shock of the outflow itself - this is because the outflow is sub - fast .
the solid line in the right panel of fig.[f4 ] shows the locus of points where the radial component of the phase velocity of ingoing alfvn waves becomes positive . inside the column
these point form a closed surface which we call the alfvn horizon of the magnetar - driven outflow .
any alfvn wave can propagate from the exterior of this surface into its interior .
one can see that the alfvn horizon is always located inside the light cylinder , which agrees with theory of steady state mhd flows @xcite .
figure [ f5 ] shows the inner region of model a by the end of the run ( @xmath77 ms ) . by this time
the size of the central cavity , that develops a diamond - like shape , is significantly larger than @xmath97 , thanks to the decreased pressure in the surrounding bubble that resulted from its expansion in the course of supernova evolution .
however , the rotational velocity in the equatorial plane is not ultra - relativistic indicating that the relativistic winding mechanism is not in operation . the velocity field ( left and right panels of fig.[f5 ] ) shows that close to the magnetar the outflow is similar to an isotropic wind but then it becomes progressively collimated in the polar direction and eventually passes through the `` bottle - neck '' nozzles at the axial corners of the diamond .
the collimation is smooth with no indication of shocks waves which is expected because the flow is always sub - fast . after passing trough the nozzle
each jet enters the channel made during the previous evolution and propagates inside a low density cocoon . in its outer layer
the cocoon contains poloidal field lines of opposite direction to that of the jet and closer to the jet one can see magnetic islands and evidence of vortex motion .
these islands are mostly remnants of plasmons injected from the dead zone of magnetar s magnetosphere ( see the right panel of fig.[f6 ] ) during its oscillatory cycle . in the middle panel of fig.[f5 ] one can see one of such recently ejected plasmons .
close inspection of animated pictures suggests the following origin of the oscillations .
when the dead zone includes the highest magnetic flux it gradually accumulates plasma supplied from the magnetar surface . this leads to increase of the centrifugal force and expanding of the dead zone .
its magnetic field lines stretch in the equatorial direction trying to open up .
the process seems to accelerate as the dead zone plasma moves further out from the star and finally escapes in eruptive manner leaving behind thin equatorial current sheet .
then the oppositely directed field lines of the sheet reconnect and this restores the initial configuration of the dead zone ( a similar behavior of pns magnetosphere was reported in bucciantini et al . , 2006 ) .
the amplitude of these oscillations is large so one gets the impression of the magnetosphere `` breathing heavily '' .
since here we integrate the equations of _ ideal _ rmhd it is the numerical resistivity that is responsible for the observed reconnection and the reliability of its time - scale is unclear .
the plots of fig.[f5 ] also reveal a rather peculiar time - dependent structure inside the diamond cavity and near to the polar axis , to which we will refer as the `` trap zone '' .
it begins at @xmath98 and continues almost up to the top corner of the cavity . in the right panel of fig.[f5 ] , that shows the ration of magnetic to gas pressure ( the inverse of the traditional magnetization parameter @xmath99 ) , it looks as the region of lower magnetization .
the density and velocity plots of fig.[f5 ] show that the trap zone has a relatively high density and low speed .
in fact , the radial velocity in the trap zone changes sign both in space and time - the density clumps seen in the trap zone exhibit oscillatory motion along the polar axis .
this explains the peculiar turns of magnetic field lines in the trap zone ( see the middle panel of fig.[f5 ] ) . from time to time
plasma clouds are ejected from this zone and the time - scale of this ejections is similar to the time - scale of global magnetospheric oscillations .
one of such clouds is seen in density plot of fig.[f5 ] at @xmath100 .
another peculiarity of this zone , to which we have no explanation , is that it is separated from the polar axis by cylindrical shell , of radius @xmath101 , with a relatively high velocity .
figure [ f6 ] provides with additional information on the magnetic structure of this solution .
one can see that the azimuthal magnetic field dominates almost everywhere in the `` diamond zone '' and in the jet , with the exception of the region immediately around the symmetry axis .
@xmath102 is particularly high in the middle of the magnetic cocoon where the poloidal field changes direction ( see the left panel of fig.[f6 ] ) .
however , the azimuthal field changes sign only in the equatorial current sheet .
this is clearly seen in the middle panel of fig.[f6 ] which shows that @xmath103 does not vanish anywhere else .
the electric current is concentrated in the jet core , the cocoon surface , and the equatorial current sheet thus showing that within most of the jet volume the magnetic field is almost force - free . the right panel of fig.[f6 ] shows the most inner part of the magnetar magnetosphere and reveals the existence of its dead zone .
the main motivation of this study was to explore the effects that the birth of a millisecond magnetar may have on the development of delayed supernova explosions .
the problem of supernova explosions is extremely complicated and we had to make a number of simplifying assumptions driven by the limitations of our numerical method .
the most problematic issue is the initial setup where we assume that the magnetic field of the magnetar generated via @xmath17-@xmath18-dynamo or other dynamo has just emerged above magnetar s surface and the large amount of energy needed to drive standard supernova explosion have just been deposited in the radiation bubble .
although the time - scales of these processes are similar ( @xmath104 ) , even small asynchronism can significantly change the initial conditions simply because of the large expansion speed of the supernova shock .
this has to be kept in mind when analyzing our numerical models , particularly their early evolution .
however , we expect the models to be more or less reliable in the second half of the simulation runs .
another important limitation of our models , as well as many other models of magnetically - driven astrophysical flows , concerns the potentially destructive role of the kink mode instabilities .
obviously , these instabilities are totally suppressed in our axisymmetric simulations and full 3d simulations are needed to investigate this issue .
we can only comment on two stabilizing factors present in the current setting .
close to the magnetar the poloidal magnetic field may provide strong `` backbone '' support for the outflows @xcite .
further away the jets propagate within a channel with `` high - inertia walls '' which could effectively dump the kink - type motions inside the channel .
finally , the condition of axisymmetry means that we can only study the exceptional case of aligned rotator .
however , we do not expect the case of oblique rotator to give dramatically different results .
for example , in the limit of magnetodynamics ( inertia - free relativistic mhd ) the spin - down power of orthogonal rotator exceeds that of aligned rotator but only by a factor of two @xcite many features of the magnetar driven outflows observed in our simulations are similar to those predicted by uzdensky & mcfadyen .
there is however a number of important differences and one of them is the nature of explosion .
none of the succesful explosions in our simulations are magnetically driven .
these explosions are produced via the deposition of thermal energy in the radiation bubble of the initial solution .
we have never seen the long phase ( many rotational periods ) of building up of magnetic pressure behind the stalled shock the key process in their scenario .
moreover , in the simulation runs where the amount of energy deposited in the buble was reduced in order to approach the conditions of failed supernova and to engage the magnetic mechanism the magnetar magnetosphere did not expand but was pressed back to the magnetar surface under the weight of neutrino - cooled layers of accreting material .
it is easy to see why .
suppose that mass @xmath105 passes through the stalled shock of failed supernova and accumulates at the radius of the light cylinder @xmath106 km ( the critical radius in the magnetic mechanism ) .
then its weight will be @xmath107 for the relatively small @xmath108 .
this is much larger than the pressure of the dipolar magnetic field at the same radius , @xmath109 ( for this estimate we assume that the at the pns surface @xmath44 g ) .
similarly , one finds that the ram pressure of the free - falling stellar material at the light cylinder is also much larger than the magnetic pressure . in principle , the magnetic field generated in the pns can be transported in the region bounded by the stalled shock of failed supernova by the large scale convective motions developing in this region and even to be locally amplified via some sort of dynamo . however , in this case the dynamics of magnetic field would be much more complicated than the one described in the simple magnetospheric model of uzdensky & mcfadyen and in order to capture it one would require much more sophysticated simulations than ours . until recently the magnetically driven outflows from astrophysical objects were classified only as `` winds '' or `` jets '' , depending on whether the outflow proceeds in all directions or it is collimated within a cone due external confinement of one sort or another .
but since the work of lynden - bell , who proposed a particular magnetic mechanism for generation of astrophysical gets , a new term , `` magnetic tower '' is gradually gaining popularity too .
in many computer simulations the development of column - like outflows have been observed and these outflows have been pronounced magnetic towers , without a detailed analysis .
in fact , collimated winds and magnetic towers have many similar properties like the magnetic braking of the central rotator , the amplification of azimuthal magnetic field , the axial compression due to hoop stress etc .
a magnetic tower occupies a finite volume which gradually increases in time whereas winds are usually associated with steady - state solutions that extend to infinity .
however , once a wind becomes super - fast its interaction with the external medium does not affect the upstream solution and thus this distinction is superficial .
a magnetic tower is prevented from lateral expansion by external pressure but some kind of external force is needed to confine any type of magnetic flow within a finite volume .
the magnetic field lines of magnetic tower have both foot - point anchored to the rotator but the same can be said about the wind from a star with , for example , a dipolar magnetic field ; the so - called open magnetic field lines of stellar winds close far in the wind zone . from our view the main distinction between a magnetic tower and a magnetically driven wind
is in the way the azimuthal magnetic field is generated in these flows . in the magnetic tower model the azimuthal field
is generated solely due the differential rotation at its base ( originally an accretion disk ) which causes winding of closed magnetic field lines whose foot points rotate with different angular velocities . thus ,
both these foot points are assumed to be in communication with each other by means of torsional alfvn waves ; the faster rotating foot - point creates stronger wave and wins in the sense that its rotation determines the sigh of the generated azimuthal field .
this also implies that the magnetic tower is a quasi - magnetostatic configuration - the vertical growth of the tower can be considered as a sequence of magnetostatic models parametrized by time .
in contrast , wind solutions are characterized by the appearance of alfvn horizon ; once the alfvn waves emitted by the rotator cross this surface they can no longer come back .
then the closed field lines of rotator s magnetic field may form two groups : ( 1 ) those that close within the horizon and ( 2 ) those that close outside of it . if a magnetic field line belongs to the first group then its foot points can communicate by means of alfvn waves . in this case
it is still important whether the rotation at the base is differential or not .
if , however , it belongs to the second group then the communication is broken and for this reason the field line can cosidered as opened .
for example , the magnetosphere of a solidly rotating star with dipolar magnetic field has two zones : the so - called `` dead zone '' where the magnetospheric plasma corotates with the star and the magnetic field is purely poloidal , and the `` wind zone '' where the magnetic field lines are `` opened '' and the magnetic field has an azimuthal component .
the magnitude of this component depends on the angular velocity of only one foot point , the upstream one , and it is created solely by the alfvn waves emitted from this point .
one may say that the winding of magnetic field in the wind zone is the feedback reaction of plasma on the magnetic field that tries to spin it up , it is there due to the finite inertia uploaded on the magnetic field lines in the wind .
there is also finite inertia in the dead zone but in the absence of outflow there is also plenty of time to enforce full corotation .
formally , the magnetic tower solutions can be constructed for a rather arbitrary distributions of external pressure @xcite .
however , for real flows with finite alfvn speed the lateral expansion due to declining external pressure can be a factor that works in favor of wind solutions .
indeed , the non - relativistic alfvn speed @xmath110 . if @xmath111 is the cylindrical radius of a magnetic surface then to zero - order approximation we have @xmath112 and , assuming constant mass loading of magnetic field lines , @xmath113 .
this gives @xmath114 and if the speed of longitudinal expansion does not decrease as fast as this then at some point the outflow will become super - alfvenic .
this would signify a transition to the wind regime .
relativity further complicates the matter as it puts an upper limit on the location of alfvn surface - it must reside inside the light cylinder @xcite .
this seems to be related to the fact that in relativistic mhd there is inertial mass associated with the magnetic ( or rather electromagnetic ) field itself . in fact , in the limit of magnetodynamics , that is relativistic mhd with vanishingly small particle inertia ( e.g. komissarov 2002 , komissarov et al.2007 ) , all of the inertia is due to electromagnetic field and these surfaces coincide this is why the light cylinder plays such an important role in the theory of pulsar winds .
in this respect we wish to comment on the `` pulsar - in - cavity '' problem of uzdensky and macfadyen .
they consider a magnetized neutron star placed in the center of an empty cavity inside a collapsing supernova progenitor .
the radius of this cavity is assumed to exceed the radius of star s light cylinder . by making the cavity so large they effectively create conditions that allow the development of a pulsar wind with the kind of magnetic winding that is specific only to winds
( notice that the star rotates as a solid body ) .
rather unfortunately , however , they still call the outflow that may eventually develope in this problem a magnetic tower .
summarizing , we wish to stress that both in analytical and numerical modeling of magnetically driven ouflows it is very important to determine whether the alfven horizon appears or not .
when the magnetic tower approximation is used this gives a key test of its validity .
when the results of computer simulations are analysed this helps to understand the physical nature of the outflow . as the thermally driven explosion ( caused by the energy deposition in the radiation bubble ) develops in our simulations a rarefaction
is created in the center and the magnetar magnetosphere begins to expand , predominantly in the equatorial direction . during this initial expansion
an alfvn horizon appears inside the magnetosphere and a magnetar wind begins to develop which injects azimuthal magnetic flux and energy into the surrounding space .
because the expansion is slow compared to the fast magnetosonic speed the fast waves quickly establish approximate magnetostatic equilibrium inside the `` magnetic cavity '' and the hoop stress of the azimuthal magnetic field ensures that this equilibrium is characterised by an axial compression . as a result the normal stress on the cavity surface develops a maximum at the poles , the cavity begins to expand predominantly in the polar direction and the polar jets are beginning to form ( see fig.[f4 ] ) .
the jets remain sub - fast up to the very end of the simulations and the termination shock never develops . however , the jet total pressure is sufficiently high to drive bow shocks into the surrounding .
in fact the structure of ouflow is quite similar to that anticipated in uzdensky and macfadyen .
there are however some qualitative differences .
first of all , the jets develop before the cavity expands beyond the light cylinder .
this is not very suprizing because the wind magnetization is not ultrarelativistic and the alfvn surface can be well inside the light cylinder .
secondly , there is no highly relativistic plasma rotation in the equatorial plane , even when the cavity expands well beyond the light cylinder .
eventually , as the central cavity becomes sufficiently large , one would expect the magnetar wind to become super - fast and the wind termination shock to appear , as it is assumed in @xcite . however , additional studies are required to determine when exactly this will occur .
this issue may be further complicated by the magnetar cooling which leads to higher magnetization of the wind and moves its fast surface further away . in any case
our results suggest that by the time of transition to super - fast regime the magnetar - driven jets will already be well established and thus the initial setup used in @xcite , where the central cavity is assumed to be large and spherical , may be somewhat unrealistic .
our results show that magnetically - driven outflows from millisecond magnetars can become highly anisotropic jets at the very early stages of supernova explosions .
assuming that their propagation speed does not decrease in time , these jets would traverse the progenitor star of radius @xmath115 cm in about 4 seconds .
after this the spin - down energy of pns will be carried away by the jets into the surrounding space and the rate of energy transfer to the supernova ejecta will drop .
the total power of long duration grb - jets is not well known but the observations of radio afterglows which are no longer subject of relativistic beaming suggest that it can be high , @xmath116erg @xcite , well in agreement with our results . since the combined total power of both jets in our simulations is @xmath117 , prior to the break out the energy transferred to the supernova ejecta will be only @xmath118 , noticeably less than what is usually derived for hypernovae .
thus , on one hand our results provide the first self - consistent numerical model of a central engine capable of producing , on a long - term basis , collimated jets with sufficient energy to explain grbs and their afterglows . on the other hand ,
the relatively small amount of energy transferred the supernova ejecta could be a problem due to the connection between high velocity sne and long grbs .
although the connection between sne and grbs is supported by strong circumstantial evidence , the direct evidence is still scarce with only handful of sne identified with grbs .
it is quite possible that the current paradigm connecting grbs with hne is simply a reflection of the observational bias towards more powerful events @xcite .
sn 2002lt / grb 021211 may the case of a grb produced by a standard sn ib / c @xcite .
the relatively high propagation speed of jets in our simulations may in part be attributed to the condition of axisymmetry .
the hydro - simulations of 2d slab jets ( e.g. komissarov & falle 2003 ) and 3d round jets @xcite show significantly lower propagation speeds of jets without enforced mirror or axial symmetry as the kink modes of current - driven or pressure - driven instabilities result in redistribution of the pressure force over larger area at the jet head ( cf .
aloy et al.,1999 ) . on the other hand ,
as the jets enter progressively less dense outer layers of collapsing star their propagation speed may actually increase .
slower propagation speed would increase the amount of energy transferred to the supernova ejecta and reduce the energy available to produce grb .
although our jets are not as fast as required by the observations of grbs the gradual cooling of pns would lead to lower mass loading and hence higher terminal speed of the jets later on .
indeed , according to the computations of pons et al .
the neutrino luminosity of pns decreases by a factor of 10 already during the first 10s and then enters the phase of rapid exponential decline , largely independent on the details of initial models and eos .
metzger et al .
argue that the magnetization @xmath119 can be reached when the the pns still has enough rotational energy to power a grb .
this corresponds to the asymptotic lorentz factor in the range of @xmath120 ( e.g. vlahakis 2004 ) .
however , the magnetic acceleration of relativistic flows is a rather slow process and we do not expect the termination speeds to be reached within the stellar boundaries .
further studies are needed to clarify the issue of magnetic acceleration .
since the cooling time of pns is comparable with the break out time , and the time - scale of long duration grbs , one would expect that that at the time of break out the mass - loading of the pns magnetosphere will still be quite high and thus the flow speed of the jets will still be relatively low , @xmath121 . only later the lorentz factor will gradually increase and eventually reach the ultra - relativistic values , @xmath122 , inferred for the grb jets .
the implications of such a non - uniform jet structure , with lorentz factor gradually decreasing with distance from the the star , for both the prompt and the afterglow emission of grbs remain to be investigated . here
we point out only few obvious points .
first of all one may expect some similarities with the model of structured jet @xcite , whose lorentz factor varies with polar angle , and the model of two - component jet @xcite . secondly , since lower lorentz factor results in weaker doppler effect one would expect softer emission from the parts of the jet emitted earlier and located further away from the star
. this could be the origin of the prompt optical and x - ray emission and early afterglows .
curiously enough , the model suggests that the zero - point of gamma - ray bursts lags behind the zero points of prompt x - ray and optical bursts . finally , while the collision between the slower earlier jet and the ism or the progenitor wind would still produce the strong forward shock usually associated with afterglows , one might expect a secondary strong forward shock where the faster late jet collides with the slower early jet
this secondary shock will propagate faster than the primary one and will eventually catch up with it .
its emission , which will be harder and beamed more strongly than that of the primary shock , may lead to distinctive features in the light curves of afterglows .
it is tempting to consider the global oscillation of magnetar s magnetosphere and the related non - stationary plasma ejection as the origin of `` inner shocks '' invoked in models of prompt grb emission @xcite .
however , it is important to check if the reconnection rate is determined mainly by the global dynamics and not by the resistivity model ( which was purely numerical in our simulations ) .
moreover , it remains to be seen if such oscillations can persist at later times ( @xmath123 few seconds ) when the the mass - loading of the magnetosphere drops and the inner `` cavity '' significantly increases in size .
later , when the jet becomes super - fast , internal shocks can be produced via interaction with inhomogeneous time - dependent cocoon and instabilities .
the inhomogeneous structure of the slow jet may also lead to variable emission from the secondary forward shock .
the results of our study show that when a core collapse of a massive star results in a birth of a millisecond proto neutron star with super - strong magnetic field ( proto - magnetar ) , this could have a spectacular effect on the supernova explosion . during the very early stage of the explosion the magnetar can produce highly collimated jets capable of puncturing the collapsing star in a matter of seconds .
the spin - down time of the proto - magnetar is an order of magnitude longer thus suggesting that a large fraction of its rotational energy , @xmath4erg , will be carried by the jets into the surrounding space .
this supports the idea that at least some long duration gamma ray bursts can have millisecond magnetars as their central engines .
the magnetic outflow is best described as a sub - fast super - alfvnic collimated wind .
the super - alfvnic nature of the flow explains the generation of azimuthal magnetic field which soon begins to dominate the flow dynamics .
the collimation is a combined effect of the inertial confinement by the stellar material and the hoop stress of the azimuthal field .
the outflow is not a magnetic tower .
it remains to be seen as to how soon the wind in the central cavity becomes super - fast and a proto - pwn is formed inside the collapsing star .
the magnetic acceleration of the jets inside the channels bored through the star and outside of the star is also an important subject for future study .
it is very likely that the grb jets produced in this fashion first emerge as only moderately relativistic flows and only later as the magnetar cools they become ultra - relativistic .
the effect of this on both the prompt and the early afterglow emission needs further investigation a non - magnetic , e.g. the delayed neutrino - driven explosion of power comparable to that of normal core - collapse supernovae is needed to turn - on the magnetic mechanism .
otherwise the pns magnetosphere is unable to expand end develop a super - alfvnic wind .
the idea that a failed supernova explosion can be revived by a millisecond magnetar does not seem to work .
we thank dmitri uzdensky and for helpful discussion of our results and the `` magnetic tower '' model , arieh konigl for many useful comments , frank timmes for help with implementing helm eos package , and the anonymous referee for constructive suggestions .
this research was funded by pparc under the rolling grant `` theoretical astrophysics in leeds '' .
akiyama s. , wheeler j.c . , meier d.l .
, lichtenstadt i. , 2003 , apj , 584 , 954 akiyama s. , wheeler j.c . , 2005 , apj , 629 , 414 aloy m.a . , ibanez j.m .
, marti j.m .
, gomez j .- l . , muller e. , 1999 , apj , 523 , l125 anderson j.m . , li z .- y . , krasnopolsky r. , blandford r.d . , 2006 , apj , 653 , l33 ardeljan n.v .
, bisnovatyi - kogan g.s . ,
moiseenko s.g . , 2005 , mnras , 359 , 333 balbus s.a . , hawley j.f
. , 1991 , apj , 376 , 214 begelman m.c . , li z .- y . , 1992 , apj , 397 , 187 berger e. , kulkarni s.r . , frail d.a . , 2004 ,
apj , 612 , 966 bethe h.a . , 1990 , rev.mod.phys . ,
62 , 801 bisnovatyi - kogan g.s . , 1970 , astron.zh . , 47 , 813 ( soviet astron . , 1971 , 14 , 652 ) bisnovatyi - kogan g.s .
, popov yu.p . , samokhin a.a . , 1976 ,
apss , 41 , 321 blondin j.m .
, mezzacappa a. , demario c. , 2003 , apj , 584 , 971 bucciantini n. , thompson t.a . , arons j. , quataert e. , del zanna l. , 2006 , mnras , 368 , 1717 bucciantini n. , quataert e. , arons j. , metzger b.b . ,
thompson t.a . , 2007a , arxiv:0705.1742 bucciantini n. , quataert e. , arons j. , metzger b.b . , thompson t.a
, 2007b , arxiv0707.2100 burrows a. , lattimer j.m . , 1986 ,
apj , 307 , 178 burrows a. , livne e. , dessart l. , livne e. , ott c.d . , murphy j. , 2006 , apj , 640 , 878 burrows a. , dessart l. , livne e. , ott c.d . , murphy j. , 2007 , apj , 664 , 416 ( astro - ph/0702539 ) cardall c.y . , fuller g.m . , 1997 ,
apj , 486 , l111 dai z.g . , gou l.j . , 2001 , apj , 552 , 72 di matteo t. , perna r. , narayan r. , 2002 , apj .
, 576 , 706 della valle m. , 2006 , chin.j.astron.astrophys . , v6 , suppl.1 , 315 duncan r.c . ,
thompson t.a . , 1992 , apj , 392 , l9 granot j. , konigl a. , piran t. , 2006 , mnras , 370 , 1946 ivanova l.n . , imshennik v.s .
, nadezhin d.k . , 1969 ,
sci.inf.astr.council.acad.sci , 13,3 jin z.p .
, yan t. , fan y.z . ,
wei d.m . , 2007 , apj , 656 , l57 khokhlov a.m. , hoflich p.a . , oran e.s . , wheeler j.c .
, wang l. , chtchelkanova a.yu . , 1999 ,
apj , 524 , l107 komissarov s.s . , 1999 , mnras , 303 , 343 komissarov s.s . ,
falle s.a.e.g , 2003 , mnras , 343 , 1045 komissarov s.s . , 2002 , mnras , 336 , 759 komissarov s.s . ,
2004a , mnras , 350 , 427 komissarov s.s .
, 2004b , mnras , 350 , 1431 komissarov s.s . , 2006 , mnras , 368 , 993 komissarov s.s . , barkov m. , lyutikov m. , 2007a , mnras , 374 , 415 .
komissarov s.s .
, barkov m.v .
, vlahakis n. , knigl a. , 2007b , mnras , online early .
kotake k. , sawai h. , yamada s. , sato k. , 2004 , apj , 608 , 391 kumar p. , granot j. , 2003 , apj , 591 , 1086 leblanc l.m .
, wilson j.r . , 1970 , apj , 161 , 541 lynden - bell d. , 1996 , mnras , 279 , 389 lynden - bell d. , 2003 , mnras , 341 , 1360 lyutikov m. , blandford r.d . , 2003 , astro - ph/0312347 macfadyen a.i .
& woosley s.e , 1999 , apj,524 , 262 masada y. , sato t. , shibata k. , 2007 , apj , 655 , 447 mszros p. , rees m.j .
, wijers r.a.m.j . , 1998 , apj , 499 , 301 mszros p. , rees m.j . , 2001 ,
apj , 556 , l37 metzger b.d . , thompson t.a . ,
& quataert e. , 2007 , apj , 659 , 561 meier d.l . ,
epstein r.e . ,
arnett w.d .
, schramm d.n . , 1976 , apj , 204 , 869 mizuno y. , yamada s. , koide s. , shibata k. , 2004 , apj , 606 , 395 mizuta a. , yamasaki t. , nagataki s. , mineshige s. , 2006 , apj , 651 , 960 moiseenko s.g .
, bisnovatyi - kogan g.s . ,
ardeljan n.v . , 2006 , mnras , 370 , 501 morsony b.j . , lazzati d. , begelman m.c . , 2006 , astro - ph/0609262 nagataki s. , takahashi r. , mizuta a. , takiwaki t. , 2007 , apj , 659 , 512 narayan r. , piran t. , kumar p. , 2001
, apj , 557 , 949 nomoto k. et al . , 2004 , in stellar collapse , ed .
c.fryer,kluwer,dordrecht , p277 norman m.,1996 , in _ energy transport in radio galaxies and quasars _
, proceedings of a workshop held in tuscaloosa , alabama , 19 - 23 sep 1995 , eds hardee p.e .
, bridle a.h . , zensus j.a . , asp conference series , sanfrancisco v.100 , p.319 obergaulinger m. , aloy m.a . , dimmelmeier h. , & muller e. , 2006 , a&a , 457 , 209 okamoto i. , 1978 , mnras , 185 , 69o panaitescu a. , 2005 , mnras , 363 , 1409 panaitescu a. , 2005 , astro - ph/0612170 piran t. , 2005a , rev.mod.phys . , 76 , 1143 piran t. , 2005b , nuovo cim .
28c , 373 pons j.a .
, reddy s. , prakash m. , lattimer j.m . , miralles j.a . , 1999 ,
apj , 513 , 780 proga d. , macfadyen a.i . ,
armitage p.j . ,
begelman m.c . , 2003 ,
apj , 599 , l5 proga d. , 2005 , apj , 629 , 397 ramirez - ruiz e. , celotti a. , rees m.j . , 2002 , mnras , 337 , 1349 rossi e. , lazzati d. , rees m.j . , 2002 , mnras , 332 , 945 ruffert m. , janka h.t . , 1999 , a&a , 344 , 573 salmonson j.d . , wilson j.r . , 1999 ,
apj , 517 , 895 schinder p.j . ,
schramm d.n . ,
wiita p.j .
, margolis s.h . ,
tubbs d.l . , 1987 , apj , 313 , 531 .
shibata m. , liu y.t . ,
shapiro s.l .
, stephens b.c . , 2006 , phys.rev d , 74 , 104026 , spitkovsky a. , 2006 , apj lett . , 648 , l51 symbalisty e.m.d . , 1984 , apj , 285 , 729 takiwaki t. , kotake k. , nagataki s. , sato k. , 2004 , apj , 616 , 1086 timmes f.x .
, swesty f.d . ,
2000 , apjss , 126 , 501 thompson t.a.,2006 , astro - ph/0611368 thompson t.a . , burrows a. , meyer b.s .
, 2001 , apj , 562 , 887 thompson c. , duncan r.c .
, 1993 , apj , 408 , 194 thompson c. , 1994 , mnras , 270 , 480 thompson c.,duncan r.c . , 1995 , mnras , 275 , 255 thompson t.a . , chang p. , quataert e. , 2004 , apj , 611 , 380 usov v. , 1992 , nature , 357 , 472 uzdensky d.a . &
macfadyen a.i . , 2006 , astro - ph/0609047 vlahakis n. , 2004 , ap&ss , 293 , 67 vlahakis n. , peng f. , konigl a. , 2003 , apj , 594 , l23 vlahakis n. , konigl a. , 2003 , apj , 596 , 1080 wang et al . , 2002 ,
apj , 579 , 671 wang l. , baade d. , hflich p. , wheeler j.c . , 2003 , apj,592 , 457 weber e.j . ,
davis l.j . , 1967 , apj , 148 , 217 wei d.m . , jin z.p . , 2003 ,
a&a , 381 , 731 wheeler j.g .
, yi i. , hoflich p. , wang l. , 2000 , apj , 537 , 810 wheeler j.g . , meier d.l . , wilson j.r .
, 2002 , apj , 568 , 807 willingale r. et al . , 2007 ,
apj , 662 , 1093 woods p.m. & thompson c. , 2004 , in compact stellar x - ray sources .
w.h.g.lewin & m. van der klis , cambridge univ .
press , cambridge , p.547 woosley s.e . ,
baron e. , 1992 , apj , 391 , 228 woosley s.e . , bloom j.s . , 2006 , ann.rev.astr.astrophys . , 44 , 507 yamada s. , sawai h. , 2004 , apj , 608 , 907 zhang w. , woosley s.e . , macfadyen a.i
, 2003 , apj , 586 , 356 | in this paper we report on the early evolution of a core - collapse supernova explosion following the birth of a magnetar with the dipolar magnetic field of @xmath0 g and the rotational period of @xmath1ms , which was studied by means of axisymmetric general relativistic mhd simulations . in this study
we use realistic eos and take into account the cooling and heating associated with emission , absorption , annihilation , and scattering of neutrinos , the neutrino transport is treated in the optically - thin regime . the supernova explosion is initiated via introducing into the initial solution the `` radiation bubble '' , whose total thermal energy is comparable with the typical energy of supernova ejecta . the numerical models exhibit highly collimated magnetically - driven jets very early on .
the jets are super - alfvnic but remain sub - fast until the end of the simulations ( t=0.2s ) .
the power released in the jets is about @xmath2erg / s which implies the spin - down time of @xmath3s .
the total rotational energy of the magnetar , @xmath4erg , is sufficient to drive a hypernova but it is not clear as to how large a fraction of this energy can be transfered to the stellar ejecta . given the observed propagation speed of the jets , @xmath5 , they are expected to traverse the progenitor in few seconds and after this most of the released rotational energy would be simply carried away by these jets into the surrounding space .
3-dimensional effects such as the kink mode instability may reduce the jet propagation speed and increase the amount of energy transferred by the jets to the supernova ejecta .
our results provide the first more or less self - consistent numerical model of a central engine capable of producing , in the supernova setting and on a long - term basis , collimated jets with sufficient power to explain long duration grbs and their afterglows .
although the flow speed of our jets is relatively low , only @xmath6 , the cooling of proto - neutron star will eventually result in much higher magnetization of its magnetosphere and ultra - relativistic asymptotic speeds of the jets .
given the relatively long cooling time - scale we still expect the jets to be only weakly relativistic by the time of break out .
this leads to a model of grb jets with systematic longitudinal variation of lorentz factor which may have specific observational signatures both in the prompt and the afterglow emission .
the simulations also reveal quasi - periodic ejection of plasma clouds into the jet on a time - scale of 20ms related to the large - scale global oscillation of magnetar s magnetosphere caused by the opening - closing of the dead zone field lines .
these kind of central engine variability may be partly responsible for the internal shocks of grb jets and the short - time variability of their gamma - ray emission .
supernovae : general stars : neutron gamma - rays : bursts methods : numerical mhd relativity |
we reformulate problem ( [ edps ] ) with conditions ( [ bord ] ) and ( initial ) as a stochastic version of an abstract cauchy problem which can be treated by using the theory of semigroups of operators:@xmath43 \right ) dt+c(u(t))dw(t ) \\
u(0 ) & = u_{0}% \end{array}% \right .
\label{8}\]]the operator @xmath44 @xmath45 @xmath46 is defined by @xmath47and the operator @xmath48 @xmath49 by@xmath50@xmath51 and @xmath52 denote usual sobolev functions spaces@xmath53 we will denote by @xmath54 and @xmath55 respectively , the scalar product and the norm in @xmath56 .
the operator @xmath57 commutes with @xmath58 and they are related by the following formula @xmath59we endow @xmath60 with the graph norm @xmath61 @xmath62 for @xmath63the operator @xmath64 is defined as follows:@xmath65 \left ( x\right ) = \dint_{\mathbb{r}}g\left ( x - y\right ) \varphi \left ( y\right ) dy .
\label{conv}\ ] ] by straightforward consequence of standard calculations , we can establish that @xmath66 , continuously so there exists a constant @xmath67 so that @xmath68note also that @xmath69 is uniformly bounded
. hence , @xmath70 is uniformly bounded . as a result of hlder s inequality
, we get @xmath71on the other hand , we have @xmath72 @xmath73 is the non linear operator , @xmath74where @xmath75 is the linear operator of multiplication by the function @xmath76 that is@xmath77 @xmath78@xmath79 denotes the space of hilbert - schmidt operators from @xmath80 to @xmath56 equipped with the norm@xmath81where @xmath82 is a complete orthonormal basis in @xmath83 suppose that the compact embedding : @xmath84then @xmath85}$ ] can be considered as a cylindrical wiener process on @xmath80 with values in @xmath86 defined on the probability space @xmath30 with a filtration @xmath87}$ ] .
[ analytique]the operator @xmath57 defined by @xmath88 is the generator of an analytic semigroup of contractions in @xmath56 , @xmath89 compact for @xmath90 .
the restrictions @xmath91 send @xmath80 into itself and are uniformly bounded in @xmath92 ( that is , there exists @xmath93 such that , @xmath94 , for @xmath95 ) .
the proof is similar to that one in @xcite .
we propose to solve , on time interval @xmath2,$ ] equation ( [ 8 ] ) in integrated form by using the stochastic generalization of the classical variation of constants formula@xmath96 ds+\dint_{0}^{t}t(t - s)c\left ( u(s)\right ) dw(s).% \end{array } \label{6}\]]we remind that a predictable @xmath56-valued stochastic process @xmath97}$ ] defined on the probability space @xmath98},p)$ ] is called a mild solution of the differential equation ( [ 8 ] ) if @xmath99 is a solution of ( [ 6 ] ) . on another hand , the stochastic process @xmath100}$ ] is called a weakened solution of ( [ 8 ] ) if @xmath99 is a solution of @xmath101 ds+\dint\limits_{0}^{t}c(u(s))dw(s ) .
\label{wkned}\]]it is well known that solutions of ( [ 6 ] ) and ( [ wkned ] ) are equivalent ( @xcite,@xcite ) .
note that the stochastic integral @xmath102 is well defined since @xmath103 $ ] .
this is due to the fact that @xmath104 is hilbert - schmidt operator and @xmath105 is linear bounded operator then , basing on the theory of hilbert - schmidt operators ( e.g. @xcite , chapter _ _
_ _ i _ _ ) , _ _
@xmath106 is hilbert - schmidt operator too@xmath107
this section is concerned with the existence of mild solutions for ( [ 8 ] ) . for this purpose , we need first to establish several propositions and lemmas .
we start by giving some useful estimates .
[ prin]1 ) there exists a constant @xmath108 such that , for all @xmath109 , @xmath110 we have @xmath111 -b\left [ vg_{g}(v)\right ] \right\vert \leq m\max ( \left\vert u\right\vert _ { { \mathcal{d}}\left ( b\right ) } \text { , } % \left\vert v\right\vert _ { { \mathcal{d}}\left ( b\right ) } ) \left\vert u - v\right\vert _ { { \mathcal{d}}\left ( b\right ) } .\]]2 ) there exists a positive constant @xmath112 such that , for all @xmath113 it holds that @xmath111 \right\vert \leq q\left\vert u\right\vert _ { { \mathcal{d}}\left ( b\right ) } \left\vert u\right\vert
.\]]3 ) there exists a positive constant @xmath114 such that , for all @xmath115 it holds that @xmath116 [ prop1]consider the following abstract problem@xmath43 \right ) dt+a(u(t))dw(t ) \\
u(0 ) & = u_{0}.% \end{array}% \right . \label{pb1}\]]suppose
that the operator @xmath117 satisfies@xmath118and there exists a constant @xmath119 such that for all @xmath120@xmath121then , for each @xmath122 @xmath123 , problem ( [ pb1 ] ) has a unique mild solution @xmath99 , @xmath124 $ ] such that @xmath125% } { sup}e(\left\vert u(t)\right\vert _ { d(b)}^{2})<\infty .$ ] we use the method of successive approximations .
let @xmath122 @xmath123 and define the sequence @xmath126
by@xmath127ds+\dint% \limits_{0}^{t}t(t - s)a(u_{n}(s))dw(s).% \end{array}%\]]let us denote by @xmath128 the banach space of all nonanticipating @xmath60-valued continuous stochastic processes @xmath129}$ ] endowed with the norm@xmath130assume that @xmath131 is bounded in @xmath132 that is , @xmath133 such that @xmath134 , @xmath135,\forall n\geq 1.$ ] hence @xmath136)_{n\geq 1}$ ] is also bounded in @xmath128 , since @xmath137 let @xmath138we have@xmath139 dw(s)\right\vert _ { d(b)}^{2 } \\ + 2e\left\vert \dint\limits_{0}^{t}t(t - s)\left ( b[u_{n}(s)g_{g}(u_{n}(s))]-b[u_{n-1}(s)g_{g}(u_{n-1}(s))]\right ) ds\right\vert _ { d(b)}^{2 } \\
\leq \underset{i}{\underbrace{2e(\dint\limits_{0}^{t}\left\vert t(t - s)\left [ a(u_{n}(s))-a(u_{n-1}(s))\right ] \right\vert _ { 2}^{2}ds ) } } \\
+ \underset{ii}{\underbrace{2e\left\vert \dint\limits_{0}^{t}t(t - s)\left ( b[u_{n}(s)g_{g}(u_{n}(s))]-b[u_{n-1}(s)g_{g}(u_{n-1}(s))]\right ) ds\right\vert _ { d(b)}^{2}}}% \end{array}%\ ] ] to calculate @xmath144 , we have@xmath145-b[u_{n-1}(s)g_{g}(u_{n-1}(s))]\right ) ds\right\vert _ { d(b ) }
\\ = \left\vert \dint\limits_{0}^{t}t(t - s)\left ( b[u_{n}(s)g_{g}(u_{n}(s))]-b[u_{n-1}(s)g_{g}(u_{n-1}(s))]\right ) ds\right\vert
\\ + \left\vert b\dint\limits_{0}^{t}t(t - s)\left ( b[u_{n}(s)g_{g}(u_{n}(s))]-b[u_{n-1}(s)g_{g}(u_{n-1}(s))]\right ) ds\right\vert \\ \leq \dint\limits_{0}^{t}m\max ( \left\vert u_{n}(s)\right\vert _ { d(b)},\left\vert u_{n-1}(s)\right\vert _ { d(b)})\left\vert u_{n}(s)-u_{n-1}(s)\right\vert _ { d(b)}ds
\\ + \dint\limits_{0}^{t}\dfrac{c}{\sqrt{t - s}}\left\vert b[u_{n}(s)g_{g}(u_{n}(s))]-b[u_{n-1}(s)g_{g}(u_{n-1}(s))]\right\vert ds \\ \leq mr(t\ + 2\sqrt{t}c)\underset{0\leq t\leq t}{\sup } \left\vert u_{n}(t)-u_{n-1}(t)\right\vert _ { d(b)}.% \end{array}%\]]therefore@xmath146hence , we obtain@xmath147e(% \underset{0\leq t\leq t}{\sup } \left\vert u_{n}(t)-u_{n-1}(t)\right\vert _ { d(b)}^{2})% \end{array}%\]]by choosing @xmath148 small enough so that @xmath149<\frac{1}{2 } , \label{condition t}\]]we obtain@xmath150@xmath151 is the general term of an absolutely convergent series in the banach space @xmath152 hence , @xmath153 is a convergent sequence in @xmath154 the limit in @xmath155 @xmath156 is a solution of ( [ pb1 ] ) for fixed @xmath6 . to complete the proof of the proposition
, we have to show that the sequence @xmath157 remains in @xmath155 i.e. @xmath158@xmath159 , \forall
t\in \lbrack 0,t].% \end{array}%\]]using @xmath148 such that ( [ condition t ] ) holds , we have@xmath160to prove the uniqueness of the solution ( up to equivalence ) , consider @xmath109 and @xmath161 two mild solutions of ( [ pb1 ] ) on @xmath2.$ ] we might show that @xmath162we have@xmath163-b[v(s)g_{g}(v(s))])ds\right% \vert _ { d(b)}^{2 } } } \\ & + \underset{(2)}{\underbrace{2e\left\vert \dint\limits_{0}^{t}t(t - s)[a(u(s))-a(v(s))]dw(s)\right\vert _ { d(b)}^{2}}}% \end{array}%\]]we obtain@xmath164to calculate @xmath165 , we have@xmath166-b[v(s)g_{g}(v(s))])ds\right% \vert _ { d(b ) } & \\
\leq mr(t+2c\sqrt{t})\underset{0\leq t\leq t}{\sup } \left\vert u(t)-v(t)\right\vert _ { d(b ) } & \end{array}%\]]then@xmath167finally , we obtain@xmath168 \\
& e(\underset{0\leq t\leq t}{\sup } \left\vert u(t)-v(t)\right\vert _ { d(b)}^{2})\text { \ \ \ \ \ \ \ \ \ } \forall t\in \lbrack 0,t]% \end{array}%\]]since @xmath169 is chosen such that @xmath170 < \frac{1}{2}$ ] , hence we have@xmath171which leads to@xmath172,\qquad p - a.s.\text{\ \ \ } \ ] ] we return to problem ( [ 8 ] ) .
we propose as in @xcite to construct a sequence of lipschitz continuous mappings @xmath173 to approximate the non - lipschitzian function@xmath174 on @xmath175 for @xmath176 , define @xmath177 by@xmath178it is clear that the sequence @xmath179 converges to @xmath180 uniformly on @xmath109 @xmath181 @xmath182 as @xmath183 let @xmath189 and @xmath82 a complete orthonormal basis in @xmath190 hence@xmath191as @xmath192 is lipschitz in @xmath109 on @xmath193 then there exists @xmath194 such that @xmath195using the hypothesis ( [ cond ] ) , we have@xmath196which leads to the fact that there exists @xmath197 such that @xmath198 we aim to prove weak convergence of the sequence @xmath202\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] by proving the tightness of @xmath203\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] and @xmath204\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] since @xmath205 is deterministic .
let the initial condition @xmath206 be a continuous deterministic mapping on @xmath207the sequences @xmath203\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] and @xmath204\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] are tight in @xmath208,d(b)).$ ] by lemma 4 , it follows that for each @xmath209 the sequence @xmath201 satisfies the conditions of proposition prop1 .
hence , there exists a unique mild solution @xmath210 $ ] for problem ( [ approx ] ) such that @xmath125}{sup}% e(\left\vert u_{n}(t)\right\vert _ { d(b)}^{2})<\infty .$ ] for given @xmath211 and @xmath212,$ ] it follows from lemma @xmath213 ( in appendix a ) that@xmath214 on another hand , by using lemma @xmath215 ( in appendix a ) , we have for given @xmath216 and @xmath212$]@xmath217\right\vert _ { d(b)}^{2}ds \\ & + 2e(\dint\limits_{t}^{t^{\prime } } \left\vert b(t(t^{\prime } -s)[u_{n}(s)g_{g}(u_{n}(s))]\right\vert _ { d(b)}^{2 } \\ & \leq 2e(\dint\limits_{0}^{t}\left\vert b(t(t - s)-t(t^{\prime } -s))\right\vert _ { d(b)}^{2}\left\vert u_{n}(s)g_{g}(u_{n}(s))\right\vert _ { d(b)}^{2}ds \\ & + 2e(\dint\limits_{t}^{t^{\prime } } \left\vert b(t(t^{\prime } -s)\right\vert _ { d(b)}^{2}\left\vert u_{n}(s)g_{g}(u_{n}(s))\right\vert _ { d(b)}^{2}ds% \end{array}%\]]using ( [ cont ] ) , we obtain@xmath218it follows that @xmath219for @xmath220 , we define @xmath221@xmath222if we let in the garsia - rodemich - rumsey lemma ( lemma 9 , in appendix a ) : @xmath223 @xmath224 @xmath225 @xmath226 @xmath227 @xmath228 and @xmath229 then@xmath230since @xmath231 @xmath232 @xmath233 thus@xmath234finally @xmath235it follows that for any @xmath236 @xmath237@xmath238,t\neq t^{\prime } } { sup}\dfrac{% \left\vert j_{n}(t,\omega ) -j_{n}(t^{\prime } , \omega \right\vert _
{ d(b)}}{% \left\vert t - t^{\prime } \right\vert ^{\overline{\delta } } } & \leq \underset{% t , t^{\prime } \in \lbrack 0,t],t\neq t^{\prime } } { \sup } \mathcal{c}% _ { 1}\left\vert t - t^{\prime } \right\vert ^{\gamma -1-\overline{\delta } % } r^{1/2 } \\ & \leq \mathcal{c}_{1}^{\prime } r^{1/2}.% \end{array}%\]]hence , for @xmath239 we can define@xmath240,d(b ) ) } & = \underset{t\in \lbrack 0,t]}{\sup } \left\vert j_{n}(t,\omega ) \right\vert _ { d(b ) } \\ & + \underset{t , t^{\prime } \in
\lbrack 0,t],t\neq t^{\prime } } { \sup } \dfrac{% \left\vert j_{n}(t,\omega ) -j_{n}(t^{\prime } , \omega ) \right\vert _ { d(b)}}{% \left\vert t - t^{\prime } \right\vert ^{\overline{\delta } } } \\ & \leq \mathcal{c}^{\prime \prime } r^{1/2}% \end{array}%\]]we have then for every @xmath220@xmath241,d(b))}\leq \mathcal{c}^{\prime \prime } r^{1/2}\right\}% \end{array}%\]]let us now show that @xmath204\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] is tight on @xmath242,d(b)).$ ] for @xmath220 , we define@xmath243,d(b))/\left\vert f\right\vert _ { c^{\overline{% \delta } } ( [ 0,t],d(b))}\leq r\right\ } .\]]@xmath244 is a compact set of @xmath242,d(b))$ ] ( see the proof b1 in appendix b ) and we have the following : @xmath245 & = \underset{n}{\sup } % p[\left\vert j_{n}(.,\omega \right\vert _
{ c^{\overline{\delta } % } ( [ 0,t],d(b))}>r ] \\ & \leq \underset{n}{\text { } \sup } p\left [ \overline{a}_{\dfrac{_{r^{2}}}{% \mathcal{c}^{^{\prime \prime } 2}}}\right ] \\ & \leq \underset{n}{\text { } \sup } p\left\ { \dint\limits_{0}^{t}\dint\limits_{0}^{t}\left ( \dfrac{j\left\vert _ { n}(t)-j_{n}(t^{\prime } ) \right\vert _ { d(b)}}{\left\vert t - t^{\prime } \right\vert ^{\gamma } } \right ) ^{2}dtdt^{\prime } > \dfrac{r^{2}}{\mathcal{c}% ^{^{\prime \prime } 2}}\right\}% \end{array}%\]]using markov inequality , we obtain@xmath246\leq \dfrac{\mathcal{c}% ^{^{\prime \prime } 2}}{r^{2}}\underset{n}{\text { } \sup } e\left\ { \dint\limits_{0}^{t}\dint\limits_{0}^{t}\left ( \frac{\left\vert j_{n}(t)-j_{n}(t^{\prime } ) \right\vert _ { d(b)}}{\left\vert t - t^{\prime } \right\vert ^{\gamma } } \right ) ^{2}dtdt^{\prime } \right\ } . \label{nad}\ ] ] since @xmath247 ( [ nad ] ) leads to@xmath246\leq \dfrac{m^{^{\prime \prime } } } { r^{2}}\text { \ } < \epsilon \text{\ } \]]for @xmath248 large enough .
this proves the tightness of @xmath249\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } .$ ] a similar argument implies that @xmath203\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] is also tight on the same space .
let the initial condition @xmath206 be a continuous deterministic mapping on @xmath207 then , there exists a probability basis @xmath250},\overline{p})$ ] on which there is a cylindrical wiener process @xmath251}$ ] and a mild solution @xmath97}$ ] of ( [ 6 ] ) in @xmath208,d(b))$ ] .
the tightness of @xmath203\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] and @xmath204\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] implies that @xmath252\}_{n\in % tcimacro{\u{2115 } } % % beginexpansion \mathbb{n } % endexpansion } $ ] converges weakly on @xmath242,d(b)),$ ]
this means that there exists a subsequence @xmath253 and a @xmath242,d(b))$]-valued random variable @xmath161 such that @xmath254,d(b)),\text { as } k\rightarrow + \infty .\]]by skorohod representation theorem , there exists a probability space @xmath255 @xmath7with a filtration @xmath256}$ ] and @xmath257,d(b))$]-valued random variables @xmath258 and @xmath109 such that as @xmath259@xmath260,d(b))\]]and @xmath261the mild solution @xmath262 of ( [ approx ] ) is also a weakened solution of ( [ approx ] ) ( see for instance @xcite ) , that is @xmath262 satisfies the following integral equation @xmath263
ds+\dint% \limits_{0}^{t}a_{n}(u_{n}(s))dw(s).% \end{array}%\]]hence , for each @xmath264,@xmath265 ds% \end{array}%\]]is a square integrable martingale with respect to @xmath266 and has the following quadratic variation@xmath267since @xmath268 ( @xmath269 denotes the probability distribution)@xmath270@xmath271 ds\]]has the same distribution as @xmath272 and hence @xmath273(we denote by @xmath274 the expectation with respect to the probability measure @xmath275)@xmath107 on another hand , since @xmath276 is a martingale , that is @xmath277and using the fact that it holds that @xmath280which implies that @xmath281 is a square integrable martingale on @xmath282 with respect to @xmath283 on another hand , since @xmath272 has the quadratic variation ( [ qv ] ) , this means that for @xmath284 @xmath285 @xmath286then@xmath287it follows that@xmath288 hence @xmath289 is an @xmath290martingale and as a consequence , @xmath291 has the unique quadratic variation process @xmath292from ( [ samelaw ] ) , since this means that @xmath294 is a sequence of uniformly integrable martingales
. therefore , there exists on @xmath282 a square integrable martingale @xmath295 such that for @xmath296 and@xmath297 ds\]]with@xmath298by the representation theorem for square integrable martingales @xcite , there exists a cylindrical wiener process @xmath299 defined on @xmath300},\overline{p})$ ] such that @xmath301.\]]then @xmath109 satisfies that @xmath101 ds+\dint\limits_{0}^{t}\sqrt{\lambda u^{+}(s)}d\overline{w}(s),\text { \ \ } t\in \lbrack 0,t]\]]or equivalently @xmath302 ds \\
+ \dint_{0}^{t}t(t - s)c\left ( u(s)\right ) d\overline{w}(s).% \end{array}%\ ] ] we represent @xmath307 in terms of the eigenvalues of @xmath308 denoted @xmath309 to which correspond the eigenvectors @xmath310@xmath311then , we have@xmath312where we have used the fact that for all @xmath313,$ ] @xmath314for @xmath315,$ ] we have @xmath316 if we set @xmath317 @xmath318),$ ] hence@xmath319similarly , for the same @xmath212 $ ] as above , we have@xmath320 * [ a2 ] * for @xmath212,$ ] there exists constants @xmath321such that@xmath322 \right\vert _
{ d(b)}^{2}du\leq c^{\prime \prime } \left\vert t - s\right\vert ^{\eta } \text { \ \ \ \ \ \ } \]]@xmath323with@xmath324 from the representation@xmath325we obtain @xmath326hence@xmath327 \right\vert _
{ d(b)}^{2}du & \\ \leq \dint\limits_{0}^{s}\dsum\limits_{j=1}^{\infty } \left\vert \exp \{-w_{j}^{2}(t - u)\}-\exp \{-w_{j}^{2}(s - u)\}\right\vert _ { d(b)}^{2}\left\vert \left\langle .,\varphi _ { j}\right\rangle \right\vert _ { d(b)}^{2}\left\vert b\varphi _ { j}\right\vert _ { d(b)}^{2}du & \\ \leq \dint\limits_{0}^{s}\dsum\limits_{j=1}^{\infty } \left\vert \exp \{-w_{j}^{2}(s - u)\}\right\vert ^{2}\left\vert 1-\exp \{-w_{j}^{2}(t - s)\}\right\vert ^{2}\left\vert \left\langle .,\varphi _ { j}\right\rangle \right\vert _ { d(b)}^{2}\left\vert b\varphi _ { j}\right\vert _ { d(b)}^{2}du & \\
\leq \dsum\limits_{j=1}^{\infty } \left\vert \left\langle .,\varphi _
{ j}\right\rangle \right\vert _ { d(b)}^{2}\left\vert b\varphi _ { j}\right\vert _ { d(b)}^{2}\left\vert w_{j}^{2}(t - s)\right\vert ^{2\lambda } \dint\limits_{0}^{s}\exp \{-2w_{j}^{2}(s - u)\}du & \\ \leq \underset{j}{\sup } \left\vert \left\langle .,\varphi _ { j}\right\rangle \right\vert _ { d(b)}^{2}\underset{j}{\sup } \left\vert b\varphi _ { j}\right\vert _ { d(b)}^{2}\left\vert t - s)\right\vert ^{2\lambda } \dsum\limits_{j=1}^{\infty } w_{j}^{-(2 - 4\lambda ) } & \\
\leq c_{2}\left\vert t - s)\right\vert ^{2\lambda } \dsum\limits_{j=1}^{\infty } w_{j}^{-(2 - 4\lambda ) } & \end{array}%\ ] ] where we have used the fact that for all @xmath313,$ ] @xmath314for @xmath315,$ ] we have @xmath316 if we set @xmath317 @xmath318),$ ] hence@xmath322
\right\vert _ { d(b)}^{2}du\leq c_{1}\left\vert t - s)\right\vert ^{\eta } .\]]similarly , for the same @xmath212 $ ] as above , we have@xmath328 [ a3](garsia - rodemich - rumsey ) let the function @xmath329 non - decreasing with @xmath330 let the function @xmath331\rightarrow \lbrack 0,1]$ ] be continuous and non - decreasing with @xmath332 set @xmath333 let @xmath334 be a continuous function on @xmath335 $ ] and suppose that @xmath336then , for all @xmath337 $ ] , we have@xmath338 * proof b1 * since @xmath339,d(b))}=\underset{t\in \lbrack 0,t]}{sup}\left\vert f(t)\right\vert _ { d(b)}+\underset{t , t^{\prime } \in
\lbrack 0,t],t\neq t^{\prime } } { sup}\dfrac{\left\vert f(t)-f(t^{\prime } ) \right\vert _ { d(b)}}{\left\vert t - t^{\prime } \right\vert ^{\overline{% \delta } } } , \ ] ] this means that @xmath340 @xmath341,t\neq t^{\prime } $ ] @xmath342,d(b))}\left\vert t - t^{\prime } \right\vert ^{\overline{\delta } } \]]as @xmath343 we have@xmath344this means that @xmath244 is uniformly equicontinuous . by the theorem of ascoli - arzela , @xmath244 is a compact of @xmath242,d(b)).$ ] f. azam and j. w. ammerman , cycling of organic matter by bacterioplankton in pelagic marine ecosystems : microenvironmental considerations . in flows of energy and material in marine ecosystems .
nato conf .
ser.@xmath351 , mar .
@xmath352 @xmath353 @xmath354 .
w. k. fitt , the role of chemosensory behavior of symbiodinium microadriatica , intermediate hosts , and host behavior in the infection of coelenterates and molluscs with zooxanthellae .
@xmath376 @xmath377 @xmath378 .
d. c. hauser , r. m. levandowsky , s. h. hunter , l chunosoff and j. s. hollwitz , chemosensory responses by the heterotrophic marine dinoflagellate crypthecodinium cohnii .
microbial ecology @xmath368 @xmath384 @xmath385 . | in this paper , we are interested in the analytical study of a nonlinear stochastic partial differential equation ( spde ) arising as a model of phytoplankton aggregation .
this spde consists in a diffusion equation with a chemotaxis term responsible of self - attraction of phytoplankton cells and a multiplicative branching noise .
existence of mild solutions is established through weak and tightness arguments .
@xmath0 keywords : phytoplankton aggregation , nonlinear stochastic partial differential equation , semigroups , chemotaxis , gaussian space - time white noise , weak convergence , tightness , skorohod representation theorem .
introduction in this paper , we are interested in the following nonlinear stochastic partial differential equation:@xmath1 ( x)\right ) + \sqrt{\lambda u_{+}(t , x)}\overset{.}{w}(t , x ) , \label{edps}\]]in @xmath2\times \omega , $ ] where @xmath3 0,l\right [ $ ] is a bounded domain with boundary @xmath4 in @xmath5 is a one dimensional coordinate , @xmath6 is time .
the motivation in studying this spde@xmath7is that equation ( [ edps ] ) arises as a model of phytoplankton aggregation ( @xcite,@xcite,@xcite,@xcite ) .
in fact , the authors in ( @xcite,@xcite , elsaadi07a,@xcite ) have investigated an individual - based model ( ibm ) of a population of phytoplankton that takes into account small scales biological mechanisms for phytoplankton cells .
these processes are : random dispersal of phytoplankton cells due to turbulence , spatial interactions between phytoplankton cells caused by chemical signals and random division and death of phytoplankton cells .
the aim of such modelling was to capture the main features of the individual dynamics at small scales that are responsible at a larger scale for the formation of aggregating patterns .
an eulerian version of the ibm has been derived in ( @xcite , elsaadi07a ) and the spde @xmath8 is obtained as a continuum limit of this ibm .
@xmath9 represents the spatio - temporal distribution density of phytoplankton on the vertical water column .
we denote by @xmath10 the positive part of @xmath9 , that is @xmath11 the diffusion term in @xmath8 takes into account the random spread of the phytoplankton cells with the coefficient of diffusion @xmath12 ; while the transport term , which is a chemotaxis term , describes the interaction mechanisms between the phytoplankton cells via the velocity @xmath13 the latter has the form of convolution as in @xcite , i.e. , @xmath14 ( x)=\int_{\mathbb{r}}g\left ( x - x^{\prime } \right ) u^{0}(t , x^{\prime } ) dx^{\prime } , \]]where @xmath15 is the attractive force given by : @xmath16 1_{\left [ -r_{1},-r_{0}\right ] \cup \left [ r_{0},r_{1}\right ] } \left ( x\right ) .\ ] ] and @xmath17the biological explanation of the interactions between phytoplankton cells is based on the following works ( @xcite,@xcite,@xcite,@xcite,@xcite,@xcite,@xcite , spero85 ) which report that species of phytoplankton such as dinoflagellates and motile algae have chemosensory abilities i.e. , they can sense the chemical field generated by the presence of other dinoflagellates and motile algae that are present at a certain distance .
more precisely , dinoflagellates and motile algae leak organic matter such as sugar and amino acids , forming regions around them having concentrations higher than average @xcite .
experiments studies on the chemosensory abilities in dinoflagellates and motile algae ( @xcite , @xcite ) show that the released products attract other dinoflagellates and motile algae that are present in a suitable neighborhood .
it has also been observed that high concentrations of these products inhibit the chemosensory behavior in dinoflagellates and motile algae @xcite.so , if a phytoplankton cell is located at a position @xmath18 , the extracellular products released by this cell form a concentration around @xmath18 , which is highest in the closest vicinity of @xmath18 ( for instance , on a radius of length @xmath19 ( @xmath20 small ) ) and then decreases progressively . via their chemosensory abilities , all dinoflagellates at positions @xmath21 such that @xmath22 ( @xmath23 and @xmath20 too small relative to @xmath24 ) detect the differences of concentration in water and hence are attracted to the cell in @xmath25 beyond @xmath24 , cells can not perceive the difference in concentration because they are sensory limited ( @xcite,@xcite , jackson89 ) .
hence , they are not attracted .
@xmath26 describes the velocity induced at the site @xmath18 by the net effects of all phytoplankton cells at various sites @xmath21 .
the kernel @xmath27 associates a strength of interaction per unit density as a function of the distance @xmath28 between any two phytoplankton cells in sites @xmath18 and @xmath21 .
@xmath15 behaves as a gradient , that is , a cell is attracted to the region of high density .
@xmath29 is a white noise in space and time defined on some probability space @xmath30 @xcite and @xmath31 is the branching rate .
the term @xmath32 describes the stochastic fluctuations of the number of phytoplankton cells as a result of the random birth and death events.here , @xmath33 represents the vertical axis oriented downward from the surface to the seabed .
the point @xmath34 is at the surface of water and @xmath35 is the limit of the `` euphotic zone '' ie .
the upper layers in the vertical water column .
the restriction of model ( [ edps ] ) to this domain is due to the fact that phytoplankton cells can survive and multiply only in the euphotic zone .
therefore , von - neumann boundary conditions are imposed at the surface @xmath34 and at @xmath35 : @xmath36\times \partial \omega .
\label{bord}\]]and the initial density is @xmath37 to our knowledge , spde ( [ edps ] ) is new and unknown in both biological and mathematical literature . from the mathematical point of view , it is a complication .
the major difficulties come from the two nonlinear terms : the chemotaxis term with a convolution and the stochastic branching term with the nonlinear multiplicative noise .
we point out that previous works ( adioui , @xcite ) have concerned the study of the deterministic part of the spde ( [ edps ] ) ( i.e. equation ( [ edps ] ) without the noise term ) with the goal of analyzing the influence of the spatial interactions between phytoplankton cells on the aggregation process . here in this paper ,
we are interested by the whole equation ( [ edps ] ) which includes the effect of the stochastic branching term .
the volatility @xmath38 of the space - time white noise is non - lipschitzian ; hence , we are not able to expect the existence of strong or mild solutions for ( [ edps ] ) .
however , we might expect the existence of a mild solution to ( [ edps ] ) in the weak sense ; that is we shall construct a probability basis @xmath39},\overline{p})$ ] , on which there exists a gaussian space - time white noise @xmath40 and a mild solution @xmath41 to ( [ edps ] ) .
this will be done by using tightness arguments which are useful in obtaining weak convergence .
the rest of the paper is organized as follows . in the next section
, we present the abstract formulation of the problem and in section 3 , we present our results on existence of mild solutions at the weak sense .
several steps are arranged for proving the main theorem .
some estimates on the semi - group generated by the operator @xmath42 are listed in the appendix . |
when the skin is pinched , wrinkles appear quite early on its surface .
the same phenomenon occurs when the skin is sheared , i.e. pinched with one finger moving in one direction and the other fixed or moving in the opposite direction . in fact , pinching is one of the tests performed by dermatologists and surgeons @xcite when trying to assess the direction of greatest tension @xcite in the neighborhood of a site of interest .
sometimes called _ lines of cleavage _ @xcite , the orientations of the lines of greatest tension are crucial to the way a scar heals . for a cut across the lines ,
the lips of a wound will be pulled away from one another during the healing process , while they will be drawn together if the cut has occurred parallel to the lines . in one case
the resulting scar can be quite unsightly , in the other it is almost invisible . in this paper
we investigate the mechanical stability of two toy models for the human skin under shear in its plane , and view the onset of small - amplitude , unstable solutions as a prototype for skin wrinkling . of course , skin is a complex , multi - faced organ , and it is not easily , nor perhaps realistically , modeled . in section [ section2 ] , we first view it as an initially isotropic , neo - hookean layer of finite thickness .
although a two - layered model of skin would be more realistic , it would greatly complicate the theoretical analysis of the shear instability properties .
therefore , we prefer to consider an epidermis of vanishing thickness on top of a hyperelastic dermis , by defining a proper surface elastic energy . in other words
we let one of the layer s faces be a material curve endowed with intrinsic elastic properties associated with extensibility , but no bending stiffness ( see @xcite for a rigorous exposition of such elastic coatings ) .
we account for the lines of greatest tension by imposing a finite plane pre - stretch in a given direction ; in other words , we simulate those lines through _ strain - induced anisotropy_. then we investigate whether surface energy and pre - stretch promote or attenuate the appearance of wrinkles when the layer is subject to simple shear in the direction of the cleavage lines . in section [ shear instability of a fibre - reinforced skin tissue ] , we then view skin as being intrinsically _ anisotropic _ that is , we switch to the point of view that lines of greatest tension are due to the presence of families of parallel bundles of stiff collagen fibres imbedded in a softer elastin matrix . the introduction of even the simplest anisotropy transverse isotropy due to a single privileged direction complicates the equations of incremental instability greatly , and we thus restrict attention to a homogeneous solid without surface tension .
we also omit finite - size effects by considering a half - space instead of a layer , and thus by focusing on the biot surface instability phenomenon @xcite . here , the anisotropic contribution to the stored energy is that recently proposed by ciarletta et al.@xcite .
its polyconvexity ensures good properties from the physical point of view , such as strong ellipticity in compression , in contrast to the standard reinforcing model used recently by destrade et al.@xcite for the same stability study .
first we consider an isotropic elastic material of finite thickness @xmath0 , undergoing an homogeneous shear . in order to mimic the response of human skin
, we incorporate the presence of a residual stretch @xmath1 along the main cleavage lines , so that the base deformation field reads @xmath2 where @xmath3 is the current position of a material point which was at @xmath4 in the reference configuration , and @xmath5 is a constant .
hence , we see that the deformation can be decomposed as a plane stretch of amount @xmath1 followed by a simple shear of amount @xmath5 , with deformation gradient @xmath6 written as @xmath7 we note that the layer s thickness @xmath0 remains unchanged through the deformation . for simplicity , we take the layer to be made of an isotropic neo - hookean incompressible material with a surface energy at the free boundary @xmath8 , so that its total strain energy @xmath9 reads @xmath10 where @xmath11 is the shear modulus , @xmath12 is the left cauchy - green deformation tensor , @xmath13 is the surface tension coefficient , @xmath14 is the elastic shear modulus per surface unit , and the comma denotes partial derivative .
this is akin to endowing one of the material boundary of the layer with a surface energy with a term proportional to changes in area ( as is often done in fluid mechanics , see e.g. @xcite ) , and another contribution depending on the elastic deformation of the surface .
this last term is proportional to a stretch measure of the surface deformation tensor , which is chosen for invariance requirements because we consider this elastic layer as a hemitropic film @xcite .
now from the constitutive assumptions in eq.([const ] ) , we find that @xmath15 , the cauchy stress tensor corresponding to the large deformation in eq . is given by @xmath16 where @xmath17 is a lagrange multiplier due to the constraint of incompressibility .
writing that the boundary at @xmath18 is traction - free fixes the value of @xmath17 as @xmath19 .
straightforward calculations reveal that principal stretches of the deformation field in eq.([def ] ) are @xmath20 ( @xmath21 ) given by @xcite @xmath22 and that the eulerian principal axes @xmath23 are obtained after an anti - clockwise rotation of angle @xmath24 of the in - plane coordinate axes about the @xmath25 axis , where @xmath26 from eq .
, we note that @xmath27 we now look for a perturbation solution in the neighbourhood of the large deformation , using the theory of incremental deformations @xcite . hence we call @xmath28 the incremental displacement field , for which the incremental incompressibility condition imposes that @xmath29 the constitutive equation for the components of the incremental nominal stress @xmath30 reads in general as @xcite , @xmath31 where @xmath32 is the increment in the lagrange multiplier , and @xmath33 is the fourth - order tensor of instantaneous moduli , i.e. the push - forward of the fixed reference elasticity tensor . in the absence of body forces
, we can therefore write the equilibrium equation of the incremental nominal stress @xmath34 as @xmath35 in the case of a neo - hookean material , it is easy to check that the components of @xmath36 are simply @xmath37 so that eq.([divs ] ) in the coordinate system aligned with the eulerian principal axes takes the following simplified form @xmath38 differentiating these incremental equilibrium equations with respect to @xmath39 , @xmath40 , and @xmath41 , respectively , and using the incremental incompressibility condition in eq.([inc1 ] ) , we find that @xmath42 that is , the incremental lagrange multiplier is a laplacian field @xcite . now , for the incremental _ boundary conditions _ , we consider that the bottom @xmath43 of the layer is fixed ( clamped condition ) : @xmath44 while the top face @xmath45 remains free of incremental traction : @xmath46 we search for solutions to eqs.([inc1],[pi],[plapl ] ) in the following form : @xmath47 corresponding to the occurrence of plane wrinkles with wavenumber @xmath48 , forming an angle @xmath49 with the direction of maximum extension .
it is easy to show that a solution in the form of eq.([sep ] ) is given by @xcite @xmath50 where @xmath51 are yet arbitrary constants , and @xmath52 is fixed by imposing eqs.([pi ] ) , as @xmath53 using eqs .
, it can be checked that only four independent boundary conditions result from eqs .. setting @xmath54^t$ ] , they can be written in the following matrix form : @xmath55 where the components of the matrix @xmath56 are @xmath57 here @xmath58 , @xmath59 are the _
characteristic capillary * and elastic * lengths _ of the material , respectively .
the resulting condition for the wrinkling instability is @xmath60 .
after lengthy manipulations not reproduced here , it is possible to show that the earliest onset of instability occurs at @xmath61 that is , when the wrinkles are aligned with the direction where the greatest stretch , @xmath62 , takes place .
there , @xmath63 , and the _ dispersion relation _ reduces to @xmath64\cosh(kh ) \right . \\ \left . -kl_\text{el}\lambda_1 ^ 2\left[2\lambda_1+(1+\lambda_1 ^ 2)\sinh(\lambda_1 kh)\sinh(kh)\right]\right\}\\ + k^2\lambda_1 ^ 2 l_\text{el}l_\text{cap}\left [ -2\lambda_1 + 2\lambda_1\cosh(kh)\cosh(\lambda_1 kh)-(1+\lambda_1 ^ 2)\sinh(kh)\sinh(\lambda_1 kh)\right]\\ + kl_\text{cap}(\lambda_1 ^ 2 - 1)\left[\cosh(kh)\sinh(\lambda_1 kh)-\lambda_1\sinh(kh)\cosh(\lambda_1 kh)\right ] \\
+ 4\lambda_1(\lambda_1 ^ 2 + 1 ) - \lambda_1(5 + 2\lambda^2_1+\lambda_1 ^ 4)\cosh(kh)\cosh(\lambda_1 kh ) \\+ ( 1 + 6\lambda^2_1+\lambda_1 ^ 4)\sinh(kh)\sinh(\lambda_1 kh)=0 . \label{disp}\end{gathered}\ ] ] this dispersive bifurcation criterion is the main result of this section , linking the material ( @xmath65 ) and geometrical ( @xmath1 and @xmath5 appearing inside the expression for @xmath66 in eq . ) parameters describing the sheared layer , to the wavelength of the expected wrinkles ( through the non - dimensional quantity @xmath67 , where @xmath68 is the wrinkles wavelength ) .
.,title="fig:",height=188 ] .,title="fig:",height=188 ] in figure [ slab ] we show the separate influences of the pre - stretch and of the surface energy .
we plot the critical amount of shear @xmath69 , at which wrinkles occur , against @xmath70 , the ratio of the layer s thickness to the wrinkles wavelength .
we find that as @xmath70 becomes small , @xmath69 increases rapidly , showing that the layer is more and more stable : that is because for a thin slab , the clamped boundary condition at the bottom takes precedence and prevents the apparition of wrinkles .
as soon as the layer s thickness becomes comparable to the wrinkles wavelength ( @xmath71 ) , the value of the critical amount of shear tends rapidly to its value for a semi - infinite solid ( surface instability ) . on figure [ slab ] ( left ) , we study the influence of @xmath1 in the absence of surface energy ( @xmath72 ) . we see that @xmath69 is increased when @xmath73 and decreased when @xmath74 . in other words ,
shearing along the direction of tension requires a greater amount of shear than shearing along the direction of compression , consistent with experimental observations , see figure [ skin2 ] . on figure [ slab ] ( right ) , we investigate the influence of surface tension ( @xmath75 ) in the absence of pre - stretch ( @xmath76 ) .
we see that as @xmath77 increases , @xmath69 also increases , indicating that surface tension makes the layer more stable .
a similar type of behaviour is obtained setting @xmath78 , but it is not shown here for the sake of brevity .
this observation is consistent with the observation that young skin does not wrinkle as early as older skin when sheared , because it is tauter . ) , the layer becomes more ( less ) stable when sheared in the direction of tension ( compression ) . here
right : in the absence of pre - stretch ( @xmath80 ) , the layer becomes more stable in shear when its top surface is endowed with a surface tension .
here @xmath81 , @xmath82 .
the horizontal line indicates the shear threshold of surface instability @xmath83 in an unstretched half - space without surface tension .
, title="fig:",height=188 ] ) , the layer becomes more ( less ) stable when sheared in the direction of tension ( compression ) . here
right : in the absence of pre - stretch ( @xmath80 ) , the layer becomes more stable in shear when its top surface is endowed with a surface tension .
here @xmath81 , @xmath82 .
the horizontal line indicates the shear threshold of surface instability @xmath83 in an unstretched half - space without surface tension .
, title="fig:",height=188 ] owing to the rapid settling of the dispersion curves to their half - space ( biot ) instability limit , we now take @xmath84 in eq . , while keeping @xmath85 , @xmath86 finite .
it then reduces to @xmath87 -\frac{\lambda_1 - 1}{\lambda_1 + 1}2\pi(l_\text{cap}/\ell)\right\}\\ + 2\pi(l_\text{cap}/\ell)(\lambda_1 + 1 ) + ( \lambda^3_1 + \lambda_1 ^ 2 + 3\lambda_1 - 1)=0 .
\label{surf}\end{gathered}\ ] ] to check for consistency , we make the link with known results . for instance , when we neglect the surface energy in eq.([surf ] ) by taking @xmath88
, we recover the surface instability criterion of plane strain @xcite , @xmath89 the corresponding _ shear threshold _ @xmath69 is found by using eq .
, as @xmath90 here the value @xmath83 , obtained in the absence of a pre - stretch ( @xmath76 ) , corresponds to the shear threshold of surface instability for a neo - hookean half - space as found both theoretically by destrade et al.@xcite and experimentally by mora et al.@xcite .
figure [ half ] confirms the trends found from the exact dispersion equation .
hence the left figure ( based on eq . )
shows that the shear threshold @xmath69 is enhanced ( half - space is more stable ) by the presence of a tensile pre - stretch and vice - versa for a compressive pre - stretch ( in tension , @xmath69 eventually reaches a maximum of 5.68 at @xmath91 , but this is way beyond the elastic limit of skin ) .
similarly , the effect of surface tension is to increase the stretchability of the half - space before it becomes unstable , as shown by the right figure ( based on eq . ) .
, the stability under shear is enhanced by a tensile pre - stretch ( @xmath73 and decreased by a compressive pre - stretch ( @xmath74 ) .
the dotted lines correspond to the shear threshold with no pre - stretch ( @xmath80 , @xmath83 ) .
right : the presence of surface tension ( @xmath92 , @xmath82 ) allows the half - space to be sheared further before the instability criterion is met ( @xmath93 , the shear threshold in the absence of surface tension.),title="fig:",height=188 ] , the stability under shear is enhanced by a tensile pre - stretch ( @xmath73 and decreased by a compressive pre - stretch ( @xmath74 ) .
the dotted lines correspond to the shear threshold with no pre - stretch ( @xmath80 , @xmath83 ) .
right : the presence of surface tension ( @xmath92 , @xmath82 ) allows the half - space to be sheared further before the instability criterion is met ( @xmath93 , the shear threshold in the absence of surface tension.),title="fig:",height=188 ] the effect of the surface energy is to fix the wavelength at threshold , as already discussed for the surface instability of compressed skin tissues @xcite . if @xmath94 and @xmath82 , in particular , a logarithmic correction can be calculated by series development of eq.([disp ] ) , reading : @xmath95 , \qquad kh= \tfrac{1}{2 } \log \left[\alpha(\lambda_\text{res } ) \frac{h}{l_\text{cap}}\right],\ ] ] with @xmath96 , @xmath97 .
+ to summarise the results of this section , we have found that the presence of a surface energy always stabilizes the free surface and fixes the morphology of the sheared surface .
the dermis of human skin is characterized by a structural arrangement of elastin and collagen type
i fibres in the extracellular matrix , leading to an anisotropic stiffening of the tissue .
the aim of this section is to investigate how such a material anisotropy affects the stability properties of the sheared skin .
let us consider a single family of fibre reinforcement oriented , in the reference configuration , along the unit vector @xmath98^t$ ] , defining the structural tensor @xmath99 so that @xmath100 represents the fibre stretch , where @xmath101 is the right cauchy - green deformation tensor . in order to build a strain measure for the fibres ,
we introduce the structural invariant @xmath102 , defined as follows : @xmath103:{\bf \widehat m}=(\lambda_\alpha-\lambda_\alpha^{-1})^2\ ] ] as discussed in @xcite , this choice provides a physically consistent deformation measure when @xmath104 _ and _ when @xmath105 , thereby allowing to account both for compression and extension of the fibres . accordingly , the strain energy density of the skin tissue is defined as : @xmath106 where @xmath107 is the anisotropic elastic modulus for the fibre reinforcement . the constitutive relation eq.([enaniso ] ) ensures strong - ellipticity of the tissue in planar deformations , a characteristic which is not met for example for the so - called standard model of fibre reinforcement chosen by @xcite .
it is a simple exercise to show that for a small tensile strain along the direction of the fibres , we have @xmath108 , where @xmath109 for the tensile stretch and @xmath110 for the lateral stretches ; then , the resulting infinitesimal stress is @xmath111 , showing that ( at least in the linear regime ) the ratio @xmath112 is a measure of the stiffness of the fibres compared to the stiffness of the matrix .
we set @xmath80 in this section for the sake of simplicity , so that the half - space is subject to _ simple shear _ only , @xmath113 with deformation gradient and principal stretches @xmath114 respectively .
it is then easy to show that @xmath115 , so that the total strain energy does not depend on fibre orientation , only on the amount of shear @xmath5 .
the corresponding cauchy stress tensor does depend on fibre orientation , as follows @xmath116 let us look for a perturbed surface wave in the form of eq.([sep ] ) ; to do so we need the components of the instantaneous moduli tensor @xmath36 in eq.([s1 ] ) .
for the anisotropic strain energy density @xmath117 defined in eq.([enaniso ] ) and @xmath6 given in eq.([fsimple ] ) we find the components @xmath118 in the coordinate system aligned with the directions of simple shear @xmath119 , @xmath120 , @xmath25 ( see appendix for explicit expressions ) .
then we take the incremental quantities to be of the form @xmath121 where the amplitudes are functions of @xmath122 only . by a well established procedure we can use eqs.([inc1]-[divs ] ) to eliminate @xmath123 and
write the incremental equations as a first - order differential system known as the _ stroh formulation
_ , @xmath124^{\mathop{t } } , \label{stroh}\ ] ] where the prime denotes differentiation with respect to the function argument @xmath122 . here
it turns out that the blocks @xmath125 , @xmath126 and @xmath127 are symmetric ( explicit expressions are given in the appendix ) .
all is in place now for a complete resolution of the surface instability problem .
there exist many strategies for this resolution , see a partial list and references in destrade et al.@xcite . here
we adopted a straightforward approach , because it turned to be tractable numerically .
noticing that the stroh matrix @xmath128 is constant , a solution of the system eq.([stroh ] ) has the form @xmath129^t , \label{sol1}\ ] ] where @xmath130 is a constant vector and @xmath52 are the eigenvalues of @xmath131 . in order for the wrinkles amplitude to decay with depth
, we retain the three roots @xmath132 , @xmath133 , @xmath134 with positive imaginary part ( i.e. @xmath135 ) .
this gives the following general solution , @xmath136 where @xmath137 , @xmath138 , @xmath139 are constants , and @xmath140 , @xmath141 are square ( @xmath142 ) matrices built from the eigenvectors @xmath143 @xmath144 , taken proportional to any column vector of the matrix adjoint to @xmath145 .
now , the traction - free boundary condition at @xmath18 can be written as : @xmath146 where @xmath147 is the surface impedance matrix .
the condition for the onset of a surface instability is thus @xmath148 as we do not know _ a priori _ in which directions the wrinkles are to appear for a given angle @xmath149 of the fibres with respect to the direction of shear , we need to span the entire plane and find the angle @xmath150 for which the corresponding amount of shear @xmath69 is minimal , indicating the earliest onset of wrinkling .
this is the main difference of instability behaviour between an isotropic material ( such as the material in the previous section ) , where the wrinkles appear _ aligned _ with a principal direction of pre - deformation , and an anisotropic material , where the wrinkles may be _ oblique _ with respect to the direction of least stretch . for our simulations ,
we chose material constants such that @xmath151 ( matrix alone ) , @xmath152 ( matrix stiffer than fibres ) , @xmath153 ( matrix as stiff as fibres ) , and @xmath154 ( matrix softer than fibres ) . for each choice of @xmath155 we found @xmath69 and @xmath150 as functions of @xmath149 . varying the angle @xmath149 can be interpreted as either varying orientation of the fibres for a shear that occurs along a fixed axis , such as the @xmath120-axis in eq .
( [ simple ] ) , or as varying the axis along which shear is taking place for a fixed fibre direction @xmath149 .
for illustrative purposes , a typical surface buckling solution is depicted in figure [ fig : buckling3d ] , where we chose @xmath156 ( fibres are twice stiffer than matrix in linear regime ) and @xmath157 ( fibres are originally almost at right angle to the direction of shear ) : there , according to figure [ fig : criticalalpha ] , we have @xmath158 and @xmath159 .
a ) as a function of the reference fibre angle @xmath149 : the presence of fibres clearly leads to earlier surface instability in shear .
( b ) the critical instability angle @xmath150 as a function of the reference fibre angle @xmath149 : these results are harder to interpret because @xmath150 is defined in the current configuration and @xmath149 in the reference configuration .
a remapping of the variations of @xmath69 and @xmath150 with the current fibre angle is shown in figure [ fig : criticalalpha*].,title="fig:",height=207 ] b ) as a function of the reference fibre angle @xmath149 : the presence of fibres clearly leads to earlier surface instability in shear .
( b ) the critical instability angle @xmath150 as a function of the reference fibre angle @xmath149 : these results are harder to interpret because @xmath150 is defined in the current configuration and @xmath149 in the reference configuration .
a remapping of the variations of @xmath69 and @xmath150 with the current fibre angle is shown in figure [ fig : criticalalpha*].,title="fig:",height=207 ] ( fibres are twice stiffer than matrix in linear regime ) and @xmath157 ( fibres are almost at right angle to the direction of shear ) , the first wrinkles appear when the amount of shear reaches @xmath158 , and the corresponding angle of the wrinkles with respect to the direction of shear is @xmath160 .
note the decay of the wrinkles amplitude with depth.,height=283 ] \a ) .
along the wavefront the material is alternatively elongated and compressed , orthogonal to it the material is neither elongated nor compressed .
b ) the fibres are shown by bold black lines.,title="fig:",height=132 ] b ) .
along the wavefront the material is alternatively elongated and compressed , orthogonal to it the material is neither elongated nor compressed .
b ) the fibres are shown by bold black lines.,title="fig:",height=136 ] to set the stage for analyzing the results in figure [ fig : criticalalpha ] , we note that the first wrinkles to appear will occupy the least energy configuration possible while satisfying the zero traction boundary condition .
when the half - space is sheared , line elements are compressed in certain directions and elongated in others .
the effect of superposing a small - amplitude wrinkle is to alternatively elongate and compress the material along the direction of the wrinkle front , i.e. in the direction @xmath49 given by eq .
( [ sji ] ) , see figure [ fig : wrinklefront](a ) .
this alternating behaviour , along with the zero traction boundary condition , makes it difficult to informally comprehend the influence of the wrinkle orientation , however we do notice a pattern . in the neo - hookean isotropic case ( @xmath161 ) ,
the wavefront is along the direction of greatest compression : hence here , wrinkling the material in a direction under compression , due to the shear , allows the material to release some potential energy . in the anisotropic case ( @xmath162 ) ,
the presence of fibres makes the wavefront of the first wrinkle tend towards being orthogonal to the fibres .
for instance , figure [ fig : wrinklefront](b ) depicts the fibre orientation for the solution in the previous figures . clearly
the _ current _ direction of fibres ( in the deformed state of finite simple shear ) , is closely linked to the value to the wavefront orientation @xmath150 . to study this relationship we re - examine the data in figure [ fig : criticalalpha ] by mapping @xmath149 to the fibre orientation in the deformed body @xmath163 , that is the angle between the spatial vector @xmath164 of the current fibre orientation and the @xmath119-axis , from which @xmath150 is also measured .
the results of this remapping are shown in figure [ fig : criticalalpha * ] .
we first turn our attention to the plots of @xmath69 against @xmath163 , see figure [ fig : criticalalpha*](a ) . on the dashed lines , the fibres are neither compressed or stretched . the ( almost straight ) continuous black lines @xmath165 and @xmath166 indicate when the fibres are aligned with the directions of greatest stretch and greatest compression , respectively .
they are given by the equations @xmath167 respectively , where the @xmath168s are given in eq . and evaluated at @xmath169 .
the @xmath165 curve helps us elucidate why there exists a point ( denoted @xmath170 ) where all anisotropic materials become unstable in shear at the same threshold shear @xmath171 as in an isotropic neo - hookean material ( where @xmath161 ) : clearly , this phenomenon occurs when the shear is such that the fibres are aligned with the direction of greatest stretch .
then , it turns out that great simplifications occur in the stroh formulation of the instability problem , and that the buckling criterion coincides with that of the neo - hookean model , see proof in the appendix .
this is an artifact of our specific choice of strain energy density in eq .. in figure [ fig : criticalalpha*](b ) , displaying the plots of @xmath150 against @xmath163 , we drew the line @xmath172 .
clearly , in a region close to @xmath170 , the wavefront is almost aligned with the fibres , as is the case in an isotropic neo - hookean material .
as @xmath173 increases , the neighborhood of this alignment widens , indicating that the stiffer the fibres are , the closer the instability curves in figure [ fig : criticalalpha*](b ) will be to the line @xmath172 and the less the wrinkles will alter the extended fibres .
the overall general conclusion is that stiffer fibres lead to earlier onset of instability ( notwithstanding the punctual fixing of all curves at point @xmath170 , due to the very special case where fibres end up being aligned with the direction of greatest stretch in the deformed configuration . )
this result is in agreement with the casual observation that old skin ( presumably with stiffer collagen bundles ) wrinkles earlier than young skin when pinched .
a ) as a function of the current angle the fibres @xmath163 with respect to the direction of shear .
( b ) the critical instability angle @xmath150 as a function of @xmath163 .
the point @xmath170 indicates a surface instability state common to all materials ( independent of the material parameters).,title="fig : " ] b ) as a function of the current angle the fibres @xmath163 with respect to the direction of shear .
( b ) the critical instability angle @xmath150 as a function of @xmath163 .
the point @xmath170 indicates a surface instability state common to all materials ( independent of the material parameters).,title="fig : " ]
in this work we have investigated the occurrence of shear instability in skin tissue within the framework of nonlinear elastic theories . in section [ section2 ] , we have considered the skin tissue as a neo - hookean layer of finite thickness , whilst the epidermis is modeled _ as a hemitropic film with given surface energy_. moreover , we have taken into account the presence of the cleavage lines of skin as preferred direction of residual stretches inside the tissue . under these assumptions , a linear stability analysis has been performed using the method of incremental elastic deformations , and an analytical form of the dispersion relation has been reported in eq .
( [ disp ] ) .
the results demonstrate that the presence of surface energy makes the layer more stable , in the sense that it needs to be sheared more for wrinkles to develop than when surface energy is absent ( figure [ slab](a ) ) .
furthermore , the surface energy fixes the surface instability wavelength at threshold at a finite value , as depicted in figure [ slab](b ) . we have also found that wrinkles appear earlier when the shear takes place perpendicular to the direction of pre - stretch than when it occurs along that direction , as confirmed by the anectodal evidences shown in figures [ skin1 ] and [ skin2 ] . in section [ shear instability of a fibre - reinforced skin tissue ] , we have investigated the effect of fibre reinforcement in the dermis layer on the shear instability characteristics . for this purpose
, we have used the polyconvex strain energy function in eq.([enaniso ] ) for modeling the transverse isotropic reinforcement along a preferential fibre direction .
a stroh formulation of the incremental elastic equations has been derived in eq .
( [ stroh ] ) , and solved numerically using an iterative technique . as shown in figures [ fig : criticalalpha ] and [ fig : criticalalpha * ] , we have found that the presence of fibres always lowers the shear threshold at which geometrical instability happens : the stiffer the fibres , the earlier the wrinkles appear in shear . considering that anisotropic stiffness of skin greatly
increase with ageing @xcite , our results are in agreement with the fact that older skin wrinkles earlier when pinched .
the presence of a universal point of instability at shear threshold @xmath174 when the fibres are aligned with the direction of greatest stretch @xmath62 , irrespective of the value of @xmath173 , can be observed on figures [ fig : criticalalpha ] and [ fig : criticalalpha * ] .
this anchor point is present for @xmath175 no matter how stiff the fibres are compared to the matrix ( in the appendix we identify its origin . )
however , it represents a very special case of shear , and when we move away from the region of influence of this point , we notice that all bifurcation curves indicate a significant lowering of the shear threshold of instability ( as soon as the fibres become at least as stiff as the matrix , @xmath176 ) . in experimental tests ( see e.g. n annaidh et al.@xcite ) , collagen fibres in human skin are determined to be at least 500 times stiffer than the elastin matrix .
we may thus deduce that our model , away from the anchor point , predicts that surface instability will occur early , at low levels of shear , in line with the visual observations of figures [ skin1 ] and [ skin2 ] . in conclusion
, this mathematical study of wrinkle formation in sheared skin confirms that pinching experiments in dermatology are useful tools to evaluate the local mechanical properties of the tissue . 99 j.c .
waldorf , g. perdikis , and s.p .
terkonda , planning incisions , _ oper .
_ * 4 * ( 2002 ) 199 - 206 .
c.j.kraissl , the selection of appropriate lines for elective surgical incisions .
_ plastic reconstr .
* 8 * ( 1951 ) 1 - 28 .
cox , the cleavage lines of the skin . _
j. surg . _
* 29 * ( 1941 ) 234 - 240 .
steigmann , and r.w .
ogden , plane deformations of elastic solids with intrinsic boundary elasticity , _ proc .
* a453 * ( 1997 ) 853 - 877 .
biot , surface instability of rubber in compression .
sci . research _ * a12 * ( 1963 ) 168 - 182 .
p. ciarletta , i. izzo , s. micera , and f. tendick , stiffening by fibre reinforcement in soft materials : a hyperelastic theory at large strains and its application , _ j. biomech .
behavior biomed .
_ * 4 * ( 2011 ) 1359 - 1368 .
m. destrade , m.d .
gilchrist , d.a .
prikazchikov , and g. saccomandi , surface instability of sheared soft tissues .
_ j. biomech .
_ * 130 * ( 2008 ) 061007 , 1 - 6 .
b. lautrup , _ physics of continuous matter _
( 2nd ed . , crc press , boca raton 2011 ) .
steigmann , and r.w .
ogden , elastic surface - substrate interactions , _ proc .
lond . _ * a455 * ( 1999 ) 437 - 474 .
m. destrade , and r.w .
ogden , surface waves in a stretched and sheared incompressible elastic material , _ int .
j. non - linear mech .
_ * 40 * ( 2005 ) 241 - 253 .
ogden , _ nonlinear elastic deformations _ ( dover , new york 1997 ) .
flavin , surface waves in pre - stressed mooney material , _
q. j. mech .
_ * 16 * ( 1963 ) 441 - 449 .
s. mora , m. abkarian , h. tabuteau , and y. pomeau , surface instability of soft solids under strain .
_ soft matter _ * 7 * ( 2011 ) 10612 - 10619 .
p. ciarletta , and m. ben amar , papillary networks in the dermal - epidermal junction of skin : a biomechanical model .
* 42 * ( 2012 ) 68 - 76 .
agache , c. monneur , j.l .
leveque , and j. de rigal , mechanical properties and young s modulus of human skin in vivo . _ arch .
* 269 * ( 1980 ) 221 - 232 .
for an incompressible anisotropic material with strain energy density @xmath117 given in eq . , there are 31 non - zero instantaneous moduli in the coordinate system aligned with the directions of simple shear @xmath119 , @xmath120 , @xmath25 in eq .. they are found from eq . as follows . @xmath177 the symmetric blocks @xmath125 , @xmath178 , and @xmath127 of the corresponding stroh matrix @xmath128 are given by @xmath179 respectively , with @xmath180 , \notag \\
& \eta= ( 3\mu + l_{1111})\cos^2\theta + 2l_{1121}\cos\theta\sin\theta + l_{2121 } \sin^2\theta , \notag \\ & \kappa= l_{1112 } + ( 3\mu + l_{1221})\cos\theta\sin\theta,\notag \\ & \nu= l_{1212}\cos^2\theta + 2l_{1121}\cos\theta\sin\theta + ( 3\mu + l_{2222})\sin^2\theta , \notag \\ & \chi= \left[\mu \sin 2\alpha + 2 \beta ( \sin 2 \alpha + \sin 2\theta)\right ] k + \left[\mu \cos^2\theta + 2 \beta ( \cos^2\theta - \cos^2 \alpha)\right ] k^2 . \notag\end{aligned}\ ] ] tremendous simplifications occur when @xmath36 and @xmath128 are expressed in the coordinate system aligned with the lagrangian principal axes and the fibres are aligned with the direction of greatest stretch @xmath62 .
then , for wrinkles aligned with that direction , we find that the stroh matrix reads @xmath181 it clearly shows that the incremental deformation in the plane of shear is uncoupled from the out - of - plane component .
further , the in - plane components do not involve @xmath182 and are identical to the components of the stroh matrix for an isotropic neo - hookean material .
it follows that these wrinkles appear at the critical amount of shear @xmath183 found by destrade et al.@xcite , independently of the value of @xmath182 . the corresponding value for the largest stretch is @xmath184 and the angle of the fibres in the reference configuration is @xmath185 rad @xmath186 . | we propose two toy - models to describe , predict , and interpret the wrinkles appearing on the surface of skin when it is sheared . with the first model , we account for the lines of greatest tension present in human skin by subjecting a layer of soft tissue to a pre - stretch , and for the epidermis by endowing one of the layer s faces with a surface tension . for the second model , we consider an anisotropic model for the skin , to reflect the presence of stiff collagen fibres in a softer elastic matrix . in both cases ,
we find an explicit bifurcation criterion , linking geometrical and material parameters to a critical shear deformation accompanied by small static wrinkles , with decaying amplitudes normal to the free surface of skin . |
in recent years , several studies of the probability density function ( pdf ) of the density field in numerical simulations of compressible turbulent flows have been advanced as a first step in its full statistical characterization .
these studies have shown that the density pdf depends on the effective polytropic exponent @xmath5 of the fluid , defined by the expression @xmath6 , where @xmath7 is the pressure and @xmath8 is the gas density .
specifically , for isothermal flows ( @xmath9 ) , the pdf is lognormal ( vzquez - semadeni 1994 ; padoan , nordlund & jones 1997 ; passot & vzquez - semadeni 1998 ; scalo et al .
1998 ; ostriker , gammie & stone 1999 ; ostriker , stone & gammie 2000 ) , while passot & vzquez - semadeni ( 1998 ) noted that , for @xmath10 ( resp .
@xmath11 ) , the pdf develops a power - law tail at high ( resp .
low ) densities ( see also scalo et al . 1998 , nordlund & padoan 1999 , and the review by vzquez - semadeni et al .
additionally , gotoh & kraichnan ( 1993 ) have reported a power - law tail at high densities for burgers flows , and porter , pouquet & woodward ( 1991 ) have reported an exponential behavior for adiabatic flows .
passot & vzquez - semadeni ( 1998 ) explained the lognormal pdf for isothermal flows as a consequence of the central limit theorem ( clt ) acting on the distribution of the logarithm of the density field .
they assumed that a given density distribution is arrived at by a succession of multiplicative density jumps , which are therefore additive in the logarithm .
since for an isothermal flow the speed of sound is spatially uniform , the density jump expected from a shock of a given strength is independent of the local density , and thus all density jumps can be assumed to follow the same distribution ( determined by the distribution of mach numbers , as studied , for example , by smith , mac low & zuev 2000 and smith , mac low & heitsch 2000 ) . finally , at a given position in space , each density jump is independent of the previous and following ones .
therefore , the clt , according to which the distribution of the sum of identically - distributed , independent events approaches a gaussian , can be applied to the logarithm of the density , and the density itself is expected to possess a lognormal pdf . however , the observationally accessible quantity is not the pdf of the mass ( or `` volume '' ) density , but rather that of the _ column density _
, i.e. , the integral ( or sum , for a discrete spatial grid ) of the density along one spatial dimension ( the `` line of sight '' , or los ) .
recently , padoan et al . ( 2000 ) and ostriker et al .
( 2000 , hereafter osg01 ) have also discussed this pdf in three - dimensional ( 3d ) numerical simulations of isothermal compressible mhd turbulence with resolutions up to @xmath12 zones . in particular , osg01 have found that the column density distribution has essentially the same shape as that of the underlying density field ( a lognormal for isothermal gas ) , although with smaller mean and width .
this result is puzzling because , according to the clt , the pdf of column density should approach a gaussian shape , the column density being proportional to the mean density along the los .
osg01 attributed the apparent inapplicability of the clt to the possible presence of intermediate - sized structures in the density field that invalidate the requirement of statistical independence of the individual zones needed for the clt .
in this paper we suggest that density `` events '' along the los can be regarded as independent if they are separated by distances larger than some `` decorrelation length '' , over which the density autocorrelation function ( acf ) has decayed by a large enough factor ( we use a factor of 10 ) .
if the column length is significantly larger than this , then convergence to a gaussian might be expected .
for 3d numerical simulations of magnetic and non - magnetic , isothermal turbulence forced at intermediate - to - large scales ( 1/4 of the box size ) , we find that the acf drops to the 10% level at relatively short separations ( @xmath13% of the box size ) . using random realizations of 3d lognormal fields ,
we show that this convergence is nevertheless very slow because of the large skewness ( asymmetry ) and kurtosis ( wing excess ) of the lognormal density pdf .
then we discuss in a speculative way the factors that may determine the shape of the density acf in molecular clouds , and suggest that its characteristic length may be inferred observationally . in [ sec : num_dat ] we describe the numerical data we use , both from simulations of isothermal compressible turbulence and from random realizations of lognormal fields . in
[ sec : pdfs ] we discuss the acfs and pdfs of the projected density fields and , in particular , the los lengths required for convergence to a gaussian . in
[ sec : discussion ] we discuss the pdf width in simulations and observations , the case of non - isothermal gas , the dependence of the correlation length on physical parameters of the turbulence , and some caveats . finally , in
[ sec : summary ] we summarize our results .
we use two different sets of data for our analysis .
the first comprises two numerical simulations of forced , compressible , isothermal , 3d turbulence , performed at a resolution of @xmath14 grid points , one non - magnetic and one magnetic .
the numerical method is pseudospectral with periodic boundary conditions , employing a combination of eighth - order hyperviscosity and second - order viscosity which allows larger turbulent inertial ranges than can be attained with second - order viscosity only .
a second - order mass diffusion operator is included as well .
we refer the reader to passot , vzquez - semadeni & pouquet ( 1995 ) and vzquez - semadeni , passot & pouquet ( 1996 ) for details . here
we just mention that for both runs the forcing rises as @xmath15 for @xmath16 , and decays as @xmath17 for @xmath18 , where @xmath19 is the wavenumber . for the non - magnetic run
the forcing is 100% compressible and has an amplitude of 25 in code units ; the hyperviscosity coefficient @xmath20 is @xmath21 , the second - order coefficient @xmath22 is @xmath23 , and the mass diffusion coefficient @xmath24 is @xmath25 .
the sound speed is @xmath26 , where @xmath27 is the velocity unit .
the mach number has an rms value @xmath28 , with maximum excursions up to @xmath29 . for the magnetic run
the forcing also peaks at @xmath30 , but is 50% compressible , and has an amplitude of 7.5 in code units ; the diffusive coefficients are @xmath31 , @xmath32 , and @xmath33 .
the sound speed is @xmath34 , giving an rms mach number @xmath35 . a uniform magnetic field
is placed initially along the @xmath36 direction , giving a @xmath37 parameter , defined as the ratio of the mean thermal to magnetic pressures , equal to 0.04 , and an rms alfvnic mach number @xmath38 .
we have chosen this rather strongly magnetized case in order to bring out the effects of the magnetic field clearly .
the differences between the two simulations are due to the fact that the magnetic simulation was not originally intended for the present study , but we do not believe this is a concern for our purposes .
our simulations are only mildly supersonic because of limitations of both the numerical scheme and the computational resources available to us , which constrain the resolution to the value mentioned above .
since at @xmath14 a projection along one axis gives a square of only @xmath39 points , column density pdfs for one single temporal snapshot contain only 10,000 data points , giving relatively poor statistics .
we thus take advantage of the fact that the simulations are statistically stationary ( although the maximum density contrast and rms mach number do fluctuate by about 50% in time ) , and choose to combine several density snapshots to produce a single column density histogram .
specifically , for the non - magnetic run we use 19 subsequent snapshots , spaced an amount @xmath40 code time units ( @xmath41 large - scale turbulent crossing times at the rms speed ) .
for the magnetic run we use 18 snapshots , spaced an amount @xmath42 code units ( @xmath43 large - scale turbulent crossing times ) . in order to overcome the limitations of the numerical simulations
, we consider a second set of data , consisting of simple realizations of random fields with lognormal pdfs , obtained by generating random numbers @xmath44 with a standard gaussian distribution ( zero mean and unit variance ) and defining a new random variable @xmath45 , where @xmath46 is a parameter that controls the width of the lognormal distribution .
we use sequences of these `` density '' values to fill `` cubes '' ( actually parallelepipeds ) with fixed `` plane of the sky '' ( pos ) dimensions @xmath47 and @xmath48 , and `` los '' lengths @xmath49 ranging from a few tens of grid cells to a few thousands .
it is important to note that we have two different sets of `` samples '' in this problem : one is the set of points along the los , whose number is given by @xmath49 ( for simplicity , @xmath49 is measured in grid cells , so that it is numerically equal to the number of contributing cells ) .
the density is effectively_averaged _ along the los .
the other sample is the set of lines of sight in the pos , whose number is given by the product @xmath50 .
this equals the number of data points in the column density pdfs .
we emphasize that the number of points in a pdf is completely independent of the los length @xmath49 , so that we can have pdfs with the same number of data points , but with different values of @xmath49 .
increasing the number of points in the pos allows us to improve the `` signal - to - noise '' ratio for the pdf , especially at the wings .
however , the functional _ shape _ of the pdf is expected to depend only on the number of points in the los . indeed , the column density is equivalent to the _ sample mean _ ( along the los ) in sampling theory , and it is well known that the statistics of the sample mean depend on the sample size ( again , the sample along the los ) . in other words ,
the column density pdfs are histograms of the sample means , of which there is one for each los . to improve the pdf signal - to - noise ratio , we consider many parallelepipeds ( actually , 50 in all cases , each with @xmath51 ) for each set of parameters ( @xmath52 ) , although this is exactly equivalent to having a single larger parallelepiped with 125,000 data points in the `` plane of the sky '' , due to the statistical independence of the data , and we only keep track of the individual parallelepipeds for analogy with the procedure of combining several temporal snapshots used in the case of the numerical simulations .
but in practice , the only relevant datum in this sense is how many data points does each pdf contain , the projected `` shape '' of the parallelepiped on the pos being completely irrelevant ( for example , it may be a square , or a straight line ) .
thus , the total number of grid cells in the larger parallelepipeds ( i.e. , their total volume ) , is 125,000@xmath53 .
we consider two subsets of data , obtained from using two different values of @xmath46 , namely @xmath54 and @xmath55 .
for both the simulation and the random data sets , we first normalize the lognormal density data as required by the clt , by defining a new variable @xmath56 , where @xmath57 is the mean density , @xmath58 is the standard deviation , and @xmath59 counts pixel position along the los . for the random data , the mean and variance of the @xmath8 distribution are related to those of the gaussian variable @xmath60 by @xmath61 and @xmath62 \exp(2\langle x\rangle + \sigma_x^2)$ ] ( see , e.g. , peebles 1987 , app .
f ) . for the simulation data ,
the mean density is 1 , but @xmath58 is not known _ a priori _ , and attempting to measure it gives large errors both because of the relatively high frequency of high - density events and because it is not constant over time .
we find empirically that the necessary values of @xmath58 to bring the column density to near unit variance ( see below ) are approximately 2 and 3 for the non - magnetic and magnetic runs , respectively .
we then project ( sum ) the normalized density along the @xmath63-axis to obtain its associated normalized ( i.e. , of zero - mean and unit - variance ) `` column density '' @xmath64 , defined by ( peebles 1987 , sec .
4.7 ) @xmath65 where the sum extends over all grid cells along the los . in the next section
we discuss the pdfs of @xmath64 .
figure [ fig : rho_pdfs ] shows the underlying density pdfs for the numerical simulation data ( _ left _ ) and for the random lognormal data ( _ right _ ) before normalization .
the density fields are seen to be exactly lognormal in the case of the random data , and approximately so in the simulation data .
the pdf of the non - magnetic simulation exhibits an excess at small densities but , since we will be focusing mostly on the high - density side , and our main conclusions will be drawn from the random data , we do not consider this excess to be a problem .
note also that the non - magnetic run has a wider density pdf even though it has a smaller mean mach number than its magnetic counterpart .
this is probably due to the fact that in the latter the forcing is only 50% compressible , and of smaller amplitude .
the density pdfs for the random data are seen to span dynamic ranges of @xmath1 and @xmath2 for @xmath54 and @xmath55 respectively .
figure [ fig : n_pdfs_num ] shows , in log-@xmath66 , lin-@xmath36 form , the time - integrated ( i.e. , adding several temporal snapshots into the same histogram ) normalized - column density ( @xmath64 ) pdfs for the magnetic and non - magnetic numerical simulations . in this graph format , a gaussian is a parabola , and an exponential is a straight line . in the two runs ,
a nearly exponential decay is apparent at moderately high @xmath64 , although the very - large-@xmath64 tail clearly exhibits an excess from this trend in the non - magnetic case and a defect in the magnetic one .
this may be an effect of the less extended underlying density pdf in the magnetic case . as already pointed out by osg01 with respect to their nearly lognormal column density pdfs
, these results are puzzling : one would expect the @xmath64-pdf to be gaussian , as the column density is essentially a sum ( or equivalently , an average ) of the density events along each los , whose distribution should approach a gaussian by virtue of the clt . as mentioned in the introduction , osg01 interpreted the deviation from gaussianity in terms of a violation of the statistical independence requirement of the clt , due to the existence of intermediate - size correlations in the density field . in order to test this hypothesis , we have computed the acf of the density field in the numerical simulations , at time @xmath67 for the non - magnetic run , and at @xmath68 for the magnetic run ( @xmath69 and 0.51 large - scale turbulent crossing times , respectively ) .
these are shown in fig.[fig : rho_corr ] as a function of spatial separation ( `` lag '' ) in grid cells .
note that we show lags only up to half the simulation size , since the periodic boundary conditions imply that the acf is symmetric about this value .
it is seen that the acf has decreased to half its maximum ( zero - lag ) value at separations of only about 7 cells , and to 10% at lags of only @xmath70 cells .
we can effectively consider the latter to be a `` decorrelation length '' for the simulations .
note that the presence of the magnetic field does not seem to have an important effect on the decorrelation length . for distances significantly larger than this decorrelation length
, the effects of density autocorrelation should be negligible , and the clt should be applicable ( see [ sec : discussion ] for a discussion of possible caveats ) .
we do not choose the more familiar @xmath71 criterion for the decorrelation length because of two reasons .
first , the @xmath71 criterion is only truly meaningful for exponential decay laws , but in general the acf does not decay in this form , and in fact crosses zero at a finite lag in the non - magnetic run .
second , we are interested in lags at which the acf has become effectively negligible compared to its zero - lag value , so that events separated by this length can be assumed to be independent , and a factor of 1/10 seems more appropriate for this purpose than one of @xmath71 . for these reasons ,
we also have chosen to refer to this as a `` decorrelation '' length , rather than a correlation one .
but in any case , this choice is essentially arbitrary . in
what follows , we denote the ratio of the column ( or cloud ) length to the decorrelation length by @xmath0 .
our simulations clearly do not have a large enough number of independent events along the los for the column density pdf to approach a gaussian , as @xmath72 in the simulation box .
we thus have chosen to study this problem using simple random realizations of lognormal fields , sacrificing the realistic hydrodynamic origin of actual density data in favor of the ability to control @xmath0 precisely , and to generate much longer loss than can be attained with even the largest presently available computational resources in numerical hydrodynamical simulations .
this approach has been used in the past for simulating turbulent velocity fields without the numerical expense of actual hydrodynamical simulations ( e.g. , dubinsky , narayan & phillips 1995 ; klessen 2000 ; brunt & heyer 2000 ) .
the main feature that is lost by doing this is the spatial correlation that is inherently present in actual mass density fields , due to the continuum nature of real flows . in any case ,
random , spatially uncorrelated fields should constitute a best - case scenario for studying the convergence to a gaussian pdf .
the presence of correlations of a certain size in grid cells should increase the required path lengths for convergence by a factor equal to this size , making convergence even slower .
for the random lognormal realizations , one decorrelation length can be thought of as a single cell , so that the integration length along the line of sight @xmath49 equals @xmath0 for the random data .
we study the convergence of the pdf to a gaussian as a function of two parameters : the width of the underlying lognormal density pdf , given by @xmath46 , and @xmath49 .
figure [ fig : n_pdfs_b1 ] shows the pdfs of @xmath64 for three realizations with @xmath73 10 , 50 and 500 grid cells .
it can be seen that at @xmath74 , the pdf of @xmath64 appears to decay exponentially for @xmath75 , but develops a concavity at larger @xmath64 . at @xmath76 ,
the high-@xmath64 side of the pdf is almost a perfect exponential , but at @xmath77 no exponential segment is left , and the curve begins to approach a gaussian .
figure [ fig : n_pdfs_b1.5 ] shows a similar sequence as that in fig .
[ fig : n_pdfs_b1 ] , but for @xmath55 . in this figure
we show realizations with @xmath73200 , 500 , 2000 and 4000 . again
a transition from concavity to convexity is seen to occur at high @xmath64 as @xmath49 increases , although in this case , even at @xmath78 an excess is seen at the largest values of @xmath64 , so the convergence is not yet complete at this path length for @xmath55 .
indeed , it is known that for very asymmetric distributions with important wing excesses , the convergence to a gaussian is fastest near the middle of the pdf , and slowest at the tails ( peebles 1987 , sec .
we conclude that , even for completely uncorrelated data , convergence to a gaussian occurs very slowly at the high-@xmath64 tail if the skewness , kurtosis and dynamic range of the underlying density data are large .
also , we can expect that , as more loss are included in the column density pdf , the extreme-@xmath64 tail will reach higher @xmath64 values , and will thus require larger lengths @xmath49 to converge
the convergence studied in the last section refers to completely uncorrelated random data , so that the correlation length is effectively one grid cell . as mentioned in [ sec : pdfs ] , the presence of a finite decorrelation length in the density data should cause the convergence to be even slower with path length , as sufficiently independent `` events '' are expected to be separated by lags of the order of the decorrelation length .
note that , if the decorrelation length is a sizeable fraction of the column ( cloud ) size , i.e. , _
if @xmath0 is not very large , then full convergence to a gaussian is not expected_. clearly , in the case of real molecular clouds , the concept of `` grid cell '' disappears , and the natural unit for measuring the path length should be the decorrelation length itself .
thus , the ratio @xmath0 serves as a measure of the path length . in our simulations , @xmath72 , and , according to the results of [ sec : pdfs ] , this is too small a length to produce a full convergence to a gaussian column density pdf even at moderate underlying density contrasts .
however , in real molecular clouds the actual value of @xmath0 is essentially unknown , and convergence to a gaussian column density pdf is plausible , if @xmath0 is large enough .
thus , it becomes important to assess this possibility .
the value of @xmath0 must be related to some characteristic scale of the _ density - fluctuation _ power spectrum .
if the spectrum has a self - similar ( power law ) dependence on wavenumber over some range , then there are no characteristic scales in this range , and the only natural characteristic scales are those where the power - law range ends at high and low wavenumbers ( analogous to the the `` inner '' and `` outer '' scales of the turbulent energy spectrum ) . in order to investigate this dependence
, we use a spectrum - modifying algorithm introduced by lazarian et al .
( 2001 ) , which allows us to modify the spectrum of any physical field without modifying its spatial distribution .
this amounts to only changing the `` contrast '' of the physical field ( armi & flament 1985 ) .
since the power spectrum only depends on the fourier amplitudes of a field , but not its fourier phases , the modification is accomplished by fourier - transforming the physical field , and then replacing the fourier amplitudes by others that satisfy the desired spectral shape , without altering the phases .
we refer the reader to lazarian et al .
( 2001 ) for details of the algorithm .
we apply the spectrum - modifying algorithm to the density field of the non - magnetic simulation , and impose on it the two spectra shown in fig.[fig : pow_acf]a . in both cases ,
the spectrum rises as @xmath79 for @xmath80 and then decreases as @xmath81 for @xmath82 , where @xmath19 is the wavenumber in units of the inverse box length , and @xmath83 is a `` peak '' wavenumber .
one case has @xmath84 ( _ solid line _ ) and the other has @xmath85 ( _ dotted line _ ) .
these spectra produce the density acfs shown in fig.[fig : pow_acf]b .
it is seen that the 10% level of the acf occurs at a lag @xmath86 , suggesting that the decorrelation length is related to the `` outer scale '' of the density power spectrum . what determines the shape of the density fluctuation power spectrum , and , in consequence , of the density acf in highly compressible turbulence is , to our knowledge , an open problem .
it should most likely be related to the energy spectrum and the forcing ( energy injection ) spectrum , but the actual forms of all these spectra in molecular clouds are unknown .
for example , the temporally and spatially intermittent energy injection in molecular clouds from embedded stellar sources and passing shocks differs significantly from the standard random forcing scheme used in most numerical simulations , which is applied everywhere in space and continuously in time ( see , e.g. , norman & ferrara 1996 and avila - reese & 2001 for related discussions at larger scales in the ism ) . in particular ,
if the energy is injected at small scales , the energy spectrum in a cloud complex may peak at scales quite smaller than the complex s size , and possibly drive the density field to a similar spectral shape . moreover , note
that , if the energy spectrum is dominated by shocks , then its form ( @xmath87 ) is of geometrical , rather than dynamical , origin ( see , e.g. , et al . 2000 ) , and in this case the density power spectrum need not have the same outer scale as the the energy spectrum .
self - gravity may be an important ingredient , too . in summary ,
the actual shapes of all the relevant spectra in molecular clouds , and thus the value of @xmath0 , remain unknown , and deserve to be studied systematically .
observationally , several workers have looked at correlation lengths in molecular gas . in a pioneering study ,
kleiner & dickman ( 1984 ) investigated the acf of column density in the taurus region , and from their plots one infers a correlation length of a few pc .
this is not too short a distance compared to the complex s size , but note , however , that this correlation length refers to the projected intensity data rather than to the underlying 3d density field .
most other observational correlation studies have focused on the acf of the line velocity centroid distribution , and are not directly applicable to our puposes . in any case , they have either reported correlation lengths of fractions of a parsec ( e.g. scalo 1984 ; kleiner & dickman 1985 ) or else find them difficult to determine unambiguously ( e.g. miesch & bally 1995 ) . in this respect , our results suggest that the column density pdf provides us with a means of observationally measuring the ratio of the cloud size to the decorrelation length @xmath0 when optically thin transitions or extinction data are used : the observed column density pdf should transit from a lognormal to an exponential and then on to a gaussian as @xmath0 increases . unfortunately , we do not know the path length a priori , but if it can be estimated by some other means at least in some cases , then the decorrelation length , and consequently the density power spectrum outer scale can be derived .
this suggests that it is necessary to investigate numerically how the decorrelation length depends on parameters of the flow such as the forcing parameters , self - gravity , the energy and magnetic spectra , etc .
another implication of the results from [ sec : pdfs ] is that , at large @xmath0 , the column density dynamic range becomes small .
figure [ fig : n_pdfs_nonstd ] shows the pdfs of the _ mean density _
( i.e. , the _ un - normalized _ column density divided by the path length ) for all loss for the two sets of random density fields .
it is seen that , while the underlying density pdfs discussed here have density contrasts of up to @xmath2 , the column density pdfs typically have dynamic ranges of at most a factor of 20 , and , for very large @xmath0 , of only factors of a few .
this is actually a trivial result , since in the limit @xmath88 , all loss would give exactly the same column density ( i.e. , the sample mean asymptotically approaches the distribution mean ) , and the column density pdf would collapse to a dirac delta function , independently of the dynamic range of the underlying density distribution .
this suggests that , if @xmath0 is large in actual clouds , then nearly constant column densities are expected , but this tells little about the dynamic range of the actual density field . in this case ,
larson s ( 1981 ) density - size relation , @xmath89 , which implies constant column density , could simply be an observational averaging effect along the los ( j.scalo 2000 , private communication ) .
on the other hand , a relatively large observational column density range would point towards relatively small values of @xmath0 .
observational studies of extinction ( e.g. , lada et al . 1994 ; kramer et al .
1998 ; cambrsy 1999 ) typically report extinction ( proportional to column density ) dynamic ranges of about a factor of 10 . comparing with the mean - density pdfs of fig .
[ fig : n_pdfs_nonstd ] , these ranges are consistent with @xmath90 and @xmath91 for underlying density ranges of @xmath1 and @xmath2 , respectively . for comparison , the column densities reported by padoan et al .
( 2000 ) from numerical simulations of mhd turbulence at a resolution of @xmath92 with underlying density fields with a dynamic range of @xmath2 , span 3 orders of magnitude , suggesting that in actual molecular clouds @xmath0 may be significantly larger than in those simulations . on the other hand , osg01 have compared the column density cumulative distribution from their simulations to that of visual extinction in cloud ic5146 ( lada , alves & lada 1999 ) , finding that the overall curve width in both data sets is comparable .
unfortunately , in order to determine whether similar pdf widths imply similar values of @xmath0 , it is necessary to know whether the dynamic ranges of the underlying density distributions are also comparable .
moreover , due to the poorly - sampled nature of the observational data , osg01 had to present the distributions in cumulative form , and in linear plots , rather than semi - logarithmic . in this format , it would be hard to distinguish , for example , between lognormal and exponential pdfs that have similar values at moderate column densities . in any case
, the results are promising , and indicate that properties of molecular cloud turbulence such as the decorrelation length may indeed be determined from the column density pdf and the dynamic range of the density field . finally , note also that our results imply that there should exist a relationship between the functional shape of the column density pdf and its width , i.e. , between its skewness and its variance .
we plan to quantify this relation in future work . in this paper
we have restricted the analysis to lognormal underlying density pdfs , in part for simplicity and in part in order to relate our results on pdfs to those from recent numerical simulations of compressible isothermal mhd turbulence ( e.g. , osg01 , padoan et al .
isothermal flows are normally considered as representative of the flow within molecular clouds .
however , it is possible that molecular is really only close to being isothermal in the density range @xmath93 ( see the discussion by scalo et al .
moreover , diffuse gas in the ism , either neutral or ionized , is in general non - isothermal , and in this case , if the flow behaves approximately barotropically ( @xmath94 , @xmath95 ) , a power - law range is expected to appear in the pdf ( passot & vzquez - semadeni 1998 ) . in this case
the clt does not necessarily apply .
indeed , let us consider a power - law range of the form @xmath96 , where @xmath97 is a constant .
if the range extends to arbitrarily large and/or small values , the variance does not exist , and therefore the clt does not apply . if the power law is truncated at low densities , and @xmath98 , then the column density pdf becomes a gamma distribution ( adams & fatuzzo 1996 ) .
however , if the power - law range has a finite extension , and beyond it the pdf drops rapidly , such as the pdfs reported by scalo et al .
( 1998 ) for non - isothermal numerical simulations of the ism , and by passot & vzquez - semadeni ( 1998 ) for polytropic flows with @xmath95 , then the variance should still exist and the clt should apply .
we expect this to be the case of observational pdfs of diffuse gas .
although the results of this paper are rather straightforward , a number of possible complications should be kept in mind .
first , it is possible that the acf fails to capture long - range correlations because the short - range ones may mask them , as small - scale structures are generally much denser .
so , even in cases where the 10% level of the acf is reached over lags much smaller than the cloud size , it is possible that the flow is not sufficiently decorrelated for the clt to apply ( j. scalo 2000 , private communication )
. numerical simulations of turbulent flows with large values of @xmath0 are necessary to test for this possibility .
second , in cases where the jeans length is close to the system size , self - gravity may promote the formation of large - scale structures , counteracting the possible action of small - scale energy injection sources , and tending to reduce the value of @xmath0 . in this case
, column densities closer to lognormal shapes , and with rather large variances might be expected .
high - resolution numerical experiments with self - gravity and realistic stellar - like forcing , even if just in 2d , similar to those of passot , & pouquet ( 1995 ) or of , ballesteros - paredes & rodrguez ( 1999 ) but with cooling functions appropriate for molecular clouds , may help resolve this issue .
finally , we have suggested that a small column density dynamic range should be taken as an indication of large values of @xmath0 .
unfortunately , small column density dynamic ranges may also arise from limitations in the sensitivity of the observations and saturation effects .
thus , the best suited observations for testing the above results are those in which these limitations are minimized .
our results can be summarized as follows : \1 .
we have proposed that the relevant parameter determining the form of the column density pdf in molecular clouds is the ratio @xmath0 of the cloud size to the decorrelation length of the density field , with the latter operationally defined in this paper as the lag at which the density autocorrelation function ( acf ) has decayed to its 10% level .
assuming that density `` events '' along the los are essentially uncorrelated , large values of this ratio imply that the central limit theorem ( clt ) can be applied to those events , and a gaussian pdf should be expected for large enough values of @xmath0 .
this parameter is essentially the number of independent events ( the `` sample size '' ) along the los , and the column density is equivalent to the `` sample mean '' along the los .
we have measured @xmath0 in two 3d numerical simulations of isothermal turbulence forced at intermediate scales , one magnetic and one non - magnetic . in both cases
we find @xmath99 , suggesting that at least partial convergence to a gaussian pdf should occur .
the column density pdfs for both runs are approximately exponential .
using simple random realizations of uncorrelated , lognormally - distributed fields , we have shown that the pdf of the _ normalized _ ( i.e. , with zero mean and unit variance ) column density @xmath64 indeed converges to a gaussian shape as @xmath0 increases , as dictated by the clt , albeit very slowly , due to the large dynamic range , skewness and kurtosis of the density lognormal distribution . for cases
in which the underlying data have a dynamic range ( `` density contrast '' ) @xmath100 , convergence to a gaussian requires @xmath101 several hundreds .
for density dynamic ranges @xmath102 , the required sample size is several thousand events .
additionally , the width ( variance ) of the column density pdf also decreases as @xmath0 increases , as expected for the distribution of the `` sample mean '' .
specifically , for @xmath90 and an underlying density dynamic range of @xmath1 , the column density dynamic range is @xmath103 , and has decreased to a factor of a few for @xmath101 a few hundred .
we have discussed the turbulent parameters that determine @xmath0 . using a spectrum - modifying algorithm ,
we have shown that the 10%-level decorrelation length appears to be given approximately by @xmath104 , where @xmath83 is the wavenumber at which the density fluctuation power spectrum peaks .
thus , the decorrelation length appears to be very close to the `` outer scale '' of the density power spectrum .
however , we believe that what determines the shape of the density spectrum in molecular cloud turbulence is still an open problem requiring much further work .
we have suggested that the slow convergence of the column density pdf , which transits from lognormal ( or a power - law , if the underlying gas behaves polytropically , but is not isothermal ) , to exponential and on to nearly gaussian shapes as @xmath0 increases , can be used to observationally determine the latter in molecular clouds .
this would provide a direct observational diagnostic of this fundamental property of the turbulence in molecular clouds .
additionally , since the variance of the column density pdf decreases with increasing @xmath0 , a functional relationship between the pdf s variance and skewness is expected to exist .
the decrease of the pdf variance with increasing @xmath0 suggests that , if @xmath0 turns out to be large in real molecular clouds , larson s ( 1981 ) density - size relation , which implies roughly constant column density , could be simply a result of this averaging along the los ( j. scalo , 2000 , private communication ) .
conversely , wide , skewed pdfs may be an indication that the clouds are not very large compared to the turbulent density decorrelation length , and larson s relation might then be the result of limited observational dynamic range ( kegel 1989 ; scalo 1990 ; et al .
1997 ) .
we also discussed briefly the case of power - law underlying density pdfs , expected when the gas is not isothermal . in this case , the clt is only expected to apply if the power laws are truncated at both low and high densities , although the convergence to a gaussian may be even slower if the power - law range is very extended , as power laws have even higher tails than a lognormal distribution .
finally , we have mentioned several possible caveats , specifically : a ) the possibility that the large - scale correlations are masked in the density acf because they involve lower - density structures , b ) the fact that self - gravity may possibly increase the decorrelation length , and c ) the fact that sensitivity and saturation problems with the observations limiting their dynamic range may incorrectly be taken to mean large values of @xmath0 .
we thank laurent cambrsy for discussing his data with us , and eve ostriker , thierry passot , luis rodrguez and john scalo for sharp comments and/or a careful reading of the manuscript . in particular , john scalo provided us with important comments about statistical distributions , limitations of the various statistical methods , and interesting implications of this work .
remarks from eve ostriker helped us uncover a misconception in an earlier version of this paper .
the turbulence simulations were performed on the cray y - mp 4/64 of dgsca , unam .
this work has made extensive use of nasa s astrophysics data system abstract service , and received partial funding from conacyt grant 27752-e to e.v .- s .
99 adams , f. c. & fatuzzo , m. 1996 , apj 464 , 256 armi , l. & flament , p. 1985 , j. geophys .
research , 90 , no .
c6 , 11779 avila - reese , v. & , e. 2001 , apj 553 , in press ( astro - ph/0101397 ) brunt , c. & heyer , m. h. 2000 , apj , in press ( astro - ph/0011200 ) cambrsy , l. 1999 , a&a 345 , 965 dubinsky , j. , narayan , r. & phillips , t. g. 1995 , apj 448 , 226 gotoh , t. & kraichnan , r. h. 1993 , phys .
fluids a 5 , 445 kleiner , s. c. & dickman , r. l. 1984 , apj 286 , 255 kleiner , s. c. & dickman , r. l. 1985 , apj 295 , 466 klessen , r. s. 2000 , apj 535 , 869 kramer , c. , alves , j. , lada , c. , lada , e. , sievers , a. , ungerechts , h. & walmsley , m. 1998 , a&a 329 , l33 lada , c. j. , lada , e. a. , clemens , d. p. & bally , j. 1994 , apj 429 , 694 lada , c.j . , alves , j. & lada , e. a. 1999 , apj , 512 , 250 larson , r. b. 1981 , mnras , 194 , 809 lazarian , a. , pogosyan , d. , , e. & pichardo , b. 2001 , apj , 554 , in press ( astro - ph/0102380 ) miesch , m. s. & bally , j. 1995 , apj 429 , 645 nordlund , & padoan , p. 1999 , in interstellar turbulence , eds . j. franco and a. carramiana ( cambridge : cambridge university press ) , p. 218
norman , c. a. & ferrara , a. 1996 , apj 467 , 280 ostriker , e. c. , gammie , c. f. & stone , j. m. 1999 , apj 513 , 299 ostriker , e. c. , stone , j. m. & gammie , c. f. 2001 , apj , 546 , 980 ( osg01 ) padoan , p. , nordlund , , & jones , b. j. t. 1997 , mnras 288 , 145 padoan , p. , juvela , m. , bally , j. & nordlund , 2000 , apj 529 , 259 passot , t. , vzquez - semadeni , e. & pouquet , a. 1995 , apj 455 , 702 passot , t. & vzquez - semadeni , e. 1998 , phys .
e 58 , no .
4 , 4501 peebles , p. z. , jr .
1987 , probability , random variables , and random signal principles , 2nd ed .
( new york : mcgraw - hill ) porter , d. h. , pouquet , a. & woodward , p. r. 1991 , in large - scale structures in hydrodynamics and theoretical physics , eds .
j. d. fournier and p. l. sulem ( berlin : springer -verlag ) , p. 105 scalo , j. , vzquez - semadeni , e. , chappell , d. & passot , t. , 1998 , apj 504 , 835 smith , m. d. , mac low , m .-
& zuev , j. m. 2000 , a&a 356 , 287 smith , m. d. , mac low , m .-
m . & heitsch , f. 2000 , a&a 362 , 333 stone , j. m. , ostriker , e. c. & gammie , c. f. 1998 , apj 508 , l99 vzquez - semadeni , e. 1994 , apj 423 , 681 vzquez - semadeni , e. , passot , t. & pouquet , a. 1996 , apj 473 , 881 , e. , ballesteros - paredes , j. & rodrguez , l.f .
1997 , apj 474 , 292 vzquez - semadeni , e. , ostriker , e. c. , passot , t. , gammie , c. & stone , j. , 2000 , in protostars & planets iv , eds . v. mannings , a. boss & s. russell ( tucson : univ . of arizona press ) , p. 3 | we discuss the probability distribution function ( pdf ) of column density resulting from density fields with lognormal pdfs , applicable to isothermal gas ( e.g. , probably molecular clouds ) . for magnetic and non - magnetic numerical simulations of compressible , isothermal turbulence forced at intermediate scales ( 1/4 of the box size ) , we find that the autocorrelation function ( acf ) of the density field decays over relatively short distances compared to the simulation size .
we suggest that a `` decorrelation length '' can be defined as the distance over which the density acf has decayed to , for example , 10% of its zero - lag value , so that the density `` events '' along a line of sight can be assumed to be independent over distances larger than this , and the central limit theorem should be applicable .
however , using random realizations of lognormal fields , we show that the convergence to a gaussian is extremely slow in the high - density tail . as a consequence
, the column density pdf is not expected to exhibit a unique functional shape , but to transit instead from a lognormal to a gaussian form as the ratio @xmath0 of the column length to the decorrelation length ( i.e. , the number of independent events in the cloud ) increases .
simultaneously , the pdf s variance decreases . for intermediate values of @xmath0 ,
the column density pdf assumes a nearly exponential decay . for cases with a density contrast of @xmath1 ( resp .
@xmath2 ) , as found in intermediate - resolution simulations , and expected from gmcs to dense molecular cores , the required value of @xmath0 for convergence to a gaussian is at least a few hundred ( resp .
several thousand ) .
we then discuss the density power spectrum and the expected value of @xmath0 in actual molecular clouds , concluding that they are uncertain since they may depend on several physical parameters .
observationally , our results suggest that @xmath0 may be inferred from the shape and width of the column density pdf in optically - thin - line or extinction studies .
our results should also hold for gas with finite - extent power - law underlying density pdfs , which should be characteristic of the diffuse , non - isothermal neutral medium ( temperatures ranging from a few hundred to a few thousand degrees ) .
finally , we note that for @xmath3 , the dynamic range in column density is small ( @xmath4 a factor of 10 ) , but this is only an averaging effect , with no implication on the dynamic range of the underlying density distribution . [
the pdf of column density in molecular clouds ] [ vzquez - semadeni & garca ] |
a deep understanding of the high redshift universe is vital in order to complete our knowledge of galaxy formation , and hence uncover the history of the universe as a whole . with the continued development of instrumentation and of new analysis techniques our knowledge of the high redshift universe is increasing rapidly .
thanks to new deep imaging and multiobject spectroscopy we now have the power to routinely look back at the universe over cosmic time to witness step by step evolution . the result of
this has been a wealth of observations of large samples of galaxies over a large redshift range in various surveys , such as within goods ( @xcite ) , cosmos and z - cosmos ( @xcite and @xcite ) and aegis ( @xcite ) .
these large samples give us the power to achieve statistically meaningful results concerning the evolution of galaxy properties . as a result
a detailed picture is beginning to form regarding when galaxy stellar mass is built up over cosmic time .
various studies ( @xcite , @xcite , @xcite , @xcite , @xcite ) have focused on the change of stellar mass density with time , and have seen generally consistent results .
these studies find that the integrated stellar mass density decreases at higher redshift , as expected since the ongoing process of star formation increases the amount of stellar mass in the universe over time .
these studies also show that roughly 50% of the mass density of the universe is in place by @xmath14 .
this implies prior to this redshift the comoving stellar mass density has a more rapid evolution .
this ties in with studies that show that the star formation rate peak is at @xmath15 ( @xcite , @xcite , @xcite ) .
while studies of the star formation history has been the traditional way to examine and probe galaxy evolution , a great deal of research has been carried out investigating the evolution of galaxies using the galaxy stellar mass function .
early examples of the measurement of the local galaxy mass function at @xmath16 have been carried out by e.g. @xcite and @xcite , using surveys such as 2mass , 2df , and the sloan digital sky survey .
these and other investigations construct the stellar mass function , and integrated stellar mass density within the local universe , and hence provide a vital benchmark for comparison with higher redshifts .
it is important to understand the stellar mass function of the local universe in itself as this traces the integrated star formation and mass assembly history over the entire universe .
however , by extending similar studies to higher redshift we can investigate not only galaxy growth as a function of stellar mass , but also the growth of integrated galaxy stellar mass with time .
that is , we can trace the evolution of stellar mass for galaxies of different stellar masses , i.e. , very massive galaxies vs. lower mass galaxies , over time .
this gives us insights into the growth of galaxies of different stellar masses due to star formation , mergers and other assembly processes .
many studies have investigated the stellar mass function up to @xmath17 using relatively large area surveys ( @xcite , @xcite , @xcite , @xcite @xcite , @xcite ) .
this has been further extended to even higher redshifts using deeper surveys , generally within a much smaller area ( @xcite , @xcite , @xcite , @xcite ) .
these studies of the stellar mass function at high redshifts allows us to form a picture of the high mass end of the stellar mass function as a function of redshift .
it is widely agreed that the dominance of star formation within massive galaxies ends much earlier than within low stellar mass galaxies by @xmath14 ( @xcite ) .
this is one form of galaxy downsizing , whereby the higher stellar mass galaxies have their star formation truncated , or gas depleted , earlier than lower mass galaxies .
this formation scenario was first observed by @xcite and subsequently observed in various studies including e.g. @xcite , @xcite , @xcite , and @xcite .
downsizing is now accepted as part of the formation scenario of galaxies , but is not yet fully understood .
what is also not fully understood is whether the related , but different process of mass downsizing , is occurring in tandem , such that the high mass galaxies form their stellar mass earlier than lower mass galaxies , and when this stellar mass differentiated galaxy formation process first reveals itself . despite the considerable work done investigating the high mass end of the stellar mass function ,
there are still many issues that are not yet fully understood .
firstly , the generally shape of the high redshift stellar mass function is not well described .
nearby stellar mass functions are well fit by a form of the schechter function , and there are various investigations regarding how the parameters of this fit change with redshift .
secondly , difficulties with obtaining deep data at high redshifts mean that the low mass end of the stellar mass function has not been explored as fully , or to as high a redshift .
more recent work has started to uncover a possible steepening with redshift of the low stellar mass end , and a dip " in the intermediate stellar mass range ( e.g. @xcite ) .
it has been suggested that this is a result of evolution of different galaxy populations driven by their mass ( @xcite , @xcite , @xcite , @xcite ) .
the exact nature and reasons behind such features of the stellar mass function are not well understood , and this can only be improved upon with deeper , more robust data .
this will then lead to a better understanding of how the populations of galaxies , as defined by stellar mass , change and evolve over time . in this paper
we use data from the goods nicmos survey ( gns ) to investigate how the stellar mass function evolves from @xmath18 to @xmath19 . by examining the stellar mass functions of galaxies ranging in stellar mass from @xmath3 to as low as @xmath20 we investigate
when , and which galaxies are forming at various epochs in the universe .
the depth of the gns data is such that we are able to probe over a factor of 10@xmath21 the stellar mass evolution up to @xmath22 and importantly trace how galaxies of different masses are evolving with time .
we find throughout this paper a differential in the stellar mass function and how it evolves , revealing a strong stellar mass dependence in the galaxy formation process .
we describe this and give some general explanations for how this differential evolution can occur due to different physical processes .
the paper is set out as follows : section [ sec : gns ] discusses the goods - nicmos survey , the galaxy sample and how the data used in this paper was obtained .
section [ sec : totalmf ] examines the galaxy stellar mass functions of all the galaxies in various redshift bins . in sections [ sec : mfforbr ] and [ sec :
mfforsf ] we split the galaxies into blue and red and star forming and non - star forming respectively .
section [ sec : massden ] describes the calculation of the stellar mass densities for the total sample and for the red and blue populations .
sections [ sec : diss ] and [ sec : summ ] contain the discussion and summary of our findings respectively . throughout this paper
we assume @xmath23 , @xmath24 and @xmath25 km s@xmath26 mpc@xmath26 . ab magnitudes and a salpeter imf are used throughout .
the galaxy sample used in this work is taken from and imaged as part of the goods - nicmos survey ( gns ) ( @xcite ) .
the gns consists of 60 pointings of the hst nicmos-3 camera utilising a total of 180 orbits .
the field of view of nicmos-3 is 51.2 arcsec x 51.2 arcsec with a pixel scale of 0.203 arcsec / pixel .
each nic3 tile was observed in six exposures , which combined give a pixel scale of 0.1 arcsec / pixel with a point spread function ( psf ) of @xmath27 0.3 arcsec full width half maximum ( fwhm ) . within the gns we find 8298 galaxies in the @xmath0 band ( f160w ) , and the 60 pointings
are designed to contain as many massive galaxies ( @xmath28 ) as possible .
these massive galaxies are in the redshift range @xmath29 and at a depth of three orbits .
the galaxies are redshift selected based on their optical to infrared colour described in @xcite .
details on the data reduction pipeline are described in @xcite .
the detections and photometry were done using sextractor ( @xcite ) . at 5@xmath30
the limiting magnitude is @xmath0=26.8 , a marked improvement on the goods ground based data at @xmath3124.5 ( @xcite ) .
further description of the gns , the pointings and the target selection is given in @xcite .
other analysis of the gns data set can be found in @xcite , @xcite , bauer et al .
( 2010 , submitted ) , @xcite , @xcite .
thanks to the large amount of optical data covering the goods fields , the @xmath0 band sources are matched to a catalogue of @xmath32 and @xmath33 band data .
this photometry is avaliable down to a limiting magnitude of [email protected] , and the matching is done within a radius of 2@xmath34 .
however the mean separation between the optical and @xmath0 band coordinates is markedly better than this with @xmath35 @[email protected] " , which is roughly the resolution of nicmos . with this multiband data template spectra
we fit to the @xmath37 photometric data points .
this was handled in two different ways to overcome the degeneracy in colour - redshift space .
the first uses hyperz ( @xcite ) which is the standard @xmath38 minimisation technique .
hyperz uses model spectra which were constructed using the evolutionary codes of @xcite . here
we use five evolutionary types , e , sa , sc , i m and a single burst scenario .
the reddening law is that of @xcite .
the most likely redshift is then computed in the age , metalicity , reddening parameter space , giving the best fit redshift , corresponding probability and several other best fit parameteres .
the second method used to obtain our photometric redshifts is the bayesian approach using bpz .
this is a similar fitting method to hyperz but employs empirical seds rather than model ones . as well as this
, bpz uses the maximum likelihood in the same parameter space as hyperz , but with the addition of empirical information regarding the likelihood of a certain combination of parameters .
this is known as prior information , or priors . here
, we used the distribution of magnitudes of different galaxy types as a function of redshift as the priors ( from hdf - n , @xcite ) . therefore in this case , not only does the code find the best fit redshift solution and spectral type , it also consideres how likely it is to find a galaxy of that type and magnitude at that redshift .
a total of 906 spectroscopic redshifts are avaliable for comparison with our photometric redshifts from the goods - n ( @xcite ) and the goods - s ( @xcite ) fields .
the reliability of photometric redshifts is defined as @xmath39 , the median error ( @xmath40 ) and the r.m.s ( @xmath30 ) are as follows . for hyperz
, the goods - n gives @xmath40 = 0.027 with @xmath30=0.04 , and for the goods - s @xmath40 = 0.043 with @xmath30=0.04 .
for bpz , the goods - n @xmath40 = 0.07 with @xmath30=0.05 and for the goods - s @xmath40 = 0.07 with @xmath30=0.06 . in both codes the extreme catastrophic outliers , as defined by @xmath41 , are around @xmath276% .
the dependence of the redshift on the @xmath0 magnitude shows the reliability of redshift within our selection methods and in this case hyperz shows a slightly better performance ( @xcite ) .
our work shows good agreement with photometric redshifts from past surveys ( e.g. fireworks ) , although our sample is bright , and it is unclear if this would be the case for fainter galaxies .
we further investigate the performance of hyperz at different redshifts , at low redshift ( @xmath6 ) and in the redshift range of @xmath42 , which is the redshift range of the galaxy sample we use in this study .
for the high redshift sample we obtain an average offset @xmath43 and a rms of @xmath44 , with a fraction of catastrophic outliers of @xmath45 . here
catastrophic outliers are defined as galaxies with @xmath46 , which corresponds to @xmath27 3 times the rms scatter .
galaxies below @xmath47 show a slightly lower , but still comparable scatter of @xmath48 , however the outlier fraction decreases dramatically to only @xmath49 .
furthermore , we simulate the effects of the photometric redshift errors on our results throughout this paper .
the stellar masses we use are obtained from the same @xmath37 catalogue as used to measure the photometric redshifts described previously , using a standard multicolour stellar population fitting technique ( e.g. , @xcite and @xcite ) . the photometry is fit to different star formation histories , based on the redshift of the galaxy , with spectroscopic redshifts used when avaliable .
this produces a distribution of likely stellar masses , rest frame optical colours and various other parameters based on a baysian approach .
the model seds are constructed from @xcite models , and the star formation history is characterised by an exponentially declining model with various ages , metalicities and dust extinctions .
the star formation rate is parameterised by an e - folding time and an age such that , @xmath50 the parameters in equation [ eq : sfrexp ] are varied randomly within the ranges ; @xmath51=0.01 to 10 gyrs , t=0 to 10 gyrs , metallicity= 0.0001 to 0.05 and the dust content is parameterised by @xmath52=0.0 , 0.5 , 1 , 2 .
these model seds are then fit to the observed photometric data points using a bayesian approach resulting in a likelihood distribution of stellar mass , age and absolute magnitude for each possible star formation history .
the stellar mass is determined based on this distribution , where the most likely stellar mass produces a peak in the distribution , and the uncertainty is the width .
the final errors produced are a result of the models used and are found to be in the range 0.2 to 0.3 dex . while several of the other parameters produced by this method are not reliable , the stellar masses and colours are robust ( see @xcite and @xcite for further explanation . ) .
it is also possible that the stellar masses are an over estimate due to the poor treatment of the tp - agb phase in a star s life .
the effects of this are much less important at the rest frame wavelengths used in this study , especially in the infrared h - band . by using the newer models of bruzual and charlot ( 2010 ) which include a more proper treatment of tp - agb stars
we find it only lowers the masses in our massive galaxy sample by @xmath530.07 dex .
this effect is much smaller than the stellar mass error and the effects of cosmic variance and it is therefore considered negligible in this work . to obtain star formation rates ( sfrs ) for our gns galaxies we use the rest - frame ultraviolet ( uv ) luminosity , which we derive from observed rest frame optical light following the procedure described in bauer et al .
( 2010 , submitted ) .
the uv luminosity is closely related to the level of ongoing star formation because it is mainly produced by short - lived o and b stars .
an advantage of using uv light to estimate star formation is that it remains largely unaffected by the age - metallicity degeneracy ( @xcite ) , but as is well known , star formation rate estimates derived from uv light are strongly affected by dust extinction .
the 2800@xmath54 rest - frame luminosity is calculated from the observed optical hubble / acs @xmath33-band flux density , which corresponds to rest - frame wavelengths of 3400 - 2125@xmath54 for @xmath55 galaxies .
to derive sfr@xmath56 we apply a simple k - correction derived from the redshift of each object as log(@xmath57 ) ( @xcite , @xcite ) and use the @xcite conversion from 2800@xmath54 luminosity to star formation rate assuming a salpeter imf : @xmath58 before dust extinction is taken into account ( section [ dust ] ) , we find at @xmath47 a limiting sfr@xmath59yr@xmath26 and at @xmath60 , we find a limit of sfr@xmath61yr@xmath26 .
the errors take into account photometric errors and the error in the conversion from a luminosity to a star formation rate ( see also bauer et al .
( 2011 , in prep . ) . to obtain reliable star formation rates from the rest - frame ultraviolet , we need to account for the amount of light obscured by dust , which is a non - trivial problem . @xcite demonstrated a correlation between dust attenuation and rest - frame uv slope , @xmath62 , for a sample of nearby starburst galaxies ( where @xmath63 ) .
updated studies of local galaxies using the _ galaxy evolution explorer _ ( galex ) near - ultraviolet band and
@xmath64 galaxies show that the uv slope from the local starburst relation can be used to recover the dust attenuation of a vast majority of moderately luminous galaxies at @xmath64 ( @xcite , @xcite , @xcite ) .
to determine the uv slope , we use the sed - fitting procedure .
we fit a spectral energy distribution ( sed ) to multi - wavelength observations from optical to mid - infrared , following the procedure described in @xcite and bauer et al .
briefly , the uv through mir seds obtained for all sources in the goods fields were fitted with stellar population synthesis models .
then , the best - fitting templates were used to get synthetic estimations of the uv emission at 1600@xmath54 and 2800@xmath54 . from the synthetic ,
model - derived uv luminosities at 1600@xmath54 and 2800@xmath54 , we calculate the spectral slope , @xmath62 .
we use the @xcite law to derive a@xmath65 from the uv spectral slope , which we then apply to the uv - derived star formation rates . using this method
we find an average extinction value of a@xmath66 1.2 mag .
we find that all galaxies in the gns exhibit a range in sfr@xmath67 between 0.2 @xmath68yr@xmath26 and @xmath272000 @xmath68yr@xmath26 .
a full description of our star formation measure is provided in bauer et al .
.the total galaxy stellar mass functions seen in figure [ totmf ] [ cols= " > , > , > " , ] we integrate over the schechter function to get the integrated stellar mass density between @xmath69 .
this is the total amount of stellar mass contained within galaxies , in a given redshift range , per comoving volume .
the results of this calculation are shown in figure [ smd ] . to calculate this
we perform a numerical integration between the limits of @xmath3 and @xmath70 .
we extended the integration beyond the lower mass limit of our survey by extrapolating the schechter function to masses beyond those which we fit .
our stellar mass densities are generally higher than what has been found in previous work , as shown in figure [ smd ] , due to our steeper values of @xmath71 .
the black dashed dot line shows the stellar mass density history obtained from the integration of the instantaneous cosmic star formation history computed by @xcite , using the same imf as in our work . they show that at @xmath72 the line does not agree well with previous work which inferred from the integration of the stellar mass at these redshifts .
the stellar mass densities that we calculate are higher than what has been found in previous works , but are still slightly systematically lower than the integrated star formation history line .
we also computed the stellar mass density with a bright lower limit of @xmath73 and a faint lower limit of @xmath74 .
the bright limit allows us to compare our computed stellar mass densities with the mass we can actually observe in all mass bins , and the faint mass limit allows a better comparison with the integrated cosmic star formation history .
we find that when we adopt these bright and faint limits we see very little difference between these the computed stellar mass densities . even in the range @xmath75 , where the largest difference is found , the stellar mass density from both the bright and faint limit still lie within the errors of the original stellar mass density .
the stellar mass densities found in this work can be compared to the local value of @xcite .
we find therefore that by @xmath76 , 40@xmath77% of the stellar mass in the local universe has formed and by @xmath78 , 41@xmath79 % has formed .
the growth of stellar mass does not alter dramatically after @xmath17 . at @xmath80 ,
30@xmath77% has formed , and we see that at @xmath81 , 30@xmath82% is in place . the final stellar mass density at @xmath83 is 14 @xmath84% of the local value , although the results for this redshift range are not as robust as the others , as discussed in section [ sec : totalmf ] .
thus , for @xmath85 , we see that the stellar mass density changes only slightly , hence there is very little change in stellar mass over this time period .
we also note the stellar mass forms quickly as roughly one third of the stellar mass is already formed when the universe is only 2.3 gyrs old .
figure [ smdbr ] shows the stellar mass density for the blue and red galaxy populations , with the total stellar mass density shown in black .
we see a large decrease in the mass density for the red galaxies , yet for the blue galaxies the stellar mass density is consistent with being constant over this redshift range .
this is seen in other studies such as @xcite , @xcite and @xcite who combined find very little evolution in the blue stellar mass functions as far out as redshift of @xmath86 . for the red population
the numbers themselves are not as robust due to the poor fitting , but the different evolution of the red galaxies is clear .
this growth in stellar mass density in the red population is consistent with other works such as @xcite and @xcite .
there is a large amount of research exploring the stellar mass function over a large range in redshifts , from @xmath87 .
what is unique about the gns , and this work in particular , is the depth and redshift range provided by the data .
a large number of surveys ( e.g. cosmos , zcosmos and combo-17 ) previously examine the stellar mass function at a lower redshift range than the gns . for @xmath88 these surveys probe a large stellar mass range
e.g. @xcite , with mass limits of @xmath89 at @xmath10 to @xmath90 at @xmath91 and @xcite , with mass limits of @xmath89 at @xmath92 to @xmath93 at @xmath94 . for @xmath95 our stellar mass functions are complete down to @xmath89 , and for @xmath96 we find we are complete to @xmath97
. other work , which probes a similar redshift range , does not probe the faint end of the mass function as low as we do here .
for example @xcite has a mass limit of @xmath97 at @xmath98 and the mass limits of @xcite range from @xmath99 to @xmath90 over @xmath2 .
we are thus able to examine stellar mass evolution at @xmath100 as a function of mass for the first time . in all of the stellar mass functions presented in this work
, we find that the high mass galaxies have formed earlier than the low mass galaxies .
for the total galaxy stellar mass functions , as shown in figure [ totmf ] , galaxies whose stellar mass is @xmath101 are close to the local value as early as @xmath96 ( for @xmath102 the number densities are @xmath103 % of the local value at this redshift ) .
these galaxies have reached the local number density by @xmath95 ( @xmath104% the local value ) .
downsizing conventionally says that the high mass galaxies have stopped star forming before the low mass galaxies .
however what we see here is galaxy mass downsizing " , where by the mass of the high mass galaxy is formed before that in the low mass galaxies .
this is occurring at @xmath105 unlike star formation downsizing which starts at @xmath6 ( bauer et al .
this means the majority of the stellar mass of a galaxy is already in place before the star formation largely stops .
this suggests that it is not the star formation making the galaxy massive , instead it is some other process such as mergers , which we do see happening often at high redshift ( @xcite , @xcite and @xcite ) .
this mass downsizing effect is also seen for the halos of galaxies ( @xcite ) .
figure [ totresid ] also shows the relative ratio between the local galaxy stellar mass function of @xcite at @xmath10 and the number densities calculated in this work .
the green horizonal line is how the ratio would appear if all the number densities were at the local value .
the black line is a linear fit to the points from the lowest to highest ratio , hence its steepness correlates to how dominant downsizing is in that redshift range .
we find that although at @xmath106 the gradient of the line is very steep , there is a generally decreasing trend with lower redshift .
we find similar results for the red and blue mass functions in figure [ brmf ] , with the red population of @xmath107 galaxies being very close to their local number density in the range @xmath75 ( @xmath108% for galaxies in this redshift range ) .
the evolution of the high mass blue galaxies towards the local value is slower than that of the red .
we find that the blue galaxies with stellar mass @xmath101 are mostly present by @xmath95 ( @xmath109% for @xmath101 for galaxies with in this redshift range ) .
this is because these blue galaxies are likely to evolve into red galaxies at some stage .
we also find downsizing in the star forming and non - star forming mass function , where the star forming/ non - star forming populations behave very similarly to the blue / red populations .
figure [ totresid ] shows various lines indicating the ratio between the local stellar mass function and the local stellar mass function with slightly varying parameters ( the coloured and dashed / dotted lines ) . for each line
only one parameter is changed at one time .
we see that between @xmath110 to @xmath111 galaxies in the mass range @xmath112 , have generally higher ratios , hence they have generally lower number densities relative to galaxies with @xmath5 and @xmath113 .
this dip feature also seems to be present for @xmath114 , but is much weaker and shifted to the stellar mass range @xmath115 .
the various coloured lines show the features of the mass function is not just an effect of the changing of the shape of the schechter function .
the dip in the total stellar mass function has been seen before and can be explained by the differential evolution of the blue and red galaxy populations .
this is consistent with studies such as @xcite and @xcite who find a dip in the total stellar mass function that can be fit by the combination of these two population .
we also find this to be the case here , as shown in figure [ totcomp ] .
we find that a double schechter function ( the black dash dotted line ) , computed using the parameters of the blue and red fits , matches the form of the total schechter function for all galaxies .
the only large discrepancy occurs at the high mass end in the range @xmath116 , and this corresponds with a poor fit to the red mass function .
@xcite find that out to a redshift of @xmath94 , a dip in the stellar mass function is observed not only in the total but also in the red and the blue stellar mass functions for @xmath11 galaxies .
they find that the blue stellar mass functions show a dip feature that appears to become more pronounced between @xmath92 and @xmath94 .
the also observe a dip in the red mass function that shows the opposite trend , becoming more pronounced at lower redshift .
the fact that we do not see this feature in either the blue or the red mass functions tentatively suggests that any causes of the dip are perhaps not present beyond @xmath94 for these colour selected populations .
figure [ totparams ] shows a general steepening of @xmath71 from low to high redshift .
it is one possibility that the low stellar mass galaxies are formed early on , then undergo mergers where we know that these merger events happen at an early epoch ( @xcite , @xcite and @xcite ) .
these mergers form higher stellar mass galaxies and hence the the number densities of lower mass galaxies would decrease over time , causing @xmath71 to become less steep , as is observed .
they could also fade , creating a less steep mass function . in this work
we generally find a steeper @xmath71 than previous investigations of the stellar mass function at high redshift .
for example figure [ totparams ] shows we are steeper than both @xcite and @xcite .
we also find a steeper slope than @xcite ( @xmath7 over @xmath13 ) and @xcite ( @xmath118 at @xmath119 then steepens to @xmath120 at @xmath121 ) . in this case , the higher stellar mass densities discussed in section [ sec : massden ] are most likely the result of the steeper slope .
the depth of the gns allows us to probe deeper into the low stellar mass region , allowing us to uncover the steeper slope and higher stellar mass density .
we test this by recalculating stellar mass densities with less steep values for the parameter @xmath71 , but keeping the remaining parameters the same as found in this work .
we find that by increasing @xmath71 by 0.1 in the redshift range @xmath122 there is a @xmath123 decrease in the integrated stellar mass density .
this decrease becomes @xmath124 when alpha is increased by 0.2 in the same redshift range .
@xcite find @xmath125 in the redshift range @xmath122 and hence compute a stellar mass density of @xmath126 .
using this value of @xmath71 we compute a stellar mass density of @xmath127 , which is within error of the value found by @xcite . since we can reproduce the lower stellar mass densities with a shallower value of @xmath71 , our higher stellar mass densities are likely a result of our steeper @xmath71 , and additional mass we see at the low mass end of the stellar mass function .
we find that the steepness , and hence the additional stellar mass we see , is dominated by the low stellar mass blue galaxies we find . in section [ sec
: err ] we test the effect of the errors on our number densities and find we can still reconstruct the schechter parameters well , thus errors calculated in this work are ruled out as being the cause of our steeper @xmath71 . for the blue mass functions in figure [ brmf ] ,
@xmath71 is consistent with being constant across the whole redshift range .
there is also very little evolution of the parameter @xmath128 , and this leads to the relatively unchanging blue stellar mass density seen in figure [ smdbr ] .
this is consistent with @xcite who show that between @xmath92 and @xmath129 both the stellar mass function and the stellar mass density show very little evolution .
this is also in agreement with @xcite who shows the parameters for the blue galaxies are virtually unchanged between @xmath130 and @xmath131 .
this suggests there is some process by which this stellar mass function is being replenished .
we find that for the total mass function , the parameter @xmath117 stays roughly constant over time .
this implies that the process which is increasing the numbers of galaxies at the high mass end , is doing so over the whole range of high masses above and around @xmath117 .
however this is not seen with the blue galaxies , thus this effect is not part of the evolution of this population .
there will be some degeneracy between the schechter function parameters , and hence by holding @xmath117 constant we may be affecting the results of @xmath71 and @xmath132 . to test the validity of the assumption of a constant @xmath117 we use a monte carlo simulation to see how the parameters @xmath71 and @xmath132 vary with @xmath117 . to do this
we took the values of @xmath117 we obtained from the first fitting described in section [ sec : fitting ] and then varied these values between the extremes of the error on @xmath117 .
this gave us a range of @xmath117 and we then repeated the fitting for each value in the range keeping @xmath71 and @xmath132 free , thus showing us how the other parameters vary .
we also recalculated the stellar mass densities using monte carlo schechter parameters , the parameters and stellar mass densities are also plotted in figure [ mcparam ] .
we only apply this analysis to the range @xmath69 as the final redshift bin is mostly excluded from the analysis as described in section [ sec : totalmf ] .
we find that although variations in @xmath117 does produce a spread in the parameter results we still see the same general trends .
we still produce steep values for @xmath71 , which in turn produce stellar mass densities that are higher than found in previous studies .
we plotted histograms of the parameters and stellar mass densities and found that they do peak at the values found from fitting with constant @xmath117 .
we construct galaxy stellar mass functions using 8298 galaxies detected in the goods nicmos survey , a hst @xmath0-band imaging survey centred around the most massive galaxies in the redshift range @xmath133 to @xmath134 .
we use the stellar masses , redshifts , colours and star formation rates calculated from the h - band imaging combined with the acs @xmath135 data to calculate stellar mass functions over this redshift range .
we calculate the mass limit of the survey and find that we are probing galaxy stellar masses down to @xmath136 in the lowest redshift bin , hence we are probing an unexplored region of the stellar mass function whilst maintaining statistically meaningful results ( @xmath27300 galaxies between @xmath96 ) .
we correct for high mass bias , consider both redshift and stellar mass errors , and fit schechter functions to examine the evolution of the parameters .
we also measure the galaxy stellar mass functions for a blue / red selection and one based on star forming / non - star forming galaxy populations .
we also calculate the corresponding total stellar mass densities by integrating over the stellar mass functions at each redshift range .
the major results of this paper are as follows : * we observe stellar mass downsizing in all of our stellar mass functions .
for the total galaxy stellar mass functions we see that galaxies at @xmath5 have reached nearly the local value by @xmath137 . even in the highest redshift bin
it is clear that the most massive galaxies are very close to being in place .
the same is true for the red and blue galaxies .
* at all redshifts the blue galaxies dominate the low mass end of the stellar mass functions . between @xmath69
the lowest stellar mass blue galaxies are close to the local value . *
the total stellar mass function shows a dip feature in the intermediate mass range , which can be explained by the differential red and blue populations evolution .
we see no such dip in the blue and the red stellar mass functions unlike some previous studies who have found this below @xmath138 .
we suggest these features are not present at high redshift . *
we find a generally steeper low mass slope for the total stellar mass functions than previous work due to the low stellar mass blue population that is probed by our deep data set .
this results in a generally higher stellar mass density .
* we find a generally higher stellar mass density , due to probing deeper into the lower stellar masses . by constructing the galaxy stellar mass function we are investigating the mass differentiation in galaxy formation , and the way in which mass drives evolution .
more work must be done to fully understand the similarities and differences between the formation of the high mass galaxies that form first , the intermediate mass galaxies who seem to show a decreased rate of formation , and low mass galaxies who are dominated by the blue population . large surveys such as candels and future spectroscopic surveys that could be performed with telescopes such as jwst and e - elt , will provide the quality of data required to continue advancing in this field .
we would like to thank the gns team for their support and work on the survey and this paper as well as the referee for their helpful comments .
we would also like to acknowledge funding from the stfc . | we use a sample of 8298 galaxies observed as part of the hst @xmath0-band goods nicmos survey ( gns ) to construct the galaxy stellar mass function both as a function of redshift and stellar mass up to @xmath1 .
our mass functions are constructed within the redshift range @xmath2 and consist of galaxies with stellar masses of @xmath3 down to nearly dwarf galaxy masses of @xmath4 in the lowest redshift bin .
we discover that a significant fraction of all massive @xmath5 galaxies are in place up to the highest redshifts we probe , with a decreasing fraction of lower mass galaxies present at all redshifts .
this is an example of ` galaxy mass downsizing ' , and is the result of massive galaxies forming before lower mass ones , and not just simply ending their star formation earlier as in traditional downsizing scenarios , whose effect is seen at @xmath6 . by fitting schechter functions to our mass functions we find that the faint end slope ranges from @xmath7 to @xmath8 , which is significantly steeper than what is found in previous investigations of the mass function at high redshift .
we demonstrate that this steeper mass function better matches the stellar mass added due to star formation , thereby alleviating some of the mismatch between these two measures of the evolution of galaxy mass .
we furthermore examine the stellar mass function divided into blue / red systems , as well as for star forming and non - star forming galaxies . we find a similar mass downsizing present for
both blue / red and star - forming / non - star forming galaxies , and further find that red galaxies dominate at the high mass end of the mass function , but that the low mass galaxies are mostly all blue , and therefore blue galaxies are creating the steep mass functions observed at @xmath9 .
we furthermore show that , although there is a downsizing such that high mass galaxies are nearer their @xmath10 values at high redshift , this turns over at masses @xmath11 , such that the lowest mass galaxies are more common than galaxies at slight higher masses , creating a ` dip ' in the observed galaxy mass function .
we argue that the galaxy assembly process may be driven by different mechanisms at low and high masses , and that the efficiency of the galaxy formation process is lowest at masses @xmath11 at @xmath12 . finally , we calculate the integrated stellar mass density for the total , blue and red populations .
we find the integrated stellar mass density of the total and blue galaxy population is consistent with being constant over @xmath13 , while the red population shows an increase in integrated stellar mass density over the same redshift range .
[ firstpage ] galaxies : evolution galaxies : general galaxies : luminosity function , mass function |
tracking the values of the fundamental constants such as the proton to electron mass ration @xmath0 and the fine structure constant @xmath3 has been shown to be an effective way of discriminating between different mechanisms for the acceleration of the expansion of the universe ( @xcite , @xcite , hereinafter t12 and tmv13 ) .
in particular rolling scalar field cosmologies where the rolling field also couples with the electromagnetic field predict changes in the values of @xmath0 and @xmath3 .
comparison of the predicted changes in @xmath0 with the observed constraints translates to equivalent constraints on the cosmologies and the new physics required by the cosmology .
a new constraint on the change in @xmath0 at a redshift of 0.89 has just been published by @xcite .
they find @xmath8 in the radio absorption spectrum of pks 1830 - 211 at a redshift of 0.88582 .
this more than a factor of six improvement over the previous limit of @xmath9 in the same object @xcite and almost a factor of 4 improvement over the previous lower limit at a redshift of 0.6847 @xcite .
the @xmath10 constraint is based on 4 methanol absorption lines observed with the 100 m effelsberg radio telescope .
two of the absorption lines are normal rotational transitions while the two other transitions are mixed torsion - rotation transitions .
the sensitivity constant @xmath11 , defined by @xmath12 for the two pure rotational transitions is -1 while the two torsion - rotation transitions have sensitivity constants of - 7.4 and -32.8 .
the differences in sensitivity constants provides a means of testing for a change in @xmath0 using transitions in a single molecular species rather than comparison between two or more molecular species .
* the connection between rolling scalar fields and changes in the values of the fundamental constants has been considered for more than 10 years , eg .
@xcite and @xcite . *
it is assumed in most rolling scalar field cosmologies that the rolling scalar field @xmath2 is also coupled with the electromagnetic field @xcite .
this coupling then alters the values of the fundamental constants as the field changes with time ( rolls ) .
this has been discussed by @xcite , hereinafter nl4 in the context of a linear coupling @xmath13 where @xmath14 , @xmath15 is the coupling constant and @xmath16 where @xmath17 is the planck mass .
the same physics applies to both @xmath0 and @xmath3 and the two are related by @xmath18 where @xmath19 is the qcd scale , @xmath20 is the higgs vacuum expectation value and r is a scalar often considered to be on the order of -40 to -50 @xcite .
nl4 further show the relationship between the rolling field and the equation of state parameter @xmath21 as @xmath22 where @xmath23 is the dark energy density . here
@xmath24 and @xmath25 indicate differentiation with respect to cosmic time and to @xmath26 respectively where @xmath27 is the scale factor of the universe . t12 and tmv13 then showed the relation between changing fundamental constants and the evolution of @xmath5 .
@xmath28 for the case of the proton to electron mass ratio equation [ eq - wmu ] establishes the connection between the product of a new physics parameter @xmath29 and a cosmological parameter @xmath30 such that @xmath31 where z is the redshift . * in creating figures [ fig - thmplot ] and
[ fig - small ] equation [ eq - omega ] for the dark energy density is used . in section [ s - examp ] dark energy densities are calculated using equations specific to the particular cosmology .
@xmath32^{-1}\ ] ] * the observational limits on @xmath33 therefore impose limits on the @xmath34 - @xmath30 parameter space that a rolling scalar field cosmology can live in .
this parameter space was first discussed in t12 , however , the new methanol observation has greatly reduced the available parameter space .
figure [ fig - con ] shows the existing constraints on @xmath33 where the most stringent constraint for each object is indicated .
the radio results are @xmath10 constraints while the optical constraints are @xmath35 measurements .
all of the measurements are from the references in table 1 of t12 except for the new pks 1830 - 211 measurement by @xcite .
it is obvious from fig .
[ fig - con ] that the current radio results are the most restrictive in terms of @xmath33 and provide a condition at @xmath36 that must be satisfied by any cosmology
. figure [ fig - thmplot ] shows the effect of the new measurement on the @xmath34 - @xmath30 parameter space along with the other radio restriction at @xmath37 @xcite and the most restrictive optical measurement at @xmath38 @xcite .
the parameter space diagram in fig .
[ fig - thmplot ] is similar to the equivalent diagram in t12 with different constraints .
the shaded areas indicates the parts of the diagram that are forbidden .
all of the area above the bottom line are forbidden at the redshift listed just above the line .
all of the light shaded area and above is forbidden by the constraint on @xmath33 at @xmath39 and all of the shaded areas and above are forbidden by the new @xmath33 measurement at @xmath40 . the dashed line in fig .
[ fig - thmplot ] indicates the expected lower limit on @xmath4 from new physics .
* @xcite use the work of @xcite to set limits on the expected value of the coupling with the fine structure constant @xmath41 of @xmath42 . with the expected value of @xmath43 for r in eq .
[ eq - amu ] this produces the expected range for @xmath44 of @xmath45 . the lower bound from @xcite was determined by setting @xmath46 , about a factor of 100 below the reported value of @xmath47 by @xcite .
* it is always possible to accommodate any value of @xmath30 by lowering the value of @xmath4 , and of course the standard model of physics expects @xmath4 to be zero .
this , however , is a test of non - standard models with rolling scalar fields that do interact with the electromagnetic field .
it is obvious from fig .
[ fig - thmplot ] that if we set @xmath4 at it lowest expected value of @xmath48 then , at a redshift of 0.89 , the value of @xmath5 is very close to @xmath7 . since the value of @xmath49 is the difference between the value of @xmath0 today and its value at @xmath50 this implies that @xmath5 has been very close to @xmath7 for the entire time between now and @xmath50 unless the scalar field deviates from its value at @xmath50 and then returns to that value at the present time .
most rolling scalar field models do not exhibit this behavior so they must conform to having @xmath5 close to @xmath7 for roughly half the age of the universe .
to better quantify the restriction fig [ fig - small ] shows an expanded view of the region near @xmath51 .
figure [ fig - small ] indicates that if we hold to the lowest expected value of @xmath4 we must place the limit of @xmath6 for all redshifts of @xmath52 or less .
this is an extreme restriction that puts the validity of current models of rolling scalar fields * that also couple with the electromagnetic field * as the driving force for the acceleration of the expansion of the universe in doubt .
the results are , however , consistent with a @xmath53cdm universe and the standard model of physics .
some areas for possible modification of the current rolling scalar field models are discussed in section [ s - ar ] .
the @xmath33 constraints at the various redshifts provide parameter space `` wickets '' through which valid cosmologies must pass .
since all of the current `` wickets '' include the space where @xmath54 , @xmath53cdm with the standard model of physics clears all of them as do most modified gravity theories . rolling scalar field models with non - standard physics
do not necessarily pass the test .
tmv13 investigated some quintessence and k - essence models to determine the parameter space that fit the observations , where in this case the parameter space included those individual parameters that define the particular model such as an initial value of @xmath5 at a given value of redshift . in the following the models examined in tmv13
are used with the new restriction and the same individual model parameters with only @xmath4 being varied to achieve a fit . as described in tmv13 the value of @xmath33 at any scale factor @xmath27
is given by @xmath55 with @xmath56 and @xmath57 determined by the details of the particular cosmology .
table 2 of tmv13 lists the values of the parameters used in each of the cosmologies .
note that the matches will be slightly different in the individual cases than in the general case discussed above since in the general case we use equation [ eq - omega ] for @xmath56 whereas in the individual cases we use the more accurate @xmath58^{-1}\ ] ] figure 5 from tmv13 plots the values of @xmath59 versus redshift for the 4 cosmologies , slow roll quintessence , hilltop quintessence , non - minimal quintessence , and k - essence , and is repeated here as fig .
[ fig - w ] for ease of reference .
it is obvious from fig .
[ fig - w ] that none of the three thawing cosmologies come close to satisfying the limit of @xmath6 at a redshift of @xmath52 .
although not obvious from the scale of the plot even the freezing slow roll quintessence cosmology does not satisfy the constraint .
since the constraints are in a parameter space that has both a cosmology and a new physics component , solutions can be sought in either area .
the easiest is to simply adjust the new physics parameter @xmath4 downward , lower than the expected value , to open up the @xmath60 parameter space .
the results of doing this are presented in fig .
[ fig - fit ] .
the required value of @xmath4 needed to make all of the cases fit the constraints is shown in each figure . in some cases , such as k - essence
the required value of @xmath4 is more than a factor of @xmath61 below the expected lower limit .
quantitative and definitive new models for rolling scalar fields are beyond the scope of this paper and will be reserved for future papers .
there are , however , obvious areas for modification which are discussed briefly below . *
the lower bound on @xmath4 as discussed in section [ s - met ] is relatively arbitrary , therefore , the most straight forward method to recover viable rolling scalar field models is to simply reduce the value of @xmath4 by a factor on the order of 10 .
this recovers most of the @xmath30 parameter space appropriate for reasonable cosmological models .
this can be accomplished by either making the two terms in equation [ eq - amu ] roughly equal to each other or by lowering the time variation of each of the terms .
the former is equivalent to reducing the value of r while the latter reduces all of the coupling between the rolling scalar field and the electromagnetic field .
further discussion of this point is beyond the scope of the present paper . * @xcite
use a simple linear form of coupling between the electromagnetic field and the scalar field ( equation [ eq - dmu ] ) .
this is quite reasonable given the lack of information on the true form of the coupling and can be considered as the first term in a taylor series expansion of the actual coupling .
the coupling , however , may be of a quite different form which could lower the effect on the electromagnetic field from the expected range .
the coupling has also been considered to be constant with time which may also be an erroneous assumption .
the new limit on @xmath33 of @xmath62 by @xcite places significant constraints on rolling scalar fields as the origin of the acceleration of the expansion of the universe .
the measured invariance of @xmath0 should be considered in the formulation of cosmologies invoking rolling scalar fields .
current results are consistent with acceleration due to a cosmological constant @xmath53 and the standard model of physics . if the expected limits on the coupling constant @xmath4 are accepted then the invariance at a redshift of 0.89 implies that the value of the equation of state variable @xmath5 has been within @xmath63 of @xmath7 for the last half of the age of the universe . * if , however , the coupling between the rolling scalar field and the electromagnetic field is weaker than expected then the constraints on @xmath5 are significantly relaxed .
this , on the other hand , has implications for new physics . *
more accurate measurements of @xmath0 at higher redshift will be very useful in the further consideration of rolling scalar fields as the origin of the late time acceleration of the expansion of the universe .
the author wishes to thank christian veillet for bringing the paper @xcite to his attention .
the author also thanks carlos martins and paolo molaro for illuminating discussions on the general topic of varying fundamental constants . *
the paper was also improved by helpful comments from an anonymous referee . * 99 avelino , p.p , martins , c.j.a.p . , nunes , n.j . & olive , k.a .
2006 , phys .
d. , 74 , 083508 bagdonaite , j. , jansen , p. , henkel , c. , bethlem , h.l . ,
menten , k.m .
& ubachs , w. 2012 , science , 13 december 2012 ( 10.1126/science.1224898 ) * chiba , t. & kohri , k. 2002 , prog . theor .
phys.,107 , 631 * copeland , e.j . , nunes , n.j . & pospelov , m. 2004 , phys . rev .
d , 69 , 023501 * dvali , g. & zaldarriaga 2002 , phys .
lett . , 88 , 091303 * ellingsen , s.p . , voronkov , m.a . ,
breen , s.l .
& lovell , e.j . 2012 , ap.j.l , 747 , l7 .
kanekar , n. 2011 , ap.j.l .
, 728 , l12 king , j. a. , webb , j. k. , murphy , m. , ubachs , w , & webb , j. 2011 , mnras , 417 , 3010 king , j. a. et al .
2012 , mon . not .
r. astron .
422 , 3370 nunes , n.j . & lidsey , j.e .
2004 , phys rev d , 69 , 123511 ( nl4 ) thompson , r.i . , 2012 mnras letters , 422 , l67 ( t12 ) thompson , r.i . , martins , c.j.a.p . & vielzeuf , p.e .
2013 , 428 , 2232 ( tmv13 ) | new pks1830 - 211 radio frequency observations of methanol at a redshift of 0.88582 have established the most stringent limits on changes in the proton to electron mass ratio @xmath0 to date .
the observations place the limit of @xmath1 which is approximately a factor of four lower than the previous lowest limit at a redshift of 0.6742 .
this stringent limit at a look back time of roughly half the age of the universe has profound implications for rolling scalar field cosmologies and the new physics that they require .
many of these cosmologies invoke a scalar field @xmath2 that is also coupled to the electromagnetic field causing the values of the fundamental constants , @xmath0 and the fine structure constant @xmath3 to roll with time .
if the lowest expected value of the coupling to @xmath0 , @xmath4 , is invoked the new limit requires a limit on the dark energy equation of state parameter @xmath5 such that @xmath6 at a redshift of 0.88582 .
this eliminates almost all of the expected parameter space for such cosmologies and new physics * that have a coupling to the electromagnetic field . in these cases
* the limit requires that @xmath5 must be extremely close to @xmath7 for the last half of the age of the universe or that the coupling of the rolling scalar field to @xmath0 and the electromagnetic field be significantly below or at the limit of its expected range .
the new observations solidify the role of fundamental constants in providing probes of the possible cosmologies and new physics to explain the acceleration of the expansion of the universe .
[ firstpage ] ( cosmology : ) cosmological parameters dark energy theory early universe . |
the analysis on singular manifolds has a long history , and the subject is to a large degree motivated by the study of partial differential equations ( with or without boundary conditions ) and generalizations of index theory to the singular setting , e.g. atiyah - singer type index theorems .
one particular approach is based on the observation first made by a. connes ( cf .
@xcite ) that groupoids are good models for singular spaces .
the outlines for a pseudodifferential calculus on singular foliations were made precise for the general case of ( longitudinally smooth ) groupoids later by b. monthubert , v. nistor , a. weinstein and p. xu ; see e.g. @xcite .
such a calculus can be defined on manifolds with singularities of various types in a unified and general setting .
a notion of global ellipticity with fredholm conditions is then also possible . on the other hand
it is important for applications in the study of pde s to also pose boundary conditions and a parametrix for general boundary value problems .
it has been a hard problem to incorporate boundary conditions in the singular context .
the construction of a refined parametrix , with appropriate fredholm conditions , in such a setting is particularly difficult for technical reasons ( the small calculus is not sufficient ) . in this work
we will enlarge the groupoid pseudodifferential calculus and develop a notion of general pseudodifferential boundary value problems for the groupoid calculus .
the most natural approach seems to be a generalization of the boutet de monvel calculus .
boutet de monvel s calculus ( e.g. @xcite ) was established in 1971 .
this calculus provides a convenient and general tool to study the classical boundary value problems . at the same time parametrices
are contained in the calculus and it is closed under composition of elements .
the elements of boutet de monvel s algebra consist of matrices of operators which act in a suitable sense as pseudodifferential operators but have non - pseudodifferential components .
it is thus our aim in this paper to first describe the calculus in terms of groupoids and groupoid actions on suitable spaces .
we start by recalling a characterization of boutet de monvel operators in terms of conormal distributions due to grubb / hrmander .
then we recover the characterization by letting groupoids act on suitable spaces .
we apply this understanding to the following problem : given a manifold with corners and a lie structure encoding geometric singularities ( the boundary _ at infinity _ ) our objective is to pose boundary conditions on _ regular _ strata which may intersect the _ singular _ boundary at infinity . a convenient way to do
this is to construct a boutet de monvel calculus adapted to this situation .
similar problems have been considered in the literature before .
the closest approach to our problem is a work by t. krainer ( cf .
there he considers from our view a special case of the problem on a manifold with polycylindrical ends and cusp type singularities .
the approach is also considerably different from our own : first a local boutet de monvel calculus is constructed using the collar neighborhood structure of the manifold with corners .
then the calculus is obtained iteratively and closedness under composition is proven via induction on the codimension of the manifold .
the inductive approach is also employed for manifolds with fibered corners in the construction of a pseudodifferential calculus in @xcite .
the inductive step in these proofs depends strongly on the particular type of singularities ( e.g. fibred or generalized cusps etc . ) . therefore this will not be the right approach in the more general setting we are considering here .
in our case we consider the following data : a lie manifold @xmath7 with boundary @xmath6 which is an embedded , transversal hypersurface @xmath8 and which is a lie submanifold of @xmath9 ( cf .
@xcite , @xcite ) .
from @xmath9 we define the double @xmath10 at the hypersurface @xmath6 which is a lie manifold @xmath11 .
transversality of @xmath6 in relation to @xmath0 is briefly described by the following equality @xmath12 for each given hyperface @xmath13 .
introduce the following notation for interior and boundary : by @xmath14 we mean the singular , stratified boundary of the manifold with corners @xmath0 .
@xmath15 the hypersurface @xmath6 is endowed with a lie structure as in @xcite : @xmath16 the goal is to construct a boutet de monvel calculus for general pseudodifferential boundary value problems adapted to this data . the pseudodifferential calculus on a lie manifold was constructed in @xcite via representations of pseudodifferential operators on a lie groupoid .
this representation also yields closedness under composition .
we enlarge the calculus to take boundary conditions into account by first working on the groupoid level . in this approach at first we proceed as usual , i.e. find a lie algebroid @xmath17 such that @xmath18 . via integrating the algebroid @xmath19
we obtain a lie groupoid @xmath20 such that @xmath21 . on the boundary lie structure on @xmath6
we also obtain a groupoid @xmath22 in the same fashion with an associated lie algebroid @xmath23 .
the other essential ingredient in our construction is a _ refined _ blow - up @xmath24 of @xmath25 with regard to the diagonal @xmath26 and @xmath27 of @xmath28 with regard to @xmath26 as manifolds with corners .
the singularities at the ( codimension @xmath29 ) intersection of @xmath6 with the singular boundary of @xmath0 essentially are _ blown - up _ and the boundary problem is posed on the blown - up version of the submanifolds .
we require there to be an isomorphism @xmath30 between these longitudinally smooth spaces @xmath24 and @xmath27 .
since the hypersurface @xmath6 divides the double @xmath10 we denote by @xmath31 the _ right _ half and by @xmath3 the _ left _ half .
these halves have corresponding lie structures and hence corresponding groupoids @xmath32 . on the symbols of pseudodifferential operators from the groupoid calculus we impose a _ fiberwise _ or @xmath33-transmission property with regard to the subgroupoids @xmath34 .
the compatibility requirements we will state particularly imply that @xmath35 have fiberwise boundaries consisting of the fibers @xmath36 for @xmath37 . the boutet de
monvel operators are defined and adapted to data given by the tuple @xmath38 .
this tuple merely depends on the initial lie structure and integrability properties of the corresponding lie algebroids . in this sense
the closedness under composition is ultimately equivalent to the fact that the lie algebroids occuring in the theory are integrable .
the boutet de monvel calculus adapted to this data should then be an algebra @xmath39 ( of order @xmath40 and type @xmath40 ) which is a subalgebra of : @xmath41 the first objective of this work is the proof of the following result .
[ theorem [ thm : bmr1 ] in section [ restricted ] . ] given a lie manifold @xmath7 with embedded hypersurface @xmath8 yielding a lie manifold @xmath9 with boundary @xmath6 such that @xmath10 , the double .
then for a pair of associated groupoids @xmath42 adapted to a boundary structure the equivariant transmission boutet de monvel calculus is closed under composition .
this means that given the order @xmath43 we have @xmath44 [ athm : bm ] in the next section we describe a _ vector - representation _ of our algebra . just as in the case of a pseudodifferential operator on a groupoid
there is a homomorphism which maps @xmath39 to an algebra @xmath45 .
the first algebra on the left consists of equivariant families on a suitable _ boundary structure_. the right hand side is the _
realization_. this latter algebra is defined to consist of matrices of pseudodifferential , trace , potential and singular green operators .
these operators are extensions from the usual operator calculus on the interior manifold with boundary @xmath46 .
hence we want to define a homomorphism @xmath47 of algebras from @xmath48 it is a non - trivial task to prove that in certain particular cases @xmath47 furnishes an isomorphism between these two algebras .
furthermore , as can already be shown by simply viewing the special case of pseudodifferential operators , it is not true in general . instead we prove an analog of a result due to ammann , lauter and nistor ( cf .
@xcite ) .
[ theorem [ thm : bmr2 ] in section [ restricted ] .
] given the vector representation @xmath47 and a boundary structure we have the following isomorphism @xmath49 [ athm : bm2 ] a priori , the inverse of an invertible boutet de monvel operator will not be contained in our calculus due to the definition via compactly supported distributional kernels . we define a completion @xmath50 of the residual boutet de monvel operators with regard to the family of norms of operators @xmath51 on sobolev spaces , cf .
define the completed algebra of boutet de monvel operators as @xmath52 the resulting algebra contains inverses and has favorable algebraic properties , e.g. it is spectrally invariant , cf .
section [ guillemin ] .
we obtain a parametrix construction after defining a notion of _ shapiro - lopatinski ellipticity_. the indicial symbol @xmath53 of an operator @xmath54 on @xmath9 is an operator @xmath55 defined as the restriction to a singular hyperface @xmath56 ( see @xcite ) . note that if @xmath57 intersects the boundary @xmath6 non - trivially we obtain in this way a non - trivial boutet de monvel operator @xmath58 defined on the lie manifold @xmath57 with boundary @xmath59 .
we say that an operator @xmath60 is _
@xmath33-elliptic _ if the principal symbol and the principal boundary symbol of @xmath54 are both pointwise invertible .
if a @xmath33-elliptic operator @xmath54 has pointwise invertible indicial symbols @xmath55 for each singular hyperface @xmath57 we call @xmath54 _
elliptic_. then we prove the following result . _
i ) _ let @xmath60 be @xmath33-elliptic .
there is a parametrix @xmath61 of @xmath54 , in the sense @xmath62 _ ii ) _ let @xmath60 be elliptic .
there is a parametrix @xmath61 of @xmath54 up to compact operators @xmath63 [ athm : parametrix ] in order to limit the size of the paper we restrict most discussions to the order-@xmath40 algebras .
future goals include the realization of this algebra by means of a representation on the @xmath33-sobolev spaces , proving basic continuity properties and studying fredholm conditions .
a further future goal is the proof of a topological index theorem for the adapted boutet de monvel calculus and generalizing the formula given by boutet de monvel in @xcite . at the end of the paper
we make some remarks on this problem .
we also provide a statement of the index problem which is independent of the calculus .
the paper is organized as follows . in section 2
we recall the definition of boutet de monvel s calculus in the standard case and study an example for our construction . in section 3
we recall the general theory of lie groupoids , groupoid actions and lie algebroids .
there we also introduce operators defined via their schwartz kernels and discuss reduced kernels .
then in section 4 we define the general setup for boundary value problems on manifolds with corners and fix the notation .
section 5 is concerned with the notion of boundary structure .
we motivate the definition by considering the special case of @xmath5-vector fields .
we prove that for the special case such a boundary structure or tuple always exists . in section 6 we introduce the extended operators of boutet de monvel type which are special instances of the operators defined in section 3 .
then we show how to compose these operators in section 7 , in section 8 we define the quantization and prove in section 9 the first version of the representation theorem for extended operators .
section 10 is concerned with the truncated boutet de monvel algebra .
the main results include closedness under composition and representation . in section 11
we recall the definition of the guillemin completion of pseudodifferential operators . after introducing order reduction operators into the calculus
, we define the boutet de monvel operators of arbitrary order and type in section 12 . in section 13
we define the guillemin completion of the boutet de monvel calculus .
section 14 is concerned with the construction of parametrices in the lie calculus . in the final section
we discuss index theory .
first and foremost i thank my advisor elmar schrohe for his invaluable support during my work on this project .
part of the research was completed while the author was visiting the university of metz lorraine , ufr mathmatiques .
i m grateful to victor nistor for his friendly hospitality and many helpful discussions .
i thank georges skandalis for helpful discussions during my visit to universit paris diderot .
additionally , magnus goffeng and julie rowlett made several suggestions which led to an improvement of the exposition .
this research is for the most part based on by my phd thesis and was conducted while i was a member of the graduiertenkolleg grk 1463 at leibniz university of hannover .
i thank the deutsche forschungsgemeinschaft ( dfg ) for their financial support .
the calculus of boutet de monvel contains the classical boundary value problems as well as their inverse if it exists and parametrices .
this calculus was developed in 1971 , see @xcite .
let for the moment @xmath9 be a smooth compact manifold with boundary denoted @xmath64 .
elements of the calculus consist of matrices of operators @xmath65 where @xmath66 is a pseudodifferential operator on a smooth neighborhood of the manifold with boundary with transmission property ( i.e. maps functions smooth up to the boundary to functions again smooth up to the boundary ) .
we denote by @xmath67 the operator obtained from @xmath66 via extension by zero @xmath68 of functions defined on @xmath9 to functions on @xmath0 and restriction @xmath69 to @xmath9 .
furthermore , @xmath70 takes the form @xmath71 for a pseudodifferential operator with transmission property @xmath72 and @xmath73 has the form @xmath74 for a pseudodifferential operator with transmission property @xmath75 . here
@xmath76 denotes the restriction to the boundary and @xmath77 is the single layer potential distribution @xmath78 .
additionally , @xmath79 is a pseudodifferential operator on the boundary . * _ features : _ * * if bundles match , i.e. @xmath80 the calculus is * _ closed under composition_*. * if @xmath81 and @xmath82 not present , the * _ classical bvp s _ * are contained in the calculus , e.g. dirichlet problem . *
if @xmath83 and @xmath84 not present , it * _ contains inverses of classical bvp s _ * ( if they exist ) .
it is non - trivial to establish the closedness under composition , which means that two boutet de monvel operators composed are again of this type , compare the standard references @xcite , @xcite . in this section
we will study a shapiro lopatinski elliptic boundary value problem adapted to a lie manifold with boundary .
specifically , we consider the appropriate generalization of the problem discussed in section 2 of @xcite .
we therefore fix the data in the introduction , a lie manifold @xmath7 with boundary @xmath85 and double @xmath11 .
these notions will be defined in the main body of the article .
let smooth hermitian vector bundles @xmath86 , restrictions @xmath87 as well as @xmath88 for @xmath89 be given .
we fix the differential operators on the double lie manifold @xmath90 for @xmath89 and @xmath91 . set @xmath92 for the _ trace operator _ and let @xmath93 .
we consider the following boundary value problem @xmath94 the problem @xmath95 in particular yields a continuous linear operator on the appropriate sobolev spaces @xmath96 assume the boundary value problem @xmath95 is shapiro - lopatinski elliptic ( in the sense of definition [ def : sl ] in section [ parametrices ] ) , then for @xmath97 @xmath96 is fredholm .
there is a parametrix @xmath98 of @xmath99 in the completed lie calculus @xmath100 ( cf .
section [ completed ] ) .
here @xmath101 is a singular green operator of order @xmath102 and @xmath103 are potential operators of order @xmath104 .
[ thm : exa ] we fix the pseudodifferential order reductions ( cf
. section [ completed ] ) @xmath105 in the completed pseudodifferential calculus @xmath106 , @xmath107 .
set @xmath108 and consider for @xmath109 the operator @xmath110 where @xmath111 .
we obtain an elliptic pseudodifferential boundary value problem contained in the completed calculus @xmath112 .
the ellipticity follows by the multiplicativity of the principal and principal boundary symbol in lemma [ lem : mult ] . by theorem
[ thm : parametrix ] in section [ parametrices ] there is a parametrix @xmath113 up to residual terms .
hence @xmath114 yields a parametrix of @xmath99 .
the fredholm property of @xmath115 follows by another application of theorem [ thm : parametrix ] , part ( b ) .
groupoids are small categories in which every morphism is invertible .
[ def : grpd ] first we will introduce and fix the notation for the rest of this paper
. then we will give the definition of a lie groupoid .
for more details on groupoids we refer the reader for example to the book @xcite .
a groupoid will be denoted @xmath116 .
we denote by @xmath117 the set of morphisms and by @xmath118 the set of objects . by a common abuse of notation
we write @xmath119 for @xmath117 .
the set of composable arrows is given as pullback @xmath120 we have the range / source maps @xmath121 such that @xmath122 is @xmath123 also denote the inversion @xmath124 and unit map @xmath125 multiplication is denoted by @xmath126 we also set @xmath127 for the @xmath128 and @xmath129 fibers and their intersection @xmath130 .
the latter is easily checked to be a group for each @xmath131 . [ not : grpd ] * axioms : * one can summarize the maps in a sequence @xmath132^{m } & { \mathcal{g}}\ar@{>->>}[r]^{i } & { \mathcal{g}}\ar@{->>}@ < 2pt>[r]^-{r , s } \ar@{->>}@<-2pt>[r]^ { } & { \mathrm{{\mathcal{g}}^{(0)}}}\ar@{>->}[r]^{u } & { \mathcal{g}}. } \ ] ] with the above notation we can give an alternative way of defining groupoids axiomatically as follows . _
( i ) _ @xmath133 . _
( ii ) _ for each @xmath122 @xmath134 _ ( iii ) _ @xmath135 . _
( iv ) _ for @xmath136 we have @xmath137 _ ( v ) _ for @xmath138 we have @xmath139 _ ( vi ) _ for each @xmath122 we have @xmath140 the @xmath141-tuple @xmath142 defines a _ lie groupoid _ if and only if @xmath116 is a groupoid , @xmath143 are @xmath144-manifolds ( with corners ) , all the maps are @xmath144 and @xmath129 is a submersion .
[ def : liegrpd ] we notice that @xmath128 is automatically a submersion due to the axiom _ iii)_. hence the pullback @xmath145_{p_2 } \ar[r]^{p_1 } & { \mathcal{g}}\ar[d]_{s } \\ { \mathcal{g}}\ar[r]^{r } &
m } \ ] ] exists in the @xmath144-category if @xmath119 is @xmath144 and thus @xmath146 is a smooth manifold as well .
[ rem : liegrpd ] given a lie groupoid @xmath20 we introduce spaces
@xmath24 which are fibred over @xmath118 and such that @xmath119 acts on @xmath24 .
this notion as well as some of the notation is adapted from the paper @xcite .
let @xmath147 be a @xmath119-space , i.e. @xmath148 is a smooth map and @xmath24 is a smooth manifold .
set @xmath149 to be the _
composable elements_. we say that @xmath119 _ acts on _ @xmath24 from the right if the following conditions hold : _ i ) _ for each @xmath150 @xmath151 _ ii ) _ for each @xmath150 and @xmath136 @xmath152 _ iii ) _ for each @xmath150 we have @xmath153 a left action of @xmath119 on a @xmath119-space @xmath154 is a right - action in the opposite category @xmath155 .
[ def : actions ] given two lie groupoids @xmath156 let @xmath154 be an @xmath157-space and @xmath147 be a @xmath119-space .
a left action is denoted by and a right action by additionally , we fix the following notation for the fibers : @xmath158 . [ not : actions ] note that for any @xmath157-space @xmath154 which is additionally _ fibered _ meaning @xmath159 is a surjective submersion then the pullback @xmath160 exists in the @xmath144-category if @xmath161 are @xmath144 .
analogously for a fibered @xmath119-space .
[ rem : actions ] consider the following actions of two lie groupoids @xmath162 : we can define a so - called left haar system on @xmath24 induced by the action of @xmath157 and analogously a right haar system induced by the action of @xmath119 .
this enables us to define left- and right - operators coming from the actions .
let @xmath163 be a haar system induced on @xmath24 by the right action of @xmath119 , see also @xcite , p. 6 .
this is a family of measures such that * the support is @xmath164 for each @xmath131 . *
the map @xmath165 + is @xmath144 .
* we have the invariance condition @xmath166 fix the _ right - multiplication _ for given @xmath122 @xmath167 this is a diffeomorphism .
the induced operators acting on @xmath144-functions are given by @xmath168 these operators @xmath169 yield @xmath170-(anti-)homomorphisms since @xmath171 is the inverse and @xmath172 . _
i ) _ a continuous linear operator @xmath173 is called a _
right @xmath24-operator _ if and only if @xmath174 is a family of continuous linear operators @xmath175 such that @xmath176 this can be expressed alternatively by requiring the following diagram to commute for each @xmath122 @xmath177^{t_{s(\gamma ) } } & c_c^{\infty}({\mathcal{x}}_{s(\gamma ) } ) \ar[d]_{r_{\gamma } } \\
c_c^{\infty}({\mathcal{g}}_{r(\gamma ) } ) \ar[r]^{t_{r(\gamma ) } } \ar[u]_{r_{\gamma^{-1 } } } & c_c^{\infty}({\mathcal{x}}_{r(\gamma ) } ) . } \ ] ] _ ii ) _ by analogy @xmath178 is a _ left @xmath24-operator _ if and only if @xmath179 is a family of continuous linear operators @xmath180 such that the diagram @xmath181^{\tilde{t}^{s(\gamma ) } } & c_c^{\infty}({\mathcal{h}}_{s(\gamma ) } ) \ar[d]_{l_{\gamma^{-1 } } } \\
c_c^{\infty}({\mathcal{x}}^{r(\gamma ) } ) \ar[u]_{l_{\gamma } } \ar[r]^{\tilde{t}^{r(\gamma ) } } & c_c^{\infty}({\mathcal{h}}_{r(\gamma ) } ) } \ ] ] commutes for each @xmath182 where @xmath183 denotes in this case the corresponding left multiplication .
[ xop ] the next proposition tells us that the family of schwartz kernels @xmath184 for a given @xmath24-operator can be replaced by a so - called reduced kernel .
this is not unlike the situation for groupoids and the pseudodifferential calculus where reduced kernels are used extensively ( cf .
@xcite ) . given a right-@xmath24-operator @xmath173 .
then for @xmath185 @xmath186 with @xmath187 depending only on @xmath188 for each @xmath189 .
[ prop : reduced ] first we can write for @xmath190 @xmath191 via the substitution @xmath192 and invariance of haar system . by use of we see that the last integral equals @xmath193 this implies the following identity by the uniqueness of the schwartz kernel for @xmath194 for each @xmath195 @xmath196 to see that @xmath197 is well - defined assume @xmath198 and @xmath199 , then @xmath200 this completes the proof .
the aim of this section is to give a definition of lie algebroids and subalgebroids .
we restrict ourselves to the bare minimum needed in the following text of the paper .
for a more detailed exposition the reader may consult e.g. @xcite . *
a _ lie algebroid _ is a tuple @xmath201 . here
@xmath202 is a vector bundle over a manifold @xmath0 and @xmath203 is a vector bundle map such that @xmath204_{\gamma(e ) } = [ \varrho \circ v , \varrho \circ w]_{\gamma(tm)}\ ] ] + and @xmath205_{\gamma(e ) } = f [ v , w]_{\gamma(e ) } + \varrho(v)(f ) w , \
f \in c^{\infty}(m ) , \ v , w \in \gamma(e).\ ] ] * given two lie algebroids @xmath206 and @xmath207 over the same manifold @xmath0 .
then a _ lie algebroid morphism _ is a map @xmath208 making the following diagram commute @xmath209_{\pi } \ar[ur]^{\varrho } \ar[r(1.8)]^{\varphi } & & \ar[dl]_{\tilde{\pi } } \ar[ul]^{\tilde{\varrho } } \tilde{{\mathcal{a } } } \\ & m & } \ ] ] + and such that @xmath210 preserves lie bracket : @xmath211_{\gamma({\mathcal{a } } ) } = [ \varrho(v ) , \varrho(w)]_{\gamma(tm)}$ ] .
[ def : algbroid ] we briefly summarize some relevant facts about the construction of lie algebroids .
* for any given lie groupoid @xmath119 we obtain an associated algebroid @xmath212 in a covariantly functorial way .
define @xmath213 the _
@xmath129-vertical tangent bundle _ as a sub - bundle of @xmath214 .
denote by @xmath215 the smooth sections and define @xmath216 as the sections @xmath217 such that @xmath218 + we then define the lie - algebroid associated to @xmath119 via the pullback @xmath219 \ar[r]^{u^{\ast } } & t^s { \mathcal{g}}\ar[d]_{\pi_{|s } } \\
m \ar[r]^{u } & { \mathcal{g}}. } \ ] ] + in other words @xmath220 .
* there is a canonical isomorphism of lie - algebras @xmath221 .
the set of smooth sections @xmath222 is a @xmath223-module with the module operation @xmath224 with @xmath225 .
* let @xmath212 be given as above and define @xmath226 by @xmath227 .
then @xmath228 so defined furnishes a lie algebroid . * a lie algebroid is said to be _
integrable _ if we can find an associated ( with connected @xmath129-fibers ) lie groupoid .
not every lie algebroid is integrable @xcite .
in the following we will be concerned with compact manifolds with corners . to this end
we will now fix our terminology and recall the main definitions related to manifolds with corners .
such manifolds are locally modelled on sets of the type @xmath229 .
we give the following _ extrinsic definition _
( it depends on the choice of boundary defining functions ) .
a hausdorff topological space @xmath0 is a _
manifold with embedded corners _ if the following conditions hold . _
i ) _ the space @xmath0 is an embedded submanifold of a smooth manifold ( without corners ) denoted by @xmath230 .
if we denote this embedding by @xmath231 then the smooth functions on @xmath0 are @xmath232 .
_ ii ) _ the boundary defining functions @xmath233 are fixed as maps @xmath234 with @xmath235 _ iii ) _ for each @xmath236 with @xmath237 for each @xmath238 it follows that @xmath239 is linearly independent .
[ def : mwc ] the sets @xmath240 will be called the _ closed _ hyperfaces of @xmath0 . a manifold with corners @xmath0 has a boundary @xmath14 which is stratified by closed ( intersecting ) hyperfaces .
the category of manifolds with boundary is a subcategory of the category of manifolds with corners .
we will not go into further details concerning the categorical structure and the structure preserving maps ( the _ @xmath5-maps _ according to r. melrose ) , but refer instead to the book @xcite , see also @xcite .
additionally , we will be concerned with _ submanifolds _ of manifolds with corners . in this case
given a compact manifold with corners @xmath0 we say that @xmath241 is a _
submanifold with corners _
if @xmath6 is a manifold with corners and each hyperface @xmath57 of @xmath6 is a connected component of a set of the form @xmath242 , where @xmath101 is a hyperface of @xmath0 which intersects @xmath6 transversally .
an important concept is submersions between manifolds with corners which are defined as follows .
a _ submersion _ between two manifolds with corners @xmath0 and @xmath243 is given by a smooth map @xmath244 such that @xmath245 is everywhere surjective and @xmath246 is an inward pointing tangent vector of @xmath0 if and only if @xmath247 is an inward pointing tangent vector of @xmath243 .
[ def : subm ] the following result is relevant in the consideration of fibered spaces such as groupoids that are longitudinally smooth ( meaning the fibers @xmath248 are smooth manifolds without corners )
. let @xmath244 be a submersion between manifolds with corners @xmath0 and @xmath243 .
then for each @xmath249 the fiber @xmath250 are smooth manifolds without corners .
[ lem : subm ] see @xcite , p.4 . for the rest of this paper
we fix the following notation and conventions .
* for any compact manifold with corners @xmath0 we denote by @xmath14 the ( stratified ) boundary at infinity and by @xmath251 the interior . if @xmath241 is a submanifold with corners we denote @xmath252 and @xmath253 .
* as stated in the introduction assume that @xmath6 is transversal to all faces of @xmath0 in the sense of on p. and that @xmath6 is of codimension one .
* then , if @xmath57 is an _ open _
hyperface in @xmath0 we denote by @xmath254 the closure in @xmath0 . *
denote by @xmath255 the _ regular boundary _ of @xmath57 . * if @xmath57 is such that @xmath256 ( we say that @xmath57 is not _ incident _ to @xmath6 ) then the regular boundary is empty . * in addition the notation @xmath257 is the boundary _ at infinity _ of the hyperface , i.e. @xmath258 * denote by @xmath259 the set of open and closed hyperfaces of the manifold with corners @xmath0 , and analogously @xmath260 the open and closed hyperfaces of @xmath6 . * also denote by @xmath261 the _ incident _ faces , i.e. @xmath262 * the boundary defining functions of @xmath6 are commonly denoted by @xmath263 with the index set @xmath264 .
denote by @xmath265 the boundary defining functions of @xmath0 with the index set @xmath266 .
[ not : mwc ] our setup can be put in rather general terms . in this description
we take the algebroid as the fundamental object .
first we define the notion of a lie subalgebroid . given a manifold @xmath0 and a submanifold @xmath267 with algebroid @xmath268 defined over @xmath0 .
then a lie algebroid @xmath269 over @xmath243 is a _ subalgebroid _ of @xmath99 iff @xmath270 is a subbundle equipped with a lie algebroid structure s.t .
the inclusion @xmath271 is a lie algebroid morphism .
[ def : subbroid ] consider the following situation .
we are given a compact manifold with corners @xmath0 and a hypersurface @xmath6 of codimension @xmath272 which is a transversal submanifold of @xmath0 as a manifold with corners .
furthermore , let @xmath268 be a lie algebroid defined on @xmath0 which is assumed to be integrable .
also we assume that to the inclusion @xmath273 there corresponds an inclusion of lie algebroids @xmath274 .
we have then the following well - known result ( see e.g. @xcite , p. 4 ) .
let @xmath275 be a lie algebroid morphism adapted to the data given above .
then @xmath276 is integrable .
[ prop : algbrd ] for the benefit of the reader we briefly sketch the idea of the proof .
first from the integrability of @xmath99 we find and fix a lie groupoid @xmath20 such that @xmath277 . via the inclusion @xmath273 and the range map of @xmath119
we define the pullback @xmath278_{\pi_1 } \ar[r ] & { \mathcal{g}}\ar[d]_{r } \\
y \ar[r]^{i_y } & m } \ ] ] then the anchor map @xmath279 of @xmath99 and the inclusion of lie algebroids @xmath275 defines a foliation @xmath280 .
we can thus consider the monodromy groupoid @xmath281 . since the action of @xmath119 on itself is a principal action it induces an action on @xmath281 .
therefore it makes sense to define the groupoid @xmath282
. then it can be shown that this groupoid integrates @xmath276 .
in fact one can explicitely write down two choices of groupoids integrating @xmath276 in this setup , see also @xcite , thm . 2.3 .
the quotient of the monodromy groupoid @xmath283 and the quotient of the holonomy groupoid @xmath284 the latter is the @xmath129-connected cover of the monodromy groupoid .
there are in general several choices of groupoids integrating @xmath276 .
consider the example of the algebroids @xmath285 . in each case
we could take two different integrating groupoids .
either the pair groupoids @xmath286 or alternatively the _ path groupoids _ ( see @xcite , example 2.9 ) @xmath287 .
this example shows that there is not necessarily any relation between the chosen groupoids @xmath288 integrating the algebroids
. we will address this issue further in the next section .
[ exa : ambi ] the following theorem from @xcite , p. 6 is relevant in this context .
let @xmath20 be a lie groupoid integrating the algebroid @xmath99 over @xmath0 .
assume @xmath22 is a subgroupoid of @xmath289 which integrates the inclusion @xmath290 .
then such a subgroupoid @xmath291 is unique up to isomorphism .
[ thm : moer ] in the following we want to consider a special case of the abstract setup of the last section . a _ lie structure _ @xmath33 on a given compact manifold with corners
@xmath0 is a @xmath223-module of vector fields on @xmath0 which is locally finitely generated , projective and closed under lie bracket .
furthermore , the vector fields in @xmath33 are assumed to be tangent to the hyperfaces of @xmath0 .
[ def : liestruct ] an example of a lie structure on a compact manifold with corners @xmath0 is @xmath292 which consists of all vector fields tangent to the hyperfaces of @xmath0 .
so in particular any lie structure is contained in @xmath293 .
the reader may consult @xcite for more details on lie manifolds .
[ exa : liestruct ] * _ we fix now the following data : _ * * a lie manifold @xmath7 which is made into a lie manifold with boundary in the following sense .
* we are given an embedded , codimension one hypersurface @xmath294 which is a submanifold with corners .
* denote by @xmath10 the double of @xmath9 at the hypersurface @xmath6 which is canonically endowed with a lie structure @xmath295 s.t .
@xmath296 + this notion of double makes sense as defined in @xcite . fix a lie algebroid @xmath297 such that @xmath298 . * the hypersurface @xmath6 is endowed with the lie structure @xmath299 as defined in .
furthermore , fix the vector bundle @xmath300 with @xmath301 . * and @xmath302 being a lie subalgebroid in the sense of def . [ def : subbroid ] . * assumption a. * we assume the algebroid @xmath99 is such that the condition @xmath303 holds . as well as @xmath304 . under isomorphisms with induced anchor map @xmath279 and @xmath305 respectively
i ) _ any lie algebroid associated with a lie manifold is integrable .
this fact can be seen as a consequence of results in the seminal work of crainic and fernandes , see @xcite . another way to see
this is from the assumption a @xmath306 where @xmath307 is dense in @xmath0 .
it was shown in @xcite that with such a condition the algebroid is always integrable .
_ ii ) _ to see that @xmath308 is in fact a lie structure , note that @xmath299 is closed under lie bracket and locally finitely generated , projective . _
iii ) _ the lie submanifold @xmath6 is in particular a submanifold with corners of codimension one . in @xcite it is shown that for such an embedding of lie manifolds the condition @xmath309 holds which reduces to the ordinary transversality condition .
[ rem : liebdy ] we fix the singular normal bundle @xmath310 from the inclusion of lie manifolds @xmath311 as described in the next lemma .
these facts are also given in @xcite and are repeated here for the convenience of the reader . _
i ) _ given a face @xmath57 of @xmath0 such that @xmath312 , @xmath313 is a submanifold with corners and we have @xmath314 computed relative to @xmath57 .
_ ii ) _ if we identify the orthogonal complement @xmath315 in @xmath316 with @xmath317 we obtain the decomposition @xmath318 for each @xmath319 the anchor @xmath320 induces an isomorphism @xmath321 [ lem : lmfbdy ] _ i ) _ this follows from .
_ ii ) _ consider the metric @xmath322 on @xmath0 which is induced from a complete ( compatible in the sense of section [ a2 ] ) riemannian metric on @xmath99 .
as stated above @xmath323 is in particular a sub vector bundle of @xmath316 .
denote by @xmath324 the corresponding quotient mapping .
then we consider the exact sequence @xmath325 & { \mathcal{a}}_{|y } \ar@{->>}[r ] & { \mathcal{a}}_{|y } / { \mathcal{a}}_{\partial } = { \mathcal{n}}. } \label{normal}\end{aligned}\ ] ] the sequence splits as a short exact sequence of vector bundles .
using the riemannian metric we can obtain a splitting as an isomorphism @xmath326 such that @xmath327 ( cf .
@xcite ) . with this isomorphism
the decomposition is then clear .
the second assertion follows immediately from the transversality condition @xmath328 ammann , lauter and nistor ( see @xcite ) have constructed a pseudodifferential calculus @xmath329 adapted to a lie manifold @xmath11 .
this can be viewed as a suitable extension of the enveloping algebra @xmath330 generated by the @xmath295 vector fields .
the calculus is closed under composition .
this is proven by a representation on the corresponding groupoid calculus @xmath331 . here
@xmath119 is some ( @xmath129-connected ) groupoid integrating the lie algebroid @xmath19 .
a pseudodifferential calculus on the lie manifold @xmath332 is thus also defined .
we summarize in the following proposition and give some details for the benefit of the reader .
there is a calculus of pseudodifferential operators @xmath333 defined on the lie manifold @xmath85 such that @xmath333 is a filtered algebra ( it is closed under composition ) .
[ prop : liestruct ] the lie structure @xmath299 is a finitely generated and projective @xmath334-module .
knowing this the vector bundle @xmath23 such that @xmath335 is obtained from the serre - swan theorem .
we also required that @xmath276 is a lie subalgebroid of @xmath316 . either by the integrability of @xmath99 and the general result [ prop : algbrd ] or simply by observing that any lie structure leads to integrable algebroids
( e.g. from @xcite ) it follows that @xmath276 is also integrable . after making use of the pseudodifferential calculus defined on a corresponding groupoid @xmath291 integrating @xmath276 and the representation theorem proven in @xcite
we infer that @xmath333 is closed under composition . _
i ) _ as mentioned in the above proof operators in the pseudodifferential lie calculus are representations of operators in the corresponding groupoid calculus .
this is proven in @xcite , thm .
3.2 and we use the notation @xmath336 for representations corresponding to the groupoids @xmath288 respectively , where @xmath119 and @xmath291 are fixed groupoids integrating the algebroids @xmath337 . then the representation theorem is the statement @xmath338 we will discuss representation theory including the definition of the maps @xmath336 further in section [ repr ] . _
ii ) _ it can be checked easily from the definition that @xmath299 is closed under lie bracket .
to be precise @xmath299 lies inside a quotient of @xmath295 .
first we have the exact sequence @xmath339 & 2 { \mathcal{v}}\ar@{->>}[r ] & 2 { \mathcal{v}}/ 2 { \mathcal{v}}i_y } \ ] ] where @xmath340 is the ideal of functions vanishing on @xmath6 .
we can apply the acyclic functor @xmath341 ( associating to a smooth vector bundle the module of vector fields ) to the exact sequence .
note that @xmath342 which yields the two exact sequences @xmath343 \\
i_y 2{\mathcal{v}}\ar@{>->}[r ] & 2{\mathcal{v}}\ar@{->>}[r ] & 2{\mathcal{v}}_y = 2{\mathcal{v}}/ 2{\mathcal{v}}i_y \ar@{->>}[d ] \\ & & \gamma({\mathcal{n } } ) . }
\ ] ] [ rem : liestruct ]
with the given lie manifold with boundary we want to associate a so - called _ boundary structure_. to motivate the definition we start with the special case of @xmath5-vector fields .
then we define a boundary structure and verify that at least for @xmath5-vector fields a boundary structure always exists .
what is necessary in the general case is a certain assumption on the groupoids @xmath344 , namely they ought to define a bimodule structure which we are going to specify . in the case of @xmath5-vector fields we verify with certain restrictive conditions on @xmath0 the groupoids are morita equivalent .
the boundary structure is in fact a good analogy for blow - ups of the corners which are the intersections of @xmath6 with the ( singular ) boundary at infinity of @xmath0 .
these blow - ups are in our setup canonically defined in terms of @xmath119 and @xmath291 , the groupoids integrating @xmath99 and @xmath276 . therefore our notion of _ boundary structure _ requires no further assumptions .
it only depends on the lie structure itself and the assumption that it leads to algebroids which are integrable via compatible groupoids .
we recall first the definition of the b - groupoid from @xcite ( see also @xcite ) . for this consider the case @xmath345 where @xmath293 denotes the module of vector fields which are tangent to all hyperfaces of @xmath9 .
fix boundary defining functions @xmath346 of @xmath0 , then the groupoid is defined as . to avoid overuse of the symbol @xmath341 we use this notation instead . ]
@xmath347 the composition and inverse is defined as follows : @xmath348 range and source maps are given by @xmath349 we define the _ b - groupoid _ @xmath350 as the union of the connected components containing the unit of each @xmath129-fiber of @xmath351 .
[ def : bgrpd ] _ i ) _ the groupoid of the boundary @xmath352 is defined analogously . here
we fix the boundary defining functions @xmath263 of @xmath6 .
we have @xmath353 and @xmath354 is defined as the @xmath129-connected envelope of @xmath355 which is the union of the @xmath129-connected components . , the @xmath129-connected envelope @xmath356 is also described as the smallest subgroupoid of @xmath119 containing the units @xmath118 . ] _ ii ) _ another way to define the @xmath5-groupoids is to use so - called _ decoupages _ as introduced by monthubert . for this
we consider the embedding of the manifold with corners @xmath0 into a smooth manifold ( without corners ) @xmath230 . inside this manifold @xmath230
there is a family of submanifolds @xmath357 .
each @xmath358 divides @xmath230 into two connected components
. the restrictions of the @xmath358 correspond to faces in @xmath0 . in @xcite
a smooth groupoid @xmath359 is defined such that @xmath360 corresponding decoupages for @xmath6 and @xmath9 are easily obtained defining the groupoids @xmath361 and @xmath354 .
they are given as follows @xmath362 these groupoids are therefore closed subgroupoids of @xmath363 .
[ rem : decoup ] in what follows we assume a decoupage @xmath364 is fixed such that @xmath350 and @xmath354 are the corresponding @xmath5-groupoids .
we define the spaces @xmath365 and @xmath366 .
additionally , on @xmath24 and @xmath27 we fix the canonical groupoid bimodule structure from the actions of the @xmath5-groupoids @xmath350 and @xmath354 .
[ def : spaces ] the canonical action of the isotropy group @xmath367 , for a given @xmath368 , on @xmath369 and @xmath370 is free and proper .
[ lem : isotropy ] consider the case of the right - action of @xmath341 on @xmath24 for some fixed @xmath368 .
the assertion for @xmath371 follows analogously .
if @xmath372 for a @xmath373 it follows from the definition of the composition in @xmath350 that @xmath322 is the identity .
therefore the action of @xmath341 is free .
secondly , note that the ( right or left ) action of @xmath341 on @xmath119 is proper . from the definition of @xmath24
, this @xmath341-action restricts to the right action of @xmath341 on @xmath24 .
we have a proper mapping @xmath374 , @xmath375 .
hence we see that the right action of @xmath341 on @xmath24 is proper .
it is important to note that the fibers of the manifolds with corners @xmath24 and @xmath371 are smooth ( no corners ) .
the spaces @xmath24 and @xmath27 are longitudinally smooth .
[ lem : lsmooth ] the assertion follows from the longitudinal smoothness of the groupoid @xmath119 where @xmath376 are surjective submersions in the sense of @xcite .
we have the actions since @xmath377 are restrictions of @xmath378 we obtain that @xmath24 has smooth fibers by the local triviality property @xcite of the groupoid @xmath119 .
the same reasoning applies to @xmath27 .
the strongest possible relation between the groupoids @xmath350 and @xmath354 is that of isomorphism in the category of lie groupoids .
an isomorphism in this category is given by ( smooth ) morita equivalence ( cf .
@xcite ) . in order to prove the next result we consider the following _ model case _ : @xmath0 is such that @xmath6 intersects every hyperface of @xmath0 in exactly one codimension two face .
_ consequence : _ note that in particular every hyperface of @xmath0 is incident , i.e. @xmath379 . in the model case
we have the equality @xmath380 . since @xmath6 is transversal and of codimension one to each face of @xmath6 there is exactly one incident hyperface in @xmath0 so @xmath381 . if @xmath6 intersects every hyperface of @xmath0 exactly once we have @xmath382 .
since every hyperface of @xmath6 arises from an intersection of @xmath6 with @xmath0 , there will be the same number of boundary defining functions for @xmath0 and @xmath6 .
we can therefore write in this case @xmath383 and @xmath263 for the boundary defining functions of @xmath0 and @xmath6 respectively .
now we can prove the following theorem . in the model case
there is a morita equivalence @xmath384 .
[ thm : mequ ] recall @xmath385 , then since @xmath386 we obtain a morita equivalence as follows .
we have the canonical left and right actions @xmath387{-90}{$\circlearrowleft$}\ { \mathcal{x}}\ \rotatebox[origin = c]{90}{$\circlearrowright$}\ { \mathcal{g}}(m)$ ] and @xmath388{-90}{$\circlearrowleft$}\ { \mathcal{x}}^t \ \rotatebox[origin = c]{90}{$\circlearrowright$}\ { \mathcal{g}}(y)$ ] .
it is immediate to see that the actions of @xmath354 and @xmath350 commute , i.e. @xmath389 for each @xmath390 and @xmath391 .
secondly , @xmath392 is in bijective correspondence with @xmath393 and @xmath394 is similarly bijective to @xmath395 . for this
we first show that @xmath159 induces the homeomorphism @xmath396 .
we show that we have @xmath397 for @xmath398 if and only if there is a ( necessarily unique ) @xmath399 such that @xmath400 .
so let @xmath401 be such that @xmath397 with @xmath402 .
we set @xmath403 and we only need to verify that @xmath404 is actually contained in @xmath350 . in order to see this note that we have two sequences @xmath405 and @xmath406 . where we have convergence in local charts of @xmath407 such that @xmath408 hence in particular we have @xmath409 such that @xmath410 similarly , one shows that @xmath411 induces a homeomorphism @xmath412 , i.e. @xmath413 if and only if there is a @xmath414 such that @xmath415 .
finally , note that @xmath119 acts on itself freely and properly .
in particular the right action on @xmath416 is free and proper . in the following
we give the axioms necessary to define a boundary structure .
at first we state a standing assumption .
this assumption will get rid of the ambiguity in choice mentioned in example [ exa : ambi ] .
* assumption b. * there is a decoupage @xmath364 such that @xmath417 and a canonical bimodule structure @xmath418 .
the groupoids @xmath42 are such that @xmath419 . a _ boundary structure _
is defined as a tuple @xmath38 consisting of a lie groupoid @xmath20 and two manifolds ( possibly with corners ) @xmath420 which are diffeomorphic via a _ flip diffeomorphism _ @xmath30 and subgroupoids @xmath32 of @xmath119 .
we impose the following axioms on this data : _ i ) _ @xmath21 as well as @xmath421 as lie algebroids .
_ ii ) _ @xmath24 and @xmath27 are @xmath119- and @xmath291-spaces each .
we have smooth maps @xmath422 and @xmath423 such that @xmath159 and @xmath424 are surjective submersions . _
iii ) _ restricted to the interior we have @xmath425 _ iv ) _ the fibers of @xmath426 are the interiors of smooth manifolds with boundary , namely : @xmath427 [ def : bdystruct ] in the following discussion our goal will be to verify the properties of a boundary structure for the case of @xmath5-vector fields .
this includes the verification of assumption b in this particular case .
they also carry over easily to the case of the @xmath428 vector fields ( generalized cusps ) as defined in @xcite . in the general case for arbitrary @xmath33 it is possible to find such a boundary structure with assumption b.
first the groupoids @xmath426 exist by integrability of the algebroid corresponding to the lie structure @xmath33 .
one can check that the actions induce longitudinally smooth manifolds @xmath420 via the general method in @xcite as in the proof of lemma [ lem : lsmooth ] . while axiom _
iv ) _ is more difficult to see in general .
it can be seen for the special case of @xmath5-groupoids in the following examples .
the essential point is the application of the definition of a lie manifold with boundary .
[ rem : bdystruct ] _ i ) _ consider a compact manifold @xmath9 with boundary @xmath64 and interior @xmath429 .
then we also fix the double @xmath10 . in this ( trivial ) case
the spaces are given by @xmath430 with the flip @xmath431 .
we have here the pair groupoids @xmath432 as well as @xmath433
. then @xmath434 are just the projections @xmath435 .
_ ii ) _ consider the example of a manifold with corners @xmath9 and the @xmath5-type vector fields @xmath345 for the lie structure and consider a regular embedded codimension one submanifold @xmath8 which is transversal in the sense of , p. .
we can consider the double @xmath10 as a lie manifold with lie structure @xmath295 . in this case since @xmath436 is a manifold with corners the pair groupoid @xmath436 is no longer fiberwise smooth .
instead we use the groupoids as given by monthubert ( cf .
@xcite , see also @xcite ) and defined in the last section .
we can obtain a boundary structure for @xmath5-type vector fields by the next result .
[ exa : bdystruct ] for a lie manifold with boundary @xmath7 where @xmath345 , the b - vector fields , there is an adapted boundary structure .
[ thm : bdystruct ] denote by @xmath437 the collections of open faces of @xmath0 and @xmath9 respectively .
introduce the following notation : @xmath438 where @xmath439 denotes the codimension of a point @xmath440 in @xmath0 .
the groupoids and spaces that we defined in definition [ def : bgrpd ] can be written _ as sets _ as follows @xmath441 for the following argument we recall the notation introduced in the subsection [ mwc ] .
we have the fixed boundary defining functions for the hyperfaces of @xmath0 and denote this family by @xmath442 . on @xmath6 there are the boundary defining functions ( relative to @xmath6 ) , denoted by @xmath263 for some index set @xmath264 .
these are the boundary defining functions of the faces from the intersections of @xmath6 with the strata of @xmath0 .
consider now the topology of @xmath24 and @xmath27 .
it is defined in local charts by the rule @xmath443 on the interior we only have the pair groupoids .
this yields the trivial actions these actions can be extended continuously to the closure of @xmath444 in @xmath119 and also of @xmath445 , and we obtain from the definition of the topology that @xmath446 the continued actions are defined axiom _ i ) _ holds because of @xcite where it was shown that the given @xmath5-groupoids integrate the lie structure of @xmath5-vector fields . then _
ii ) _ was verified in lem .
[ lem : lsmooth ] and _ iii ) _ follows from the definition of the action we just gave .
also note that the flip diffeomorphism @xmath447 is defined by @xmath448 it remains to verify condition _
iv)_. for this we define @xmath449 and prove that this groupoid has the required property . thus we want to show that @xmath450 the boundary is possibly empty ( for @xmath440 not incident to the hypersurface @xmath6 ) .
we have to distinguish two cases : @xmath440 in the interior and @xmath440 on the boundary of @xmath0 .
the groupoid fiber @xmath451 for @xmath440 in the interior @xmath307 trivializes to the pair groupoids and this case is thus immediate .
we need to consider the case of a point on the boundary of @xmath0 .
assume that @xmath452 for some open face @xmath57 of @xmath0 which is incident to @xmath6 ( i.e. shares a hyperface with @xmath6 ) . by the local triviality property of groupoids ( see @xcite )
we have @xmath453 the same follows by definition for @xmath24 , i.e. @xmath454 where @xmath455 denotes the face of @xmath57 such that @xmath456 . via the definition of the lie manifold with boundary ( cf .
@xcite ) we obtain that the component @xmath457 is the interior of a manifold with boundary .
in particular we see that @xmath458 in summary , we obtain @xmath459
the next goal is to define potential , trace and singular green operators on the groupoid level
. these operators should be equivariant families of operators on the fibers , similar to the case of pseudodifferential operators on groupoids .
the singular green , trace and potential operators are ordinarily defined so as to act like pseudodifferential operators in the cotangent direction and as convolution operators in the normal direction .
this somewhat complicated behaviour is difficult to realize in the groupoid setting .
we start from a different but equivalent definition .
our approach is inspired by the ordinary case of a smooth , compact manifold with boundary as explained in the appendix . here
the trace , potential and singular green operators are extended to the double of the manifold and can be understood as conormal distributions with rapid decay along the normal direction . in our general
setting we would therefore like to consider conormal distributions on @xmath25 and @xmath28 as well as @xmath436 .
since we are working in the setting of manifolds with corners we will desingularize these manifolds and pull back the integral kernels to the desingularized versions .
this is were the previously introduced notion of a boundary structure enters . for the cases of @xmath436 and @xmath460
this is realized through the groupoids @xmath119 and @xmath291 respectively and the pseudodifferential operators on groupoids .
we introduce additional blowups @xmath24 and @xmath27 with good properties ( fibered over the manifolds @xmath6 and @xmath0 ) with regard to @xmath119 and @xmath291 .
then we define the trace , potential and singular green operators as distributions on these spaces and @xmath119 conormal to the diagonal @xmath26 . *
_ from now on we fix : _ * a boundary structure @xmath461 satisfying assumption b adapted to our lie manifold @xmath462 with boundary @xmath6 and its double @xmath463 .
we then fix the groupoid actions which are summarized in the following picture . in the first column
the dotted arrows indicate morphisms in the category of lie groupoids ( correspondences ) which are implemented by the actions in the second column .
fix also haar systems on the groupoids and fibered spaces as follows .
@xmath464 in each case the system is a ( left / right)-haar system if the corresponding action is a ( left / right)-action . in order to define the operators on groupoids and actions
as given in the last section we have to introduce the local charts .
the charts are given by diffeomorphisms which preserve the @xmath129-fibers , see also @xcite , p. 3 .
fix the dimensions @xmath465 . *
a chart of @xmath119 is an open subset @xmath466 which is diffeomorphic to two open subsets of @xmath467 .
choose two open subsets @xmath468 and @xmath469 .
then choose two diffeomorphisms @xmath470 and @xmath471 .
additionally , we require that these diffeomorphisms are _ fiber - preserving _ in the sense that @xmath472 for @xmath473 and @xmath474 for @xmath475 .
hence we have the following commuting diagrams : @xmath476 & \ar[l]_-{\psi_r } \ar[ld ] \omega \ar[rd ] \ar[r]^-{\psi_s } & s(\omega ) \times w_s \ar[rd ] & \\ v_r & \ar[l ] r(\omega ) & & s(\omega ) \ar[r ] & v_s } \end{aligned}\ ] ] * similarly , the charts for @xmath24 are given by the sets of the form @xmath477 for charts @xmath478 of @xmath119 fitting into the following commuting diagrams : @xmath479 & \ar[l ] \ar[ld ] \tilde{\omega } \ar[rd ] \ar[r ] & v_{p^t } \times w_{p^t } \ar[rd ] & \\ w_{q } & \ar[l ] q(\tilde{\omega } ) & & p^t(\tilde{\omega } ) \ar[r ] & v_{p^t } } \end{aligned}\ ] ] * analogously , @xmath480 are charts with the actions reversed and hence in this case we have the commuting diagrams : @xmath481 & \ar[l ] \ar[ld ] \tilde{\omega } \ar[rd ] \ar[r ] & v_{q^t } \times w_{q^t } \ar[rd ] & \\ w_{p } & \ar[l ] p(\tilde{\omega } ) & & q^t(\tilde{\omega } ) \ar[r ] & v_{q^t } } \end{aligned}\ ] ] _ i ) _ a family @xmath482 of operators @xmath175 is a _
differentiable family of trace type _
iff the following holds . given any chart @xmath466 with fiber preserving diffeomorphism , @xmath483 for some @xmath484 open .
moreover for @xmath485 such that @xmath486 we have a fiber - preserving diffeomorphism @xmath487 , and for each @xmath488 the operator @xmath489 has a schwartz kernel @xmath490 the operator @xmath491 for each @xmath492 corresponds to the schwartz kernel @xmath493 via the diffeomorphisms @xmath494 and @xmath495 . _ ii ) _ analogously , we define a family @xmath496 of operators @xmath497 with the charts reversed .
this is called _ differentiable family of potential type_. _ iii ) _ a _ differentiable family of singular green type _
@xmath498 is a family of operators @xmath499 defined as follows .
given any chart @xmath466 with fiber preserving diffeomorphism @xmath483 for some @xmath484 open and @xmath500 .
then for each @xmath501 the operator @xmath502 has a schwartz kernel @xmath503 furthermore , @xmath504 for each @xmath492 corresponds to the schwartz kernel @xmath493 via the diffeomorphism @xmath495 .
[ def : operators ] fix the following operations @xmath505 whenever defined .
a trace type family @xmath73 has a family of schwartz kernels @xmath506 .
define the support of @xmath73 as @xmath507 the _ reduced support _ of @xmath73 is written @xmath508 the analogous definitions for potential type operators @xmath70 and green type operators @xmath101 are given by @xmath509 * an _ extended trace operator _ is a differentiable family @xmath482 of trace type which is a right @xmath24-operator ( see definition [ xop ] , p. ) such that the reduced support of @xmath73 is a compact subset of @xmath24 .
* an _ extended potential operator _ is a differentiable family @xmath496 of potential type which is a left @xmath371-operator such that the reduced support of @xmath70 is a compact subset of @xmath24 .
* an _ extended singular green operator _ is a differentiable family @xmath510 of singular green type which is equivariant and such that the reduced support of @xmath101 is a compact subset of @xmath119 .
[ def : extops ] _ i ) _ since we also have a right action of @xmath119 on @xmath24 and @xmath24 is diffeomorphic ( via @xmath30 ) to @xmath27 we obtain that being a left @xmath27-operator is equivalent to the equivariance condition with regard to the right action of @xmath119 on @xmath24 given in equation on p. .
hence a potential operator is also a right operator with regard to @xmath24 in this sense , which furnishes by the proof of prop .
[ prop : reduced ] a reduced kernel for extended potential operators .
ii ) _ note that we obtain the reduced kernels for pseudodifferential operators on @xmath119 and extended singular green operators with an argument completely analogous to the proof of prop .
[ prop : reduced ] .
[ rem : extops ] _ i ) _ given an extended trace operator @xmath73 the reduced kernel @xmath197 ( see proposition , p. ) is a compactly supported distribution on @xmath24 conormal to @xmath26 .
_ ii ) _ analogously an extended potential operator @xmath70 has reduced kernel @xmath511 a compactly supported distribution on @xmath27 conormal to @xmath26 . furthermore , @xmath70 is the adjoint of an extended trace operator . _
iii ) _ an extended singular green operator @xmath101 has a reduced kernel @xmath512 being a compactly supported distribution on @xmath119 conormal to @xmath26 .
[ prop : conormal ] we give a proof of conormality for the case _
i ) _ of extended trace operators .
the other cases are the same . given a family of schwartz kernels for @xmath506 contained in @xmath513 for each @xmath195 .
rewrite this as @xmath514 here @xmath515 is the map @xmath516 and @xmath517 then we need to show that : @xmath518 . to this end
let @xmath519 and @xmath520 a cutoff function such that @xmath210 is equal to @xmath272 in a neighborhood of @xmath26 and equal to @xmath40 in a neighborhood containing @xmath521 .
then @xmath522 restricted to @xmath523 yields @xmath524 and this is @xmath144 because @xmath525 by definition .
hence @xmath526 is @xmath144 , but this implies that @xmath527 is smooth as well .
this proves conormality .
finally , we show that a trace operator is the adjoint of a potential operator and vice versa .
let @xmath482 be an extended trace operator and let @xmath506 be the corresponding family of schwartz kernels .
the adjoint @xmath528 is given by @xmath529 such that for @xmath530 we have @xmath531 define the family of operators @xmath496 by @xmath532 and @xmath533 .
we obtain a family @xmath534 of distributions on @xmath535 conormal to @xmath536 for each @xmath195 .
in addition @xmath70 is equivariant with regard to the right action of @xmath119 which is by remark [ rem : extops ] equivalent to being a left @xmath371-operator .
hence @xmath70 is an extended potential operator .
the same argument shows that the adjoint of an extended potential operator is an extended trace operator .
we fix the notation for the reduced kernels and denote by @xmath537 the space of reduced kernels of extended trace operators of order @xmath538 , by @xmath539 the reduced kernels of extended potential operators of order @xmath538 and by @xmath540 the space of reduced kernels of singular green operators of order @xmath538 .
for the pseudodifferential operators on @xmath119 of order @xmath538 we use the notation @xmath541 for the space of operators and @xmath542 for the reduced kernels . with the schwartz kernel theorem
it can be proven that the spaces @xmath541 and @xmath542 are isomorphic , see @xcite , p. 24 .
[ not : reduced ] we will also use the notation @xmath543 for these classes of extended trace , potential and singular green operators , respectively .
the @xmath544 in each case are the appropriate isomorphisms from the schwartz kernel theorem .
hence the operators defined previously act as follows .
the mapping @xmath545 is for @xmath546 given by @xmath547 analogously for the potential operators we have @xmath548 which for @xmath549 is given by @xmath550
lastly , for the singular green operators @xmath551 we have for @xmath552 @xmath553 [ rem : schwartz ] with any fibered space , longitudinally smooth via an action of a nice enough groupoid one can associate an equivariant calculus of pseudodifferential operators .
we want to define such a calculus on @xmath24 and @xmath27 .
the following definition can in somewhat greater generality also be found in @xcite .
a family of pseudodifferential operators of order @xmath538 on @xmath24 is defined as @xmath554 such that _ i ) _ each @xmath555 is contained in @xmath556 .
_ ii ) _ for each chart of @xmath24 given by @xmath557 there is a smooth function @xmath558 such that for each @xmath559 we have @xmath560 via identifying @xmath561 with @xmath562 . here
we denote by @xmath564 the set of pseudodifferential families on @xmath24 .
[ def : bdypsdo ] this leads immediately to a definition of equivariant pseudodifferential operators on @xmath24 and @xmath27 .
the space of equivariant pseudodifferential operators @xmath565 on @xmath24 consists of elements @xmath554 of @xmath566 such that the following equivariance condition holds @xmath567 by analogy we define the equivariant pseudodifferential operators @xmath568 on @xmath27 coming from the left action of @xmath119 .
the equivariance condition in this case is given as in definition [ xop ] , _ ii ) _ on p. .
[ def : bdypsdo2 ] the operators defined here are in each case families parametrized over the double @xmath0 .
we have to clarify what role the pseudodifferential operators defined on @xmath291 play .
we have the following exact sequence @xmath569 & \psi^{\bullet}({\mathcal{x}})^{{\mathcal{g } } } \ar@{->>}[r]^{{\mathcal{r}}_y^{{\mathcal{g } } } } & \psi^{\bullet}({\mathcal{g}}_{\partial } ) } \ ] ] where @xmath570 is a well - defined restriction of families @xmath571 . here
@xmath572 are the smooth functions on @xmath0 that vanish on @xmath6 .
[ prop : bdypsdo ] first note that @xmath291 acts ( from the left and the right ) on itself .
extend this action to the set of families @xmath573 with @xmath574 .
invariance under this action is just the usual equivariance condition for pseudodifferential operators . together with the uniform support condition we therefore recover the class of pseudodifferential operators , denoted @xmath575 .
the exactness of the sequence @xmath576 & \psi^{\bullet}({\mathcal{x } } ) \ar@{->>}[r]^{{\mathcal{r}}_y } & \psi^{\bullet}({\mathcal{x}}_{|y } ) } \ ] ] for the restriction operator @xmath577 defined by @xmath578 is immediate . here
note that @xmath579 by assumption .
note also that by the previous remarks @xmath580 .
this furnishes the exact sequence of equivariant pseudodifferential operators with a well - defined restriction map @xmath570 @xmath569 & \psi^{\bullet}({\mathcal{x}})^{{\mathcal{g } } } \ar@{->>}[r]^-{{\mathcal{r}}_y^{{\mathcal{g } } } } & \psi^{\bullet}({\mathcal{x}}_{|y})^{{\mathcal{g}}_y^y } \cong \psi^{\ast}({\mathcal{g}}_{\partial } ) . } \ ] ]
in order to prove the main theorem we first establish a lemma about compositions of conormal distributions . the classes of extended boutet de monvel operators are closed under compositions induced by groupoid actions and convolution .
more precisely we have the following compositions : @xmath581 [ lem : conormal ] we have the equivalences @xmath582 , @xmath582 and @xmath582 by prop . [
prop : conormal ] , _
ii)_. since the argument in each case goes along the same lines we only treat the first 3 cases of compositions exemplarily . _
i ) _ we consider first the case of the composition . consider a family of extended trace operators @xmath482 and extended potential operators @xmath496 .
denote the corresponding family of schwartz kernels by @xmath583 as well as @xmath584 for @xmath195 .
we make the following computation involving an interchange of integration we still have to justify via a reduction to local coordinates .
let @xmath585 then @xmath586 the kernels of the composition @xmath587 would therefore take the form @xmath588 this corresponds to the convolution of reduced kernels @xmath589 which is immediately defined from the actions .
first we check the support condition of the composed operator .
the reduced support is compact via the inclusion @xmath590 here the inversion of elements in the spaces @xmath24 and @xmath371 is performed inside the groupoid @xmath119 where it is always defined .
fix the projections @xmath591 then by the uniform support condition the family @xmath592 is in particular properly supported .
this means for compact sets @xmath593 we have that @xmath594 is compact .
we make use of this property for the following argument .
next we check the smoothness property of compositions .
let @xmath595 be given , we will show that @xmath596 .
assume that @xmath592 has a schwartz kernel @xmath197 contained in @xmath597 .
then @xmath197 is @xmath144 on the closed subset @xmath598 of @xmath599 and via fiber preserving diffeomorphisms we obtain a @xmath144-atlas .
the function @xmath600 yields a smooth function because we integrate the kernels @xmath601 which are smooth functions .
hence we can interchange integration and differentiation .
therefore @xmath596 for @xmath602 .
consider a general extended trace operator @xmath73 .
let @xmath603 and @xmath466 be a chart with fiber preserving diffeomorphism @xmath604 .
we can assume that @xmath484 is convex , open , @xmath605 and that @xmath478 is a neighborhood of @xmath606 such that @xmath607 gets mapped to @xmath606 via the diffeomorphism .
we also set @xmath608 and @xmath500 with a fiber preserving diffeomorphism @xmath609 ( recall the fact that @xmath610 by assumption ) . by the previous remarks the family @xmath73 is properly supported , which implies in particular that each @xmath194 is properly supported for @xmath195 . hence we obtain that the kernels @xmath601 of @xmath73 satisfy the support estimate @xmath611 therefore the fact that @xmath596 reduces to a computation in local coordinates .
similar reasoning applies to potential operators .
using the same argument as above we deduce that @xmath612 for @xmath613 if @xmath70 is smoothing .
for a general @xmath70 we note that each @xmath614 is properly supported for @xmath195 and hence we obtain that the kernels @xmath615 of @xmath70 satisfy the support estimate @xmath616 the smoothness of @xmath617 reduces to a computation in local coordinates .
consider the general composition @xmath618 for @xmath73 an arbitrary extended trace operator and @xmath70 an arbitrary extended potential operator .
then make the suitable support estimates as above to show that @xmath619 are compositions of smooth families of conormal distributions which act on the sets @xmath620 .
it then follows from a general theorem of hrmander about compositions of conormal distributions , see @xcite , thm .
25.2.3 , p. 21
that the composition @xmath621 yields a family of operators each of which is a conormal distribution in the sense of definition [ def : operators ] , _
iii)_. we therefore obtain a properly supported family @xmath618 which is by the above argument uniformly supported since @xmath70 and @xmath73 are each uniformly supported .
finally , we check the equivariance property for the new family of operators @xmath622 .
since @xmath171 we can write the equivariance condition from definition [ xop ] in the form @xmath623 since @xmath70 is equivariant with regard to the right action of @xmath119 on @xmath24 by remark [ rem : extops ] , _ i ) _ , the equivariance condition for @xmath70 reads @xmath624 for @xmath122 we calculate @xmath625 and hence @xmath618 has the required equivariance property with regard to the right action of the groupoid @xmath119 .
we have thus verified all the properties of an extended green operator . _
ii ) _ consider the next composition @xmath626 which is again for @xmath190 and @xmath627 given by @xmath628 the kernel @xmath629 is written @xmath630 now we can argue again analogously to _
i ) _ that the composition has the right support condition and via a reduction to local charts the formal computation can be made precise .
we therefore obtain a family of kernels @xmath631 .
_ iii ) _ the third composition @xmath632 gives a family of extended trace operators .
we obtain for @xmath633 @xmath634 we obtain the kernel @xmath635 we proceed by making the analogous argument as in _
i ) _ , _ ii)_. the right support condition holds on @xmath636 e.g. via the identification from [ prop : bdypsdo ] .
the rest of the reasoning then yields a family of kernels @xmath637 with the correct support condition .
one can also obtain an equivalent definition of these operator classes based on explicit quantization for conormal distributions .
we recall this definition with the help of the normal bundles and fibrations introduced in the next section .
hence the reader should compare the notation used here to the beginning of section [ repr ] as well as remark [ rem : conormal ] .
let us consider the extended trace operator exemplarily and fix the normal bundle @xmath638 to the inclusion @xmath639 . with the normal fibration
we obtain a map @xmath640 which is a diffeomorphism from open neighborhoods @xmath641 to open neighborhoods of the zero section in @xmath642 . with the normal fibration of the embedding of @xmath26 in @xmath24
the open neighborhoods can be chosen such that the following diagram commutes @xmath643_{r_{\partial } = p } \ar[r]^-{\psi } & { \mathcal{n}}^{{\mathcal{x } } } \delta_y \ar[d ] \\
\delta_y \ar[r ] & \delta_y . }
\ ] ] moreover the condition @xmath644 holds
. then we have the following result .
the space of extended trace operators on @xmath24 identifies with the space of distributional kernels @xmath645 which are given by the integral @xmath646 with the fourier modes @xmath647 and a fixed cutoff function @xmath648 . here
the symbol @xmath649 is contained in the hrmander symbol class @xmath650 .
[ prop : char ] the representation follows from hrmander @xcite , section 18.2 where the distributional kernel @xmath645 takes the form @xmath651 for a symbol @xmath649 contained in @xmath650 . from [ prop : veciso ] we have an isomorphism @xmath652 . additionally , for a given @xmath190 we can trivialize the normal bundle @xmath653 .
this yields the desired form . with the above form of distributional kernels
a trace operator is therefore with @xmath627 written @xmath654 using some more notation from section [ repr ] we can write the potential operators in this form as well .
for the potential operators we have for an open neighborhood @xmath655 the analogous commuting square @xmath643_{s_{\partial } = p^t } \ar[r]^-{\psi^t } & { \mathcal{n}}^{{\mathcal{x}}^t } \delta_y \ar[d ] \\
\delta_y \ar[r ] & \delta_y . }
\ ] ] the representation of distributional kernels is given by @xmath656 [ rem : char ] we return to the special case of a compact manifold @xmath9 with boundary @xmath657 as well as the double @xmath10 .
then we note that the equivariant pseudodifferential operators with regard to the pair groupoid @xmath460 identify as follows @xmath658 this follows because the equivariance implies that the families parametrized over @xmath195 on the left and over @xmath319 on the right are simply constant families . by use of the previous representation of extended operators we can calculate the composition of a pseudodifferential operators on the boundary with a trace operator using only the groupoid action .
hence fix the distributional kernels @xmath659 and @xmath660 . since the definition is invariant under changes of coordinates we loosely identify @xmath661 in the following calculation .
write the kernel of a trace operator for @xmath662 @xmath663 we trivialize the normal bundle @xmath664 .
consider the following definition @xmath665 this involves an interchange of integration which we justify by the rapid decay property along the normal direction ( compare appendix [ a1 ] ) . here
@xmath666 is given for a fixed @xmath667 by @xmath668 we keep in mind that @xmath669 is a filtered algebra with regard to convolution , and use the action of @xmath460 on @xmath25 to calculate for @xmath670 with fixed cutoff @xmath671 @xmath672(z ' , w ' ) e^{i \theta_n w_n } \,d\theta_n.\end{aligned}\ ] ] for @xmath673 fixed we see from this convolution that @xmath674 .
this example is a reflection of the fact that a trace operator acts in the tangential direction like a pseudodifferential operator on the boundary . with the additional rapid decay in the normal direction we can therefore reduce the composition @xmath675 to the form of a composition of pseudodifferential operators on the boundary .
[ exa : comp ]
let us denote by @xmath676 the properly supported extended boutet de monvel operators of order @xmath538 , defined on the interior .
in this section we introduce an algebra of extended operators @xmath677 on the double lie manifold @xmath0 .
this is defined by extending the distributional kernels in @xmath678 to take the lie structure into account .
then we show that this class of operators are in a certain sense the representations of operators in the groupoid calculus adapted to the boundary structure from [ bdystruct ] .
we also fix the actions with corresponding notation from [ actions ] .
next we fix a small tubular neighborhood @xmath679 and partition the manifold accordingly @xmath680 introduce the singular normal bundles for the inclusions @xmath681 as well as @xmath682 .
the singular normal bundles are denoted by @xmath683 respectively .
restricted to the interior we have by axiom _ iii ) _ in def .
[ def : bdystruct ] the isomorphisms @xmath684 here we denote by @xmath685 the normal bundle to the inclusion @xmath686 and the same for the others .
recall the identifications from [ a1 ] of the normal bundles , see also @xcite , p.227 .
we can identify the normal bundles on the interior as subsets of @xmath687 and @xmath688 respectively as follows .
let @xmath689 be the adjoint to the injection @xmath690 .
denote by by @xmath691 the mapping @xmath692 then we write @xmath693 analogously : @xmath694 it is not hard to see that @xmath642 can be identified with @xmath316 which is isomorphic to @xmath695 .
hence we can summarize .
there are ( non - canonical ) isomorphisms @xmath696 [ prop : veciso ] _ i ) _ on the singular normal bundles we define the hrmander symbols spaces @xmath697 such that for @xmath698 open with @xmath699 we have the estimates for @xmath700 @xmath701 for each @xmath702 .
note that we have by hrmander s results a correspondence between the spaces of symbols on the normal bundle to a smooth manifold and conormal distributions on the space ( at least in the smooth case , cf .
@xcite , thm 18.2.11 ) : @xmath703 where @xmath704 is the obligatory correction of order @xmath705 we will ignore this order convention in the following discussions .
note that our earlier definition of smooth families of operators defined as conormal distributions suggests immediately a quantization which we state next .
we can require additionally the ( local ) rapid decay property stated earlier and in the appendix , then we use the notation @xmath706 for these symbol spaces . and analogously the spaces @xmath707 _ ii ) _
a second definition we will need is that of conormal distributions on the normal bundles themselves .
first given the normal and conormal bundles @xmath708 ( and analogously for @xmath709 ) define the _ fiberwise fourier transform _
@xmath710 @xmath711 the inverse is given by duality @xmath712 here we use the notation @xmath713 for the spaces of rapidly decreasing functions on the normal and conormal bundle respectively , see also @xcite , chapter 1.5 .
then the spaces of conormal distributions are defined as : @xmath714 and @xmath715 analogously .
[ rem : conormal ] as a final preparation we fix the notation for the next type of fourier and partial fourier transform we need to consider .
fix the projections @xmath716 .
first given a cutoff function @xmath717 with a function @xmath718 then we set @xmath719 above @xmath720 denotes the exponential on @xmath276 ( by restriction on @xmath721 ) .
then the ( partial ) fourier transform is defined by @xmath722 for the definition of the quantization rule we need some further notation .
let @xmath723 where @xmath724 is the ( positive ) injectivity radius of @xmath0 . * first for the case of trace operators .
we set @xmath725 + as well as @xmath726 + fix the restriction @xmath727 + we denote by @xmath640 the normal fibration of the inclusion @xmath686 such that @xmath640 is the local diffeomorphism mapping an open neighborhood of the zero section @xmath728 onto an open neighborhood @xmath729 ( cf .
@xcite , thm .
then we have the induced map on conormal distributions @xmath730 + also let @xmath731 be a cutoff function which acts by multiplication @xmath732 * for potential operators we use the analogous notation : @xmath733 * finally , in the singular green case we have the induced normal fibration @xmath734 + and the fiberwise fourier transform @xmath735 + the restriction and cutoff is denoted by @xmath736 + and @xmath737 _ i ) _ define @xmath738 such that for @xmath700 we have @xmath739 where @xmath740 _ ii ) _ define @xmath741 such that for @xmath742 we have @xmath743 _ iii ) _ define @xmath744 such that for @xmath745 we have @xmath746 [ def : quant ] the fibrations @xmath747 and @xmath748 define properly supported schwartz kernels .
[ prop : proper ] consider exemplarily the trace operators . since @xmath749 is properly supported we find that @xmath750 defines a properly supported operator .
it is clear from the definition that @xmath751 has the schwartz kernel @xmath752 . instead of the cutoff function @xmath731 we may choose a cutoff function @xmath717 and require rapid decay on part of the normal bundle .
it is immediate to see that these conditions are interchangeable .
the quantization rule for the calculus is described close to the boundary in the tubular neighborhood @xmath753 as follows . _
i ) _ let @xmath754 and @xmath717 a cutoff function .
then the quantization for extended trace operators is given as follows @xmath755 _ ii ) _ let @xmath756 and @xmath717 a cutoff function .
then quantization for extended potential operators is given as follows @xmath757 _ iii ) _ let @xmath758 and @xmath717 a cutoff function .
the quantization for extended singular green operators is then denoted by @xmath759 _ i ) _ we remark that a boutet de monvel calculus consisting of the classes @xmath760 will not form an algebra .
this fails already in the case of only pseudodifferential operators .
the basic reason is that two different lie structures can yield the same metric on @xmath307 ( cf .
hence we need to introduce special smoothing terms to obtain an algebra .
_ ii ) _ the above quantization can be rewritten as follows .
first trivialize the normal and conormal bundles @xmath761 .
then a substitution yields for a fixed @xmath717 @xmath762 where @xmath763 is the integral kernel on @xmath764 .
hence this kernel has the form for @xmath765 @xmath766 _ iii ) _ let @xmath767 be the differentiation in the normal direction , i.e. @xmath768 for some fixed @xmath769 with support in our fixed tubular neighboorhood , which is differentiation in the normal direction close to @xmath6 . then note the formal similarity between our quantization and the quantization of boundary pseudodifferential operators : @xmath770 in fact this can also be shown for the restricted calculus , but we will not do so here .
[ rem : quant ] the quantizations @xmath771 are in each case independent of the choice of cutoff functions up to smoothing errors .
[ prop : cutoff ] we show this exemplarily for the trace operators .
let @xmath754 and @xmath772 .
then the kernel of @xmath773 can be written : @xmath774 since the behaviour in the normal direction is taken care of by the rapid decay condition we only need to regularize the integral in the cotangent direction .
note that the phase function @xmath775 has only critical points for @xmath776 .
for @xmath777 we can therefore find a vector field @xmath704 such that @xmath778 .
hence write for each @xmath779 such that @xmath780 @xmath781 this shows @xmath782 is smoothing . from the compactness of @xmath0 and @xmath6 we can associate to each vector field in @xmath1 respectively
@xmath299 a _ global flow _
then consider the diffeomorphisms evaluated at time @xmath784 @xmath785 and fix the corresponding group actions on functions which we denote by @xmath786 the upshot of this is a definition of the suitable smoothing terms for our calculus which we state next . _
i ) _ the class of @xmath33-trace operators is defined as @xmath787 here @xmath788 consists of the extended operators from the previous definition .
the residual class is defined as follows @xmath789 _ ii ) _ the class of @xmath33-potential operators is defined in the same fashion @xmath790 with residual class @xmath791 _ iii ) _
lastly , the class of @xmath33-singular green operators is defined as @xmath792 with residual class @xmath793 _ iv ) _ then the calculus @xmath794 of extended operators consists of matrices of the form @xmath795 for @xmath796 and @xmath101 an extended singular green operator , @xmath70 extended potential and @xmath73 extended trace operator .
[ def : extended ]
the extended calculus is related to our algebra defined in the boundary structure in a very strong sense .
in fact in many cases there is a canonical isomorphism which is furnished by the so - called _
vector representation_. _ i ) _ recall the definition of the vector representations @xmath279 associated to the groupoid @xmath119 and @xmath305 associated to the groupoid @xmath291 .
these are homomorphisms of ( filtered ) algebras ( cf .
@xcite ) @xmath797 and @xmath798 these can be viewed as suitable extensions of the anchor maps @xmath799 and @xmath800 ( abusing notation by using the same symbols ) .
the vector representations are uniquely characterized by the equations @xmath801 for each @xmath802 . as well as @xmath803 for each @xmath804 and @xmath805 .
_ ii ) _ we want to define the following vector representation @xmath806 which will turn out to be a well - defined surjective map and a homomorphism of algebras for @xmath807 . the homomorphism @xmath47 is represented as a matrix @xmath808 it is uniquely determined by the defining property @xmath809 for each @xmath810 and @xmath811 . from this we can define linear mappings which represent trace , potential and singular green operators individually . so given a trace operator @xmath73 , a potential operator @xmath70 and a singular green operator @xmath101 ( equivariant families ) we define @xmath812 for the pseudodifferential operator on the boundary we simply recover the definition of the representation @xmath305 . [
rem : extended ] given a @xmath33-boundary structure the previously defined vector representation @xmath47 furnishes the isomorphism @xmath813 [ thm : bm2 ] we have to check two inclusions for each matrix component of the homomorphism @xmath47 .
for pseudodifferential operators we can refer to the result in @xcite , thm .
the case of singular green operators @xmath814 follows the same lines of reasoning ( in fact , it is an easier case than the trace and potential operators ) .
the case of trace operators @xmath815 is equivalent to the case of potential operators @xmath816 by duality [ prop : conormal ] , _ ii ) _ ( each component of @xmath47 is adjoint preserving ) . therefore we will establish the assertion for the trace operators .
we need to show that ( setting @xmath817 ) @xmath818 first we will establish the commutativity of the following diagram : @xmath819^{\chi } & i_{(r)}^{m}({\mathcal{n}}^{{\mathcal{x } } } \delta_y , \delta_y ) \ar[d]_{\alpha_{\ast } } \ar[r]^-{{\mathcal{r } } } & i_{(r)}^{m}(n^{y_0 \times m_0 } \delta_{y_0 } , \delta_{y_0 } ) \ar[d]_{\psi_{\ast } } \\ { \mathscr{t}}^{m,0}({\mathcal{g } } , { \mathcal{x } } ) \ar[d]_{e_x } \ar@{>->>}[r ] & i_{c}^{m}({\mathcal{x } } , \delta_y ) \ar[d]_{\mu^{\ast } } \ar[r]^-{l_{\ast } } & i^{m}(y_0 \times m_0 , \delta_{y_0 } ) \ar@{=}[d ] \\ { \mathscr{t}}_{prop}^{m,0}({\mathcal{g}}_x ; { \mathcal{x}}_x)^{\gamma } \ar@{>->>}[r ] & i_{prop}^{m}({\mathcal{x}}_x \times { \mathcal{g}}_x , \delta_x)^{\gamma } \ar[d]_{{\mathcal{j } } } \ar[r]^{\widetilde{(p \otimes r)}_{\ast } } & i^{m}(y_0 \times m_0 , \delta_{y_0 } ) \ar[d]_{{\mathcal{j } } } \\ & { \mathscr{t}}_{prop}^{m,0}({\mathcal{g}}_x ; { \mathcal{x}}_x)^{\gamma } \ar[r ] & { \mathscr{t}}_{prop}^{m,0}(m_0 , y_0 ) } \end{aligned}\ ] ] each of the maps in this diagram are defined as follows .
fix @xmath820 which is stricly smaller than the injectivity radius of the manifold @xmath0 .
then we have by the tubular neighborhood theorem ( see @xcite , p. 53 ) an open embedding @xmath821 which induces the map on conormal distributions @xmath822 the function @xmath823 is a fixed cutoff function and the arrow means a multiplication by this function and @xmath824 denotes restriction .
furthermore , we fix a local diffeomorphism @xmath640 which maps diffeomorphically a neighborhood @xmath825 of the zero section onto a neighborhood of the diagonal @xmath826 . restricted to the interior we have by the previous identifications of the normal bundles as subbundles of @xmath827 the definition of the exponential map
then we recover the previously mentioned normal fibration for the inclusion on the interior @xmath686 .
hence we then denote by @xmath829 the induced ( local ) mapping on conormal distributions .
the map @xmath830 denotes the already defined fiberwise fourier transform , @xmath831 is always the suitable isomorphism from the schwartz kernel theorem , and @xmath832 is the induced mapping coming from the multiplication @xmath833 .
the illucidation of the mappings @xmath834 and @xmath835 requires a little more care . for the remainder of the argument we fix a @xmath368 and set @xmath836 for the isotropy group .
on @xmath837 we have the induced metric @xmath838 from the compatible metric @xmath322 on @xmath99 .
hence we consider on @xmath839 the riemannian metric induced from @xmath838 on @xmath837 .
we have that @xmath840 and @xmath841 are surjective submersions .
furthermore , @xmath842 and @xmath843 are isometries with regard to the metric @xmath838 .
it follows that @xmath844 are local diffeomorphisms . additionally , the discrete group @xmath341 acts freely on @xmath837 from both sides and on @xmath839 from the right .
hence we consider the right action of @xmath341 on @xmath523 .
we have that @xmath845 and @xmath846 .
thus @xmath844 are each covering maps with covering group @xmath341 .
consider the @xmath341-invariant functions @xmath847 and @xmath848 .
for a given @xmath849 and @xmath850 we have that @xmath851 is a @xmath341-invariant function on @xmath837 and @xmath852 is a @xmath341-invariant function on @xmath839 .
hence for a given @xmath849 the function @xmath853 is defined since @xmath73 is properly supported and there is a @xmath850 such that @xmath854 the operator @xmath855 is then given by @xmath856 .
we can therefore rewrite the vector representation @xmath857 as @xmath858 and by the commutativity of the left - most rectangle obtain @xmath859 thus the operator @xmath835 is completely determined by the commutativity of the bottom rectangle .
the map @xmath834 in the diagram is defined by @xmath860 it remains to prove that @xmath861 for this we need an alternative description of @xmath834 which is given as follows .
the space @xmath862 denotes the space of @xmath341-invariant conormal distributions on @xmath523 .
since @xmath837 is a covering of @xmath307 and @xmath839 is a covering of @xmath863 with group @xmath341 we have the identifications @xmath864 denote by @xmath865 the map @xmath866 which is then also a covering map .
this makes it possible to identify a distribution with small support in @xmath867 with a distribution with support in a small subset of @xmath444 .
the identifications extend by summation along the fibers of @xmath868 . to any distribution @xmath869 on @xmath870
whose support intersects only finitely many components of @xmath871 for any locally trivializing open set @xmath872 we associate a distribution @xmath873 .
the morphism @xmath835 identifies then with @xmath874 .
we observe that @xmath875 then define @xmath876 as follows : restrict a distribution @xmath877 to @xmath878 and apply the pushforward given by @xmath879 . since @xmath880 by assumption a on p. it follows @xmath881 hence @xmath882 and we have an alternative description of @xmath834 .
this establishes since @xmath883 is injective on @xmath884 and @xmath885 are isometric coverings which preserve the exponential maps .
the equality stated in which we left open can now be established via the commutativity of the diagram . consider the first inclusion @xmath886 .
we have by commutativity @xmath887 where @xmath888 and @xmath889 .
hence every operator of the form @xmath890 is in the range of @xmath279 .
we have thus shown that @xmath891 .
now let @xmath892 be a vector field and lift it to a vector field @xmath893 on @xmath119 . by integrating this vector field
let @xmath894 be the family of diffeomorphisms of each @xmath837 .
we obtain @xmath895 where @xmath896 and @xmath897 , since a regularizing distribution multiplied with an operator induced by a diffeomorphism is regularizing
. hence the first inclusion follows .
consider the opposite inclusion @xmath898 : let @xmath899 be given and set @xmath900 .
fix a cutoff function @xmath901 supported in @xmath902 and let @xmath823 be such that @xmath903 on the support of @xmath904 . setting @xmath905 we write @xmath906 for @xmath907 where @xmath908 then we obtain @xmath909 this concludes the proof .
from the extended calculus one can easily define the corresponding restricted boutet de monvel operators . for this
we introduce the truncation operators on the manifold level and the groupoid level .
the restriction @xmath910 to the interior @xmath911 and the extension by zero operator @xmath912 are given on the manifold level by @xmath913^{r^{+ } } & \ar@/-0pc/[l]^-{e^{+ } } l^2(\mathring{x}_0 ) } \ ] ] with @xmath914 and @xmath915 being a projection onto a subspace of @xmath916 .
on the groupoid level we use the same symbols since it will be clear from context which is meant .
so we define the operators @xmath917^{r^{+ } } & \ar@/-0pc/[l]^-{e^{+ } } l^2({\mathcal{g}}^{+ } ) } \ ] ] with @xmath918 and @xmath915 being a projection onto a subspace of @xmath919 .
first we are going to introduce an important propery of the pseudodifferential operators on the groupoid @xmath920 .
namely we require that each element of the family @xmath921 has the _ transmission property _ with regard to the the boundary @xmath839 .
the operator @xmath922 has the transmission property if the symbol @xmath923 . here the class of hrmander symbols @xmath923 consists of families @xmath924 such that each symbol @xmath925 has the transmission property with regard to @xmath926 .
in particular the operators @xmath927 map functions smooth up to the boundary @xmath839 to functions which have the same property .
[ def : tprop ] * notice first that if @xmath368 is an interior point we have that @xmath928 and we recover the transmission property on the interior manifold @xmath929 with boundary @xmath863 . * in our trivial case @xmath930 and @xmath10 , @xmath9 a compact manifold with boundary
@xmath64 we recover the transmission property meaning @xmath931 .
the operation of _ truncation _ itself is given as a linear operator . _
i ) _ on the groupoid calculus this operator is given by @xmath932 _ ii ) _ on the extended calculus we define @xmath933 now we will give a definition of the _ restricted calculus _ defined on the boundary structure as well as the _ representable operators _ on the lie manifold with boundary @xmath934 . _
i ) _ the restricted calculus on the boundary structure is defined as the set of operators for @xmath935 @xmath936 _
ii ) _ the class of _ representable operators _ is for @xmath935 defined as @xmath937 the _ vector representation _
@xmath938 is defined by @xmath939 for @xmath940 and @xmath941 .
[ def : restricted ] the next result is now a consequence of our previous preparations . the restricted operators @xmath942 are closed under composition .
[ thm : bmr1 ] this is an immediate consequence of the proof of lemma [ lem : conormal ] .
the truncated operators are boutet de monvel operators on a smooth manifold .
the support estimates in the proof of lemma [ lem : conormal ] therefore reduce the calculations to the ordinary known case of properly supported operators . from the right action of the groupoid @xmath119 on @xmath24
we obtain the induced action of the subgroupoid @xmath943 .
the right action @xmath944{90}{$\circlearrowright$}\ { \mathcal{g}}^{+}$ ] defines the equivariance of the families in the calculus @xmath942 .
the range and source maps in @xmath945 are the restrictions of the range and source maps in @xmath119 , hence the representation @xmath938 is well - defined .
thereby we obtain that the truncation maps are linear , equivariant maps and the following diagram commutes as linear operators @xmath946^{\tilde{\varrho}_{bm } } { \mathcal{b}}^{m,0}({\mathcal{g}}^{+ } , { \mathcal{x } } ) \\ { \mathcal{b}}_{2{\mathcal{v}}}^{m,0}(m , y ) \ar@{->>}[u]^{{\mathcal{c } } } & \ar[l]^{\varrho_{bm } } { \mathcal{b}}^{m,0}({\mathcal{g } } , { \mathcal{x } } ) \ar@{->>}[u]^{\tilde{{\mathcal{c}}}}. } \ ] ] [ rem : restricted ] the vector representation induces an isomorphism for @xmath935 @xmath947 in particular @xmath45 is closed under composition .
[ thm : bmr2 ] this follows from remark [ rem : restricted ] combined with theorem [ thm : bm2 ] .
since @xmath948 and @xmath47 are surjective we obtain the surjectivity of @xmath938 as follows .
let @xmath949 then by surjectivity of @xmath47 and @xmath950 we lift this to an element @xmath951 .
then @xmath952 is the required preimage . by commutativity
we have @xmath953 hence in this case @xmath938 is surjective .
it is also immediate that it is a well - defined homomorphism of algebras .
this yields the closedness under composition .
finally , we are going to show that the operators in our algebra are continuous on @xmath954-spaces . for higher orders this result can be easily generalized to sobolev spaces defined on the groupoids or lie manifolds as they were introduced e.g. in @xcite .
let @xmath955 such that @xmath956 where @xmath957 a pseudodifferential operator with transmission property , @xmath73 is a restriced trace operator and @xmath70 a restricted potential operator .
@xmath101 is a restricted singular green operator , and @xmath958 is a pseudodifferential operator on the boundary
. then we have the following continuous extensions @xmath959 [ thm : continuity ] _ i ) _ the continuous extension property of @xmath960 follows from the continuity on each fiber @xmath961 and the fiberwise transmission property .
the continuity of @xmath79 is also clear . _
ii ) _ consider a trace operator @xmath73 and denote by @xmath532 the @xmath954-adjoint which is a potential operator ( cf .
[ prop : conormal ] ) . by theorem [ thm : bmr1 ]
we have @xmath621 is a pseudodifferential operator in @xmath962 . since @xmath621 is of order zero @xmath963 continuously ( via @xmath30 ) .
hence @xmath964 and thus @xmath965 is continuous . the same way we show continuity of the trace operators .
_ iii ) _ since the truncated operator @xmath101 is a family of singular green operators we know that we can write these operators in the form @xmath618 . where @xmath70 is a potential operator of order @xmath40 and
@xmath73 a trace operator of order and type @xmath40 .
then the continuity of @xmath101 is a consequence of the continuity of @xmath73 and @xmath70 . * vector bundles * up until now we have only considered scalar operators of boutet de monvel type .
it does only require minor modifications to consider operators acting on smooth sections of smooth vector bundles , see also @xcite . to this effect
let @xmath966 be smooth vector bundles on @xmath9 and @xmath967 smooth vector bundles on @xmath6 .
we can pull back these bundles to @xmath945 via @xmath968 .
similarly , the actions allow us to pull back the bundles @xmath969 to @xmath24 and obtain @xmath970 .
it is not difficult to modify our construction for operators acting on the smooth sections such that @xmath971 is a continuous linear operator @xmath972 similarly , @xmath973 is a continuous linear operator @xmath974
in this section we make preparations for the parametrix constructions for elliptic elements in our algebra . for this we will define a suitable notion of ellipticity in the next section . at the outset it is required to consider an enlarged groupoid calculus of pseudodifferential operators .
we present one approach based on non - canonical completions .
this ensures that inverses and parametrices are contained in the calculus .
the material of this section is mostly combined from @xcite and @xcite .
we will nevertheless provide some details for the benefit of the reader .
first we make a common assumption on the groupoid . +
* assumption c. * the groupoid @xmath20 is hausdorff . +
this assumption has several simplifying consequences .
the vector representation @xmath975 yields an isomorphism @xmath976 [ lem : vr ] the injectivity of @xmath975 is a consequence of the hausdorff condition on @xmath119 .
we consider for a fixed @xmath977 the evaluation morphism @xmath978 given by @xmath979 .
the evaluation morphism @xmath978 is injective for @xmath977 . in order to prove this
we first establish the _ claim _ : let @xmath980 , then for any @xmath981 the map @xmath982 is continuous .
_ proof of claim : _ let @xmath981 , then there is a @xmath983 such that @xmath984
. then @xmath985 by a choice of haar system ( using the smoothness property of haar systems ) and the hausdorff property of @xmath119 .
this proves the claim . to see that @xmath986 is injective for a fixed @xmath977 let @xmath987 .
we need to show that then @xmath988 for each @xmath989 , i.e. @xmath990 . by right invariance and the fact that @xmath991 it follows that @xmath988 for each @xmath992 .
let @xmath989 be arbitrary , then we show that @xmath993 for any @xmath994 .
let @xmath981 be such that @xmath995 which is possible since @xmath996 is closed in the locally compact hausdorff space @xmath119
. then by the claim @xmath997 is continuous and on @xmath992 the function vanishes .
since @xmath307 is dense in @xmath0 it follows @xmath998 for each @xmath989 . hence @xmath986 is indeed injective .
using the canonical diffeomorphism @xmath999 we obtain a bijection @xmath1000 .
the composition of @xmath986 with @xmath1001 coincides with the representation @xmath975 .
since @xmath975 is known to be surjective by the representation theorem ( compare remark [ rem : liestruct ] , _ i ) _ ) this proves that @xmath975 is an isomorphism . _
i ) _ under certain additional conditions on the lie manifold ( considering so - called cm - manifolds ) we can in fact show that assumption c implies that the induced representation @xmath1002 is also injective .
therefore in this case @xmath119 is automatically amenable .
we refer to @xcite for the details . _
ii ) _ note that the lemma implies also that @xmath938 furnishes an isomorphism @xmath1003 under this assumption .
[ rem : vr ] given @xmath1004 we set @xmath1005 such that @xmath1006 .
we define the order reducing operators @xmath1007 and @xmath1008 for each @xmath1009 in the closure of the domain of the operator @xmath1010 define @xmath1011 this norm gives rise to sobolev spaces henceforth denoted by @xmath1012 .
it is equivalent to the sobolev spaces introduced in @xcite .
these order reductions are insufficient even though the symbol is clearly elliptic for the simple reason that the parametrix is not contained in the calculus . to get to the actual order reduction we need to do some more work .
the guillemin completion is now defined as @xmath1013 where @xmath1014 is completed with regard to the seminorms @xmath1015 secondly , we set @xmath1016 let @xmath1017 be a domain and denote by @xmath1018 the algebra of holomorphic functions on @xmath478 with values in a frchet space @xmath1019 . from @xcite
we cite the following lemma on holomorphic families . _
i ) _ let @xmath1020 a holomorphic family of elliptic operators . then there is a @xmath1021 such that @xmath1022 _ ii ) _ there is a @xmath1023 with @xmath1024 and @xmath1025 [ lem : holfam ] _ i ) _ the construction of @xmath1026 via asymptotic expansion is standard if we have shown that holomorphic families are compatible with asymptotic expansions in the sense of the proof of prop . 4.1 in @xcite . _
ii ) _ apply _ i ) _ to the family @xmath1027 .
this puts us in a position to show that parametrices are contained in the guillemin completion and construct elliptic invertible elements ( order reductions ) .
additionally , the following theorem contains the basic algebraic properties : @xmath1028-algebra and spectral invariance . _
i ) _ @xmath1029 is a frchet algebra and @xmath1030 is a @xmath1028-algebra . _
ii ) _ for a given @xmath1031 there is a @xmath1032 such that @xmath1033 . _
iii ) _ let @xmath1034 and @xmath1035 then @xmath1036 . _
iv ) _ for @xmath1037 with @xmath1038 which is elliptic and invertible , as a possibly unbounded operator on @xmath954 , we have @xmath1039 .
[ thm : guil ] _ i ) _ we check that the family of seminorms @xmath1040 as defined above is submultiplicative . since they generate a frchet topology
this proves that @xmath1029 is a frchet algebra .
let @xmath1041 and @xmath1042 then @xmath1043 for the @xmath1044-property we refer to the general result @xcite , prop .
_ ii ) _ make use of the lemma and set @xmath1045 hence @xmath1046 . for @xmath1047 with @xmath1048 sufficiently small @xmath1049 is invertible on @xmath916 .
thence @xmath1050 for @xmath1047 sufficiently small .
choose @xmath779 such that @xmath1051 .
this yields that @xmath1052^k$ ] has the required property . _
iii ) _ the operator @xmath1053 is invertible if @xmath1054
. then since @xmath1055 is open it follows that inversion is continuous .
the integral @xmath1056 exists and hence @xmath1057 . _
@xmath1058 ) _ let @xmath1059 be such that @xmath1060 which is possible by _
ii)_. then @xmath1061 is elliptic , injective and bounded . from the ellipticity of @xmath1062
we obtain that the range of @xmath72 is @xmath1063 and by ellipticity of @xmath66 we obtain that the domain of @xmath66 is also @xmath1064 .
thus @xmath1065 is also surjective and hence invertible on @xmath1066
. therefore we can consider @xmath1067 as an @xmath954-bounded operator . _
@xmath1068 ) _ by the previous argument we can wlog assume that @xmath807 and that @xmath1069 is invertible as an operator on @xmath954 .
we show first the _ claim : _ there exists @xmath1070 such that @xmath1071 with regard to @xmath1072 . clearly @xmath1073 is bounded and invertible . by continuity of the functional calculus
, we obtain a sequence @xmath1074 of polynomials such that @xmath1075 set @xmath1076 , then we have @xmath1077 this proves the claim .
set @xmath1078 .
with the claim it follows that @xmath1079 set @xmath1080 . for @xmath1081
sufficiently large this is a parametrix with @xmath1082 this yields @xmath1083 and @xmath1084 .
the following sections mark the start of the analysis of the represented side of our calculus ( the algebra @xmath45 ) . in order to limit the size of the paper and
preserve its readability most proofs are referred to the comprehensive literature on boutet de monvel s calculus , where these constructions have already been given and which need only be adapted to our setup in a straightforward manner .
we partition the manifold @xmath0 as in on page and set @xmath1085 , then close to the boundary ( in the fixed tubular neighborhood ) we define the symbol @xmath1086 for @xmath1031 , some constant @xmath1087 and @xmath1088 such that @xmath1089 .
then @xmath1090 is an elliptic symbol and has the transmission property ( cf .
@xcite , proposition 1.3 ) .
we construct a global classical symbol @xmath1091 as follows .
fix a normal cover @xmath1092 of @xmath0 with local trivializations @xmath1093 . * if @xmath1094 ( an _ interior chart _ ) set @xmath1095 . *
if @xmath1096 ( a _ boundary chart _ ) set @xmath1097 .
let @xmath1091 be the order reducing symbol as defined above and set @xmath1098^k$ ] for @xmath779 chosen suitably large as in the proof of [ thm : guil ] , part _ ii)_. [ def : reduction ] we can use the same argument as in @xcite , proposition 1.7 to prove the following result .
the operator @xmath1099 extends to a linear isomorphism @xmath1100 for each @xmath1101 .
[ prop : reduction ] denote by @xmath1102 the restriction operator @xmath1103 .
then @xmath1104 is continuous for each @xmath1031 , see @xcite , theorem 4.7 .
we also fix the operator @xmath1105 which is supported in @xmath1106 and near @xmath6 coincides with @xmath1107 for @xmath1108 . for @xmath1109 with @xmath1110
we obtain @xmath1111 is continuous .
[ lem : partial ] the extension by zero operator @xmath1112 is a priori continuous for @xmath1113 .
since any differential operator has the transmission property we obtain that @xmath1114 is continuous for @xmath1110 by @xcite , theorem 1.14 .
set for a given @xmath1115 with support in @xmath1116 @xmath1117 given @xmath1118 denote by @xmath1119 the distribution given by @xmath1120 [ def : gammaj ] for @xmath1121 we have @xmath1122 is continuous .
[ prop : gammaj ] the differential operator @xmath1123 is continuous .
denote by @xmath1124 a continuous extension operator depending on @xmath129 . by the previous remarks
@xmath1125 is continuous provided @xmath1110 . in the fixed tubular neighborhood
we can write @xmath1126 .
this proves the claim by the continuity of @xmath1127 and @xmath1128 .
we are now in a position to define the operators @xmath1129 of arbitrary order @xmath1031 and type @xmath1130 . _
i ) _ a _ trace operator _ of order @xmath1031 and type @xmath1130 is defined as @xmath1131 if there is a sequence @xmath1132 such that @xmath1133 a residual trace operator of type @xmath1130 is defined in the same way for a sequence @xmath1134 .
_ ii ) _ a _ singular green operator _ of order @xmath1031 and type @xmath1130 is defined as @xmath1135 if there is a sequence @xmath1136 such that @xmath1137 a residual singular green operator of type @xmath1130 is defined in the same way for a sequence @xmath1138 .
[ def : bmd ] in @xcite , p. 149
we have the following standard representations for elements in boutet de monvel s calculus .
a trace operator @xmath1139 takes the form @xmath1140 for a @xmath1141 and a potential operator @xmath1142 takes the form @xmath1143 for a @xmath1144 .
we should remark that in our framework we can prove these properties in the same way . by an elementary but tedious computation
we can show directly that the formal adjoint of a trace operator is a potential operator also in the order @xmath1145 cases .
[ rem : stdrepr ] _ i ) _ a trace operator @xmath1131 extends to a continuous linear operator @xmath1146 for @xmath1147 .
_ ii ) _ a potential operator @xmath1148 extends to a continuous linear operator @xmath1149 for @xmath1150 . _
iii ) _ a singular green operator @xmath1135 extends to a continuous linear operator @xmath1151 for @xmath1147 .
[ thm : cont2 ] _ i ) _ let @xmath1131 and decompose @xmath73 as follows @xmath1152 where @xmath1139 and @xmath1153 can be written @xmath1154 ( see e.g. @xcite for the proof ) .
hence the continuity follows from proposition [ prop : gammaj ] .
_ ii ) _ this follows because @xmath70 is the formal adjoint of a trace operator @xmath73 . _ iii ) _ by the same proof as in @xcite , lemma 2.2.14 we can write @xmath1155 the first term on the right hand side is continuous by proposition [ prop : gammaj ] and _ ii)_. the second term has the following series representation by @xcite , lemma 7 , p.149 @xmath1156 for a sequence of potential operators @xmath1157 and trace operators @xmath1158 .
hence the continuity follows by _
i ) _ and _ ii)_.
in the sequel we will study the corresponding guillemin completion for boutet de monvel s calculus . by analogous arguments as in section [ guillemin ]
it follows that the completed calculus has the same favorable algebraic properties .
most importantly , inverses ( if they exist ) are contained in the calculus .
define the completion @xmath50 of the residual boutet de monvel operators with regard to the family of norms of operators @xmath51 on sobolev spaces .
let @xmath1159 be the _ completed _ calculus of boutet de monvel operators .
[ def : completed ] let @xmath1160 with @xmath1161 and @xmath1162 order reductions in the pseudodifferential calculi .
define @xmath1163 , then @xmath1164 .
[ prop : red ] let @xmath935 , then @xmath1165 implies that @xmath1166 where @xmath1167 [ prop : adjoint ] we have @xmath1168^{\ast } = r^{+ } { \operatorname{op}}(a)^{\ast } e^{+}$ ] immediately because @xmath935 and @xmath954-continuity .
the adjoint of a trace operator is a potential operator and vice versa and the adjoint of a singular green operator is a again a singular green operator by definition .
finally , @xmath1169 since the lie calculus is closed under adjoints , see @xcite , corollary 3.3 . for the following theorem we use a common trick based on order reductions ,
see for example @xcite , theorem 6.3.6 .
let @xmath1170 with @xmath1171 . _
i ) _ with @xmath935 and @xmath1172 it follows that @xmath1173 _ ii ) _ for @xmath1145 it follows that @xmath1174 _ iii ) _ if @xmath1175 is invertible for some @xmath1176 as an operator @xmath1177 then the inverse @xmath1178 . [ thm : inverse ] the properties _ i ) _ and _ ii ) _ follow immediately from proposition [ prop : adjoint ] . in order to prove _ iii ) _ we consider the pseudodifferential order reductions @xmath1179 and @xmath1180 of order @xmath1181 and @xmath1182 respectively .
then @xmath1183 is contained in @xmath1184 and is invertible .
therefore the assertion reduces to the order @xmath1185 case .
we apply the @xmath1044-property of the @xmath1185-algebra which is proved analogously to theorem [ thm : guil ] .
in this section we will introduce the principal and principal boundary symbol of an operator in our calculus .
we will define the notion of ellipticity and show that a parametrix exists under the previously stated conditions on the calculus .
a major technical problem is that in the lie calculus already inverses of invertible operators are not necessarily contained .
this makes a parametrix construction difficult and we state here a version of such a result .
there are at least two approaches to overcome the problem of inverses : _ ( 1 ) _ using a larger calculus of pseudodifferential operators ( with asymptotics ) or _ ( 2 ) _ completing the algebra of pseudodifferential operators ( non - canonically ) such that inverses are contained . in the last section we outlined the second approach
. we will henceforth consider the completed boutet de monvel calculus as defined in section [ completed ] .
fix the smooth , hermitian vector bundles @xmath1186 and recall the notation for the boundary algebroid and its co - bundle @xmath1187 .
we define the principal symbol and principal boundary symbol on @xmath1188 for @xmath935 .
* set @xmath1189 for the vertical tangent bundle over @xmath24 .
* let @xmath1190 be a @xmath144-family of boutet de monvel operators .
* then for each @xmath1191 we have a boutet de monvel operator @xmath1192 .
* hence the principal symbol @xmath1193 and the principal boundary symbol @xmath1194 are defined invariantly on @xmath1195 and @xmath1196 respectively .
* by right - invariance of the family @xmath54 these symbols descend to a principal symbol @xmath1197 and principal boundary symbol @xmath1198 defined invariantly on @xmath1199 and @xmath1200 respectively . *
introduce the _ indicial symbol _
@xmath1201 for a face @xmath1202 as the restriction @xmath1203 . by noting that @xmath1204 we make the following definition for principal and principal boundary symbol on the represented algebra @xmath1205 for @xmath935 . to an element
@xmath1206 we associate the two principal symbols . _
i ) _ the _ principal boundary symbol _ @xmath1207 is defined as @xmath1208 this yields a section of the infinite dimensional bundle @xmath1209 here @xmath1210 is a bundle with fiber @xmath1211 on the inward pointing normal direction .
in particular the restriction of @xmath1212 to the interior @xmath46 agrees with the principal boundary symbol on the interior . _
ii ) _ the _ principal symbol _
@xmath1213 which is the principal symbol of the pseudodifferential operator in the upper left corner of the matrix @xmath54 .
this yields a section in @xmath1214 _ iii ) _ define by @xmath1215 the space consisting of pairs of principal symbols @xmath1216 .
these are homogenous or @xmath1217-homogenous sections of the bundles @xmath1218 , respectively , with canonical compatibility condition .
[ def : bdysmb ] given @xmath1219 we have @xmath1220 [ lem : mult ] as proven in section [ guillemin ] the vector representation furnishes an isomorphism @xmath1221 since the principal and principal boundary symbol are each defined invariantly on the bundles @xmath1222 the computation reduces to the equivariant families of boutet de monvel operators . hence given @xmath1223 in @xmath1224 we have @xmath1225 in the same way we obtain multiplicativity of the principal symbol . since we will in the following
only be concerned with represented operators we will simply write @xmath1226 and @xmath1227 for @xmath1228 and @xmath1212 . for the following result and proof in the standard case see e. g. @xcite .
the following sequence is exact @xmath1229 & { \mathcal{b}}_{{\mathcal{v}}}^{0,0}(x , y ; e_1 , e_2 , j_{\pm } ) \ar@{->>}[r]^{\sigma \oplus \sigma_{\partial } } & \sigma_{{\mathcal{v}}}^{0,0}(x , y ; e_1 , e_2 , j_{\pm } ) . } \ ] ] [ thm : exact ] _ i ) _ the same exact sequence holds for the interior calculus . since the principal symbols are extensions of the interior we immediately obtain that @xmath1230 .
_ ii ) _ to prove surjectivity let @xmath1231 .
since we also have an analogous exact sequence for the class of pseudodifferential operators @xmath1232 it suffices to find singular green , trace and potential operators in the preimage . in a fixed small tubular neighborhood
@xmath1233 trivialize the singular normal bundle @xmath317 .
let @xmath1234 be a normal covering of @xmath6 ( assumed finite by compactness of @xmath6 ) and let @xmath1235 be a subordinate partition of unity .
we also fix a boundary defining function @xmath1236 such that @xmath1237 .
this is defined by the tubular neighborhood theorem for lie manifolds ( cf .
hence for the diffeomorphism of tubular neighborhoods @xmath1238 we have @xmath1239 let @xmath1240 be a cutoff function such that @xmath1241 close to @xmath40 .
hence we can locally on each trivialization @xmath1242 construct singular green , potential and trace operators . via our partition of unity
we obtain corresponding global symbols .
this furnishes a global right inverse morphism @xmath1243 .
the notion of ellipticity we introduce here is the usual condition of shapiro - lopatinski type .
it is sufficient to obtain a parametrix .
we obtain fredholm operators if we additionally request the invertibility of the indicial symbols . an operator @xmath1244 is called _ i ) _ _ @xmath33-elliptic _ iff @xmath1245 is pointwise bijective , _ ii ) _ _ elliptic _ if @xmath54 is @xmath33-elliptic and for each @xmath1202 the indicial symbol @xmath58 is invertible .
[ def : sl ] the elements of the symbol algebra @xmath1246 can be written more simply in terms of the _ action in the normal direction_. let us define this action for a given symbol @xmath1247 .
we set @xmath1248 then in local trivializations of the bundles @xmath1249 and @xmath1250 we have @xmath1251 where @xmath779 is the fibre dimension of @xmath1249
. we will alternatively consider these maps as operators acting on @xmath1252 , the closure of @xmath1211 . in @xcite the action in the normal direction
is defined for the boundary symbols which act more generally as _ wiener - hopf operators _ on the spaces @xmath1253 and their respective @xmath954-closures .
we will not need this formulation in the present work .
denote by @xmath1254 the canonical projection of the sphere bundle .
let @xmath54 be @xmath33-elliptic and consider the principal symbol @xmath1255 of the pseudodifferential operator @xmath66 in the upper left corner .
since the infinite dimensional bundles are trivial the action in the normal direction @xmath1256 yields a fredholm family , parametrized over @xmath1257 which preserves the index , cf .
@xcite , prop . 5 , p.95 .
we can express the principal boundary symbol ( with analogously defined actions in the normal direction for green , potential and trace operators ) @xmath1258 as an element of @xmath1259 .
[ rem : sl ] _ ( a ) _ let @xmath1260 be @xmath33-elliptic , then there is a parametrix @xmath1261 which means that @xmath1262 _ ( b ) _ let @xmath60 be elliptic . there is a parametrix @xmath61 of @xmath54 up to compact operators @xmath63 [ thm : parametrix ] _ a ) _ first we have to show that the symbol algebra ( for the completed calculus ) @xmath1263 is closed under inverse .
specifically , let @xmath1264 be @xmath33-elliptic and invertible .
denote by @xmath66 the pseudodifferential operator in the upper left corner of @xmath54 . by @xmath33-ellipticity of @xmath54
we fix a @xmath1265 such that @xmath1266 . set @xmath1267 for a symbol @xmath1268 .
we want to show that there is a an elliptic boundary symbol of the form @xmath1269 such that @xmath1270 . for this we can adapt an approach to the problem the idea of which goes back to boutet de monvel , @xcite ( the index bundle , p.35 ) and which can be found in detail in the reference @xcite .
we divide the argument into three parts . _
i ) _ we have @xmath1271 with constants independent of the parameter @xmath1272 .
then there is a finite dimensional trivial subbundle @xmath562 in @xmath1273 such that @xmath1274 for each @xmath1275 .
we obtain the index element @xmath1276 which depends on the homotopy class of the fredholm family .
let @xmath1277}$ ] be a homotopy of elliptic symbols in @xmath1278 , then @xmath1279 hence for @xmath1255 we obtain @xmath1280 - [ p^{\ast } j_{-}].\ ] ] by the same argument as in @xcite , prop .
11 on page 199 it follows that @xmath1281 _ claim : _ by there is a green symbol @xmath1282 and bundles @xmath1283 such that @xmath1284 for the proof see @xcite , p. 201 .
_ ii ) _ the construction of the boundary symbol is a parameter dependent construction of @xmath144-boundary symbol ( an isomorphism ) @xmath1285 since @xmath1286 is a parametrix of @xmath1287 in the sense of fredholm families we have by part _
i ) _ @xmath1288 - [ p^{\ast } j_{-}].\ ] ] in order to find an isomorphism of the form @xmath1289 it suffices to find a green symbol @xmath1290 with @xmath1291 we know by the claim in part _
i ) _ that there is a green symbol @xmath1282 with @xmath1292 for suitable vector bundles @xmath1293 and @xmath243 such that @xmath1294 - [ { \mathbb{c}}^n ] = [ p^{\ast } j_{+ } ] - [ p^{\ast } j_{-}]\ ] ] by lemma 5 , p. 203 @xcite . for @xmath243 sufficiently large
there are decompositions @xmath1295 for vector bundles @xmath1296 with isomorphisms @xmath1297 where @xmath1298 induces an isomorphism @xmath1299 .
there is a green symbol @xmath1300 so that the isomorphism @xmath1068 is induced by @xmath1301 and that @xmath1302 vanishes on the complement of @xmath1303 in @xmath1273 .
then @xmath1304 is a green symbol with the desired property . _
iii ) _ the inverse of @xmath1305 can be calculated separately for each @xmath1272 ( cf .
2.1.2.4 , prop .
6 , p. 110 ) .
what is left to show is the smoothness of the inverse symbol depending on @xmath1272 .
for @xmath1289 the smoothness is clear .
additionally , the composition of smooth symbols yields smooth symbols .
thus @xmath1306 is in @xmath1307 . hence if we can show that @xmath1308 is @xmath144 we obtain that @xmath1309 is @xmath144
this is a lenghy but elementary calculation for which we refer to prop .
6 , p. 204 - 206 in @xcite .
the parametrix is obtained as follows . _
i ) _ from the exact sequence given in theorem [ thm : exact ] there is a @xmath1261 such that @xmath1310 then by the multiplicativity of the principal symbol and the principal boundary symbol it follows that @xmath1311 applying the exact sequence once more we have that @xmath1312 . hence @xmath1026 is a right parametrix of @xmath54 of order @xmath272 . setting @xmath1313
we obtain @xmath1314 with @xmath1315 .
thus @xmath1316 is a right parametrix of @xmath54 of order @xmath779 .
ii ) _ let @xmath1317 be a sequence of right parametrices of @xmath54 of orders @xmath1318 . by asymptotic completeness we can find @xmath1319 such that @xmath1320 .
hence we have found a right parametrix up to residual terms .
_ iii ) _ fix a right parametrix up to residual terms @xmath1321 of @xmath54 .
analogously to _
i ) _ and _ ii ) _ we can find a left parametrix @xmath1322 of @xmath54 such that @xmath1323 rewrite the operator @xmath1324 as follows @xmath1325 hence @xmath1326 so that @xmath1327 the left hand side of the previous equation equals @xmath1328 and note that this is @xmath1329 .
hence @xmath1322 is also a right parametrix equal up to residual terms to @xmath1330 .
analogously one proves that @xmath1321 is also a left parametrix equal up to residual terms to @xmath1331 .
it follows that @xmath1332 and therefore there is an operator @xmath1261 equal to @xmath1333 and @xmath1334 such that @xmath1335 _ b ) _ consider the pullback algebra @xmath1336 which is defined as the restricted direct sum of @xmath1246 and the indicial algebras @xmath1337 .
we can summarize the situation in the following pullback diagram .
@xmath1338_{\pi_2 } \ar@{->>}[r(1.7)]^-{\pi_1 } & & \overline{\sigma}_{{\mathcal{v}}}^{0,0 } \ar@{- > > } [ d(1.8)]^{(r_f)_{f \in { \mathcal{f}}_1(m ) } } & & \\ & \ar@{->>}[ul]_{\sigma \oplus \sigma_{\partial } ( \oplus_{f } { \mathcal{r}}_f ) } \ar@{->>}[dl]_{\oplus_{f } { \mathcal{r}}_f } \overline{{\mathcal{b}}}_{{\mathcal{v}}}^{0,0 } \ar@{->>}[ur]^{\sigma \oplus \sigma_{\partial } } \ar@{->>}[dr ] _ { } & \\ \oplus_{f \in { \mathcal{f}}_1(x ) } \overline{{\mathcal{b}}}_{{\mathcal{v}}(f)}^{0,0}(f , f_{reg } ) \ar@{->>}[r(1.4)]^-{\oplus_{f \in { \mathcal{f}}_1(x ) } \sigma_f \oplus \sigma_{\partial|f } } & & \oplus_{f \in { \mathcal{f}}_1(x ) } \overline{\sigma}_{{\mathcal{v}}(f)}^{0,0}(f , f_{reg } ) & & } \ ] ] _ claim : _ the following sequence is exact @xmath1339 & { \mathcal{b}}_{{\mathcal{v}}}^{0,0}(x , y ) \ar@{->>}[r(2.0)]^-{\sigma
\oplus \sigma_{\partial } ( \oplus_f { \mathcal{r}}_f ) } & & & \sigma_{bm}. } \ ] ] the surjectivity is immediate . let @xmath1340 , then we prove that @xmath1341 is compact if and only if @xmath1342 and @xmath1343 for each @xmath1202 .
the following is a generalization of the argument in @xcite .
assume that @xmath1344 .
from the short exact sequence in theorem [ thm : exact ] it follows that @xmath1345 .
let @xmath1346 be the collection of boundary defining functions of the faces at infinity of @xmath9 .
setting @xmath1347 we have that @xmath1348 for some @xmath1349 . hence @xmath1350 is continuous . by the generalized kondrachov s theorem ,
see theorem 3.6 in @xcite , we obtain that @xmath1351 is compact .
also @xmath1352 is compact .
we obtain that @xmath1341 is compact .
for the other direction assume that @xmath54 is compact
. then @xmath1353 and @xmath1354 .
assume for contradiction that there is a face @xmath1202 such that @xmath1355 and fix such an @xmath57 .
we identify bijectively @xmath1356 .
the restriction @xmath1201 is defined on the level of families by @xmath1357 and by the hausdorff property of the groupoid @xmath119 we have that @xmath1358 is a continuous function for each @xmath1359 .
let @xmath452 and @xmath1360 be given such that @xmath1361 and let @xmath1362 be extensions to functions in @xmath1363 and @xmath1364 respectively .
then we have @xmath1365 for @xmath1366 .
these restrictions identify as functions @xmath1367 and @xmath1368 via the canonical isometries @xmath1369 .
the supports of @xmath1370 increase via @xmath1371 for @xmath1366 .
this implies weak convergence in sobolev spaces @xmath1372 . by continuity
it follows @xmath1373 this yields a contradiction to the compactness of @xmath54 .
hence the proof of the claim is finished .
set @xmath1374 and denote by @xmath1375 the calkin algebra with quotient map @xmath1376 .
we want to show that @xmath54 is elliptic if and only if @xmath54 is fredholm on the hilbert space @xmath157 . by lemma [ lem : vr ] and
remark [ rem : vr ] we see that the vector representation furnishes an injective @xmath170-representation @xmath1377 an operator @xmath1378 is fredholm if and only if @xmath1379 is invertible .
apply the standard exact sequence for the indicial symbol for order @xmath1380 ( note that by invariance of each singular hyperface @xmath57 the indicial symbol is surjective ) @xmath1381 & { \mathcal{b}}_{{\mathcal{v}}}^{-1,0}(x , y ) \ar@{->>}[r]^-{{\mathcal{r}}_f^{-1 } } & { \mathcal{b}}_{{\mathcal{v}}}^{-1,0}(f , f_{reg } ) = : { \mathcal{b}}_{f}^{-1,0}. } \ ] ] consider the diagram @xmath1382 \ar@{>->}[d ] \ar@{==}[r ] & { \mathcal{k}}({\mathcal{h } } ) \ar@{>->}[d ] \ar@{>->}[r ] & { \mathcal{l}}({\mathcal{h } } ) \ar@{->>}[r]^{q_{{\mathcal{c } } } } & { \mathcal{c}}({\mathcal{h } } ) \\
{ \mathcal{b}}_{{\mathcal{v}}}^{-1,0 } \ar@{->>}[d ] \ar@{>->}[r ] & { \mathcal{b}}_{{\mathcal{v}}}^{0,0 } \ar@{->>}[d]^-{\sigma \oplus \sigma_{\partial } \oplus ( \oplus_f { \mathcal{r}}_f ) } \ar@{>->}[ur]^{\pi } \ar@{->>}[r]^{\sigma \oplus \sigma_{\partial } } & \sigma_{{\mathcal{v } } } & \\
\oplus_f { \mathcal{b}}_{f}^{-1,0 } & \sigma_{bm } & & } \ ] ] since @xmath1383 we obtain with the standard exact sequence in theorem [ thm : exact ] and the above diagram an injective map
@xmath1384 it follows that @xmath1385 is invertible in @xmath1386 if and only if @xmath1387 is pointwise invertible for each @xmath1202 .
we briefly indicate how to state an index problem for lie manifolds with boundary . although we assume here that we are given a boundary structure , we state the problem independently of the calculus .
we expect that the boutet de monvel calculus or more precisely a suitable completion thereof will play a crucial role in the resolution of the problem .
the following approach is a generalization of the statement of the index problem for compact manifolds with boundary from @xcite .
this is based on the tangent groupoid of a. connes , cf .
@xcite .
first we introduce the _ semi - algebroid _ as follows .
we fix an invariant connection @xmath1388 on the algebroid @xmath1389 of the lie manifold with boundary @xmath7 .
define the associated exponential map @xmath1390 and the _ semi - algebroid _ @xmath1391 then construct a semi - groupoid @xmath1392 $ ] which is the analog of connes s tangent groupoid . here
we glue the groupoid @xmath1393 $ ] to the semi - algebroid @xmath1394 . as a set this
is written @xmath1395 \cup \tilde{{\mathcal{a } } } \times \{0\}.\ ] ] note that restricted to the interior we get ( since @xmath1396 by assumption ) @xmath1397 \cup t x_0 \times \{0\}\ ] ] which recovers the tangent groupoid which is also the adiabatic groupoid @xmath1398 .
in general we consider the reduced @xmath1399-algebra @xmath1400 and define the evaluations at @xmath40 and @xmath272 @xmath1401 we make the assumption that @xmath945 is amenable ( see also remark [ rem : vr ] , _ i ) _ ) , hence @xmath1402 .
then we obtain the short exact sequence @xmath1403 \otimes c^{\ast}({\mathcal{g}}^{+ } ) \ar@{>->}[r ] & c_r^{\ast}(\tilde{{\mathcal{g } } } ) \ar@{->>}[r]^{e_0 } & c_r^{\ast}(\tilde{{\mathcal{a } } } ) . } \ ] ] note that @xmath1404 \otimes c^{\ast}({\mathcal{g}}^{+})$ ] is contractible .
then apply the six - term exact sequence in @xmath70-theory .
we deduce from contractibility that the induced maps in @xmath70-theory @xmath1405 are isomorphisms @xmath1406 therefore we can define the analytic index as the map @xmath1407 making the following diagram commute @xmath1408_{{\operatorname{ind}}_a^{+ } } \ar[r]^{(e_0)_{\ast}^{-1 } } & k_0(c_r^{\ast}(\tilde{{\mathcal{g } } } ) ) \ar[dl]_{(e_1)_{\ast } } \\ k_0(c^{\ast}({\mathcal{g}}^{+ } ) ) . } \ ] ] we also obtain an analytic index on the interior pair groupoid with the same technique .
it remains to prove that @xmath1409 generalizes the fredholm index for boundary value problems .
moreover , an appropriate topological interpretation needs to be found .
in particular one should be able to relate the @xmath70-theory of @xmath1410 with the @xmath70-theory of @xmath1411 .
in the special case of a compact manifold with boundary the k - theories are isomorphic as was shown in @xcite .
we recall a definition of the boutet de monvel calculus on which all of the preceding discussions are based .
this particular theorem can be found in the standard reference @xcite , p.227 .
first we set up the necessary notation .
here we are in the setting of a compact manifold with boundary .
this means our discussion here can be seen as an elaboration of example [ exa : bdystruct ] , _ i ) _ of a boundary structure in the simplest case .
let @xmath9 be a compact manifold with boundary @xmath64 and denote by @xmath10 the double . since all the following calculations are local we identify @xmath1412 .
then we have the setup @xmath1413 as pair groupoids , @xmath1414 .
we first recall the symbol calculus for boutet de monvel operators .
let @xmath1415 be the symbols of singular green , trace and potential operators of order @xmath538 and type @xmath40 respectively .
to each symbol we assign a continuous linear operator @xmath1416 since trace operators and potential operators are adjoints of each other , we restrict discussion to trace operators and singular green operators
. it will be more convenient to describe the estimates for the associated _ symbol kernels _
. these are given by @xmath1418 the symbol kernels then are required to satisfy the @xmath954-estimates @xmath1419 for each @xmath1420 and @xmath1421 and @xmath1422 for each @xmath1423 .
we are interested in the boutet de monvel operators as described by their schwartz kernels .
these are defined as follows for @xmath1424 @xmath1425 and for @xmath1426 we have @xmath1427 furthermore , we define the sets @xmath1428 for the trace operators . for the potential operators
@xmath1429 similarly , we have the sets for singular green operators @xmath1430 then we have the following result ( grubb , @xcite , p.227 ) _ i ) _ given a trace operator @xmath1431 of order @xmath538 and type @xmath40 we can extend the integral kernel @xmath1432 to the double @xmath1433 .
call this extended operator and kernel @xmath1434 and @xmath1435 .
then @xmath1435 is a conormal distribution in @xmath1436 which has rapid decay along the normal direction . _
ii ) _ given a potential operator @xmath1437 of order @xmath538 the same procedure gives a conormal distribution @xmath1438 in @xmath1439 .
such that @xmath1438 has rapid decay along the normal direction .
_ iii ) _ a green operator @xmath1440 of order @xmath538 and type zero can be extended to a yield a kernel @xmath1441 in @xmath1442 . with rapid decay along the normal direction .
[ lem : grubb ] although the proof can be found in grubb s book , we include some arguments for the benefit of the reader . _
i ) _ first as in @xcite we apply the seeley extension operator to extend the integral kernel of the trace operator to the double @xmath0 and thereby obtain the extended trace operator @xmath1443 .
write this as @xmath1444 with integral kernel @xmath1445 using plancherel s theorem in as well as the elementary inequality @xmath1446 @xmath1447 this simplifies to @xmath1448 for @xmath1449
. therefore @xmath1450 belongs to @xmath1451 by @xcite , thm .
. then we also obtain that @xmath1452 .
the rapid decay property as @xmath1453 is immediate from the last estimate . _
ii ) _ given a potential operator @xmath70 of order @xmath538 .
this furnishes by the same analysis an extended kernel @xmath1438 contained in the space of conormal distributions @xmath1454 . _
iii ) _ we extend the schwartz kernel @xmath1455 by applying the seeley extension operator in the @xmath1456 direction .
we have the new kernel @xmath1457 with @xmath1458 . then corresponds to @xmath1459 applying the plancherel theorem and the elementary inequality the same way again we obtain the estimate @xmath1460 for each @xmath1461 . by use of the elementary inequality @xmath1462
we obtain from the last estimate @xmath1463 for each @xmath1464 and a constant @xmath1465 . then from @xcite , thm .
18.2.11 we obtain that @xmath1441 is a conormal distribution in @xmath1466 , and for @xmath1467 the kernel decays rapidly . with this we can state the main theorem .
first fix the restriction and extension by zero operators @xmath69 and @xmath68 @xmath1468^{r^{+ } } & \ar@/-0pc/[l]^-{e^{+ } } l^2(\mathring{x } ) } \ ] ] with @xmath1469 and @xmath915 being a projection onto a subspace of @xmath1470 . _
i ) _ an element in the boutet de monvel calculus @xmath1471 is given by @xmath1472 here the operators are entirely described by their schwartz kernels which are conormal distributions ; @xmath1473 and @xmath1474 for the pseudodifferential operators on @xmath0 and the boundary @xmath6 , respectively .
additionally , @xmath66 is required to fulfill the _ transmission property _ ( mapping functions smooth up to @xmath6 to functions also smooth up to @xmath6 ) , and @xmath1475 and rapidly decreasing along @xmath1476 .
furthermore , @xmath1477 is rapidly decreasing along @xmath1478 , and @xmath1479 is rapidly decreasing along @xmath1480 . _
ii ) _ the zero calculus @xmath1481 consisting of extended operators @xmath1482 is closed under composition . furthermore , the truncation @xmath1483 is a linear , surjective operator .
[ thm : bm3 ] _ i ) _ this is the characterization of boutet de monvel s operators in terms of conormal distributions given in the previous lemma .
_ ii ) _ the compositions in the extended calculus are furnished by convolutions of the integral kernels which are conormal distributions .
one can apply the trivial actions on the pair groupoids and the corresponding convolution product as well as a direct calcululation then shows that @xmath1484 yields again a conormal distribution which describes a trace operator .
the other cases can be handled by direct calculation as well .
alternatively , one may invoke the general result due to hrmander @xcite , thm .
the surjectivity of @xmath950 is implied in the proof of lemma [ lem : grubb ] , since we can always seeley extend each operator in @xmath1485 which gives a preimage in @xmath1481 .
in the main part of the paper we have made use of the definition of _ lie manifold with boundary _ as well as the tubular neighborhood theorem . for completeness sake
we briefly recall these notions .
let @xmath1486 be a lie manifold , i.e. a compact manifold with corners @xmath0 endowed with a lie structure ( a @xmath223-module of vector fields , locally finitely generated , projective ) .
by the serre - swan theorem there is a vector bundle @xmath1487 such that @xmath1488 .
this carries the structure of a lie algebroid ( cf .
@xcite ) .
a riemannian metric on @xmath0 is called _ compatible _ if @xmath322 is a suitable restriction of a metric defined on the algebroid @xmath99 such that the following holds : for each @xmath195 there is a neighborhood @xmath1489 in @xmath0 and a local basis of vector fields @xmath1490 from @xmath33 such that @xmath1490 forms an orthonormal frame with regard to the metric @xmath322 restricted to @xmath1489 .
* @xmath1491 is of bounded geometry for a compatible metric @xmath1282 on @xmath307 .
this means @xmath307 has bounded , positive injectivity radius and bounded covariant derivatives of the curvature tensor .
* @xmath1491 is of infinite volume and complete .
next fix the exponential map @xmath1492 coming from the invariant connection defined by the metric on @xmath99 .
notice that from the the splitting in and the identification @xmath1493 we in particular obtain the normal exponential map @xmath1494 the metric @xmath1282 restricts to a product - type metric .
this is just a tubular neighborhood theorem for lie manifolds .
such a lie submanifold with this property is called _ tame _ in the above cited reference and there it is also shown that a submanifold of codimension @xmath272 is always tame .
99 j. aastrup , r. nest , e. schrohe , _ a continuous field of @xmath1399-algebras and the tangent groupoid for manifolds with boundary _ , journal of functional analysis , volume 237 , issue 2 , pages 482 - 506 .
b. ammann , c. carvalho , v. nistor , _ regularity for eigenfunctions of schrdinger operators _ , letters in math .
physics , vol .
101 , issue 1 , pp .
b. ammann , a. ionescu , v. nistor , _ sobolev spaces on lie manifolds and regularity for polyhedral domains _ , doc .
11 , 161 - 206 ( 2006 ) .
b. ammann , r. lauter , v. nistor , _ pseudodifferential operators on manifolds with a lie structure at infinity _ , ann . of math .
165 , 717 - 747 ( 2007 ) .
b. ammann , r. lauter , v. nistor , a. vasy , _ complex powers and non - compact manifolds _ , comm .
part . diff .
5/6 671 - 705 ( 2004 ) .
u. battisti , _ zeta functions of pseudodifferential operators and fourier integral operators on manifolds with boundary _ , phd thesis , hannover 2015 .
k. bohlen , _ boutet de monvel s calculus via groupoid actions _ , phd thesis , hannover 2015 .
louis boutet de monvel , _ boundary problems for pseudo - differential operators _ , acta math . , 126(1 - 2):11 - 51 , 1971 .
a. connes , _ noncommutative geometry _ , academic press , 1994 .
m. crainic , r. l. ferandes , _ integrability of lie brackets _ , ann . of math . , 157 ( 2003 ) , 575 - 620 . c. debord ,
_ holonomy groupoids of singular foliations _ , j. differential geometry , 58 ( 2001 ) , 467 - 500 . c. debord , j .-
lescure , f. rochon , _ pseudodifferential operators on manifolds with fibred corners _ ,
g. grubb , _ functional calculus of pseudodifferential boundary problems _ , birkhuser ; 2nd edition , 1986 .
g. harutjunjan , b .- w .
schulze , _ reduction of orders in boundary value problems without transmission property _
, j. math .
japan , vol .
56 , number 1 ( 2004 ) , 65 - 85 .
l. hrmander , _ the analysis of linear partial differential operators iii _ , springer - verlag , berlin heidelberg , 1985
. l. hrmander , _ the analysis of linear partial differential operators iv _ , springer - verlag , berlin heidelberg , 1985 .
t. krainer , _ elliptic boundary problems on manifolds with polycylindrical ends _ , journal of functional analysis vol . 244 issue 2 march 15 , 2007 .
p. 351 - 386 .
r. lauter , b. monthubert , v. nistor , _ pseudodifferential analysis on continuous family groupoids _ , documenta math .
5 ( 2000 ) , 625 - 655 .
r. lauter , b. monthubert , v. nistor , _ spectral invariance for certain algebras of pseudodifferential operators _ , j. de linst .
jussieu 4 ( 2005 ) , issue 03 , 405 - 442 .
r. lauter , v. nistor , _ analysis of geometric operators on open manifolds : a groupoid approach _
, progress in mathematics , vol .
198 , 2001 , pp 181 - 229 .
r. melrose , _ the atiyah - patodi - singer index theorem _ , a. k. peters , ltd . ,
boston , mass . , 1993 .
i. moerdijk , j. mrcun , _ introduction to foliations and lie groupoids _ , cambridge univ .
press , 2003 .
i. moerdijk , j. mrcun , _ on the integrability of lie subalgebroids _ , adv . in math .
204 , issue 1 , 2006 , 101 - 115 .
b. monthubert , _ pseudodifferential calculus on manifolds with corners and groupoids _ , proc .
127 , 10 , 2871 - 2881 .
p. s. muhly , _ coordinates in operator algebra _ ,
v. nistor , _ pseudodifferential operators on non - compact manifolds and analysis on polyhedral domains _ , proceedings of the workshop on spectral geometry of manifolds with boundary and decomposition of manifolds , roskilde university , 307328 , contemporary mathematics , ams , rhode island , 2005
. v. nistor , a. weinstein , p. xu , _ pseudodifferential operators on differential groupoids _ , pacific j. math . 189
( 1999 ) , 117 - 152 .
a. l. t. paterson , _
groupoids , inverse semigroups , and their operator algebras _ , progress in mathematics vol .
170 , birkhuser , 1999 .
a. l. t. paterson , _ the equivariant analytic index for proper groupoid actions _ , k - theory , 32(2004 ) , 198 - 230 .
s. rempel , b .- w .
schulze , _ index theory of elliptic boundary problems _ , math .
lehrbcher monogr .
ii abt . math .
55 , akademie - verlag , berlin , 1982 .
e. schrohe , b .- w .
schulze , _ boundary value problems in boutet de monvel s algebra for manifolds with conical singularities i _ , advances in partial differential equations 1 , pp .
s. r. simanca , _ pseudo - differential operators _ , pitman research notes in mathematics 236 , 1990 . | we consider general pseudodifferential boundary value problems on a lie manifold with boundary .
this is accomplished by constructing a suitable generalization of the boutet de monvel calculus for boundary value problems .
the data consists of a compact manifold with corners @xmath0 which is endowed with a lie structure of vector fields @xmath1 , a so - called lie manifold as introduced by bernd ammann , robert lauter and victor nistor .
the lie manifold @xmath0 is split into two equal parts @xmath2 and @xmath3 each of which are lie manifolds which intersect in an embedded hypersurface @xmath4 . in this setup
our goal is to describe a transmission boutet de monvel calculus for boundary value problems .
starting with the example of @xmath5-vector fields we show that there are two groupoids integrating the lie structure on @xmath0 and @xmath6 respectively which form a bimodule structure ( a groupoid correspondence ) and in mild cases these groupoids are isomorphic inside the category of lie groupoids ( morita equivalent ) . with the help of the bimodule structure and
canonically defined manifolds with corners , which are blow - ups in particular cases , we define a class of extended boutet de monvel operators .
then we describe the restricted transmission boutet de monvel calculus by truncation of the extended operators .
we define the representation for restricted operators and show closedness under composition with the help of an analog of the ammann , lauter , nistor representation theorem .
finally , we analyze the parametrix construction and in the last section state the index problem for boundary value problems on lie manifolds .
* keywords : * boutet de monvel s calculus , groupoid , lie manifold . |
one major obstacle to the fulfillment of the promise of quantum computing is the current scarcity of quantum algorithms .
quantum computing researchers simply have not yet found enough quantum algorithms to determine whether or not future quantum computers will be general purpose or special purpose computing devices . as a result , much more research
is crucially needed to determine the algorithmic limits of quantum computing .
one of the most promising and versatile approaches to creating new quantum algorithms is based on the quantum hidden subgroup ( qhs ) paradigm , originally suggested by alexei kitaev @xcite .
this class of quantum algorithms encompasses the deutsch - jozsa , simon , shor algorithms , and many more . in this paper ,
our strategy for finding new quantum algorithms is to decompose shor s quantum factoring algorithm into its basic primitives , then to generalize these primitives , and finally to show how to reassemble them into new qhs algorithms .
taking an `` alphabetic building blocks approach , '' we will use these primitives to form an `` algorithmic toolkit '' for the creation of new quantum algorithms , such as wandering shor algorithms , continuous shor algorithms , the quantum circle algorithm , the dual shor algorithm , a qhs algorithm for feynman integrals , free qhs algorithms , and more . toward the end of this paper ,
we show how grover s algorithm is most surprisingly almost a qhs algorithm , and how this suggests the possibility of an even more complete `` algorithmic tookit '' beyond the qhs algorithms .
before discussing how shor s algorithm can be decomposed into its primitive components , let s take a quick look at an example of the execution of shor s factoring algorithm . as we discuss this example , we suggest that the reader , as an exercise , try to find the basic qhs primitives that make up this algorithm .
can you see them ?
shor s quantum factoring algorithm reduces the task of factoring a positive integer @xmath0 to first finding a random integer @xmath1 relatively prime to @xmath0 , and then next to determining the period @xmath2 of the following function @xmath3{ccl}\mathbb{z } & \overset{\varphi}{\longrightarrow } & \mathbb{z}\operatorname{mod}n\\ x & \longmapsto & a^{x}\operatorname{mod}n\text { , } \end{array}\ ] ] where @xmath4 denotes the additive group of integers , and where @xmath5 denotes the integers @xmath6 under multiplication with @xmath7 is found by selecting a random integer , and then applying the euclidean algorithm to determine whether or not it is relatively prime to @xmath0 .
if not , then the @xmath8 is a non - trivial factor of @xmath0 , and there is no need to proceed futher .
however , this possibility is highly unlikely if @xmath0 is large . ] .
since @xmath4 is an infinite group , shor chooses to work instead with the finite additive cyclic group @xmath9 of order @xmath10 , where @xmath11 and with the `` approximating '' map @xmath3{ccll}\mathbb{z}_{q } & \overset{\widetilde{\varphi}}{\longrightarrow } & \mathbb{z}\operatorname{mod}n & \\ x & \longmapsto & a^{x}\operatorname{mod}n\text { , } & 0\leq x < q \end{array}\ ] ] shor begins by constructing a quantum system with two quantum registers @xmath12 the left intended for holding the arguments @xmath13 of @xmath14 , the right for holding the corresponding values of @xmath14 .
this quantum system has been constructed with a unitary transformation @xmath15 implementing the approximating map @xmath14 . as an example , let us use shor s algorithm to factor the integer @xmath16 , assuming that @xmath17 has been randomly chosen .
thus , @xmath18 .
unknown to us , the period is @xmath19 , and hence , @xmath20 .
we proceed by executing the following steps : * initialize@xmath21 * apply the inverse fourier transform @xmath22 to the left register , where @xmath23 is a primitive @xmath24-th root of unity , to obtain @xmath25 * apply the unitary transformation @xmath26 to obtain @xmath27 * apply the fourier transform @xmath28 to the left register to obtain @xmath29 where @xmath30 * measure the left register .
then with probability @xmath31 the state will `` collapse '' to @xmath32 with the value measured being the integer @xmath33 , where @xmath34 . a plot of @xmath35 is shown in figure 1 .
( see @xcite and @xcite for details .
) fig1.ps + * figure 1 . a plot of * @xmath36**. *
* the peaks in the above plot of @xmath35 occur at the integers @xmath37 the probability that at least one of these six integers will occur is quite high .
it is actually @xmath38 .
indeed , the probability distribution has been intentionally engineered to make the probability of these particular integers as high as possible .
and there is a good reason for doing so .
the above six integers are those for which the corresponding rational @xmath39 is closest to a rational of the form @xmath40 . by closest we mean that @xmath41 in particular , @xmath42 are rationals respectively closest to the rationals @xmath43 the six rational numbers @xmath44 are `` closest '' in the sense that they are convergents of the continued fraction expansions of @xmath45 , respectively . hence , each of the six rationals @xmath44 can be found using the standard continued fraction recursion formulas .
but ... , we are not searching for rationals of the form @xmath40 . instead
, we seek only the denominator @xmath19 .
unfortunately , the denominator @xmath19 can only be obtained from the continued fraction recursion when the numerator and denominator of @xmath40 are relatively prime .
given that the algorithm has selected one of the random integers @xmath46 , the probability that the corresponding rational @xmath40 has relatively prime numerator and denominator is @xmath47 , where @xmath48 denotes the euler phi ( totient ) function .
so the probability of finding @xmath19 is actually not @xmath38 , but is instead @xmath49 .
as it turns out , if he repeats the algorithm @xmath50 times , we will obtain the desired period @xmath2 with probability bounded below by approximately @xmath51 . however , this is not the end of the story .
once we have in our possession a candidate @xmath52 for the actual period @xmath19 , the only way we can be sure we have the correct period @xmath2 is to test @xmath52 by computing @xmath53 .
if the result is @xmath54 , we are certain we have found the correct period @xmath2 .
this last part of the computation is done by the repeated squaring algorithm via the expression @xmath55 where @xmath56 is the radix 2 expansion of @xmath52 . ] .
now that we have taken a quick look at shor s algorithm , let s see how it can be decomposed into its primitive algorithmic components .
we will first need to answer the following question : _ what is a quantum hidden subgroup algorithm ? _ but before we can answer the this question , we need to provide an answer to an even more fundamental question : _ what is a hidden subgroup problem ?
_ a map @xmath57 from a group @xmath58 into a set @xmath59 is said to have * hidden subgroup structure * if there exists a subgroup @xmath60 of @xmath58 , called a * hidden subgroup * , and an injection @xmath61 , called a * hidden injection * , such that the diagram@xmath3{ccc}g & \overset{\varphi}{\longrightarrow } & s\\ \nu\searrow & & \nearrow\iota_{\mathbf{\varphi}}\\ & g / k_{\mathbf{\varphi } } & \end{array}\ ] ] is commutative .
the notion generalizes in an obvious way to more complicated diagrams . ] , where @xmath62 denotes the collection of right cosets of @xmath63 in @xmath58 , and where @xmath64 is the natural surjection of @xmath58 onto @xmath65 .
we refer to the group @xmath58 as the * ambient group * and to the set @xmath59 as the * target set*. if @xmath60 is a normal subgroup of @xmath58 , then @xmath66 is a group , called the * hidden quotient group * , and @xmath67 is an epimorphism , called the * hidden epimorphism*. we will call the above diagram the * hidden subgroup structure * of the map @xmath68 .
( see @xcite,@xcite . )
the underlying intuition motivating this formal definition is as follows : given a natural surjection ( or epimorphism ) @xmath64 , an `` archvillain with malice of forethought '' hides the algebraic structure of @xmath69 by intentionally renaming all the elements of @xmath65 , and `` maliciously tossing in for good measure '' some extra elements to form a set @xmath59 and a map @xmath57 .
the hidden subgroup problem can be stated as follows : * hidden subgroup problem ( hsp ) . *
@xmath68 _ be a map with hidden subgroup structure .
the problem of determining a hidden subgroup _
@xmath63 _ of _ @xmath58 _ is called a * hidden subgroup problem ( hsp)*. an algorithm solving this problem is called a * hidden subgroup algorithm*. _ the corresponding quantum form of this hsp is stated as follows : * hidden subgroup problem ( quantum version ) . *
@xmath57 _ be a map with hidden subgroup structure . construct a quantum implementation of the map _
@xmath70 _ as follows : let _ @xmath71 _ and _ @xmath72 _ be hilbert spaces defined respectively by the orthonormal bases _
@xmath73__and _ _ @xmath74 _ and let _
@xmath75 _ _ , where _ _ @xmath54 _ denotes the identity .
] of the ambient group _
@xmath76__. finally , let _ _
@xmath77 _ _ be a unitary transformation such that__@xmath3{ccc}\mathcal{h}_{g}\otimes\mathcal{h}_{s } & \longrightarrow & \mathcal{h}_{g}\otimes\mathcal{h}_{s}\\ \left\vert g\right\rangle \left\vert s_{0}\right\rangle & \mapsto & \left\vert g\right\rangle \left\vert \varphi(g)\right\rangle \end{array}\ ] ] _ determine the hidden subgroup _ @xmath63 _ with bounded probability of error by making as few queries as possible to the blackbox _
@xmath78__. a quantum algorithm solving this problem is called a * quantum hidden subgroup ( qhs ) algorithm*. _ _
we are now in a position to construct one of the fundamental algorithmic primitives found in shor s algorithm .
let @xmath57 be a map from a group @xmath58 to a set @xmath59 with hidden subgroup structure .
we assume that all representations of @xmath58 are equivalent to unitary representations .
let @xmath79 denote a * complete set of distinct irreducible unitary representations * of @xmath58 .
using multiplicative notation for @xmath58 ,
we let @xmath54 denote the * identity * of @xmath58 , and let @xmath80 denote its image in @xmath59 .
finally , let @xmath81 denote the * trivial representation * of @xmath58 .
if @xmath58 is abelian , then @xmath82 becomes the * dual group * of characters .
the generic qhs algorithm is given below : quantum subroutine qrand@xmath83 * initialization@xmath84 * application of the inverse fourier transform @xmath85of @xmath58 to the left register@xmath86 where @xmath87 denotes the cardinality of the group @xmath58 .
* application of the unitary transformation @xmath88@xmath89 * application of the fourier transform @xmath90 of @xmath58 to the left register@xmath91 where @xmath92 denotes the degree of the representation @xmath93 , where @xmath94 denotes the contragradient representation ( i.e. , @xmath95 ) , where @xmath96 , and where @xmath97 . * measurement of the left quantum register with respect to the orthonormal basis@xmath98 thus , with probability@xmath99 the resulting measured value is the entry @xmath100 , and the quantum system `` collapses '' to the state@xmath101 * step 5 .
output @xmath102 , and stop .
but shor s algorithm consists of more than the primitive qrand . for many ( but not all ) hidden subgroup problems ( hsps ) @xmath68 , the corresponding generic qhs algorithm qrand either is not physically implementable or is too expensive to implement physically .
for example , the hsp @xmath70 is usually not physically implementable if the ambient group is infinite ( e.g. , @xmath58 is the infinite cyclic group @xmath4 ) , and is too expensive to implement if the ambient group is too large ( e.g. , @xmath58 is the symmetric group @xmath103 ) . in this case , there is a standard generic way of `` tweaking '' the hsp to get around this problem , which we will call * pushing*. let @xmath57 be a map from a group @xmath58 to a set @xmath59 .
a map @xmath104 from a group @xmath105 to the set @xmath59 is said to be a * push * of @xmath70 , written @xmath106 provided there exists an epimorphism @xmath107 from @xmath58 onto @xmath105 , and a transversal be an epimorphism from a group @xmath76 to a group @xmath108 .
then a transversal @xmath109 of @xmath69 is is a map @xmath110 such that @xmath111 is the identity map @xmath112 .
( it immediately follows that @xmath109 is an injection . ) in othere words , a transversal @xmath109 of an epimorphism @xmath69 is a map which maps eacl element @xmath113 of @xmath108 to an element of @xmath76 contained in the coset @xmath113 , i.e. , to a coset representative of @xmath113 . ]
@xmath114 of @xmath69 such that @xmath115 , i.e. , such that the following diagram is commutative@xmath3{ccc}g & \overset{\varphi}{\longrightarrow}\quad & s\\ \quad\uparrow\tau & \quad\nearrow\widetilde{\varphi}\quad & \\ \widetilde{g } & & \end{array}\ ] ] if the epimorphism @xmath116 and the transversal @xmath109 are chosen in an appropriate way , then execution of the generic qhs subroutine with input @xmath117 , i.e. , execution of @xmath118 will with high probability produce an irreducible representation @xmath119 of the group @xmath105 which is sufficiently close to an irreducible representation @xmath93 of the group @xmath58 .
if this is the case , then there is a polynomial time classical algorithm which upon input @xmath119 produces the representation @xmath93 .
obviously , much more can be said about pushing .
but unfortunately that would take us far afield from the objectives of this paper .
for more information on pushing , we refer the reader to @xcite
. it would be amiss not to mention that the above algorithmic primitive of pushing suggests the definition of a second primitive which we will call * lifting*. let @xmath57 be a map from a group @xmath58 to a set @xmath59 .
a map @xmath120 from a group @xmath121 to the set @xmath59 is said to be a * lift * of @xmath70 , written @xmath122 provided there exists an morphism @xmath123 from to @xmath58 such that @xmath124 , i.e. , such that the following diagram is commutative@xmath3{cl}\underline{g } & \\ \ \
\eta\downarrow\quad & \quad\searrow\underline{\varphi}\quad\\ g & \quad\overset{\varphi}{\longrightarrow}\quad s \end{array}\ ] ] fig2.eps + * figure 2 . pushing and lifting hsps . *
we are now in position to describe shor s algorithm in terms of its primitive components . in particular , we are now in a position to see that shor s factoring algorithm is a classic example of a qhs algorithm created from the push of an hsp .
let @xmath0 be the integer to be factored .
let @xmath4 denote the additive group of integers , and @xmath125 denote the integers @xmath6 under multiplication .
shor s algorithm is a qhs algorithm that solves the following hsp @xmath3{rrc}\varphi:\mathbb{z } & \longrightarrow & \mathbb{z}_{n}^{\times}\\ m & \longmapsto & a^{m}\operatorname{mod}n \end{array}\ ] ] with unknown hidden subgroup structure given by the following commutative diagram @xmath3{ccc}\mathbb{z } & \overset{\varphi}{\longrightarrow } & \mathbb{z}_{n}^{\times}\\ \nu\searrow & & \nearrow\iota\\ & \mathbb{z}/p\mathbb{z } & \end{array } \text { \ , } \ ] ] where @xmath1 is an integer relatively prime to @xmath0 , where @xmath2 is the hidden integer period of the map @xmath126 , where @xmath127 is the additive subgroup of all integer multiples of @xmath2 ( i.e. , the hidden subgroup ) , where @xmath128 is the natural epimorpism of the integers onto the quotient group @xmath129 ( i.e. , the hidden epimorphism ) , and where @xmath130 is the hidden monomorphism .
an obstacle to creating a physically implementable algorithm for this hsp is that the domain @xmath4 of @xmath70 is infinite .
as observed by shor , a way to work around this difficulty is to push the hsp .
in particular , as illustrated by the following commutative diagram@xmath3{ccl}\mathbb{z\qquad } & \overset{\varphi}{\longrightarrow } & \qquad\mathbb{z}_{n}^{\times}\\ \mu\searrow\nwarrow\tau & & \nearrow\varphi = push\left ( \varphi\right ) = \varphi\circ\tau\\ & \mathbb{z}_{q } & \end{array } \text { \ \ , } \ ] ] a push @xmath117 is constructed by selecting the epimorphism @xmath131 of @xmath4 onto the finite cyclic group @xmath9 of order @xmath132 , where the integer @xmath132 is the unique power of @xmath133 such that @xmath134 , and then choosing the transversal is an injection @xmath135 such that @xmath136 is the identity map on @xmath9 , i.e. , a map that takes each element of @xmath9 onto a coset representative of the element in @xmath4 . ]
@xmath3{rrc}\tau:\mathbb{z}_{q } & \longrightarrow & \mathbb{z}\\ m\operatorname{mod}q & \longmapsto & m \end{array } \text { \ , } \ ] ] where @xmath137 .
_ this push _
@xmath138 _ is called _ * shor s oracle*. shor s algorithm consists in first executing the quantum subroutine qrand@xmath139 , thereby producing a random character @xmath140 of the finite cyclic group @xmath9 .
the transversal @xmath109 used in pushing has been engineered in such a way as to assure that the character @xmath141 is sufficiently close to a character @xmath142 of the hidden quotient group @xmath143 . in this case `` sufficiently close '' means that @xmath144 which means that @xmath40 is a continued fraction convergent of @xmath39 , and thus can be found found by the classical polynomial time continued fraction algorithm and @xmath145 can in the obvious way be identified with points of in the unit circle in the complex plane . with this identification
, we can see that this inequalty is equivalent to saying the the chordal distance bewteen these two rational points on the unit circle is less than or equan to @xmath146 .
hence , shor s algorithm is using the topology of the unit circle . ] .
now let s use the primitives described in sections 3 , 4 , and 5 to create other new qhs algorithms , called wandering shor algorithms .
wandering shor algorithms are essentially qhs algorithms on free abelian finite rank @xmath147 groups @xmath76 which , with each iteration , first select a random cyclic direct summand @xmath4 of the group @xmath76 , and then apply one iteration of the standard shor algorithm to produce a random character of the approximating finite group @xmath148 , called a * * group probe * * , we mean an epimorphic image of the ambient group @xmath76 . ] .
three different wandering shor algorithms are created in @xcite .
the first two wandering shor algorithms given in @xcite are quantum algorithms which find the order @xmath2 of a maximal cyclic subgroup of the hidden quotient group @xmath149 .
the third computes the entire hidden quotient group @xmath149 .
the first step in creating a wandering shor algorithm is to find the right generalization one of the primitives found in shor s algorithm , namely , the transversal @xmath150 of shor s factoring algorithm . in other words ,
we need to construct the `` correct '' generalization of the transversal from @xmath9 to a free abelian group @xmath76 of rank @xmath147 . for this reason ,
we have created the following definition : let @xmath76 be the free abelian group of rank @xmath147 , let @xmath151 onto the cyclic group @xmath9 of order @xmath132 with selected generator @xmath152 .
a transversal be an epimorphism from a group @xmath76 to a group @xmath108 .
then a transversal @xmath109 of @xmath69 is is a map @xmath110 such that @xmath153 is the identity map @xmath154 .
( it immediately follows that @xmath109 is an injection . ) in othere words , a transversal @xmath109 of an epimorphism @xmath69 is a map which maps eacl element @xmath113 of @xmath108 to an element of @xmath76 contained in the coset @xmath113 , i.e. , to a coset representative of @xmath113 . ]
@xmath155 of @xmath69 is said to be a * shor transversal * provided that : * @xmath156 for all @xmath157 * for each ( free abelian ) basis @xmath158 of @xmath76 , the coefficients @xmath159 of @xmath160 satisfy @xmath161 .
later , when we construct a generalization of shor transversals to free groups of finite rank @xmath147 , we will see that the first condition simply states that a shor transversal is nothing more than a 2-sided schreier transversal .
the second condition of the above definition simply says that @xmath162 maps the generator @xmath152 of @xmath9 onto a generator of a free direct summand @xmath4 of @xmath76 .
( for more details , please refer to section 12 of this paper . ) in @xcite , we show how to use the extended euclidean algorithm to construct the epimorphism @xmath163 and the transversal @xmath155
. fig3.eps + * figure 3 .
flowchart for the first wandering shor algorithm ( a.k.a . , a vintage shor algorithm ) .
this algorithm finds the order * @xmath2 * * of a maximal cyclic subgroup of the hidden quotient group * * @xmath149**. * * flow charts for the three wandering shor algorithms created in @xcite are given in figures 3 through 5 . in @xcite ,
these were also called * vintage shor algorithms*. fig4.eps + * figure 4 .
flowchart for the second wandering shor algorithm ( a.k.a . , a vintage shor algorithm ) .
this algorithm finds the order * @xmath2 * * of a maximal cyclic subgroup of the hidden quotient group * * @xmath149**. * * fig5.eps + * * figure 5 .
flowchart for the third wandering shor algorithm , a.k.a . , a vintage shor algorithm .
this algorithm finds the entire hidden quotient group * * @xmath149*. * the algorithmic complexities of the above wandering shor algorithms is given in @xcite .
for example , the first wandering shor algorithm is of time complexity@xmath164 where @xmath147 is the rank of the free abelian group @xmath76 .
this can be readily deduced from the abbreviated flowchart given in figure 6 .
fig6.ps + * figure 6 .
abbreviated flowchart for the first wandering shor algorithm . *
in in @xcite and in @xcite , the algorithmic primitives found in above sections of this paper were used to create a class of algorithms called continuous shor algorithms . by a * continuous variable shor algorithm * ,
we mean a quantum hidden subgroup algorithm that finds the hidden period @xmath2 of an admissible function @xmath165 from the reals @xmath166 to itself . by an admissible function , we mean a function belonging to any sufficiently well behaved class of functions .
for example , the class of functions which are lebesgue integrable on every closed interval of @xmath166 . there , are many other classes of functions that work equally as well .
actually , the papers @xcite , @xcite give in succession three such continuous shor algorithms , each successively more general than the previous .
for the first algorithm , we assume that the unknown hidden period @xmath2 is an integer .
the algorithm is then constructed by using rigged hilbert spaces@xcite , @xcite , linear combinations of dirac delta functions , and a subtle extension of the fourier transform found in the generic qhs subroutine qrand@xmath167 , which has been described previously in section 4 of this paper . in step 5 of qrand@xmath167 ,
the observable@xmath168 is measured , where @xmath132 is an integer chosen so that @xmath169 .
it then follows that the output of this algorithm is a rational @xmath170 which is a convergent of the continued fraction expansion of a rational of the form @xmath171 .
the above quantum algorithm is then extended to a second quantum algorithm that finds the hidden period @xmath2 of functions @xmath165 , where the unknown period @xmath2 is a rational .
finally , the second algorithm is extended to a third algorithm which finds the hidden period @xmath2 of functions @xmath165 , when @xmath2 is an _ arbitrary real number_. we point out that for the third and last algorithm to work , we must impose a very restrictive condition on the map @xmath165 , i.e. , the condition that the map @xmath70 is continuous .
we have shown in previous sections how the mathematical primitives of pushing and lifting can be used to create new quantum algorithms . in particular , we have described how pushing and lifting can be used to derive new hsps from an hsp @xmath57 on an arbitrary group @xmath58 . we now see how group duality can be exploited by these two primitives to create even more quantum algorithms
. fig7.eps + * figure 7 . using duality to create new qhs algorithms . * to this end
, we assume that @xmath58 is an _ abelian _ group .
hence , its dual group of characters @xmath79 exists is non - abelian , then its dual is not a group , but instead the representation algebra @xmath172 over the group ring @xmath173 . the methods described in this section can also be used to create new quantum algorithms for hsps @xmath174 on on the representation algebra @xmath172 . ] .
it now follows that pushing and lifting can also be used to derive new hsps from an arbitrary hsp @xmath175 on the dual group @xmath79 . in @xcite
, this method is used to create a number of new quantum algorithms derived from shor - like hsps @xmath176 .
a roadmap is shown in figure 8 of the developmental steps taken to find and to create a new qhs algorithm on @xmath9 , which is ( in the sense described below ) dual to shor s original algorithm .
we call the algorithm developed in the final step of figure 8 the * dual shor algorithm*. fig8.eps + * figure 8
. roadmap for creating the dual shor algorithm . * as indicated in figure 5 , our first step is to create an intermediate qhs algorithm based on a shor - like hsp @xmath176 from the additive group of integers @xmath4 to a target set @xmath59 .
the resulting algorithm `` lives '' in the infinite dimensional space @xmath177 defined by the orthonormal basis @xmath178 .
this is a physically unemplementable quantum algorithm created as a first steping stone in our algorithmic development sequence .
intuitively , this algorithm can be viewed as a `` distillation '' or a `` purification '' of shor s original algorithm . as a next step ,
* duality * is used to create the * quantum circle algorithm*. this is accomplished by devising a qhs algorithm for an hsp @xmath179 on the dual group @xmath180 of the additive group of integers @xmath4 .
( by @xmath180 , we mean the * additive group of reals * @xmath181 , which is isomorphic to the multiplicative group @xmath182 , i.e. , the * unit circle * in the complex plane . ) once again , this is probably a physcally unemplementable quantum algorithm .
but its utility lies in the fact that it leads to the physically implementable quantum algorithm created in the last and final developmental step , as indicated in figure 8 . for in the final step ,
a physically implementable qhs algorithm is created by * lifting * the hsp @xmath179 to an hsp @xmath183 . for the obvious reason , we call the resulting algorithm a * dual shor algorithm*. for detailed descriptions of each of these quantum algorithms , i.e. ,
the `` distilled '' shor , the quantum circle , and the dual shor algorithms , the reader is referred to @xcite and @xcite .
we give below brief descriptions of the quantum circle and the dual shor algorithms .
for the * quantum circle algorithm * , we make use of the following spaces ( each of which is used in quantum optics ) : * the rigged hilbert space @xmath184 with orthonormal basis @xmath185 . by orthonormal we mean that @xmath186 , where @xmath187 denotes the dirac delta function .
the elements of @xmath184 are * formal integrals * of the form @xmath188 .
( the physicist dirac in his classic book@xcite on quantum mechanics refers to these integrals as infinite sums .
see also @xcite and @xcite . ) * the complex vector space @xmath189 of formal sums@xmath190 with orthonormal basis @xmath191 . by orthonormal we mean that @xmath192 , where @xmath193 denotes the kronecker delta .
we can now design an algorithm which solves the following hidden subgroup problem : * hidden subgroup problem for the circle . * _ let _ @xmath194 _ be an admissible function from the circle group _ @xmath180 _ to the complex numbers _ @xmath195 _ with hidden rational period _ @xmath196 _ _ , where _ _ @xmath197 _ denotes the rational circle , i.e. , the rationals _
@xmath198__. _ _ by an admissible function , we mean a function belonging to any sufficiently well behaved class of functions .
for example , the class of functions which are lebesgue integrable on @xmath180 .
there , are many other classes of functions that work equally as well . if @xmath199 ( with @xmath200 ) is a rational period of a function @xmath201 , then @xmath202is also a period of @xmath203 .
hence , the minimal rational period of @xmath203 is always a reciprocal integer @xmath198 .
the following quantum algorithm finds the reciprocal integer period of the function @xmath203 .
circle - algorithm@xmath204 * initialization@xmath205 * application of the inverse fourier transform @xmath206@xmath207 * step 2 .
application of the unitary transformation @xmath208@xmath209 * application of the fourier transform @xmath210 remark . letting @xmath211 , we have@xmath212 where @xmath213 is the unknown reciprocal period .
but@xmath214{ll}a & \text{if \ }
n=0\operatorname*{mod}a\\ 0 & \text{otherwise}\end{array } \right.\ ] ] hence,@xmath215 * measurement of@xmath216 with respect to the observable@xmath217 to produce a random eigenvalue @xmath218 .
the above quantum circle algorithm can be extended to a quantum algorithm which finds the hidden period @xmath219 of a function @xmath194 , when @xmath219 is an arbitrary real number @xmath198 .
but in creating this extended quantum algorithm , a very restrictive condition must be imposed on the map @xmath194 , namely , the condition that @xmath203 be continuous .
we now give a brief description of the * dual shor algorithm*. the dual shor algorithm is a qhs algorithm created by making a discrete approximation of the quantum circle algorithm .
more specifically , it is created by lifting the qhs circle algorithm for @xmath220 to the finite cyclic group @xmath9 , as illustrated in the commutative diagram given below:@xmath3{ll}\;\,\mathbb{z}_{q } & \\ \mu\downarrow\;\ ; & \;\;\searrow\widetilde{\varphi}=push\left ( \varphi \right ) = \varphi\circ\mu\\ \mathbb{r}\mathbf{/}\mathbb{z } & \longrightarrow\;\;s \end{array}\ ] ] intuitively , just as in shor s algorithm , the circle group @xmath180 is `` approximated '' with the finite cyclic group @xmath9 , where the group @xmath9 is identified with the additive group@xmath221 and where the hidden subgroup @xmath222 is identified with the additive group@xmath223 with @xmath224 .
this is a physically implementable quantum algorithm . in a certain sense
, it is actually faster than shor s algorithm . for the last step of shor s algorithm
uses the standard continued fraction algorithm to determine the unknown period . on the other hand ,
the last step of the dual shor algorithm uses the much faster euclidean algorithm to compute the greatest common divisor of the integers @xmath225 , thereby determining the desired reciprocal integer period @xmath213 . for more details
, please refer to @xcite and @xcite .
we now discuss a qhs algorithm based on feynman path integrals .
this quantum algorithm was developed at the mathematical sciences research institute ( msri ) in berkeley , california when one of the authors of this paper was challenged with an invitation to give a talk on the relation between feynmann path integrals and quantum computing at an msri conference on feynman path integrals . until recently , both authors of this paper thought that the quantum algorithm to be described below was a highly speculative quantum algorithm .
for the existence of feynman path integrals is very difficult ( if not impossible ) to determine in a mathematically rigorous fashion .
but surprisingly , jeremy becnel in his doctoral dissertation @xcite actually succeeded in creating a firm mathematical foundation for this algorithm .
we should mention , however , that the physical implementability of this algorithm is still yet to be determined . definition .
let paths be the real vector space of all continuous paths @xmath226 \longrightarrow\mathbb{r}^{n}$ ] which are @xmath227 with respect to the inner product@xmath228 with scalar multiplication and vector sum defined as * @xmath229 * @xmath230 we wish to create a qhs algorithm for the following hidden subgroup problem : * hidden subgroup problem for * paths .
@xmath231paths@xmath232 _ be a functional with a hidden subspace _
@xmath233 _ of _ paths _ _ such that__@xmath234 our objective is to create a qhs algorithm which solves the above problem , i.e. , which finds the hidden subspace @xmath233 .
let @xmath235 be the rigged hilbert space with orthonormal basis @xmath236 , and with bracket product @xmath237
. we will use the following observation to create the qhs algorithm : * observation .
* paths@xmath238 _ , where _ @xmath239 _ denotes the orthogonal complement of the hidden vector subspace _
@xmath233__. _ _ the qhs algorithm for feynman path integral is given below :
feynman@xmath83 * initialize@xmath240 * apply @xmath206@xmath241 * apply @xmath242@xmath243 * apply @xmath210@xmath3{lll}\left\vert \psi_{3}\right\rangle & = & { \displaystyle\int\limits_{\text{\textsc{paths } } } } \mathcal{d}y{\displaystyle\int\limits_{\text{\textsc{paths } } } } \mathcal{d}x\ e^{-2\pi ix\cdot y}\left\vert y\right\rangle \left\vert \varphi\left ( x\right ) \right\rangle \\ & = & { \displaystyle\int\limits_{\text{\textsc{paths } } } } \mathcal{d}y\ \left\vert y\right\rangle { \displaystyle\int\limits_{\text{\textsc{paths } } } } \mathcal{d}x\ e^{-2\pi ix\cdot y}\left\vert \varphi\left ( x\right ) \right\rangle \end{array}\ ] ] but @xmath244 however,@xmath245 so , @xmath246 * measure@xmath247 with respect to the observable@xmath248 to produce a random element of @xmath239 the above algorithm suggests an intriguing question .
can the above qhs feynman integral algorithm be modified in such a way as to create a quantum algorithm for the jones polynomial ?
in other words , can it be modified by replacing paths with the space of gauge connections , and making suitable modifications ?
this question is motivated by the fact that the integral over gauge transformations@xmath249 looks very much like a fourier transform , where@xmath250 denotes the * wilson loop * over the knot @xmath251 .
in this and the following section of this paper , our objective is to show that a free group is the the most natural domain for qhs algorithms .
in retrospect , this is not so surprising if one takes a discerning look at shor s factoring algorithm . for in section 6
, we have seen that shor s algorithm is essentially a qhs algorithm on the free group @xmath4 which has been pushed onto the finite group @xmath9 . in particular ,
let @xmath57 be a map with hidden subgroup structure from a finitely generated ( f.g . )
group @xmath58 to a set @xmath59 .
we assume that the hidden subgroup @xmath251 is a normal subgroup of @xmath58 of finite index
. then the objectives of this section are to demonstrate the following : * every hidden subgroup problem ( hsp ) @xmath57 on an arbitray f.g .
group @xmath58 can be lifted to an hsp @xmath252 on a free group @xmath253 of finite rank .
* moreover , a solution for the lifted hsp @xmath252 is for all practical purposes the same as the solution for the original hsp @xmath57 . _ thus , one need only investigate qhs algorithms for free groups of finite rank ! _ before we can describe the above results , we need to review a number of definitions .
we begin with the definition of a free group : [ universal definition]a group @xmath253 is said to be * free * of finite rank @xmath147 if there exists a finite set of @xmath147 generators @xmath254 such that , for every group @xmath58 and for every map @xmath255 of the set @xmath256 into the group @xmath58 , the map @xmath257 extends to a morphism @xmath258 .
we call the set @xmath256 a * free basis * of the group @xmath253 , and frequently denote the group @xmath253 by @xmath259 , .
it follows from this definition that the morphism @xmath260 is unique .
the intuitive idea encapsulated by this definition is that a free group is an unconstrained group ( very much analogous to a physical system without boundary conditions . ) in other words , a group is free provided it has a set of generators such that the only relations among those generators are those required for @xmath253 to be a group . for example , * @xmath261 is an allowed relation * @xmath262 is not an allowed relation for @xmath263 * @xmath264 is not an allowed relation as an immediate consequence of the above definition , we have the following proposition : let @xmath58 be an arbitrary f.g .
group with finite set of @xmath147 generators @xmath265 , and let @xmath266 be the free group of rank @xmath147 with free basis @xmath267 . then by the above definition , the map @xmath268 @xmath269 induces a uniques epimorphism @xmath270 from @xmath253 onto @xmath58 . with this epimorphism ,
every hsp @xmath68 on the group @xmath58 uniquely lifts to the hsp @xmath271 on the free group @xmath253 .
moreover , if @xmath251 and @xmath272 are the hidden subgroups of the hsps @xmath70 and @xmath273 , respectively , the corresponding hidden quotient groups @xmath274 and @xmath275 of these two hsps are isomorphic . hence , every solution of the hsp @xmath252 immediately produces a solution of the original hsp @xmath57 .
we close this section with the defintion of a group resentation , a concept that will be needed in the next section for generalizing shor s algorithm to free groups .
let @xmath58 be a group .
a * * group presentation**@xmath276 for @xmath58 is a set of free generators @xmath277 of a free group @xmath253 and a set of words @xmath278 in @xmath259 , called * relators * , such that the group @xmath58 is isomorphic to the quotient group @xmath279 , where @xmath280 , called the * consequence * of @xmath278 , is the smallest normal subgroup of @xmath259 containing the relators @xmath278 .
the intuition captured by the above definition is that @xmath277 are the generators of @xmath58 , and @xmath281 is a complete set of relations among these generators , i.e. , every relation among the generators of @xmath58 is a * consequence * of ( derivable from ) the relations @xmath281 . for example , * @xmath282 and @xmath283 are both presentations of the free group @xmath259 * @xmath284 and @xmath285 are both presentations of the cyclic group @xmath9of order @xmath132 , where @xmath1 and @xmath113 are integers such that @xmath286 . *
@xmath287 is a presentation of the symmetric group @xmath288 on three symbols .
the objective of this section is to generalize shor s algorithm to free groups of finite rank of rank 1 constructed by a push onto the cyclic group @xmath9 . in light of this and of the results outlined in the previous section ,
it is a natural objective to generalize shor s algorithm to free groups of finite rank . ] .
the chief obstacle to accomplishing this goal is finding a correct generalization of the shor transversal @xmath3{cccc}\mathbb{z}_{q } & \overset{\tau}{\longrightarrow } & \mathbb{z } & \\ n\operatorname{mod}q & \longmapsto & n & \left ( \text{\ } 0\leq n < q\right ) \end{array}\ ] ] unfortunately , there appear to be few mathematical clues indicating how to go about making such a generalization .
however , as we shall see , the generalization of the shor transversal to the transversal found in the wandering shor algorithm does provide a crucial clue , suggesting that a generalized shor transversal must be a 2-sided schreier transversal .
( see section 7 . )
we begin by formulating a constructive approach to free groups : let @xmath259 be a free group with free basis @xmath277 .
then a * word * is a finite string of the symbols @xmath289 . a * reduced word * is a word in which there is no substring of the form @xmath290 or @xmath291 .
two words are said to be * equivalent * if one can be transformed into the other by applying a finite number of substring insertions or deletions of the form @xmath290 or @xmath291 .
we denote an * arbitrary word * @xmath292 by @xmath293 , where each @xmath294 .
the * length * @xmath295 of a word @xmath296 is number of symbols @xmath297 that appear in @xmath292 , i.e. , @xmath298 . for example
, @xmath299 is a word of length @xmath300 which is equivalent to the reduced word @xmath301 of length @xmath302 .
it easily follows that : a free group @xmath259 is simply the set of reduced words together with the obvious definition of product , i.e. , concatenation followed by full reduction .
we can now use this constructive approach to create a special kind of transversal @xmath303 of an epimorphism @xmath270 , called a 2-sided scheier transversal@xcite : a set @xmath304 of reduced words in a free group @xmath266 is said to be a * 2-sided schreier system * provided * the empty word @xmath54 lies in @xmath304 .
* @xmath305 , and * @xmath306 given an epimorphism @xmath307 of the free group @xmath253 onto a group @xmath58 , a * 2-sided schreier transversal * @xmath308 for @xmath69 is a transversal of @xmath69 for which there exists a 2-sided schreier system such that @xmath309 .
a 2-sided schreier transversal is said to be * minimal * provided the length of each word @xmath292 is less than or equal to the length of each reduced word in the coset @xmath310 , where @xmath311 denotes the kernel of the epimorpism @xmath69 .
the wandering shor algorithm found in section 7 suggests that a correct generalization of the shor transversal @xmath312 @xmath313 must at least have the property that it is a minimal 2-sided schreier transversal .
whatetever other additional properties this generalization must have is simply not clear . in @xcite , we construct and investigate a number of different qhs algorithms on free groups that arise from the application of various additional conditions imposed upon the minimal 2-sided schreier transversal requirement . in this section ,
we only give a descriptive sketch of the simplest of these algorithms , i.e. , a qhs algorithm on free groups with only the minimal 2-sided schreier transversal requirement imposed .
let @xmath266 be the free group of finite rank @xmath147 with free basis @xmath254 , and let @xmath314 be an hsp on the free group @xmath253 .
we assume that the hidden subgroup @xmath251 is normal and of finite index in @xmath253 .
( please note that @xmath315 . )
* choose a finite group probe @xmath58 with presentation @xmath316 , where the subscript @xmath69 denotes the epimorphism @xmath307 induced by the map @xmath317 .
* choose a minimal 2-sided schreier transversal @xmath303 of the epimorphism @xmath307 .
* finally , construct the push @xmath318 our generalized shor algorithm for the free group @xmath253 consists of the following steps : * call qrand@xmath319 to produce a word @xmath320 in @xmath253 close to a word @xmath321 lying in @xmath322 . * with input @xmath320 , use a polytime classical algorithm to determine @xmath321 .
( see @xcite . ) * repeat steps 1 and 2 until enough relators @xmath321 s are found to produce a presentation @xmath323 of the hidden subgroup @xmath324 , then output the presentation @xmath325 , and stop .
obviously , much more needs to be said . for , example
, we have not explained how one chooses the relators @xmath326 so that @xmath327 is a good group probe .
moreover , we have not explained what classical algorithm is used to transform the words @xmath320 into the relators @xmath321 . for more details , we refer the reader to @xcite .
in this section , our objective is to factor grover s algorithm into the qhs primitives developed in the previous sections of this paper . as a result
, we will show that grover s algorithm is more closely related to shor s algorithm than one might at first expect .
in particular , we will show that grover s algorithm is a qhs algorithm in the sense that it solves an hsp @xmath328 , which we will refer to as the * grover hsp . * however
, we will then show that the standard qhs algorithm for this hsp can not possibly find a solution .
we begin with a question : _ does grover s algorithm have symmetries that we can exploit ? _ the problem solved by grover s algorithm @xcite , @xcite , @xcite
, @xcite is that of finding an unknown integer label @xmath329 in an unstructured database with items labeled by the integers : @xmath330 given the oracle@xmath331{ll}1 & \text{if \ } j = j_{0}\\ 0 & \text{otherwise}\end{array } \right.\ ] ] let @xmath332 be the hilbert space with orthonormal basis @xmath333 .
grover s oracle is essentially given by the unitary transformation@xmath334 where @xmath335 is inversion in the hyperplane orthogonal to @xmath336 .
let @xmath337 denote the hadamard transformation on the hilbert space @xmath338 .
then grover s algorithm is as follows : * ( initialization)@xmath339 * loop until @xmath340@xmath341 * measure @xmath342 with respect to the standard basis@xmath343 to obtain the unknown state @xmath344 with@xmath345 but where is the hidden symmetry in grover s algorithm ?
let @xmath346 be the symmetric group on the symbols @xmath347 .
then grover s algorithm is invariant under the * hidden subgroup * @xmath348,called the * stabilizer subgroup * for @xmath329 , i.e. , grover s algorithm is invariant under the group action@xmath349 moreover , if we know the hidden subgroup @xmath350 , then we know @xmath329 , and vice versa . in other words ,
the problem of finding the unknown label @xmath329 is informationally the same as the problem of finding the hidden subgroup @xmath350 .
let @xmath351 denote the permutation that interchanges integers @xmath352 and @xmath353 , and leaves all other integers fixed .
thus , @xmath354is a transposition if @xmath263 , and the identity permutation @xmath54 if @xmath355 .
the set @xmath356 is a complete set of distinct coset representatives for the hidden subgroup @xmath350 of @xmath346 , i.e. , the coset space @xmath357is given by the following complete set of distinct cosets:@xmath358 we can now see that grover s algorithm is a hidden subgroup algorithm in the sense that it is a quantum algorithm which solves the following hidden subgroup problem : * grover s hidden subgroup problem .
@xmath359 _ be a map from the symmetric group _
@xmath346 _ to a set _ @xmath360 _ _ with hidden subgroup structure given by the commutative diagram__@xmath3{ccc}s_{n } & \longrightarrow & \quad\;\;s\\ \nu_{j_{0}}\searrow & & \nearrow\iota\\ & s_{n}/stab_{j_{0 } } & \end{array } \text { \ , } \ ] ] _ where _ @xmath361 _ is the natural surjection of _
_ on to the coset space _ @xmath357 _ _ , and where _ _ @xmath3{ccc}\iota:\;s_{n}/stab_{j_{0 } } & \longrightarrow & s\\ \quad\left ( jj_{0}\right ) stab_{j_{0 } } & \longmapsto & j \end{array}\ ] ] _ is the unknown relabeling ( bijection ) of the coset space _
@xmath357__onto the set _ _ @xmath59__. find the hidden subgroup _ _
@xmath350__with bounded probability of error . _ _
now let us compare shor s algorithm with grover s . from section 6 ,
we know that shor s algorithm @xcite , @xcite , @xcite , @xcite solves the hidden subgroup problem @xmath362 with hidden subgroup structure@xmath3{ccc}\mathbb{z } & \longrightarrow & \quad\;\;\mathbb{z}_{n}\\ \;\;\nu\searrow & & \nearrow\iota\\ & \mathbb{z}/p\mathbb{z } & \end{array}\ ] ] moreover , as stated in section 6 , shor has created his algorithm by pushing the above hidden subgroup problem @xmath362 to the hidden subgroup problem @xmath363 ( called shor s oracle ) , where the hidden subgroup structure of @xmath14 is given by the commutative diagram@xmath3{ccc}z & \longrightarrow & \quad\;\;z_{n}\\ \qquad\;\;\alpha\searrow\nwarrow\tau & & \nearrow\widetilde{\varphi}=\varphi\circ\tau\\ & z_{q } & \end{array } \text { \ , } \ ] ] where @xmath219 is the natural epimorphism of @xmath4 onto @xmath9 , and where @xmath109 is shor s chosen transversal for the epimorphism @xmath219 .
surprisingly , grover s algorithm , viewed as an algorithm that solves the grover hidden subgroup problem , is very similar to shor s algorithm . like shor
s algorithm , grover s algorithm solves a hidden subgroup problem , i.e. , the grover hidden subgroup problem @xmath359 with hidden subgroup structure@xmath3{ccc}s_{n } & \longrightarrow & \quad\;\;s\\ \;\;\nu\searrow & & \nearrow\iota\\ & s_{n}/stab_{j_{0 } } & \end{array } \text { \ , } \ ] ] where @xmath360 denotes the set resulting from an unknown relabeling ( bijection ) @xmath364 of the coset space@xmath365 also , like shor s algorithm , we can think of grover s algorithm as one created by pushing the grover hidden subgroup problem @xmath359 to the hidden subgroup problem @xmath366 , where the pushing is defined by the following commutative diagram@xmath3{ccc}s_{n } & \longrightarrow & \qquad\qquad\;\;s = s_{n}/stab_{j_{0}}\\ \qquad\;\;\alpha\searrow\nwarrow\tau & & \nearrow\widetilde{\varphi}=\varphi\circ\tau\\ & s_{n}/stab_{0 } & \end{array } \text { \ \ , } \ ] ] where @xmath367 denotes the natural surjection of @xmath346 onto the coset space @xmath368 , and where @xmath369denotes the transversal of @xmath219 given by@xmath3{ccc}s_{n}/stab_{0 } & \longrightarrow & s_{n}\\ \left ( j0\right ) stab_{0 } & \longmapsto & \left ( j0\right ) \end{array } \text { \ \ .}\ ] ] again also like shor s algorithm , the map @xmath14 given by @xmath370 is ( if @xmath371 ) actually a disguised grover s oracle .
for the map @xmath14 can easily be shown to simply to@xmath372{ll}(j0)stab_{j_{0 } } & \text{if \ } j = j_{0}\\ stab_{j_{0 } } & \text{otherwise\ \ , } \end{array } \right.\ ] ] which is informationally the same as grover s oracle@xmath331{ll}j & \text{if \ } j = j_{0}\\ 1 & \text{otherwise}\end{array } \right.\ ] ] hence , we can conclude that grover s algorithm is a quantum algorithm very much like shor s algorithm , in that it is a quantum algorithm that solves the grover hidden subgroup problem .
however , , this appears to be where the similarity between grover s and shor s algorithms ends . for the standard non - abelian qhs algorithm for
@xmath346 can not find the hidden subgroup @xmath350 for each of following two reasons : * since the subgroups @xmath373 are not normal subgroups of @xmath346 , it follows from the work of hallgren et al @xcite , @xcite that the standard non - abelian hidden subgroup algorithm will find the largest normal subgroup of @xmath346 lying in @xmath350 .
but unfortunately , the largest normal subgroup of @xmath346 lying in @xmath373 is the trivial subgroup of @xmath346 . *
the subgroups @xmath374 are mutually conjugate subgroups of @xmath346 .
moreover , one can not hope to use this qhs approach to grover s algorithm to find a faster quantum algorithm . for zalka @xcite
has shown that grover s algorithm is optimal .
the arguments given above suggest that grover s and shor s algorithms are more closely related that one might at first expect .
although the standard non - abelian qhs algorithm on @xmath346 can not solve the grover hidden subgroup problem , there does remain an intriguing question : * question .
* _ is there some modification of ( or extension of ) the standard qhs algorithm on the symmetric group _
@xmath346 _ that actually solves grover s hidden subgroup problem ? _ for a more in - depth discussion of the results found in this section ,
we refer the reader to @xcite .
in this paper , we have decomposed shor s quantum factoring algorithm into primitives , generalized these primitives , and then reassembled them into a wealth of new qhs algorithms . but as the results found in the previous section suggest , this list of quantum algorithmic primitives is far from complete .
this is expressed by the following question : _ where can we find more algorithmic primitives to create a more well rounded toolkit for quantum algorithmic development ? _
the previous section suggests that indeed all quantum algorithms may well be hidden subgroup algorithms in the sense that they all find hidden symmetries , i.e. , hidden subgroups .
this is suggestive of the following meta - procedure for quantum algorithm development : * explicitly state the problem to be solved . *
rephrase the problem as a hidden symmetry problem . *
create a quantum algorithm to find the hidden symmetry .
_ can this meta - procedure be made more explicit ? _ perhaps some reader to this paper will be able to answer this question .
this work is partially supported by the defense advanced research projects agency ( darpa ) and air forche research laboratory , air force materiel command , usaf , under agreement number f30602 - 01 - 2 - 0522 .
government is authorized to reproduce and distribute reprints for governmental purposes notwithstanding any copyright annotation thereon . this work also partially supported by the institute for scientific interchange ( isi ) , torino , the national institute of standards and technology ( nist ) , the mathematical sciences research institute ( msri ) , the isaac newton institute for mathematical sciences , and the l - o - o - p fund .
hallgren , sean , alexander russell , amnon ta - shma , * the hidden subgroup problem and quantum computation using group representations , * proceedings of the thirty - second annual acm symposium on theory of computing , portland , oregon , may 2000 , 627 - 635 .
lomonaco , samuel j. , jr . , ( ed . ) ,
* `` quantum computation : a grand mathematical challenge for thetwenty - first century and the millennium , '' * proceedings of the symposia of applied mathematics , vol .
58 , american mathematical society , providence , rhode island , ( 2002 ) .
( 358 pages)(http://www.ams.org / bookstore?fn=20&arg1=whatsnew&item = psapm-58)(http://www.csee.umbc.edu / lomonaco / ams / lecture_notes.html ) lomonaco , samuel j. , jr . , and howard e. brandt , ( eds . ) , * `` quantum computation and information , '' * ams contemporary mathematics , vol . 305 , american mathematical society , providence , ri , ( 2002 ) .
( 310 pages)(http://www.csee.umbc.edu/@xmath375lomonaco / ams / special.html ) lomonaco , samuel j. , jr . , and louis h. kauffman , * continuous quantum hidden subgroup algroithms * , spie proceedings on quantum information and computation , vol .
5105 , 11 , ( 2003 ) , 80 - 89 .
( http://arxiv.org/abs/quant-ph/0304084 ) lomonaco , samuel j. , jr . , and , louis h. kauffman , * quantum hidden subgroup algorithms : the devil is in the details , * 2004 proceedings of spie proceedings on quantum information and computation , ( 2004 ) , 137 - 141 .
http://arxiv.org/abs/quant-ph/0403229 mosca , michelle , and artur ekert , * the hidden subgroup problem and eigenvalue estimation on a quantum computer , * proceedings of the 1st nasa international conference on quantum computing and quantum communication , springer - verlag , ( 2001 ) .
( http://xxx.lanl.gov/abs/quant-ph/9903071 ) shor ,
peter w. , * polynomial time algorithms for prime factorization and discrete logarithms on a quantum computer , * siam j. on computing , 26(5 ) ( 1997 ) , pp 1484 - 1509 .
( http://xxx.lanl.gov/abs/quant-ph/9508027 ) | one of the most promising and versatile approaches to creating new quantum algorithms is based on the quantum hidden subgroup ( qhs ) paradigm , originally suggested by alexei kitaev .
this class of quantum algorithms encompasses the deutsch - jozsa , simon , shor algorithms , and many more . in this paper ,
our strategy for finding new quantum algorithms is to decompose shor s quantum factoring algorithm into its basic primitives , then to generalize these primitives , and finally to show how to reassemble them into new qhs algorithms .
taking an `` alphabetic building blocks approach , '' we use these primitives to form an `` algorithmic toolkit '' for the creation of new quantum algorithms , such as wandering shor algorithms , continuous shor algorithms , the quantum circle algorithm , the dual shor algorithm , a qhs algorithm for feynman integrals , free qhs algorithms , and more . toward the end of this paper ,
we show how grover s algorithm is most surprisingly almost a qhs algorithm , and how this result suggests the possibility of an even more complete `` algorithmic tookit '' beyond the qhs algorithms . |
let @xmath0 be a knot or a link in @xmath1 and @xmath2 a punctured sphere in the complement @xmath3 obtained from a bridge sphere of @xmath0
. then the following natural question arises .
[ question0 ] ( 1 ) which essential simple loops on @xmath2 are null - homotopic in @xmath3 ?
\(2 ) for two distinct essential simple loops on @xmath2 , when are they homotopic in @xmath3 ?
a refined version of the first question for @xmath4-bridge spheres of @xmath4-bridge links was proposed in the second author s joint work with ohtsuki and riley ( * ? ? ?
* question 9.1(2 ) ) , in relation with epimorphisms between @xmath4-bridge links .
it may be regarded as a special variation of a question raised by minsky ( * ? ? ?
* question 5.4 ) on essential simple loops on heegaard surfaces of @xmath5-manifolds .
the purpose of this note is to announce a complete answer to question [ question0 ] for @xmath4-bridge spheres of @xmath4-bridge links established by the series of papers @xcite and to explain its application to the study of character varieties and mcshane s identity @xcite .
the key tool for solving the question is small cancellation theory , applied to two - generator and one - relator presentations of @xmath4-bridge link groups .
we note that it has been proved by weinbaum @xcite and appel and schupp @xcite that the word and conjugacy problems for prime alternating link groups are solvable , by using small cancellation theory ( see also @xcite and references in it ) .
moreover , it was shown by sela @xcite and praux @xcite that the word and conjugacy problems for any link group are solvable .
a characteristic feature of our work is that it gives complete answers to special ( but also natural ) word and conjugacy problems for the link groups of 2-bridge links , which form a special ( but also important ) family of prime alternating links .
( see @xcite for the role of @xmath4-bridge links in kleinian group theory . )
this note is organized as follows . in sections [ statements ] , [ application1 ] and [ application2 ]
, we describe the main results , applications to character varieties and mcshane s identity .
the remaining sections are devoted to explanation of the idea of the proof of the main results . in section [ group_presentation ]
, we describe the two - generator and one - relator presentation of the @xmath4-bridge link group to which small cancellation theory is applied , and give a natural decomposition of the relator , which plays a key role in the proof . in section [ sequences ] , we introduce a certain finite sequence associated with the relator and state its key properties . in section [ small_cancellation ] ,
we recall small cancellation theory and present a characterization of the `` pieces '' of the symmetrized subset arising from the relator . in sections [ outline_proof ] and [ outline_proof2 ] ,
we describe outlines of the proofs of the main results .
the authors would like to thank norbert acampo , hirotaka akiyoshi , brian bowditch , danny calegari , max forester , koji fujiwara , yair minsky , toshihiro nakanishi , caroline series and ser peow tan for stimulating conversations .
for a rational number @xmath6 , let @xmath7 be the @xmath4-bridge link of slope @xmath8 , which is defined as the sum @xmath9 of rational tangles of slope @xmath10 and @xmath8 ( see figure [ fig.trefoil ] ) .
the common boundary @xmath11 of the rational tangles is identified with the _ conway sphere _
@xmath12 , where @xmath13 is the group of isometries of the euclidean plane @xmath14 generated by the @xmath15-rotations around the points in the lattice @xmath16 .
let @xmath17 be the @xmath18-punctured sphere @xmath19 in the link complement @xmath20 .
any essential simple loop in @xmath17 , up to isotopy , is obtained as the image of a line of slope @xmath21 in @xmath22 by the covering projection onto @xmath17 .
the ( unoriented ) essential simple loop in @xmath17 so obtained is denoted by @xmath23 .
we also denote by @xmath23 the conjugacy class of an element of @xmath24 represented by ( a suitably oriented ) @xmath23 . then the _ link group _
@xmath25 is identified with @xmath26 , where @xmath27 denotes the normal closure .
@xmath9 with @xmath28 . here
@xmath29 and @xmath30 , respectively , are the inside and the outside of the bridge sphere @xmath31 . ]
let @xmath32 be the _
farey tessellation _
, whose ideal vertex set is identified with @xmath33 . for each @xmath34 , let @xmath35 be the group of automorphisms of @xmath32 generated by reflections in the edges of @xmath32 with an endpoint @xmath8 , and let @xmath36 be the group generated by @xmath35 and @xmath37 .
then the region , @xmath38 , bounded by a pair of farey edges with an endpoint @xmath10 and a pair of farey edges with an endpoint @xmath8 forms a fundamental domain of the action of @xmath36 on @xmath39 ( see figure [ fig.fd ] ) .
let @xmath40 and @xmath41 be the closed intervals in @xmath42 obtained as the intersection with @xmath42 of the closure of @xmath38 .
suppose that @xmath8 is a rational number with @xmath43 .
( we may always assume this except when we treat the trivial knot and the trivial @xmath4-component link . )
write ( 230,70 ) ( 0,48)@xmath44[0pt]{$m_2 \ , + \ , $ } \raisebox{-10pt}[0pt][0pt]{$\ , \ddots \ $ } \raisebox{-12pt}[0pt][0pt]{$+ \ , \cfrac{1}{m_k}$ } } } \ = : [ m_1,m_2 , \dots , m_k],}$ ] where @xmath45 , @xmath46 , and @xmath47
. then the above intervals are given by @xmath48 $ ] and @xmath49 $ ] , where @xmath50 & \mbox{if $ k$ is odd,}\\ [ m_1 , m_2 , \dots , m_{k-1 } , m_k-1 ] & \mbox{if $ k$ is even , } \end{cases}\\ r_2 & = \begin{cases } [ m_1 , m_2 , \dots , m_{k-1 } , m_k-1 ] & \mbox{if $ k$ is odd,}\\ [ m_1 , m_2 , \dots , m_{k-1 } ] & \mbox{if $ k$ is even . }
\end{cases}\end{aligned}\ ] ] a fundamental domain of @xmath51 in the farey tessellation ( the shaded domain ) for @xmath52 $ ] . ]
we recall the following fact ( ( * ? ? ?
* proposition 4.6 and corollary 4.7 ) and ( * ? ? ?
* lemma 7.1 ) ) which describes the role of @xmath36 in the study of @xmath4-bridge link groups .
[ previous_results ] ( 1 ) if two elements @xmath53 and @xmath54 of @xmath33 belong to the same orbit @xmath36-orbit , then the unoriented loops @xmath23 and @xmath55 are homotopic in @xmath20 .
\(2 ) for any @xmath21 , there is a unique rational number @xmath56 such that @xmath53 is contained in the @xmath36-orbit of @xmath57 . in particular , @xmath23 is homotopic to @xmath58 in @xmath20 . thus if @xmath59 , then @xmath23 is null - homotopic in @xmath20
. thus the following question naturally arises ( see ( * ? ? ?
* question 9.1(2 ) ) ) . [ question ]
( 1 ) does the converse to proposition [ previous_results](2 ) hold ? namely , is it true that @xmath23 is null - homotopic in @xmath20 if and only if @xmath53 belongs to the @xmath36-orbit of @xmath10 or @xmath8 ?
\(2 ) for two distinct rational numbers @xmath60 , when are the unoriented loops @xmath23 and @xmath55 homotopic in @xmath20 ? the following theorem proved in @xcite gives a complete answer to question [ question](1 ) .
[ first_result ] the loop @xmath23 is null - homotopic in @xmath61 if and only if @xmath53 belongs to the @xmath36-orbit of @xmath10 or @xmath8 . in other words , if @xmath62 , then @xmath23 is not null - homotopic in @xmath20 .
this has the following application to the study of epimorphisms between @xmath4-bridge link groups ( see ( * ? ? ? * section
2 ) for precise meaning ) .
[ first_cor ] there is an upper - meridian - pair - preserving epimorphism from @xmath63 to @xmath64 if and only if @xmath53 or @xmath65 belongs to the @xmath36-orbit of @xmath8 or @xmath10 .
the following theorem proved in @xcite gives a complete answer to question [ question](2 ) .
[ second_result ] suppose that @xmath8 is a rational number such that @xmath66 . for distinct @xmath60 , the unoriented loops @xmath23 and @xmath55 are homotopic in @xmath20 if and only if one of the following holds . 1 .
@xmath67 , where @xmath68 is an integer , and @xmath69 and @xmath70 satisfy @xmath71 and @xmath72 , where @xmath73 is a pair of relatively prime positive integers .
2 . @xmath74 , namely @xmath7 is the whitehead link , and the set @xmath75 equals either @xmath76 or @xmath77 .
the proof of theorem [ second_result ] reveals the structure of the normalizer of an element of @xmath64 represented by @xmath23 .
this enables us to show the following .
[ peripheral_results ] let @xmath8 be a rational number such that @xmath66 .
suppose @xmath7 is hyperbolic , i.e. , @xmath78 and @xmath79 , and let @xmath53 be a rational number contained in @xmath80 .
\(1 ) the loop @xmath23 is peripheral if and only if one of the following holds . 1 .
@xmath81 and @xmath82 or @xmath83 .
2 . @xmath84 for some integer @xmath85 , and @xmath86 .
3 . @xmath87 for some integers @xmath88 , and @xmath89 .
\(2 ) the conjugacy class @xmath23 is primitive in @xmath64 with the following exceptions
. 1 . @xmath81 and @xmath90 or @xmath91 . in this case
@xmath23 is the third power of some primitive element in @xmath64 .
2 . @xmath92 and @xmath90 . in this case
@xmath23 is the second power of some primitive element in @xmath64 .
3 . @xmath93 and @xmath94 . in this case
@xmath23 is the second power of some primitive element in @xmath64 . at the end of this section
, we describe a relation of theorem [ first_result ] with the question raised by minsky in ( * ? ? ?
* question 5.4 ) .
let @xmath95 be a heegaard splitting of a @xmath5-manifold @xmath96 .
let @xmath97 be the mapping class group of @xmath98 , and let @xmath99 be the kernel of the map @xmath100 .
identify @xmath99 with a subgroup of @xmath101 , and consider the subgroup @xmath102 of @xmath101 .
now let @xmath103 be the set of ( isotopy classes of ) simple loops in @xmath2 which bound a disk in @xmath98 .
let @xmath104 be the set of essential simple loops in @xmath2 which are null - homotopic in @xmath96 .
note that @xmath104 contains @xmath103 and invariant under @xmath102 .
in particular , the orbit @xmath105 is a subset of @xmath104 .
then minsky posed the following question .
[ minsky_question ] when is @xmath104 equal to the orbit @xmath105 ?
the above question makes sense not only for heegaard splittings but also bridge decompositions of knots and links .
in particular , for @xmath4-bridge links , the groups @xmath37 and @xmath35 in our setting correspond to the groups @xmath106 and @xmath107 , and hence the group @xmath36 corresponds to the group @xmath108 . to make this precise ,
recall the bridge decomposition @xmath9 , and let @xmath109 ( resp .
@xmath110 ) be the mapping class group of the pair @xmath30 ( resp .
@xmath29 ) , and let @xmath111 be the kernel of the natural map @xmath112 ( resp .
@xmath113 ) . identify @xmath111 with a subgroup of the mapping class group @xmath114 of the @xmath18-times punctured sphere @xmath17 .
recall that the farey tessellation @xmath32 is identified with the curve complex of @xmath17 and there is a natural epimorphism from @xmath114 to the automorphism group @xmath115 of @xmath32 , whose kernel is isomorphic to @xmath116 .
then the groups @xmath117 and @xmath118 , respectively , are identified with the images of @xmath119 and @xmath120 by this epimorphism .
moreover , the sets @xmath121 and @xmath122 , respectively , correspond to the sets @xmath123 and @xmath124 .
theorem [ first_result ] says that the set @xmath104 of simple loops in @xmath17 which are null - homotopic in @xmath20 is equal to the orbit @xmath125 .
thus theorem [ first_result ] may be regarded as an answer to the special variation of question [ minsky_question ] .
in this section and the next section , we assume @xmath78 , where @xmath126 and @xmath127 are relatively prime positive integers such that @xmath79 .
this is equivalent to the condition that @xmath7 is hyperbolic , namely the link complement @xmath20 admits a complete hyperbolic structure of finite volume .
let @xmath128 be the @xmath129-representation of @xmath24 obtained as the composition @xmath130 where the last homomorphism is the holonomy representation associated with the complete hyperbolic structure .
now , let @xmath131 be the once - punctured torus obtained as the quotient @xmath132 , and let @xmath133 be the orbifold @xmath134 where @xmath135 is the group generated by @xmath15-rotations around the points in @xmath136 .
note that @xmath133 is the orbifold with underlying space a once - punctured sphere and with three cone points of cone angle @xmath15 .
the surfaces @xmath131 and @xmath17 , respectively , are @xmath137-covering and @xmath116-covering of @xmath133 , and hence their fundamental groups are identified with subgroups of the orbifold fundamental group @xmath138 of indices @xmath4 and @xmath18 , respectively .
the @xmath129-representation @xmath128 of @xmath24 extends , in a unique way , to that of @xmath138 ( see ( * ? ? ?
* proposition 2.2 ) ) , and so we obtain , in a unique way , a @xmath129-representation of @xmath139 by restriction .
we continue to denote it by @xmath128 .
note that @xmath140 is _ type - preserving _
, i.e. , it satisfies the following conditions . 1
. @xmath128 is irreducible , i.e. , its image does not have a common fixed point on @xmath141 .
2 . @xmath128 maps a peripheral element of @xmath139 to a parabolic transformation . by extending the concept of a geometrically infinite end of a kleinian group ,
bowditch @xcite introduced the notion of the end invariants of a type - preserving @xmath129-representation of @xmath139 .
tan , wong and zhang @xcite ( cf .
@xcite ) extended this notion ( with slight modification ) to an arbitrary @xmath129-representation of @xmath139 .
( to be precise , @xcite treats @xmath142-representations .
however , the arguments work for @xmath129-representations . ) to recall the notion of end invariants , let @xmath143 be the set of free homotopy classes of essential simple loops on @xmath131 .
then @xmath143 is identified with @xmath33 , the vertex set of the farey tessellation @xmath32 by the following rule . for each @xmath21 , let @xmath144 be the essential simple loop on @xmath131 obtained as the image of a line of slope @xmath53 in @xmath22 .
then the correspondence @xmath145 gives the desired identification @xmath146 .
the projective lamination space @xmath147 is then identified with @xmath148 and contains @xmath143 as the dense subset of rational points .
[ def_end_invariant ] let @xmath149 be a @xmath129-representation of @xmath139 .
\(1 ) an element @xmath150 is an _ end invariant _ of @xmath149 if there exists a sequence of distinct elements @xmath151 such that @xmath152 and such that @xmath153 is bounded from above .
\(2 ) @xmath154 denotes the set of end invariants of @xmath149 . in the above definition
, it should be noted that @xmath155 is well - defined though @xmath156 is defined only up to sign .
note also that the condition that @xmath153 is bounded from above is equivalent to the condition that the hyperbolic translation lengths of the isometries @xmath157 of @xmath158 are bounded from above .
tan , wong and zhang @xcite showed that @xmath154 is a closed subset of @xmath147 and proved various interesting properties of @xmath154 , including a characterization of those representations @xmath149 with @xmath159 or @xmath147 , generalizing a result of bowditch @xcite .
they also proposed an interesting conjecture ( * ? ? ? * conjecture 1.8 ) concerning possible homeomorphism types of @xmath154 .
the following is a modified version of the conjecture of which tan @xcite informed the authors .
[ conj_twz ] suppose @xmath154 has at least two accumulation points . then either @xmath160 or a cantor set of @xmath147 .
they constructed a family of representations @xmath149 which have cantor sets as @xmath154 , and proved the following supporting evidence to the conjecture .
[ discrete_twz ] let @xmath161 be discrete in the sense that the set @xmath162 is discrete in @xmath163 .
then if @xmath154 has at least three elements , then @xmath154 is either a cantor set of @xmath147 or all of @xmath147 .
the above theorem implies that the end invariants @xmath164 of the representation @xmath128 induced by the holonomy representation of a hyperbolic @xmath4-bridge link @xmath7 is a cantor set .
but it does not give us the exact description of @xmath164 . by using the main results stated in section [ statements ]
, we can explicitly determine the end invariants @xmath164 . to state the theorem , recall that the _ limit set _
@xmath165 of the group @xmath36 is the set of accumulation points in the closure of @xmath39 of the @xmath36-orbit of a point in @xmath39 . for a hyperbolic @xmath4-bridge link @xmath7 ,
the set @xmath164 is equal to the limit set @xmath165 of the group @xmath36 .
we would like to propose the following conjecture .
let @xmath166 be a type - preserving representation such that @xmath167 .
then @xmath149 is conjugate to the representation @xmath128 .
in his ph.d .
thesis @xcite , mcshane proved the following surprising theorem .
[ mcshane ] let @xmath168 be a type - preserving fuchsian representation .
then @xmath169 in the above identity , @xmath170 denotes the translation length of the orientation - preserving isometry @xmath171 of the hyperbolic plane .
this identity has been generalized to cusped hyperbolic surfaces by mcshane himself @xcite , to hyperbolic surfaces with cusps and geodesic boundary by mirzakhani @xcite , and to hyperbolic surfaces with cusps , geodesic boundary and conical singularities by tan , wong and zhang @xcite . a wonderful application to the weil - petersson volume of the moduli spaces of bordered hyperbolic surface
was found by mirzakhani @xcite .
bowditch @xcite ( cf .
@xcite ) showed that the identity in theorem [ mcshane ] is also valid for all quasifuchsian representations of @xmath139 , where @xmath170 is regarded as the complex translation length of the orientation - preserving isometry @xmath171 of the hyperbolic @xmath5-space .
moreover , he gave a nice variation of the identity for hyperbolic once - punctured torus bundles , which describes the cusp shape in terms of the complex translation lengths of essential simple loops on the fiber torus @xcite .
other @xmath5-dimensional variations have been obtained by @xcite . as an application of the main results stated in section [ statements ] ,
we can obtain yet another @xmath5-dimensional variation of mcshane s identity , which describes the cusp shape of a hyperbolic @xmath4-bridge link in terms of the complex translation lengths of essential simple loops on the bridge sphere .
this proves a conjecture proposed by the first author in @xcite . to describe the result , note that each cusp of the hyperbolic manifold @xmath20 carries a euclidean structure , well - defined up to similarity , and hence it is identified with the quotient of @xmath163 ( with the natural euclidean metric ) by the lattice @xmath172 , generated by the translations @xmath173 $ ] and @xmath174 $ ] corresponding to the meridian and ( suitably chosen ) longitude respectively .
this @xmath175 does not depend on the choice of the cusp , because when @xmath7 is a two - component link there is an isometry of @xmath20 interchanging the two cusps .
we call @xmath175 the _ modulus _ of the cusp and denote it by @xmath176 .
[ conj : mcshane ] for a hyperbolic @xmath4-bridge link @xmath7 with @xmath78 , the following identity holds : @xmath177 further the modulus @xmath176 of the cusp torus of the cusped hyperbolic manifold @xmath20 with respect to a suitable choice of a longitude is given by the following formula : @xmath178 the main results stated in section [ statements ] are used to establish the absolute convergence of the infinite series .
in the remainder of this note , @xmath126 and @xmath127 denote relatively prime positive integers such that @xmath179 and @xmath78 .
theorems [ first_result ] and [ second_result ] are proved by applying the small cancellation theory to a two - generator and one - relator presentation of the link group @xmath64 . to recall the presentation ,
let @xmath180 and @xmath181 , respectively , be the elements of @xmath182 represented by the oriented loops @xmath183 and @xmath184 based on @xmath185 as illustrated in figure [ fig.generator ] .
then @xmath186 is identified with the free group @xmath187 .
note that @xmath188 intersects the disk , @xmath189 , in @xmath190 bounded by a component of @xmath191 and the essential arc , @xmath192 , on @xmath193 of slope @xmath194 , in figure [ fig.generator ] .
@xmath195 , where @xmath180 and @xmath181 are represented by @xmath183 and @xmath184 , respectively . ]
the line of slope @xmath196 gives @xmath197 and @xmath198 . since the inverse image of @xmath199 ( resp .
@xmath200 ) in @xmath22 is the union of the single arrowed ( resp .
double arrowed ) vertical edges , a positive intersection with a single arrowed ( resp .
double arrowed ) edge corresponds to @xmath180 ( resp .
@xmath181 ) . ] to obtain an element , @xmath201 , of @xmath187 represented by the simple loop @xmath202 ( with a suitable choice of an orientation and a path joining @xmath202 to the base point @xmath185 ) , note that the inverse image of @xmath199 ( resp .
@xmath200 ) in @xmath22 is the union of the single arrowed ( resp .
double arrowed ) vertical edges in figure [ line ] .
let @xmath203 be the line in @xmath14 of slope @xmath8 passing through the origin , and let @xmath204 be the line in @xmath22 obtained by slightly modifying @xmath203 near each of the lattice points in @xmath203 so that @xmath204 takes an upper circuitous route around it , as illustrated in figure [ line ] . pick a base point @xmath205 from the intersection of @xmath204 with the second quadrant , and consider the sub - line - segment of @xmath204 bounded by @xmath205 and @xmath206 .
then the image of the sub - line - segment in @xmath17 is homotopic to the loop @xmath23 .
let @xmath201 be the word in @xmath207 obtained by reading the intersection of the line - segment with the vertical lattice lines (= the inverse images of @xmath199 and @xmath200 ) as in figure [ line ] .
then @xmath208 is represented by the simple loop @xmath202 , and we obtain the following two - generator one - relator presentation .
@xmath209 to describe the explicit formula for @xmath201 , set @xmath210 where @xmath211 is the greatest integer not exceeding @xmath212
. then we have the following ( cf .
* proposition 1 ) ) .
let @xmath213 where @xmath211 is the greatest integer not exceeding @xmath212 . 1 .
if @xmath126 is odd , then @xmath214 where @xmath215 .
2 . if @xmath126 is even , then @xmath216 where @xmath217 . in the above formula , @xmath218
is obtained from the open interval of @xmath203 bounded by @xmath219 and @xmath220 . the decomposition of the relator @xmath221 we now describe a natural decomposition of the word @xmath201 , which plays a key role in the proof of the main results .
let @xmath222 ( @xmath223 ) be the rational number introduced in section [ statements ] .
then @xmath224 and the parallelogram in @xmath14 spanned by @xmath219 , @xmath225 , @xmath226 and @xmath220 does not contain lattice points in its interior .
consider the infinite broken line , @xmath227 , obtained by joining the lattice points @xmath228 which is invariant by the translation @xmath229 .
let @xmath230 be the topological line obtained by slightly modifying @xmath227 near each of the lattice points in @xmath227 so that @xmath230 takes an upper or lower circuitous route around it according as the lattice point is of the form @xmath231 or @xmath232 for some @xmath233 , as illustrated in figure [ line2 ] .
we may assume the base points @xmath205 and @xmath234 in @xmath204 also lie in @xmath230 .
then the sub - arcs of @xmath204 and @xmath230 bounded by @xmath205 and @xmath234 are homotopic in @xmath22 by a homotopy fixing the end points . moreover , the word @xmath201 is also obtained by reading the intersection of the sub - path of @xmath230 with the vertical lattice lines .
pick a point @xmath235 whose @xmath212-coordinate is @xmath236 , and set @xmath237 and @xmath238 .
let @xmath239 be the sub - path of @xmath230 bounded by @xmath240 and @xmath241 ( @xmath242 ) , and consider the subword , @xmath243 , of @xmath201 corresponding to @xmath239 .
then we have the decomposition @xmath244 where the lengths of the subwords @xmath243 are given by @xmath245 and @xmath246 .
this decomposition plays a key role in the following section .
we begin with the following observation . 1 . the word @xmath201 is _ reduced _ , i.e. , it does not contain @xmath247 or @xmath248 for any @xmath249 .
it is also _ cyclically reduced _ ,
i.e. , all its cyclic permutations are reduced .
2 . the word @xmath201 is _ alternating _ , i.e. , @xmath250 and @xmath251 appear in @xmath201 alternately , to be precise , neither @xmath252 nor @xmath253 appears in @xmath201 .
it is also _ cyclically alternating _ ,
i.e. , all its cyclic permutations are alternating .
this observation implies that the word @xmath201 is determined by the @xmath2-sequence defined below and the initial letter ( with exponent ) .
[ def : alternating ] \(1 ) let @xmath254 be a nonempty reduced word in @xmath207 .
decompose @xmath254 into @xmath255 where , for each @xmath256 , all letters in @xmath257 have positive ( resp .
negative ) exponents , and all letters in @xmath258 have negative ( resp .
positive ) exponents .
( here the symbol @xmath259 means that the two words are not only equal as elements of the free group but also visibly equal , i.e. , equal without cancellation . )
then the sequence of positive integers @xmath260 is called the _
@xmath2-sequence of @xmath261_. \(2 ) let @xmath262 be a nonempty reduced cyclic word in @xmath263 represented by a word @xmath254 .
decompose @xmath262 into @xmath264 where all letters in @xmath257 have positive ( resp .
negative ) exponents , and all letters in @xmath258 have negative ( resp .
positive ) exponents ( taking subindices modulo @xmath265 )
. then the _ cyclic _ sequence of positive integers @xmath266 is called the _ cyclic @xmath2-sequence of @xmath262_. here the double parentheses denote that the sequence is considered modulo cyclic permutations . in the above definition , by a _ cyclic word _ , we mean the set of all cyclic permutations of a cyclically reduced word . by @xmath267
, we denote the cyclic word associated with a cyclically reduced word @xmath261 .
[ def4.1(3 ) ] for a rational number @xmath8 with @xmath268 , let @xmath201 be the word in @xmath207 defined in section [ group_presentation ] .
then the symbol @xmath269 ( resp .
@xmath270 ) denotes the @xmath2-sequence @xmath271 of @xmath201 ( resp .
cyclic @xmath2-sequence @xmath272 of @xmath273 ) , which is called the _ s - sequence of slope @xmath8 _ ( resp .
cyclic s - sequence of slope @xmath8 _ ) .
we can easily observe the following .
@xmath274 where @xmath275 is the natural decomposition of @xmath201 obtained at the end of the last section .
it is also not difficult to observe @xmath276 and @xmath277 . by setting @xmath278 and @xmath279
, we have the following key propositions . [ sequence ] the decomposition @xmath280 satisfies the following . 1 .
each @xmath281 is symmetric , i.e. , the sequence obtained from @xmath281 by reversing the order is equal to @xmath281 .
( here , @xmath282 is empty if @xmath283 . ) 2 .
each @xmath281 occurs only twice in the cyclic sequence @xmath270 .
3 . set @xmath284 .
then @xmath269 consists of only @xmath285 and @xmath286 , and @xmath282 begins and ends with @xmath286 , whereas @xmath287 begins and ends with @xmath285 .
[ connection ] let @xmath288 be as in proposition [ sequence ] .
for a rational number @xmath53 with @xmath289 , suppose that the cyclic @xmath2-sequence @xmath290 contains both @xmath282 and @xmath287 as a subsequence
. then @xmath291 .
a subset @xmath38 of the free group @xmath187 is called _ symmetrized _ , if all elements of @xmath38 are cyclically reduced and , for each @xmath292 , all cyclic permutations of @xmath254 and @xmath293 also belong to @xmath38 .
suppose that @xmath38 is a symmetrized subset of @xmath187 .
a nonempty word @xmath261 is called a _ piece _ if there exist distinct @xmath294 such that @xmath295 and @xmath296 .
small cancellation conditions @xmath297 and @xmath298 , where @xmath126 and @xmath127 are integers such that @xmath68 and @xmath299 , are defined as follows ( see @xcite ) . 1 .
condition @xmath297 : if @xmath292 is a product of @xmath300 pieces , then @xmath301 .
condition @xmath298 : for @xmath302 with no successive elements @xmath303 an inverse pair @xmath304 mod @xmath305 , if @xmath306 , then at least one of the products @xmath307 @xmath308 , @xmath309 is freely reduced without cancellation .
the following proposition enables us to apply small cancellation theory to the group presentation @xmath310 of @xmath64 .
[ small_cancellation_condition ] let @xmath8 be a rational number such that @xmath311 , and let @xmath38 be the symmetrized subset of @xmath312 generated by the single relator @xmath201 of the group presentation @xmath313 .
then @xmath38 satisfies @xmath314 and @xmath315 .
this proposition follows from the following characterization of pieces , which in turn is proved by using proposition [ sequence ] .
[ cor : max_piece_2 ] ( 1 ) a subword @xmath254 of the cyclic word @xmath316 is a piece if and only if @xmath317 does not contain @xmath282 as a subsequence and does not contain @xmath287 in its interior , i.e. , @xmath317 does not contain a subsequence @xmath318 for some @xmath319 .
\(2 ) for a subword @xmath254 of the cyclic word @xmath316 , @xmath254 is not a product of two pieces if and only if @xmath317 either contains @xmath320 as a proper initial subsequence or contains @xmath321 as a proper terminal subsequence .
let @xmath38 be the symmetrized subset of @xmath312 generated by the single relator @xmath201 of the group presentation @xmath322 .
suppose on the contrary that @xmath23 is null - homotopic in @xmath20 , i.e. , @xmath323 in @xmath64 , for some @xmath62 .
then there is a _ van kampen diagram _ @xmath96 over @xmath324 such that the boundary label is @xmath325 . here
@xmath96 is a simply connected @xmath4-dimensional complex embedded in @xmath14 , together with a function @xmath326 assigning to each oriented edge @xmath327 of @xmath96 , as a _ label _ , a reduced word @xmath328 in @xmath207 such that the following hold . 1
if @xmath327 is an oriented edge of @xmath96 and @xmath329 is the oppositely oriented edge , then @xmath330 .
2 . for any boundary cycle @xmath331 of any face of @xmath96
, @xmath332 is a cyclically reduced word representing an element of @xmath38 .
( if @xmath333 is a path in @xmath96 , we define @xmath334 . )
we may assume @xmath96 is _ reduced _ , namely it satisfies the following condition : let @xmath335 and @xmath336 be faces ( not necessarily distinct ) of @xmath96 with an edge @xmath337 , and let @xmath338 and @xmath339 be boundary cycles of @xmath335 and @xmath336 , respectively .
set @xmath340 and @xmath341 .
then we have @xmath342 .
moreover , we may assume the following conditions : 1 .
@xmath343 for every vertex @xmath344 .
2 . for every edge @xmath327 of @xmath345 , the label
@xmath328 is a piece .
3 . for a path @xmath346 in @xmath345 of length @xmath347
such that the vertex @xmath348 has degree @xmath4 for @xmath349 , @xmath350 can not be expressed as a product of less than @xmath300 pieces . since @xmath38 satisfies the conditions @xmath314 and @xmath315 by proposition [ small_cancellation_condition ]
, @xmath96 is a @xmath351$]-map , i.e. , 1 .
@xmath352 for every vertex @xmath344 ; 2 .
@xmath353 for every face @xmath354 . here
, @xmath355 , the _ degree of @xmath261 _ , denotes the number of oriented edges in @xmath96 having @xmath261 as initial vertex , and @xmath356 , the _ degree of @xmath357 _ , denotes the number of oriented edges in a boundary cycle of @xmath357 .
now , for simplicity , assume that @xmath96 is homeomorphic to a disk .
( in general , we need to consider an extremal disk of @xmath96 . )
then by the curvature formula of lyndon and schupp ( see ( * ? ? ?
* corollary v.3.4 ) ) , we have @xmath358 by using this formula , we see that there are three edges @xmath359 , @xmath360 and @xmath361 in @xmath345 such that @xmath362 and @xmath363 , where @xmath364 for each @xmath365 . since @xmath366 is not expressed as a product of two pieces , we see by proposition [ cor : max_piece_2 ] that the boundary label of @xmath96 contains a subword , @xmath254 , with @xmath367 or @xmath368 .
this in turn implies that the @xmath2-sequence of the boundary label contains both @xmath282 and @xmath287 as subsequences .
hence , by proposition [ connection ] , we have @xmath369 , a contradiction .
suppose , for two distinct @xmath60 , the unoriented loops @xmath23 and @xmath55 are homotopic in @xmath20 . then there is a reduced annular @xmath38-diagram @xmath96 such that @xmath325 is an outer boundary label and @xmath370 is an inner boundary label of @xmath96 .
again we can see that @xmath96 is a @xmath351$]-map and hence we have the following curvature formula : @xmath371 by using this formula , we obtain the following very strong structure theorem for @xmath96 , which plays key roles throughout the series of papers @xcite . [ cor : structure ] figure [ layer](a ) illustrates the only possible type of the outer boundary layer of @xmath96 , while figure [ layer](b ) illustrates the only possible type of whole @xmath96 .
( the number of faces per layer and the number of layers are variable . ) in the above theorem , the _ outer boundary layer _ of the annular map @xmath96 is the submap of @xmath96 consisting of all faces @xmath357 such that the intersection of @xmath372 with the outer boundary of @xmath96 contains an edge , together with the edges and vertices contained in @xmath372 .
the first paper @xcite of the series devoted to proof of theorem [ second_result ] treats the case when the 2-bridge link is a @xmath373-torus link , the second paper @xcite treats the case of 2-bridge links of slope @xmath374 and @xmath375 , where @xmath376 is an arbitrary integer , and the third paper @xcite treats the general case .
the two families treated in the second paper play special roles in the project in the sense that the treatment of these links form a base step of an inductive proof of the theorem for general @xmath4-bridge links .
we note that the figure - eight knot is both a 2-bridge link of slope @xmath374 with @xmath377 and a 2-bridge link of slope @xmath375 with @xmath378 .
surprisingly , the treatment of the figure - eight knot , the simplest hyperbolic knot , is the most complicated .
this reminds us of the phenomenon in the theory of exceptional dehn filling that the figure - eight knot attains the maximal number of exceptional dehn fillings .
h. akiyoshi , h. miyachi and m. sakuma , _ variations of mcshane s identity for punctured surface groups _ , proceedings of the workshop `` spaces of kleinian groups and hyperbolic 3-manifolds '' , london math .
soc . , lecture note series * 329 * ( 2006 ) , 151185 . | the purpose of this note is to announce complete answers to the following questions .
( 1 ) for an essential simple loop on a 2-bridge sphere in a 2-bridge link complement , when is it null - homotopic in the link complement ? ( 2 ) for two distinct essential simple loops on a 2-bridge sphere in a 2-bridge link complement , when are they homotopic in the link complement ?
we also announce applications of these results to character varieties and mcshane s identity . |
estimation of population covariance matrices from samples of multivariate data has draw many attentions in the last decade owing to its fundamental importance in multivariate analysis . with dramatic advances in technology in recent years , various research fields , such as genetic data , brain imaging , spectroscopic imaging , climate data and so on , have been used to deal with massive high - dimensional data sets , whose sample sizes can be very small relative to dimension . in such settings , the standard and
the most usual sample covariance matrices often performs poorly @xcite .
fortunately , regularization as a class of new methods to estimate covariance matrices has recently emerged to overcome those shortages of using traditional sample covariance matrices .
these methods encompass several specified forms , banding @xcite , tapering @xcite and thresholding @xcite for instance . moreover , there are many cases where the model is known to be structured in several ways at the same time . in recent years , one of research contents is to estimate a covariance matrix possessing both sparsity and positive definiteness .
for instance , rothman @xcite gave the following model : @xmath8 where @xmath9 is the sample covariance matrix , @xmath10 is the frobenious norm , @xmath11 is the element - wise @xmath0-norm and @xmath12 . from the optimization viewpoint , ( [ rest ] ) is similar to the graphical lasso criterion @xcite which also has a log - determinant part and the element - wise @xmath0-penalty .
rothman @xcite derived an iterative procedure to solve ( [ rest ] ) .
while xue , ma and zou @xcite omitted the log - determinant part and considered the positive definite constraint @xmath13 for some arbitrarily small @xmath14 : @xmath15 they utilized an efficient alternating direction method ( adm ) to solve the challenging problem ( [ xest ] ) and established its convergence properties .
most of the literatures , e.g.,@xcite , required the population covariance matrices being positive definite , and thus there is no essence of pursuing the low - rank of the estimator .
by contrast , newly appeared research topic is to consider simultaneously the sparsity and low - rank of a structured model , which implies that the population covariance matrices are no longer restricted to the positive definite matrix cone and can be relaxed to the positive semidefinite cone .
in addition , the models with structure of being simultaneously the sparsity and low - rank are widely applied into practice , such as sparse signal recovery from quadratic measurements and sparse phase retrieval , see @xcite for example . moreover , richard et al .
@xcite showed that both sparse and low - rank model can be derived in covariance matrix when the random variables are highly correlated in groups , which means this covariance matrix has a block diagonal structure . with stimulations of those ideas ,
we construct the following convex model encompassing the @xmath0-norm and nuclear norm for estimating the covariance matrix : @xmath16 where @xmath17 are tuning parameters .
the @xmath0-norm penalty @xmath18 is also called lasso - type penalty and is used to encourage sparse solutions . the nuclear norm , @xmath19 with @xmath20 being the eigenvalue of @xmath21 , is the trace norm when @xmath22 and ensures low - rank solutions of ( [ est ] ) . here
we inroduce the _ approximate rank _ to interpret the low - rank , which is defined as @xmath23 being the smallest number such that @xmath24 where @xmath25 are the singular value of @xmath26 with @xmath27 , and @xmath28 could be chosen based on the needs , throughout our paper we fix @xmath29 for simplicity .
the contributions of this paper mainly center on two aspects .
for one thing , being different from @xcite , we establish the theoretical statistical theory under different assumptions rather than giving the generalized error bound of the estimation .
especially , we acquire the estimation rate @xmath30 under the frobenious norm error , which improves the optimal rate @xmath31 where the low - rank property of the estimator does not be considered @xcite and @xmath32 is the samples dimension with @xmath33 . for another , we take advantage of the alternating direction method of multipliers ( admm ) , also can be seen in @xcite , to combat our problem ( [ est ] ) .
the organization of this paper is as follows . in section [ sec2 ]
we will present some theoretical properties of the estimator derived by the proposed model ( [ est ] ) .
after that the alternating direction method of multipliers ( admm ) is going to be introduced to combat the problem , and numerical experiments are projected to show the performance of this method in sections [ sec3 ] and [ sec4 ] respectively .
we make a conclusion in the last section .
before the main part , we hereafter introduce some notations . @xmath34 and @xmath35 denote the expectation of @xmath36 and the probability of the incident @xmath26 occurring respectively .
@xmath37 is the number of entries of the set @xmath38 .
normal distribution with mean @xmath39 and covariance @xmath21 is written as @xmath40 .
say @xmath41 if for every @xmath42 , there is a @xmath43 such that @xmath44 for all @xmath45 , and say @xmath46 if @xmath47 .
if there are two constants @xmath48 such that @xmath49 , we write as @xmath50 . for given
observed independently and identically distributed ( i.i.d . for short ) @xmath32-variate random variables
@xmath51 @xmath52 with covariance matrix @xmath53 and @xmath54 , the goal is to estimate the unknown matrix @xmath53 based on the sample @xmath55 .
this problem is called covariance matrix estimation which is of fundamental importance in multivariate analysis . given a random sample @xmath56 from @xmath57 ( without loss of generality ) and a population covariance matrix @xmath58 , the sample covariance matrix is @xmath59 where @xmath60 .
denote @xmath38 the support set of the population covariance matrix @xmath61 as @xmath62 [ ass1 ] for all @xmath32 , @xmath63 , where @xmath64 is a constant .
assumption [ ass1 ] is a common used condition in covariance matrix estimation , on which a useful lemma based is recalled here for the sequel analysis .
one can also refer it in @xcite . [ zz ]
let @xmath65 be i.i.d .
@xmath66 and assumption [ ass1 ] holds , _ _ ( _ _ i.e .
, for all @xmath32 , @xmath67__)__. then , if @xmath68 , @xmath69 where constants @xmath70 and @xmath71 depend on @xmath72 only .
another lemma which plays an important role in our main results is stated below .
[ i m ] suppose that assumption [ ass1 ] holds , @xmath73 .
then for @xmath14 sufficiently small , @xmath74 where @xmath75 is defined as @xmath76 .
* proof * first make the eigenvalue decomposition of @xmath61 ( @xmath77 ) as @xmath78 where @xmath79 with @xmath80 is the matrix composed of eigenvectors , @xmath81 is a diagonal matrix generated by eigenvalues with @xmath82 and @xmath83 . + by denoting @xmath84 with @xmath85 and @xmath86 , which implies @xmath87 , we consider the model @xmath88 clearly , from ( [ est ] ) we have @xmath89 which implies @xmath90 . for a given @xmath14 sufficiently small and any @xmath91 ( i.e. , @xmath92 ) , we compute @xmath93 for convenience we denote @xmath94
. then for i , it holds @xmath95 for ii , we obtain by noting @xmath96 and @xmath97 that , @xmath98 from the h@xmath99lder inequality , one can prove that @xmath100 for iii , combining with ( [ hd1 ] ) we get that @xmath101 since @xmath102 and ( [ hd2 ] ) , @xmath103 therefore , by @xmath73 @xmath104 hence we prove that if @xmath73 and @xmath102 , for any @xmath91 , it holds @xmath105 in addition , from ( [ delta ] ) , we have @xmath106 which implies that @xmath107 .
otherwise , we suppose @xmath108 , then @xmath109 . since for any @xmath91 , it follows @xmath110 which is contradicted with the fact @xmath111 is a convex function and @xmath112 , because @xmath113 finally @xmath107 indicates that @xmath114 due to @xmath115 . hence the desired result is obtained.@xmath116 then in order to acquiring the rate of the estimation , the two following commonly used assumptions are needed to introduced , and
also can be seen @xcite .
assumption [ ass2 ] holds , for example , if @xmath117 are gaussian .
[ ass2 ] @xmath118 hold for all @xmath119 and @xmath120 , where @xmath121 and @xmath122 are two constants .
[ ass3 ] @xmath123 hold for all @xmath124 , where some @xmath125 and @xmath126 is a constant . built on the two assumptions ,
we give our main results with regard to rates of the estimator of ( [ est ] ) .
[ th1 ] for some @xmath127 , let @xmath128 be a sufficiently large constant and suppose assumptions [ ass1 ] and [ ass2 ] hold . if @xmath129 and @xmath130 , then @xmath131 * proof * since assumptions [ ass1 ] and [ ass2 ] hold , a fact employed by rothman et al .
@xcite is that for @xmath132 sufficiently small , @xmath133 where @xmath134 and @xmath135 are some constants .
we then apply the bound @xmath130 and lemma [ i m ] with @xmath136 to obtain that @xmath137 evidently , @xmath138 can be arbitrarily close to one by choosing @xmath128 sufficiently large.@xmath116 [ co1 ] for some @xmath127 , let @xmath128 be a sufficiently large constant such that @xmath139 , and suppose assumptions [ ass1 ] , [ ass2 ] hold .
if @xmath140 and @xmath141 , then @xmath142 clearly , if the @xmath143 in assumption [ ass1 ] , the better rate @xmath144 would reduce to @xmath145 .
it is worth mentioning that under the the assumption [ ass1 ] , the minimax optimal rate of convergence under the frobenius norm in theorem 4 of @xcite is @xmath145 which also has been obtained by @xcite .
however , to attain the same rate in the presence of the log - determinant barrier term ( [ rest ] ) , rothman @xcite instead would require that @xmath146 , the minimal eigenvalue of the true covariance matrix , should be bounded away from zero by some positive constant , and also that the barrier parameter should be bounded by some positive quantity .
@xcite illustrated this theory requiring a lower bound on @xmath146 is not very appealing .
[ th2 ] let @xmath147 be a sufficiently large constant and suppose that assumptions [ ass1 ] and [ ass3 ] hold . if @xmath148 and @xmath149 , then @xmath150 * proof * since assumptions [ ass1 ] and [ ass3 ] hold , one can modify a result of bickel & levina ( 2008a ) and show that for @xmath132 sufficiently small @xmath151 where @xmath152 is a constant .
we then apply the bound @xmath149 and lemma [ i m ] with @xmath153 to get @xmath154 apparently , the bound @xmath155 can be arbitrarily close to one by taking @xmath147 sufficiently large.@xmath116
in this section , we will construct the alternating direction method of multipliers ( admm ) to solve problem ( [ est ] ) . by introducing an auxiliary variable @xmath156 , problem ( [ est ] )
can be rewritten as @xmath157 the constraint @xmath158 can be put into the objective function by using an indicator function : ( 0)= 0 , 0 + + , .
this leads to the following equivalent reformulation of ( [ est1 ] ) : @xmath159 recently , the alternating direction method of multipliers ( admm ) has been studied extensively for solving ( [ est2 ] ) . a typical iteration of admm for solving ( [ est2 ] )
can be described as [ ad1]^k+1:= _ ( ^k,,^k ) + [ ad2]^k+1:= _ ( , ^k+1,^k ) + [ ad3]^k+1:=^k-(^k+1-^k+1 ) , where the augmented lagrangian function @xmath160 is defined as @xmath161 in which @xmath162 is the lagrange multiplier and @xmath163 is a penalty parameter .
note that admm ( [ ad1]-[ad3 ] ) can be written explicitly as + [ ad11]^k+1:=_0 _ * + -(^k+^k)_f^2 + [ ad22]^k+1:= _ _ 1 + -(_n+^k+1-^k)_f^2 + [ ad33]^k+1:=^k-(^k+1-^k+1)/. we now show that the two subproblems ( [ ad11 ] ) and ( [ ad22 ] ) can be easily solved . for the subproblem ( [ ad11 ] ) , @xmath164 where @xmath165 denote the projection of a matrix @xmath36 onto the convex positive semidefinite cone @xmath166 . namely @xmath167 , where @xmath168 and @xmath169 the solution of the second subproblem ( [ ad22 ] ) is given by the @xmath0-shrinkage operation @xmath170 where @xmath171 and @xmath172 is a sign function . therefore , combining with ( [ ad33])-([shr1 ] ) , the whole algorithm is written as follows >
l * admm : alternating direction method of multipliers * + initialize @xmath173 + * repeat * + compute @xmath174 ; + compute @xmath175 + compute @xmath176 + * untill * convergence + * return * + to end this section , we prove that the sequence @xmath177 produced by the alternating direction method of multipliers ( table [ admm ] ) converges to @xmath178 , where @xmath179 is an optimal solution of ( [ est2 ] ) and @xmath180 is the optimal dual variable .
now we label some necessary notations for the ease of presentation .
let @xmath181 be a @xmath182 matrix defined as @xmath183 the weighted norm @xmath184 stands for @xmath185 and the corresponding inner product @xmath186 is @xmath187 . before presenting the main theorem with regard to the global convergence of admm
, we introduce the following lemma .
[ conle ] assume that @xmath188 is an optimal solution of @xmath189 and @xmath180 is the corresponding optimal dual variable associated with the equality constraint @xmath190
. then the sequence @xmath191 produced by admm satisfies @xmath192 where @xmath193 and @xmath194 .
based on the lemma above , the convergent theorem can be derived immediately .
[ conth]the sequence @xmath191 generated by algorithm 1 from any starting point converges to an optimal solution of @xmath189 .
in this section we will exploit the proposed method admm to tackle two examples , one of which possessed the block structured population covariance matrix , and another utilized the banded population covariance matrix .
actually as the constraint @xmath22 , our proposed model ( [ est ] ) is equivalent to @xmath195 so similar to the method in @xcite , one can solve the soft - thresholding estimator @xmath196 to initialize the @xmath197 . if the derived @xmath198 then the recovered sparse and low - rank semidefinite estimator @xmath199 . in our stimulation
, we uniformly initialize @xmath200 as the matrix with all entries being 1 , @xmath197 as zero matrix and @xmath201 respectively . unlike @xmath202 and @xmath203
, @xmath39 does not change the final covariance estimator , thus we fixed @xmath204 just for simplicity and the stop criteria is set as @xmath205 for the sample dimensions , we always take @xmath206 and @xmath207 .
analogous to the model , modified slightly here , emerged in @xcite who synthesized @xmath4 samples @xmath208 for a block diagonal population covariance matrix @xmath209 , we will use @xmath210 blocks of random sizes , and each block is generated by @xmath211 where the entries of @xmath212 are drawn i.i.d . from the uniform distribution on @xmath213 $ ] .
evidently , the rank of @xmath61 produced in the way is @xmath214 .
corresponding matlab code of generating @xmath215 is @xmath216 , thereby deriving the sample covariance matrix @xmath217 what is worth mentioning is that if @xmath61 is a positive definite matrix , the solution our method obtains would be a positive definite matrix with full rank .
but fortunately , compared to some largest singular values of @xmath61 , the left are relatively small so that can be ignored . here
, therefore , we consider the approximate rank ( [ apr ] ) . in addition , we say the sparsity of a matrix @xmath218 by @xmath219 captypefigure captypefigure apart from the approximate @xmath220 and the sparsity @xmath221 of the sparse and low - rank semidefinite estimator @xmath222 , we also take advantage of other two types of errors to show the selection performance of our proposed method admm : @xmath223 where @xmath224 stands for the false positive rate , which means the rate of significant variables that are unselected over the whole zero entries , and @xmath225 denotes the true positive rate , which implies the ratio of significant variables that are selected over the entire none zero elements . for more visualized purpose , we plot the population covariance , sample covariance an the recovered covariance . from figures [ fig1 ] and [ fig2 ] ,
the left population covariance is @xmath61 , the median sample covariance stands for @xmath9 and the right recovered covariance denotes @xmath75 .
the yellow region stands for the sparse area in which the values are quite close ( or most of them equal ) to zero , while the green zone is the place where the none zero entries locate .
moreover the deeper the green color is , the larger the value stands .
evidently the recovered covariance matrices are quite dependent on the sample covariance matrix .
captypefigure captypefigure
we then report average results over 100 runs .
timing ( in seconds ) was carried out on a cpu 2.6ghz desktop . in computation , the sparse parameter @xmath202 and the low - rank parameter @xmath203 are given in corresponding stimulations respectively .
@xmath197 ia taken as the zero matrix .
as we mentioned before the rank of the produced @xmath61 actually is given , i.e. , @xmath226 .
table [ tab1 ] shows the performance of our approach under different dimensions @xmath207 and various blocks @xmath214 .
obviously , all the @xmath227 nearly tends to the true rank @xmath214 .
the values of @xmath224 and @xmath225 are quite desirable , which manifests that the selection performance of admm is very well .
moreover , the cpu time reveals the method runs relatively fast .
in addition , with the @xmath214 rising , for example @xmath228 , the @xmath224 is decreasing while @xmath224 is increasing , and the time spent by the method is also ascending .
in addition , for small dimension such as @xmath229 and large block such as @xmath230 , admm performs unstably because various stimulations can not be recovered .
c c c c c c c c + & @xmath32 & @xmath227 & @xmath231 & @xmath232 & @xmath224&@xmath225 & time + & 100 & 5.0 & 0.2356 & 0.1148 & 0.1392 & 0.9863 & 0.775 + @xmath233&200 & 5.0 & 0.2362 & 0.1121 & 0.1414 & 0.9911 & 3.364 + & 500&5.4&0.2225 & 0.0990 & 0.1399 & 0.9748 & 32.49 + & 200 & 9.90 & 0.1228 & 0.0565 & 0.0716 & 0.9766 & 3.580 + @xmath234&500&10.8 & 0.1163 & 0.0598 & 0.0649 & 0.9371 & 33.03 + & 1000&12.0&0.1199 & 0.0604 & 0.0692 & 0.9092 & 321.6 + & 200 & 19.3 & 0.0601 & 0.0339 & 0.0324 & 0.8571 & 4.165 + @xmath235&500 & 20.5 & 0.0569 & 0.0268 & 0.0328 & 0.9368 & 38.81 + & 1000&23.5 & 0.0564 & 0.0291 & 0.0316 & 0.8821 & 368.7 + & 200 & 181.0 & 0.0364 & 0.0189 & 0.0201 & 0.8850 & 16.09 + @xmath230&500&47.0 & 0.0223 & 0.0122 & 0.0127 & 0.8006 & 49.43 + & 1000&54.2&0.0221 & 0.0134 & 0.0123 & 0.7447 & 449.5 + to simply observe the performance of our proposed method under different parameters @xmath236 and initialized @xmath197 , we fix @xmath237 and @xmath238 . from table
[ tab11 ] , results of left columns of @xmath239 and time are generated from the initialized @xmath240 , and results of right columns are produced with @xmath241 .
one can easily to discern that when @xmath242 and @xmath241 , the performance of the method is relatively bad regardless of what @xmath203 is taken due to the @xmath227 and @xmath225 are undesirable . by contrast , when @xmath243 and @xmath244 , it behaves much better , particularly when @xmath245 .
moreover , with the increasing of @xmath202 , the rate @xmath224 of significant variables that are unselected is rising , even though the percentage @xmath225 of significant variables that are selected is ascending as well . by comparing the effectiveness of those two initialized point @xmath240 and @xmath201 , as shown in the table , @xmath246 and time generated from @xmath240
are basically same , which means method with zero starting point performs more stable .
but when @xmath243 and @xmath244 , admm with @xmath241 generates larger @xmath225 than that from @xmath240 , moreover it needs less computational time in all stimulations regardless of the parameters .
c c c c c c + @xmath202 & @xmath203 & @xmath227 & @xmath224&@xmath225 & time + & 0.05&5.0 12.0&0.1365 0.0565 & 1.0000 0.5666 & 3.379 2.716 + @xmath247&0.25&5.5 8.5&0.1313 0.0507 & 0.8813 0.4783 & 4.857 3.803 + & 0.50&5.0 7.5&0.1303 0.0615 & 0.9967 0.4273 & 3.604 4.040 + & 0.05&5.0
5.0&0.1371 0.1362 & 0.9980 0.9843 & 3.584 1.885 + @xmath248&0.25&5.0 5.0&0.1367 0.1365 & 0.9972 0.9972 & 3.963 2.897 + & 0.50&5.2 5.2&0.1318 0.1317 & 0.9653 0.9661 & 4.134 3.477 + & 0.05&5.0 5.0 & 0.1375 0.2008 & 1.0000 1.0000 & 3.319 1.822 + @xmath244&0.25&5.4 5.8&0.1575 0.2169 & 0.9432 1.0000 & 4.068 2.580 + & 0.50&5.0
6.0&0.1451 0.2035 & 0.9881 1.0000 & 3.933 3.791 + in this part we consider the population covariance matrix with banded structure which has been emerged in @xcite . to be more exact , the population covariance matrix @xmath249 has the following formula @xmath250 we first report average results over 100 replicators and take the sparse parameter @xmath251 and the low - rank parameter @xmath252 respectively .
information listed in table [ tab2 ] shows the performance of our approach under different dimensions @xmath207 and two distinct starting point @xmath253 and @xmath201 .
c c c c c c c c c + & @xmath32 & @xmath254 & @xmath227 & @xmath231 & @xmath232 & @xmath224&@xmath225 & time + & 100&90 & 33.3 & 0.1810 & 0.1209 & 0.0718 & 0.9750 & 1.421 + @xmath240&200&176&66.6&0.0927 & 0.0606 & 0.0351 & 0.9851 & 7.560 + & 500&487&327.0 & 0.0376 & 0.0211 & 0.0174 & 0.9791 & 63.06 + & 1000&847&334.2&0.0189&0.0133&0.0067&0.9233&851.3 + & 100&90 & 33.3 & 0.1810 & 0.1209 & 0.0718 & 0.9750 & 1.417 + @xmath241&200&176 & 66.6 & 0.0927 & 0.0606 & 0.0351 & 0.9851 & 7.576 + & 500&487&327.0 & 0.0376 & 0.0211 & 0.0174 & 0.9791 & 60.02 + & 1000&847&334.2&0.0189&0.0133&0.0067&0.9233&866.2 + as we can discern in table [ tab2 ] , compared with @xmath61 , the @xmath227 and @xmath232 are relatively small , and the former ascends while the latter descends with the rise of @xmath32 .
in addition , in example [ ex1 ] the block structured @xmath61 whose @xmath255 leads to the estimator the rank of @xmath75 is also close to @xmath214 .
being distinct with that , in this example , the @xmath254 increases with the dimension @xmath32 and is not low - rank , but the recovered solution @xmath75 has been rendered the relatively low - rank property .
the values of @xmath224 and @xmath225 are both quite desirable , which manifests that the selection performance of admm is very well in this example .
moreover , the cpu time reveals the method runs extremely fast as well .
in addition , under such parameters z@xmath256 , admm behaves nearly identically even though the starting point @xmath197 are different .
captypefigure captypefigure then from figure [ fig3 ] , one can check that the recovered covariance matrices @xmath75 are quite dependent on the sample covariance matrix @xmath9 , and the selection performance are relatively well because @xmath224 is pretty small while @xmath225 is close to @xmath257 . to simply observe the behavior of admm under different parameters @xmath236 and initialized @xmath197 , we fix @xmath237 and @xmath258 . as indicated in table
[ tab21 ] , results of left columns of @xmath239 and time are generated from the initialized point @xmath240 , and results of right columns are produced with @xmath241 .
it is clear that the cpu time cost by the method with starting point @xmath201 is almost less than that from zero initialization . with the @xmath202 rising ,
tpr generated by the method with @xmath241 is increasing to 1 , while that with @xmath240 basically stabilizes at 0.97 . in terms of the @xmath227 , the proposed method with starting point @xmath201
will not create a lower rank solution , comparing with @xmath240 .
the reason for this phenomenon probably is that @xmath201 is a sparse point but not low rank ; from ( [ este ] ) if @xmath201 is an approximately semidefinite positive matrix , algorithm will stop after a few iterations which results in the solution is not a desired low - rank one .
c c l c c c + @xmath202 & @xmath203 & @xmath227 & @xmath224&@xmath225 & time + & 0.25&64.2 126.2 & 0.0356 0.0168 & 0.9835 0.4383 & 7.439 5.725 + @xmath248&0.50 & 66.0 95.2 & 0.0345 0.0126 & 0.9888 0.3877 & 8.682 8.474 + & 0.75 & 62.6 78.0 & 0.0381 0.0102 & 0.9780 0.3986 & 7.852 11.05 + & 0.25 & 67.4 103.0&0.0323 0.0377 & 0.9528 0.9580 & 8.996 4.818 + @xmath244&0.50&65.8 80.0&0.0370 0.0385 & 0.9834 0.9845 & 8.445 7.148 + & 0.75&67.8 67.8&0.0353 0.0353 & 0.9783 0.9783 & 8.728 8.451 + & 0.25 & 67.8 167.0&0.0336 0.0726 & 0.9792 1.0000 & 9.508 2.510 + @xmath259&0.50&64.0 157.8 & 0.0351 0.0750 & 0.9807 1.0000 & 8.596 1.802 + & 0.75&68.0 166.4&0.0322 0.0731&0.9696 1.0000&8.470 2.183 +
we have acquired a positive semidefinite estimator , being simultaneously sparse and low - rank , from samples of the covariance matrices through utilizing @xmath0 norm and nuclear norm penalties .
the theoretical properties manifest that in high - dimensional settings the estimator we have constructed performs very well .
meantime , the efficient admm with global convergence has possessed several merits illustrated by the numerical simulations , such as less computational time and beautiful recovered effectiveness .
the work was supported in part by the national basic research program of china ( 2010cb732501 ) , the national natural science foundation of china ( 11171018 , 71271021,11301022 ) .
77 bickel , p. and levina , e. : regularized estimation of large covariance matrices .
statist . *
36 * , 199 - 227 ( 2008 ) .
bickel , p. and levina , e. : covariance regularization by thresholding .
statist . *
36 * , 2577 - 2604 ( 2008 ) .
cai , t. and liu , w. : adaptive thresholding for sparse covariance matrix estimation . j.
assoc . * 106 * , 1 - 13 ( 2011 ) .
cai , t. , zhang , c. : and zhou , h. , optimal rates of convergence for covariance matrix estimation .
statist . *
38 * , 2118 - 2144 ( 2010 ) .
cai t t , yuan m. : adaptive covariance matrix estimation through block thresholding .
ann.statist . * 40 * , 2014 - 2042 ( 2012 ) .
cai , t. and zhou , h. : minimax estimation of large covariance matrices under @xmath0-norm .
sinica * 22 * , 1319 - 1378 ( 2012 ) .
cai , t. and zhou , h. : optimal rates of convergence for sparse covariance matrix estimation . ann.statist . * 40 * , 2389 - 2420 ( 2012 ) .
el karoui , n. : operator norm consistent estimation of large dimensional sparse covariance matrices .
statist . *
36 * , 2717 - 2756 ( 2008 ) .
friedman , j. , hastie , t. and tibshirani , r. , sparse inverse covariance estimation with the graphical lasso .
* 9 * , 432 ( 2008 ) .
furrer , r. and bengtsson , t. : estimation of high - dimensional prior and posterior covariance matrices in kalman filter variants .
j. multivariate anal . * 98 * , 227 - 255 ( 2007 ) .
johnstone , i. : on the distribution of the largest eigenvalue in principal components analysis .
, 295 - 327 ( 2001 ) .
liu h , wang l , zhao t. : sparse covariance matrix estimation with eigenvalue constraints .
journal of computational and graphical statistics , ( 2013 accepted ) .
oymak , s. , jalali , a. , fazel , m. , eldar , y. c. and hassibi , b. : simultaneously structured models with application to sparse and low - rank matrices .
arxiv preprint arxiv:1212 - 3753 ( 2012 to appear ) .
richard , e. , savalle , p. a. , and vayatis , n. : estimation of simultaneously sparse and low rank matrices .
arxiv preprint arxiv : 64 - 74 ( 2012 to appear ) .
rothman , a. : positive definite estimators of large covariance matrices .
biometrika * 99 * , 733 - 740 ( 2012 ) .
rothman , a. , levina , e. , and zhu , j. : generalized thresholding of large covariance matrices .
* 104 * , 177 - 186 ( 2009 ) .
wu , w. and pourahmadi , m. : nonparametric estimation of large covariance matrices of longitudinal data .
biomet . * 90 * , 831 - 844 ( 2003 ) .
xue , l. , ma , s. and zou , h. : positive definite @xmath0 penalized estimation of large covariance matrices .
assoc . , * 107 * , 1480 - 1491 ( 2012 ) .
yuan x.m . : alternating direction method of multipliers for covariance selection models .
journal of scientific computing * 51 * , 261 - 273 ( 2012 ) .
* proof of lemma [ conle ] * for convenience , we denote that @xmath260 clearly , @xmath261 is a convex function .
since @xmath188 is an optimal solution of @xmath189 , which satisfies the following kkt conditions : @xmath262 note that the optimality conditions for the first subproblem in admm , i.e. , the subproblem with respect to @xmath156 in @xmath263 , are given by @xmath264 this together with @xmath265 , i.e. , @xmath266 , we have @xmath267 combining @xmath268 and @xmath269 and using the fact that @xmath270 is a monotone operator , we get @xmath271 the optimality conditions for the second subproblem ( i.e. , the subproblem with respect to @xmath21 ) in @xmath272 are given by @xmath273 this together with @xmath265 , i.e. , @xmath266 , we have @xmath274 similarly , combining @xmath275 and @xmath276 , using the fact that @xmath277 is a monotone operator , we get @xmath278 the summation of @xmath279 and @xmath280 gives @xmath281 where the equality because of @xmath282 and @xmath283 simple algebraic derivation from @xmath284 yields the following inequality : @xmath285 rearranging the right hand side of @xmath286 using @xmath287 and @xmath288 , then @xmath286 can be reduced to @xmath289 using the notation of @xmath290 and @xmath291 , the inequality above can be rewritten as @xmath292 combining @xmath293 with the following identity @xmath294 we get @xmath295 now , using @xmath276 for @xmath296 instead of @xmath297 , we get , @xmath298 combining @xmath299 and using the fact that @xmath300 is a monotone function , we obtain , @xmath301 which means @xmath302 by substituting @xmath303 into @xmath304 , we get the desired result @xmath305.@xmath116 * proof of theorem [ conth ] * from lemma [ conle ] , we can get that + ( a ) @xmath306 ; + ( b ) @xmath307 lies in a compact region ; + ( c ) @xmath308 is monotonically non - increasing and thus converges .
+ connecting with notations of @xmath309 and @xmath310 , it holds that @xmath311 and @xmath312 , which together with @xmath265 imply that @xmath313 and @xmath314 . from @xmath315
, @xmath307 has a subsequence @xmath316 that converges to @xmath317 , i.e. , @xmath318 and @xmath319 .
also we have @xmath320 from @xmath314 .
therefore , @xmath321 is a limit point of @xmath191 .
note that @xmath322 and @xmath323 respectively imply that @xmath324 those together with @xmath325 , it reduces that @xmath326 @xmath327 and @xmath325 mean that @xmath321 is an optimal solution to @xmath189 .
therefore , we showed that any limit point of @xmath191 is an optimal solution to @xmath189.@xmath116 | this paper aims at achieving a simultaneously sparse and low - rank estimator from the semidefinite population covariance matrices.we first benefit from a convex optimization which develops @xmath0-norm penalty to encourage the sparsity and nuclear norm to favor the low - rank property . for the proposed estimator
, we then prove that with large probability , the frobenious norm of the estimation rate can be of order @xmath1 under a mild case , where @xmath2 and @xmath3 denote the number of sparse entries and the rank of the population covariance respectively , @xmath4 notes the sample capacity .
finally an efficient alternating direction method of multipliers with global convergence is proposed to tackle this problem , and meantime merits of the approach are also illustrated by practicing numerical simulations .
[ section ] [ theorem]lemma [ theorem]corollary [ theorem]remark [ theorem]proposition [ theorem]assumption [ theorem]definition [ theorem]construction [ theorem]example [ theorem ] [ claim_nb]claim a shenglong zhou@xmath5{0.98,0.00,0.00}{*}}$ ] , naihua xiu@xmath5{0.98,0.00,0.00}{*}}$ ] , ziyan luo@xmath5{0.98,0.00,0.00}{+}}$ ] , lingchen kong + @xmath6{0.98,0.00,0.00}{^*}~$]department of applied mathematics + @xmath7{0.98,0.00,0.00}{^+}}$ ] state key laboratory of rail traffic control and safety + @xmath7{0.98,0.00,0.00}{^{\underset{+}{*}}}}$ ] beijing jiaotong university , beijing 100044 , p. r. china + * keywords : * covariance matrices , sparse , low - rank , rate of estimation , alternating direction method of multipliers |
the dominant formation mechanism of early - type galaxies remains one of the long - standing debates of modern astrophysics . the classical monolithic model ( e.g. * ? ? ?
* ; * ? ? ?
* ) suggests that early - type galaxies form in highly efficient starbursts at high redshifts and evolve without much residual star formation from that point onward . on the other hand ,
the popular lambda cold dark matter ( @xmath9cdm ) paradigm strongly suggests a hierarchical merger scenario for massive early - type galaxies ( e.g. * ? ? ? * ) . in this model ,
early - type galaxies form as the result of successive mergers and are thought to have continued or episodic star formation events .
numerous tests discriminate between these scenarios .
for example , the optical colour - magnitude relations ( cmrs ) of early - type galaxies consistently reveal a small scatter around the mean relation ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , apparently supporting the monolithic scenario .
deep imaging surveys have however shown that many early - types possess shells , tidal features ( e.g. * ? ? ?
* ) and signatures of ongoing or recent star formation ( rsf ; e.g. ngc 2865 , * ? ? ?
* ; * ? ? ?
integral - field spectroscopy now also allows to obtain spatially resolved maps of various galaxy properties .
for example , we have previously highlighted the importance of central stellar discs and kinematically - decoupled cores in the sauron sample ( e.g. , @xcite , hereafter @xcite ; @xcite , hereafter @xcite ) .
more recently , the _ galaxy evolution explorer _ ( _ galex _ ) satellite opened up the ultraviolet ( uv ) window , allowing to probe the recent star formation history of galaxies with much greater accuracy than was previously possible with only optical information .
the rest - frame uv is one order of magnitude more sensitive than the optical to the presence of hot stars , easily revealing populations younger than @xmath10 gyr .
@xcite used _ galex _ medium imaging survey data to construct the first near ultraviolet ( nuv ) cmr of early - type galaxies classified by the sloan digital sky survey ( sdss ) .
they found a remarkably large scatter towards blue colours and interpreted it as evidence of recent star formation . @xcite and
@xcite subsequently found that the fraction of early - type galaxies that experienced recent star formation in the last billion years can be greater than @xmath11 per cent .
@xcite further concluded that the star formation rates of nearby early - type galaxies derived from mid - infrared imaging are in good agreement with the uv results .
there is thus clear evidence for residual star formation in the local early - type galaxy population and for its influence on global scaling relations , at least colour - magnitude relations .
the fundamental plane ( fp ) is another key scaling relation of early - type galaxies , a two - dimensional plane in the three - dimensional manifold of their global structural parameters ( effective radius , surface brightness and stellar velocity dispersion ; e.g.@xcite ) , and may also be affected by rsf . under the assumption of structural homology ( i.e. that all early - type galaxies have the same mass distribution and kinematics ) , the virial theorem implies that the fp parameters should scale in a specific manner , but observations reveal a _ tilt _ of the fp away from the virial prediction .
numerous studies have investigated this , and the observed tilt has been attributed to a variation of the mass - to - light ratio across the sequence of early - type galaxies and/or to the breaking of the homology assumption ( see , e.g. , @xcite ; @xcite , hereafter @xcite ; @xcite ) .
nevertheless , much debate still exists on the origin of the tilt and the scatter about the mean relation ( e.g.@xcite and references therein ) .
spatially - resolved galaxy data provide valuable information on the details of the stellar population distributions , especially when combined with uv data .
hence , we aim here to obtain a uv database matching the sauron integral - field data , and present spatially - resolved uv and optical imaging of most early - type galaxies in the sauron survey .
numerous investigations can be performed with such a database , but we will focus here on revisiting the effects of star formation on the scaling relations of early - types , particularly colour - magnitude relations and the fundamental plane .
longstanding enigmas such as the uv - upturn phenomenon ( see @xcite and references therein ) are closely related . in this paper , we thus present and discuss our uv imaging of sauron early - type galaxies obtained with _ galex _ , along with ground - based optical imaging from the mdm observatory . in [ sec : obs ] , we present a brief summary of the sauron survey and our optical observations , and then describe at length the _ galex _ observations and surface photometry . in [ sec : discussion ] , we discuss the uv colour - magnitude relations , the first uv fundamental planes , and make an attempt at interpreting radial uv@xmath3optical colour profiles .
we summarize our findings in [ sec : conclusions ] .
the sauron observations are aimed at determining the two - dimensional stellar kinematics ( @xcite , hereafter @xcite ) , stellar linestrengths ( @xcite , hereafter @xcite ) and ionised gas kinematics ( @xcite , hereafter @xcite ) of @xmath1 nearby early - type galaxies and @xmath12 spiral bulges in the field and clusters ( see @xcite ) .
the observations and data reduction are described in the relevant papers .
detailed dynamical modeling is also available ( see , e.g. , @xcite ; @xcite , hereafter @xcite ; @xcite , hereafter @xcite ) .
+ + + + + + + ground - based optical imaging observations in the _ hubble space telescope _ ( _ hst _ ) filter f555w ( similar to johnson @xmath13 ) were obtained using the mdm observatory 1.3-m mcgraw - hill telescope , as part of a large survey targeting the whole sauron galaxy sample . the mdm observations are described in detail in falcn - barroso et al .
( in prep . ) and were reduced and calibrated in the standard manner
. the field - of - view of the mdm images is @xmath14 with @xmath15 pixels , allowing for accurate sky subtraction and proper sampling of the seeing .
the seeing during the observations was typically @xmath16 arcsec .
the total @xmath13 magnitudes we use in this paper are also from falcn - barroso et al .
( in prep . ) .
we obtained both fuv ( @xmath17@xmath18 ) and nuv ( @xmath18@xmath19 ) images of @xmath0 e / s0 galaxies using the medium - depth imaging mode of _ galex _ , as part of an ongoing uv imaging survey of the sauron sample ( _ galex _ guest investigator programmes gi1109 and gi3041 ) and the _ galex _ nearby galaxy survey ( ngs ; * ? ? ?
of the @xmath1 early - type galaxies in the sauron sample , @xmath20 are too close to uv - bright stars to be observed with _ galex _ and @xmath21 from our programmes have not been observed yet , similarly for @xmath22 galaxies from other guest investigator programmes , explaining the current sample of @xmath0 .
the typical exposure time per field was one orbit ( @xmath23 s ) but occasionally more .
details of the _ galex _ instruments , pipeline and calibration are described in @xcite and @xcite .
we note in particular a possible systematic error in the fuv and nuv zero - points of up to @xmath24 mag .
this is not included in the uncertainties quoted in the current paper as we are mostly interested in the relative colours of our objects .
the spatial resolutions of the images are approximately @xmath25 and @xmath26 fwhm in fuv and nuv , respectively , sampled with @xmath27 pixels .
[ cols= " < , > , > , > , > , > , > , > , > , > , < " , ] for the first time , we present in figures [ fig : fuvfp ] and [ fig : nuvfp ] the fundamental planes in the fuv and nuv bands , using again the same symbols as in figure [ fig : cmrs ] .
table [ tab : fpresults ] lists the fp coefficients and scatter at both fuv and nuv , obtained from a least - square fit by minimizing the variations to the planes .
fits including and excluding rsf galaxies are listed . as mentioned above ,
rsf galaxies tend to have lower velocity dispersions and smaller effective radii and thus smaller masses than the bulk of the sample galaxies .
crucially , rsf galaxies systematically deviate from the best - fit planes so as to create shallower slopes , and they significantly increase the scatters .
this is clearly shown by both figures [ fig : fuvfp ] and [ fig : nuvfp ] and the fp parameters listed in table [ tab : fpresults ] .
although our sample is limited , this indicates that lower mass galaxies , forming a larger amount of young stars , more greatly ( and systematically ) deviate from the plane defined by the quiescent galaxies .
the long - wavelength @xmath28-band slope is less sensitive to contamination from young stellar populations than optical bands ( @xmath29 ; see @xcite ) .
considering our nuv fundamental plane fit for the entire sample , our slope is significantly different from both the virial expectation and the @xmath28-band value .
however , comparing our nuv - derived slope of @xmath30 for quiescent galaxies only , we recover all of the @xmath28-band slope and almost all of the tilt away from the virial expectation .
we thus conclude that a significant fraction of the fp tilt and scatter is due to low - mass early - type galaxies with stellar populations significantly younger that those of high - mass galaxies , thus reducing their mass - to - light ratios .
the fuv fundamental plane is more complicated to interpret because of the additional presence of old but hot stars ( uv upturn phenomenon ) , but a similar if weaker trend is observed .
the tilt of the fundamental plane was also considered in @xcite using @xmath13-band photometry and sophisticated dynamical modeling .
the main result to emerge was a clear , continuous and monotonic increase of the total dynamical mass - to - light ratio with velocity dispersion , consistent with the fp tilt .
as can be seen here in figure [ fig : cmrs2 ] , the nuv blue galaxies affecting the tilt of the uv fundamental planes also have some of the smallest velocity dispersions and lowest virial masses . furthermore ,
although there is non - negligible scatter , clear ( nuv@xmath8)-@xmath31 and ( fuv@xmath3nuv)-@xmath31 ( or ( nuv@xmath8)-@xmath32 and ( fuv@xmath3nuv)-@xmath32 ) relations are observed . the current results thus seem entirely consistent with that of @xcite .
we are simply picking up the galaxies with the largest and/or most recent star formation events here with our integrated nuv@xmath33 criterion .
many studies have also exploited the projection of the fundamental plane onto the plane defined by @xmath34 and @xmath35 , i.e. the kormendy relation @xcite . a clear advantage of this relation is that it is based on photometric parameters only , which are easily accessible even at high redshifts .
all of our sample galaxies follow the kormendy relation , and we note that our uv kormendy relation slopes ( @xmath36 and @xmath37 at fuv and nuv , respectively , see figures [ fig : fuvfp ] and [ fig : nuvfp ] ) are in rough agreement with the optical one ( @xmath38 in @xcite )
. it may however be significant that , compared to the mean trends , rsf galaxies tend to be slightly offset toward smaller and/or fainter objects .
we use the radial colour profiles to discuss the stellar populations of the sample galaxies .
we discuss our criteria to classify them in groups exhibiting similar properties below , and then discuss in more detail galaxies with signs of recent star formation .
first , our main tool are the nuv@xmath8 radial colour profiles . as for integrated magnitudes ,
we classify a galaxy as having had recent star formation if its nuv@xmath8 profile is below @xmath39 for all or a significant portion of the radii probed .
galaxies included in this category are ngc 474 , ngc 1023 , ngc 2974 , ngc 3032 , ngc 4150 , ngc 4459 , ngc 4526 , ngc 4550 and ngc 7457 .
having said that , nuv@xmath3optical colours normally show negative slopes ( the central regions being redder ) due to variations in the properties of the underlying stellar populations , mainly age and metallicity .
it is well - known that there is a significant age - metallicity degeneracy ( see , e.g. , * ? ? ?
* ) , whereby it is difficult to distinguish the effects of a small change in age from those of a small change in metallicity .
therefore , and even though we recognise the qualitative aspect of such a criterion , if only a small part usually outside the effective radius of the uv@xmath3optical colour profiles goes smoothly and only a little below the nuv@xmath40 cut , we consider it as having been caused by a simple metallicity gradient rather than residual star formation . in other words ,
localised regions with nuv@xmath33 are only classified as recently star - forming if the changes in colour are abrupt .
the galaxies exhibiting this behavior include ngc 821 , ngc 2699 , ngc 4546 , ngc 4570 , ngc 5308 , ngc 5813 , ngc 5831 and ngc 5845 .
their integrated colours are all redder than the nuv@xmath40 colour threshold .
second , we consider the fuv@xmath3nuv profiles . if a galaxy without evidence of recent star formation has a central region with fuv@xmath3nuv@xmath41 , we identify it as a uv - upturn galaxy .
these galaxies include ngc 4374 , ngc 4486 , ngc 4552 , ngc4621 and ngc 5846 . as expected , these galaxies are among the most luminous and most massive in our sample .
all except ngc 4621 are slow - rotators ( see @xcite ) .
significantly , some uv - upturn galaxies have extreme fuv@xmath3nuv and fuv@xmath8 colours at small radii , bluer than the integrated aperture colours of classic uv - upturn galaxies dating from the _ international ultraviolet explorer _ ( _ iue _ ) era .
as these galaxies provided important constraints on stellar evolutionary models of the uv - upturn , it will be interesting to revisit the models with the current data .
it is also probably worthwhile to point out that a number of galaxies exhibit evidence for both a uv - upturn ( fuv@xmath3nuv@xmath41 in the centre ) and a substantial age or metallicity gradient ( nuv@xmath33 in the outer parts ) .
these galaxies are ngc 2695 , ngc 4278 , ngc 4564 , ngc 5198 and ngc 5982 .
they probably harbour milder forms of the uv - upturn phenomenon , where neither the uv - upturn nor large - scale stellar population gradients dominates .
finally , we note that some galaxies exhibit no evidence for recent star formation or a uv - upturn , and no substantial large - scale age and/or metallicity gradient .
these galaxies are ngc 524 , ngc 2768 , ngc 4387 , ngc 4458 , ngc 4473 , ngc 4477 and ngc 5838 .
of the nine galaxies with convincing evidence for recent star formation , five ( ngc 474 , ngc 1023 , ngc 2974 , ngc 4459 and ngc 4526 ) show blue uv@xmath3optical colours in some regions only ( as opposed to all radii ) .
interestingly , this appears connected to kinematic ( sub-)structures detected in sauron and other studies .
the case of ngc 474 , known for its shells ( e.g. * ? ? ? * ) and strong misalignment between the kinematic and photometric major axes ( see @xcite ) , shows blue uv@xmath3optical colours and enhanced h@xmath5 line strength at large radii .
ngc 1023 , with a prominent twist in the centre of the velocity map ( @xcite ) , reveals a uv blob on the eastern side of the galaxy and blue uv@xmath3optical colours in the same region , probably related to an interacting gas - rich dwarf galaxy ( see @xcite for hi observations ) .
the presence of a uv - prominent outer ring related to a bar in ngc 2974 was discussed by @xcite .
blue uv@xmath3optical colours in this galaxy are observed in the central and outer regions . in the case of ngc 4459 ,
blue uv@xmath3optical colours are observed only in the centre .
@xcite shows that this galaxy has strong h@xmath5 absorption within a central dust ring associated with a decoupled stellar and ionised gas disc ( @xcite and @xcite , respectively ) .
ngc 4526 is a similar object appearing more edge - on .
both ngc 4459 and ngc 4526 also harbor central molecular gas discs @xcite . on the other hand , of the nine galaxies with recent star formation , ngc 3032 , ngc 4150 , ngc 4550 and ngc 7457 show _ overall _ blue uv@xmath3optical colours .
ngc 3032 shows the bluest uv@xmath3optical colour in our sample and has a prominent peak in the h@xmath5 map suggesting recent and ongoing star formation .
@xcite identified two types of kinematically - decoupled components ( kdcs ) among the sauron sample : compact kdcs , which are often young and occur in fast - rotating galaxies , and kiloparsec - scale kdcs , with homogeneously old populations in slow - rotating galaxies ( see also @xcite ) .
ngc 3032 , ngc 4150 and ngc 7457 all have small - scale kdcs ( @xcite ) and are considered fast rotators ( @xcite ) .
ngc 3032 and ngc 4150 further have substantial molecular gas discs @xcite . in the case of ngc 4550 , with two counter - rotating stellar discs ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ) , the h@xmath5 line strength is elevated in the central region along the major axis ( @xcite ) . its molecular gas content and detailed star formation history
are discussed in @xcite .
ngc 7457 has a controversial co detection discussed previously ( see @xcite and @xcite ) .
we have presented _ galex _ fuv and nuv imaging along with ground - based f555w imaging from the mdm observatory for @xmath0 early - type galaxies from the sauron survey sample .
nine of them show extended blue uv@xmath3optical colours , hinting at recent star formation .
five of these are also classified as rsf early - types by our integrated colour - magnitude relation criterion . supporting the findings from the uv colour - magnitude relation technique ,
four out of the five candidate rsf galaxies also show enhanced h@xmath5 absorption line strengths , a widely - used post - starburst signature .
the only exception ( ngc 474 ) is noted for its outer shells .
roughly @xmath42 per cent of the early - type galaxies in our sample are therefore classified as rsf early types . considering that the uv flux from a starburst is only detectable for roughly @xmath10 gyr
, this implies that residual star formation has been common in early - type galaxies even up to the present day .
this star formation interpretation is also consistent with the results of molecular gas surveys , as the co detection rate is roughly @xmath43 per cent in rsf early - type galaxies , much higher than in quiescent systems .
the velocity dispersions of these galaxies with evidence of recent star formation are among the lowest in the sample , and none has @xmath44 km s@xmath7 , consistent with the independent work of @xcite .
recent star formation early types also tend to have smaller effective radii and thus smaller ( virial ) masses . despite the limited number statistics ,
a key result from the present study is that rsf early types change the slopes of scaling relations ( colour - magnitude relations and fundamental planes ) and dominate the scatters in them .
most notably , much of the fp tilt and scatter can now be explained by the fact that the properties of a substantial fraction of early - type galaxies are influenced by rsf , systematically biasing their mass - to - light ratios , especially at low masses .
the uv fundamental planes become significantly closer to the virial expectation ( and tighter ) when only quiescent early - type galaxies are considered .
the same must be true at optical wavelengths , although the effect will be smaller .
our result appears consistent with the m / l-@xmath31 relation derived in @xcite , although the current colour threshold only picks up the most extreme objects .
the radial uv@xmath3optical colour profiles not only reveal recent star formation galaxies , but also a number of galaxies with smooth large - scale age and/or metallicity gradients .
similarly , a number of galaxies exhibiting the uv - upturn phenomenon are identified .
some have bluer fuv@xmath3nuv and fuv@xmath8 colours at small radii than the integrated aperture colours of classic uv - upturn galaxies .
a number of galaxies show both a central uv - upturn and large - scale gradients .
this diversity of behaviours will , in due time , need to be explained and reproduced by stellar population models .
early - type galaxies are no longer thought to be simple .
many of them have kinematic anomalies and sub - structures that are non - trivial to interpret .
we have shown here that they are also composed of multiple generations of stars widely separated in age .
this is consistently found by other short - wavelength as well as far - infrared studies , mainly aided by space experiments .
the presence of young stars in seemingly old populations is no longer debatable , yet it is still unclear what causes and regulates this residual star formation .
we believe that our database , spatially resolving nearby early - type galaxies , will be key to answering some of those questions .
the authors thank the anonymous referee for useful comments which led to improvements in the paper . this work was supported by the korea research foundation grant funded by the korean government ( krf - c00156 ) to sky .
mb acknowledges support from nasa through galex guest investigator program galexgi04 - 0000 - 0109 .
mb and sky are grateful to the royal society for an international joint project award ( 2007/r2 ) supporting this work .
rld acknowledges support from the royal society in the form of a wolfson merit award .
mb and rld are also grateful for postdoctoral support through stfc rolling grant pp / e001114/1 .
the stfc visitors grant to oxford also supported joint visits .
gvdv acknowledges support provided by nasa through hubble fellowship grant hst - hf-01202.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 .
mc acknowledges support from a stfc advanced fellowship ( pp / d005574/1 ) .
based on observations made with the nasa galaxy evolution explorer .
galex is operated for nasa by the california institute of technology under nasa contract nas5 - 98034 .
photometric data were also obtained using the 1.3 m mcgraw - hill telescope of the mdm observatory .
part of this work is based on data obtained from the eso / st - ecf science archive facility .
this project made use of the hyperleda database ( http://leda.univ-lyon1.fr ) and the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
bender r. , burstein d. , faber s. m. , 1992 , apj , 399 , 462 bernardi m. et al .
, 2003 , aj , 125 , 1866 bolton a. s. , treu t. , koopmans l. v. e. , gavazzi r. , moustakas l.a . , burles s. , schlegel d. j. , wayth r. , 2008 , apj , 684 , 248 bower r. g. , lucey j. r. , ellis r. , 1992 , mnras , 254 , 589 burstein d. , bertola f. , buson l. m. , faber s. m. , lauer t. r. , 1988 , apj , 328 , 440 cairs l. m. , caon n. , vlchez j. m. , gonzlez - prez j. n. , muo - tun c. , 2001 , apjs , 136 , 393 cardelli j. a. , clayton g. c. , mathis j. s. , 1989 , apj , 345 , 245 cappellari m. et al .
, 2006 , mnras , 366 , 1126 [ paper iv ] cappellari m. et al . , 2007 , mnras , 379 , 418 [ paper x ] ciotti l. , bertin g. , 1999 , a&a , 352 , 447 combes f. , young l. m. , bureau m. , 2007 , mnras , 377 , 1795 crocker a. f. et al . , 2009 , mnras , 393 , 1255 de carvalho r. r. , djorgovski s. g. , 1992 , apjl , 389 , l49 djorgovski s. , davis m. , 1987 , apj , 313 , 59 dressler a. , lynden - bell d. , burstein d. , davies r. l. , faber s. m. , terlevich r. j. , vegner g. , 1987 , apj , 313 , 42 de zeeuw p. t. et al . , 2002 , mnras , 329 , 513 [ paper ii ] de zeeuw p. t. , franx m. , 1991 , ara&a , 29 , 239 donas j. et al . , 2007 , apjs , 173 , 597 donofrio m. , valentinuzzi t. , secco l. , caimmi r. , bindoni d. , 2006 , newa rev .
, 50 , 447 eggen o. j. , lynden - bell d. , sandage a. r. , 1962 , apj , 136 , 748 ellis r. s. et al .
, 1997 , apj , 483 , 582 emsellem e. et al . , 2004 ,
mnras , 352 , 721 [ paper iii ] emsellem e. et al . , 2007 ,
mnras , 379 , 401 [ paper ix ] faber s. m. , jackson r. e. , 1976 , apj , 204 , 668 faber s. m. , wegner g. , burstein d. , davies r. l. , dressler a. , lynden - bell d. , terlevich r. j. , 1989 , apjs , 69 , 763 faber s. m. et al . , 1997 ,
aj , 114 , 1771 gil de paz a. et al . , 2007 , apjs , 173 , 185 guzmn r. , lucey j. r. , bower r. g. , 1993 , mnras , 265 , 731 hau g. k. t. , carter d. , balcells m. , 1999 , mnras , 306 , 437 hudson m. j. , lucey j. r. , smith r. j. , steel j. , 1997 , mnras , 291 , 488 jedrzejewski r. i. , 1987 , mnras , 226 , 747 jeong h. , bureau m. , yi s. k. , krajnovi d. , davies r. l. , 2007 , mnras , 376 , 1021 jrgensen i. , franx m. , kjrgaard p. , 1996 ,
mnras , 280 , 167 kaviraj s. et al . , 2007 , apjs , 173 , 619 kelson d. d. , illingworth g. d. , van kokkum p. g. , franx m. , 2000 , apj , 531 , 184 khochfar s. , burtert a. , 2005 , mnras , 359 , 1379 kormendy j. , 1977 , apj , 217 , 406 krajnovi d. et al .
2008 , mnras , 390 , 93 hamabe m. , kormendy j. , 1987 , in iau symposium 127 , structure and dynamics of elliptical galaxies , ed .
t. de zeeuw ( dordrecht : reidel ) , 379 kormendy j. , bender r. , 1996 , apj , 464 , l119 kuntschner h. et al . , 2006 , mnras , 369 , 497 [ paper vi ] larson r. b. , 1974 , mnras , 166 , 385 martin d. c. et al . , 2005 ,
apj , 619 , l1 mcdermid r. m. et al . , 2006 , mnras , 373 , 906 [ paper viii ] morganti r. et al . , 2006 , mnras , 371 , 157 morrissey p. et al .
, 2005 , apj , 619 , l7 morrissey p. et al . , 2007 , apjs , 173 , 682 nolan l. a. , dunlop j. s. , panter b. , jimenez r. , heavens a. , smith g. , 2007 , mnras , 375 , 371 oconnell r. w. , 1999 , ara&a , 37 , 603 pahre m. a. , djorgovski s. g. , de carvalho r. r. , 1998 , aj , 116 , 1591 pahre m. a. , de carvalho r. r. , djorgovski s. g. , 1998 , aj , 116 , 1606 prugniel ph . , simien f. , 1996 , a&a , 309 , 749 raimann d. , storchi - bergmann t. , quintana h. , hunstead r. , wisotzki l. , 2005 , mnras , 364 , 1239 rix h .- w . , franx m , fisher d. , illingworth g. , 1992 , apj , 400 , l5 rubin v. c. , graham j. a. , kenney j. d. p. , 1992 ,
apj , 394 , l9 sarzi m. et al . , 2006 , mnras , 366 , 1151 [ paper v ] schawinski k. et al . , 2006 , nature , 442 , 888 schawinski k. et al . , 2007a , apjs , 173 , 512 schlegel d. j. , finkbeiner d. p. , davis m. , 1998 , apj , 500 , 525 schweizer f. , seitzer p. , 1992
, aj , 104 , 1039 scodeggio m. , 1997 , ph.d .
thesis , cornell univ .
stanford s. a. , eisenhardt p. r. m. , dickinson m. , 1998 , apj , 492 , 461 toomre a. , toomre j. , 1972 , apj , 178 , 623 treu t. , koopmans l. v. e. , bolton a. s. , burles s. , moustakas l. a. , 2006 , apj , 640 , 662 trujillo i. , burkert a. , bell e. , 2004 , apj 600 , 39 turnbull a. j. , bridges t. j. , carter d. , 1999 , mnras , 307 , 967 van den bosch f. c. , emsellem e. , 1998 , mnras , 298 , 267 van dokkum p. g. , franx m. , fabricant d. , illingworth g. d. , kelson d. d. , 2000 , apj , 541 , 95 welch g. a. , sage l. j. , 2003 , apj , 584 , 260 wernli f. , emsellem e. , copin y. , 2002 , a&a , 396 , 73 worthey g. , 1994 , apjs , 95 , 107 wyder t. k. et al . , 2005 , apjl , 619 , l15 yi s. , demarque p. , oemler a. j. , 1997 , apj , 486 , 201 yi s. k. et al .
, 2005 , apjl , 619 , l111 young l. m. , bureau m. , cappellari m. , 2008 , apj , 676 , 317 young l. m. , bendo g. j. , lucero d. m. , 2009 , aj , 137 , 3053 zepf s. e. , whitmore b. c. , 1993 , apj , 418 , 72
we briefly comment here on the _ galex _ and mdm surface brightness profiles of the e / s0 galaxies presented in this paper .
individual comments on the stellar kinematics , the emission - line maps and the line strength structures are given in @xcite , @xcite and @xcite , respectively .
this galaxy ( arp 227 ) , noted for its shell structures ( e.g. * ? ? ? * ) , shows blue nuv@xmath8 colours except in the central regions . the h@xmath5 absorption line strength rises toward larger radii ( ( see * ? ? ?
considering the uv@xmath8 colours , it seems that young stellar populations are present in the outer parts .
this sb0 galaxy shows a tilt in the fuv and nuv surface brightness profiles around @xmath45 arcsec , a central concentration in all metal lines ( ( see * ? ? ?
* ) ) and significantly negative values of both @xmath46 and @xmath47 ( ( see * ? ? ?
we discovered a strong nuv blob , constituting strong evidence for recent / ongoing star formation , just outside the galaxy .
this is likely the result of a tidal interaction between ngc 1023 and a neighboring gas - rich galaxy ( see * ? ? ?
this galaxy shows blue fuv@xmath3nuv colours in the central regions with a high amplitude in @xmath46 .
the uv@xmath8 colours become bluer toward larger radii , albeit with large uncertainties .
we thus consider this a red galaxy .
this galaxy with a uv outer ring shows a blue colour in the outer regions , suggesting young stellar populations associated with a large - scale bar ( see * ? ? ?
* ) , although the h@xmath5 line strength map appears relatively flat ( ( see * ? ? ?
blue uv@xmath8 colours are observed everywhere , suggesting a strong ongoing starburst .
this galaxy also shows strong h@xmath5 absorption with negative mg@xmath48 and fe5270@xmath49 line strength gradients ( ( see * ? ? ?
* ) ) .
this galaxy with a prominent jet shows blue uv@xmath8 colours in the central regions , almost certainly related to nuclear activity .
the jet also induces large errors in the surface brightness profiles .
the uv@xmath8 plots of this galaxy exhibit blue colours in the centre , coincident with strong h@xmath5 absorption ( ( see * ? ? ?
* ) ) and a stellar velocity dispersion drop due to a star - forming fast - rotating disc ( ( see * ? ? ?
* ) and @xcite ) .
this object has a disc - like component similar to ngc 4564 ( ( see * ? ? ? * ) ) .
a counter - rotating component is also detected inside @xmath52 arcsec @xcite .
blue fuv@xmath3nuv colours are observed in the centre .
like ngc 5813 , this galaxy with a kinematically - decoupled core shows blue nuv@xmath8 colours in the outer regions .
there is a small bump in both fuv and nuv surface brightness profiles around @xmath53 arcsec . | we present _ galaxy evolution explorer _ ( _ galex _ ) far ( fuv ) and near ( nuv ) ultraviolet imaging of @xmath0 nearby early - type galaxies from the sauron representative sample of @xmath1 e / s0 galaxies , all of which have ground - based optical imaging from the mdm observatory .
the surface brightness profiles of nine galaxies ( @xmath2 per cent ) show regions with blue uv@xmath3optical colours suggesting recent star formation .
five of these ( @xmath4 per cent ) show blue integrated uv@xmath3optical colours that set them aside in the nuv integrated colour - magnitude relation .
these are objects with either exceptionally intense and localised nuv fluxes or blue uv@xmath3optical colours throughout .
they also have other properties confirming they have had recent star formation , in particular h@xmath5 absorption higher than expected for a quiescent population and a higher co detection rate .
this suggests that residual star formation is more common in early - type galaxies than we are used to believe .
nuv - blue galaxies are generally drawn from the lower stellar velocity dispersion ( @xmath6 km s@xmath7 ) and thus lower dynamical mass part of the sample .
we have also constructed the first uv fundamental planes and show that nuv blue galaxies bias the slopes and increase the scatters .
if they are eliminated the fits get closer to expectations from the virial theorem .
although our analysis is based on a limited sample , it seems that a dominant fraction of the tilt and scatter of the uv fundamental planes is due to the presence of young stars in preferentially low - mass early - type galaxies .
interestingly , the uv@xmath3optical radial colour profiles reveal a variety of behaviours , with many galaxies showing signs of recent star formation , a central uv - upturn phenomenon , smooth but large - scale age and metallicity gradients , and in many cases a combination of these .
in addition , fuv@xmath3nuv and fuv@xmath8 colours even bluer than those normally associated with uv - upturn galaxies are observed at the centre of some quiescent galaxies .
four out of the five uv - upturn galaxies are slow rotators .
these objects should thus pose interesting challenges to stellar evolutionary models of the uv - upturn .
galaxies : elliptical and lenticular , cd galaxies : evolution galaxies : fundamental parameters galaxies : photometry galaxies : structure ultraviolet : galaxies . |
a companion article @xcite in this volume investigates curves of finite total curvature .
this class includes both smooth and polygonal curves , and allows a unified treatment of curvature .
here we briefly review the theory of smooth curves from the point of view we will later adopt for surfaces .
the @xmath1 ( which we usually assume is parametrized by its arclength @xmath2 ) are the local properties of its shape , invariant under euclidean motions .
the only first - order information is the tangent line ; since all lines in space are equivalent , there are no first - order invariants . second - order information ( again , independent of parametrization )
is given by the osculating circle ; the one corresponding invariant is its curvature @xmath3 .
( for a plane curve given as a graph @xmath4 let us contrast the notions of curvature @xmath5 and second derivative @xmath6 . at a point @xmath7 on the curve , we can find either one by translating @xmath7 to the origin , transforming so the curve is horizontal there , and then looking at the second - order behavior .
the difference is that for curvature , the transformation is a euclidean rotation , while for second derivative , it is a shear @xmath8 .
a parabola has constant second derivative @xmath6 because it looks the same at any two points after a shear .
a circle , on the other hand , has constant curvature because it looks the same at any two points after a rotation . )
a plane curve is completely determined ( up to rigid motion ) by its ( signed ) curvature @xmath9 as a function of arclength @xmath2 . for a space curve , however , we need to look at the third - order invariants ; these are the torsion @xmath10 and the derivative @xmath11 , but the latter of course gives no new information .
curvature and torsion now form a complete set of invariants : a space curve is determined by @xmath9 and @xmath12 .
generically speaking , while second - order curvatures usually suffice to determine a hypersurface ( of codimension @xmath13 ) , higher - order invariants are needed for higher codimension . for curves in @xmath14 , for instance , we need @xmath15 generalized curvatures , of order up to @xmath16 , to characterize the shape .
let us examine the case of space curves @xmath17 in more detail .
at every point @xmath18 we have a splitting of the tangent space @xmath19 into the tangent line @xmath20 and the normal plane .
a framing along @xmath1 is a smooth choice of a unit normal vector @xmath21 , which is then completed to the oriented orthonormal frame @xmath22 for @xmath0 , where @xmath23 . taking the derivative with respect to arclength
, we get a skew - symmetric matrix ( an infinitesimal rotation ) that describes how the frame changes : @xmath24 here , @xmath25 is the _ curvature vector _ of @xmath1 , while @xmath10 measures the twisting of the chosen orthonormal frame .
if we fix @xmath21 at some basepoint along @xmath1 , then one natural framing is the or _ bishop frame _ @xcite defined by the condition @xmath26 .
equivalently , the vectors @xmath27 are parallel - transported along @xmath1 from the basepoint , using the riemannian connection on the normal bundle induced by the immersion in @xmath0 .
one should note that this is not necessarily a closed framing along a closed loop @xmath1 ; when we return to the basepoint , the vector @xmath21 has been rotated through an angle called the of @xmath1 .
other framings are also often useful .
for instance , if @xmath1 lies on a surface @xmath28 with unit normal @xmath29 , it is natural to choose @xmath30
. then @xmath31 is called the _ cornormal _ vector , and
@xmath32 is the ( adapted to @xmath33 ) .
the curvature vector of @xmath1 decomposes into parts tangent and normal to @xmath28 as @xmath34 . here
, @xmath35 measures the _ normal curvature _ of @xmath28 in the direction @xmath36 , and is independent of @xmath1 , while @xmath37 , the _ geodesic curvature _ of @xmath1 in @xmath28 , is an intrinsic notion , showing how @xmath1 sits in @xmath28 , and is unchanged if we isometrically deform the immersion of @xmath28 into space . when the curvature vector of @xmath1 never vanishes , we can write it as @xmath38 , where @xmath39 is a unit vector , the _ principal normal _ , and @xmath40
. this yields the orthonormal @xmath41 , whose twisting @xmath10 is the _ torsion _ of @xmath1 .
the _ total curvature _ of a smooth curve is @xmath42 . in
@xcite we review a number of standard results : for closed curves , the total curvature is at least @xmath43 ( fenchel ) and for knotted space curves the total curvature is at least @xmath44 ( fry / milnor ) . for plane curves , we can consider instead the signed curvature , and find that @xmath42 is always an integral multiple of @xmath43 .
suppose ( following milnor ) we define the total curvature of a polygonal curve simply to be the sum of the turning angles at the vertices .
then , as we explain in @xcite , all these theorems on total curvature remain true .
our goal , when defining curvatures for polyhedral surfaces , will be to ensure that similar integral relations remain true .
given a ( two - dimensional , oriented ) surface @xmath28 smoothly immersed in @xmath0 , we understand its local shape by looking at the @xmath45 given by the unit normal vector @xmath46 at each point @xmath47 .
the derivative of the / map at @xmath7 is a linear map from @xmath48 to @xmath49 .
since these spaces are naturally identified , being parallel planes in @xmath0 , we can view the derivative as an endomorphism @xmath50 .
the map @xmath51 is called the ( or weingarten map ) .
the shape operator is the complete second - order invariant ( or curvature ) which determines the original surface @xmath28 .
( this statement has been left intentionally a bit vague , since without a standard parametrization like arclength , it is not quite clear how one should specify such an operator along an unknown surface . ) usually , however , it is more convenient to work not with the operator @xmath51 but instead with scalar quantities .
its eigenvalues @xmath52 and @xmath53 are called the _ principal curvatures _ , and ( since they can not be globally distinguished ) it is their symmetric functions which have the most geometric meaning .
we define the @xmath54 as the determinant of @xmath51 and the _ mean curvature _ @xmath55 as its trace . note that the sign of @xmath56 depends on the choice of unit normal @xmath29 ,
so often it is more natural to work with the _ vector mean curvature _ ( or mean curvature vector ) @xmath57 .
furthermore , some authors use the opposite sign on @xmath51 and thus @xmath58 , and many use @xmath59 , justifying the name `` mean '' curvature .
our conventions mean that the mean curvature vector for a convex surface points inwards ( like the curvature vector for a circle ) . for a unit sphere oriented with inward normal ,
the / map @xmath29 is the antipodal map , @xmath60 , and @xmath61 .
the / curvature is an intrinsic notion , depending only on the pullback metric on the surface @xmath28 , and not on the immersion into space .
that is , @xmath62 is unchanged by bending the surface without stretching it .
for instance , a developable surface like a cylinder or cone has @xmath63 because it is obtained by bending a flat plane .
one intrinsic characterization of @xmath64 is obtained by comparing the circumferences @xmath65 of ( intrinsic ) @xmath66-balls around @xmath7 to the value @xmath67 in @xmath68 .
we get @xmath69 mean curvature , on the other hand , is certainly not intrinsic , but it has a nice variational interpretation . consider a variation vectorfield @xmath70 on @xmath28 ; for simplicity assume @xmath70 is compactly supported away from any boundary .
then @xmath71 in the sense that @xmath72 with respect to the @xmath73 inner product @xmath74 on vectorfields , the vector mean curvature is thus the negative gradient of the area functional , often called the first variation of area : @xmath75 .
( similarly , the negative gradient of length for a curve is its curvature vector @xmath76 . ) just as @xmath5 is the geometric version of second derivative for curves , mean curvature is the geometric version of the laplacian @xmath77 .
indeed , if a surface @xmath28 is written locally near a point @xmath7 as the graph of a height function @xmath78 over its tangent plane @xmath79 , then @xmath80 . alternatively , we can write @xmath81 , where @xmath82 is the position vector in @xmath0 and @xmath83 is the laplace / beltrami operator , the intrinsic surface laplacian . we can flow a curve or surface to reduce its length or area , by following the gradient vectorfield @xmath76 or @xmath84 ; the resulting parabolic ( heat ) flow is slightly nonlinear in a natural geometric way .
this so - called _ mean - curvature flow _ has been extensively studied as a geometric smoothing flow .
( see , among many others , @xcite for the curve - shortening flow and @xcite for higher dimensions . )
for surfaces , we will consider various integral curvature relations that relate area integrals over a region @xmath85 to arclength integrals over the boundary @xmath86 .
first , the //bonnet theorem says that , when @xmath87 is a disk , @xmath88 here , @xmath89 is the vector line element along @xmath1 , and @xmath90 is again the cornormal . in particular , this theorem implies that the total / curvature of @xmath87 depends only on a collar neighborhood of @xmath1 : if we make any modification to @xmath87 supported away from the boundary , the total curvature is unchanged ( as long as @xmath87 remains topologically a disk ) .
we will extend the notion of ( total ) / curvature from smooth surfaces to more general surfaces ( in particular polyhedral surfaces ) by requiring that this property remain true .
our other integral relations are all proved by stokes s theorem , and thus require only that @xmath1 be the boundary of @xmath87 in a homological sense ; for these @xmath87 need not be a disk .
first consider the @xmath91 the right - hand side represents the total vector area of any surface spanning @xmath1 , and the relation shows this to depend only on @xmath1 ( and this time not even on a collar neighborhood ) .
the integrand on the left - hand side depends on a choice of origin for the coordinates , but because we integrate over a closed loop , the integral is independent of this choice .
both sides of this vector area formula can be interpreted directly for a polyhedral surface , and the equation remains true in that case .
we note also that this vector area @xmath92 is one of the quantities preserved when @xmath1 evolves under the hasimoto or smoke - ring flow @xmath93 .
( compare @xcite . ) a simple application of the fundamental theorem of calculus to the tangent vector of a curve @xmath1 from @xmath7 to @xmath94 shows that @xmath95 this can be viewed as a balance between elastic tension forces trying to shrink the curve , and sideways forces holding it in place .
it is the key step in verifying that the vector curvature @xmath76 is the first variation of length .
the analog for a surface patch @xmath87 is the mean - curvature equation @xmath96 again this represents a balance between surface tension forces acting in the conormal direction along the boundary of @xmath87 and what can be considered as pressure forces ( espcially in the case of constant @xmath56 ) acting normally across @xmath87 .
we will use this equation to develop our analog of mean curvature for discrete surfaces .
the force balance equation can be seen to arise from the translational invariance of curvatures .
it has been important for studying surfaces of constant mean curvature ; see for instance @xcite .
the rotational analog is the following : @xmath97 somewhat related is the following equation : @xmath98 it gives , for example , an interesting expression for the area of a minimal ( @xmath99 ) surface .
for us , a discrete or polyhedral surface @xmath100 will mean a triangulated surface with a continuous map into space , linear on each triangle . in more detail , we start with an abstract combinatorial a simplicial complex representing a @xmath101-manifold with boundary .
we then pick positions @xmath102 for all the vertices , which uniquely determine a linear map on each triangle ; these maps fit together to form the polyhedral surface .
the union of all triangles containing a vertex @xmath7 is called @xmath103 , the _ star _ of @xmath7 . similarly , the union of the two triangles containing an edge @xmath104 is @xmath105 .
it is well known how the notion of / curvature extends to such discrete surfaces @xmath28 .
( banchoff @xcite was probably the first to discuss this in detail , though he notes that hilbert and cohn - vossen @xcite had already used a polyhedral analog to motivate the intrinsic nature of / curvature . )
any two adjacent triangles ( or , more generally , any simply connected region in @xmath28 not including any vertices ) can be flattened
developed isometrically into the plane .
thus the / curvature is supported on the vertices @xmath47 .
in fact , to keep the //bonnet theorem true , we must take @xmath106 here , the angles @xmath107 are the interior angles at @xmath7 of the triangles meeting there , and @xmath108 is often known as the angle defect at @xmath7 .
if @xmath87 is any neighborhood of @xmath7 contained in @xmath103 , then @xmath109 ; when the triangles are acute , this is most easily seen by letting @xmath110 be the path connecting their circumcenters and crossing each edge perpendicularly .
analogous to our intrinsic characterization of / curvature in the smooth case , note that the circumference of a small @xmath66 ball around @xmath7 here is exactly @xmath111 .
this version of discrete / curvature is quite natural , and seems to be the correct analog when / curvature is used intrinsically .
but the / curvature of a smooth surface in @xmath0 also has extrinsic meaning ; for instance the total absolute / curvature is proportional to the average number of critical points of different height functions . for such considerations , brehm and khnel @xcite suggest the following :
when a vertex @xmath7 is extreme on the convex hull of its star , but the star itself is not a convex cone , then we should think of @xmath7 as having both positive and negative curvature .
we let @xmath112 be the curvature at @xmath7 of the convex hull , and set @xmath113 .
then the absolute curvature at @xmath7 is @xmath114 , which is greater than @xmath115 .
this discretization , is of course , also based on preserving an integral curvature relation a different one .
( see also @xcite . )
we can use the same principle as before preserving the //bonnet theorem to define / curvature , as a measure , for much more general surfaces .
for instance , on a piecewise smooth surface , we have ordinary @xmath62 within each face , a point mass ( again the angle defect ) at each vertex , and a linear density along each edge , equal to the difference in the geodesic curvatures of that edge within its two incident faces .
indeed , clothes are often designed from pieces of ( intrinsically flat ) cloth , joined so that each vertex is intrinsically flat and thus all the curvature is along the edges ; corners would be unsightly in clothes .
returning to polyhedral surfaces , we note that @xmath108 is clearly an intrinsic notion ( as it should be ) depending only on the angles of each triangle and not on the precise embedding into @xmath0 .
sometimes it is useful to have a notion of combinatorial curvature , independent of _ all _ geometric information .
given just a combinatorial triangulation , we can pretend that each triangle is equilateral with angles @xmath116 .
( such a euclidean metric with cone points at certain vertices exists on the abstract surface , independent of whether or not it could be embedded in space .
see the survey @xcite . )
the curvature of this metric , @xmath117 , can be called the _ combinatorial ( / ) curvature _ of the triangulation . ( see @xcite for combinatorial applications of this notion . ) in this context , the global form @xmath118 of //bonnet amounts to nothing more than euler s formula @xmath119 .
( we note that forman has proposed a combinatorial ricci curvature @xcite ; although for smooth surfaces , ricci curvature is / curvature , for discrete surfaces forman s combinatorial curvature does not agree with ours , so he fails to recover the //bonnet theorem . )
our discrete / curvature @xmath108 is of course an integrated quantity .
sometimes it is desirable to have instead a curvature density , dividing @xmath108 by the surface area associated to the vertex @xmath7 .
one natural choice is @xmath120 , but this does not always behave nicely for irregular triangulations .
one problem is that , while @xmath108 is intrinsic , depending only on the cone metric of the surface , @xmath121 depends also on the choice of which pairs of cone points are connected by triangle edges .
one fully intrinsic notion of the area associated to @xmath7 would be the area of its intrinsic voronoi cell in the sense of bobenko and springborn @xcite ; perhaps this would be the best choice for computing / curvature density .
the formula @xmath122 needs no special interpretation for discrete surfaces : both sides of the equation make sense directly , since the surface normal @xmath29 is well - defined almost everywhere .
however , it is worth interpreting this formula for the case when @xmath87 is the star of a vertex @xmath7 .
more generally , suppose @xmath1 is any closed curve ( smooth or polygonal ) , and @xmath87 is the cone from @xmath7 to @xmath1 ( the union of all line segments @xmath123 for @xmath124 ) .
fixing @xmath1 and letting @xmath7 vary , we find that the volume enclosed by this cone is an affine linear function of @xmath102 , and thus @xmath125 is independent of the position of @xmath7 .
we also note that any such cone @xmath87 is intrinsically flat except at the cone point @xmath7 , and that @xmath126 is the cone angle at @xmath7 .
the mean curvature of a discrete surface @xmath28 is supported along the edges .
if @xmath104 is an edge , and @xmath127 , then we set @xmath128 here @xmath129 is the normal vector to the triangle @xmath130 , and the operator @xmath131 rotates by @xmath132 in the plane of that triangle .
note that @xmath133 , where @xmath134 is the exterior dihedral angle along the edge , defined by @xmath135 .
no nonplanar discrete surface has @xmath136 along every edge .
but this discrete mean curvature can cancel out around vertices .
we set @xmath137 the area of the discrete surface is a function of the vertex positions ; if we vary only one vertex @xmath7 , we find that @xmath138 .
this mirrors the variational characterization of mean curvature for smooth surfaces , and we see that a natural notion of discrete minimal surfaces is to require @xmath139 for all vertices @xcite .
suppose that the vertices adjacent to @xmath7 , in cyclic order , are @xmath140 ,
. then we can express @xmath142 and @xmath143 explicitly in terms of these neighbors .
we get @xmath144 and similarly @xmath145 where @xmath146 and @xmath147 are the angles opposite edge @xmath148 in the two incident triangles .
this latter equation is the famous `` '' @xcite which also arises naturally in a finite - element discretization of the laplacian .
suppose we change the combinatorics of a discrete surface @xmath28 by introducing a new vertex @xmath7 along an existing edge @xmath104 and subdividing the two incident triangles .
then @xmath143 in the new surface equals the original @xmath149 , independent of where along @xmath104 we place @xmath7 .
this allows a variational interpretation of @xmath149 . a somewhat different interpretation of mean curvature for convex polyhedra is found in the context of minkowski s theory of mixed volumes .
( in this simple form , it dates back well before minkowski , to steiner @xcite . )
if @xmath150 is a smooth convex body in @xmath0 and @xmath151 denotes its @xmath152-neighborhood , then its steiner polynomial is : @xmath153 here , by //bonnet , the last integral is always @xmath44 .
when @xmath150 is instead a convex polyhedron , we already understand how to interpret each term except @xmath154 .
the correct replacement for this term , as steiner discovered , is @xmath155 .
this suggests @xmath156 as a notion of total mean curvature for the edge @xmath104 .
note the difference between this formula and our earlier @xmath157 .
either one can be derived by replacing the edge @xmath104 with a sector of a cylinder of length @xmath158 and arbitrary ( small ) radius @xmath159 .
we have @xmath160 , so that @xmath161 the difference is explained by the fact that one formula integrates the scalar mean curvature while the other integrates the vector mean curvature .
again , these two discretizations both arise through preservation of ( different ) integral relations for mean curvature .
see @xcite for a more extensive discussion of the analogous situation for curves :
although as we have mentioned , the sum of the turning angles @xmath162 is often the best notion of total curvature for a polygon , in certain situations the `` right '' discretization is instead the sum of @xmath163 or @xmath164 .
the interpretation of curvatures in terms of the mixed volumes or steiner polynomial actually works for arbitrary convex surfaces .
( compare @xcite in this volume . ) using geometric measure theory and a generalized normal bundle called the normal cycle one can extend both / and mean curvature in a similar way to quite general surfaces .
see @xcite . a smooth surface which minimizes area under a volume constraint has constant mean curvature
; the constant @xmath56 can be understood as the lagrange multiplier for the constrained minimization problem .
a discrete surface which minimizes area among surfaces of fixed combinatorial type and fixed volume will have constant discrete mean curvature @xmath56 in the sense that at every vertex , @xmath165 , or equivalently @xmath166 .
in general , of course , the vectors @xmath143 and @xmath142 are not even parallel : they give two competing notions of a normal vector to the discrete surface at the vertex @xmath7 . still , @xmath167 gives a better notion of mean curvature density near @xmath7 than , say , the smaller quantity @xmath168 where @xmath120 .
suppose we want to discretize the elastic bending energy for surfaces , @xmath169 , known as the .
the discussion above shows why @xmath170 ( which was used in @xcite ) is a better discretization than @xmath171 ( used fifteen years ago in @xcite ) .
several related discretizations are by now built into brakke s evolver ; see the discusion in @xcite .
recently , bobenko @xcite has described a completely different approach to discretizing the willmore energy , which respects the mbius invariance of the smooth energy . as mentioned above
, we can define a _ discrete minimal surface _ to be a polyhedral surface with @xmath139 .
an early impetus to the field of discrete differential geometry was the realization ( starting with @xcite ) that discrete minimal surfaces are not only critical points for area ( fixing the combinatorics ) , but also have other properties similar to those of smooth minimal surfaces . for instance , in a conformal parameterization of a smooth minimal surface , the coordinate functions are harmonic . to interpret this for discrete surfaces ,
we are led to the question of when a discrete map should be considered conformal .
in general this is still open .
( interesting suggestions come from the theory of circle packings , and this is an area of active research .
see for instance @xcite . )
however , we should certainly agree that the identity map is conformal .
a polyhedral surface @xmath28 comes with an embedding @xmath172 which we consider as the identity map .
indeed , we then find ( following @xcite ) that @xmath28 is discrete minimal if and only if @xmath173 is . here a polyhedral map @xmath174 is called _ discrete harmonic _ if it is a critical point for the dirichlet energy , written as the following sum over the triangles @xmath36 of @xmath28 : @xmath175 we can view @xmath176 as a measure of nonconformality . for the identity map , @xmath177 and @xmath178 , confirming that @xmath28 is minimal if and only if @xmath173 is harmonic .
we now give a general definition of vector bundles and connections on polyhedral manifolds ; this leads to another interpretation of the / curvature for a polyhedral surface .
a polyhedral @xmath179-manifold @xmath180 means a cw - complex which is homeomorphic to an @xmath179-dimensional manifold , and which is regular and satisfies the intersection condition .
( compare @xcite in this volume . )
that is , each @xmath179-cell ( called a _ facet _ )
is embedded in @xmath180 with no identifications on its boundary , and the intersection of any two cells ( of any dimension ) is a single cell ( if nonempty ) . because @xmath180 is topologically a manifold , each @xmath181cell ( called a _ ridge _ ) is contained in exactly two facets . a _ discrete vector bundle _
@xmath182 of rank @xmath183 over @xmath180 consists of a vectorspace @xmath184 for each facet @xmath78 of @xmath185 .
a _ connection _ on @xmath182 is a choice of isomorphism @xmath186 between @xmath187 and @xmath188 for each ridge @xmath189 of @xmath185 .
we are most interested in the case where the vectorspaces @xmath187 have inner products , and the isomorphisms @xmath186 are orthogonal .
consider first the case @xmath190 , where @xmath185 is a polygonal curve . on an arc ( an open curve )
any vector bundle is trivial . on a loop ( a closed curve ) ,
a vector bundle of rank @xmath183 is determined ( up to isomorphism ) simply by its around the loop , an automorphism @xmath191 .
now suppose @xmath180 is linearly immersed in @xmath14 for some @xmath16 .
that is , each @xmath183-face of @xmath185 is mapped homeomorphically to a convex polytope in an affine @xmath183-plane in @xmath14 , and the star of each vertex is embedded .
then it is clear how to define the discrete tangent bundle @xmath36 ( of rank @xmath179 ) and normal bundle @xmath39 ( of rank @xmath192 ) .
namely , each @xmath193 is the @xmath179-plane parallel to the affine hull of the facet @xmath78 , and @xmath194 is the orthogonal @xmath195plane .
these inherit inner products from the euclidean structure of @xmath14 .
there are also natural analogs of the s on these bundles .
namely , for each ridge @xmath159 , let @xmath196 be the exterior dihedral angle bewteen the facets @xmath197 meeting along @xmath159 .
then let @xmath198 be the simple rotation by this angle , fixing the affine hull of @xmath159 ( and the space orthogonal to the affine hull of the @xmath197 ) .
we see that @xmath186 restricts to give maps @xmath199 and @xmath200 ; these form the connections we want .
( note that @xmath201 is a trivial vector bundle over @xmath180 , but the maps @xmath186 give a nontrivial connection on it . )
consider again the example of a closed polygonal curve @xmath202 .
the tangent bundle has rank @xmath13 and trivial holonomy .
the holonomy of the normal bundle is some rotation of @xmath203 . for @xmath204 this rotation of the plane
@xmath68 is specified by an angle equal ( modulo @xmath43 ) to the of @xmath205 .
( to define the writhe of a curve as a real number , rather than just modulo @xmath43 , requires a bit more care , and requires the curve @xmath185 to be embedded . )
next consider a two - dimensional polyhedral surface @xmath206 and its tangent bundle . around a vertex @xmath7
we can compose the cycle of isomorphisms @xmath207 across the edges incident to @xmath7 .
this gives a self - map @xmath208 .
this is a rotation of the tangent plane by an angle which it is easy to check equals the discrete / curvature @xmath108 .
now consider the general case of the tangent bundle to a polyhedral manifold @xmath209 .
suppose @xmath7 is a codimension - two face of @xmath185 .
then composing the ring of isomorphisms across the ridges incident to @xmath7 gives an automorphism of @xmath210 which is the local holonomy , or _
curvature _ of the levi - civita connection around @xmath7 .
we see that this is a rotation fixing the affine hull of @xmath7 . to define this curvature ( which can be interpreted as a sectional curvature in the two - plane normal to @xmath7 ) as a real number and not just modulo @xmath43
, we should look again at the angle defect around @xmath7 , which is @xmath211 where the @xmath147 are the interior dihedral angles along @xmath7 of the facets @xmath78 incident to @xmath7 . in the case of a hypersurface @xmath212 in @xmath14 ,
the one - dimensional normal bundle is locally trivial : there is no curvature or local holonomy around any @xmath7 .
globally , the normal bundle is of course trivial exactly when @xmath185 is orientable .
thomas f. banchoff and wolfgang khnel , _ tight submanifolds , smooth and polyhedral _ , tight and taut submanifolds ( berkeley , ca , 1994 ) , math .
32 , cambridge univ .
press , cambridge , 1997 , pp .
51118 .
george francis , john m. sullivan , robert b. kusner , kenneth a. brakke , chris hartman , and glenn chappell , _ the minimax sphere eversion _ , visualization and mathematics ( h .- c .
hege and k. polthier , eds . ) , springer , heidelberg , 1997 , pp
. 320 .
karsten groe - brauckmann , robert b. kusner , and john m. sullivan , _ triunduloids : embedded constant mean curvature surfaces with three ends and genus zero _ , j. reine angew . math . * 564 * ( 2003 ) , 3561 ; arxiv : math.dg/0102183 .
mark meyer , mathieu desbrun , peter schrder , and alan h. barr , _ discrete differential - geometry operators for triangulated @xmath101-manifolds _ , visualization and mathematics iii ( h .- c .
hege and k. polthier , eds . ) , springer , berlin , 2003 , pp .
3557 .
ulrich pinkall and boris a. springborn and steffen weissmann , _ a new doubly discrete analogue of smoke ring flow and the real time simulation of fluid flow _
, j. phys .
a , 2007 , to appear ; arxiv.org/0708.0979 .
jakob steiner , _ ber parallele flchen _ , ( monats)bericht akad
berlin ( 1840 ) , 114118 ; ges .
werke ii , 2nd ed .
( ams / chelsea , 1971 ) , 171176 ; bibliothek.bbaw.de/bbaw/bibliothek-digital/digitalequellen/schriften/anzeige/index_html?band=08-verh/1840&seite:int=114 .
kenneth stephenson , _ introduction to circle packing _
, cambridge univ . press , cambridge , 2005 . | we discuss notions of gauss curvature and mean curvature for polyhedral surfaces .
the discretizations are guided by the principle of preserving integral relations for curvatures , like the gauss / bonnet theorem and the mean - curvature force balance equation .
= /gauss # 1#2 # 1figure [ fig:#1 ] the curvatures of a smooth surface are local measures of its shape . here
we consider analogous quantities for discrete surfaces , meaning triangulated polyhedral surfaces .
often the most useful analogs are those which preserve integral relations for curvature , like the //bonnet theorem or the force balance equation for mean curvature . for simplicity ,
we usually restrict our attention to surfaces in euclidean three - space @xmath0 , although some of the results generalize to other ambient manifolds of arbitrary dimension .
this article is intended as background for some of the related contributions to this volume .
much of the material here is not new ; some is even quite old .
although some references are given , no attempt has been made to give a comprehensive bibliography or a full picture of the historical development of the ideas . |
the investigation of heavy - quark ( bottom or charm ) production in @xmath2 collisions is a useful test of perturbative quantum chromodynamics ( qcd ) since the heavy quark mass provides a hard scale that allows to perform calculations within perturbation theory . at leading order ( lo ) , boson - gluon fusion , @xmath3 is the dominant process for bottom - quark production .
when the negative squared four - momentum of the initial and final electron , @xmath4 , is small , the process @xmath5 is treated as photoproduction , in which a quasi - real photon emitted by the initial electron interacts with the partons in the proton ( direct contribution ) . for @xmath6 , in addition to the direct contributions , there are also resolved contributions , where the incoming photon changes into an initial quark or gluon , which interact with partons from the incoming proton . bottom photoproduction has been measured using several different methods by the h1 @xcite and zeus @xcite collaborations at hera . in most of the measurements ,
the cross section was obtained using semi - leptonic decays into muons or electrons .
all these measurements have been compared with predictions based on the fixed - flavor - number scheme ( ffns ) @xcite . in this scheme ,
the bottom quark is generated in the hard scattering - process and appears only in the final state .
predictions for the photoproduction cross section are calculated at next - to - leading - order ( nlo ) taking the finite mass of the bottom quark explicitly into account .
reasonable agreement of the data with ffns predictions has been obtained . in the large-@xmath7 region ,
characterized by @xmath8 , the so - called massless or zero - mass variable flavor - number - scheme ( zm - vfns ) @xcite is considered more appropriate .
this is the conventional parton model approach , where the @xmath0 quark , considered massless like any other incoming parton , is also an incoming parton from the proton and , for the resolved contribution , from the incoming photon .
this gives rise to additional contributions from hard - scattering subprocesses with @xmath0 quarks in the initial state , besides those with @xmath9 , @xmath10 , @xmath11 and @xmath12 quarks and the gluon ( @xmath13 ) .
predictions in this approach are reliable only in the large @xmath7-region where terms of the order @xmath14 are negligible .
calculations at nlo automatically resum leading and next - to - leading logarithmic terms @xmath15 . at the same time , all non - logarithmic terms through @xmath16 are retained for @xmath17 . with the usual choice of renormalization ( @xmath18 ) and factorization scales ( @xmath19 and @xmath20 for initial and final - state factorization , resp . ) , @xmath21 , the results are dominated by contributions from the @xmath0-quark pdf of the proton down to @xmath22 . the neglect of the terms of order @xmath14 in the hard scattering cross section is remedied in the general - mass variable - flavor - number scheme ( gm - vfns ) @xcite .
this theoretical framework combines the ffn and the zm - vfn schemes .
the mass - dependent terms of the ffns are added to the zm - vfns by applying subtractions in such a way that in the limit @xmath23 the hard - scattering cross sections in the zm - vfns are recovered . however , with the conventional choice of scales indicated above , the results in the gm - vfns are still dominated by the contributions of the @xmath0-quark pdf down to @xmath24 . therefore , there is no smooth transition from the gm - vfns at large @xmath7 values to the ffns in the small-@xmath7 range , and the gm - vfns in its original definition fails to describe the small-@xmath7 cross sections .
the dominance of contributions with @xmath0-quarks in the initial state at small @xmath7-values is connected to the fact that this part is treated in the massless scheme .
a calculation of the @xmath0-quark initiated subprocesses in a scheme with massive partons is not available for photo- and for hadroproduction .
in addition it would be necessary to correctly describe the effects of the finite bottom - quark mass @xmath25 in the dglap evolution .
attempts in this direction have been presented in @xcite .
the cross section with massless partons is divergent for @xmath26 . for a realistic description
it is therefore necessary to eliminate the contribution with massless @xmath0-quarks in the initial state in the small-@xmath7 region .
for inclusive @xmath27-meson production in @xmath28 collisions this has been achieved in recent work together with kniehl and schienbein @xcite by a suitable choice of scales @xmath19 and @xmath20 . in this
modified gm - vfns the ffns results with the exact @xmath25 dependence are recovered at small @xmath7 .
this way reasonable results were obtained for the inclusive @xmath27-meson production cross section measured with cdf at the fermilab tevatron and with lhcb at the cern lhc .
it is the aim of this work to apply the same prescription for the choice of the factorization scales to @xmath0-quark photoproduction . for the resolved part
the modified prescription works as described in @xcite for hadron - hadron scattering .
for the direct part a corresponding calculation has not been done yet .
we shall do this here and compare our results to recent zeus and h1 cross section data in order to study whether the modifications of the gm - vfns can lead to a better agreement with presently available experimental data .
cross sections for inclusive charmed - hadron photoproduction have been calculated in the past @xcite , but without the scale modifications described in @xcite .
the corresponding experimental data are not available for very small @xmath7 , where the modifications would be most clearly visible . for somewhat larger @xmath7 ,
very good agreement of the various observables for inclusive @xmath29 photoproduction has been found @xcite .
the outline of this paper is as follows . in sect .
2 we introduce our strategy for the transition to the ffns and compare our predictions with the zeus and h1 data . our conclusions are presented in sect .
in this section , we shall consider a viable framework for inclusive @xmath30 production , where @xmath30 is any hadron containing a @xmath0-quark .
our aim is to unify prescriptions for the calculation of theoretical predictions at small and at large @xmath7-values .
we shall compare with the differential cross section @xmath31 measured by the zeus and the h1 collaborations at hera .
we take the @xmath0-quark pole mass to be @xmath32 gev , evaluate @xmath33 at nlo with @xmath34 and @xmath35 mev if @xmath36 and with @xmath37 and @xmath38 mev if @xmath39 . for the proton pdfs
we use the parametrization cteq6.6 @xcite while for the photon pdf we use afg04_bf @xcite as our default , but we will also consider other pdf choices for comparison . we start with results to be compared with the zeus data @xcite shown in fig .
[ fig:1 ] together with our theoretical predictions .
the data have been measured in various different analysis .
only five data points were reported in @xcite .
the other 13 points are from earlier publications .
their origin can be traced back from fig .
8 of @xcite .
the zeus data are for the differential cross section for @xmath0-quark production as a function of @xmath7 ( between 5 and 30 gev ) with the kinematic constraints @xmath40 gev@xmath41 for the momentum transfer squared , @xmath42 for the inelasticity @xmath43 ( where @xmath44 is the bjorken scaling variable and @xmath45 the total energy in the center - of - mass reference frame ) and @xmath46 for the @xmath0-quark pseudo - rapidity .
the data points at @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 gev from @xcite have in average smaller experimental errors than the earlier measurements since they are measured with the largest integrated luminosity of 133 pb@xmath52 . note that not all points are consistent with each other , i.e. for some points the systematic errors are presumably larger than given in @xcite . in most of the previous measurements of bottom photoproduction at hera , the cross section was determined using semi - leptonic decays into muons or electrons . in
@xcite the @xmath0-quark photoproduction cross sections were determined for the production of jets containing @xmath0-quarks .
the cross sections for the production of @xmath0-jets have been converted into cross sections for @xmath0-quark production with the help of the fmnr program @xcite . in fig .
8 of ref .
@xcite the data have been compared with predictions of the ffns @xcite . in fig .
[ fig:1 ] ( upper left frame ) , we compare the zeus data with nlo predictions in the gm - vfns . our framework has been described in detail in refs .
@xcite for photoproduction of @xmath29 mesons .
the full line in fig .
[ fig:1 ] ( upper left frame ) shows the result for the old default choice of scales , @xmath53 with @xmath54 for @xmath55 , @xmath56 , @xmath57 where @xmath58 , @xmath56 , and @xmath57 denote renormalization , initial - state and final - state factorization scales .
the dashed lines represent an estimate of the theoretical error by varying @xmath59 up and down by a factor of 2 .
13 data points agree with the theoretical prediction within experimental and theoretical errors , 5 points lie outside this range .
the central curve in fig . [ fig:1 ] ( upper left frame ) , as well as the curve for the upper error estimate , show the characteristic behavior of a monotonic increase of @xmath31 in the limit @xmath60 .
this behavior is caused by the scale choice with @xmath61 .
the curve for the lower error estimate with @xmath62 , however , shows a turn - over towards small @xmath7 with a maximum at @xmath63 gev .
this is the same characteristic behavior as in the ffns @xcite and is caused in our case by the fact that at around @xmath64 gev the factorization scales @xmath65 fall below @xmath25 where the @xmath0-quark pdf and ff are zero .
the same typical turn - over was also found in our previous study of inclusive @xmath27-meson production for @xmath66 and @xmath67 collisions @xcite .
actually , the @xmath0-quark cross section is calculated from the @xmath27-meson cross section using an evolved fragmentation function ( ff ) for @xmath68 taken from @xcite and dividing by the branching fraction @xmath69 .
the effect of the evolved ff is visible at larger values of @xmath70 gev where the cross section @xmath31 is smaller than without a ff .
we emphasize that a consistent theoretical prescription of the gm - vfns requires to include a ff describing the @xmath71 transition since final state singularities have to be subtracted from the bare cross section in analogy to the calculation in the zm - vfns .
unfortunately all the zeus measurements are for @xmath72 gev , so that any turn - over of the cross section @xmath31 towards @xmath22 is not measured . all data points agree reasonably well also with the gm - vfns predictions with the original default scale @xmath73 .
a monotonic increase of the cross section @xmath31 for the production of a massive quark towards small @xmath7 is unphysical . from our previous work on @xmath27-meson production in @xmath66 and @xmath67 collisions @xcite we know how the choice of the factorization scales must be modified in order to obtain cross sections for @xmath60 which show the expected behavior .
therefore , as in @xcite , we choose as the new default scales @xmath74 and @xmath75 such that both the @xmath0-quark pdf and the ff for the @xmath71 transition vanish for @xmath76 already at non - zero @xmath7 . with @xmath77 , @xmath78 as the new default scale choice
, we estimate the theoretical error in the usual way by varying the renormalization scale parameter @xmath18 by a factor 2 up and down about the default scale .
we have not introduced an extra prescription to freeze @xmath18 below @xmath25 because the choice of @xmath18 is not related to switching off @xmath0-quark initiated subprocesses .
the resulting error band is shown in fig .
[ fig:1 ] ( upper right frame ) together with the full curve for the new default choice and the zeus data @xcite .
the agreement with the data is only marginally better than in the upper left frame of fig .
[ fig:1 ] .
since the data have @xmath7 values larger than 5 gev the turn - over of @xmath31 towards @xmath60 is not tested by the zeus measurements . in the lower part of fig .
[ fig:1 ] we compare the zeus data with results obtained in the ffns .
here we have fixed the scales to @xmath79 .
the ffns prediction is only slightly below the gm - vfns calculation at low @xmath7 , but lies above it by about a factor of two at the largest @xmath7 .
the difference at large @xmath7 is mainly due to the fact that the ffns calculation does not include an evolved ff .
we observe that the ffns calculation over - shoot the high - precision data points of zeus .
similar data are available from the h1 collaboration at hera .
the most recent publication @xcite describes results for photoproduction of @xmath0-quarks measured in the process @xmath80 .
the decay channel @xmath81 is selected by identifying the semi - electronic decays of the @xmath0-quarks .
the production cross section is measured in the kinematic range where the photon virtuality is small , @xmath82 gev@xmath41 , the inelasticity is restricted to the range @xmath83 and the pseudorapidity of the @xmath0-quarks is in the range @xmath84 .
the measured cross sections are converted into single - inclusive @xmath0-quark production cross sections @xmath31 for four @xmath7 bins in the range @xmath85 gev .
we have calculated the @xmath86 cross section @xmath31 as a function of @xmath7 in this @xmath7 range and with the same kinematic constraints as in @xcite in the original gm - vfns with default scales @xmath87 and with the new default scales @xmath77 , @xmath88 . the results , together with scale variations as above to obtain an error estimate , are shown in fig .
[ fig:3 ] together with the h1 data .
the upper left panel of fig .
[ fig:3 ] shows results for the original scale setting , the upper right frame is for the new scale choice .
the experimental data for @xmath89 gev agree with the predictions in both schemes .
only the cross section @xmath31 in the lowest-@xmath7 bin , @xmath90 gev , is lower than the predictions in both cases . in this @xmath7-bin
the measured cross section is smaller than the theory prediction by approximately a factor of 6 in the upper left frame of fig .
[ fig:3 ] and by a factor 3 to 4 in the upper right frame of fig .
[ fig:3 ] .
although the @xmath7-dependence of the differential cross section @xmath31 exhibits the expected turn - over at @xmath91 gev if the new default scales are chosen , the predicted cross section is still larger than the experimental data , even when taking into account the experimental uncertainty .
a comparison of the data with the ffns prediction is shown in the lower frame of fig .
[ fig:3 ] , again choosing scales as @xmath79 and without folding with a ff . as above ,
the ffns results are slightly below those of the gm - vfns at low @xmath7 , but above at large @xmath7 .
we note that the nlo corrections in the ffns are positive and lead to an increase of the cross section by about a factor of 1.5 at low @xmath7 .
the h1 data with its relatively large errors are unable to discriminate between the two schemes .
we should remark that the h1 data are presented as a function of the average transverse momentum of two produced bottom quarks , while our calculation is for the one - particle inclusive cross section where the independent kinematic variable is the transverse momentum of one observed @xmath0-quark .
this may be the reason that the theoretical predictions shown in @xcite do not agree with our ffns evaluation . left frame : ratios of the differential cross sections @xmath31 for bottom photoproduction using different proton pdf parametrizations with respect to the default cteq6.6 @xcite pdf .
the full curve is for ct10 @xcite , the dashed curve for herapdf1.5 @xcite .
right frame : ratio of the differential cross section @xmath31 for bottom photoproduction as a function the transverse momentum @xmath7 for the photon pdf grv ho @xcite with respect to the photon pdf afg04 @xcite . in both figures , the kinematic constraints of the zeus data and the new default scales @xmath74 , @xmath75 have been used .
, title="fig:",width=302 ] left frame : ratios of the differential cross sections @xmath31 for bottom photoproduction using different proton pdf parametrizations with respect to the default cteq6.6 @xcite pdf .
the full curve is for ct10 @xcite , the dashed curve for herapdf1.5 @xcite .
right frame : ratio of the differential cross section @xmath31 for bottom photoproduction as a function the transverse momentum @xmath7 for the photon pdf grv ho @xcite with respect to the photon pdf afg04 @xcite . in both figures , the kinematic constraints of the zeus data and the new default scales @xmath74 , @xmath75 have been used .
, title="fig:",width=302 ] in order to study the influence of the choice of the proton pdf we present also results for two other pdfs : ct10 @xcite and herapdf1.5 ( nlo ) @xcite .
the results are shown in fig .
[ fig:2 ] ( left frame ) as ratios to the prediction obtained for the default choice cteq6.6 .
the differences to cteq6.6 are very small , in particular for ct10 , as to be expected .
for herapdf1.5 the difference is somewhat larger . in the range @xmath92 gev
the cross section with herapdf1.5 is larger than for cteq6.6 and for larger @xmath7 values , @xmath70 gev , the ratio is smaller than 1 and decreases towards 0.9 for @xmath93 gev .
we include also a comparison with two different choices of the photon pdfs , shown in the right frame of fig .
[ fig:2 ] .
again , we present the results as a ratio of cross sections and choose the grv ho @xcite photon pdf as an alternative to the afg04 parametrization .
the cross section ratio is very close to 1 . only for the smallest considered @xmath7 value the ratio reaches values slightly above @xmath94 .
as is well - known , a complete calculation of the photoproduction cross section has to take into account two parts , the direct contribution and the resolved contribution .
it is interesting to study whether both parts show the same dependence on @xmath7 . to see this in some detail
we have calculated these two parts separately , denoted @xmath95 and @xmath96 .
results for the h1 kinematic conditions are shown in fig .
[ fig:4 ] . for the sake of a more clear visibility
we display the ratios of @xmath97 and @xmath98 where @xmath99 , res and @xmath100 is the sum of the direct and resolved contributions as shown fig .
[ fig:3 ] ( upper right frame ) . from fig .
[ fig:4 ] we see that the resolved cross section varies between @xmath101 and @xmath102 for @xmath103 gev , while it stays near @xmath101 for the larger @xmath7 values .
this means that even when the resolved part would vanish for all @xmath7 the direct part would change by less than @xmath102 .
this is too small to be identified unambiguously in the data . in particular , a different @xmath7 dependence of the direct part
can not be made responsible for the low cross section measured by h1 at @xmath104 gev .
we have presented predictions for @xmath0-quark photoproduction at hera at next - to - leading order in the general - mass variable - flavor - number scheme .
in contrast to previous calculations , we have fixed the factorization scale parameters in such a way that @xmath0-quark initiated contributions are eliminated at low transverse momenta . with this new prescription
we revocer the typical low-@xmath7 behaviour of a fixed - flavor - number scheme .
the comparison with experimental data from the zeus collaboration shows reasonable agreement over the whole range of transverse momenta , while we observe a discrepancy in the lowest-@xmath7 bin of the h1 measurement . | we present predictions for @xmath0-quark production in photoprodcution and compare with experimental data from hera .
our theoretical predictions are obtained at next - to - leading - order in the general - mass variable - flavor - number scheme , an approach which takes into account the finite mass of the @xmath0 quarks . we use realistic evolved nonperturbative fragmentation functions obtained from fits to @xmath1 data .
we find in general good agreement of data with both the gm - vfns and the ffns calculations , while the more precise zeus data seem to prefer the gm - vfns predictions .
+ + pacs : 12.38.bx , 12.39.st , 13.85.ni , 14.40.nd |
a comprehensive theory of vortex dynamics in quantum fluids must address phenomena occurring at three different length scales . at the smallest , most microscopic scale , a fully quantum mechanical treatment of a vortex , including its internal structure ,
interaction with elementary excitations such as phonons and rotons , and interaction with disorder , is required at the next
what we shall refer to as intermediate
length scale , one would like to regard a vortex as a classical object , subject to a classical equation of motion . of course , there is no guarantee that we can do this , and , indeed , certain pathologies result from our insistence in doing so , but experience has shown that this classical picture is extremely successful.@xcite there are different ways to formulate the classical approach . if we let @xmath0 denote the position in the @xmath1 plane of the center of an isolated straight vortex line , say , as a function of time , we could hypothesize a phenomenological equation of motion of the form @xmath2 here we have taken the circulation vector @xmath3 of the vortex , a vector parallel to the vortex with a magnitude equal to the circulation , to be along the @xmath4 direction .
the first two terms on the right - hand - side of ( [ equation of motion ] ) are to include all forces linear in the vortex velocity .
the coefficients @xmath5 , @xmath6 , and @xmath7 , which describe the vortex effective mass per unit length , viscous damping force per unit length , and nondissipative transverse force per unit length , respectively , are to be determined from a microscopic theory , as are the `` pinning '' and `` driving '' forces per unit length @xmath8 and @xmath9 .
the latter may depend on the normal and superfluid densities and velocities , and therefore on the vortex position @xmath10 , but do not depend on the vortex velocity . in a superfluid , @xmath8 might describe the force exerted by an externally imposed wire , and @xmath9 would include the superfluid - velocity - dependent part of the magnus force ( see below ) and possibly other vortex - velocity - independent contributions . at the most macroscopic scale one needs to understand how the forces arising at the intermediate scale act to determine the bulk , experimentally observable properties of quantum fluids . the length scale that usually defines this regime is the characteristic inter - vortex distance , and one is interested in coarse - grained properties of the quantum fluid at scales larger than that distance . in superfluids ,
this is the regime where the concept of mutual friction applies .
mutual friction refers to a momentum transfer , with both longitudinal and transverse components , between the normal and superfluid parts of a quantum fluid.@xcite although such an interaction is absent in the two - fluid model itself , the presence of vortices mediate a bulk momentum exchange
. similarly , in superconductors one needs to understand how the lorentz force , pinning forces , and other intermediate - scale forces conspire to determine , say , the hall effect in the mixed state , which depends on the collective motion of a macroscopic number of vortices.@xcite our recent work has focused mostly on the intermediate length - scale regime , namely , the determination of the various forces that act on vortices . after a brief general review of that work we shall discuss our new results on the problem of the iordanskii and lifshitz - pitaevskii forces , which have been subjects of considerable controversy .
the thouless - ao - niu ( tan ) theory@xcite and its generalization to ultraclean type - ii superconductors@xcite start with the appropriate exact many - body hamiltonian and include a pinning potential centered at @xmath10 , which is also the position of the vortex .
the vortex is then dragged with a velocity @xmath11 , and a transverse force @xmath12 is found . here @xmath13 is the canonical momentum density , and the line integral is taken around a large circle enclosing the vortex . evidently , the transverse force is independent of the detailed microscopic structure of the vortex and its interaction with the normal fluid . in a neutral bose or fermi superfluid , @xmath14 and therefore @xmath15 in the normal fluid component ,
viscosity causes vortex - like circulation to diffuse away to the outer boundary , so @xmath16 vanishes in equilibrium , and @xmath17 in a charged superfluid or superconductor , however , the hamiltonian contains a current - current interaction term that modifies the velocity operator . the canonical momentum density can be expressed in terms of the physical , gauge - invariant momentum density @xmath18 , resulting in @xmath19 the second equality in ( [ sc case ] ) follows from the meissner effect , which causes @xmath20 to vanish at large distances , and from flux quantization .
therefore , @xmath21 where @xmath22 is the total mass density of the fluid .
we turn now to a brief discussion of some of the controversies in the theory of intermediate - scale vortex dynamics , focusing on transverse forces , and organized according to the complexity of the quantum fluid in question .
it seems appropriate to start by recalling a result of classical hydrodynamics , the magnus force : when a vortex with circulation @xmath3 is dragged with velocity @xmath11 through an ideal fluid of mass density @xmath23 , a transverse force per unit length @xmath24 acts on the object doing the dragging .
the force is similar to the lift force on an airplane wing .
it is clear from galilean invariance that if the fluid far from the vortex is not at rest , but has a velocity @xmath25 , then the force is instead @xmath26 by analogy , it is reasonable in a neutral bose superfluid at zero temperature to expect that @xmath27 where @xmath28 is the quantized circulation .
of course , writing @xmath29 here instead of @xmath23 is arbitrary , because these are equal at zero temperature .
the controversy begins in the finite - temperature neutral bose superfluid , because galilean invariance allows for a transverse force of the form @xmath30 where @xmath31 and @xmath32 are parameters . the first term in ( [ general sf magnus force ] )
has a simple classical interpretation : it describes a magnus - type force originating from the superfluid component of the fluid ( remembering that it is the superfluid that is circulating ) .
thus classical reasoning would suggest that @xmath33 , and wexler has recently given an elegant proof of this.@xcite to our knowledge there have been no criticisms of wexler s theory . from this classical point - of - view ,
again keeping in mind that it is the superfluid that is circulating here , the second term in ( [ general sf magnus force ] ) would be of a nonhydrodynamic origin .
if present , it would describe a transverse interaction between the vortex and the excitations phonons and rotons of the fluid .
iordanskii@xcite predicted just such an interaction with phonons , and lifshitz and pitaevskii@xcite ( following earlier work by hall and vinen@xcite ) predicted one with rotons .
note , however , that in these works @xmath34 and @xmath35 refer to flow velocities _ near _ the vortex line , whereas our quantities are asymptotic values .
but the tan result ( [ sf transverse force ] ) implies @xmath36 .
when combined with wexler s theory , this implies @xmath37 , i.e. , that there are no transverse iordanskii and lifshitz - pitaevskii forces .
this apparent discrepancy has motivated us to understand better the interaction between a quantized vortex and the normal fluid in a neutral bose system , and to calculate the coefficient @xmath32 in ( [ general sf magnus force ] ) directly .
we shall return to this direct calculation below in section [ two - fluid ] the next controversy concerns the transverse force in a neutral fermi superfluid , also described in the tan theory . according to kopnin and kravtsov,@xcite low - energy quasiparticles in the vortex core also contribute to the transverse force linear in @xmath11 , a contribution not found by tan .
vortices in superconductors inherit all of the above controversies and have additional complexity of their own.@xcite we will not have space to discuss them further .
two ingredients are needed for a microscopic theory of the coefficient @xmath32 in eqn .
( [ general sf magnus force ] ) .
first , one has to solve a vortex - excitation scattering problem .
it is probably correct to use the gross - pitaevskii equation to do this , even though mean field theory is not expected to hold inside the vortex , because the scattering potential is long - ranged .
in fact , the potential is sufficiently long - ranged , and the forward scattering sufficiently singular , that there are several mathematically incorrect scattering calculations present in the literature.@xcite in the scale of lengths greater than the size of the vortex core and less than the mean free path for scattering of excitations by one another , if there is such a range , the dynamics of excitations moving in the slowly varying background of the rotating superfluid is well defined , and in this region the traditional arguments@xcite give a transverse force on the excitations equal to @xmath38 , where @xmath35 is the average normal fluid velocity at a distance of the order of the mean free path from the vortex core , as well as a dissipative longitudinal force . in the region well beyond a mean free path from the vortex core , flow velocities are varying slowly and two - fluid hydrodynamics , which incorporates the basic conservation laws , provides an accurate description .
in particular , the two - fluid version of the navier stokes equation gives the force momentum balance . in this region we have studied how forces generated within a mean free path of the vortex core can be transmitted to large distances.@xcite within a linearized approximation to the two - fluid model the force due to the superfluid flow relative to the vortex is just the ordinary superfluid magnus force @xmath39 , where @xmath34 is the asymptotic value of the superfluid velocity .
this is given for superfluids , as it is for ordinary fluids , by the cross terms , between circulation and the superfluid flow , in the momentum flow tensor and the bernoulli pressure .
there seems to be no possible modification of this even for a fermionic superfluid , such as is suggested by the work of kopnin and kravtsov,@xcite if there is no bulk interaction with a stationary background . for the normal fluid contributions our considerations are quite similar to those of hall and vinen.@xcite in the linear regime the force
is transmitted by the viscous force ( and an accompanying pressure term ) , which is generated by a flow velocity in the direction of the force that increases as the logarithm of the distance from the vortex core .
the asymptotic value of the normal fluid velocity is given by a combination of the value @xmath35 close to the core , and this logarithmically growing term .
the logarithmically growing term has to be cut off at a distance @xmath40 which is determined by the reynolds number of the flow , by the spacing between vortices , or by the diffusion length at the frequency of vortex oscillation . in the limit considered by tan , with vanishingly small normal fluid velocity , and a single vortex in an infinite medium , this
cut - off goes to infinity , and @xmath35 is negligibly small in comparison , so the asymptotic velocity is determined by the logarithmic term , and the force is parallel to the normal fluid flow .
realistically , the logarithm which the cut - off introduces is not necessarily large compared with other parameters in the problem .
even when the iordanskii force is not put in explicitly at the inner boundary , there is some transverse force due to normal fluid flow , but this is proportional to @xmath41^{-2 } \ , v_{\rm n}$ ] , where @xmath42 is the mean free path .
it is not clear how to match conditions close to the core with conditions in the hydrodynamic region , but if it is assumed that the iordanskii force causes @xmath35 close to the core to match the viscous force in the hydrodynamic region , and to be parallel to it , the relative importance of the logarithmic term will depend on the ratio of the kinematic viscosity @xmath43 to the quantum of circulation a large kinematic viscosity will diminish the importance of the logarithmic term in the flow velocity .
this work was supported in part by nsf grant no .
dmr-9813932 , by a research innovation award from the research corporation , and by a sarah moss fellowship . | it has been known since the pioneering work of onsager and feynman that the statistical mechanics and dynamics of vortices play an essential role in the behavior of superfluids and superconductors .
however , the theory of vortices in quantum fluids remains in a most unsatisfactory state , with many conflicting results in the literature . in this paper
we review the theory of thouless , ao and niu , which gives an expression for the total transverse force acting on a quantized vortex that is in apparent disagreement with the work of iordanskii and of lifshitz and pitaevskii .
in particular , no transverse force proportional to the asymptotic normal fluid velocity was found .
we use two - fluid hydrodynamics to study this discrepancy .
pacs numbers : 67.40.vs , 74.60.ge , 47.37.+q |
two of the most basic problems in the theory of packings and coverings is to find the most efficient packing and covering by replicas of a given set @xmath0 in either @xmath1-dimensional euclidean space or @xmath1-dimensional hyperbolic space ( [ fek ] ) .
the term ` most efficient ' is intentionally ambiguous as different circumstances have led to different interpretations .
generally , the density of a packing of euclidean space is defined as the limit of the relative fraction of volume occupied by the bodies in the packing in a ball of radius @xmath4 centered at some point @xmath5 as @xmath4 tends to infinity .
it can be shown that if this limit exists for some point @xmath5 , it exists for every point @xmath5 and has the same value .
a densest packing is then one which maximizes this density .
it is easy to prove the existence of a densest packing in euclidean space .
the density of a covering can be defined similarly and a thinnest covering is one that minimizes the density .
one disadvantage of this definition is that is does not depend on local structure .
for example , if a finite number of bodies in a densest packing are removed , it remains a densest packing .
another disadvantage is that it is not clear that this definition carries over well to hyperbolic space .
for instance , the above limit may exist for every point in hyperbolic space yet take on different values for different points .
for these and other reasons , the authors of [ fkk ] introduced the concept of a completely saturated packing .
for such a packing , it is not possible to replace a finite number of bodies of the packing with a greater number of bodies and still remain a packing .
similarly a completely reduced covering is one in which it is not possible to replace a finite number of bodies of the covering with a smaller number of bodies and still remain a covering . in [ fkk ] , it was proven that any convex body of euclidean space admits a completely saturated packing ( and more generally any body with the strict nested similarity property ) ( see also [ kup ] ) .
the authors conjectured that completely saturated packings and completely reduced coverings exist for any body @xmath0 in either @xmath1-dimensional euclidean or hyperbolic space . in section 4 , we prove this conjecture .
the proof requires few properties of either euclidean or hyperbolic space and easily extends to a more general setting ( see remark after theorem 2.2 ) .
our techniques are based on those developed in [ bor ] . in that paper ,
a new notion of efficiency was introduced which intuitively applies to a class of measures on a space of packings rather than to individual packings .
such measures are often easier to analyze but at the same time they carry information about the ` average ' properties of some packings . in particular , birkhoff s ergodic theorem ( in the euclidean case ) and recent results due to nevo and stein ( in the hyperbolic case ) show that this new notion ( to be explained in section 3 ) coincides ( in a measure - theoretic way ) to the usual notion of density given by limits of volumes in expanding balls .
this will enable us to construct a peculiar family of bodies @xmath0 that admit tilings of hyperbolic space but such that they also admit completely saturated packings with arbitrarily low density . * acknowledgments *
: i would like to thank my advisor charles radin for many useful conversations and suggestions .
let @xmath6 denote either @xmath1-dimensional hyperbolic or @xmath1-dimensional euclidean space with a distinguished point @xmath7 which we call the origin .
let @xmath8 be the group of orientation - preserving isometries of @xmath6 .
a * body * @xmath0 is a nonempty connected compact subset of @xmath6 which equals the closure of its interior .
let @xmath9 .
a * packing * by a body @xmath0 is a collection @xmath10 of congruent copies of @xmath0 that pairwise have disjoint interiors .
note that if @xmath11 for some @xmath12 , then @xmath13 for all @xmath14 since @xmath15 implies @xmath16 .
conversely , if @xmath17 then @xmath18 .
so we let @xmath19 be the subset of the space @xmath20 of left cosets defined by @xmath21 if any only @xmath22 .
we note that @xmath19 uniquely determines @xmath23 ( given @xmath0 ) so if @xmath24 , we will say that @xmath25 determines the packing @xmath23 if @xmath26 .
later , we will use this correspondence to define a metric on the space of packings .
a * covering * by a body @xmath0 is a collection @xmath27 of congruent copies of @xmath0 such that for every point @xmath28 , @xmath29 for some @xmath30 . as
above , we let @xmath31 be the subset of the space @xmath20 of left cosets defined by @xmath32 if any only @xmath33 .
a * tiling * by @xmath0 is a packing that is also a covering .
a * completely saturated * packing @xmath23 is such that for no packing @xmath34 does there exist another finite packing @xmath35 with more bodies that @xmath36 such that @xmath37 is a packing .
a * completely reduced * covering @xmath38 is such that for no finite subset @xmath39 does there exist another finite set @xmath40 with fewer elements that @xmath36 such that @xmath41 determines a covering . note that this is the equivalent to the definition given earlier .
our main result is the following . for any body @xmath0 in @xmath6
, there exists a completely saturated packing by @xmath0 and a completely reduced covering by @xmath0 .
* remark * : the theorem extends with almost no changes to the case in which @xmath42 where @xmath8 is a unimodular lie group , @xmath43 is a compact subgroup of @xmath43 and @xmath8 satisfies the following property . for every compact subset @xmath44 of @xmath8 there exists a discrete cocompact subgroup @xmath45 such that there exists a fundamental domain @xmath46 for @xmath47 and @xmath48 . in this
setting , ` congruent copies of a body @xmath49 ' means subsets of the form @xmath50 where @xmath51 and @xmath8 acts on subsets of @xmath6 in the natural way .
+ + we will only prove the existence of completely saturated packings for @xmath0 as the other assertion can be proven similarly .
for any point @xmath28 , we let @xmath52 denote the closed ball of radius @xmath1 centered at @xmath5 .
also we let @xmath53 denote the usual measure on @xmath6 .
finally for any packing @xmath23 we let @xmath54 be the closed subset associated to @xmath23 .
we obtain the following counterintuitive result : let @xmath6 be @xmath1-dimensional hyperbolic space for some @xmath55 .
let @xmath2 .
then there is a body @xmath0 in @xmath6 that admits a tiling @xmath56 and a completely saturated packing @xmath23 with the following properties .
for every point @xmath28 , the limit @xmath57 exists and is independent of @xmath5 .
furthermore this common limit is less than @xmath3 .
the paper is organized as follows . in section 3
, we introduce the main tools of our study . in section 4 , we prove theorem 2.1 . in section 5 , we prove theorem 2.2 .
let @xmath58 be the stabilizer @xmath59 of the origin @xmath7 .
it is isomorphic to @xmath60 .
let @xmath61 be the canonical projection identifying @xmath62 with @xmath6 such that @xmath63 for every @xmath51 .
let @xmath64 be the normalization of haar measure on @xmath8 such that @xmath53 , the usual measure on @xmath6 , has the following relationship with @xmath64 ( see [ rat ] for the hyperbolic case ) .
for any borel set @xmath65 , @xmath66 .
we note the general fact ( see [ lan ] ) that the haar measure on any locally compact group @xmath67 is determined uniquely up to scalar multiplication by the property that it is a nonzero borel measure that is invariant under left multiplication by every element @xmath68 ( i.e. @xmath69 for every @xmath70 and borel set @xmath71 where @xmath72 is a haar measure of @xmath67 ) .
let @xmath73 be the usual distance function on @xmath6 .
let @xmath74 be any left - invariant distance function on @xmath8 satisfying @xmath75 and @xmath76 .
the left - invariant of @xmath74 means that for any @xmath77 , @xmath78 . for any body @xmath0 in @xmath6
, we define a distance function @xmath79 on @xmath20 by @xmath80 . in other words
, @xmath79 is the pushforward of @xmath74 .
let @xmath81 and @xmath82 denote the closed balls of radius @xmath1 centered at @xmath7 in @xmath6 , @xmath83 in @xmath20 and the identity in @xmath8 respectively . if @xmath84 is a subset of @xmath20 and @xmath2 we let @xmath85 denote the open @xmath3-neighborhood of @xmath84 in @xmath20 .
let @xmath86 denote the set of packings of @xmath6 by @xmath0 .
first , for every @xmath87 , define a pseudometric @xmath88 on @xmath86 by @xmath89 next we define a metric @xmath90 on @xmath86 by : @xmath91 it can be checked that @xmath86 is a compact metric space on which @xmath8 acts jointly continuously in the obvious way , i.e. for @xmath92 and @xmath93 , @xmath94 ( cf [ raw ] ) .
we note that in euclidean space , if @xmath95 is translation by @xmath96 , then @xmath97 is commonly denoted by @xmath98 .
although the metric @xmath90 will be convenient in what is to follow , we are really only concerned with the topology that it induces .
the idea is that two packings are close in @xmath86 if in a large ball about the origin , the bodies of one packing can by small rigid motions be made to coincide with the bodies of the other packing .
let @xmath99 be the set of borel probability measures on @xmath86 .
we let @xmath100 be the subset of @xmath101 consisting of all those measures that are @xmath8-invariant .
in other words , @xmath102 is in @xmath103 if and only if @xmath104 for every @xmath51 and borel set @xmath105 .
although we will not use it , we show here how to construct a natural class of measures in @xmath103 .
if @xmath23 is any packing such that the symmetry group @xmath106 of @xmath23 is such that @xmath20 is compact then there is a canonical measure @xmath107 associated to @xmath23 .
there is a homeomorphism @xmath108 from the orbit @xmath109 to @xmath110 defined by @xmath111 .
this homeomorphism commutes with the action by @xmath8 where @xmath8 acts on @xmath110 by left multiplication .
there is a natural measure on @xmath110 induced from @xmath64 since the quotient map from @xmath8 to @xmath110 is a covering map in which the covering transformations are all measure - preserving ( in fact they are isometries of @xmath8 ) .
this measure is invariant under the action of @xmath8 since @xmath64 is invariant .
now a measure @xmath112 on @xmath113 can be induced ( via @xmath108 ) from the measure on @xmath110 .
this measure can then be extended to a measure @xmath114 on all of @xmath86 by @xmath115 .
thus for every packing whose symmetry group is cocompact , there is a natural measure in @xmath103 associated to that packing . for any packing @xmath23 by a body @xmath0 ,
we let @xmath116 be the closed subset associated to @xmath23 .
the density of a measure @xmath117 , @xmath118 is defined by @xmath119 in other words , the density of @xmath120 is the probability ( with respect to @xmath120 ) that the origin lies in a body of a random packing .
we define the ( measure - theoretic ) * optimal density * of a body @xmath0 to be @xmath121 .
a measure @xmath117 is said to be * optimally dense * ( for @xmath0 ) if @xmath122 and @xmath120 is ergodic . by ergodic we mean
that for any @xmath105 such that @xmath123 for all @xmath51 , @xmath124 .
we will sketch a proof that when @xmath6 is euclidean , the optimal density of @xmath0 is equal to the usual notion .
to be precise , for a packing @xmath23 and any point @xmath28 , let @xmath125 be defined by @xmath126 whenever this limit exists .
let @xmath127 .
note that @xmath128 does not depend on the point @xmath5 chosen since @xmath129 where @xmath51 such that @xmath130 .
we will show that @xmath131 .
groemer [ gro ] proved that , when @xmath6 is euclidean , for any body @xmath132 there exists a packing @xmath23 by @xmath0 such that @xmath133 for all points @xmath28 and such that this limit converges uniformly in @xmath5 . by standard ergodic theory ,
there exists a measure @xmath117 whose support is contained in the closure of the orbit @xmath134 . since convergence is uniform in @xmath5 , @xmath135 for all @xmath136 in the closure of @xmath113 . by birkhoff
s ergodic theorem , @xmath137 for @xmath120-almost every packing @xmath136 .
thus @xmath138 and so @xmath139 .
for the other way , let @xmath120 be an optimally dense measure ( the existence of which is proven below ) .
then birkhoff s ergodic theorem implies that for @xmath120-almost every packing @xmath23 , @xmath140 .
thus @xmath141 .
so the two notions are equal . in the hyperbolic case
, we can only show that @xmath141 ( see corollary 5.2 ) .
in fact , for any given @xmath3 we will construct a body @xmath0 such that there exists a tiling by @xmath0 ( and hence @xmath142 ) but @xmath143 .
this immediately implies that the subspace of @xmath86 consisting of all tilings by @xmath0 has measure zero with respect to any measure in @xmath103 .
so , by restricting our attention to invariant measures , many ` pathological ' packings ( or tilings ) are ignored .
it may seem odd to some readers that we call @xmath144 the optimal density in spite of the fact @xmath128 may be greater than @xmath144 .
so we emphasize that @xmath144 is just the _ measure theoretic _ optimal density .
the main theorem of this paper is : let @xmath145 be the set of all packings by @xmath0 that are not completely saturated .
then for any optimally dense measure @xmath120 for @xmath0 , @xmath146 .
theorem 2.1 follows immediately from this one and the following existence theorem . for any body @xmath49
, there exists an optimally dense measure @xmath120 for @xmath0 . a proof of this theorem appears in [ bor , theorem 3.2 ] .
we include the same proof here for completeness .
we let @xmath103 have the weak * topology .
equivalently @xmath147 if and only if for every continuous function @xmath148 , @xmath149 . by standard functional analysis ,
since @xmath86 is a compact metric space , @xmath103 is compact .
let @xmath150 .
let @xmath151 be the characteristic function of @xmath152 . because @xmath151 is upper semicontinuous there exists a decreasing sequence @xmath153 of continuous real valued functions on @xmath86 which converge pointwise to @xmath151
choose a sequence @xmath154 such that @xmath155 as @xmath156 , and , using the compactness of @xmath157 , assume without loss of generality that @xmath158 converges to some @xmath159 .
then @xmath160 as @xmath161 , and @xmath162 as @xmath163 . since @xmath164 and @xmath165 as @xmath163 , @xmath166 . from the krein - milman theorem
there exists an ergodic measure @xmath167 for which @xmath168 , and thus @xmath169 .
but then from the definition of @xmath144 , @xmath170 .
before proving the theorem , we outline the case in which @xmath6 is the euclidean plane .
let @xmath171 be the usual square tiling by squares of side length @xmath172 .
we assume that the origin is in the center of a tile of @xmath171 for all @xmath172 .
let @xmath173 be the group of translations which fix @xmath171 .
let @xmath174 be the set of packings which are not saturated in the square of @xmath171 that contains the origin . to be precise , @xmath175 if and only if there is a finite packing @xmath34 and a finite packing @xmath35 such that @xmath35 has more elements than @xmath36 , for all @xmath176 , @xmath50 is contained in the square of @xmath171 that contains the origin and @xmath177 is a packing . for each @xmath172
, there exists a borel map @xmath178 satisfying : 1 .
@xmath22 and @xmath50 intersects an edge of @xmath171 if and only if @xmath179 and @xmath50 intersects an edge of @xmath171 ; 2 . for any given square in @xmath171 ,
the number of bodies of @xmath180 that are contained in that square is as large as possible given the above constraint ; 3 .
@xmath181 commutes with @xmath173 .
the idea is that @xmath181 replaces @xmath92 with a @xmath180 that is saturated with respect to the squares of @xmath171 .
because @xmath181 commutes with @xmath173 , it induces a map @xmath182 such that if @xmath117 then @xmath183 is invariant under @xmath173 . since @xmath173 is cocompact and discrete we can average @xmath183 over a fundamental domain for @xmath173 in @xmath8 to obtain a measure that is invariant under all of @xmath8 .
this new measure will have a density greater than @xmath120 unless @xmath184 . since @xmath185 is the set of all non - completely saturated packings
, this shows that an optimally dense measure must satisfy @xmath146 which proves the theorem . by a fundamental domain @xmath46 of a subgroup @xmath186 ,
we shall mean a connected set in @xmath6 equal to the closure of its interior such that @xmath187 is a tiling by @xmath46 . for the general case ,
let @xmath188 be a sequence of discrete cocompact subgroups of @xmath8 such that there exist fundamental domains @xmath189 ( in @xmath6 ) for @xmath173 with @xmath190 .
we will also assume that @xmath189 and @xmath173 have been chosen so that for all @xmath191 , @xmath192 .
for the euclidean case , we could , for example , let @xmath173 be the cubic lattice generated by the translations @xmath193 for @xmath194
. then we could choose @xmath189 to be the cube of side length @xmath172 whose center is the origin and whose faces are parallel to the coordinate planes .
for the hyperbolic case , we refer to [ fkk , theorem 4.1 ] for the existence of @xmath188 . for @xmath92 and @xmath195 ,
let @xmath196 be the packing consisting of all elements @xmath22 such that the interior of @xmath50 intersects @xmath46 nontrivially . also let @xmath197 denote the boundary of @xmath46 , i.e. the intersection of @xmath46 with the closure of its complement .
let @xmath198 with the subspace topology .
for any @xmath199 , a filling @xmath200 for @xmath96 is a packing by @xmath0 such that @xmath201 , @xmath202 and the number of elements of @xmath200 is as large as possible given these constraints . for each @xmath203
, there exists a borel map @xmath204 such that for each @xmath199 , @xmath205 is a filling for @xmath96 .
let @xmath203 be fixed and let @xmath206 .
given @xmath207 , @xmath200 a filling for @xmath96 and @xmath208 , let @xmath209 there exists a filling @xmath210 for @xmath211 with @xmath212 . from the compactness of @xmath189 it follows that given @xmath208 , there exists a finite number of packings @xmath213 and fillings @xmath214 for @xmath215 such that @xmath216 .
thus for a given integer @xmath208 , there exists a finite partition @xmath217 of @xmath84 such that each @xmath218 is borel and contained in some @xmath219 .
we will assume that @xmath220 refines @xmath221 ( i.e. for every @xmath222 , there exists an @xmath223 such that @xmath224 ) .
for each @xmath208 and @xmath225 , choose @xmath226 and @xmath227 a filling for @xmath228 so that the following are satisfied : 1 .
@xmath229 for all @xmath208 .
2 . for every @xmath208 , @xmath230 , @xmath231
there is a filling @xmath200 for @xmath232 such that @xmath233 .
this is possible since for each @xmath218 there exists @xmath207 and a filling @xmath200 for @xmath96 with @xmath234 .
3 . if for some @xmath235 and @xmath236 , @xmath237 , then @xmath238 .
we define a map @xmath239 by @xmath240 if @xmath241 .
each @xmath242 is borel ( since each @xmath218 is borel ) .
we claim that @xmath243 converges pointwise as to a map @xmath244 .
let @xmath207 . then by the third condition above , if @xmath245 then @xmath246 as @xmath247 .
so the sequence @xmath248 is cauchy and therefore converges in @xmath86 to an element @xmath249 .
since all the maps @xmath242 are borel , @xmath244 must be borel , too .
we claim that @xmath249 is a filling for @xmath96 for each @xmath207 .
so let @xmath250 and for each @xmath208 choose a filling @xmath251 for @xmath96 such that @xmath252 .
by the definition of @xmath218 and the second condition of @xmath227 listed above , such an @xmath251 exists for all @xmath253 . by the triangle inequality , @xmath254 which goes to zero as @xmath235 goes to infinity .
hence the sequence @xmath255 converges to @xmath249 .
since @xmath251 is a filling for @xmath96 for each @xmath235 , it follows that @xmath249 is also a filling for @xmath96 .
now we are done .
we choose maps @xmath181 satisfying the conclusion of the above lemma .
define @xmath178 by the following properties : 1 . for any @xmath92 , @xmath256
; 2 . for any @xmath257 and @xmath92 , @xmath258 .
it should be clear from this definition that for each @xmath172 , @xmath181 is borel .
let @xmath117 be given .
let @xmath259 , i.e. for every borel set @xmath105 , @xmath260 . by the above , @xmath261 is a borel probability measure that is invariant under @xmath173 .
let @xmath262 be defined by @xmath263 for any borel @xmath105 .
it is easy to show that @xmath264 is @xmath8-invariant and therefore really is in @xmath103 .
the next two lemmas will provide tools for calculating @xmath118 and @xmath265 with respect to the relative density within @xmath189 .
@xmath266 let @xmath267 .
let @xmath151 denote the characteristic function of @xmath152 . by definition of density and of @xmath264 , @xmath268
hence we will be done once we show that for any @xmath92 , @xmath269 so , @xmath270 equation ( 18 ) holds because @xmath64 is inversion - invariant ( see [ rat ] for the hyperbolic case ) and because @xmath271 .
the last equation holds since for any borel set @xmath272 , @xmath273 .
@xmath274 by equation ( 13 ) in the previous lemma , @xmath275 the third inequality holds because @xmath120 is @xmath8-invariant . _
( of theorem 3.1 ) _ let @xmath174 be the set of all packings in @xmath86 that are unsaturated relative to @xmath189 , i.e. for any @xmath175 , there is a packing @xmath136 such that @xmath276 but @xmath277 has more elements than @xmath278 .
@xmath279 the first equality is lemma 4.2 , the second comes from the definition of @xmath261 and the last equality uses lemma 4.3 . if @xmath120 is optimally dense then by definition @xmath280 .
so , @xmath184 for all @xmath172 .
hence @xmath281 .
since @xmath190 for all @xmath172 , @xmath282 is the set of all packings that are not completely saturated .
in this section we will prove theorem 2.2 .
we assume from now on that @xmath6 is @xmath1-dimensional hyperbolic space .
we will need a lemma that follows from much more general ergodic theory results of nevo and stein .
it is an immediate corollary of theorem 3 in [ nes ] . if @xmath8 acts continuously on a compact metric space @xmath84 such that there is a borel probability measure @xmath120 on @xmath84 that is invariant and ergodic under this action , then for every function @xmath283 ( @xmath284 ) the following holds for @xmath120-a.e . @xmath285 .
* remark * : it is not yet known whether @xmath0 can be chosen to be convex .
it is known , however , that there exist convex polygons @xmath294 which admit tilings of @xmath295 but such that @xmath296 .
for example , the 5-gon with vertices at @xmath297 , @xmath298 , @xmath299 , @xmath300 , and @xmath301 ( in the upperhalf plane model ) is one such .
+ + to prove this theorem we will construct a body with some indentations and protrusions such that in any tiling by @xmath0 every protrusion in any copy of @xmath0 must fit into an indentation of another copy of @xmath0 .
there will be more indentations than protrusions implying at least , that it does not admit a tiling of a closed manifold . to show that any invariant measure in @xmath86 has small density
, we will show that the probability that the origin is in a protrusion is about equal to the probability that the origin is in any given indentation .
since there are many more indentations than protrusions , the probability that the origin is in an indentation but not in a protrusion will be high , implying that the density is low .
we will need some notation and a lemma ( which is similar to a key part of the proof of proposition 2 in [ bor ] ) .
let @xmath0 be a body whose symmetry group @xmath83 is trivial .
for any @xmath302 , let @xmath303 there exists @xmath304 such that @xmath305 .
let @xmath0 be a body with @xmath306 .
let @xmath307 such that @xmath308 . if @xmath309 is a borel subset of @xmath6 such that for every @xmath310 there exists a unique @xmath304 such that @xmath311 then @xmath312 note that if for some @xmath319 , @xmath316 , @xmath51 and @xmath320 is in the image of @xmath321 then @xmath322 . since @xmath323 , @xmath324 .
hence @xmath325 .
so if @xmath95 is in the image of @xmath326 and @xmath327 is in the image of @xmath321 then @xmath328 .
we define a measure @xmath329 on @xmath8 by the following .
if @xmath330 is a borel subset of @xmath8 and there exists a @xmath51 with @xmath331 contained in the image of @xmath326 for some @xmath332 , then @xmath333 . by the above statement and the @xmath8-invariance of @xmath120 ,
this is well - defined .
note that the image of @xmath334 equals @xmath335 and thus contains an open set ( since @xmath336 does ) .
so for any borel subset @xmath71 there exists a countable partition @xmath337 of @xmath330 such that for each @xmath338 , there exists @xmath339 such that @xmath340 is contained in the image of @xmath334 .
then we define @xmath341 .
if @xmath344 and @xmath309 satisfy the hypotheses of the above lemma and @xmath309 is the disjoint union of @xmath345 and @xmath346 , then both @xmath345 and @xmath346 satisfy the hypotheses of above lemma and @xmath347 is the disjoint union of @xmath348 and @xmath349 . _
( of theorem 5.3 ) _ for simplicity we will only prove the dimension 2 case .
we will identify @xmath6 with the upperhalf plane model of @xmath295 and thus regard it as a subset of the complex plane @xmath350 ( see [ rat ] ) .
let @xmath356 be the euclidean rectangle with vertices @xmath297 , @xmath357 , @xmath358 , and @xmath359 .
let @xmath360 be the euclidean rectangle with vertices @xmath361 , @xmath362 , @xmath363 , and @xmath364 .
finally , let @xmath365 be the euclidean rectangle with vertices @xmath366 , @xmath367 , @xmath368 and @xmath369 . in the figure above , @xmath376
, @xmath0 is in heavy outline and @xmath294 , @xmath377 , @xmath360 , @xmath378 , @xmath365 , and @xmath379 are in light outline although @xmath365 and @xmath379 are not labeled .
we let @xmath380 . since @xmath381 and @xmath382 as @xmath383 , we can ( and do ) choose @xmath384 so that @xmath385 we let @xmath386 and @xmath387 for every @xmath388 in the upperhalf plane .
we will assume that @xmath389 has been chosen small enough so that for any packing @xmath92 such that the @xmath0 is in @xmath23 and for any @xmath22 such that @xmath390 , @xmath391 .
in other words , unless the protrusion @xmath392 of @xmath50 fits into @xmath360 , no part of @xmath50 overlaps @xmath360 .
this assumption implies @xmath393 is such that for every @xmath394 , there exists a unique @xmath304 such that @xmath395 .
thus lemma 5.3 applies .
* remark * : it follows from the above that the space of tilings by copies of @xmath0 does not admit a @xmath8-invariant measure .
however , this fact can be proven more directly by noting that there is an equivariant continuous map from the space of tilings by @xmath0 onto the space at infinity of the hyperbolic plane .
the latter admits no @xmath8-invariant borel probability measure .
_ ( of theorem 2.2 ) _ let @xmath2 .
let @xmath0 be as in theorem 5.3 .
let @xmath120 be an optimally dense measure for @xmath0 . since @xmath120 is ergodic , corollary 5.2 shows that the set @xmath408 of all @xmath92 satisfying on the other hand , theorem 3.1 shows that there exists a subset @xmath415 of full @xmath120-measure such that every packing in @xmath416 is completely saturated .
then @xmath417 has full @xmath120-measure , in particular it is non - empty .
any @xmath418 satisfies the conclusion of theorem 2.2 . | we prove the following conjecture of g. fejes toth , g. kuperberg , and w. kuperberg : every body @xmath0 in either @xmath1-dimensional euclidean or @xmath1-dimensional hyperbolic space admits a completely saturated packing and a completely reduced covering .
also we prove the following counterintuitive result : for every @xmath2 , there is a body @xmath0 in hyperbolic @xmath1-space which admits a completely saturated packing with density less than @xmath3 but which also admits a tiling .
* msc * : 52c17 , 52a40 , 52c26 + * keywords : * completely saturated , completely reduced , packings , coverings . |
we thank michele arneodo , alessia bruni and risto orava for valuable discussions .
mgr and vak thank , respectively , the ippp at the university of durham and the aspen center for physics for hospitality .
this work was supported by the royal society , by intas grant 05 - 103 - 7515 , by grant rfbr 04 - 02 - 16073 and by the federal program of the russian ministry of industry , science and technology ss-1124.2003.2 .
kaidalov , v.a .
khoze , yu.f .
pirogov and n.l .
ter - issakyan , phys . lett .
* b45 * ( 1973 ) 493 ; + a.b .
kaidalov and k.a .
ter - martirosyan , nucl .
b75 * ( 1974 ) 471 ; + r.d .
field and g.c .
fox , nucl .
b80 * ( 1974 ) 367 .
ter - martirosyan , sov .
jetp * 17 * ( 1963 ) 233 ; nucl . phys .
* b68 * ( 1964 ) 591 ; + i.a .
verdiev , o.v .
kancheli , s.g .
matinyan , a.m. popova and k.a .
ter - martirosyan , sov .
jetp * 19 * ( 1964 ) 148 .
anderson et al . , phys .
* 38 * ( 1977 ) 263 ; + na50 collaboration : b. alessandro et al .
* b553 * ( 2003 ) 167 ; + fnal e866/nusea collaboration : m.j .
leitch et al . , phys .
lett . * 84 * ( 2000 ) 3256 .
j. hfner , yu.p .
ivanov , b.z .
kopeliovich and a.v .
tarasov , phys .
* d62 * ( 2000 ) 094022 ; + l. frankfurt , l. gerland , m. strikman and m.zhalov , phys . rev .
* c68 * ( 2003 ) 044602 ; + l. gerland , l. frankfurt , m. strikman and h. stocker , phys . rev .
* c69 * ( 2004 ) 014904 .
abramovsky , a.v .
dmitriev and a.a .
schneider , arxiv : hep - ph/0512199 ; + a. capella , j. kaplan and j. tran thanh van , nucl .
b105 * ( 1976 ) 333 ; + v.a .
abramovsky and r.g .
betman , sov .
* 49 * ( 1989 ) 747 ; + k. goulianos and j. montanha , phys . rev .
* d59 * ( 1999 ) 114017 ; + s. ostaptchenko , phys . rev .
* d74 * ( 2006 ) 014026 ; phys . lett .
* b636 * ( 2006 ) 40 . | the triple - pomeron coupling lies at the heart of the predictions for high energy diffractive processes .
we explain why the existing determinations , which use single - particle inclusive hadronic data @xmath0 , underestimate the value of the bare coupling , due to the neglect of soft rescattering which populates the rapidity gaps .
we describe how data for the process @xmath1 can be used to give a much more reliable estimate of the bare coupling .
we use the existing , fragmentary , @xmath2 data from hera to show that the triple - pomeron coupling is probably about 3 times larger than the previous determinations .
we emphasize the importance of an explicit measurement of the mass spectrum of the @xmath3 system which accompanies @xmath2 production at hera .
the consequences for ultra high energy cosmic ray showers are mentioned .
ippp/06/65 + dcpt/06/130
+ 29th september 2006 + * the extraction of the bare triple - pomeron vertex ; a crucial ingredient for diffraction * v.a .
khoze@xmath4 , a.d . martin@xmath5 and m.g .
ryskin@xmath4 + @xmath5 department of physics and institute for particle physics phenomenology , + university of durham , dh1 3le , uk + @xmath6 petersburg nuclear physics institute , gatchina , st .
petersburg , 188300 , russia + the triple - pomeron coupling is a crucial ingredient for understanding diffraction .
the value and the @xmath7-dependence of the vertex determine the asymptotic high energy behaviour of total cross sections and the cross section of diffractive dissociation .
moreover , in turn , the calculation of the rapidity gap survival factors , @xmath8 , for diffractive processes depend on the properties of the triple - pomeron vertex .
the factor @xmath8 depends on the particular diffractive process , as well as the values of its kinematic variables , and is found to be of size from about 0.01 to 1 .
an example of an analysis of soft high energy hadronic scattering , with the consequent calculation of the @xmath8 factors , and their relation to the triple - pomeron vertex can be found in @xcite .
the existing determinations of the vertex , which were made many years ago , were based on the analyses of the cross sections for single particle hadronic inclusive processes a+b c+y , [ eq : hh ] in the triple pomeron region where the leading hadron , @xmath9 , carries a large fraction @xmath10 of the incoming momentum , see fig .
[ fig:1 ] .
the production of the leading hadron with @xmath11 selects pomeron exchange with trajectory @xmath12 , providing the quantum numbers of hadrons @xmath13 and @xmath9 are the same . for simplicity
we take @xmath14 , which is generally the case .
in addition the production of a system @xmath3 of high mass @xmath15 is described by pomeron exchange with trajectory @xmath16 ; leading to a diagram with a triple - pomeron vertex , which is shown on the right - hand - side of fig .
[ fig:1 ] . in this formalism
the @xmath0 cross section is given by @xcite m_y^2 = g_a^2(t ) g_b(0 ) g_3p(t ) ( ) ^1+_p(0)-2_p(t ) ( ) ^_p(0)-1 + rrp+ppr terms [ eq : trf ] with @xmath17 and where @xmath18 is the centre - of - mass energy of the incoming @xmath19 system .
here we use sd to denote single - particle diffractive dissociation . the triple - pomeron term ( ppp )
is shown explicitly in ( [ eq : trf ] ) .
there are also contributions rrp , ppr , ... in which secondary reggeons r replace the upper and/or lower pomeron exchanges .
the rrp term may give some contribution as @xmath10 decreases away from 1 , and the ppr term may give some effect if the mass @xmath15 is not sufficiently large .
unfortunately , the determinations of the triple - pomeron coupling @xmath20 from data for the hadronic @xmath21 processes @xcite are , at best , just estimates .
the problem is that no account is taken of the soft rescattering . in such a rescattering
the leading hadron will produce new secondary particles and , as a consequence , its value of @xmath10 will be diminished .
nowadays this effect is called the rapidity gap survival factor @xmath8 , since the secondaries populate and destroy the rapidity gap , @xmath22 ln @xmath23 , between the leading hadron @xmath9 and the hadrons in the system @xmath3 .
since the @xmath8 was not accounted for explicitly , the determinations give only an _ effective _ value of the triple - pomeron vertex , which implicitly embodies an @xmath8 factor .
however the probability of soft rescattering , and hence the @xmath8 factor , depends on the nature of the incoming and leading hadrons .
moreover the value of @xmath8 depends on the kinematical variables of the process @xcite .
clearly it is important to find a way to measure the universal _ bare _ triple - pomeron vertex , free from these rescattering effects . , which is described by a diagram with a triple - pomeron vertex .
we take @xmath14 , which is always the case for the observed triple - pomeron processes.,height=188 ] a good illustration of the situation is an analogous problem which arises in the analysis of leading neutron production at hera @xcite . there , to understand the process we need to allow for rescattering ( absorptive ) corrections to pure regge @xmath24 exchange . the rescattering modifies the predictions of not only the absolute value , but also the @xmath7 dependence of leading neutron production .
quantitatively it reduces the cross section by a factor of 0.4 , and changes the @xmath7 slope , @xmath25 by 1 @xmath26 , where @xmath27 .
the @xmath7 dependence of the triple - pomeron vertex plays a crucial role in the solution of the so - called finkelstein - kajantie problem @xcite .
that is , if we neglect the @xmath7 dependence of the vertex and the gap survival factor , then it turns out , even in the case @xmath28 , that the cross section for multigap events grows as a power of the energy , and so violates the froissart limit .
each additional gap brings a ln@xmath29 factor arising from the integral over the gap size .
the sum of these ln@xmath29 factors leads to the power behaviour .
two solutions were proposed .
first , the _ weak coupling _ solution in which the vertex _ vanishes _ as @xmath30 @xcite . in this case the ln@xmath29 factor arising from the gap size is compensated by a 1/ln@xmath29 caused by the shrinkage of the @xmath7 distribution with increasing energy .
interestingly , it predicts that , at ultra high energies , the total cross sections @xmath31 will tend to a universal constant independent of @xmath13 and @xmath32 @xcite . an alternative possibility is to suppress the multigap events by a strong absorptive correction , that is by decreasing the gap survival factors .
in other words a large multigap cross section , calculated using a bare triple - pomeron vertex which is _ non - vanishing _ as @xmath30 , is multiplied by a gap survival factor which decreases faster with energy than the bare cross section .
this leads to the froissart - like black disc limit , and is called the
_ strong coupling _
solution @xcite .
the old hadron - hadron data of ( [ eq : hh ] ) show no indication that the triple - pomeron vertex vanishes as @xmath30 . however these data are inconclusive regarding the @xmath7 behaviour .
the reasons are as follows .
first , if it were to vanish , this behaviour would only be apparent at rather small @xmath7 , @xmath33 , where @xmath34 is the typical transverse size of the incoming hadrons .
unfortunately , the data are not precise enough in this small @xmath7 region .
second , the effect is masked by the pomeron cut contribution soft rescattering washes out the momentum transferred through the individual pomeron , and zero @xmath35 of the leading hadron does not mean that the momentum transfer in the bare triple - pomeron vertex is @xmath36 .
it has been shown @xcite that after accounting for multi - pomeron effects , the old hadron - hadron data can not eliminate the possibility that the bare triple - pomeron vertex vanishes as @xmath30 .
photoproduction , @xmath37 , which is described by a diagram with a triple - pomeron vertex in which the rescattering effects are small .
the dotted line would mean the diagram became an enhanced diagram this contribution is small .
, height=188 ] it is clearly important to study the triple - pomeron interaction in a process where the rescattering effects are absent or strongly suppressed .
such a process , which is experimentally accessible , is proton dissociation in diffractive @xmath38 photo- ( or electro- ) production + p j/ + y. [ eq;psi ] the process is shown schematically in fig .
[ fig:2 ] . by detecting the @xmath2 in the final state , we select the charm component of the photon wave function , which has a small size of @xmath39 .
such a component has a small absorptive cross section .
therefore the probability of an additional rescattering is much weaker . even in the case of a nuclear target
the probability of @xmath2 rescattering is small @xcite .
of course , there may be some ` enhanced ' absorptive corrections ( of the type indicated by the dotted pomeron - exchange line on the diagram on the right - hand - side of fig .
[ fig:2 ] ) .
however the phenomenological analysis of leading neutron data @xcite demonstrates that such corrections are small at the available hera energies .
otherwise the probability to observe a leading neutron would strongly depend on the initial @xmath40 energy , @xmath41 , in contradiction with the data .
moreover , to provide enough phase space for such an ` enhanced ' correction , we would need to have sufficiently large rapidity intervals ( either side of the dotted line ) between the different vertices @xcite .
this corresponds to an extremely small value of @xmath42 , which is beyond the reach of the present experiments .
another advantage of the process @xmath37 is that , thanks to the small size of the @xmath2 meson , the vanishing of the triple - pomeron vertex ( if true ) would reveal itself over a more extended region of @xmath7 .
this possibility is practically already eliminated by the present data , see fig .
4 of @xcite .
thus we can conclude that the strong coupling solution of the triple - pomeron interaction is confirmed by the hera @xmath2 data .
though the process @xmath37 is experimentally accessible at hera , so far no dedicated measurements are available in the literature .
however some information does exist , as the process has been considered as a background to elastic @xmath2 production , @xmath43 .
this somewhat fragmentary information allows us to make the following observations .
we use @xmath44 and @xmath45 to denote the two @xmath2 production cross sections . * the energy dependence of @xmath2 production with proton dissociation and for the elastic process are consistent with each other . if we write _
i w^_i where @xmath46 is the photon - proton centre - of - mass energy , then @xcite @xmath47 next , the @xmath15 dependence @xcite for a fixed value of the ratio @xmath48 is consistent with the behaviour of hadronic total cross sections , that is _ y ( m_y^2)^ with 0.08 . [
eq : tot ] the above results are the properties expected from the triple - pomeron term and @xmath12 in ( [ eq : trf ] ) , that is in the triple - pomeron diagram in fig .
[ fig:2 ] , are not the same .
the lower pomeron @xmath16 in fig .
[ fig:2 ] is the usual ` soft ' pomeron ; whereas the upper ones , with @xmath12 , include dglap evolution from a low initial scale @xmath49 up to a rather large scale @xmath50 at the @xmath38 production vertex .
the summation of the double logarithms @xmath51 leads to a steeper @xmath52-dependence and hence to a larger effective intercept for the trajectory @xmath12 of the upper ` hard ' pomeron .
thus in ( [ eq : el ] ) and ( [ eq : tot ] ) we have @xmath53 , where @xmath54 and @xmath55 . ] in ( [ eq : trf ] ) .
* the @xmath7 slope of @xmath2 production by proton dissociation is observed @xcite to be rather small , @xmath56 . in some sense , we can regard @xmath57 to be driven by the pomeron form factor . on the other hand ,
the slope for elastic @xmath2 production , which is driven by the proton form factor , is measured @xcite to be @xmath58 .
this indicates that the size of the pomeron and the triple - pomeron vertex are much less than that for the proton .
this conclusion about the small size of the triple - pomeron vertex is confirmed by the data on @xmath59 production , where the difference of slopes , @xmath60 @xcite .
these properties justify the idea that the pure triple - pomeron coupling may be extracted from a triple - regge analysis of data for @xmath2 production with proton dissociation .
the small size of the pomeron indicates that the rescattering corrections are indeed small .
this information comes from a consideration of proton dissociation as a background process in the presentation of elastic @xmath2 production data .
however there is more detailed information on the @xmath7 dependence of the proton dissociation process in fig . 4 of @xcite , that we mentioned previously . *
a comparison of the cross sections is also revealing . for @xmath2 production ,
an approximate estimate gives @xcite _ y / _ el 1 0.3 , while for @xmath59 production we have @xcite _ y / _ el 0.6 0.2 .
this difference in the values of the ratios is anticipated because strong absorption is expected in the case of @xmath44 for @xmath59 production .
the above summary of the existing , admittedly fragmentary , data on @xmath37 certainly supports a dominant triple - pomeron behaviour .
so we may attempt an extraction of the _ bare _ triple - pomeron coupling , @xmath20 .
we assume that the data for @xmath44 , collected @xcite in the interval .
] of @xmath15 from 2.5 gev up to @xmath61 , is described just by the triple - pomeron term in ( [ eq : trf ] )
. then dm_y^2 / = g^bare_3p(0)j / g_n(0 ) , where we estimate that the integration over @xmath15 gives the factor @xmath62 .
in this way we find g^bare_3p(0 ) / g_n(0 ) 1 / 3 .
[ eq : bare ] this has to be compared with the ratio extracted @xcite many years ago from hadron - hadron initiated data g^effective_3p(0 ) / g_n(0 ) 1 / 10 .
the distortion of the coupling in the old estimate , due to the survival factors , is clearly evident .
the fact that including absorptive effects gives a larger value of the triple - pomeron coupling , is not new .
larger values of @xmath63 were obtained , for example , in @xcite . however all the previous estimates are model dependent .
we can not sum up all the reggeon diagrams which describe the absorptive effects .
therefore the value of @xmath63 depends on the class of multi - pomeron diagrams chosen for resummation , on the assumptions about the behaviour of the multi - pomeron vertices , and sometimes on the threshold factor in the multi - pomeron vertices , etc .
the advantage of the present evaluation of @xmath63 is that we study a process where the absorptive corrections are small , and so affect the final result much less . of course
, in the integration over @xmath15 , we can not neglect the possible contributions of secondary reggeons in ( [ eq : trf ] ) , such as the rrp term near the upper limit of the integral and the ppr term for @xmath15 close to @xmath15(min ) .
the value of @xmath64 is therefore expected to be a bit smaller than in ( [ eq : bare ] ) .
it is thus crucially important to have @xmath37 data with an explicit measurement of the @xmath15 spectrum in order to perform a full triple - regge analysis in which we quantify the different triple - regge contributions .
the discussion in this paper has implications for all diffractive processes .
we mention two topical examples , both of which illustrate the importance of the energy dependence of the survival factor @xmath8 . in the existing monte carlos for high energy multiparticle production ,
the distributions of leading particles are parametrized in the triple - regge form . in comparison with the limiting fragmentation or feymann scaling hypothesis @xcite , where the normalized cross section does not depend on energy , we now have = ( 1-x_l)^_p(0)-2_p(t ) s_sd^2(s , t )
, [ eq : sss ] which depends , not only on @xmath10 , but on the energy ) does not allow for the effects of migration which , as we have seen in the case of the leading neutrons @xcite , give a noticeable contribution for @xmath65 .
therefore ( [ eq : sss ] ) is only valid in its present form for @xmath66 . also ( [ eq : sss ] )
contains just the ppp - contribution . for @xmath67 allowance
should be made for a rrp contribution with its own @xmath68 . ] .
the energy dependence comes entirely from the survival factor @xmath69 .
so , before we discuss the two examples , we compute the energy dependence of @xmath70 . as before , we use sd to denote single - particle diffractive dissociation . , for single ( soft and hard ) diffractive dissociation , via the triple - pomeron interaction .
the gzk energy , @xmath71 ev , is the theoretical upper ( greisen - zatsepin - kuzmin @xcite ) limit on the energy of cosmic rays which arises from the large cross section for the interaction of the protons with relic photons due to @xmath72 isobar production.,height=453 ] in fig .
[ fig:3 ] we show the expected energy dependence of the survival factor @xmath69 for the cross section integrated over @xmath7 . this was calculated in a simplified two - channel eikonal model with the @xmath7 dependence of the elastic pomeron - proton vertex , @xmath73 , given by the proton electromagnetic form factor @xmath74 , that is @xmath75 .
the parameters of the pomeron trajectory were tuned to describe the cern - isr , cern - sps and tevatron elastic data .
it is clear from the figure that the energy dependence of @xmath69 is quite pronounced .
moreover , we find that the predictions for @xmath69 obtained using a two - channel eikonal are stable decrease and the values at ultra high energies become too low and unstable . ] to changing the details of the model , after the ` soft ' parametrization is tuned to fit the elastic data .
the continuous curve in fig . [ fig:3 ] is for _ soft _ diffractive dissociation . here
the process takes place at relatively large impact parameters @xmath76 , where the absorption is not so strong .
suppose , instead , we were to consider _ hard _ diffractive dissociation , for example where a pair of high @xmath77 jets @xcite or a heavy quark pair or @xmath46 or @xmath78 boson or some other heavy object is produced within the system @xmath3 .
in these cases the interaction takes place at much smaller @xmath76 . here , in the central @xmath76 region
, the absorption is stronger and the corresponding @xmath69 factor is smaller , especially at ultra high energies where the proton opacity becomes very close to the black disc limit .
therefore , for completeness , we also show the predictions for the energy dependence of @xmath69 for hard diffractive dissociation , by the dashed curve in fig .
[ fig:3 ] .
our first topical example , where the energy dependence of @xmath69 plays an important role , is the evaluation of the so - called pile - up backgrounds to forward physics studies at the lhc , see , for instance , @xcite . at high lhc luminosities , the soft pile - up single ( and even double )
diffractive processes can fake the signal from hard diffractive events due to overlap with the non - diffractive hard processes observed in the central detector
. however , when we allow for the energy dependence of @xmath69 ( for soft diffraction ) , we decrease the predicted leading proton spectra at the lhc energies , as compared to the leading proton spectrum measured at hera @xcite . a preliminary estimate found @xcite for leading neutrons at hera . for leading protons the value of @xmath8 may be a bit smaller , since a pomeron - exchange interaction has smaller @xmath76 than for @xmath24-exchange leading neutron production . to obtain a quantitative estimate of the reduction factor it will be necessary to perform a careful analysis of the leading protons produced in ddis @xcite . ]
gives a reduction by about a factor of 3 .
therefore the impact of pile - up is be suppressed by the energy dependence of @xmath69 .
the second example illustrates the importance of the energy dependence of @xmath69 to the analysis of ultra high energy cosmic ray data .
the observed structure of the extensive cosmic ray air showers depends sensitively on the spectrum of the leading hadrons . from the continuous curve in fig .
[ fig:3 ] we see that in going from hera energies to the gzk cut - off energy @xcite , the survival probability @xmath69 decreases by almost an order of magnitude . of course
, the small values of @xmath69 at ultra high energies do not mean that the leading hadron disappears . instead ,
after the rescattering , the large @xmath10 hadron migrates to a lower @xmath10 and larger @xmath35 domain @xcite , thereby changing the shape of the extensive air shower . |
this work focuses on stability conditions and their possible relations to different fermi surface topologies in a superconductor with unequal number densities of fermions ( or unequal chemical potentials)@xcite .
the physics of paired fermion systems with unequal densities of the two fermion species is of interest in the study of two physical systems : ( 1 ) ultracold atomic fermionic gases , where one can freely choose populations in two hyperfine states of the fermionic atom .
experimental work is currently being conducted with these systems .
( 2 ) quark matter in the interior of neutron stars , which is believed to be a color superconductor .
there , the mismatch in quark fermi surfaces is driven by differences in quark masses and electric and color neutrality conditions . in both cases , the ground state of such
a system is the subject of current debate .
the study of fermionic superfluids with imbalanced populations is a novel subject , because until recently , experimental evidence for this situation was elusive . in a superconductor ,
an imbalance in population between spin - up and spin - down electrons can be created by a magnetic field .
however , due to the meissner effect , the magnetic field is either completely shielded from the superconductor bulk or enters in the form of quantized flux lines or vortices .
cold fermionic atomic gases provide new possibilities in the experimental exploration of fermionic superfluids with unequal mixtures of fermions . in cold atoms ,
feshbach resonance permits tuning of the interaction from ( 1 ) attraction and a resulting superfluid of loosely bound pairs ( the bcs regime at @xmath1 ) ; to ( 2 ) repulsion and a resulting bose - einstein condensate of tightly bound molecules ( the bec regime at @xmath2 ) , where @xmath3 is the size of a pair and @xmath4 is the interparticle separation ( or , more precisely , the mean free path ) . with equal mixtures , the bec bcs crossover is smooth , with no phase transition . with asymmetric densities ,
one or more phase transitions are expected , and a more complex phase diagram may result . in the weak coupling regime , a bcs superfluid remains stable as long as the difference in chemical potential is small compared to the pairing gap , @xmath5 ; the gap prevents the excess unpaired atoms from entering the superfluid state . by either increasing the mismatch or reducing the binding energy ( and hence decreasing the gap ) , a quantum phase transition from the superfluid to normal state takes place , and superflidity ceases .
the point at which the phase transition occurs is known as the clogston limit , which can be estimated as @xmath6 , where @xmath7 is the fermi energy .
thus only an exponentially small population imbalance is allowed in weak coupling . in strong coupling , as one approaches feshbach resonance , the situation is quite different . on the bec side , superfluidity with imbalanced population
is robust over a wide range of parameter space .
surprisingly , on the bcs side , macroscopic imbalance is also possible in a superfluid state , possibly due to the formation of a gapless superfluid at strong coupling .
this gapless superfluid incorporates large numbers of unpaired fermions which reside at the secondary fermi surfaces .
here , we consider stability conditions for gapless states with different fermi surface topologies .
we consider two species of nonrelativistic fermions , @xmath8 with the same mass but with different chemical potentials , @xmath9 .
there is an attractive interaction only between different species of fermions , @xmath10 .
the order parameter is @xmath11 where @xmath12 , it defines pairing in the singlet channel , spin @xmath13 , @xmath14\times [ 2]=[1]+[3]$ ] .
the order parameter breaks the original group associated with conservation of particle numbers of each species down to the diagonal subgroup associated with conservation of difference in particle numbers , @xmath15 which is invariant under simultaneous rotation @xmath16 .
total number of particles is not conserved .
formally , one can express this pattern of symmetry breaking by gauging the theory .
we introduce two external gauge fields , and couple each species of fermions to its gauge field , @xmath17 couples to @xmath18 with @xmath19 and @xmath20 couples to @xmath21 with @xmath22 , that is reflected by the generators of the gauge group , @xmath23 and @xmath24 .
according to the symmetry breaking pattern , gauge fields mix in a superconductor , rotated sum of fields @xmath25 is screened with non - zero meissner mass via andersen - higgs mechanism , @xmath26 corresponding to @xmath27 , and rotated difference of fields @xmath28 propagates in a superconductor unscreened , meissner mass is zero @xmath29 corresponding to @xmath30 .
there is a different mixing of gauge fields in case of debye masses . in qcd , with color and electromagnetic gauge groups , @xmath31 , mixing of gauge fields
is more complicated . we treat four - fermion interaction on a mean field level .
we integrate over the fermions , obtain fermion determinant , @xmath32\ , .
\label{part}\ ] ] we use the nambu - gorkov formalism with particle - hole basis , @xmath33 . inverse fermion propagator and gauge field are @xmath34 matrices in this basis , including the nambu - gorkov and two fermion species indices , @xmath35^{-1 } & \phi^- \\ \phi^+ & [ g_0 ^ -]^{-1 } \end{array}\right ) \ , , \ ] ] where the inverse free fermion propagators are @xmath36^{-1 } = i\partial_t \pm \frac{\nabla^2}{2 m } \pm \mu$ ] , and we have abbreviated for the gauge fields @xmath37 with @xmath38 and @xmath39 . performing derivative expansion and collecting terms quadratic in the gauge field , we produce fermion loops , @xmath40 , @xmath41 , @xmath42 , @xmath43 .
debye mass in one - loop is defined by the temporal component of polarization operator , meissner mass is given by the spatial component , @xmath44 and @xmath45 , where @xmath46 .
screening masses are given@xcite @xmath47 \ , , \nonumber\\ m_{m , ab}^2&= & \frac{1}{2m}\lim_{p\to 0}\frac{t}{v}\sum_k \left(\phantom{\frac{k^2}{m}}\hspace{-0.5 cm } \delta_{ab}{\rm tr}[s(k)\bar{\gamma}_a^2]\right . \label{screening } \\ & + & \left.\frac{k^2}{2m}\,[1-(\hat{p}\cdot\hat{k})^2]\,{\rm tr}[s(k)\gamma_a^+s(k - p)\gamma_b^+]\right)\ , , \nonumber % \label{screening}\end{aligned}\ ] ] where @xmath48 is the fermion propagator and we have introduced the following matrices in nambu - gorkov space , @xmath49 and @xmath50 . physically , debye mass to all loops , including ladder diagrams , is equivalent to compressibility of a system , and meissner mass can be associated with the density of superconducting fermions .
these @xmath51 matrices in two - fermion space shall be evaluated in the following section in order to obtain stability conditions for gapless superconductors .
one may derive the pressure from the partition function in ( [ part ] ) using the cornwall - jackiw - tomboulis formalism .
the pressure is the negative of the effective potential at its stationary point ( i.e. , with the propagators determined to extremize the effective potential ) .
the fermionic part of the pressure is @xmath52 + \gamma_2[{\cal s}]$ ] , where @xmath53 is the tree - level fermion propagator in nambu - gorkov space and @xmath54 $ ] is the sum of all two - particle irreducible diagrams .
the number densities is defined @xmath55 .
the number susceptibility @xmath56 is defined as the derivative of the number density with respect to the chemical potential ( at constant volume and temperature ) . using the expression for the pressure , we obtain@xcite @xmath57 \nonumber\\ & -&\,\frac{1}{2g_a}\frac{t}{v}\sum_k{\rm tr}\left[\gamma_a^-{\cal s}(k)\frac{\partial \sigma(k ) } { \partial\mu_b}{\cal s}(k)\right ] \
, , \end{aligned}\ ] ] where @xmath58 is the fermion self - energy , @xmath59 . the first term on the right - hand side of this equation is given by the one - loop result for the electric screening mass , cf . eq . ( [ screening ] ) . for the second term , we assume that the self - energy @xmath58 depends on @xmath7 only through the gap , then we obtain@xcite @xmath60 in general , the self - energy @xmath58 contains terms of any number of fermion loops .
consequently , the number susceptibility contains terms of arbitrary many fermion loops too , corresponding to the exact debye mass including all possible perturbative insertions .
remarkably , the free fermion result for @xmath56 , i.e. @xmath61 , gives the one - loop result for @xmath62 .
susceptibility is equivalent to compressibility of a system .
we shall use eq .
( [ chidef ] ) in the following section to compute the number susceptibility .
as this equation shows , it goes beyond the one - loop result for the electric screening mass .
we consider three cases distinguished by how many zeros quasiparticle dispersion has , @xmath63 , where @xmath7 is the average chemical potential and @xmath64 is the difference between potentials , fig .
[ figoccupation ] . in this case ,
number of zeros is equivalent to number of the fermi surfaces .
no zeros corresponds to the fully gapped state . in case of one fermi surface ,
momenta outside the fermi surface contribute to the pairing , while the excess of fermions resides inside the fermi ball . in case of two fermi surfaces
, the excess of fermions resides between the two fermi surfaces in momentum space .
as was noticed by son , apart from number of zeros , dispersion can have two different characteristic behaviors , distinguished by the position of the minimum .
minimum is located at nonzero momentum for positive @xmath65 , and corresponds to bcs , and minimum shifts to @xmath66 for negative @xmath7 , and corresponds to bec .
this behavior manifests itself in stability conditions .
we depict different topologies which are distinguished by the number of fermi surfaces on the phase diagram in dimensionless average chemical potential @xmath67 and difference in chemical potentials @xmath68 , the gap is the energy scale , in fig .
[ figglobal ] regions between the solid lines .
we have fully gapped @xmath69 , and gapless states with one @xmath70 and two @xmath71 fermi surfaces . at small mismatch
, there is a fully gapped state @xmath69 , which is at positive @xmath65 and @xmath5 , bcs , and at small and negative @xmath65 , @xmath72 , bec .
increasing mismatch , when mismatch is at least larger than the gap , there is a gapless state with one fermi surface @xmath70 when @xmath7 is around the feshbach resonance , @xmath73 , and gapless state with two fermi surfaces @xmath71 for positive @xmath7 restricted from below , @xmath74 .
we therefore expect that @xmath70 exists at strong coupling , while @xmath71 probably exists only at weak coupling .
the question we are solving here is , `` what is the ground state in a degenerate fermi system with asymmetric number densities of fermions ? '' .
we avoid solving for the ground state explicitly .
instead , we check stability criteria , positive definite eigenvalues of screening masses and number susceptibility .
if stability conditions are satisfied , homogeneous superconducting gapless state is indeed the ground state in this parameter range .
if stability conditions are not satisfied , the alternative state may be realized as the ground state .
these include loff , spatially separated mixtures , normal ( not superconducting ) states .
we analyze stability conditions , eq .
( [ screening ] ) and eq .
( [ chidef ] ) , in all parameter space @xmath75 , and depict stable / unstable regions together with topology regions , fig .
[ figglobal ] left panel . debye mass e.v .
are positive in all parameter space , hence debye mass does not impose a constraint .
all entries for the meissner mass matrix are the same , it is trivial to diagonalize .
we obtain@xcite @xmath29 corresponding to the unbroken sector @xmath30 , and @xmath76 corresponding to the broken group @xmath27 , where @xmath77 , which defines the dashed - dotted ( blue online ) curve in fig .
[ figglobal ] left panel .
it renders all states between the dashed - dotted curve and the solid vertical line unstable .
there is a strip left in gapless superconductor state with two fermi surfaces @xmath71 which is stable , gray area . @xmath69 and @xmath70 are stable with respect to @xmath78 everywhere . in magnetic sector , mixing does not depend on chemical potentials and it is defined by the pattern of symmetry breaking . in electric sector , mixing depends on chemical potentials , e.g. in qcd mixing depends on @xmath64 . mixing in electric and magnetic sectors is the same only for the fully gapped case . analyzing the number susceptibility matrix , eq .
( [ chidef ] ) , we obtain@xcite the expression defining the sign of e.v
. , which is very similar in structure to that for the meissner mass , see expression for @xmath79 , @xmath80 . here , we defined @xmath81 and @xmath82 with elliptic integrals @xmath83^{3/2}$ ] and @xmath84^{3/2}$ ] , and @xmath85 are zeros of @xmath86 with @xmath87 , @xmath88 . in expression for @xmath89 one should put @xmath90 when applied to @xmath70 region . it defines the dashed ( red online ) line , and it renders all states between dashed and solid vertical lines unstable .
thus all @xmath71 states are unstable , and there is a strip in @xmath70 which is stable , dark gray area .
@xmath69 is stable everywhere .
stable states with respect to @xmath56 correspond to local maximum of pressure .
we obtained stable regions which are local maxima of pressure , fig .
[ figglobal ] left panel .
now we consider global maxima , fig .
[ figglobal ] right panel .
for this we compare pressure of the superconducting and normal states .
superconducting states which pressure is higher than that of the normal state are stable , @xmath91 .
the dashed - dotted ( blue on line ) line is @xmath92 , it renders all states above and to the right of it unstable .
all unstable regions with respect to the screening masses and number susceptibility are subset of unstable region with respect to the pressure . in a weak coupling , the vertical dashed - dotted line reproduces the known clogston limit , @xmath93 , above which bcs is unstable .
the global stability line cuts through the stable strip of @xmath70 state , below is a stable superconducting state , above is a metastable state .
both lines coincide at large mismatches .
currently , experiments are being performed with unequal mixtures of fermions to map superfluid regions as a function of population imbalance , interaction strength and temperature .
the experimental signature of superfluidity is the existence of vortices , which prove phase coherence in a sample . | gapless superconductivity can arise when pairing occurs between fermion species with different fermi surface sizes , provided there is a sufficiently large mismatch between fermi surfaces and/or at sufficiently large coupling constant . in gapless states ,
secondary fermi surfaces appear where quasiparticle excitation energy vanishes .
this work focuses on homogeneous and isotropic superfluids in the @xmath0-wave channel , with either zero ( conventional superconductor ) , one , or two spherical fermi surfaces .
the stability conditions for these candidate phases are analyzed .
it is found that gapless states with one fermi surface are stable in the bec region , while gapless states with two fermi surfaces are unstable in all parameter space .
the results can be applied to ultracold fermionic atom systems . |
low mass x - ray binaries ( lmxbs ) are close binary systems consisting of a degenerate primary black hole or neutron star ( ns ) accreting matter from a low mass ( @xmath7 1 m@xmath4 ) secondary , via roche lobe overflow .
x - ray transients ( xrt ) are lmxbs which undergo periodic outbursts , explained by the disc instability model ( e.g. * ? ? ?
* ) , while persistently bright systems are permanently in the outbursting state .
one subclass of lmxbs are the accretion - powered millisecond pulsars ( amsps ) , the first of which , sax j1808.4@xmath03658 , was discovered in 1998 @xcite .
seven more of these systems have been discovered since then @xcite . in general , the outburst duration for amsps ranges from a few weeks to months .
amsps generally contain a weakly magnetized ( @xmath610@xmath810@xmath9 g ) ns with spin frequencies between 180 and 600 hz .
the orbital periods range between 40 min and 5 h. the secondary star in these systems is either a white or brown dwarf ( * ? ? ?
* and references therein ) .
@xcite have measured the predicted decrease in the ns spin period for the amsp system igr j00291@xmath105934 , supporting the idea that these amsps are in fact old ns , which have over time been spun up to millisecond periods by acquiring angular momentum from the accretion of material from the secondary . as such ,
they provide the missing link between lmxbs and old , isolated millisecond radio pulsars . in comparison to brighter lmxbs , relatively little
is known about the optical properties of amsps . even in outburst ,
the eight so far discovered are comparatively faint , the brightest being sax j1808.4@xmath03658 at @xmath1 @xmath11 16.2 mag @xcite . in quiescence , these systems are generally very faint optically , the brightest measured again being sax j1808.4@xmath03658 at @xmath1 @xmath11 20.9 mag .
hete j1900.1@xmath02455 is the seventh amsp , discovered by @xcite on 2005 june 14 by the _ high energy transient explorer ii _
( _ hete ii _ ) .
several type - i bursts have since been observed , and assuming that the peak flux observed during the brightest of these is eddington - limited , hete j1900.1@xmath02455 is at a distance of @xmath64.3 kpc @xcite .
the pulsation frequency was determined with the _ rossi x - ray transient explorer _ to be 377.3 hz @xcite .
pulse timing analysis by @xcite revealed a circular orbit with a period ( @xmath12 ) of @xmath683.3 minutes ( 4995.258 @xmath13 0.005 s ) , and a projected primary semimajor axis , @xmath14 , of 18.41 @xmath13 0.01 @xmath15 10@xmath16 light second .
these parameters combine to provide a mass function @xmath17 .
@xcite report that pulses were detected intermittently in the first two months after discovery .
no pulses have been reported since then .
unusually for an xrt , this system has remained active for more than 2 years , at a level of @xmath62 @xmath15 10@xmath18 erg s@xmath3 . several times during 2007 march / february , the x - ray flux fell by an order of magnitude , but each time returned to outburst levels within @xmath61 week @xcite . in 2007
may , the source was observed to decline to its lowest level since discovery ( by a factor of @xmath19 10@xmath20 ) , lasting for several weeks , prompting speculation that hete j1900.1@xmath02455 was returning to quiescence @xcite .
however , the system returned to its outburst state within @xmath623 weeks @xcite .
the optical counterpart was found by @xcite with an @xmath1-band magnitude of @xmath618.4 mag .
subsequent observations by @xcite found the @xmath1-band magnitude to be 18.02 @xmath13 0.03 mag , and the @xmath21 colour ( dereddened ) to be @xmath22 mag , with spectroscopy revealing a broad he@xmath23ii @xmath244686 emission line .
in this paper we present optical photometry and spectroscopy of hete j1900.1@xmath02455 in its outburst state .
these observations were motivated by several goals , key amongst them the aim of measuring the radial velocity of the secondary , and hence constraining the mass of the compact object . for the longer period transient systems , such measurements are possible via radial velocity studies of absorption lines from the secondary once the system has returned to quiescence . for the shorter period amsps
however , the secondary is likely to be so faint as to make such observations extremely difficult , if not impossible .
however , @xcite and @xcite have shown how radial velocity measurements of the secondary are still possible for x - ray bright systems , by studying the bowen blend emission from the irradiated face of the secondary : this technique may offer the only way of measuring the velocity of the secondary in amsps ( i.e. whilst they are still x - ray bright ) , and usefully constraining the mass of the neutron star .
our data consist of photometry obtained with the 2.3 m advanced technology telescope ( att ) at siding spring observatory , australia , the 3.5 m wisconsin - indiana - yale - noao ( wiyn ) telescope at kitt peak , arizona , and the 8.4 m large binocular telescope ( lbt ) at mount graham , arizona .
our spectroscopy was obtained using the high resolution chelle spectrometer ( hires ) on the 10 m keck i telescope at mauna kea , hawaii .
we obtained photometry on 2 nights in 2006 september with the 2.3 m att , using the ccd4240 e2v detector , mounted at the @xmath25 nasmyth focus .
the detector has 2148 @xmath15 2148 13.5 @xmath26 m pixels , with 0.34 pixel@xmath3 . on 2006 september 21
, we obtained 16 @xmath1-band exposures , covering slightly more than 1 orbital period , with exposure times of 60 s ( @xmath27 ) , 300 s ( @xmath28 ) and 240 s ( @xmath29 ) . on the following night , we obtained @xmath1 , @xmath30 and @xmath31-band exposures , 2 @xmath15 300 s exposures of each . on both nights , exposures were taken of the photometric standard field sa-112 @xcite .
seeing was @xmath62 on both nights , but deteriorated towards the end of the second night .
three 300 s @xmath1-band exposures were also taken with the 3.5 m wiyn telescope on 2006 september 19 , using the wiyn mini - mosaic imager , consisting of 2 site ccds , each with 4096 @xmath15 2048 15 @xmath26 m pixels , with 0.14 pixel@xmath3 .
the seeing was @xmath62 for all 3 exposures .
exposures were also taken of the standard star field pg031@xmath10051 @xcite .
after the decline / rebrightening episode of 2007 may / june , we obtained additional observations using the lbt . on 2007
june 13 and june 24 we obtained 200 s sloan @xmath32-band exposures of hete j1900.1@xmath02455 with the large binocular camera ( blue ) ( giallongo et al .
, in preparation ; @xcite ) at the prime focus of the lbt ( 20 exposures per night ) .
the eev - blue detector consists of 4 ccds with 2048 @xmath15 4608 13.5 @xmath26 m pixels , with 0.@xmath33224 pixel@xmath3 . the seeing on 2007 june 13 was @xmath60.851.15 , while on june 24 was @xmath611.5 .
the hete j1900.1@xmath02455 field is shown in fig . [ 1900 ] ( from lbt on 2007 june 13 , with seeing @xmath60.85 ) .
2455 ( sloan @xmath32-band ) , 2.3 @xmath15 1.6 .
the black arrow indicates the position of hete j1900.1@xmath02455 . ]
all the photometry data were reduced in the same way .
the frames were firstly bias corrected , trimmed and flat - fielded using the routines in .
photometry was performed using the @xcite point spread function fitting package in .
the magnitude of hete j1900.1@xmath02455 was measured relative to several bright stars .
these relative magnitudes were then calibrated by comparison with the standard star frames .
note that the wiyn data was used for photometric calibration only .
the att data from 2006 september gives @xmath34 mag , @xmath35 mag and @xmath36 mag .
the errors quoted are an estimate of the systematic uncertainty .
we note that the measured @xmath1-band magnitude is consistent with that reported by @xcite .
on 2007 june 13 , @xmath37 mag , while on 2007 june 24 , @xmath38 mag . taking the value for the hydrogen column density to the source , @xmath39 , as 1.6 @xmath13 0.4 @xmath15 @xmath40 @xmath41 @xcite , and using the method detailed by @xcite yields @xmath42 .
the resulting dereddened colours are ( @xmath43)@xmath44 @xmath13 0.08 mag , and ( @xmath21)@xmath45 @xmath13 0.06 mag ( again , consistent with the value reported by * ? ? ?
* ) . in the @xmath1-band
, hete j1900.1@xmath02455 is blended with a faint star 2 to the northwest ( see fig . [ 1900 ] ) .
this star was fitted simultaneously with hete j1900.1@xmath02455 , and was found to have an @xmath1-band magnitude of 20.1 @xmath13 0.1 mag , which was constant , within these errors , in all our observations .
this star was not visible in the @xmath31 or @xmath30-band frames .
this is consistent with the observations of @xcite , who note that this blended star is @xmath62 magnitudes fainter ( in the @xmath1-band ) than hete j1900.1@xmath02455 . in the lbt data , in frames with good seeing ,
an additional faint star is observed @xmath61 northwest of hete j1900.1@xmath02455 , with an @xmath1-band magnitude @xmath65 mag fainter than hete j1900.1@xmath02455 . in fig .
[ lc ] we plot the three @xmath1-band light curves , folded on the pulse timing ephemeris of @xcite .
the phase error at the epochs of these optical observations , extrapolated from the the error in the period of the pulse timing ephemeris , is @xmath60.01 .
each light curve exhibits a similar modulation , but that of 2007 june 24 has an amplitude @xmath63 times greater than for the other two .
clearly , the phasing of the three light curves is not consistent , implying that the period of this modulation is different to the orbital period .
our spectroscopy was obtained using the hires spectrometer @xcite at the right nasmyth focus of the 10 m keck i telescope , with a 52.68 lines mm@xmath3 chelle grism and the hiresr 250 lines mm@xmath3 long wavelength cross disperser .
the slit width was 0.86 with a slit length of 7 .
the chelle grating angle was 0@xmath46 and the cross disperser grating angle was 0.2405 , providing a wavelength coverage of @xmath4739008500 , spread over a mosaic of 3 mit - lincoln labs ccds ( `` red '' , `` green '' and `` blue '' ) , each with 2048 @xmath15 4096 15 @xmath26 m pixels , with 0.12 pixel@xmath3 in the spatial dimension .
the images were binned by 2 pixels in the spatial dimension .
15 chelle spectra were obtained with hires on 2006 october 11 .
appropriate bias , flat , trace star , thar arclamp and flux standard ( g191b2b ) exposures were also taken .
the object exposure times were 300 s ( @xmath48 ) , 360 s ( @xmath49 ) and 420 s ( @xmath50 ) , with the 15 spectra covering slightly more than one orbital period . the spectra were extracted using the ( mauna kea chelle extraction ) hires reduction package , which takes the two dimensional raw images , and produces optimally extracted , wavelength calibrated , one dimensional spectra .
the lines of interest to us were on the `` blue '' and `` green '' ccds .
the wavelength calibration was performed using the thar arc lamp exposures , by fitting a sixth order polynomial to an average of 30 lines per chelle order ( minimum 20 ) ; this resulted in a dispersion of 0.0150.025 pixel@xmath3 for the `` blue '' ccd and 0.0210.032 pixel@xmath3 for the `` green '' ccd , and an rms scatter of @xmath7 0.003 ( average @xmath60.002 ) . to confirm the extraction method implemented in
, the spectra were also extracted using the softwarexavier / hiredux/ ] , and the package in : however the spectra extracted using all three methods were very similar . the main emission features present in our spectra are h@xmath2 @xmath246563 , he@xmath23ii @xmath244686 and the bowen blend of n@xmath23iii and c@xmath23iii lines centred at @xmath6@xmath244640 .
h@xmath51 is present as a weak emission line within a broader absorption feature .
the chelle orders containing these features were exported for further processing with the spectroscopic data analysis package , and , a package used for performing maximum entropy doppler tomography .
the spectra were re - binned onto a velocity scale of 40 km s@xmath3 pixel@xmath3 , yielding a signal - to - noise ( s / n ) ratio of 616 pixel@xmath3 .
the s / n varied across each order , particularly on the `` blue '' ccd , with the lowest s / n near the edges of the orders . for each spectrum , the continuum was fitted with a 2nd order polynomial , after masking the emission regions .
the fit was then divided into the spectrum , and the remaining continuum subtracted .
the average processed spectrum for h@xmath2 is shown in fig .
[ ha_avg ] , and that for the bowen blend and he@xmath23ii @xmath244686 is in fig .
[ bowen_heii_avg ] .
note that the data in fig .
[ bowen_heii_avg ] has been boxcar smoothed , and shifted to the rest frame of the system . from the averaged spectra , the equivalent width for h@xmath2 is @xmath52 @xmath13 0.3 , for he@xmath23ii @xmath244686 is @xmath53 @xmath13 0.3 , and for the bowen blend is @xmath54 @xmath13 0.2 .
this is in contrast to the observations of @xcite , who report that although their spectroscopy was contaminated by a nearby star , the main feature observed was a broad emission line from he@xmath23ii @xmath244686 and balmer emission was at most very weak . , normalized to the local continuum . ]
ii @xmath244686 , normalized to the local continuum .
the spectrum has been shifted to the rest frame of the system , based on the systemic velocity determined from the h@xmath2 emission line in section 3.2 .
the spectrum has also been boxcar smoothed . ]
the three light curves in fig . [ lc ] exhibit a sinusoidal modulation , classically attributed to x - ray irradiation of the inner face of the secondary , where the modulation is due to the phase dependent visibility of this higher temperature region .
the phasing of the att light curve is approximately consistent with this scenario , with the brightest measurements occurring near phase 0.5 .
however , the phasing of the other light curves is completely different , suggesting that the observed modulation does not vary on the orbital period . in particular ,
the two lbt light curves are separated by only 11 nights , and in that time the phasing of the light curves , folded on the orbital period , differs by @xmath60.2 .
[ ha_avg ] shows the average h@xmath2 spectrum .
this broad emission line exhibits a hint of a double peak ( a classic signature of an accretion disc ) , although the peaks are not as distinctive as those seen in the spectra of other lmxbs .
the distribution of the flux appears to be symmetrical about a central wavelength , suggesting that , on average , the emission from the disc is axisymmetrical . to constrain the systemic velocity , @xmath55
, we fitted a gaussian to the wings of the averaged spectrum , where the emission is from disc regions closer to the centre , and presumably less susceptible to corruption from the hotspot and other non - uniformities in the outer disc @xcite .
this gives a value for @xmath55 of 132 @xmath13 8 km s@xmath3 .
we investigated the emission line distribution of the system by examining a doppler tomogram of the h@xmath2 line ( fig .
[ ha ] ) , constructed using @xmath56 km s@xmath3 . as with the photometry , the spectra were phased using the pulse timing ephemeris . for this work
, we used the maximum entropy method , implemented in : see @xcite , @xcite and @xcite for further details and applications of doppler tomography .
enhanced emission is present in the ( @xmath57,@xmath58 ) quadrant , extending into the ( -@xmath59,-@xmath60 ) quadrant , with a faint ring of emission .
such a ring of emission is seen in doppler tomograms of many lmxbs , and is attributed to emission from the outer disc @xcite .
this emission ring lies between radii of @xmath6200400 km s@xmath3 . plotting only the axisymmetric component of this tomogram reveals that the brightest emission is from an annulus at a velocity of @xmath6230 km s@xmath3 , but with significant emission extending to @xmath6150 km s@xmath3 .
since the secondary must be moving slower than the outer disc , we choose 200 km s@xmath3 as an upper limit to @xmath61 , the velocity of the secondary projected onto the line of sight . for illustration , we plot in fig .
[ ha ] the roche lobe of the secondary , the gas stream trajectory and accretion disc velocity curve , computed using @xmath62 km s@xmath3 . since @xmath63 is already accurately known from the pulse timing analysis ( @xmath64 km s@xmath3 ; * ? ? ?
* ) , @xmath65 ( @xmath66)@xmath67 .
@xmath61 and @xmath65 are the only parameters upon which the locations of the roche lobe and the gas stream trajectory depend . at this value of @xmath61
, the predicted gas stream misses the hotspot location .
lower values of @xmath61 would cause the gas stream to be closer to the hotspot .
however , we also note that doppler tomograms of the he@xmath23ii @xmath244686 emission line in the two ns lmxbs v801 ara and v926 sco also show the peak of emission in the ( @xmath68,@xmath69 ) quadrant , close to the @xmath68 axis , and for these systems this emission is attributed to an extended disc bulge @xcite .
as will be discussed in section 4 , the location of the hotspot in hete j1900.1@xmath02455 may be explained by a warped , irradiated accretion disc . .
the roche lobe of the secondary is plotted .
the lower curve represents the gas stream velocity ( the free - fall ballistic trajectory from the inner lagrangian point , @xmath70 ) and the upper curve is the keplerian velocity of the accretion disc along the gas stream .
both curves are marked with a circle when the gas stream moves 0.1r@xmath71 ( 0.1 of the distance between the primary and @xmath70 ) closer to the primary .
the gas stream trajectory and disc velocity along the stream were calculated using @xmath62 km s@xmath3 and @xmath72 . ] using the standard mass function equation , and with @xmath73 it is possible to calculate @xmath65 and @xmath61 for various orbital inclinations ( @xmath74 ) and primary masses ( @xmath75 ) .
[ kvi ] shows a plot of @xmath61 vs. @xmath74 for a primary mass of 1.4 m@xmath4 . if @xmath61 @xmath7 200 km s@xmath3 , @xmath74 @xmath5 20 , for all @xmath75 @xmath19 1.4 m@xmath4 . ) vs. orbital inclination angle ( @xmath74 ) for a primary mass of 1.4 m@xmath4 . the upper limit on @xmath61
is set by the velocity of the h@xmath2 emission ring . ]
hence our upper limit on @xmath61 provides an upper limit on @xmath74 of @xmath620@xmath46 .
this low inclination is consistent with the fact that the double peak in h@xmath2 is not as clear as in systems with higher inclinations , and thus higher projected disc velocities ( see for example the lmxb xte j1118@xmath10480 , with @xmath74 @xmath670 ; * ? ? ?
@xcite discovered that emission from the bowen blend ( a blend of n@xmath23iii , c@xmath23iii and o@xmath23ii lines , centred at @xmath6@xmath244640 ) traced the motion of the irradiated secondary star in the lmxb sco x-1 . some of the major features seen in the bowen blend in x - ray binaries are the n@xmath23iii lines at @xmath244634.13 , @xmath244640.64 and @xmath764641.85/4641.96 , and the c@xmath23iii lines at @xmath244647.4 , @xmath244650.1 and @xmath244651 @xcite .
our average spectrum of the bowen blend reveals a broad range of emission between @xmath244620 and @xmath244650 ( fig .
[ bowen_heii_avg ] ) . superimposed on this broad emission
, there appear to be several narrower emission features , three of which correspond to the expected wavelengths of the n@xmath23iii @xmath244634.13 line , the n@xmath23iii blend near @xmath244641 , and the @xmath244647.4 c@xmath23iii line . because of the low s / n , we performed doppler tomography of all of the lines in the bowen blend simultaneously , with the intention of finding emission from the irradiated face of the secondary .
although this map contains a single sharp spot of emission , it is in the ( @xmath77,@xmath78 ) quadrant , @xmath645 away from the expected location of secondary emission on the @xmath78 axis .
the significance of this emission is uncertain because of the low s / n of the individual exposures .
the he@xmath23ii @xmath244686 emission is contaminated in the wings , particularly on the red side .
while the lower s / n in the wings might explain some of this variability , is does not account for all of the systematic artefacts seen in particular , the 2 dips on the red side , near @xmath244704 and @xmath244710 , can be seen in averages of just a few spectra , and so are not due to random noise . as well as this , the average line appears asymmetrical , with an excess of flux between @xmath7646954705 .
the s / n ratio throughout the order is also lower than that for h@xmath2 . despite this , a gaussian fit to the wings of this line ( masking the line core , and the excess on the red side ) gives a systemic velocity consistent with that seen in the h@xmath2 emission .
with its maximum brightness near phase 0.5 , we attempted to fit the first of our light curves ( fig .
[ lc](a ) ) using the eclipsing light curve ( ) code @xcite , under the assumption that the modulation was due to x - ray heating .
this provided a satisfactory fit , although because of the degeneracy between inclination angle , disc temperature and disc flaring angle , it was not possible to extract a unique parameter set , and a wide range of parameter values yielded equally good fits .
however , the phasing of the subsequent light curves ( fig .
[ lc](b),(c ) ) suggests that the observed modulation is not due to x - ray heating of the secondary : the most likely explanation for this is that the observed modulation is due to superhumps ( see below ) .
therefore the apparent alignment of the first light curve is most likely coincidental , and the elc code ( which does not incorporate superhumps ) can not be used to extract reliable system parameters .
superhumps are periodic optical modulations , originally seen during superoutbursts of su uma dwarf novae ( dne ) , but also seen in some lmxbs in outburst ( * ? ? ?
* and references therein ) .
the phenomenon has also been seen in persistently bright systems .
the superhump period is typically a few percent longer than @xmath12 .
the current understanding is that in systems where the mass ratio is less than @xmath60.25 , the tidal truncation radius lies outside the 3:1 resonance radius .
when the accretion disc radius extends out to beyond the 3:1 resonance radius , the disc becomes eccentric , and the perturbation from the secondary causes a prograde precession . for precessing discs in lmxbs , @xcite proposed that the disc area changes on the superhump period , causing the area of the disc intercepting the x - rays to increase .
the increased reprocessing causes the disc to brighten , giving rise to the observed superhumps in lmxbs .
several systems exhibit superhumps with almost exact sinusoidal modulations like those seen in hete j1900.1@xmath02455 , in particular another lmxb , xte j1118@xmath10480 @xcite .
the peak - to - peak amplitude of the modulation seen in hete j1900.1@xmath02455 was @xmath60.1 mag on 2006 september 21 and 2007 june 13 , but increased to @xmath60.3 mag on 2007 june 24 .
the latter amplitude is quite high , but as pointed out by @xcite superhumps due to irradiated discs are likely to appear more prominent at _ low _ inclinations : hence the large superhump amplitude supports our estimate of the inclination in section 3.2.1 . for an x - ray luminosity of @xmath610@xmath18 erg
s@xmath3 the disc is likely to also suffer irradiation - driven warping ( * ? ? ? *
in particular their model number 1 ) and the disc shape continuously changes under the combination of prograde apsidal precession and retrograde warped precession .
such models may also explain the factor of @xmath63 increase in modulation amplitude over a relatively short timescale ( haswell , private communication ) .
the location of the hotspot in the h@xmath2 doppler tomogram may be interpreted in terms of this irradiated precessing disc model : as the warped disc precesses , the gas stream / accretion disc impact point varies .
how this occurs can be seen clearly in the @xcite simulations .
these simulations suggest that multi - epoch doppler tomography of hete j1900.1@xmath02455 should reveal a hotspot whose location varies between the ( @xmath68,@xmath78 ) and ( @xmath68,@xmath69 ) quadrants .
previous work has shown that , once the superhump period ( @xmath79 ) is identified , then the period excess , @xmath80 (= ( @xmath81)/@xmath12 ) , can be directly related to the mass ratio @xcite .
a fourier transform of our photometry , however , does not allow us to identify a unique superhump period , because our sampling is too sparse . a more comprehensive photometry campaign is required to measure @xmath79 , and estimate @xmath65 . using the empirical relationship between the absolute visual magnitude of an irradiated accretion disc ( @xmath82 ) , the x - ray luminosity of the central source and orbital period @xcite , we find @xmath83 mag . at a distance of 4.3 kpc @xcite
, such a disc should have an apparent magnitude of 16.7 @xmath13 0.5 mag .
this is consistent with the measured @xmath30-band magnitude of @xmath617.1 mag , implying that the optical light , in the @xmath30-band , is dominated by an x - ray heated accretion disc .
thus , x - ray heating of the disc can explain the observed optical brightness of hete j1900.1@xmath02455 .
@xcite estimated the mass of the secondary to be between 0.016 and 0.07 m@xmath4 , for primary masses between 1.4 and 2.2 m@xmath4 , but this was based on an assumption of inclination angle : 90 for the lower mass limit , and a uniform a priori distribution for the 95 per cent confidence upper limit .
we have seen that for @xmath61 @xmath7 200 km s@xmath3 , @xmath84 20 ( see fig . [ kvi ] ) . at this upper limit of @xmath61 , @xmath72 , and
the minimum value of @xmath85 ( for @xmath86 m@xmath4 ) is therefore 0.05 m@xmath4 . @xcite
have computed a mass - radius relation for the secondaries in short period cataclysmic variables ( cvs ) , finding that in general , the mass - radius curve lies above that for zero - age main sequence stars .
an empirical equation relates @xmath85 to @xmath12 : @xmath87 m@xmath4 , where @xmath2 is the factor relating the radius of the secondaries in these cvs to that of main sequence stars with the same mass , and @xmath12 is in hours .
@xcite find that @xmath2 @xmath11 1.18 .
assuming that the secondary in hete j1900.1@xmath02455 is similar to the secondary in these cv systems , this equation predicts @xmath88 m@xmath4 . with the lower limit on @xmath65 of 0.035 ,
this sets an upper limit to the primary mass of @xmath89 m@xmath4 .
the secondary may also be a brown dwarf , in which case @xmath85 is poorly constrained , as is @xmath75 . finally , we briefly examine the radial velocity of hete j1900.1@xmath02455 relative to the local galactic radial velocity . based on its position and estimated distance , hete
j1900.1@xmath02455 is @xmath63.6 kpc from the galactic centre , @xmath61 kpc below the galactic plane . using the galactic rotation curve of @xcite
, we find that the radial velocity of hete j1900.1@xmath02455 is @xmath690 km s@xmath3 relative to the local galactic rotation .
combined with future proper motion studies , this velocity may be used to derive the space velocity of hete j1900.1@xmath02455 and compute its orbit around the galactic centre ( e.g. * ? ? ?
we have presented a detailed optical study of the amsp hete j1900.1@xmath02455 .
the @xmath1-band light curves reveal a modulation whose phase is inconsistent with simple heating of the secondary star / accretion disc : we interpret this as due to superhumps associated with a precessing accretion disc .
fitting to the wings of the averaged h@xmath2 emission line gives the systemic velocity of 132 @xmath13 8 km s@xmath3 , and a similar value is found from the he@xmath23ii line .
doppler tomography of the h@xmath2 emission line suggests a upper limit of @xmath6200 km s@xmath3 on the projected secondary velocity , implying that hete j1900.1@xmath02455 has a low inclination , @xmath5 20 , and a minimum secondary mass of 0.05 m@xmath4 . assuming that the secondary in hete j1900.1@xmath02455 is similar to those seen in short period cvs
, we estimate a secondary mass of 0.085 m@xmath4 , which implies an upper limit to the ns mass of 2.4 m@xmath4 ( again for @xmath61 @xmath5 200 km s@xmath3 ) . if the secondary is a brown dwarf , then this constraint is relaxed .
we acknowledge the use of the 2.3 m advanced technology telescope of the australian national university at siding spring observatory .
the wiyn observatory is a joint facility of the university of wisconsin - madison , indiana university , yale university , and the national optical astronomy observatory .
some of the data presented herein were acquired using the large binocular telescope ( lbt ) .
the lbt is an international collaboration among institutions in the united states , italy and germany .
lbt corporation partners are : the university of arizona on behalf of the arizona university system ; istituto nazionale di astrofisica , italy ; lbt beteiligungsgesellschaft , germany , representing the max - planck society , the astrophysical institute potsdam , and heidelberg university ; the ohio state university , and the research corporation , on behalf of the university of notre dame , university of minnesota and university of virginia .
we thank lbt director r. green for allocation of dd time , and d. thompson , lbto for acquiring some of the data .
the spectroscopic data presented herein were obtained at the w. m. keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w. m. keck foundation .
this research made use of nasa s astrophysics data system , and the simbad database , operated at cds , strasbourg , france .
x - ray quick - look results provided by the asm / rxte team .
we acknowledge the use of and software packages developed by t. r. marsh , university of warwick .
we thank j. x. prochaska , uco / lick observatory , for reducing our keck spectroscopy using .
pe and pjc acknowledge support from science foundation ireland .
elebert p. , callanan p. j. , torres m. a. p. , 2006
, in meurs e. j. a. , fabbiano g. , eds . , proceedings of the 230th symposium of the international astronomical union , cambridge university press , pp .
5758 patterson j. , thorstensen j. r. , kemp j. , skillman d. r. , vanmunster t. , harvey d. a. , fried r. a. , jensen l. , cook l. m. , rea r. , monard b. , mccormick j. , velthuis f. , walker s. , martin b. , bolt g. , pavlenko e. , odonoghue d. , gunn j. , novk r. , masi g. , garradd g. , butterworth n. , krajci t. , foote j. , beshore e. , 2003 , pasp , 115 , 1308 patterson j. , kemp j. , harvey d. a. , fried r. e. , rea r. , monard b. , cook l. m. , skillman d. r. , vanmunster t. , bolt g. , armstrong e. , mccormick j. , krajci t. , jensen l. , gunn j. , butterworth n. , foote j. , bos m. , masi g. , warhurst p. , 2005 , pasp , 117 , 1204 ragazzoni r. , giallongo e. , pasian f. , baruffolo a. , bertram r. , diolaiti e. , di paola a. , farinato j. , gentile g. , hill j. , lombini m. , pedichini f. , speziali r. , smareglia r. , vernet e. , 2006 , in stepp , l. m. , ed . , society of photo - optical instrumentation engineers ( spie ) conference , volume 6267 , ground - based and airborne telescopes suzuki m. , kawai n. , tamagawa t. , yoshida a. , nakagawa y. e. , tanaka k. , shirasaki y. , matsuoka m. , ricker g. r. , vanderspek r. , butler n. , lamb d. q. , graziani c. , pizzichini g. , sato r. , arimoto m. , kotoku j. , maetou m. , yamauchi m. , 2007 , pasj , 59 , 263 vogt s. s. , allen s. l. , bigelow b. c. , bresee l. , brown b. , cantrall t. , conrad a. , couture m. , delaney c. , epps h. w. , hilyard d. , hilyard d. f. , horn e. , jern n. , kanto d. , keane m. j. , kibrick r. i. , lewis j. w. , osborne j. , pardeilhan g. h. , pfister t. , ricketts t. , robinson l. b. , stover r. j. , tucker d. , ward j. , wei m. z. , 1994 , spie , 2198 , 362 | we present phase resolved optical photometry and spectroscopy of the accreting millisecond pulsar hete j1900.1@xmath02455 .
our @xmath1-band light curves exhibit a sinusoidal modulation , at close to the orbital period , which we initially attributed to x - ray heating of the irradiated face of the secondary star .
however , further analysis reveals that the source of the modulation is more likely due to superhumps caused by a precessing accretion disc .
doppler tomography of a broad h@xmath2 emission line reveals an emission ring , consistent with that expected from an accretion disc . using the velocity of the emission ring as an estimate for the projected outer disc velocity , we constrain the maximum projected velocity of the secondary to be 200 km s@xmath3 , placing a lower limit of 0.05 m@xmath4 on the secondary mass . for a 1.4 m@xmath4 primary
, this implies that the orbital inclination is low , @xmath5 20 . utilizing the observed relationship between the secondary mass and orbital period in short period cataclysmic variables , we estimate the secondary mass to be @xmath60.085 m@xmath4 , which implies an upper limit of @xmath62.4 m@xmath4 for the primary mass .
[ firstpage ] accretion , accretion discs techniques : photometric techniques : spectroscopic pulsars : individual : hete j1900.1@xmath02455 x - rays : binaries |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.